1A LEGEND

HYDROGEN 1.0079

G.l* H 1 Name Symbol Mass Number

I
ts' Density (gm cm-3) MAGNESIUM/204.305
3.71 HEX (Common Crystal Phase) ..........

14.0 110
2A 1.74 Mg 12
r-- Atomic Number
Lattice Constant, a ""- 2 ~ Atomic Configuration
[Ne]3s -
UTHIUM 6.9041 BERYLLIUM 9.0122 (A)
3,21 HEX
O.A U 3 1.15 Be 4

2
Melting Temperature _
922 t 318 t-- Mean Debye Temperature (LTSignifies a Low Temperature Determination)
ls225' ls~,z (K)
I
1.48 BeC 2.28 HEX Crystal Type (Common Crystal Phase)

453 400 1550 1000 FCC - Face-eentered Cubic ORC - Orthorhombic

SODIUM 22.9898 MAGNESIUM 24.305 BeC - Body-eentered Cubic HEX - Hexagonal
SC - SlmpleCublc DIA - Diamond
0.17 Na 11 1.74 Mg 12
CUB- Cubic RHl - Rhombohedral
3 [Ne]3s' [Ne] 3,z TET - Tetragonal MeL - Monoclinic

4.23

371.0
BeC
150 922
3.21 HEX

318
38 48 58 8B 78 . 8
POTASSIUM 39.09 CALCIUM 40.06 SCANDIUM 44.956 mANIUM 47.90 VANADIUM 50.942 CHROMIUM 52.00 MANGANESE 504.938 IRON 55.85 COBALT 58.93

0... K 18 1.14 Ca 20 2.. Se 21 4.51 TI 22 1.1 V 23 7.18 Cr 24 7.43 Mn 25 7." Fe 261.8 Co 27

4 [Ar]4s' [Ar]4s2 [Ar]3d'4s 2 [Ar] 3d 24s2 [Ar] 3d s4s 2 [Ar]3cf54s 1 [Ar] 3d 54,z [Ar]3cf14s2 [Ar] 3d 74,z
In BCC u. FCC 3,31 HEX 2.11 HEX 3.02 BeC 2M Bce .... CUB 2.17 BeC 2.51 HEX

337 100 1111 230 1812 359 LT 1933 380 2163 390 2130 460 1518 400 1808 420 1768 3B5
RUBIDIUM 85.47 STRONTIUM 87.62 YTTRIUM 88.91 ZIRCONIUM 91.22 NIOBIUM 92.91 MOl YBOENUM 95.94 TECHNETIUM 98.91 RUTHENIUM 101.07 RHODIUM 102.90

1.u Rb 37 2.10 Sr 38 UI Y 3St •.• Zr 401.4 Nb 41 10.2 Mo 42 11.5 Tc 4312.2 Ru 44 12.4 Rh 45

5 [Kr]Ss' [Kr] s,z [Kr]4d'5s 2 [Kr] 4dZS,z [Krj4(145s ' [Kr] 4d 55s' [Kr] 4d 5SS2 [Kr]4d 7Ss' [Krj4deSs'
5,H BeC '.01 FCC 3M HEX 3,23 HEX 3.30 acc 3.11 acc 2.74 HEX 2.70 HEX 3.10 FCC
T 212S
312 56 LT 1043 14rtT 1796 2set 250 2741 275 2890 380 2445 2583 J82lT 2239 35()lT

CESIUM 132.91 BARIUM 137.304 lANTHANUM 138.91 HAFNIUM 178.49 TANTAlUM 180.95 TUNGSTEN 183.85 RHENIUM 186.2 OSMIUM 190.20 IRIDIUM 192.22
(WOLFRAM)
1.10 C. 553.5 Sa 58 '.17 La 57 13.1 Hf 721... Ta 7318.3 W 74 21.0 R. 75 22.1 Oa 78 22.5 Ir rr
6 [Xe]6s' [Xe] 6s 2 [Xe]5~6s2 <lIlI [Xe] 4"45cf6s 2 [Xe) 4f'45d3E),z [Xe] 4f'4Sd~S2 [Xe) 4f'45dS&S2 [Xe] 4,,4Sd'6s2 [Xef 4f'~d76s2

1.05 BeC 5.02 BeC 3.75 HEX 3.20 HEX 3,31 BeC 3.1' acc 2.76 HEX 2.74 HEX 3.14 FCC

302 4QlT 998 110&-T 1193 132 2495 225 3683 310 3453 416 LT 3318 4()()lT 2683 430
FRANCIUM 223 RADIUM 226 ACTINIUM 227

Fr 87 (5.0) Ra 88 10.1 Ac ~
7 [Rn]7s' [Rn]7s 2 [Rn]6d'7s 2

(300)
(BeC)
973
1.31

1323
FCC
... RARE EARTHS
CERIUM 140.12 PRASEODYMIUM NEODYMIUM 14-4.24 PROMETHIUM 145 SAMARIUM 150.35
140.91
~ ~ 58~ ~ H~ ~ H ~ ~~ ~ ~

LANTHANIDES 6 [Xe] 4fscfl6s 2 [Xe]4f~£106s2 [Xe] 4f 4SdOSs2 [Xe] 4f 55cfl6s 2

[Xe] 4f'S(106,z

5.1' FCC 3.17 HEX 3." HEX '.00 RHL

1071 1JgLT 1204 152 LT 1283 15~T (1350) 1345 166
THORIUM 232.04 PROTACTINIUM 231 URANIUM 238.03 NEPTUNIUM 237.05 PLUTONIUM 244

11.7 Th 80 15.4 Pa 81 18.07 U 82 20.3 Np 93 11.8 Pu M

ACTINIDES 7 [Rn]6d~s2 [Rn} Sf3&d'7s 2 [Rn] SfSSdo7s2

5.oe FCC 2.15 ORC 4.72 ORC

2020 100 1470 1406

 NOBLE
ELEMENTS
HELIUM 4.0026

0.171 He 2

15 2
3.17 HEX
3A 4A SA 6A 7A
-1.0 (26 Aim) 26 LT
BORON 10.81 CARBON 12.01 NITROGEN 14.007 OXYGEN 15.999 FLUORINE 18.998 NEON 20.18

2.34 B 5 2.21. C 8 1.03 N 7 1.43 0 8 U7{a) F 8 1.1. Ne 10

15 22r2p' ls22s~ ls22r2 p 3 lr2s~4 1$~s22p5 ls 22522'"

.. 73 TET 3.17 DIA 4.031 HEX 1.13 CUB Mel 4,43 FCC

2600 1250 (4300) 1860 63.3 (~)7g..T 54.7 (y)46LT 53.S 24.5 63
ALUMINUM 26.982 SILICON 28.086 PHOSPHORUS 30.974 SULFUR 32.084 CHLORINE 35.453 ARGON 39.iW8

2.70 AI 13 2,33 SI 141.82(.....)P 15 2.07 S 18 2.oe CI 17 1.71 Ar 1.

[Ne] 3s 23p' [Ne]3s 23p2 [Ne] 3s 23p 3 [Ne] 3s 23p4 [Ne] 3s 23p5 [Ne]3s23p'

. 1B 2B
4.01

933
FCC 1.43

394 1683
DIA 7.17
625 317.3
CUB 10.47 ORC

386
2".
1. • 1.24 ORC
172.2
1.324
0.71'
I .• FCC
83.9 85
NICKEL 58.71 COPPER 63.55 ZINC 65.38 GAllIUM 69.72 GERMANIUM 72.59 ARSENIC 74.922 SELENIUM 78.96 BROMINE 79.91 KRYPTON 83.80
... HI 28 .... Cu 28 7.14 Zn 30 U1 Ga 31 l.az Ge 32 1.72 Aa 33 4.71 se 34 4.10 Br 35 3.07 Kr 31

[Ar}3d'4s2 [Ar] 3d,04s 1 [Ar] 3d,04s2 [Ar] 3d 104s23p' [Ar) 3d,04r4'' [Ar] 3d,04r4p3 [Ar) 3d,04s24Jt [Ar] 3d,04s24p5 [Ar]3d'04s24p'
3.12 FCC 3.11 FCC 2.• HEX 4.11 ORC I .• DIA 4.13 RHL 4.11 HEX • .17 ORC 5.72 FCC
1726~ 375 1356 315 893 234 303 240 1211 360 1090 285 490 15QlT 266 116.5 73 LT
PAlLADIUM 106.40 SILVER 107.87 CADMIUM 112.40 INDIUM 114.82 TIN 118.69 ANTIMONY 121.75 TELLURIUM 127.60 IODINE 126.90 XENON 131.30

12.0 Pd 4I1o.a Ag 47 .... Cd 41 7.31 In 49 7.30 Sn 50 U2 Sb 51 1.24 Te 52 4JM I 53 3.77 Xe 54

[Kr] 4d,05so [Kr]4d'OSs' [Kr] 4d 105s2 [Kr] 4dl05s~p' [Kr) 4d'OSs25" [Kr] 4d 1OSs25p 3 [Kr}4d'05rSp4 [Kr] 4dl05s~p5 [Kr] 4d'OSS25p'
3. FCC 4.oe FCC 2.• HEX 4.H TET 1.12 TET 4.11 RHl 4.... HEX 7Z7 ORC '.20 FCC
1825 275 1234 215 594 120 429.8 129 505 170 904 200 723 l:J9lT 387 161.3 5SLT
PLATINUM 195.09 GOlD 196.97 MERCURY 200.59 THALLIUM 204.31 lEAD 207.19 BISMUTH 208.98 POlONIUM 210 ASTATINE 210 RADON 222
21.4 PI 7811.3 Au 78 13.' Hg 80 11.11 T1 81 11.4 Pb 82.., BI 831.4 Po 84 At 85 (4,4) An 81
[Xe] 4f 145d'OSso (Xe]4f I4Sd1OSs' [Xe] 4f 14Sd'OSS2 [Xe] 4f'45d'OSs2E)p' [Xe}4f'4Sd 1OSs2E)p2 [Xe]4f'45d'OSs2Eip3 [Xe] 4f'45dlOSS~Jt [Xe] 4fI45d'OSs~p5 [Xe] 4fI45d'OSs~'"
3.12 FCC 4.oe FCC 2.• RHl 3M HEX 4.11 FCC 4.75 RHL 3.35 SC (FCC)
2045 230 1337 170 234.3 100 sn 96 601 88 544.S 120 S27 (57S) (202)

EUROPIUM 151.96 GADOLINIUM 151.25 TERBIUM 158.92 DYSPROSIUM 162.50 HOlMIUM 164.93 ERBIUM 187.26 THULIUM 168.93 YTTERBIUM 173.04 LUTETIUM 174.97

7.10 Eu 53 1.23 Get 84 1.54 Tb 15 '.71 Dy 66 '.01 Ho 67 1.37 Er 68 1.31 Tm 68 1.17 Vb 70 ..,4 Lu 71

[Xe) 4f 7SdOSr [Xe) 4f 7Sd16s2 [Xe) 4f 1l5dOSs2 [Xe] 4f,oSdOSs2 [Xe] 4f115~6r [Xe] 4f 12SdOSs2 [Xe] 4r~do6s2 [Xe) 4"45~6s2 [Xe} 4f'4Sd'6s2
4.11 BeC 3..14 HEX 1.80 HEX 3.51 HEX U8 HEX 3.1. HEX 3.14 HEX I .• FCC 3.51 HEX
1095 10rL T 1585 1761.T 1633 1~T 1680 186 LT 1743 191 LT 1795 195 LT 1818 2O()lT 1097 llBl T 1929 207 LT
AMERICIUM 243 CURIUM 247 BERKELIUM 241 CAliFORNIUM 251 EINSTEINIUM 254 FERMIUM 257 MENDELEVIUM 256 NOBELIUM 254 LAWRENCIUM 257

11.' Am 85 Cm 88 Bk 87 Cf 81 e- 88 Fm 100 Md 101 No 102 Lw 103

[Rn] Sf 76cJ&7s 2 [Rn} 5f 76cJ17s 2 [Rn]5f76d~s2 [Rn] 5faad'7r

1267 1600
 INTRODUCTORY
QUANTUM MECHANICS
FOURTH EDITION

Richard L. Liboff
Distinguished Professor ofPhysics
University of Central Florida

A
~
Addison

-- Wesley

San Francisco Boston New York

Capetown Hong Kong London Madrid Mexico City
Montreal Munich Paris Singapore Sydney Tokyo Toronto
Acquisitions Editor: Adam Black
Project Editor : Nancy Gee
Production Editor : Joan Marsh
Text Designer: Leslie Galen
Cover Illustrator: Kenneth X. Probst
Cover Designer: Blakeley Kim
Marketing Manager: Christy Lawrence
Manufacturing Coordinator: Vivian McDougal
Project Coordination and Electronic Page Makeup : Integre Technical Publishing Co., Inc.

Fifth printing with revisions

Copyright © 2003 Pearson Education, Inc., publishing as Addison Wesley, 1301 Sansome St., San
Francisco, CA 94111. All rights reserved . Manufactured in the United States of America. This
publication is protected by Copyright and permission should be obtained from the publi sher prior to
any prohibited reproduction, storage in a retrieval system , or transmission in any form or by any
means, electronic, mechanical, photocopying, recording , or likewise. To obtain pennission(s) to use
material from this work, please submit a written request to Pearson Education , Inc., Permissions
Department, 1900 E. Lake Ave., Glenview, IL 60025 . For information regarding permissions, call
847/486/2635.

Liboff, Richard L.
ISBN 0-8053-8714-5 (hardcover)

ISBN 0-8053-8714-5

TT
Addison
.
- Wesley
6 7 8 9 10 -MAL- 10 09 08 07 06
www.aw-bc.com/physics
 "I do not know what I may appear to the world; but to myself I seem
to have been only like a boy playing on the seashore , and diverting
myself in now and then finding a smoother pebble or a prettier shell
than ordinary, whilst the great ocean of truth lay all undiscovered
before me."

This statement by Isaac Newton shortly before his death in 1727 eloquently
reflects the sentiments of all mature scientists from the ancient past to the present.
To Myra
"She openeth her mouth with wisdom; and in her tongue is the law of
kindness . . . ."
 Preface

Since earlier editions of this text, it remains the case that physics continues to
evolve in esoteric and pragmatic directions . In the present edition, a new chapter
addressing quantum computing has been added that well represents this theme,
as quantum computing is founded on basic elements of quantum mechanics but
is thought to represent a new concept for computers. Components of this chapter
include : Binary Numbers ; Logic Gates; Turing Machine and Complexity Classes;
Qubits and Quantum Logic Gates. The chapter concludes with a description of
Grover's algorithm, which comes into play in the 'search problem .'
Twenty seven problems are included in this new chapter, many of which, in
the spirit of earlier editions, carry solutions. Some of the topics included in these
problems are: Classical and quantum logic gates, Boolian relations , Factoring
problems and Euclid's algorithm . There are a total of 870 problems in this edition.
A new appendix is included in this edition that describes the Harmonic Oscilla-
tor in Spherical Coordinates and a number of inserts are included in the appendix
on Physical Constants and Equivalence Relations .
A total of nine new problems has been added to previous chapters that address
in part: Properties of the commutator, reduced form of the square of angular mo-
mentum, parallel relations for the cubical quantum box, quantum confinement.
A number of corrections have been made throughout the text, mostly due to
input from students and teachers throughout the world, whose suggestions I take
pleasure in acknowledging.
I take this opportunity to express my deep gratitude to the many individuals
throughout the world who have communicated with me regarding typos and sug-
gestions for this text. More locally, in addition to all other colleague s who have
contributed to the success of this book, the following individuals have proved to be
of invaluable assistance in preparation of this new edition: Toby Berger, Bradley
Minch, Rajit Manohar, Eric Sakk, Ian Rippke, Brian E. Moritti, James Snyder,
Andy Martwick, and Igor Devetak. My special thanks goes to David Mermin for
his expert assistance in the preparation of the new chapter.
It is my pleasure also to declare again my appreciation to the many individuals
who have taught from prior editions of this work and the many who have leamed
from it. I trust that these kind individuals will find this new edition equally re-
warding.

Ithaca, 2002 R.L. Liboff

lY'~?~,n

v
 Piping down the valleys wild,
Piping songs ofpleasant glee,
On a cloud I saw a child,
And he laughing said to me;
"Pipe a song about a Lamb! "
So I piped with merry chear.
"Piper, pipe that song again;"
So I piped; he wept to hear.
"Drop thy pipe, thy happy pipe;
Sing thy songs of happy chear; "
So I sung the same again,
While he wept with joy to hear.
"Piper, sit thee down and write
In a book, that all may read."
So he vanish'd from my sight,
And I pluck'd a hollow reed,
And I made a rural pen,
And I stain 'd the water clear,
And I wrote my happy songs
Every child may joy to hear.

Introduction to Songs of Innocence

-William Blake (1757-1827)
 Contents

Preface vii

Topical Problems xvii

PART I • ELEMENTARY PRINCIPLES AND APPLICATIONS TO

PROBLEMS IN ONE DIMENSION 1

1 • Review of Concepts of Classical Mechanics 3

1.1 Generalized or "Good" Coordinates 3
1.2 Energy, the Hamiltonian, and Angular Momentum 6
1.3 The State of a System 19
1.4 Properties of the One-Dimensional Potential Function 24

2 • Historical Review: Experiments and Theories 30

2.1 Dates 30
2.2 The Work of Planck. Blackbody Radiation 31
2.3 The Work of Einstein. The Photoelectric Effect 36
2.4 The Work of Bohr. A Quantum Theory of Atomic States 39
2.5 Waves versus Particles 43
2.6 The de Broglie Hypothe sis and the Davisson-Germer
Experiment 46
2.7 The Work of Heisenberg. Uncertainty as a Cornerstone of Natural
Law 53
2.8 The Work of Born. Probability Waves 55
2.9 Semiphilosophical Epilogue to Chapter 2 57
viii Contents

3 • The Postulates of Quantum Mechanics. Operators,

Eigenfunctions, and Eigenvalues 68
3.1 Observables and Operators 68
3.2 Measurement in Quantum Mechanics 74
3.3 The State Function and Expectation Values 76
3.4 Time Development of the State Function 80
3.5 Solution to the Initial-Value Problem in Quantum Mechanics 84

4 • Preparatory Concepts. Function Spaces and

Hermitian Operators 90
4.1 Particle in a Box and Further Remarks on Normalization 90
4.2 The Bohr Correspondence Principle 94
4.3 Dirac Notation 97
4.4 Hilbert Space 98
4.5 Hermitian Operators 104
4.6 Properties of Hermitian Operators 108

5 • Superposition and Compatible Observables 115

5.1 The Superposition Principle 115
5.2 Commutator Relations in Quantum Mechanics 130
5.3 More on the Commutator Theorem 137
5.4 Commutator Relations and the Uncertainty Principle 140
5.5 "Complete" Sets of Commuting Observables 143

6 • Time Development, Conservation Theorems, and Parity 152

6.1 Time Development of State Functions 152
6.2 Time Development of Expectation Values 168
6.3 Conservation of Energy, Linear and Angular Momentum 171
6.4 Conservation of Parity 176

7 • Additional One-Dimensional Problems. Bound and

Unbound States 187
7.1 General Properties of the One-Dimensional Schrodinger
Equation 187
7.2 The Harmonic Oscillator 190
7.3 Eigenfunctions of the Harmonic Oscillator Hamiltonian 198
7.4 The Harmonic Oscillator in Momentum Space 211
7.5 Unbound States 216
7.6 One-Dimensional Barrier Problems 222
 Contents ix

7.7 The Rectangular Barrier. Tunneling 228

7.8 The Ramsauer Effect 235
7.9 Kinetic Properties of a Wave Packet Scattered from a Potential
Barrier 241
7.10 The WKB Approximation 243
7.11 Principle of Least Action and Feynman's Path Integral
Formulation 268

8 • Finite Potential Well, Periodic Lattice, and Some Simple

Problems with Two Degrees of Freedom 278
8.1 The Finite Potential Well 278
8.2 Periodic Lattice. Energy Gaps 289
8.3 Standing Waves at the Band Edges 307
8.4 Brief Qualitative Description of the Theory of Conduction in
Solids 313
8.5 Two Beads on a Wire and a Particle in a Two-Dimensional Box 317
8.6 Two-Dimensional Harmonic Oscillator 324
8.7 Linear Combination of Atomic Orbitals (LCAO)
Approximation 331
8.8 Density of States in Various Dimensions 336

PART II • FURTHER DEVELOPMENT OF THE THEORY

AND APPLICATIONS TO PROBLEMS IN
THREE DIMENSIONS 347

9 • Angular Momentum 349

9.1 Basic Properties 349
9.2 Eigenvalues of the Angular Momentum Operators 358
9.3 Eigenfunctions of the Orbital Angular Momentum Operators
L2 and i; 367
9.4 Addition of Angular Momentum 386
9.5 Total Angular Momentum for Two or More Electrons 396

10 • Problems in Three Dimensions 404

10.1 The Free Particle in Cartesian Coordinates 404
10.2 The Free Particle in Spherical Coordinates 410
10.3 The Free-Particle Radial Wavefunction 415
10.4 A Charged Particle in a Magnetic Field 430
10.5 The Two-Particle Problem 436
10.6 The Hydrogen Atom 446
x Contents

10.7 Elementary Theory of Radiation 463

10.8 Thomas-Fermi Model 472

11 • Elements of Matrix Mechanics. Spin Wavefunctions 480

11.1 Basis and Representations 481
11.2 Elementary Matrix Properties 488
11.3 Unitary and Similarity Transformations in Quantum Mechanics 492
11.4 The Energy Representation 499
11.5 Angular Momentum Matrices 504
11.6 The Pauli Spin Matrices 512
11.7 Free-Particle Wavefunctions, Including Spin 517
11.8 The Magnetic Moment of an Electron 519
11.9 Precession of an Electron in a Magnetic Field 527
11.10 The Addition of Two Spins 536
11.11 The Density Matrix 543
11.12 Other "Pictures" in Quantum Mechanics 553
11.13 Polarization States. EPR Revisited 558
11.14 The Transfer Matrix 571

12 • Application to Atomic, Molecular, Solid-State, and

Nuclear Physics. Elements of Quantum Statistics 579
12.1 The Total Angular Momentum , J 579
12.2 One-Electron Atoms 584
12.3 The Pauli Principle 597
12.4 The Periodic Table 602
12.5 The Slater Determinant 612
12.6 Application of Symmetrization Rules to the Helium Atom 614
12.7 The Hydrogen and Deuterium Molecules 623
12.8 Brief Description of Quantum Models for Superconductivity and
Superfluidity 630
12.9 Impurity Semiconductors and the p-n Junction 641
12.10 Elements of Nuclear Physics. The Deuteron and Isospin 669

13 • Perturbation Theory 681

13.1 Time-Independent, Nondegenerate Perturbation Theory 681
13.2 Time-Independent, Degenerate Perturbation Theory 692
13.3 The Stark Effect 700
13.4 The Nearly Free Electron Model 703
13.5 Time-Dependent Perturbation Theory 709
13.6 Harmonic Perturbation 712
 Contents xi

13.7 Application of Harmonic Perturbation Theory 718

13.8 Selective Perturbations in Time 727
13.9 Atom-Radiation Interaction 739
13.10 Hartree-Fock Model 757

14 • Scattering in Three Dimensions 762

14.1 Partial Waves 762
14.2 S-WaveScattering 770
14.3 Center-of-Mass Frame 774
14.4 The Born Approximation 777
14.5 Atomic-Radiative Absorption Cross Section 782
14.6 Elements of Formal Scattering Theory. The Lippmann-Schwinger
Equation 785

15 • Relativistic Quantum Mechanics 793

15.1 Preliminary Remarks 793
15.2 Klein-Gordon Equation 798
15.3 Dirac Equation 800
15.4 Electron Magnetic Moment 806
15.5 Covariant Description 810

16 • Quantum Computing 817

16.1 Binary Number System 817
16.2 Logic Gates 823
16.3 Turing Machine and Complexity Classes 830
16.4 Qubits and Quantum Logic Gates 832

List of Symbols 843

APPENDIXES 847

A • Additional Remarks on the xand pRepresentations 849

B • Spin and Statistics 853

C • Representations of the Delta Function 857

D • Differential Vector Relations 861

xii Contents

E • Harmonic Oscillator in Spherical Coordinates 865

F • Physical Constants and Equivalence Relations 867

Index 871
 List of Tables

1.1 Hamiltonian of a free particle in three coordinate frames 17

3.1 Examples of operators 69
7.1 The first six eigenenergies and eigenstates of the simple
harmonic oscillator Hamiltonian 201
7.2 Transmission coefficients for three elementary potential barriers 237
7.3 Properties of Airy functions 251
9.1 The first few normalized spherical harmonics and corresponding
associated Legendre functions 373
9.2 Properties of the Legendre polynomials 374
9.3 Properties of the associated Legendre functions 375
9.4 Normalized relations between 1.+, 1._ , Lx, Ly and the states 11m) 379
9.5 Nine common eigenstates Ilm11/2) of the operators £.2, t.; 1. 12,
2
1. 2 for two p electrons 394
10.1 Properties of the spherical Bessel and Neumann functions 418
10.2 Solutions to the three fundamental box problems in quantum
mechanics 424
10.3 Eigenfunctions of hydrogen in terms of associated Laguerre
polynomials 450-451
10.4 Allowed values of I and mi for n = 1,2, 3 451
10.5 Several normalized time-independent eigenstates of hydrogen 455
11.1 Matrix properties 490
11.2 Spin wavefunction for two electrons in the uncoupled representation 536
11.3 Spin wavefunctions for two electrons in the coupled representation 539
11.4 Elements of Schrodinger and Heisenberg pictures 555
12.1 Diagrammatic enumeration of states available in the first three
atomic shells 603
12.2 Distribution of electrons in the atoms from Z = 1 to Z = 36 605-606
12.3 States allowed by the exclusion principle of the L-S coupling scheme 607
12.4 The periodic table 609
12.5 Properties of the nuclear component wavefunctions for D2 and H2 628
12.6 Semiconductor relations 654
12.7 Two-nucleon coupled isospin states 675
13.1 Examples of perturbation Hamiltonians 682
13.2 Transition probabilities for time-dependent perturbations 734
16.1 Boolean postulates in 0 and 1 827
16.2 Boolean postulates in 0, 1 and A 827
16.3 Boolean postulates in A, Band C 827
xiii
xiv List of Tables

Threshold frequencies 38
Connection formulas 253
First few zeros of in (p) 425
Example of basis 481
Term notation of orbital angular momentum 582
Separation of states 627
First four nP - IS transitions 747
 Topical Problems

1.21 Complex variables

1.23 Dumbbell molecule
1.24 Classical reversibility
2.28 Compton effect
2.40 Adiabatic expansion of the universe
2.45 Jellium and Wigner crystallization
2.46 Energy filter for neutron beam
2.49 Summational and product probabilities
3.6 Delta-function relations
4.28 Variational principle
5.43 Wronskian
5.46 Uncertainty relation in three dimensions
5.49 Nondegeneracy of one-dimensional bound states
6.28 Interlaced zeros of eigenstates
6.31 Ehrenfest's principle
6.33 Time reversibility
7.62 Fowler-Nordheim equation
7.63 a-particle decay
7.70,71,85 State of minimum uncertainty
7.78 Scattering amplitude
7.84 Complex potential
7.94 Potential well with no reflection
8.43 Supersymmetry
8.47 The linear triatomic molecular ion
8.57 Dyadic form of effective mass
9.16 Angular momentum-angle commutation relations
9.46 Parity of wavefunctions in spherical coordinates
9.47 x , y , z in spherical coordinates
9.53 The nonclassical notion of spin angular momentum
xv
xvi Topical Problems

9.54 Normalization of associated Legendre polynomials

9.55 Canonical properties of angular momentum components

10.11 Plane wave expansion in spherical harmonics

10.16,17 Quantum dot; quantum wire
10.31 Spherical well of finite depth
10.52-54 Selection rules
10.58 Runge-Lenz vector
10.61 Interacting spheres
10.67 Annihilation and creation operators for spherical Bessel functions
10.68 Property of radial bound states
10.70 Larmor radius and the corresondence principle
10.73, 11.90 Momentum operator in curvilinear coordin ates

11.65 Double S-G experiment

11.76 Rotation of spinors
11.81 EPR configuration
11.87 et seq. Heisenberg picture
11.91 Mass correction to S-G experiment
11.92,95 Spin states for three electrons
11.96 Lorentz transformation
11.97 Bohm correspondence
11.102 Interaction picture
11.105 Bell's inequality
11.113 Quantum description of the rigid sphere

12.13 Quantum defect

12.25 Hund 's rules
12.28 Ground state of helium
12.48 Quantization of circulation
12.50 Heisenberg model for ferrom agnetism
12.52 Klein-Gordon equation
12.53 States of the deuteron
12.55 Ground state of lithium
12.64 Ground state of H- ion, electronegativity
12.81 Earnshaw's theorem
12.84 Number-phase uncertainty, squeezed states
12.85 21-cm line of atomic hydrogen

13.11 Polarizability of hydrogen

13.37,38 Elements of second quantization
Topical Problems xvii

13.44 Radioactive tritium

13.48 Pauli equation
13.49 Radioactive decay of ell
13.53 Schmidt orthogonalization procedure
13.54 Neumann-Liouville expansion, time-ordering operator
13.56 Positronium atom in magnetic field
13.58 Fourier expansion of even and odd potentials
13.77 Degenerate material
14.11 Lippmann-Schwinger equation
15.18 Lorentz invarianceof proper time
15.22 Properties of r matrices in relativistic quantum mechanics
16.10 Particle spin and base 2,3 numbers.
 PART
I

Elementary Principles
and Applications to Problems
in One Dimension
CHAPTER

1 Review of Concepts of
Classical Mechanics

1.1 Generalized or "Good " Coordinates

1.2 Energy, the Hamiltonian, and Angular Momentum
1.3 The State ofa System
1.4 Properties ofthe One-D imensional Potential Function

This is a preparatory chapter in which we review fundamental concepts of clas-

sical mechanics important to the development and understanding of quantum
mechanics. Hamilton 's equations are introduced and the relevance of cyclic co-
ordinates and constants ofthe motion is noted. In discussing the state ofa system,
we briefly encounter our first distinction between classical and quantum descrip-
tions. The notions offorbidden domains and turning points relevant to classical
motion, which find application in quantum mechanics as well, are also described.
The experimental motivation and historical background of quantum mechanics
are described in Chapter 2.

1.1 • GENERALIZED OR "GOOD" COORDINATES

Our discussion begins with the concept of generalized or good coordinates.

A bead (idealized to a point particle) constrained to move on a straight rigid
wire has one degree offreedom (Fig. 1.1). This means that only one variable (or
parameter) is needed to uniquely specify the location of the bead in space. For the
problem under discussion, the variable may be displacement from an arbitrary but
specified origin along the wire.
A particle constrained to move on a flat plane has two degrees of freedom . Two
independent variables suffice to uniquely determine the location of the particle in
space. With respect to an arbitrary, but specified origin in the plane, such variables

I •
x =O x

FIGURE 1.1 A bead constrained to move on a straight wire has one degree of freedom.

3
4 Chapter 1 Review of Concepts of Classical Mechanics

y y

• (x, y)

x x

(a) (b)

FIGURE 1.2 A particle constrained to move in a plane has two degrees of freedom.
Examples of coordinates are (x, y) or (r, 8).

x=O

FIGURE 1.3 Two beads on a wire have two degrees of freedom. The coordinates Xl and
X2denote displacements of particles 1 and 2, respectively.

might be the Cartesian coordinates (x, y) or the polar coordinates (r, 8) of the
particle (Fig. 1.2).
Two beads constrained to move on the same straight rigid wire have two de-
grees of freedom. A set of appropriate coordinates are the displacements of the
individual particles (Xl, X2) (Fig. 1.3).
A rigid rod (or dumbbell) constrained to move in a plane has three degrees
of freedom. Appropriate coordinates are the location of its center (x, y) and the
angular displacement of the rod from the horizontal, 8 (Fig. 1.4).
Independent coordinates that serve to uniquely determine the orientation and
location of a system in physical space are called generalized or canonical or good
coordinates. A system with N generalized coordinates has N degrees offreedom.

FIGURE 1.4 A rigid dumbbell in a plane has three degrees of freedom. A good set of
coordinates are (x, y), the location of the center, and (), the inclination of the rod with the
horizontal.
Problems 5

The orientation and location of a system with, say, three degrees of freedom are
not specified until all three generalized coordinates are specified . The fact that
good coordinates may be specified independently of one another means that, given
the values of all but one of the coordinates, the last coordinate remains arbitrary.
Having specified (x, y) for a point particle in 3-space, one is still free to choose z
independently of the assigned values of x and y .

PROBLEMS

1.1 For each of the following systems, specify the number of degrees of freedom and a set
of good coordinates.
(a) A bead constrained to move on a closed circular hoop that is fixed in space.
(b) A bead constrained to move on a helix of constant pitch and constant radius.
(c) A particle on a right circular cylinder.
(d) A pair of scissors on a plane.
(e) A rigid rod in 3-space.
(f) A rigid cross in 3-space .
(g) A linear spring in 3-space.
(h) Any rigid body with one point fixed.
(i) A hydrogen atom .
(j) A lithium atom.
(k) A compound pendulum (two pendulums attached end to end) .
1.2 Show that a particle constrained to move on a curve of any shape has one degree of
freedom .
Answer
A curve is a one-dimensional locus and may be generated by the parameterized equa-
tions

x = x(I), Y = Y(I), Z = z(l)

Once the independent variable I) (e.g., length along the curve) is given, x , y , and Z are
specified.
1.3 Show that a particle constrained to move on a surface of arbitrary shape has two de-
grees of freedom .
Answer
A surface is a two-dimensional locus . It is generated by the equation

u(X,Y,Z)=o

Any two of the three variables x, y, Z determine the third . For instance, we may solve
for z in the equation above to obtain the more familiar equation for a surface (height z
at the point x, y) .

z=z(X ,y)

In this case, x and y may serve as generalized coordinates.

6 Chapter 1 Review of Concepts of Classical Mechanics

1.4 How many degrees of freedom does a classical gas composed of 1023 point particles
have?

1.2 • ENERGY, THE HAMILTONIAN, AND ANGULAR MOMENTUM

These three elements of classical mechanics have been singled out because they
have direct counterparts in quantum mechanics. Furthermore, as in classical me-
chanics, their role in quantum mechanics is very important.
Consider that a particle of mass m in the potential field V(x, y, z) moves on
the trajectory

x = x(t)

y = y(t) (1.1)

z = z(t)
At any instant t, the energy of the particle is

The velocity of the particle is v. Dots denote time derivatives. The force on the
particle F is the negative gradient of the potential.

(1.3)

The three unit vectors (ex, ey, ez) lie along the three Cartesian axes.
Here are two examples of potential. The energy of a particle in the gravitational
force field,

F = -ezmg = -Vmgz
is

(1.4)

The particle is at the height z above sea level. For this example,

V =mgz

An electron charge q and mass m, between capacitor plates that are maintained
at the potential difference <1>0 and separated by the distance d (Fig. 1.5), has po-
tential
q<l>o
V=--z
d
 1.2 Energy, the Hamiltonian, and Angular Momentum 7

1
L_-_J~o
FIGURE 1.5 Electron in a uniform capacitor field.

The displacement of the electron from the bottom plate is z. The electron's energy
is

1 (x· 2 + y·2
E -- -m + Z.2) + --z
q <Po (1.5)
2 d
In both examples above, the system (particle) has three degrees of freedom.
The Cartesian coordinates (x, y, z) of the particle are by no means the only
"good" coordinates for these cases. For instance, in the last example, we may
express the energy of the electron in spherical coordinates (Fig. 1.6):

E = ~m(r2 + r 2e2 + r 2ciJ2 sin2 e) + q<l>o r cos e (1.6)

2 d
In cylindrical coordinates (Fig. 1.7) the energy is

1 .2 2 ·2 .2 q<Po
E == - m (p + p ¢ + z ) + - z (1.7)
2 d

z <r cos 8

(',8,4»
I
I
I
I
I
J
I
I
I
:y =r sin 8 sin 4>
/
/
/

, /
/
, /
<,
, /
/

, /
-. I //
Capacitor plate
_~~!_Si~ ~ .:~s_~ ~,-J,//

(a) (b)

FIGURE 1.6 Spherical coordinates.

8 Chapter 1 Review of Concepts of Classical Mechanics

(p,Z,<!»

y =p sin ¢

_osQ
I.....
'/\> p
'f.~ _

Capacitor plate
(a) (b)

FIGURE 1.7 Cylindrical coordinates.

The hydrogen atom has six degrees of freedom. If (Xl, YI, Zl) are the coordi-
nates of the proton and (X2, Y2, Z2) are the coordinates of the electron, the energy
of the hydrogen atom appears as

(1.8)

(Fig. 1.8). The mass of the proton is M and that of the electron is m. In all the
cases above, the energy is a constant of the motion. A constant of the motion is a

z
m

(x l : Yl' Z 1 )

vex 1 - X 2)2 + (y 1 - Y 2)2 + (z I - Z 2)2

FIGURE 1.8 The hydrogen atom has six degrees of freedom. The Cartesian coordinates
of the proton and electron serve as good generalized coordinates.
1.2 Energy, the Hamiltonian, and Angular Momentum 9

dynamical function that is constant as the system unfold s in time. For each of these
cases, whatever E is initially, it maintains that value, no matter how complicated
the subsequent motion is. Constants of the motion are extremely useful in classical
mechanics and often serve to facilitate calculation of the trajectory.
A system that in no way interacts with any other object in the universe is called
an isolated system. The total energy, linear momentum, and angular momentum
of an isolated system are constant. Let us recall the definition of linear and angular
momentum for a particle. A particle of mass m moving with velocity v has linear
momentum

p=mv (1.9)

The angular momentum of this particle, measured about a specific origin, is

L= r x p (1.10)

where r is the radius vector from the origin to the particle (Fig. 1.9).
If there is no component of force on a particle in a given (constant) direction,
the component of momentum in that direction is constant. For example, for a
particle in a gravitational field that is in the z direction, Px and Py are constant.
If there is no component of torque N in a given direction, the component of
angular momentum in that direction is constant. This follows directly from New-
ton's second law of angular momentum.

N= dL (1.11)
dt
For a particle in a gravitational field that is in the minus z direction, the torque on
the particle is

N = r x F = -r x ezmg

L=rxp

FIGURE 1.9 Angular momentum of a particle with momentum p about the origin O.
10 Chapter 1 Review of Concepts of Classical Mechanics

x
FIGURE 1.10 The torque r x F has no component in the z direction.

The radius vector from the origin to the particle is r (Fig. 1.10). Since ez x r has
no component in the Z direction (ez • ez x r = 0), it follows that

Lz = XPy - YPx = constant (1.12)

Since Px and Pv are also constants, this equation tells us that the projected orbit
in the x y plane is a straight line (Fig. 1.11).

FIGURE 1.11 The projected motion in the xy plane is a straight line. Its equation is
given by the constant z component of angular momentum: L z = XPy - YPx.
1.2 Energy, the Hamiltonian, and Angular Momentum 11

Hamilton's Equation
The constants of motion for more complicated systems are not so easily found.
However, there is a formalism that treats this problem directly. It is Hamilto-
nian mechanics . Consider the energy expression for an electron between capaci-
tor plates (1.5). Rewriting this expression in terms of the linear momentum p (as
opposed to velocity) gives

I (Px
, . ') ~ H( x,y,z,Px ,Py,Pz) = 2m
E( x ,y,z,x,y,z 2 +Py 2 +pz2) +q<l>o
TZ
(l.l3)

The energy, written in this manner, as a function of coordinates and momenta is

called the Hamiltonian, H . One speaks of Px as being the moment conjugate to
x; py is the momentum conjugate to y; and so on.
The equations of motion (i.e., the equations that replace Newton's second law)
in Hamiltonian theory are (for a point particle moving in three-dimensional space)

oH , oH .
-=-Px -=X
ox opx
oH . oH .
-
oy- -
- Py -=y (l.l4)
Opy
oH . oH .
- = - Pz -=z
oz Opy

Cyclic Coordinates
For the Hamiltonian (1.13) corresponding to an electron between capacitor plates,
one obtains
oH oH
-=-=0 (l.lS)
ox oy
The Hamiltonian does not contain x or y . When coordinates are missing from
the Hamiltonian, they are called cyclic or ignorable. The momentum conjugate to
a cyclic coordinate is a constant of the motion. This important property follows
directly from Hamilton 's equations, (l.l4). For example, for the case at hand, we
see that oH/ox = 0 implies that Px = 0, so Px is constant; similarly for Py·
(Note that there is no component of force in the x or y directions .) The remaining
four Hamilton 's equations give

Px = mx , Pv = my, Pz =mz (l.l6)

The last three equations return the definitions of momenta in terms of velocities.
The first equation is the z component of Newton's second law. (For an electron ,
q = -Iq I· It is attracted to the positive plate.)
12 Chapter 1 Reviewof Concepts of Classical Mechanics

Consider next the Hamiltonian for this same electron but expressed in terms of
spherical coordinates. We must transform E as given by (1.5) to an expression in-
volving r, e, if>, and the momenta conjugate to these coordinates. The momentum
conjugate to r is the component of linear momentum in the direction of r. If e, is
a unit vector in the r direction, then

r·p .
Pr = --
r
= Cr • P = mCr • v = mr (1.17)

e
The momentum conjugate to the angular displacement is the component of an-
e
gular momentum corresponding to a displacement in (with r and if> fixed). The
moment arm for this motion is r. The velocity is r8. It follows that

(1.18)

(Fig. 1.12).
The momentum conjugate to if> is the angular momentum corresponding to a
displacement in if> (with rand e fixed). The moment arm for this motion is r sin e.
The velocity is r¢ sin e (Fig. 1.13). The angular momentum of this motion is

(1.19)

Since such motion is confined to a plane normal to the z axis, PIP is the z compo-
nent of angular momentum. This was previously denoted as L z in (1.12) .

x
FIGURE1.12 Motion of a particle in spherical coordinates with rand ¢J fixed: Vii = rO,
Pli = rmvli = mr 20. The moment ann is r .
1.2 Energy, the Hamiltonian, and Angular Momentum 13

x
FIGURE 1.13 Motion of a particle with rand () fixed: v</> = r sin ()¢. The moment ann
is r sin (), P</> = (r sin ()mv</> = mr 2¢ sin2 ().

In terms of these coordinates and momenta, the energy expression (1.6) be-
comes
2 2
Pr P(? N q <1>0
H(r, (), ¢, p-, po, N) = 2m + 2mr 2 + 2mr 2 sin2 () + T r cos () (1.20)

Hamilton's equations for a point particle, in spherical coordinates , become

aH . aH .
-=-po -=()
a(} ap~
aH . aH .
-=-P</> -=¢ (1.21)
a¢ aN
aH . aH
-=r
.
-=-Pr
ar apr

From the form of the Hamiltonian (1.20) we see that ¢ is a cyclic coordinate . That
is,

(1.22)

It follows that N, as given by (1.19), is constant. Thus, the component of angu-

lar momentum in the z direction is conserved . The torque on the particle has no
component in this direction .
14 Chapter 1 Review of Concepts of Classical Mechanics

Again the momentum derivatives of H in (1.20) return the definitions of mo-

menta in terms of velocities. For example, from (1.20),

oH . PO
-=()=- (1.23)
opo mr 2

which is (1.18). Hamilton's equation for fir is

oH . po 2 p,p2 q<1>o
- - = Pr = - -3 + - --cos() (1.24)
or mr mr> sin2 () d

The first two terms on the right-hand side of this equation are the components of
centripetal force in the radial direction, due to () and ¢ displacements , respectively.
The last term is the component of electric force -e zq<1>o/d in the radial direction.
Hamilton's equation for PO is

oH . p,p 2 cos () q <1>0 •

- -o() = PO = mr 2 sirr' ()
+ - d r SIn () (1.25)

The right-hand side is a component of torque. It contains the centripetal force

factor due to the ¢ motion (Pt/> 2 / mr 2 sirr' ()) and a moment arm factor, r cos (). At
any instant of time this component of torque is normal to the plane swept out by
r due to () motion alone.
A very instructive example concerns the motion of a free particle. A free par-
ticle is one that does not interact with any other particle or field. It is free of all
interactions and is an isolated system. A particle moving by itself in an otherwise
empty universe is a free particle . In Cartesian coordinates the Hamiltonian for a
free particle is

(1.26)

All coordinates (x, y, z) are cyclic. Therefore, the three components of momenta
are constant and may be equated to their respective initial values at time t = O.

Px = Px(O)
Pv = Py(O) (1.27)
P: = p z(O)

Combining these with the remaining three Hamilton 's equations gives

mx = Px(O)
my = Py(O) (1.28)
m i; = p z(O)
1.2 Energy, the Hamiltonian, and Angular Momentum 15

These are simply integrated to obtain

x (t ) = Px(O)t +x(O)
m
yet) = Py(O)t + yeO) (1.29)
m
z(t) = pz(O) t + z(O)
m
which are parametric equations for a straight line.
Let us calculate the y component of angular momentum of the (free) particle.

t., = zpx - xpz = [Z(O) + P~O) t] Px(O) - [xeD) + Px~O) t] pz(O) (1.30)

Canceling terms, we obtain

L y = z(O)Px(O) - x(O)Pz(O) = L y(O) (1.31)

and similarly for L, and L z . It follows that

(1.32)

for a free particle.

Investigating the dynamics of a free particle in Cartesian coordinates has given
us immediate and extensive results. We know that p and L are both constant. The
orbit is rectilinear.
We may also consider the dynamics of a free particle in spherical coordinates.
The Hamiltonian is
2 2 2
H=!!!...-+~+ P<P (1.33)
2m 2 2mr 2mr 2 sin2 ()
Only tP is cyclic, and we immediately conclude that p<p (or equivalently, L z ) is
constant. However, Pr and pe are not constant. From Hamilton's equations , we
obtain
2 2
. _ pI)
Pr- - -
+ p<p
mr 3 mr 3 sin2 ()
(1.34)
. p,,/ cos()
pI) =
mr 2 sin3 ()
These centripetal terms were interpreted above. In this manner we find that the
rectilinear, constant-velocity motion of a free particle, when cast in a spherical
coordinate frame, involves accelerations in the rand () components of motion.
These accelerations arise from an inappropriate choice of coordinates. In simple
language: Fitting a straight line to spherical coordinates gives peculiar results .
16 Chapter 1 Review of Concepts of Classical Mechanics

A comparison of the Hamiltonian for a free particle in Cartesian, spherical, and

cylindrical coordinates is shown in Table 1.1.

Canonical Coordinates and Momenta

While the reader may feel some familiarity with the components of linear momen-
tum (Px, Py, pz) and angular momentum (PI} , P<t», it is clear that these intuiti ve
notions are exhausted for a system with , say, 17 degrees of freedom. If we call the
seventeenth coordinate ql7, what is the momentum Pl7 conjugate to ql7? There is
a formal procedure for determining the momentum conjugate to a given general-
ized coordinate. For example, it gives PI} = mr 2(} as the momentum conjugate to
() for a particle in spherical coordinates. This procedure is described in any book
on graduate mechanics. I
The coordinates of a system with N degrees of freedom, (ql' q2, q3, ... , qN),
and conjugate momenta (PI , P2, p3,· · · , PN) are also called canonical coordi-
nates and momenta. A set of coordinates and momenta are canonical if with the
Hamiltonian, H(ql, .. . ,qN, PI , · · · , PN, t), Hamilton's equations

aH . (l = 1, ... , N ) (1.35)
-=-Pt,
aqt

are entirely consistent with Newton's laws of motion. We have seen this to be the
case for all the problems considered above. (Time-dependent Hamiltonians are
considered in Chapter 13.)
Other important functions and concepts of classical mechanics include the La-
grangian, action integral, and Hamilton 's principle . These topics are discussed in
Section 7.11, which addresses the Feynman path integral.

PROBLEMS

1.5 Show that the z component of angular momentum for a point particle

L z = XPy - YPx

when expressed in spherical coordinates, becomes

L z = P<t> = mr 2cjJ' sin

. 2
()

(Hint: Recall the transformation equations

z = r cos e

Y = r sin ()sin cjJ

x = r sin s cos e.)
ISee , for example, H. Goldstein, Classical Mechan ics, 3d ed., Addi son-Wesley, Reading, Ma ss., 200 I.
TABLE 1.1 Hamiltonian of a free particle in three coordinate frames
Cartesian Spherical Cylindrical
Coordinates Coordinates Coordinates

Frames z

(x, Y. z) (r. O. 4» z (P. z. 4»

Hamiltonian tu». y, z, r«. Py. pz) Htr, e. r/>, Pr , PO. PIP) H(p.z,r/>,Pp.pz.P~)

I
= -(Px 2 + pi
2m
+ pz 2)
= 2m
I
[ Pr
2 I
+ r2 (
2 PIP
Po + sin22e) ] = 2m
I
(Pp
2 2 PIP
+ P: + p2 2)
2
= -2mI ( Pr 2 + -L )
r2

Momenta Px = mx Pr = mi P» =mp
Py =my PO = mr 2 tJ P: =mz
pz =mz PIP = mr 2J> sin2 e PIP = mp2J>
Cyclic x ,y, z z, r/>
coordinates
Constant Px. Py, pz pz·PIP
momenta
18 Chapter 1 Review of Concepts of Classical Mechanics

1.6 (a) Calculate Pr, po, and PI/> as explicit functions of time for the following motion
of a particle.

Y = Yo, z = zo , x = vot
(b) For what type of free-particle orbit are the following conditions obeyed ?
(1) Pr =0
(2) PrJ =0
(3) PI/> = 0
(4) Pr = PO = PI/> = 0
(c) Describe an experiment to measure p-, at a given instant, for the motion of
part (a).
1.7 Show that the energy of a free particle may be written
2 2
Pr L
H=-+--
2m 2mr 2
where L =r x p. [Hint: use the vector relation

together with the definition Pr = (r > p) /r.]

1.8 Show that angular momentum of a free particle obeys the relation
2
L2 - L 2 + L 2 + L 2 _ P 2 + sin
PI/>
- x y Z - 0 2 ()

(Hint: Employ the results of Problem 1.7.)

1.9 A particle of mass m in the environment of a force field with components

r, = -Kz , Fx =0, Fy =0

with K constant.
(a) Write down the Hamiltonian of the particle in Cartesian coordinates. What are
the constants of motion ?
(b) Use the fact that the Hamiltonian itself is also constant to obtain the orbit.
(c) What is the Hamiltonian in cylindrical coordinates? What are the constants of
motion ?
1.10 Suppose that one calculates the Hamiltonian for a given system and finds a coordinate
missing. What can be said about the symmetry of the system ?
l.ll A particle of mass m is attracted to the origin by the force

F=-Kr

Write the Hamiltonian for this system in spherical and Cartesian coordinates. What
are the cyclic coordinates in each of these frames ? [Hint: The potential for this force,
V(r), is given by F = -Kr = -VV(r) .]
 1.3 The State of a System 19

1.12 A "spherical pendulum" consists of a particle of mass m attached to one end of a

weightless rod of length a. The other end of the rod is fixed in space (the origin).
The rod is free to rotate about this point. If at any instant the angular velocity of the
particle about the origin is w, its energy is

(The motion of inertia is I .) What is the Hamiltonian of this system in spherical

coordinates? (Hint: Recall the relation L = I w.) What is the Hamiltonian of the
pendulum in a gravity field? (Let 9 denote the acceleration of gravity.)

1.3 • THE STATE OF A SYSTEM

To know the values of the generalized coordinates of a system at a given instant

is to know the location and orientation of the system at that instant. In classical
physics we can ask for more information about the system at any given instant.
We may ask for its motion as well. The location , orientation, and motion of the
system at a given instant specify the state of the system at that instant. For a point
particle in 3-space, the classical state r is given by the six quantities (Fig. 1.14)

r= (x, y, z, x, y, z) (1.36)

In terms of momenta,

r= (x, y, z, Px- Pv- Pz) (1.37)

More generally, the state of a system is a minimal aggregate of information about

the system which is maximally informative. A set of good coordinates and their
corresponding time derivatives (generalized velocities) or corresponding mo-
menta (canonical momenta) always serves as such a minimal aggregate which
is maximally informative and serves to specify the state of a system in classical
physics.

x
FIGURE 1.14 The classical state of a free particle is given by six scalar quantities
(x, y, z. Px, Py, Pz)·
20 Chapter 1 Review of Concepts of Classical Mechanics

The state of the system composed of two point particles moving in a plane is
given by the eight parameters

(1.38)

Just as the set of generalized coordinates one assigns to a given system is not
unique, neither is the description of the state r. For instance, the state of a point
particle moving in a plane in Cartesian representation is

r= (x, y, Px, p y) (1.39)

In polar representation it is

r= (r, e, Pr- po) (lAO)

All representations of the state of a given system in classical mechanics contain

an equal number of variables. If we think of r as a vector, then for a system
with N degrees of freedom, r is 2N -dimensional. In classical mechanics change
of representation is effected by a change from one set of canonical coordinates
and momenta (q, p) to another valid set of canonical coordinates and momenta
(q', p') .

One form of canonical transformation results simply from a change in coordi-

nates. For example, the transformation from Cartesian to polar coordinates for a
particle moving in a plane effects the following change in representation:

rex, y, Px, p y) -+ I'(r, e, Pr, po)

Representations in Quantum Mechanics

Next, we tum briefly to the form these concepts take in quantum mechanics. The
specification of parameters that determines the state of a system in quantum me-
chanics is more subtle than in classical mechanics. As will emerge in the course
of development of this text, in quantum mechanics one is not free to simultane-
ously specify certain sets of variables relating to a system. For example, while the
classical state of a free particle moving in the x direction is given by the values
of its position x, and momentum Px- in quantum mechanics such simultaneous
specification cannot be made. Thus, if the position x of the particle is measured
at a given instant, the particle is left in a state wherein the particle's momentum
is maximally uncertain. If on the other hand the momentum Px is measured, the
particle is left in a state in which its position is maximally uncertain. Suppose it
is known that the particle has a specific value of momentum. One may then ask if
there are any other variables whose values may be ascertained without destroying
the established value of momentum. For a free particle one may further specify
the energy E; that is, in quantum mechanics it is possible for the particle to be
1 .3 The State of a System 21

in a state such that measurement of momentum definitely finds the value Px and
measurement of energy definitely finds the value E . Suppose there are no further
observable properties of the free particle that may be specified simultaneously
with those two variables. Consequently, values of Px and E comprise the most
informative statement one can make about the particle and these values may be
taken to comprise the state of the system of the particle

As remarked above, if the particle is in this state, it is certain that measurement of

momentum finds Px and measurement of energy finds E . Such values of Px and
E are sometimes called good quantum numbers . As with their classical counter-
part, good quantum numbers are an independent set of parameters which may be
simultaneously specified and which are maximally informative.
For some problems in quantum mechanics it will prove convenient to give the
state in terms of the Cartesian components of angular momentum: Lx, L y , and
L z • We will find that specifying the value of L z , say, induces an uncertainty in
the accompanying components of L, and L y , so that, for example, it is impossi-
ble to simultaneously specify L z and L ; for a given system. One may, however,
simultaneously specify L z together with the square of the magnitude of the total
momentum, L 2 . For a particle moving in a spherically symmetric environment,
one may also simultaneously specify the energy of the particle. This is the most
informative/ statement one can make about such a particle, and the values of en-
ergy, L 2 and L z , comprise a quantum state of the system.

(1.41)

The values of E, L 2 , and L , are then good quantum numbers. That is, they are an
independent set of parameters which may be simultaneously specified and which
are maximally informative.
Just as change in representation, as discussed above, plays an important role in
classical physics, so does its counterpart in quantum mechanics. A representation
in quantum mechanics relates to the observables that one can precisely specify in
a given state . In transforming to a new representation, new observables are speci-
fied in the state . For a free point particle moving in 3-space, in one representation
the three components of linear momentum Px, Py, and P: are specified while
in another representation the energy p 2 / 2m, the square of the angular momen-
tum L 2 , and any component of angular momentum, say L z , are specified. In this
change of representation,

(1.42)

When treating the problem of the angular momentum of two particles (Lj and
L2, respectively) in one representation, (LI 2, L2 2 , L zl' L Z2 ) are specified while
2More precisely, r includes the parity of the system. This is a purelyquantummechanical notion and
will be discussed more fully in Chapter6.
22 Chapter 1 Review of Concepts of Classical Mechanics

in another representation (LI 2, L2 2, L 2, L z) are specified. Here we are writing L

for the total angular momentum of the system L = LI + L2. In this change of
representation,

(1.43)

Finally, in this very brief introductory description, we tum to the concept of

the change of the quantum state in time. In classical mechanics, Newton's laws of
motion determine the change of the state of the system in time. In quantum me-
chanics, the evolution in time of the state of the system is incorporated in the wave
(or state) function and its equation of motion, the Schrodinger equation. Through
the wavefunction, one may calculate (expected) values of observable properties
of the system, including the time development of the state of the system.
These concepts of the quantum state-its evolution in time and change in
representation---comprise principal themes in quantum mechanics. Their under-
standing and application are important and are fully developed later in the text.

PROBLEMS

1.13 Write down a set of variables that may be used to prescribe the classical state for
each of the 11 systems in problem 1.1.

Answer (partial)
(e) A rigid rod in 3-space: Since the system has five degrees of freedom, the classical
state of the system is given by 10 parameters. For example,

[Note: The quantum state is less informative. For example, such a state is pre-
scribed by five variables (x , y, z. 0, ep). Another specification of the quantum
state is given by five momenta (Px, Py, Pz» P(), Pt/» . However, simultaneous
specification of, say, x and Px is not possible in quantum mechanics.]

1.14 (a) Use Hamilton's equations for a system with N degrees of freedom to show that
H is constant in time if H does not contain the time explicitly. [Hint: Write

dH sn N (aH
-=-+L . sn , )
-q/+-P/ .]
dt at /=1 aq/ apt

(b) Construct a simple system for which H is an explicit function of the time.

1.15 For a system with N degrees of freedom, the Poisson bracket of two dynamical func-
tions A and B is defined as
Problems 23

(a) Use Hamilton's equations to show that the total time rate of change of a dynam-
ical function A may be written

dA aA
- = - +{A,H}
dt at

where H is the Hamiltonian of the system.

(b) Prove the following: (1) If a(q, p) does not contain the time explicitly and
{A, H} = 0, then A is a constant of the motion. (2) If A contains the time
explicitly, it is constant if aAjat = {H, A}.
(c) For a free particle moving in one dimension, show that

pt
A=x--
m

satisfies the equation

aA
at = -{A, H}

so that it is a constant of the motion. What does this constant correspond to

physically?

1.16 How many degrees of freedom does the compound pendulum depicted in Fig. 1.15
have? Choose a set of generalized coordinates (be certain they are independent).
What is the Hamiltonian for this system in terms of the coordinates you have chosen?
What are the immediate constants of motion?

FIGURE 1.15 Compound pendulum composed of two masses connected by weightless

rods of length a. The motion is in the plane of the paper. (See Problem 1.16.)

1.17 How many constants of the motion does a system with N degrees of freedom have?

Answer
Each of the coordinates {qz} and momenta {pz} satisfies the first-order differential
equation in time (i.e., Hamiltonian's equations). Every such equation has one con-
stant of integration. These comprise 2N constants of the motion.
24 Chapter 1 Review of Concepts of Classical Mechanics

Vex)

A B C D x

FIGURE 1.16 Arbitrary potential function.

1.4 • PROPERTIES OF THE ONE-DIMENSIONAL POTENTIAL FUNCTION

Consider a particle that is constrained to move in one dimension, x . The particle

is in the potential field V (x) depicted in Fig. 1.16. What is the direction of force at
the point x = A? We can calculate the gradient (in the x direction) and conclude
that the direction of force at A is in the + direction . There is a simpler technique .
Imagine that the curve drawn is the contour of a range of mountain peaks . If a ball
is placed at A, it rolls down the hill. The force is in the +x direction . If placed at
B (or C), it remains there. If placed at D, it rolls back toward the origin; the force
is in the -x direction . This technique always works (even for three-dimensional
potential surfaces) because the gravity potential is proportional to height z, so the
potential surface for a particle constrained to move on the surface of a mountain
is that same surface.
The one-dimensional spring potential, V = Kx 2/2, is depicted in Fig. 1.17.
If the particle is started from rest at x = A, it oscillates back and forth in the
potential well between x = +A and x = - A .
Motion described by a potential function is said to be conservative. For such
motion, the energy

E=T+V (1.44)

K 2
V= Z-x

A x

FIGURE 1.17 Spring potential.

Problems 25

FIGURE 1.18 Forbidden domains at energy E .

is constant. In terms of the kinetic energy T,

mv 2
T=-=E-V (l.45)
2
Forbidden Domains
From (1.45) we see that if V > E, then T < 0 and the velocity becomes imag-
inary. In classical physics, particles are excluded from such domains. They are
calledforbidden regions. Again consider a one-dimensional problem with poten-
tial V(x) shown in Fig. 1.18. The constant energy E is superimposed on this
diagram. Segments AB and CD are forbidden regions. Points A, B, C, and D are
stationary or turning points. Since E = V at these points, T = 0 and x = O. Sup-
pose that a particle is started from rest from the point C. What is the subsequent
motion? The particle is trapped in the potential well between Band C. It accel-
erates down the hill, slows down in climbing the middle peak, then slows down
further in climbing to B, where it comes to rest and turns around. This periodic
motion continues without end.
The one-dimensional potential depicted in Fig. 1.18 can be effected by appro-
priately charging and spacing a linear array of plates with holes bored along the
axis. The potential depicted in Fig. 1.18 is seen by an electron constrained to move
along this (x) axis.

PROBLEMS

1.18 A particle constrained to move in one dimension (x) is in the potential field
VO(x - a)(x - b)
V (x) = ----"----'---=---'- (0 < a < b < c < (0)
(x - c)2

(a) Make a sketch of V .

(b) Discuss the possible motions, forbidden domains and turning points. Specifically,
if the particle is known to be at x = -00 with
26 Chapter 1 Review of Concepts of Classical Mechanics

3Vo
E= --(b-4a+3c)
c-b
at which value of x does it reflect?
1.19 A particle of mass m moves in a "central potential ," V (r), where r denotes the radial
displacement of the particle from a fixed origin .
(a) What is the (vector) force on the particle? Recall here the components of the V
operator in spherical coordinates.
(b) Show that the angular momentum L of the particle about the origin is constant.
(Hint: Calculate the time derivative of L = r x p and recall that p = mr.)
(c) Show that the energy of the particle may be written
2 2
Pr L
E = 2m + 2mr 2 + V(r)
(d) From Hamilton's equations obtain a "one-dimensional" equation for Pr, in the
form
• iJ
Pr = --
iJr
Veff(r)

where Veff denotes an "effective" potential that is a function of r only.

(e) For the case of gravitational attraction between two masses (M, m), V =
<Gm M]», where G is the gravitational constant. Make a sketch of Veff versus
r for this case . Use this sketch to establish the conditions for circular motion
(assume that M is fixed in space) for a given value of L 2 .
1.20 Complex variables play an important role in quantum mechanics . The following two
problems are intended as a short review.
If

= 11/11 expueq )
1/1
X = IxI exp( ia2)

show that

1.21 Use the expression

ei8 = cos 8 + i sin8

to derive the following relations
(a) cos(81 + 82) = cos 81 cos 82 - sin 81 sin 82
(b) sin(81 + 82) = cos 81 sin 82 + sin 81 cos 82
(c) 2 sin 81 cos 82 = sin(81 - 82) + sin(81 + 82)
(d) 2cos81 cos 82 = cos(81 + 82) + cos(81 - 82)
(e) 2 sin 81 sin 82 = cos(81 - 82) - cos(81 + 82)
(I) 2 cos 2 8 = 1 + cos 28
(g) 2 sin 2 8 =1- cos 28
Problems 27

(h) eiO - I = 2i eiO/ 2 sin(O/2)

(i) ~leiOI +e i0212 = ~(ejOl +ei(2)(eiOI +e i( 2)* = I +COS(OI -(2)
2 2
(j) 2 Re z = z + z"
(k) 2i Irn z = z - z*
(I) (expz)" = exp z"
(m) [exp d = exp(2 Re z)
(n) ii = e- 1f / 2 , e- 5rr / 2 , e- 9rr / 2 , •..

l
(0) L exp( - i2:rtn/ I) = 0
n=1
where I > I is an integer. [Hint: Rewrite the given relation as a summation over
the complex unit vector z(O) = exp(iO) .]

1.22 Vector calculus plays in important role in quantum mechanics. The following is in-
tended as a short review.
(a) Write down V2cp in Cartesian (x, y, z), spherical (r, 0, rjJ) , and cylindrical
(p, rjJ, z) coordinates.
(b) If (dA, dV) represents differential of (area, volume) , working in spherical coor-
dinates, show that

dV = r 2drdQ
where dQ is the differential of solid angle

dQ = sinO dO drjJ

(c) Calculate the value of

A == Is dQ

where the integral is over a closed surface, S, about the origin. The units of solid
angle are steradians. Note that, in two dimensions, the corresponding expression
is

AL = f dO = 2:rt (radians)
(d) Show that

vx (Vcp) =0
(e) Show that

V· (V x A) =0
Hint: The latter two properties are most easily established with Gauss ' and
Stokes' integral laws.
28 Chapter 1 Review of Concepts of Classical Mechanics

(0 What is the value of the line integral

The integral is along a curve C which connects the end points rj , r2 .

1.23 The dumbbell molecule plays an important role in later discussions of angular mo-
mentum . Consider a dumbbell molecule of reduced mass /-L and atomic separation a .
Let I denote the moment of inertia of the molecule with respect to rotation of the
dumbbell about its center of mass. If one writes I = ureff2, obtain an expression for
reff in terms of m \> m 2, and the separation a.

Answer
Respective displacements of masses from the center of mass of the dumbbell are
am2
q=M'
where M =m I + m2. Thus,
mla2m22 +m2 a2m 21
I = ----"----.::.--:-=------''-
M2

mlm2 a2M 2
M2 =/-La

so that reff = a.
1.24 Time reversibility in classical physics is described by the invariance of Hamilton 's
equations (1.35) to the transformation (t --+ -t , P --+ -p) . What implications does
reversibility have on:
(a) The Hamiltonian of the system?
(b) The dynamical solutions x(t) , p(t)?
1.25 The transformation of coordinates and momenta

is said to be canonical, providing these variables satisfy the fundamental Poisson-

bracket relations

where (,} denotes Poisson brackets (Problem 1.15) evaluated with respect to
(ql , ' " , a»: PI , .. . ,PN) variables.
(a) Consider a rigid body that rotates about a fixed interior point in the absence of
external forces, with energy given by

where I denotes moment of inertia. Is this energy a valid Hamiltonian form?

Problems 29

(b) Now consider a one-dimensional rigid rod that rotates about a fixed point on
the rod in the absence of external forces, with energy L 2 /2/. Is this a valid
Hamiltonian fonn?

Answer (partial)
(a) This Hamiltonian is not valid as the angular momenta (Lx, L y , L z) satisfy the
relations {Lx, L y } = L z, etc., and consequently do not comprise a canonical set
of momenta.
CHAPTER

Historical Review:
2 Experiments and Theories

2.1 Dates
2.2 The Work ofPlanck. Blackbody Radiation
2.3 The Work of Einstein. The Photoelectric Effect
2.4 The Work ofBohr. A Quantum Theory ofAtomic States
2.5 Waves versus Particles
2.6 The de Broglie Hypothesis and the Davisson-Germer Experiment
2.7 The Work ofHeisenberg. Uncertainty as a Cornerstone ofNatural Law
2.8 The Work ofBorn. Probability Waves
2.9 Semiphilosophical Epilogue to Chapter 2

The following sections summarize experiments and theories formulated during

the early decades of the century. These observations and theories contribute to
the genesis of quantum mechanics. The important concept of the wavefunction
is introduced and the Born interpretation of this function in terms ofprobability
density is described. A more formal presentation of the postulates of quantum
mechanics appears in Chapter 3.

2.1 • DATES

Physics at the tum of the century was in a state of turmoil. There was a Pandora's
box of experimental observations which, on the grounds of otherwise firmly estab-
lished classical theory, was totally inexplicable. One by one, all these perplexing
questions were answered-with the drama and flair of a story told by a mas-
terful raconteur. Out of the turmoil came a new philosophy of science. A new
way of thinking was called for. At the very core of natural law lay subjective
probability-not objective determinism.
What were some of these perplexing observations ? Light exhibits interference
and therefore may be assumed to be a wave phenomenon. However, if we try
to explain the photoelectric effect (light hitting a metal surface ejects electrons)
on the basis of the wave nature of light, we obtain erroneous results. It is found
that the energy of an emitted electron is dependent only on the frequency of the
incident radiation, not on the intensity as might be expected from the classical
theory of light.

30
 2.2 The Work of Planck. Blackbody Radiation 31

In 1911 it was established by Rutherford that an atom has a positive central

core and satellite electrons. Hydrogen, for instance, has a proton at its center and
one outer electron. But such a circulating (and therefore accelerating) electron
radiates and soon should collapse into the nucleus. So why do we not see a burst
of ultraviolet radiation emitted as the electron spirals into the nucleus? Why is the
frequency spectrum of light emitted form an atom a discrete line spectrum and
not a continuous spectrum?
Another dilemma lay in the observations of the spectrum of radiant energy in
a cavity whose walls are maintained at a fixed temperature. Theory (on the basis
of the wave nature of light) was unable to account for the observed frequency
distribution of radiant energy.
Outlined below are the discoveries and events that occurred near the tum of
the century that removed the enigmas and led naturally to the development of
quantum mechanics.
1898 Mme . Curiel Radioactive polonium and radium
1901 Planck Blackbody radiation
1905 Einstein Photoelectric effect
1911 Rutherford/ Model of the atom
1913 Bohr Quantum theory of spectra
1922 Compton Scattering photons off electrons
1924 Pauli Exclusion principle
1925 de Broglie Matter waves
1926 Schrodinger Wave equation
1927 Heisenberg Uncertainty principle'
1927 Davisson and Germer Experiment on wave properties of electrons
1927 Born Interpretation of the wavefunction
1928 Dirac Relativistic wave equation; prediction of
existence of the positron
In the remainder of this chapter we will outline these topics in more detail,
except for the work of Schrodinger, which is formally presented in Chapter 3,
and the work of Pauli, which is presented in Chapter 12. The Compton effect is
discussed in Problem 2.28, and the Dirac equation is studied in Chapter 15.

2.2 • THE WORK OF PLANCK. BLACKBODY RADIATION

Place a closed, evacuated container (with a small window in the wall) in an oven
of uniform temperature. Wait until all components of the experiment reach the
same temperature (thermal equilibrium). At a sufficiently high temperature, vis-
ible light emerges from the window of the container cavity. The cavity contains
radiant energy, which is in thermal equilibrium with the cavity walls . Suppose that
IMarja Sklodowska (1867-1934), born in Poland .
ZEmest Rutherford (1871-1937), born in poverty in New Zealand to parents who were potato farmers.
Scholarship of the Great Exhibition of 1851 awards him study at Cambridge.
32 Chapter 2 Historical Review: Experiments and Theories

the total radiant energy per unit volume in the cavity (at any instant) is U . How
much of this energy is in electromagnetic waves with frequency between v and
v + dv? Let us call the answer u(v) dv. The function u(v) then gives the energy
per frequency interval per unit volume. The total energy per unit volume in the
radiation field in the cavity is

U = 100

u(v) dv (2.1)

The radiation is called blackbody radiation because it is assumed that any light
falling on the window is totally absorbed . The window acts as a perfect radiator
and a perfect absorber. This propert y is characteristic of ideal black surfaces. At
any given temperature, no object emits or absorbs radiation more efficiently than
does an (ideal) blackbody.
The experimentally observed curve of u(v) is shown in Fig. 2.1. Classical elec-
trodynamic and thermodynamic theory give two properties of the spectral distri-
bution of a radiation field in equilibrium at the temperature T . The Rayleigh-
Jeans (1900) approximation

is appropriate for low frequencies. In this formula kB is Boltzmann's constant,

ks = 1.381 x 10- 16 erg K

Radio and microwaves

'i
'2
::l
1.4. ~-----lnfrared -----:l~---t~

FIGURE 2.1 Spectrum of blackbody radiation . The curves have been distorted to bring
out some important features. In reality, the curve at 10,000 K is about 37,000 times higher
than the curve at 300 K. Also, the radio and microwave domain is only about 1 30,000 of
the v axis depicted .
Problems 33

and c is the speed of light. While this approximation is valid at low frequencies,
it is seen to diverge at larger frequencies, where as shown in Fig. 2.1, the correct
spectral distribution falls off to zero. Wien's law (1893) specifies that u, as a
function of wavelength A = c/ v , is of the form

where W is an arbitrary continuous function of the product of wavelength A and

temperature T. Although this formula is valid over the whole spectrum of wave-
lengths, it is incomplete in that W(AT) is undetermined. The complete explicit
form for the spectral distribution U cannot be obtained from classical physics . A
quantum hypothesis must be invoked. Such was the assumption made by Planck
to obtain a uniformly valid formula for u(v) . It implied that energy of radiation
with frequency v exists only in multiples of hv, where h is a constant of nature
(Planck 's constant) . A quantum of radiation of energy h v is called a photon.

E =hv (2.2)

The correct formula for u(v) that results is (see Problems 2.36 and 2.37)
3
u(v) = 8rrhv 1 (2.3)
c3 ehv /kBT - 1

h = 6.626 x 10-27 erg-s

This expression precisely matches the experimental curves shown in Fig. 2.1.

PROBLEMS

2.1 (a) Show that for photons of frequency v and wavelength A:

(1) dv = -CdA /A 2
(2) U(A) d): = -u(v) dv
(3) U(A) dA = u(v)c dA/A2
(b) Show that the Rayleigh-Jeans spectral distribution of blackbody radiation,
UR J (v) , is of the form required by Wien'slaw,

W(AT)
UW(A)=~

(c) Obtain the correct form of Wien's undetermined function W(AT) from Planck's
formula.
2.2 A spherical enclosure is in equilibrium at the temperature T with a radiation field
that it contains. Show that the power emitted through a hole of unit area in the wall
of enclosure is
1
P = -cU
4
34 Chapter 2 Historical Review: Experiments and Theories

r = (r, (), ¢)
Radiation Field

Solid angle subtended

by the area A about
the point r

FIGURE 2.2 The power radiated by an electromagnetic field in equilibrium at temper-

ature T is due to photons that lie in a hemisphere of radius c, centered at the hole. (See
Problem 2.2.)

Answer
Let the cavity be very large, so that its walls can be considered to be flat. The energy
that flows through a hole in the wall, of unit area, in 1 s is the power radiated. This
energy is due to photons that lie in a hemisphere of radius c, centered at the hole
(Fig. 2.2). The energy in the volume element dr about the point r is U dr. Owing to
the isotropy of the radiation field, the amount of this energy that passes through the
hole is U dr times the ratio of solid angle Q subtended by the area of the hole about
the point r, to 4JT, the total solid angle about the point r.

Q A' e, · AU dr U cos B
dP == -U dr == - - U dr == == - - dr
4JT 4JTr 2 4JTr 2 4JTr 2
U d¢cosBdcosBr 2dr
4JTr 2
The radiation energy that passes through the hole in 1 s is from all volume elements
in the hemisphere is the total power radiated per unit area.

P==
1hemisphere
dP==-
U
4JT
1 1 27T

a
d¢
a
1
cosBdcosB 1a
c
dr==-cU
1
4

2.3 Show that the energy density U (T) of a radiation field in equilibrium at the temper-
ature T is directly proportional to T 4 . The corresponding expression for the emitted
power is

P == CJT 4
Problems 35

where o is the Stefan-Boltzmann constant

[Hint: Nondimensionalize the integration over (2.3) through the variable x _

hv/kBT .]

2.4 Use (2.3) to prove Wien's displacement law

Amax T = constant = 0.290 cm K

The wavelength Amax is such that U(Amax) is maximum. [Hint: Differentiate U(A)
with respect to the variable x == he] kT A and set equal to zero.]
2.5 From the sketch of U versus v given in Fig. 2.1, make a sketch of U versus A, where
VA =c.
2.6 What is the photon flux (pbotons/cm/ s) at a distance of 1 km from a light source
emitting 50 W of radiation in the visible domain, with wavelength 6000 A?

2.7 The average energy in a unit volume in the v frequency mode of a blackbody radiation
field is
hv
(U ) = ehv/kBT _ 1

What does (U) reduce to in the limit (a) v -+ O? (b) T -+ oo?

2.8 As discussed above, the radiation field interior to a closed cavity whose walls are in
thermal equilibrium (i.e., at the same temperature) with the radiation field is called
blackbody radiation. Prove that blackbody radiation has the following properties by
showing that if any of these properties are not true, a device can be constructed which
violates the second law of thermodynamics.
(a) The flux of radiation is the same in all directions. (The radiation field is
isotropic. )
(b) The energy density is the same at all points inside the cavity. (The radiation field
is homogeneous.)
(c) The energy density interior to the cavity is the same (function of frequency) at a
given temperature, regardless of the material of the cavity wall.

2.9 Prove that the radiation emitted by the surface of an ideal blackbody at the temper-
ature T is the same as that which travels in one direction inside a closed isothermal
cavity at the same temperature.

Answer
Immerse an ideal black cube inside the isothermal container. The radiation that falls
on any face of the cube is completely absorbed. For equilibrium to be maintained,
the radiation emitted must be balanced by that absorbed, so that the radiation emitted
is precisely that which flows into the face.
If, on the other hand, the cube is not ideally black, equilibrium is maintained by
balancing the absorbed radiation by the reflected plus emitted radiation. Since energy
density in the cavity is the same as in the case above (both experiments are at the same
36 Chapter 2 Historical Review: Experimentsand Theories

temperature), the radiation emitted by the nonblack surface is less intense than that
emitted by the ideally black surface.

2.10 One of the theories of the origin of the universe is that it was contained in a primeval
fireball which began its expansion about 1010 years ago. As it expanded, it cooled.
Measurements of the energy spectrum of cosmic photons suggest a (blackbody) tem-
perature of 3 K. At what frequency is maximum energy observed?

2.11 Suppose that you are inside a blackbody radiation cavity which is at temperature T .
Your job is to measure the energy in the radiation field in the frequency interval 1014
to 89 x 1014 Hz. You have a detector that will do the job. For best results, should the
temperature of the detector T' be T' > T , T' = T, T' < T, or T' = 0; or is the
temperature of the detector irrelevant to the measurement?

2.3 • THE WORK OF EINSTEIN. THE PHOTOELECTRIC EFFECT

The experimental setup that exhibits the photoelectric effect is depicted in

Fig. 2.3. The observation is as follows. A metal plate (e.g., copper) is irradi-
ated with light of a given frequency. Electrons are ejected from the photo cathode
and current is registered in the ammeter A. As the potential on the collecting
plate is made more negative, the current diminishes, until finally at the potential
VS10 P , current ceases. The energy that an electron must have in order to climb
the potential hill imposed by the negative bias V is e V. Only the most kinetic
electrons reach the plate near VS10 p ' At VS10 P the electrons with maximum kinetic
energy IE have been repelled. Then

IE = eVS10 p (2.4)

e-- Collecting
plate

Photocathode
v

'---+-tll t-_---J
FIGURE 2.3 Photoelectric experiment.
2.3 The Work of Einstein. The Photoelectric Effect 37

v
FIGURE 2.4 Typical data showing energy of most energetic electrons as a function of
frequency v in the photoelectric experiment.

At a given frequency v, one makes a measurement of lE and plots a point on an

lE versus lJ graph (Fig. 2.4). If the intensity of light is increased while v is held
fixed, lE remains constant. On the other hand, when lJ is increased, lE increases. A
typical collection of data is shown in Fig. 2.4.
To explain this effect Einstein hypothesized that light is composed of localized
bundles of electromagnetic energy called photons. At frequency v, the energy of
a photon is h v. When striking the metal surface, the photon interacts with an
electron and ejects it from the metal. Let us consider the Sommerfeld model of
a conductor (Fig. 2.5). The conductor is composed of fixed positive sites (e.g.,
Cu 2 + ions in copper) and free electrons. The positive ions generate a potential
well in which the electrons are trapped. The electrons have energy from 0 to EF,
the Fermi energy. The minimum work to remove an electron from the metal is
W - E F, which is called the work function, <1>. The depth of the well is W.
Electrons distribute themselves in accordance with the Pauli exclusion
principle. This principle precludes more than one electron existing in the same
quantum state. For example, the distribution of electron energies shown in Fig. 2.5
is maintained at 0 K. At this temperature, electrons fall to lowest allowable ener-
gies. They cannot all fall to the single lowest level, owing to the Pauli principle.
Once this level is occupied, the next electron must seek the next higher level. The
maximum value of energy so reached is the Fermi energy E F .

+
[E

----w t
<I>

Fermi sea
} of electrons

o
FIGURE 2.5 Sommerfeld model for energy distribution of electrons in a metal.
38 Chapter 2 Historical Review: Experiments and Theories

Suppose that a photon of energy hv hits an electron and ejects it with kinetic
energy lE. The most energetic electrons come from the top of the Fermi sea. The
energy lE of such an electron ejected by a photon of energy hv is given by

lE = hv - <I> (2.5)

If we plot lE versus v from this equation, we obtain the curve shown in Fig. 2.4.
Note that the slope ofthe curve is Planck's constant h, and the v intercept gives the
work function (of the photocathode metal). If <I> == hVth, Vth is called the threshold
frequency. A few typical values are:

Metal vth (Hz) EF (eV)

Silver 1.14 x 1015 5.5

Potassium 0.51 x 1015 2.1
Sodium 0.56 x 1015 3.1

Millikan in 1916 used the photoelectric experiment to obtain a value of

Planck's constant, h [see (2.3)].

Contact Potential
The preceding description may also be used to explain the phenomenon of con-
tact potential, the finite potential that develops between two dissimilar metals
which are brought into contact with each other. To describe this effect we con-
sider a parallel-plate capacitor with one plate made of metal A and the other made
of metal B. When the plates are isolated and displaced far from each other, the
common zero in potential of both metals corresponds to zero free-particle kinetic
energy (Fig. 2.6a).
Now let the metals be brought into contact with each other. Electrons then
"fall" from the Fermi level of metal A, which has the smaller work function, to the
deeper-lying Fermi level of metal B, until the tops of the two electron energy dis-
tributions are equalized. Having lost electrons, metal A is left electropositive with
respect to metal B and a potential difference exists between the plates (Fig. 2.6b).
This description leads to the conclusion that the contact potential difference
Vc between two metals should be well approximated by the difference in work
functions :

The validity of this relation is borne out by experiment.

PROBLEMS

2.12 (a) A monochromatic point source of light radiates 25 W at a wavelength of 5000 A. A

plate of metal is placed 100 em from the source. Atoms in the metal have a radius
of I A. Assume that the atom can continually absorb light. The work function of
the metal is 4 eV. How long is it before an electron is emitted from the metal?
 2.4 The Work of Bohr. A Quantum Theory of Atomic States 39

Zero free-particle
kinetic energy

<PA
----EF,A <PB

EF.B------

(a)

(b)

FIGURE 2.6 The difference in work functions causes electrons to fall to the lower Fermi
level thereby creating a contact potential. (a) The metals far removed from each other.
(b) The metals in contact with each other. The sloping curve represents the potential seen
by an electron.

(b) Is there sufficient energy in a single photon in the radiation field to eject an electron
from the metal?

2.13 The photoelectric threshold of tungsten is 2300 A. Determine the energy of the electrons
ejected from the surface by ultraviolet light of wavelength 1900 A.

2.14 The work function of zinc is 3.6 eV. What is the energy of the most energetic photo-
electron emitted by ultraviolet light wavelength 2500 A?

2.15 Photoelectrons emitted from a cesium plate illuminated with ultraviolet light of wave-
length 2000 A are stopped by a potential of 4.21 V. What is the work function of ce-
sium?

2.4 • THE WORK OF BOHR. A QUANTUM THEORY OF ATOMIC STATES

Consider a discharge tube filled with hydrogen gas. At sufficient voltage the gas
glows. If the light is examined in a spectroscope, it is seen that only a discrete
set of frequencies-a line spectrum-is emitted. Bohr was able to account for the
discrete emission spectra in an analysis based on two postulates:
40 Chapter 2 Historical Review: Experiments and Theories

1. Hydrogen exists in discrete energy states. These states are characterized by

discrete values of the angular momentum as given by the relation

f pede = nh (2.6)

with n an integer greater than zero. In these states the atom does not radi-
ate. The line integral follows the electron in one complete orbit about the
nucleus .
2. When an atom undergoes a change in energy from En to Em, electromag-
netic radiation (a photon) is emitted at a frequency v given by

hv = En - Em (2.7)

Let us recall how condition (2.6) leads to a discrete set of energies {En}. The
energy of a (stationary) hydrogen atom whose electron is moving in circular mo-
tion is
1 e2
E = -mv 2 (2.8)
2 r
The radius r obeys the centripetal condition

(2.9)
r
so that, with (2.6)

e2 pe 2 n21i2
-;: = mr 2 = mr 2 (2.10)

(2.11)

These are the quantized values of r at which the electron persists without radiat-
ing. The values of the energy at these radii are

2 21i2 2)2
E __ pe __ n (me
n - 2mr2 - 2m n21i2
(2.12)

where lR is the Rydberg constant:

4
me -11
lR = 21i2 = 2.18 x 10 erg = 13.6 eV (2.13)

The negative quality of the energy reflects the fact that we are dealing with bound
states. When n = 1, the atom is in the ground state and has energy, - R To ionize
2.4 The Work of Bohr. A Quantum Theory of Atomic States 41

the atom when it is in this state takes +lRergs of energy. The value of r when the
atom is in the ground state is

!i 2
r1 == ao == -2 == 5.29 x 10- 9 ern (2.14)
me
This is a fundamental length in physics. It is called the Bohr radius.
When the electron and proton are infinitely far removed and at rest, Tn == 00.
From (2.11) we see that this corresponds to n == 00. In this state En == 0; there is
no kinetic energy and no potential energy. If the electron is given a tap, it becomes
a free particle. The composite system of proton plus electron then has positive
energy (kinetic only), with all (unquantized) positive values of energy allowed
(Fig. 2.7).
The emission spectra of hydrogen is generated by the values for En (2.12) and
the second postulate (2.7). The frequencies so generated (with some minor refine-
ments, e.g., accounting for the motion of the proton) agree to a high degree of
accuracy with the data. Characteristically, the spectrum divides into various series
of lines. The Lyman series is comprised of frequencies generated by transitions to
the ground state:

(n > 1) (UV) (2.15)

The Balmer series is generated by transitions to the first excited state:

(n > 2) (UV and visible) (2.16)

and so forth (Fig. 2.8). Remaining series are Paschen, Brackett, and Pfund (all
IR).

Energy } Continuum
n free states
00
E4 0 -lR/16 4
E3 -lR/9 3

E2 -lR/4 2

Bound
states

E1 -lR

FIGURE 2.7 Bohr spectrum.

42 Chapter 2 Historical Review: Experimentsand Theories

n~£=O
5
4
3

2
(
/! f--A oo

Balmer series
=0 .365 11m

Al = 0.656 11m

Al =

-.
0.122 11m _ Aoo = 0.091 11m

£=
-13 .58 eV
IV
Lyman series

FIGURE 2.8 First two series of emission spectrum for hydrogen. Wavelengths of radia-
tion are given in units of microns (10- 4 em).

PROBLEMS

2.16 (a) Consider the spherical pendulum described in Problem 1.12. Use the Bohr for-
mula (2.6) to obtain the quantum energies of this system. Note the identity
PO = L.
(b) Suppose that the pendulum is comprised of a proton attached to a weightless rod
of length a = 2 A. What is the ground rotational state of this system (in eV)?
(See Problem 10.40.)

2.17 (a) What is the formula for the frequency v of radiation emitted when the hydrogen
atom decays from state n to state n'? Give your answer in terms of R n, n' , and
h only.
(b) What is the corresponding formula for the wavelength A emitted in the same
transition? Now your formula will contain the additional constant c, the speed of
light.
2.18 The angular momentum of an isolated system is constant (when referred to any ori-
gin). Derive an expression for the angular momentum PO carried away by a photon
emitted when a hydrogen atom decays from the state n to the state n' (in the Bohr
model).
2.19 In classical electromagnetic theory, an accelerating charge e radiates energy at the
rate
 2.5 Waves versus Particles 43

2 a 2e2
W = :3 --;;3 ergs/s
The acceleration is a and c is the speed of light. At time t = 0, a hydrogen atom has
a radius 1 A. Assuming classical circular motion:
(a) What is a initially?
(b) What is the initial frequency of radiation that the atom emits?
(c) How long does it take for the radius to collapse from 1 A to 0.5 A? (Assume that
a is constant.)
(d) What is the frequency of radiation at the radius 0.5 A?
2.20 The dimensionless number

e2 1
cx=-=---
he 137.037
is called the fine-structure constant.
(a) Show that the Rydberg constant may be written lR = !cx2mc2 .
(b) If the rest-mass energy of the electron is = 0.511 MeV, calculate lR in eV.
mc 2
(c) Obtain an expression for the Bohr energies En in terms of a and mc 2 .

2.5 • WAVES VERSUS PARTICLES

Suppose that a disturbance propagates from one point in space to another point in
space . What is propagating, waves or particles? A principal distinguishing char-
acteristic is that waves exhibit interference, particles do not.
Consider the two-slit experiment shown in Fig. 2.9. A continuou s spray of par-
ticles is fired from the source S. They strike the wall or pass through the two slits
A and B . An intensity II (number/unit area - second) emerges from A and inten-

Wall Screen

o~
s""",

FIGURE 2.9 Particle double-slit experiment. Particle intensities add.

44 Chapter 2 Historical Review: Experiments and Theories

sity I: emerges from B. When striking the screen, the two streams of particles
superimpose and the net intensity measured is

I = II +h (2.17)

This is nothing more than the statement that numbers of particles add.
Now consider the same experimental setup, but instead of a source of particles,
let S represent a source of waves, say water waves (Fig. 2.10). Waves are charac-
terized by an amplitude function 1/J such that the absolute square of this function
gives the intensity I :

(2.18)

(Absolute values are taken from complex amplitudes.) Let the two propagat-
ing wave disturbances have (complex) scalar (as opposed to vector) amplitudes
1/JI (r, t) and 1/J2(r, t) , respectively. These functions have the representations

(2.19)

where ex is the phase of the wave, which in general is also a function of (r, t) . The
intensities of these waves are

(2.20)

At a common value of rand t, the two wave amplitudes superimpose to give the
resultant amplitude:

1/J = 1/JI + 1/J2 (2.21)

Wall Screen

y
l
A )))

9))) J)))

FIGURE 2.10 Wave double-slit experiment. Amplitudes add.

Problems 45

The corresponding resultant intensity is

I == IVJ 12 == IVJl + VJ21 2 == (VJl + VJ2) (VJl + VJ2) * (2.22)

== IVJ11 2 + IVJ21 2 + IVJIVJ21[ei(a1-a2) + e- i (a 1- a 2 ) ]
== II + 12 + 2)/1 /2 cos(al - (2)

Comparing this with the resultant intensity I for the particle case (2.17), we note
that wave intensity carries the additional term

~ == 2J/1/2cos(al - (2) (2.23)

This is an interference term. As the y component of r traverses the screen in

Fig. 2.10, ~ oscillates and gives a pattern of the form depicted.
Hence, we have uncovered an operational, distinguishing characteristic be-
tween particles and waves. Waves exhibit interference, particles do not. Consider
the example of a propagating electric field E(r, t). The intensity of the wave (en-
ergy flux) is proportional to the time average of IEI 2. If two electric waves El and
E2 are superimposed, the new value of the electric field becomes

E == El + E2 (2.24)

The intensity is proportional to the time average of squared amplitude, lEI + E21 2.
So we have the following important rule: When two noninteracting beams of
particles combine in the same region of space, intensities add; when waves inter-
act, amplitudes add. The intensity is then proportional to the time average of the
absolute square of the resultant amplitude.

PROBLEMS

2.21 In a given wave double-slit experiment, a detector traces across a screen along a
straight line whose coordinate we label y. If one slit is closed, the amplitude

is measured. If the other slit is closed, the amplitude

v: =
is measured. What is the intensity pattern along the y axis if both slits are open?
46 Chapter 2 Historical Review: Experiments and Theories

2.6 • THE DE BROGLIE HYPOTHESIS AND THE

DAVISSON-GERMER EXPERIMENT

In preceding sections we have seen that for a consistent explanation of certain ex-
periments it is necessary to ascribe particle (photon) behavior to light. The energy
of such a photon of frequency v is E = h v. Its momentum is
E hv
p=-=- (2.25)
c c
This formula can also be written in terms of wavelength A. The relation between
A and v for light is particularly simple. It is

AV = c (2.26)

In terms of wavenumber k (em -I) and angular frequency w,

w = 2rrv, k = 2rr (2.27)

A
Equations (2.25) appear as

E = 1Uv, p = hk, w=ck (2.28)

The last of these three equations is called a dispersion relation. It reveals a linear
dependence between wand k. The significance of this is that the phase velocity
(w / k) of a monochromatic wave of frequency co is independent of to or k. It is the
constant c (speed of light) . If a wave packet composed of a collection of waves of
different wavelengths (or, equivalently, different wavenumbers) is constructed, it
propagates with no distortion (dispersion) . All component waves have the same
speed, c.
The first two equations of (2.28) reveal that photons , which are in essence
particles , are identified by two wave parameters : wavenumber k and frequency co.
Now in what sense is a photon different from other more familiar particles (e.g.,
electrons, protons, etc.)? A photon is special in that it has zero rest mass and
travels only at the speed of light. The more familiar particles with finite rest mass
also have wave properties. For a (nonrelativistic) particle of kinetic energy
p2
E=- (2.29)
2m
the wavelength for the corresponding ("matter") wave is
h
A= - or p = hk (2.30)
P
which we see from (2.25) and (2.26) is equally relevant to photons . Equa-
tion (2.30) is, in essence , the de Broglie hypothesis. It ascribes a wave prop-
erty to particles . While the Planck hypothesis, which assigned a particle quality
2.6 The de Broglie Hypothesis and the Davisson-Germer Experiment 47

to electromagnetic waves, had strong experimental motivation, the de Broglie

hypothesis , when first introduced in 1925, had little. Such motivation lay to a
large degree in the mystery that surrounded the Bohr recipe for the hydrogen
atom. What was the physical basis of the first rule for stationary orbits (2.6)? For
circular orbits of radius r, with electron momentum p , this rule gives

'Iu rp = nh (2.31)

In terms of the de Broglie wavelength A, the last equation reads

2:rrr = n): (2.32)

The stationary orbits in the Bohr model have an integral number of wavelengths
precisely fitting the circumference (Fig. 2.11). This is the classical criterion for
the existence of (standing) waves on a circle.
Thus, we see that the de Broglie hypothesis returns the stationary orbit radii
of the Bohr theory. This result lends support to the idea that the electron has
something "wavy" associated with it, this property being characterized by the
de Broglie wavelength (2.30). It was not until two years later (1927) that M. Born
suggested what is believed today to be the correct interpretation of this wave prop-
erty (see Section 2.8).
If electrons (in some respect) propagate as waves, they should exhibit inter-
ference. This is the essence of the Davisson-Germer experiment (1927). Reflect
a beam of electrons with well-defined momentum (therefore , wavelength) off a
crystal surface whose ion sites are separated by a distance a (the lattice constant)
which is of the order of the de Broglie wavelength of the electrons. In the actual
experiment, low-energy (,...., 200-eV) electrons were reflected from the face of a

FIGURE 2.11 De Broglie wavelength Aand n = 4 Bohr orbit of the hydrogen atom.
48 Chapter 2 Historical Review: Experimentsand Theories

-l
a
T
The phase difference ~
~ = 2a cos 8 =nX

FIGURE 2.12 Reflection of plane waves from a lattice. Conditions stated are for con-
structive interference, with n an integer.

nickel crystal (a = 3.52 x 10- 8 em). An interference pattern was observed which
could most consistently be interpreted as the diffraction of plane waves (with
de Broglie wavelength) by the regularly spaced atoms of the crystal (Fig. 2.12).
To bring out the full physical interpretation of these results we will consider
a simpler experiment in which the same principles are involved. We revert to the
two-slit configuration depicted in Fig. 2.10. The source S is able to eject single
electrons with well-defined momenta- p = hk. This is a vector normal to the
diffracting wall. The distance between S and the diffracting wall is large compared
to the distance between slits. The screen is composed of scintillation material.
When an electron hits it, there is a localized flash at the point of impact.
In any single run of this experiment, one sees a single localized flash on the
screen. There is no interference pattern. If we record the number and location
(idealized to one dimension , y) of these flashes, the results of 5 runs are shown in
Fig. 2.13a; 10 runs in Fig. 2.13b; 50 runs in Fig. 2.13c; 10,000 runs in Fig. 2.13d.
The solid curve is the theoretically calculated diffraction pattern obtained with the
de Broglie wavelength.
The electrons begin to distribute themselves in an interference pattern. It fol-
lows that if we change the source to eject a current pulse containing many elec-
trons, the scintillation plate will show an interference pattern.
Similarly, for a source of light we can use a detection plate made of many
photomultipliers. If the source emits a single photon, a single pulse from one of
the photomultiplier tubes is registered. There is no diffraction pattern. A single
particle under any circumstance always gives a single localized "flash." Wherein
lies the wave quality of particles? Clearly, it is centered in a statistical interpreta-
tion of data. Such description was first presented by Born in 1927. Before turning
to this analysis, we will give a brief account of a discovery by Heisenberg, which
was to throw the well-established philosophical dogma of the seventeenth to nine-
teenth centuries into disarray.

3The consistency of this arrangement with the uncertainty principle is discussed in the next section.
 N

y
(a)

y
(b)

y
(c)

y
(d)

FIGURE 2.13 Number and location of flashes in electron double-slit experiment de-
picted in Fig. 2.10. Each point is an average of flashes over a unit interval. The solid curve
is the theoretical interference pattern corresponding to the de Broglie wavelength.
49
50 Chapter 2 Historical Review: Experiments and Theories

PROBLEMS

2.22 A photon of energy hv collides with a stationary electron of rest mass m . Show that
it is not physically possible for the photon to impart all its energy to the electron.
Answer
We must do this problem relativistically. Let us assume that the photon does give up
all its energy to the electron. Conservation of energy and momentum then give

hv + me2 = mye 2
hv
- =myf3e
e
where
v
f3 =-,
e
m = rest mass
The speed of the electron after collision is v. Eliminating hv from the conservation
equations gives

yf3 = y - 1

whose only (real) solution is f3 = 0 (y = I) . This is a contradiction.

2.23 Show that the de Broglie wavelength of an electron of kinetic energy E (eV) is

12.3 x 10- 8
Ae = £1 /2 em

and that of a proton is

0.29 x 10- 8
Ap = £1 /2 em
2.24 At what speed is the de Broglie wavelength of an ex particle equal to that of a lO-keV
photon ?
2.25 Show that in order to associate a de Broglie wave with the propagation of photons
(electromagnetic radiation), photons must travel with the speed of light e and their
rest mass must be zero. (Do relativistically.)
Answer
For a de Broglie wave associated with a particle of rest mass m,
h h
A=-=-
P my v
For a photon with rest mass m,
e he he
A=-=-=--
v hv mye2
Equating these relations gives

v=e

which gives a noninfinite mass, ym, only for m = O.

Problems 51

2.26 The relativistic kinetic energy T of a particle of rest mass m and momentum p =
ymvis

(a) Show that

(b) Show that in the limit f3 « 1,

(c) Show that the relativistic expression for T above gives the correct energy-
momentum relation for a photon if m = O.
(d) What is the total relativistic energy, E (i.e., including rest-mass energy) of a
particle of mass m?
(e) What is the total relativistic energy of a particle moving in a potential field V (x) ?
What is the corresponding Hamiltonian, H (p, x)?

Answers (partial)
(d) E = ymc Z = ,j'p"Zc""""Z-+- m
- "Zc·4 .
(e) E = ymcZ + V(x); H = ,j'p"Zc""""Z-+-m-,,-Zc...4 + Vex) .
(An extensive discussion of relativistic quantum mechanics is presented in Chap-
ter 15.)

2.27 Assuming the sun to be a blackbody with a surface temperature of 6000 K, (a) calcu-
late the rate at which energy is radiated from it. (b) Determine the loss in solar mass
per day due to this radiation.

2.28 In 1922, A. H. Compton applied the photon concept of electromagnetic radiation to

explain the scattering of x rays from electrons. In the analysis it is assumed that a
photon of energy h v and momentum h vI c = hf): is incident on a stationary but
otherwise free electron of rest mass m. The photon scatters from the electron. Its
new momentum, hi)..', makes an angle {} with the incident (old) momentum. The
momentum of the recoiling electron makes an angle tP with the incident momentum
(Fig. 2.14). Ifthe system of electron and photon is an isolated system, its energy and
total momentum are constant. Conservation of energy reads

hv + mc Z = hv' + mycZ
Conservation of momentum (the whole collision occurs in a plane) gives

~ = myf3ccostP + (~) cos e

0= -myf3csintP + (~) sin{}
52 Chapter 2 Historical Review: Experiments and Theories

Scattered photon
Energy = hv'
Momentum:
Incident photon p; = (hi"") cos e
Energy =hv
)' t p:. = (hi"") sin e

i:r
Momentum: I r

:
Px=hvlc I
P =0 I Electron
Y I at rest

J- 1/
---f--------
I
I
I
/Energy = In C 2
Momentum = 0
--------t~~--~
I "1()
I
I
I I Recoil electron
I I Energy =-Ylnc2
Momentum:
P;
=1n-y(3c cos If,
P;
=1n-y(3c sin ¢
(a) (b)

FIGURE 2.14 Angles () and ¢ in the Compton scattering of photons from electrons . (See
Problem 2.28.)

Using these three conservation equations, derive the Compton effect equation for the
difference in wavelengths :

A' - A = ACO - cos e)

h
AC = the Compton wavelength = - = 2.43 x 10- 10 cm
me

2.29 The "classical radius of the electron," ro is obtained by setting the potential e2 / ro
equal to the rest-mass energy of the electron, mc 2 .

e2
- =mc 2
ro

Show that the successive powers of the fine-structure constant

e2
(X = he
are measures of the Bohr radius to the Compton wavelength ; and of the Compton
wavelength to the classical radius of the electron . That is, show that

ao: Xc : ro = I : (X : (X2
 2.7 The Work of Heisenberg. Uncertainty as a Cornerstone of Natural Law 53

2.7 • THE WORK OF HEISENBERG. UNCERTAINTY AS A CORNERSTONE

OF NATURAL LAW

It is an essential feature of Newton 's second law that, given the initial coordinates
and velocity of a particle , r(O) and r(O), respectively, and knowing all the forces
on the particle, the orbit r(t) is uniquely determined. The same holds true for a
system of particles . This is the essence of determinism. Laplace , in the eighteenth
century, took the implications of the latter statements to their extreme : The entire
universe consists of bodies moving through space and obeying Newton's laws.
Once the interaction between these bodies is precisely known and the position
and velocities of all the bodies at any given instant are known, these coordinates
and velocities are determined (through Newton's second law) for all time.
Quantum mechanics was to bring down the walls of this deterministic philos-
ophy. The instrument of destruction was the Heisenberg uncertainty principle.
What Heisenberg put forth in 1927 implied the following : If the momentum of
a particle is known precisely, it follows that the position (location) of that same
particle is completely unknown . Quantitatively, if an identical experiment involv-
ing an electron is performed many times, and in each run of the experiment the
position (x) of the electron is measured, then although the experimental setup is
identical (same electron momentum) in each run, measurement of the position of
the electron does not give the same result. Let the average of these measurements
be (x) . Then we can form the mean-square deviation

(2.33)

The standard deviation is labeled LlX. If LlX is small compared to some typical
length in the experiment, one is more certain to find the value x = (x) in any
given run. If LlX is large, it is not certain what the measurement of x will yield
(Fig. 2.15) . For this reason LlX is also called the uncertainty in x .

P(x) P(x)

x < x> x

(a) (b)

FIGURE 2.15 (a) Large uncertainty in x : (Llx)2 = (x 2) - (x )2. (b) Small uncertainty
inx .
54 Chapter 2 Historical Review: Experiments and Theories

Similarly, one may speak of an uncertainty in any physically observable quan-

tity : magnetic field :B, energy E , momentum p, and so forth.

t:.:Bx = J((:Bx - (:Bx})2), t:.:B y = ...

t:.E = J((E - (E})2) (2.34)

t:.px = J((px - (Px})2), t:.Py = ...

Heisenberg's uncertainty principle relation for momentum Px and position x

(parallel components) appears as

(2.35)

If it can be said with certainty what the position of a particle is (t:.x = 0), then
there is total uncertainty regarding the momentum of the particle (t:.px = 00) .
Observable parameters obeying a relation such as (2.35) are called complementary
variables. Examples include (1) coordinates and momenta (x, Px); (2) energy
and time (E, t) ; and (3) any two Cartesian components of angular momentum
(L x , L y ) . Later in the text a formal technique is presented to determine if two
observables are complementary (as opposed to compatible).
When an electron (or photon) exists within a well-defined locality of space
(momentum is ill defined), it acts very much like a particle. This is the case since
in a double-slit experiment such a localized disturbance would only go through
one slit; and we can therefore follow it in time, so it is very much like a true
particle . When the electron does not exist in a well-defined locality of space, its
momentum can be defined more precisely. Under such circumstances the wave
character of the electron manifests itself. We cannot follow it . A whole wave is
propagating. Nevertheless, it should be borne in mind that when a scintillation
screen is put across its path one gets a single flash-although one has little idea
of when or where the event will occur.

PROBLEMS

2.30 Consider a particle with energy E = p 2/2m moving in one dimension (x). The
uncertainty in its location is Ilx . Show that if Ilxllp > h; then IlElll > h, where
(p lm)6.t = Ilx .

2.31 The size of an atom is approximately 10- 8 em. To locate an electron within the
atom, one should use electromagnetic radiation of wavelength not longer than, say,
10- 9 em. (a) What is the energy of a photon with such a wavelength (in eV)?
(b) What is the uncertainty in the electron's momentum if we are certain about its
position by 10-9 em?
 2.8 The Work of Born. ProbabilityWaves 55

2.8 • THE WORK OF BORN. PROBABILITY WAVES

When discussing the double-slit wave diffraction experiment and the Davisson-
Germer experiment, we found it appropriate to introduce an amplitude function
1/1, the square of whose modulus , 11/1 12 , was set equal to the intensity (2.18) of the
wave.
Born suggested in 1927 that, when referred to the propagation of particles,
11/1 12 is more appropriately termed a probability density. The function 1/1 is called
the wavefunction (also the state function or state vector) of the particle . Quanti-
tatively, the Born postulate states the following (in Cartesian space). The wave-
function for a particle 1/1 (x , y, Z, t) is such that

11/11 2 dx dy d; = P dx dy d z (2.36)

where P dx dy d z is the probability that measurement of the particle's position at

the time t finds it in the volume element dx dy dz about the point x , y , z.
This statement is quite consistent with the discussions above relating to the
interference of photons or electrons . In all cases an interference pattern exhibits
itself when an abundance of particles is present. The wavefunction 1/1 generates
the interference pattern . Where 11/11 2 is large, the probability that a particle is
found there is large. When enough particles are present, they distribute themselves
in the probability pattern outlined by the density function 11/11 2 •
The rules of quantum mechanics (Chapter 3) give a technique for calculating
the wavefunction 1/1 to within an arbitrary multiplicative constant. The equation
one solves to find 1/1 is called the Schrodinger equation. This is a homogenous
linear equation. Suppose that we solve it and obtain a function 1/1. Then A 1/1 is
also a solution , where A is any constant. The Born postulate specifies'' A. For
problems where it can be said with certainty that the particle is somewhere in a
given volume V,

Iv 11/11 dx dy dz =
2
1 (2.37)

This is a standard property that probability densities satisfy. It is the mathematical

expression of the certainty that the particle is in the volume V.
As an example , consider the following one-dimensional problem. A particle
that is known somewhere on the x axis has the wavefunction

(2.38)

The frequency UJ and length a are known constants. The (real) constant A is to
be determined. Since it is certain that the particle is somewhere in the interval

4If A is complex it may be determinedonly to withinan arbitrary phase factor, e ia , where a is a real
number.
56 Chapter 2 Historical Review: Experiments and Theories

-00 < x < +00, it follows that

(2.39)

The nondimensional variable '1 = x]a. This calculation gives

(2.40)

The normalized wavefunction is therefore

(2.41)

For the stated problem, 11/1 12 as obtained from (2.41) is the correct probability
density.

PROBLEMS

2.32 The wavefunction for a particle in one dimension is given by

Another state that the particle may be in is

A third state the particle may be in is

Normalize all three states in the interval -00 < y < +00 (i.e., find AI, A2, and A3) .
Is the probability of finding the particle in the interval a < y < 1 when the particle
is in the state 1/13 the same as the sum of the separate probabilities for the states 1/11
and 1/12? Answer the same question for the interval -1 < y < + 1.

2.33 The energy density (ergs/cm-') of electromagnetic radiation is proportional to 8 2,

where 8 is the electric field. Present an argument to demonstrate that 8 2 dr is a
measure of the probability of finding a photon in the volume element dr. Assume a
monochromatic radiation field.

2.34 Suppose that in a sample of 1000 electrons, each has a wavefunction

 2.9 Semiphilosophical Epilogue to Chapter 2 57

Measurements are made (at a specific time, t = t') to determine the locations of
electrons in the sample. Approximately how many electrons will be found in the
interval -! .s x ::: !? A graphical approximation is adequate.
2.35 A beam of monochromatic electromagnetic radiation incident normally on a totally
absorbing surface exerts a pressure on the surface of

82
p=u=-
8)1

where 8 is the amplitude of the electric field vector. If P = 3 x 10-6 dyne/cm 2 and
the wavelength of radiation is A = 8000 A, what is the photon flux (cm- 2/s) striking
the surface?

2.9 • SEMIPHILOSOPHICAL EPILOGUE TO CHAPTER 2

The wavefunction 1/1 affords information related to experiments on, say, an elec-
tron. Consider once again the double-slit experiment of Fig. 2.10. Again, we sup-
pose that the source is able to fire single electrons with well-defined momenta.
There is a corresponding (propagating) wavefunction 1/I(r, t) that is diffracted by
the slits. When measurements are made, the scintillation screen gives a single
flash (for a single electron). If we calculate 11/11 2 at the screen, we find an inter-
ference pattern. What is the significance of this pattern? Suppose that the electron
is a bullet. We can playa quantum mechanical Russian roulette. The game is to
stand at the screen so that the bullet misses you. The first thing to do is solve
the Schrodinger equation and calculate 11/11 2 at the screen. Stand where it is mini-
mum. But this, of course, does not guarantee that the bullet does not find its mark.
The laws of nature do not provide a more definite knowledge of the electron 's
trajectory.
Now when a pulse of electrons (assume that they are all independent of one
another) is fired at the slits, the scintillation screen registers an interference pat-
tern. Eventually (i.e., when a sufficient number are fired), the electrons begin to
follow the dictates of 11/1 21 and fall into place (Fig. 2.13). It is interesting that all
the information in 1/1 cannot be extracted from an experiment on one electron.
To get this information one has to do many experiments , each of which involves
many more than one electron.
At this point the reader may well ask: If electrons are particles, why not follow
their trajectories through the slits (an electron can only go through one slit at a
time) and onto the screen? One could then add the intensities of particles stem-
ming from each individual slit and obtain the (noninterference) pattern depicted
in Fig. 2.9. Well, we can do exactly that, and the interference pattern does vanish.
In the process of "watching" the electrons, the interference is destroyed .
This is seen as follows. Let us see if we can discern which slit the electron
goes through. The uncertainty in measurement of its y coordinate must obey the
inequality
58 Chapter 2 Historical Review: Experiments and Theories

d
8.y « 2" (2.42)

If the interference pattern is not to be destroyed, the uncertainty in an elec-

tron's y momentum 8.Py, induced by encounter with a photon, must be substan-
tially smaller than that which would displace the electron from a maximum in
the interference pattern to a neighboring minimum . With the aid of Fig. 2.16 this
condition is
e h
8.py « 2"Px = 2d (2.43)

In the latter equality we have recalled the de Broglie relation (2.30).

The first inequality for 8.y , (2.42), enables one to observe which slit the elec-
tron goes through. The second inequality for 8.Py, (2.43), guarantees the preser-
vation of the interference pattern. Combining these two inequalities gives the re-
lation

(2.44)

which is in contradiction to the Heisenberg uncertainty principle.

~ ~-::=================[---_l~~~::::=--l--
p..
- -.... d

d sin 8

Illuminating
photons

FIGURE 2.16 First maximum at 0 = 0; second maximum at sinO ~ 0 = A/d . The

angle between the first minimum and the second maximum = 0/2 = A/2d.
2.9 Semiphilosophical Epilogue to Chapter 2 59

We conclude that if it is possible to observe which slit electrons go through,

their interference pattern is destroyed. In observing the positions of the electrons,
their wave quality (e.g., interference-producing mechanism) diminishes. When
the light (whose photons are illuminating the electrons' path) is switched off, the
interference pattern reappears.
In general, we may note the fundamental rule that quantum mechanics does
not delineate the trajectory of a single particle. One may calculate the probability
that an electron is in some region of space, but this is again a probability and not
a guarantee that the electron will be found there. To realize this probability, one
must in principle observe many experiments on the same system with identical
initial conditions obeyed in each experiment. Average results then fall to the dic-
tates of quantum mechanics. In this regard , Einsteirr' has remarked that quantum
mechanics is incapable of describing the behavior of a single system (such as an
electron).

Hidden Variables
There is another, somewhat philosophical school of thought (due primarily to
Bohm and de Broglie") which holds that the impossibility of quantum mechanics
to predict with certainty the outcome of a given measurement on an individual
system stems from one 's inability to know the exact values of certain hidden vari-
ables relating to the system . In this description, the wavefunction is viewed as a
mathematical object that contains all the information one possesses regarding an
incompletely known system. Quantum formulas should emerge as averages over
the hidden parameters in much the same way as the laws of classical physics do
in fact follow in averaging over the quantum equations,"
Bethe 8 has argued that the existence of such hidden variables for an elec-
tron would imply electronic degrees of freedom which are not specified in atom
physics . However, the success of the present theory in formulation of the periodic
table indicates that this is not the case (i.e., no further degrees of freedom exist). 9
In Chapter 3 we discuss the postulates of quantum mechanics. These are clear-
cut formal statements whose mastery enables the student to treat many problems
in the quantum mechanical domain . In addition, a deeper understanding is gained
of some of the questions raised in this semiphilosophical epilogue. The notion
of hidden variables is returned to in Section 11.13 in discussion of the Einstein-
Podolsky-Rosen paradox .

Sp. A. Schlipp, ed., Albert Einstein : Philosopher-Scientist. Harper & Row, New York, 1959.
6D. Bohm , Phys. Rev. 85, 166, 180 (1952) : L. de Broglie, Physicien at Penseur, Albin Michel , Paris,
1953. For further discussion and reference , see J. S. Bell. Rev. Mod. Phys. 38,447 (1966) ; F. J. Belin-
fante, A Survey ofHidden- Variable Theories , Pergamon Press. New York, 1973.
7This relation between quantum and classical mechanics is called Ehrenfest 's principle and is dis-
cussed fully in Chapter 6.
8H. A. Bethe and R. W. Jackiw, lntermediate Quantum Mechanics , 2d ed., W. A. Benjamin, New
York, 1968.
9Experimental evidence obtained by S. Freedman and J. Clauser, Phys. Rev. Lett. 28, 938 (1972), also
appears to point against a hidden-variable theory.
60 Chapter 2 Historical Review: Experiments and Theories

PROBLEMS

2.36 In deriving the Planck radiation formula (2.3), one first sets

u( v) = hvn(v)

where n(v) dv is the density of photons in the frequen cy interval v, v + dv . An

expression for n (v) is obtained through the relation

n(v) = 9(V)fBE(V)
where 9 (v) dv is the density of modes (i.e., vibrational states) in the said frequency
interval and f BE, the Bose-Einstein factor, gives the average number of photons per
mode at the frequency v. Use Planck 's hypothesis to obtain the expression

I
fBE(V) = ehv /kBT _ I

[The calculation of 9 (v) is considered in Problem 2.37.]

Answer
We seek the average number of photons per mode at the frequency v. At this fre-
quency the modes of excitation of the radiation field have energies hv; 2hv , 3hv, .. ..
Let us assume that the probability that the Nth energy mode is excited is given by
the Boltzmann distribution

p(N) = e- NX/ f
N=O
e- Nx

There are N photons of frequency v in the Nth mode . Averaging over N gives

fBE = (N) = LNP(N)

N

= LNe-NX/Le-NX = -~In Le-Nx

N N b N

= -~ In L(e-x)N
ax N
= --a I a
- - = -In(l - e
-x
)
ax In -
I - e- axX

2.37 In Problem 2.36 we noted that the number of photons per unit volume in the fre-
quency interval v, v + dv , is given by

n(v) = 9(V)fBE (~)

kBT
Problems 61

(a) Calculate the density of states 9 (v) , assuming that the blackbody radiation field
consists of standing waves in a cubical box with perfectly reflecting walls and
edge length a.
(b) Obtain the Rayleigh-Jeans law for the radiant energy density URJ (v) , assuming
the classical equipartition hypothesis for the electromagnetic field: that is, each
mode of vibration contains k B T ergs of energy.
(c) Make a sketch of u R J (v ) and compare it to the Planck formula for u (v) . In what
frequency domain do the two theories agree?
(d) What property of the vibrational energy levels of the radiation field (at a given
frequency) allows the classical description (i.e., URJ) to be valid?

Answers (partial)
(a) The spatial components of a standing electric field in a cubical box of volume
V = a 3 , with perfectly reflecting walls, are

Ex = Acoskxxsinkyysinkz Z
E y = B sin kxx cos kyy sin kzz
Ez = Csinkxxsinkyycoskz z

These fields have the required property that the tangential component of E van-
ishes at all six walls provided that

kya = nyJr,
where n x , ny, and nz assume positive integer values. There is a mode of vibration
for each triplet of values (nx, ny , nz) . We seek the number of such modes in the
frequency interval v, v + dv . First note that for each mode the square sum

n2 = nx 2 + ny 2 + nz2 = (a-;)2 (kx 2 + ky 2 + k z2)

is proportional to the square of v, the frequency of vibration (2Jrv = ck). Next

consider Cartesian n space with axes n x , ny , and nz . Each point in this space
corresponds to a mode of vibration . It is clear that all points which fallon a
spherical surface of radius (2av/ c) correspond to modes at the frequency v. It
follows that the number of modes in the frequency interval v, v + dv , is given by
the volume in n space of a spherical shell of thickness dn and radius n (Fig. 2.17):

The factor 2 enters because of the two possible polarizations of an electric field
in a given mode. The factor ~ is due to the fact we wish only to consider positive
frequencies so that only that portion of the shell in the first octant is counted.
62 Chapter 2 Historical Review: Experiments and Theories

Volume of spherical
i
shell = x 4lTn 2 dn

FIGURE 2.17 Cartesian n space for the enumeration of standing electromagnetic wave
states in a box of edge length a. All points that fall in the shell of thickness dn and radius
(2av lc) correspond to modes with frequencies in the interval v,v + dv . (See Problem
2.37.)

This gives the desired result:

8Jfv 2
g(v) = -3-
c
(b) If there is kBT energy per mode, the spectral energy density is

8Jfv 2
URJ = -3-kBT
c

2.38 For a gas of N noninteracting particles (an ideal gas) in the thermodynamic equilib-
rium at temperature T and confined to volume V, the pressure P is given by

PV = N (pxvx)
The momentum of a particle is p and v is its velocity. The average is taken over all
particles in the gas. Show that for a gas of photons, this relation gives
I
PV= -E
3
while for a gas of mass points it gives
2
PV =-E
3
Problems 63

The total energy of the gas is E . (Note: The principle of equipartition of energy
ascribes equal portions of energy, on the average, to each degree of freedom of a
particle. Thus, if the average energy of a mass particle is c, then (~m vx 2) = e 13.)
2.39 The energy density U = E 1V of a blackbody radiation field is a function only of
temperature , U = U (T) . Using this fact together with the result of the last problem,
E = 3PV, show that Stefan's law, U = (4(1 Ic)T 4 , follows from purely thennody-
namic arguments, thereby establishing the law as a classical result.
Answer
The first two laws of thennodynarnics give

TdS =dE + PdV

The second law defines the entropy S, while the first law gives the conservation of
energy statement in the form: heat added = increase in internal energy + work done.
Using the given relations permits this equation to be rewritten
V 4U
dS = -dU + --dV
T 3T
We recognize this equation to be in the form

dS= -as I dR+ -as I dV

au v sv U

It follows that

aV;6TI =~aU(T)ITI
av U(T) 3 au v
which integrates to the desired result.
2.40 An adiabatic process is one in which a system exchanges no heat with its environ-
ment.
(a) What form does the first law of thennodynarnics assume for an adiabatic pro-
cess?
(b) Using the form obtained in part (a), show that for an adiabatic expansion of a
blackbody radiation field,

V T3 = constant
(c) Consider that the primeval fireball described in Problem 2.10 contains mass
M = p V and radiation energy E = U V, where V is volume. Show that in an ex-
panding universe, the radiant density U decreases faster than the mass density p .
Thus, although it is believed that the radiation density of the primeval fireball
far exceeded its mass density, the fact that in our present universe mass density
dominates over radiation density is seen to be consistent with an adiabatically
expanding universe in which Stefan's law holds.

Answers (partial)
(a) Since d S = 0 for an adiabatic process, the first law becomes

dE+PdV=O
64 Chapter 2 Historical Review: Experiments and Theories

(b) Hint: To find P = P(T) , use E = 3P V in conjunction with Stefan's law.

(c) Hint: Compare p(V) at constant M with U(V)adb '

2.41 Two plates of an ideal parallel-plate capacitor are made of platinum and silver, re-
spectively. When the plates are brought to the displacement 10- 3 em, electrons "tun-
nel" through 10 the potential barrier, thereby creating a contact potential. What is the
electric field (V/cm) between the plates? Which metal is left positive?
2.42 As described in Section 2.3, in a sample of metal at absolute zero, electrons com-
pletely fill the lowest levels such that no more than one electron occupies each state.
All levels are filled from zero energy to EF , the Fermi energy (see Fig. 2.5). Let
9 (E) dE represent the number of energy states that are available for occupation in
the interval E , E + d E; per unit volume. Since each state from 0 to E F is occupied,
the number of free electrons, n, per unit volume is given by the number of available
states per unit volume in this interval. That is,

rEF
n= 10 g(E)dE

For free electrons, the density ofstatesg (E) (see Section 8.8) is given by

(a) Using the expressions above, obtain an explicit formula for E F for a metal with
n free electrons per unit volume.
(b) Given that EF(CU) = 7.0 eV and EF(Na) = 3.1 eV, use your formula to obtain
the density (cm- 3) ofCu and Na nuclei, respectively, in samples at absolute zero.
(The periodic chart is given in Table 12.4.)

Answer (partial)
(a) EF = (h 2 /2m)(3n /81f)2 /3 .
2.43 (a) Show that when expressed in terms of angular frequency w, the radiant energy
density relevant to a blackbody radiation field is given by

1iw3 / 1f2C3
u(w) = e/iw/kBT _ I

(b) Show that the corresponding density of states is given by

w2
g(w) =-n
1f C

(c) Answers to the preceding questions may be found by setting

u(w) dw = u(v) dv
What is the physical meaning of this equality? (See also Problem 2.1.)
2.44 (a) What is the wavelength of an electromagnetic wave of frequency v propagating
in vacuum?
IOThe mechanism of quantum mechanical tunneling is described in Section 7.7.
Problems 65

(b) What is the de Broglie wavelength of a photon of frequency v?

(c) How are the wavelengths of parts (a) and (b) related?

2.45 An idealized model of a plasma (i.e., a fluid of electrons and ions) is given by a so-
called one-component plasma (OCP) . In this configuration electrons move in a uni-
formly distributed charge-neutralizing background. This background is sometimes
referred to as "jellium," It was originally suggested by E. P. Wigner that a high-
density OCP will crystallize at some critical value of electron density as this param-
eter is relaxed from its extreme high value.
(a) What is the potential energy of an electron situated at the center of a sphere of
radius a which is uniformly filled with positive neutralizing charge?
(b) What is the uncertainty in momentum of an electron confined to a region of
dimension a? What is the corresponding kinetic energ y Ek ?
(c) Assuming that crystallization will occur when potential energy per particle ex-
ceeds uncertainty energy, obtain a criterion for crystallization in terms of the
distance between electrons a and the Bohr radius ao.

Answers
(a) The charge density of the neutralizing medium is

e
p=--
i n a3
3

The potential of interaction between electron and positive jellium in a shell of

thickness dr at r is

ep4nr 2 dr
dV=....:....---
r
Thus
2
v= fa ep4nr dr = 3e2
Jo r 2a

(b) From the uncertainty relation (2.35) we have tsp ~ li/a so that

(Llp)2 li2
Ek~--=--
2m 2ma 2

(c) The relation V > Ek gives the order-of-magnitude criterion,

a > ao

where ao is the Bohr radius . Thus , Wigner crystallization can be expected to

occur when interparticle spacing exceeds the Bohr radius .

2.46 In a nuclear reactor, uranium atoms break apart and release several high-energy neu-
trons . These neutrons have an initial energ y of 2 MeV, but after many collisions with
other atoms they slow down to an average energy of 0.025 eV. Thus there will be a
continuous distribution of neutrons with energies ranging from 2 MeV to less than
0.025 eV.
66 Chapter 2 Historical Review: Experiments and Theories

(a) What are the de Broglie wavelengths of the 2-MeV neutrons and the 0.025-eV
neutrons ? Give your answer in A.
(b) A "low-pass" energy filter can be constructed by passing the neutrons through
a long piece of polycrystalline graphite. Explain briefly why low-energy "par-
ticles" can travel straight through the block but high-energy "particles" are re-
flected out the sides.
(c) What is the shortest de Broglie wavelength of the neutrons that are totally trans-
mitted through the crystal? Give your answer in A. Spacing between lattice
planes in graphite is a :::: 2 A.

Answer (partial)
(c) The condition for constructive interference from adjacent planes (Fig. 2.12) is

n): = 2acosB ::: 2a

Thus for reflection , A ::: 2a = Amax .

2.47 de Broglie waves are incident on a crystal with lattice planes separated by a. Show
that constructive interference of waves reflected from planes separated by 2a is ruled
out by parallel reflection from planes in between .

2.48 As described in Section 7.10, a criterion which discerns if a given configuration is

classical or quantum mechanic al may be stated in terms of the de Broglie wave-
length A. Namely, if L is a scale length characteristic to the configuration at hand,
then one has the following criteria :

A« L: Classical
A ?:, L: Quantum mechanical

Use these criteria to describe which physics is relevant to the following configura-
tions :
(a) An atomic electron . For the typical length choose the Bohr radius. For typical
energy choose the Rydberg .
(b) A proton in a nucleus . For nuclear size choose c; 10- 13 em (1 fermi) . For energy
choose c; 10 MeV .
(c) An electron in a vacuum tube operating at 10 kV.
(d) An electron gas of densit y 1020 em -3 and temperature 300 K.

!
2.49 A particle constrained to move on the x axis has a probabilit y of for being in the
interval (-d - a , -d + a) and ~ for being in the interval (d - a, d + a), where
d » a.
(a) Sketch the wavefunction that describes this situation .
(b) Call the normalized wavefunction for the "left" interval rp_(x) and that for
the "right" interval rp+(x) . What is the normalized wavefunction rp(x) for the
particle?
(c) Again, with d » a, what is the probability density P(x) for the particle ? What
does the integral over all x of P(x) give?
Problems 67

Answers (partial)

(b) rp± = Ifrp- ± ~rp+

(c) P(x) = Irp±I Z = !lrp_I Z + ~Irp+e
Note that for d » a, rp+rp_ = O.
1 4
1
00
P(x)dx = -+ - = 1
-00 5 5
Note: For mutually exclusive events a and b, the summational probability PI (a) +
PZ(b) gives the probabilit y of a or b occurring ; whereas for independent events,
the product probability PI (a)PZ (b) gives the probability of a and b occurring (and
is called the joint probability .) Note further that mutually exclusive events are not
independent.
2.50 Explain the meaning of the following statement: Interference between photons does
not occur. A photon can only interfere with itself. II
2.51 Current in a copper wire from a 100 V source carries power to a household appli-
ance. Is an electron in this current classical or quantum mechanical ? [Hint: Calculate
the deBroglie wavelength of the electron and compare it to the lattice constant for
copper.]

II For further discussion see P. A. M. Dirac, The Principles ofQuantum Mechanics , 4th ed., Section 3,
Oxford University Press, New York, 1958.
CHAPTER
The Postulates of Quantum

3 Mechanics. Operators,
Eigenfunctions, and Eigenvalues

3.1 Observables and Operators

3.2 Measurement in Quantum Mechanics
3.3 The State Function and Expectation Values
3.4 Time Development ofthe State Function
3.5 Solution to the Initial- Value Problem in Quantum Mechanics

In this chapter we consider four basic postulates of quantum mechanics, which

when taken with the Born postulate described in Section 2.8, serve toformalize the
rules of quantum mechanics. Mathematical concepts material to these postulates
are developed along with the phys ics. The postulates are applied throughout the
text. We choose the simplest problems first to exhibit their significance and method
of application-that is, problems in one dimension .

3.1 • OBSERVABLES AND OPERATORS

Postulate II
This postulate states the following : To any self-consistently and well-defined ob-
servable in physics (call it A), such as linear momentum, energy, mass, angular
momentum, or number of particles, there corresponds an operator (call it A) such
that measurement of A yields values (call these measured values a) which are
eigenvalues of A. That is, the values, a, are those values for which the equation

AqJ = aqJ an eigenvalue equation (3.1)

has a solution tp, The function qJ is called the eigenfunction of A corresponding to

the eigenvalue a.
Examples of mathematical operators, which are not necessarily connected to
physics, are offered in Table 3.1. (Labels such as D, G, and M are of no special
significance.) An operator operates on a function and makes it something else
(except for the identity operator 1).
Let us now tum to operators that correspond to physical observables. Two very
important such observables are the momentum and the energy.
1The orde r in which these postulates appear is by no means conventional.

68
3.1 Observables and Operators 69

TABLE 3.1 Examples of operators

v = a/ax Vrp(x) = arp(x) /ax
.& = -a 2/ax 2 = _v2 '&rp(x) = -a 2rp(x) /ax 2
M = a2/ax ay Mrp(x , y) = a2rp(x, y) /ax ay
i = operation that leaves rp unchanged irp=rp
Q = fJ dx ' Qrp(x) = fJ dx 'rp(x')
F = multiplication by F(x) Frp(x) = F(x)rp(x)
iJ = division by the number 3 iJrp(x) = 1rp(x)
8 = operator that annihilates rp 8rp =0
P = operator that changes rp to a Prp = rp3 _ 3rp2 - 4
specific polynomial of rp
G= operator that changes rp
to the number 8

The Momentum Operator p

The operator that corresponds to the observable linear momentum is

p= -ihV (3.2)

What are the eigenfunctions and eigenvalues of the momentum operator? Con-
sider that the particle (whose momentum is in question) is constrained to move in
one dimension (x) . Then the momentum has only one nonvanishing component,
P»- The corresponding operator is

Px = -In-
.~ a (3.3)
ax
The eigenvalue equation for this operator is

. a
-Ih-f{! = Pxf{! (3.4)
ax
The values Px represent the possible values that measurement of the x component
of momentum will yield. The eigenfunction f{! (x) corresponding to a specific value
of momentum (Px) is such that If{! 12 dx is the probability of finding the particle
(with momentum Px) in the interval x, x + dx. Suppose we stipulate that the
particle is afree particle. It is unconfined (along the x axis). For this case there is
no boundary condition on f{! and the solution to (3.4) is

T
f{! = A exp ( iPxX ) = Ae'ikx (3.5)
70 Chapter 3 The Postulates of Quantum Mechanics

where we have labeled the wavenumber k and have deleted the subscript x.

(3.6)

The eigenfunction given by (3.5) is a periodic function (in x). To find its wave-
length ).., we set

(3.7)
1= e i kA
= ccs k); + i sink)"
which is satisfied if

cos k): = 1 (3.8)

sinH = 0

The first nonvanishing solution to these equations is

H = 2Jr (3.9)
which (with 3.6) is equivalent to the de Broglie relation

h
p=- (3.10)
)..

We conclude that the eigenfunction of the momentum operator corresponding to

the eigenvalue p has a wavelength that is the de Broglie wavelength hip.
In quantum mechanics it is convenient to speak in terms of wavenumber k
instead of momentum p. In this notation one says that the eigenfunctions and
eigenvalues of the momentum operator are

p = hk (3.11 )

The subscript k on rpk denotes that there is a continuum of eigenfunction and

eigenvalues, hk, which yield nontrivial solutions to the eigenvalue equation, (3.4).

The Energy Operator it

The operator corresponding to the energy is the Hamiltonian iJ, with the momen-
tum p replaced by its operator counterpart, p. For a single particle of mass m, in
a potential field V (r),

(3.12)

The eigenvalue equation for iJ,

iJ rp(r) = Erp(r) (3.13)
3.1 Observables and Operators 71

is called the time-independent Schriidinger equation. It yields the possible ener-

gies E which the particle may have. Again consider the free particle. The energy
of a free particle is purely kinetic, so

A2 ",2
A P n. 2
H=-=--"ij (3.14)
2m 2m

Constraining the particle to move in one dimension, the time-independent

Schrodinger equation becomes

(3.15)

In terms of the wave vector

(3.16)

(3.15) appears as

(3.17)

The subscript x denotes differentiation. For a free particle there are no boundary
conditions and we obtain/

(3.18)

This is the eigenfunction of iJ which corresponds to the energy eigenvalue

(3.19)

We have found above (3.11) that the momentum of a free particle is hk, This is
clearly the same hk that appears in (3.19), since for a free particle

(3.20)

Note also that the ei~enfunction of iJ (3.18), with B = 0, is also an eigenfunction

of p (3.11) . That Hand p for a free particle have common eigenfunctions is
a special case of a more general theorem to be discussed later.' The following
simple argument demonstrates this fact. Let

pcp = hksp (3.21)

2The solution to (3.17) with boundary conditions imposed is discussed in Section 4.1.
3The commutator theorem, Chapter 5.
72 Chapter 3 The Postulates of Quantum Mechanics

Let us see if ep is also an eigenfunction of iI (for a free particle) .

iIep = L(pep) = p(nkep) = nk pep = (nk)2 ep (3.22)

2m 2m 2m 2m

It follows that ep is also an eigenfunction of iI.

Both the energy and momentum eigenvalue s for the free particle comprise a
continuum of values:

(3.23)

That is, these are valid eigenvalues for any wavenumber k. The eigenfunctions (of
both iI and p) corresponding to these eigenvalues is

(3.24)

If the free particle is in this state, measurement of its momentum will definitely
yield hk, and measurement of its energy will definitely yield (n 2k 2 12m).
Suppose that we measure its position x ; what do we find? Well, where is the
particle most likely to be? Again we call on the Born postulate. If the particle is
in the state epk, the probability density relating to the probability of finding the
particle in the interval x, x + dx , is

(3.25)

The probability density is the same constant value for all x. That means we would
be equally likely to find the particle at any point from x = -00 to x = +00 . This
is a statement of maximum uncertainty which is in agreement with the Heisenberg
uncertainty principle . In the state epk, it is known with absolute certainty that mea-
surement of momentum yields hk, Therefore, for the state epk, !:lp = 0, whence
!:lx = 00.
We mentioned in Section 2.7 that E and t are complementary variables; that
is, they obey the relation !:lE!:lt 2: It. Specifically, this means that if the energy
is uncertain by amount !:lE, the time it takes to measure E is uncertain by !:It 2:
nl!:lE. Now for the problem at hand, in the state epk, it is certain that measurement
of E yields n2k2/2m . Therefore !:lE = 0. To measure E we have to let the particle
interact with some sort of energy-measuring apparatus, say a plate with a spring
attached to measure the momentum imparted to the plate when the particle hits it
head on. Well, if the plate with attached spring is placed in the path of the particle,
how long must we wait before we detect something? We can wait 10-8 s-or we
can wait 1010 yr. The uncertainty !:It is infinite in the present case, since there is
an infinite uncertainty in x .
Problems 73

PROBLEMS

3.1 For each of the operators listed in Table 3.1 (D, ~,Nt, etc .), construct the square, that
A2 A2
is, D , t. , . . . .

Answer (partial)

Pip = iip = ip
Qip =Q hI dx'ip(x') = hI hIdx" dx'ip(x')

Pip = F 2ip
A2 1
B ip= 9ip

p2ip = P(Pip) = (ip3 _ 3ip2 _ 4)3 _ 3(ip3 _ 3ip2 _ 4)2 - 4

3.2 The inverse of an operator Ais written A-I . It is such that

Construct the inverses of D, i, fr, 8, e, G, provided that such inverses exist.

3.3 An operator o is linear if

where a and b are arbitrary constants. Which of the operators in Table 3.1 are linear
and which are nonlinear?

3.4 The displacement operator iJ is defined by the equation

iJf(x) = f(x + n
Show that the eigenfunctions of iJ are of the form
ipp = e Px 9 (x)
where

g(x + n = g(x)
and f3 is any complex number. What is the eigenvalue corresponding to sp B?
3.5 An electron moves in the x direction with the de Broglie wavelength 10- 8 ern,
(a) What is the energy of the electron (in eV) ?
(b) What is the time-independent wavefunction of the electron?
74 Chapter 3 The Postulates of Quantum Mechanics

3.2 • MEASUREMENT IN QUANTUM MECHANICS

Postulate II
The second postulate" of quantum mechanics is, measurement of the observable
A that yields the value a leaves the system in the state CPa , where CPa is the eigen-
function of A that corresponds to the eigenvalue a.
As an example, suppose that a free particle is moving in one dimension. We
do not know which state the particle is in. At a given instant we measure the
particle's momentum and find the value p = hk (with k a specific value, say 1.3 x
10 10 cm"), This measuremenf leaves the particle in the state CPko so immediate
subsequent measurement of p is certain to yield hk,
Suppose that one measures the position of a free particle and the position
x = x' is measured . The first two postulates tell us the following . (l) There is
an operator corresponding to the measurement of position, call it x. (2) Measure-
ment of x that yields the value x' leaves the particle in the eigenfunction of x
corresponding to the eigenvalue x'.
The operator equation appears as

x 8(x - x') = x'8(x - x') (3.26)

Dirac Delta Function

x
The eigenfunction of has been written? 8(x - x ') and is called the Dirac delta
function . It is defined in terms of the following two properties. The first are the
integral properties

I: f(x')8(x - x') dx' = f(x)

I: 8(x -x')dx' = 1
(3.27)

or equivalently, in terms of the single variable y

I: f( y)8(y) dy = f(O)

I: 8( y)dy = 1
(3.28)

4This postulate has been the source of some discussion among physicists. For further reference, see
B. S. DeWitt, Phys. Today 23,30 (September 1970).
5Measurement is taken in the idealized sense. More formal discussion on the theory of measurement
may be found in K. Gottfried, Quantum Mechanics, W. A. Benjamin, New York, 1966; J. Jauch,
Foundations of Quantum Mechanics, Addison-Wesley, Reading, Mass., 1968; and E. C. Kemble, The
Fundamental Principles of Quantum Mechanics with Elementary Applications, Dover, New York,
1958.
x
6More accurately one says that 8(x - x') is an eigenfunction of in the coordinate representation .
This topic is returned to in Section 7.4 and in Appendix A.
3.2 Measurement in Quantum Mechanics 75

The second defining property is the value

8(y) = 0 (for y =1= 0) (3.29)

A sketch of 8(y) is given in Fig. 3.1. Properties of 8(y) are usually proved with
the aid of the defining integral (3.27). For instance , consider the relation

y8'(y) = -8(y) (3.30)

To establish this relation we employ a test function f (y) and perform the follow-
ing integration by parts.

00 f(y)y8'( y) dy = 100
-00 dyd (fy8) dy (3.31)
1 -00

d
100
00
- 8-(yf)dy
-00 dy

= -1 8(y) (yd
f
+ f) dy

= - i:
-00 dy

8( y)f( y)dy

which establishes (3.30).

The student should not lose sight of the fact that x,
when operating on a
function f (x), merely represents multiplication by x . For example, f (x) = x
xf(x) . These topics will be returned to in Chapter 11 and discussed further in
Appendix A.

1If
b(y)=O ,y*O

Jb(y)dy = 1
bey)

Jb (y)[(y)dy =[(D)

FIGURE 3.1 Dirac delta function 8(y) . The curve is distorted to bring out essential
features. A more accurate picture is obtained in the limit E --+ O.
76 Chapter 3 The Postulates of Quantum Mechanics

PROBLEMS

3.6 Establish the following properties of o(y) :

(a) o(y) = o(-y)
(b) 0' (y)= -8'( -y)
(c) yo(y) =0
(d) o(ay) = lal-Io(y)
(e) o(yZ - a Z) = 12al- I[0(y - a) + o(y + a)]
(1) J~oo o(a - y)o(y - b) dy = o(a - b)
(g) f(y)o(y - a) = f(a)o(y - a)
(h) yo' (y) = -o(y)
• "O(y-Yi)
(I) o[g(y)] = z:
Yi
1'( ')1 '
g y,
g(Yi) =0
3.7 Show that the following are valid representations of o(y) :
(a) 2Jl'0(y) = J~oo ei ky dk
. sin ny
(b) Jl'o(y) = 11m - -
1]-->00 Y
Note: In mathematics, an object such as o(y), which is defined in terms of its in-
tegral properties, is called a distribution. Consider all X(y) defined on the interval
(-00,00) for which

L: Ix(y)IZdy < 00

Then two distributions, 01 and OZ, are equivalent if for all X(y),

L: L:XOI dy = xoZdy

When one establishes that a mathematical form such as J~oo exp(iky) dk is a repre-
sentation of o(y), one is in effect demonstrating that these two objects are equivalent
as distributions.

3.8 Show that the continuous set of eigenfunctions {o (x - x')} obeys the "orthonormality"

L:
condition

o(x - x')o(x - x") dx = o(x' - x'')

3.9 (a) Show that 0(.JX) = O.

(b) Evaluate8(Jx Z - a Z).

3.3 • THE STATE FUNCTION AND EXPECTATION VALUES

Postulate III
The third postulate of quantum mechanics establishes the existence of the state
function and its relevance to the properties of a system: The state of a system a!
3.3 The State Function and Expectation Values 77

any instant of time may be represented by a state or wave function 1/1 which is
continuous and differentiable. All information regarding the state of the system is
contained in the wavefunction. Specifically, if a system is in the state 1/I(r, t) , the
average of any physical observable C relevant to that system at time t is

(C) = f 1/1*(;1/1 dr (3.32)

(The differential of volume is written dr.) The average, (C), is called the expec-
tation value of C .
The physical meaning of the average of an observable C involves the following
type of (conceptual) measurements. The observable C is measured in a specific
experiment, X. One prepares a very large number (N) of identical replicas of X .
The initial states 1/I(r, 0) in each such replica are all identical. At the time t, one
measures C in all these replica experiments and obtains the set of values CI ,
Cz, . . . , C N. The average of C is then given by the rule
1 N
(C)=-LC; (N )> 1) (3.33)
N ;=1

The postulate stated above claims that this experimentally calculated average
(3.33) is the same as that given by the integral in (3.32) . Another way of defining
(C) is in terms of the probability P(C;) . This function gives the probability that
measurement of C finds the value C; . For (C) , we then have

(C) = L C; P(C;) (3.34)

all C,

This is a consistent formula if all the values C may assume comprise a discrete set
(e.g., the number of marbles in a box) . In the event that the values C may assume
comprise a continuous set (e.g ., the values of momentum of a free particle), (C)
becomes .

(C) = f CP(C)dC (3.35)

The integration is over all values of C . Here P (C) d C is the probability of finding
C in the interval C, C + dC .
The quantity (C) is also called the expectation value of C because it is repre-
sentative of the value one expects to obtain in any given measurement of C . This
will be especially true if the deviation of values of C from the mean value (C) is
not large. As discussed in Section 2.7, a measure of this spread of values about
the value (C) is given by the mean-square deviation !:lC, defined through

(3.36)

In order to become familiar with the operational use of postulate III. we work
out the following one-dimensional problem. A particle is known to be in the state
78 Chapter 3 The Postulatesof Quantum Mechanics

- (x - XO)2] (iPOX)
1/I(x,t) = Aexp
[
4a 2 exp T
exp(-iwo t) (3.37)

The lengths Xo and a are constants, as are the momentum PO and frequency woo
The (real) constant A is determined through normalization. This then ensures that
1/1*1/1 is a numerically correct probability density.

(3.38)

The nondimensional "dummy" variable TJ and constant TJo are such that
x - xo
TJ=--
a
x = a(1] + 1]0) (3.39)
Xo
1]0 = -
a
Having obtained A, we may now calculate the expectation of x:

(x) = i: 1/1*£1/1 dx = i: 1/1* x 1/1 dx (3.40)

= A 2a2 i: r
e- ? /2(1] + 1]o)dTJ = a1]O (aA
2
i: 2
e-T/2 / d 1])

which, with the normalization condition (3.38), gives

(x) = aTJo = xo (3.41)

[Note that integration of the odd integrand 1]exp( _TJ2 /2) in (3.40) vanishes .] That
Xo is the proper value for (x) is evident from the sketch of 11/1 12 shown in Fig. 3.2.
If we call

(3.42)

the probability of finding the particle in the interval x, x + dx, then

(x) = i: xP(x)dx (3.43)

This is consistent with definition (3.35).

The probability density

P(x) =
I
r;c exp
[-(X - 2XO)2]
a v2rr 2a
3.3 The State Function and Expectation Values 79

hW =P(x)

FIGURE 3.2 Gaussian probability density with variance a 2 . The variance measures the
spread of P(x) about the mean, (x) = xo. In quantum mechanics the square root of vari-
ance is called the uncertainty in x and is denoted as t..x, so for the case under discussion,
a = Sx = J(x 2) - (x)2 .

is called the Gaussian or normal distribution, and a 2 is called the variance of x .

It is a measure of the spread of P (x) about the mean value

(x) = xo
As shown in Problem 3.10, the variance of x is the same as the mean-square
deviation , (~x)2.
(~x)2 = (x 2) _ (x}2 = a 2 + xo 2 - xo 2 = a 2

If it is known that a particle is in the state y, (x) at a given instant of time, and that
in this state (x) = xo , one may then ask: With what certainty will measurement of
x find the value xo? A measure of the relative uncertainty is given by the square
root of the variance, ~x . If this value is large (compared to (x)), one may say with
little certainty that measurement will find the particle at xo. If, on the other hand,
~x is small, one is more certain that measurement will find the particle at x = xo .
In quantum mechanics ~x is called the uncertainty in x, introduced previously in
Section 2.7.
Next, we calculate the expectation of the momentum for a particle in the state
y" (3.37).

(p) = 1 00

y,* fiy, dx = 1 (-i n!-)

00

y,*
ax y, dx (3.44)

2
-00

= A a

= PO
i: -00

(PO + ~: rJ) e-T/2 /2 dn = PO ( A a

2
i: e-T/2 /2 drJ)

It follows that the parameter PO which appears in the state function y, is the aver-
age value of p. Note in particular that the values, (x) = Xo and (p) = PO, refer to
a specific time, t . As time increases beyond t , the "Gaussian" wavefunction (3.37)
80 Chapter 3 The Postulates of Quantum Mechanics

propagates and, because of various momentum components of this wavefunction ,

disperses. For further discussion, see (6.40).

PROBLEMS

3.10 For the state 1fr, given by (3.37), show that

(t..x)2 = a2
Argue the consistency of this conclusion with the change in shape that 11fr1 2 suffers
with a change in the parameter a.
3.11 Calculate the uncertainty Sp for a particle in the state 1fr given by (3.37). Do you
find your answer to be consistent with the uncertainty principle? (In this problem one
must calculate (fi2) . The operator fi2 = -!i2 a 2 /ax 2 .)
3.12 Let s be the number of spots shown by a die thrown at random.
(a) Calculate (s) .
(b) Calculate Ss ,
3.13 The number of hairs (NI) on a certain rare species can only be the number i (I =
0,1,2, . . . ). The probability of finding such an animal with i hairs is e- I / I! What
is the expectation, (N) ? What is t..N?

3.4 • TIME DEVELOPMENT OF THE STATE FUNCTION

Postulate IV
The fourth postulate of quantum mechanics specifies the time development of
the state function 1/J(r, t) : the state function for a system (e.g., a single particle)
develops in time according to the equation

a ~
ili-1/J(r, t) = H1/J(r, r) (3.45)
at
This equation is called the time-dependent Schriidinger equation," The operator
fI is the Hamiltonian operator. For a single particle of mass m, in a potential field
V (r), it is given by (3.12). If fI is assumed to be independent of time, we may
~w .
(3.46)

Under these circumstances, one is able to construct a solution to the time-

dependent Schrodinger equation through the technique of separation of variables.
7A formulation of the Schrodinger equation that has its origin in the classical principle of least action
has been offered by R. P. Feynrnan, Rev. Mod. Phys. 60. 367 (1948) . An elementary description of this
derivation may be found in S. Borowitz Quantum Mechan ics, W. A. Benjamin . New York, 1967.
3.4 Time Development of the State Function 81

We assume a solution of the form

1/J(r, t) = cp(r)T(t) (3.47)

Substitution into (3.45) gives

(3.48)

The subscript t denotes differentiation with respect to t . Equation (3.48) is such

that the left-hand side is a function of t only, while the right-hand side is a function
of r only. Such an equation can be satisfied only if both sides are equal to the same
constant, call it E (we do not yet know that E is the energy) .

ifcp(r) = Ecp(r) (3.49)

'E)T(t) =
( ata + ~ 0 (3.50)

The first of these equations is the time-independent Schrodinger equation (3.13).

This identification serves to label E, in (3.49), the energy of the system. That is,
E , as it appears in this equation, is an eigenvalue of if. But the eigenvalues of if
are the allowed energies a system may assume, and we again conclude that E is
the energy of the system .
The second equation (3.50) is simply solved to give the oscillating form

T(t)=Aexp
iEt)
(-Ii: (3.51)

Suppose that we solve the time-independent Schrodinger equation and obtain

the eigenfunctions and eigenvalues

(3.52)

For each such eigensolution, there is a corresponding eigensolution to the time-

dependent Schrodinger equation

(3.53)

In equations (3.52) and (3.53) the index n denotes the set of integers n = 1,2, . . . .
This notation is appropriate to the case where the solution to the time-independent
Schrodinger equation gives a discrete set of eigenfunctions, {CPn}. Such is the case
for problems that pertain to a finite system, such as a particle confined to a finite
domain of space. We will encounter this property in Chapter 4 when we solve
the problem of a bead constrained to move on a straight wire strung between two
impenetrable walls.
82 Chapter 3 The Postulates of Quantum Mechanics

In the one-dimensional free-particle case treated in Section 3.1, one obtains a

continuum of eigenfunctions fPdx) and, correspondingly, a continuum of eigen-
values, Ei , To repeat, these values are

(3.54)

fPk = A exp(ikx) , (3.55)

For each such time-independent solution, there is a solution to the time-dependent

Schrodinger equation

1/rdx, t) = Aei(kx-wt) (3.56)

where we have labeled

(3.57)

The structure of the solution (3.56) is characteristic of a propagating wave. More

generally, any function of x and t of the form

f(x , t) = f( x - vt) (3.58)

represents a wave propagating in the positive x direction with velocity v. To see

this, we note the following property of f:

f(x + v!:lt, t + !:It) = f(x , t) (3.59)

At any given instant t , one may plot the x dependence of f (Fig. 3.3). If t increases
to t +!:It , this curve is displaced to the right (as a rigid body) by the amount v St .
We conclude from these arguments that the disturbance f (3.58) propagates with
the wave speed v.
Now let us return to the free-particle eigenstate, (3.56), and rewrite it in the
form

1/rdx, t) = A exp [ ik (x - It)] (3.60)

Comparison with the waveform (3.58) indicates that (1) 1/r is a propagating wave
(moving to the right), and (2) the speed of this wave is

co luo p 2/2m p VCL

V=-=-=---=-=- (3.61)
k hk p 2m 2

The velocity vc L represents the classical velocity of a particle of mass m and

momentum p . Thus we find that the wave speed of the state function of a particle
with well-defined momentum, p = hk, is half the classical speed, Vc L = pi m .
This discrepancy is due to the following fact. Suppose that we calculate the
probability density corresponding to the state given in (3.56). We obtain the result
3.4 lime Development of the State Function 83

(a) x

f(x . t + AI)

x
, I
I• l'A t ------1
(b)

FIGURE 3.3 Propagating wave, [t», I) = f(x - VI) : (a) at time I; (b) at time 1+ 6./ .

"'(X,I)

FIGURE 3.4 Wavepacket at a given instant of time I.

that it is uniformly probable to find the particle anywhere along the x axis. This is
not a typical classical property of a particle. The state function that better repre-
sents a classical (localized) particle is a wave packet. The shape of such a function
is sketched in Fig. 3.4. Such a state may be constructed as a sum of eigenstates
of the form given in (3.56) (a Fourier series) . The velocity with which the packet
moves is called the group velocity,8

vg = -
aw (3.62)
ak
8The concepts of phase and group velocities are returned to in Section 6.1.
84 Chapter 3 The Postulates of Quantum Mechanics

For a wave packet composed of free-particle eigenstates, v9 takes the value

(3.63)

= VeL

The value of k that enters the formula for v9 is the value about which there is
a superabundance of llrk component waves. These topics will be more fully devel-
oped in Chapter 6. For the moment we are concerned only with the identification
given in (3.63).

PROBLEMS

3.14 Describe the evolution in time of the following wavefunctions:

0/1 = A sinwtcosk(x +ct)

0/2 = A sin(lO-5 kx) cosk(x - cr)
0/3 = A cos k(x - ct) sin[IO- 5k(x - ct)]

3.15 What is the expectation of momentum (p) for a particle in the state

o/(x , t) = Ae-(xja)2 e- iwt sinkx?

3.5 • SOLUTION TOTHE INITIAL-VALUE PROBLEM

IN QUANTUM MECHANICS

Functions of Operators
The time-dependent Schrodinger equation permits solution of the initial-value
problem : given the initial value of the state function 1/r(r, 0), determine 1/r(r, t) .
We will formulate the solution to the problem for a time-independent Hamilto-
nian. The more general case is given as an exercise (Problem 3.18).
First we rewrite (3.45) in the form
a if!
at 1/r(r,t) + r;1/r(r, t) = 0 (3.64)

Next, we multiply this equation (from the left) by the integrating factor 0- 1
A_I
U =exp
(itf!)
T (3.65)

which is the inverse of

U =exp
(itf!)
-T (3.66)
3.5 Solution to the Initial-Value Problem in Quantum Mechanics 85

This function of the operator, H, is itself an operator. It is defined in terms of its

Taylor series expansion.

U
A_I
=exp -
itHA) itHA 1 (A)2
=1+-+- -
itH + .. . (3.67)
( h h 2! h

More generally for any operator A , the function operator f(A) is defined in terms
of a series in powers of A. A few examples are provided in the problems .
Let us return to the problem under discussion . Multiplying the time-dependent
Schrodinger equation through by the integrating factor (3.65), one obtains the
equation

a [(itH)
at exp h 1{t(r, t) ] = 0 (3.68)

Integrating over the time interval (0, t) gives

itH)
exp
( h 1{t(r, t) -1{t(r, 0) = 0 (3.69)

Multiplying this equation through by (; gives the desired result:

t) = itH)
1{t(r, exp
( - h 1{t(r,O) = U1{t(r, 0) (3.70)
A

Here we have used the fact that

U -I = exp (-itH)
UAA -h- exp (itH)
h = IA (3.71)

where j is the identity operator.

Suppose that in solution (3.70) we choose the initial state to be an eigenstate
of k Call it qJn, so that
qJn (r)
1{t(r,O) = (3.72)
HqJn = EnqJn
By virtue of the theorem presented in Problem 3.16,

1{tn(r, t) - itH) exp (-iEnt)

= exp
( -h- qJn = -h- qJn = e-rit» t qJn(r)
n

(3.73)

This is the solution of the time-dependent Schrodinger equation, derived in Sec-

tion 3.4 by the technique of separation of variables. The solution given in (3.70)
86 Chapter 3 The Postulates of Quantum Mechanics

is more general. It exhibits the development of an arbitrary initial state 1/r(r, 0) in

time. It will be used extensively in the chapters to follow, where the student will
gain a more workable understanding of the equation .
As a final topic of discussion in this chapter we note the following. Suppose
that a system is in an eigenstate of the Hamiltonian at t = 0, described by (3.72).
At this (initial) time the expectation of an observable A is

(A)r=O = f 1/r*(r, O)A1/r(r, 0) dr = f f{!n *A({Jn dr (3.74)

What is (A) at a later time, t > O? The state of the system at t > 0 is given by
(3.73):

(3.75)

so that at t > 0 (assuming that aA/at = 0),

(A)t = f 1/r*(r, t)A1/r(r, t) dr = e+iwnle-iwnt f f{!n *Af{!n dr

= f f{!n * A({Jn dr = (A)t=o (3.76)

I (A)t >o = (A)t=o in a stationary state

The expectation of any observable is constant in time, if at any instant in time the
system is in an eigenstate of the Hamiltonian . For this reason eigenstates of the
Hamiltonian are called stationary states.

a stationary state (3.77)

In the first three sections of this chapter we encountered functions relevant to a

system which are eigenfunctions of operators corresponding to observable prop-
erties of that same system. In what sense are these eigenfunctions related to the
state function of the system? From postulate II we know that ideal measurement
of A leaves the system in the eigenstate of A corresponding to the value of A that
was found in measurement. Thus, the state function of the system immediately
after measurement is this same eigenstate of A. The state function then evolves in
time according to (3.70).

PROBLEMS

3.16 Let the eigenfunctions and eigenvalues of an operator A be {f{!n} and {an} , respec-
tively, so that
Problems 87

Let the function f(x) have the expansion

00
l
f(x) = Lblx
1=0

Show that IfIn is an eigenfunction of f(A) with eigenvalue f(a n) . That is,

3.17 If P is the momentum operator in the x direction, and f(x) is an arbitrary "well-
behaved" function, show that

. ~ ') f(x) =
exp ( I': f(x +~)

The constant ~ represents a small displacement. In this problem the student must
demonstrate that the left-hand side of the equation above is the Taylor series expan-
sion of the right-hand side about ~ = o.

3.18 If His an explicit function of time, show that the solution to the initial-value problem
(by direct differentiation) is

1/J(r, t) = exp [-~ L dt' H(t')J 1/J(r, 0)

You may assume that H(t)H(t') = H(t')H(t) .

3.19 What is the effect of operating on an arbitrary function f(x) with the following two
operators?
(a) 01 == (a 2Iax 2 ) - 1 + sin2(a3Iax 3 ) + cos 2(a3 lax 3 )
(b) 02 == cos(2alax) + 2sin 2(alax) + f: dx
3.20 (a) The time-dependent Schrodinger equation is of the form

a a1/J = H1/J
at
Consider that a is an unspecified constant. Show that this equation has the fol-
lowing property. Let H be the Hamiltonian of a system composed of two inde-
pendent parts, so that

and let the stationary states of system 1 be 1/JI (XI , t) and those of system 2 be
1/J2 (X2 , t). Then the stationary states of the composite system are

That is, show that this product form is a solution to the preceding equation for
the given composite Hamiltonian.
Such a system might be two beads that are invisible to each other and move
on the same straight wire. The coordinate of bead I is xI and the coordinate of
bead 2 is x2.
88 Chapter 3 The Postulates of Quantum Mechanics

(b) Show that this property is not obeyed by a wave equation that is second order in
time, such as

(c) Arguing from the Born postulate, show that the wavefunction for a system com-
posed of two independent components must be in the preceding product form,
thereby disqualifying the wave equation in part (b) as a valid equation of motion
for the wavefunction 1/1.

Answer (partial)
(c) If the two components are independent of each other, the joint probability density
describing the state of the system is given by

This, in turn, guarantees that the probability density associated with compo-
nent I,

is independent of the form of P2(X2) (and vice versa). The product form for Pl2
is guaranteed by the product structure for the wavefunction 1/1 (Xl , X2).

3.21 It is established in Problem 3.20 that for the joint probability for two independent
systems to be consistently described by the time-dependent Schrodinger equation,
this equation must be of the form

a1/1
a-=H1/I
A

at
where a is some number. Show that for this equation to imply wave motion, a must
be complex . You may assume that H has only real eigenvalues.

Answer
Following development of the general solution (3.70), we find that the given equation
implies the solution

1/I(r, t) = (t:)
exp 1/I(r, 0)

Since H has only real eigenvalues, the time dependence of 1/I(r, r) is nonoscillating.
It modulates 1/I(r, 0) in time and does not give propagation. Thus, is a is real, 1/1 can-
not represent a propagating wave. (Note: The fact that a is complex implies that 1/1 is
complex . These last two problems illustrate the necessity of complex wavefunctions
in quantum mechanics.)

3.22 Consider the wavefunction

1/1 = Aei(kx+wt)
Problems 89

where k is real and (J) > 0 and is real. Is this wavefunction an admissible quantum
state for a free particle? Justify your answer. If your answer is no, in what manner
would you change the given function to describe a free particle moving in the -x
direction?
3.23 (a) A free particle of mass m moves in one-dimensional space in the interval 0 :s x,
with energy E . There is a rigid wall at x = O. Write down a time-independent
wavefunction, ip(x) , which satisfies these conditions , in terms of x and k, where
k is the wave vector of the motion. State the relation between k and E for this
wavefunction .
(b) Show explicitly that the wavefunction you have found in part (a) is an eigenfunc -
tion of the Hamiltonian for this system.
(c) What is the time-dependent state, ip(x, z), corresponding to the wavefunction,
ip(x) ?
CHAPTER

4 Preparatory Concepts. Function

Spaces and Hermitian Operators

4.1 Particle in a Box and Further Remarks on Normalization

4.2 The Bohr Correspondence Principle
4.3 Dirac Notation
4.4 Hilbert Space
4.5 Hermitian Operators
4.6 Properties of Hermitian Operators

In this and the follow ing two chapters, we continue development ofphysical prin-
ciples and mathematical groundwork important to quantum mechanical descrip-
tions. Included in the present chapter are the notions of Hilbert space and Her-
mitian operators. First, we obtain wavefun ctions relevant to a particle in a one-
dimensional box. These, together with previously derived free-particle wavefunc-
tions, then serve as simply understood references for subsequent descriptions of
Hilbert space and Hermitian operators.

4.1 • PARTICLE IN A BOX AND FURTHER REMARKS ON NORMALIZATION

In Chapter 3 we solved the quantum mechanical free-particle problem . We recall

that the free-particle Hamiltonian generates a continuous spectrum of eigenvalue s,
!i 2k 2 / 2m , and eigenfun ctions, C{Jk = A exp(ikx), as given in (3.55).
The second one-dimensional problem we wish to treat is that of a point mass
m , constrained to move on an infinitely thin, frictionless wire that is strung tightly
between two impenetrable walls a distance a apart (see Fig. 4.1). The correspond-
ing potential has the values

Vex) = 00 (x ::;O, x:::a) (domain 1)

(4.1)
Vex) = 0 (0 < x < a) (domain 2)
and is depicted in Fig. 4.2. This configuration is known as the one-dimensional
box. I
1A mathematically more accurate description of the one-dimensional box would characterize it as an
infinitesimally thin, flat sheet of infinite extent and finite mass which moves between two walls of
infinite extent. The two walls and sheet are all parallel and the velocity of the sheet is normal to the
walls. Every point in space is then characterized by one coordinate, the normal displacement of the
sheet from either of the walls.

90
4.1 Particle in a Box and Further Remarks on Normalization 91

Impenetrable walls

FIGURE 4.1 One-dimensional "box."

The Hamiltonian for this problem is the following operator:

AZ
i/i=L+oo=
2m
oo (x :5 0, x 2: a) (domain 1) (4.2)

AZ
Hz= L2m (0 < x < a) (domain 2) (4.3)

In domain 1 the time-independent Schrodinger equation gives ip = 0. For any

finite eigenenergy E, in this domain the time-independent Schrodinger equation
read s

(4.4)

Since tp and E are finite, the right-hand side is finite. Therefore, the left-hand side
is finite and qJ must vanish in this domain.

Domain I Domain 2 Domain 1

o a x

FIGURE 4.2 Potential corresponding to the one-dimensional box.

92 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

The fact that tp = 0 in domain 1 implies that there is zero probability that
the particle is found there (lqJl 2 = 0). This is in agreement with the discussion
in Chapter 1 on "forbidden domains." These, we recall, are domains E < V .
Certainly, this is the case in domain 1 for any finite energy E.
In domain 2 the time-independent Schrodinger equation is

Ii? a2
- 2m ax2 qJn = EnqJn (4.5)

The subscript n is in anticipation of a discrete spectrum of energies En and eigen-

functions qJn .
Since qJn is a continuous function , it must have the values

qJn(O) = qJn(a) = 0 (4.6)

First we rewrite (4.5) in the form

a2qJn 2
ax 2 + kn qJn = 0 (4.7)

2 2m En
kn =---yj,2 (4.8)

This is merely a change of variables from energy En to wavenumber k n. The

solution to (4.7) appears as

qJn = A sin knx + B cos knx (4.9)

The boundary conditions (4.6) give

B=O (4.10)
A sinkna = 0 (4.11)

The second of these equations serves to determine the eigenvalues k n .

n=0,1 ,2, . .. (4.12)

This is seen to be equivalent to the requirement that an integral number of half-

wavelengths, n (A/2), fit into the width a.
The spectrum of eigenvalues and eigenfunctions is discrete. To find the con-
stant A in (4.11), we normalize qJn.

(4.13)

The dummy variable () = nnx ]a.

 4.1 Particle in a Box and Further Remarks on Normalization 93

If follows that the eigenenergies En and normalized eigenfunctions qJn for the
one-dimensional box problem are

(4.14)

qJn =
. (mrx)
[i SIn
V-;; -a- (4.15)

The eigenstate corresponding to n = 0 is tp = O. This, together with the solution

in domain I, gives qJ = 0 over the whole x axis. There is zero probability of
finding the particle anywhere . This is equivalent to the statement that the particle
does not exist in the n = 0 state. Another argument that disallows the n = 0

t: ({!3

x x

({J2

£2 = 4£) 1 - - - - - - 1
x x

.,~,
£)1------1
o a x o a x

(a) (b) (c)

FIGURE4.3 (a) Eigenenergies for the one-dimensional box problem. (b) Eigenstates for
the one-dimensional box problem:

CPn = V(2~ sin (mrx)

-a-

(c) Probabiliy densities for the one-dimensional box problem:

r; = ICPn I2 = 2
~ sin
2 (n7TX)
---;;-
94 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

state follows from the uncertainty principle. The energy corresponding to n = 0

is E = O. Since the energy in domain 2 is entirely kinetic, this, in tum, implies
that the particle is in a state of absolute rest (D.p = 0), an illegitimate state of
affairs for a particle constrained to move in a finite domain.
The eigenenergies and eigenfunctions given by (4.14) and (4.15) , together with
the corresponding probability densities l<Pn 12 , are sketched in Fig . 4.3.

The Arbitrary Phase Factor

In concluding this section we note the following important fact. As described
in Section 3.3, the wavefunction lfr gives information about a system through
calculation of averages of observable properties of that system , according to the
rule

(C) = f lfr*Clfr dx

This equation, as well as the normalization condition

f lfr*lfr dx = 1

are invariant under the transformation lfr ---+ eia lfr , where a is any real number.
That is, a wavefunction is determined only to within a constant pha se factor of
the form eia . Although associated with all wavefunctions, this arbitrary property
has no effect upon any physical results .f

PROBLEMS

4.1 What are the energy eigenfunctions and eigenvalues for the one-dimensional box prob-
lem described above if the end s of the box are at -a/2 and +a /2 ? [Check your answer
with (6.100) .]

4.2 For what values of the real angle 8 will the constant C = i (ei O - I) have no effect in
calculations involving the modulu s ICo/l?

4.2 • THE BOHR CORRESPONDENCE PRINCIPLE

Let us now consider the classical motion of a particle in a one-dimensional box .

As described previously, this configuration is effected by a bead sliding with no
friction on a taut wire strung between two impenetrable walls a distance a apart.
If the particle is given a velocity v, its motions (between collisions with the wall)

20n the other hand, component phase factors for a composite wavefunction such as that discussed in
Section 2.5 do contribute to measurable effects, such as interference .
 4.2 The Bohr Correspondence Principle 95

is

x = Xo + vt
Now suppose that the initial position Xo is completely unknown . What is the prob-
ability P dx of finding the particle in the interval x , x + dx ; at a subsequent time?
The answer is, the fraction of time d t f T it spends in this interval. .

P d x = dt = v dt = dx (4.16)
T a a

so that

1
P = - = constant (4.17)
a

It is uniformly probable to find the particle at any position on the wire. If we

make a large number of replicas of this one-dimensional system, measurement (at
random times) of the coordinate x of the bead will find all values (0 :s x :s a)
occurring equally often (Fig. 4.4).
On the other hand, in the quantum mechanical case, if the particle is in the state
CP3, say, the probability density P is peaked at x = (a I6 , a 12, 5a I6) ; see Fig. 4.3.
In this case, measurement on an abundant number of replica systems finds the
particle spending much of its time in the neighborhood of these three values of x.
This situation is quite different from the classical case described above. Suppose
we move to higher quantum states. At what values of x is the probability density
P peaked ? The solution is left to Problem 4.3, where one obtains that ICPn 12 is

P(x) P"(x)

aI r - - - - - - - - - - - ,

a x a x
(a) (b)

FIGURE 4.4 (a) Classical probability density for the one-dimensional box. (b) Quantum
mechanical probability density

for the one-dimensional box problem, for the case n » 1. The probability Pn vanishes
n + 1 times in the interval (0, a) .
96 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

peaked at the values

2j + 1
xj=--a, j = 0,1,2, .. . (4.18)
2n
As n becomes very large, the probability density oscillates with so large a fre-
quency that it begins to assume a uniform quality. For any n, one can divide the
interval (0, a) into n strips of equal width Sx such that in each strip the proba-
bility lC'fJn 21!:i.x of finding the particle is equal. For the classical case the number
of such strips is infinite. In the quantum mechanical case, the same situation is
approached in the limit n -+ 00.
One encounters this transition to classical physics from the quantum mechan-
ical domain in many problems. Bohr was the first to analyze this transition and
offered the general rule that a quantum mechanical result must reduce to its clas-
sical counterpart in the classical domain. Since classical formulas do not contain
n
ti, such a transition should be realized in the limit that becomes small. For many
problems this limit is attained in passing to high quantum numbers (n -+ 00).
This rule is called the Bohr correspondence principle.
Classical physics includes the dynamics of macroscopic bodies. An aggregate
of particles (e.g., a gas) obeys classical laws when the de Broglie wavelength,
A, of a typical particle is small compared to all relevant lengths. For example,
if the density of particles (number/em') is n, the gas obeys classical statistics if
A « n- 1/ 3 (the mean distance between particles is n- 1/ 3) . In the classical limit,
fluctuations about the average become small and the probabilities indigenous to
quantum mechanics reduce to certainties.
A rule of thumb in this area is that any quantum mechanical result that does not
n
contain is in essence a classical result. The first (fortuitous) example of this rule
was Rutherford's classical calculation of the Coulomb cross section, relevant to
the scattering of charged particles. The correct quantum mechanical calculation of
this parameter is found not to contain h. Rutherford's classical calculation yields
the same result.
More examples of the correspondence principle will arise in the course of de-
velopment of the text. Coulomb scattering is further described in Section 14.4.

PROBLEMS

4.3 For the one-dimensional box problem, show that P = l'Pn 12 is maximum at the values
x = Xj given by

2j +1 j =0,1,2, .. . ,n-l
xj=~a ,
 Problems 97

4.3 • DIRAC NOTATION

In this section we introduce a notation that proves to be an invaluable tool in

calculation, called the Dirac notation. It gives a monogram to the integral of the
product of two state functions, lfr(x) and qJ(x), which appears as

(4.19)

In Dirac notation , the integral on the right is written in the form shown on the left.
More generally, the integral operation (lfr IqJ} denotes: (1) take the complex
conjugate of the object in the first slot (lfr ---+ lfr*) and then, (2) integrate the
product (lfr*qJ) . This operation has the following simple properties. If a is any
complex number and the functions lfr and tp are such that

(4.20)

the following rules hold:

(lfrlaqJ} = a (lfr IqJ} (4.21)

(alfrlqJ} = a*(lfrlqJ} (4.22)
(lfrlqJ}* = (qJllfr} (4.23)

(qJ + lfr I = (lfr I + (qJl (4.24)

f (lfrl + lfr2)*(qJI + qJ2) dx = (lfrl + lfr2!qJ, + qJ2} (4.25)

= «(lfr, I + (lfr2IHlqJ,} + IqJ2})

= (lfrilqJI} + (lfrllqJ2} + (lfr2IqJI} + (lfr2IqJ2}

The object (lfr I (called a "bra vector") joins in a product form with a ("ket vector")
IqJ} , to form the "bra-ket," (lfrlqJ}.
A more fundamental description of Dirac notation in quantum mechanics is
given in Appendix A.

PROBLEMS

4.4 Write the following equations for the state vectors f, g, and so on, in Dirac notation.
(a) f(x) = g(x)
(b) c = J g*(x')h(x')dx'
98 Chapter 4 Preparatory Concepts. Function Spaces and Herm itian Operators

(c) f(x) =L
n
rpn(x) f rpn *(x')f(x') dx'

(d) 0 = v(x) f dx'rp*(x')

(e) ~
ax f(x) = h(x) f h*(x ')g (x ') dx'

4.5 Consider the operator 0 = Irp} (v i and the arbitrary state function f(x) . Describe the
following forms.
(a) (flO
(b) Ol!)
(c) (f101!)
(d) (fIOlv)

Answer (partial)
•
(a) (flO is the bra vector C(vl, where the constant C = (flrp) = foo
-00 f*rpdx .

4.4 • HILBERT SPACE

In this section we introduce the concept of a space of functions . Specifically we

will deal with a Hilbert space . This serves the purpose of giving a geometrical
quality to some of the abstract concepts of quantum mechanics.
We recall that in Cartesian 3-space a vector V is a set of three numbers, called
components (Vx , Vy , Vz ). Any vector in this space can be expanded in terms of
the three unit vectors ex, e y , ez (Fig. 4.5). Under such conditions one terms the
triad ex, ey , ez , a basis.

(4.26)

The vectors ex, e y , ez are said to span the vector space.

The inner ("dot") product of two vectors (U and V) in the space is defined as

(4.27)

The length of the vector V is Jv . V.

A Hilbert space is much the same type of object. Its elements are functions
instead of three-dimensional vectors. The similarity is so close that the functions
are sometimes called vectors. A Hilbert space 5) has the following properties:

1. The space is linear. A function space is linear under the following two con-
ditions : (a) If a is a constant and tp is any element of the space, then acp is
also an element of the space. (b) If tp and 1fr are any two elements of the
space, then tp + 1fr is also an element of the space.
2. There is an inner product, (1frlcp), for any two elements in the space. For
function s defined in the interval a :s x :s b (in one dimension), we may
4.4 Hilbert Space 99

FIGURE 4.5 Vector V in Cartesian 3-space and its components (Vx • Vy • Vz ). The or-
thogonal triad (ex. ey• ez) spans the space.

take

(4.28)

3. Any element of 5) has a norm ("length") that is related to the inner product
as follows:

(4.29)

4. 5) is complete. Every Cauchy sequence of functions in 5) converges to an

element of 5). A Cauchy sequence {CPn} is such that IICPn - CPIII ~ 0 as nand
I approach infinity. (See Problem 4.24 .) Loosely speaking, a Hilbert space
contains all its limit points.

An example of a Hilbert space is given by the set of functions defined on the

interval (0 ::: x ::: a) with finite norm

(4.30)

Another example is the space of functions commonly referred to by mathemati-

cians as "L 2 space." This is the set of square-integrable functions defined on the
whole x interval.
100 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

Sh (4.31)

Let us see how the preceding concept of inner product (4.28) is similar to the
definition of the inner product between two finite-dimensional vectors (4.27). To
see this we interpret the function ({l(x) as a vector with infinitely many compo-
nents. These components are the values that ({l assumes at each distinct value of
its independent variable x . Just as the inner product between U and V is a sum
over the products of parallel components, so is the inner product between tp and 1ft
a sum over parallel components. This sum is nothing but the integral of the prod-
uct of tp and 1ft. The reason we complex-conjugate the first "vector" is to ensure
that the "length" (square root of the inner product between a "vector" tp and itself)
of a vector ({l is real.
Thus we see that Hilbert space is closely akin to a vector space. Mathe-
maticians ' call it that-an infinite-dimensional vector space (also, a complete,
normed, linear vector space.) Elements of this space have length and one can
form an inner product between any two elements. The vector quality of Hilbert
space can be pushed a bit further. We recall that if two vectors U and V in
three-dimensional vector space are orthogonal to each other, their inner product
vanishes . In a similar vein, two vectors in Hilbert space, tp and 1ft, are said to be
orthogonal if

(1ftI({l) = 0 (4.32)

Furthermore, we recall that the three unit vectors ex, ey, and e, "span" 3-space.
Similarly, there is a set of vectors that "spans" Hilbert space. For instance, the
Hilbert space whose elements all have the property given by (4.30) is spanned by
the .sequence of functions {({In}, which are the eigenfunctions of the Hamiltonian
relevant to the one-dimensional box problem (4.15). This means that any function
tp in this Hilbert space may be expanded in a series of the sequence {({In} .

00

((l(x) = I:>n({ln(X) (4.33)

n=1

The geometrical interpretation of this relation is depicted in Fig. 4.6. The coef-
ficient an is the projection of tp onto the vector ({In. To see this, first we state a fact
to be illustrated in the next section. The basis vectors {({In} comprise an orthogonal
set. That is,

(n =1= n') (4.34)

3A more mathematically accurate presentation of function spaces may be found in C. Goffman and
G. Pedrick, First Course in Functional Analysis, Prentice-Hall, Englewood Cliffs, N.J., 1965. Another
book in this area, but more directly related to quantum mechanics, is T. F. Jordan , Linear Operators
for Quantum Mechanics, Wiley, New York, 1969.
4.4 Hilbert Space 101

Projection of '"
on to "'3 = < "'31", >

"'8------.,,;

Projection of
'" onto "'\ = <"'\ I", >

FIGURE 4.6 Projection of rp onto an orthonormal set of eigenfunctions in Hilbert space.

Furthermore, Cf!n is a unit vector ; that is, it has unit "length"

(4.35)

These latter two statements may be combined into the single equation

(4.36)

The symbol on,n' is called the Kronecker delta and is defined by

On.n' = 0 for n i= n' , On ,n' =1 for n = n' (4.37)

Any sequence of functions that obeys (4.36) is called an orthonormal set.

To show that an is the projection of Cf! and Cf!n, we first rewrite (4.33) in Dirac
notation .

(4.38)

Then we multiply from the left by lCf!n'} and use the relation (4.36) .

(Cf!n,ICf!) = L (Cf!n' lanCf!n)

n

= Lan (Cf!n'lCf!n) = Lanon,n' = an' (4.39)

n n
102 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

The coefficient an' is the inner product between the basis vector CPn' and the vec-
tor cp o Since CPn' is a "unit" vector, an' is the projection of cP onto CPn' (Fig. 4.6).
The student should recognize (4.33) to be a discrete Fourier series representation
of tp; in terms of the trigonometric sequence (4.15).

Delta-Function Orthogonality
We will continue with the use of the labels Sj 1 and Sj2 to denote the two Hilbert
spaces defined by (4.30) and (4.31), respectively. As stated previously, the se-
quence {CPn} given by (4.15) "spans" Sjl. The sequence {CPn} is a basis of Sj I. What
are the vectors that span Sj2? The answer is, the eigenfunctions of the momentum
operator p,

(4.40)

Let us see if this (continuous) set of functions is an orthogonal set. Toward these
ends we form the inner product

(CPklcpk') = - 1
2rr
1 00

-00
e ix(k' - k) dx = 8(k , - k) (4.41)

It follows that the inner product between any two distinct eigenvectors of the
operator p vanishes.
Any function in Sj2 may be expanded in terms of the eigenvectors {CPk}. Since
this sequence is a continuous set, the expansion is not a discrete sum as in (4.33),
but an integral. If cp(x) is any element of Sj2, then since {cpd spans this space, one
may write

cp(x) = i: b(k)cpk(X) dk (4.42)

This is the Fourier integral representation of cp(x). Again, the coefficient of ex-
pansion b(k) is the projection of cp(x) onto CPk . To exhibit this fact, we first rewrite
the last integral in the form

(4.43)

Again, if this equation is compared to (4.38), we see how the sum over discrete
an values is replaced by an integration over the continuum of b(k) values. If we
now multiply (4.43) from the left with (cpk ,l, there results

(cpk'lcp) = i: dk(cpk'lb(k)CPk) = i: dkb(k)(cpk'lcpk) (4.44)

= i: dkb(k)8(k' - k) = b(k')
Problems 103

The coefficient of expansion b(k') is the inner product between Cf!k' and tp; hence
it may be termed a projection of tp onto Cf!k' , But Cf!k' does not appear to be a "unit"
vector. Indeed, the vector Cf!k is infinitely long.

(4.45)

Although this disqualifies the set {Cf!d for membership in Sh, they nevertheless
span the space. They comprise a valid set of basis vectors and the projection of
any function in Sh onto any member of the basis {Cf!k} gives a finite result. If tp is
any function in 5)2, then

(4.46)

The functions {Cf!k} may, through proper renormalization, be cast in a form which
allows them to be members of 5)2. (See Problem 4.6.)

PROBLEMS

4.6 Consider the functions

f(Jk
1 l'kx
= _e
.;a
defined over the interval (-a I2 , +a I2) .
(a) Show that these functions are all normalized to unity and maintain this normal-
ization in the limit a --+ 00.
(b) Show that these functions are an orthogonal set in the limit a --+ 00.
4.7 State to which space each of the functions listed belong, n) or n2.
(a) !I = (x S-x 4-ax4+ax 3) /(x-2a)
(b) h = (sinx)e- X2
(c) h = Jln[x(x - a) + 1]
(d) 14 = sin 2mr[x(x - a) + 1], n = 0, 1,2, . . .
(e) Is = e iax (x 2 + a 2 )- 1
(0 16 = x lOe- x2
(g) h = II sin kx

4.8 The function

9 (x) = x(x - a)e ikx

is in n) . Calculate the coefficients of expansion, an, of this function , in the series

representation (4.33), in terms of the constants a and k. Use the basis functions (4.15).
4.9 Two vectors 1/1 and f(J in Hilbert space are orthogonal. Show that their lengths obey
the Pythagorean theorem,
104 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

4.10 Consider a free particle moving in one dimension. The state functions for this particle
are all elements of n2. Show that the expectation of the momentum (Px) vanishes in
any state that is purely real (1/1* = 1/1) . Does this property hold for (H)? Does it hold
for (x)?

4.5 • HERMITIAN OPERATORS

The average of an observable A for a system in the state 1ft (x , t) is given by (3.32).
In Dirac notation this equation appears as (in one dimension)

(A ) = f 1ft*(x, t)A1ft(x, t) dx = (1ftIA1ft) (4.47)

Since t is a fixed parameter in this equation, we may conclude that the formula
gives the expectation of A at the time t . Now one may ask: What are the possible
state functions for a particle moving in one dimension at a given instant of time?
The answer is: any function in 5)2. For example, the particle could be in any of
the following states at some specified time:

Ce ikx
1ft2 = - - , (4.48)
X

where B , C , and D are normalization constants . Again consider the observable A.

If the average of this observable is calculated in any of these states (that is, any
member of 5)2), the result must be a real number. This is a property that we de-
mand an operator have if it is to qualify as the operator corresponding to a phys-
ical observable. The object (1ft IA 1ft) must be real for all 1ft in 5)2. When working
with the one-dimensional box problem , (1ft IA 1ft ) must be real for all 1ft in 5)1. For
example, if iI is the operator corresponding to energy, then

(E) = (1ftIH1ft) = -
r
10
1ft*li2
2m
a2
ax21ft dx (4.49)

must be real for any state function 1ft in 5) 1.

These observations give rise to the following rule: In quantum mechanics one
requires that the eigenvalues of an operator corresponding to a physical observable
be real numbers . In this section we discuss the class of operators that have this
property. They are called Hermitian operators and are a cornerstone in the theory
of quantum mechanics.

The Hermitian Adjoint

To understand what a Hermitian operator is, we must first understand what the
Hermitian adjoint of an operator is. Consider the operator A. The Hermitian ad-
joint of A is written At. Under most circumstances, it is an entirely different
operator from A. For instance, the Hermitian adjoint of the complex number c is
4.5 Hermitian Operators 105

the complex conj ugate of c. That is,

(4.50)

How is the Hermitian adjoint defined ? First , let us agree that an operator is defined
over a specific Hilbert space, Sj. Also if A is the operator and Vr is any element of
Sj, then AVr is also in Sj. For any two elements of this space, say Vrl and Vrn , we
can form the inner product

(4.51)

Suppose there is another operator, At, also defined over Sj, for which
(4.52)

Suppose further that this equality holds for all ur; and Vrn in Sj. Then At is called
the Hermitian adjoint of A. To find the Hermitian adjoint of an operator A, we
have to find the object At that fits (4.52) for allVrl and Vrn . Consider A = a, a
complex number. Then

(4.53)

Equating the first and the last terms , we see that at = a" , As a second example,
consider the operator

D=-
a (4.54)
ax

defined in Sj2. then

(VrLlDVrn)
A

=
/00 dxwi" ~Vrn
a (4.55)
-00 uX

= * +00 -
[Vrl VrnLoo -00 dx (aax Vrl *) Vrn
/00
= (- DVrLlVrn)
The "surface" term is zero since Vrl and Vrn are elements of Sj2. Thus we find

Dt =-D (4.56)

For some cases we will find that the Hermitian adjoint of an operator is the oper-
ator itself. For such an operator A, we may write

(4.57)
106 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

In terms of the defining equation (4.52) , this implies that for all 1/11 and 1/In in S)
(over which A is defined) ,

(4.58)

Operators that have this property are called Hermitian operators. The simplest
example of a Hermitian operator is any real number a, since

(4.59)

If A and B are two Hermitian operators, is the produ ct operator AB Hermitian?

This is most simply answered with the aid of Problem 4.11(b), according to which

(4.60)

If A and B are Hermitian, then

(4.61)

and AB is not (necessarily) Hermitian. What about AB + BA?

(AB + BA)t = BtA t + AtBt = BA + AB (4.62)
=AB+BA
It follows that if A and B are both Hermitian, so is the bilinear form (AB + BA).
Is the square of a Hermitian operator Hermitian?

(4.63)

The answer is yes. Another way of doing this problem is as follows . Look at the
inner product.

(4.64)

The first equality follows because A1/1n is in S) and A is Hermitian, while the
second equality follows simply because A is Hermitian. Comparing the first and
third terms shows that A2 is Hermitian.

The Momentum and Energy Operators

Let us test the momentum operator p and see if it is Hermitian. For the free-
particle case, p is Hermitian if for all 1/11 and 1/In in S)2,

(4.65)

Developing the left-hand side, we have

Problems 107

This technique is, by and large, the principal method by which a specific operator
is shown to be Hermitian.
Having shown that p is Hermitian, it follows that the free-particle Hamiltonian,
fI , is Hermitian.
~2
~ P
H=- (4.67)
2m
P~ 2 )t P~ 2
Ht
~
= (2m = 2m = H
~
(4.68)

[Recall (4.63) .] For a particle in a potential field V (x),

~2
~ P
H = - + V(x) (4.69)
2m
Since V (x) is a real function that merely multiplies (say in Sh), it is Hermitian.

(l/rIIVVrn) = i: VrI*VVrn dx = i: VVrl *Vrn dx (4.70)

= !(VVrI)*Vrn dx = (VVrIIVrn)

It follows that fI as given by (4.69) is Hermitian.

PROBLEMS

4.11 (a) Show that (aA + bB) t = a* At + b" Bt.

(b) Show that (A B) t = Bt At.
(c) What is the Hermitian adjoint of the real number a?
(d) What is the Hermitian adjoint of b 2 ? [See (4.54).)
(e) What is the Hermitian adjoint of (AB - BA)?
(I) What is the Hermitian adjoint of (AB + BA)?
(g) What is the Hermitian adjoint of i(AB - BA)?
(h) What is (A t)t?
(i) What is (At A) h

4.12 If Aand Bare both Hermitian, which of the following three operators are Hermitian ?
(a) i (AB - BA)
1U8 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

(b) (AB-BA)
(c) (AB;BA)
(d) If Ais not Hermitian, is the product At AHermitian?
(e) If Acorresponds to the observable A, and B corresponds to B, what is a "good"
(i.e., Hermitian) operator that corre sponds to the physically observable product
AB?
4.13 If A is Hermitian, show that

Answer (in ShY

(A2 ) = L: 1/f*A 21/fdx = L:

(A1/f)*A1/fdx

= L: IA1/r1 2dx ~0

4.14 If Ais Hermitian, show that (A) is real ; that is, show that (A) * = (A) .
x
4.15 For a particle moving in one dimen sion, show that the operator p is not Hermitian.
Construct an operator which corresponds to this physically observable product that
is Hermitian.

4.6 • PROPERTIES OF HERMITIAN OPERATORS

The first property of Hermitian operators we wish to establish is that their eigen-
values are real. Let A be a Hermitian operator. Let {«Pn} and {an} represent, re-
spectively, the eigenfunctions and eigenvalues of the operator A.
A \

A«pn = an«Pn (4.71)

In Dirac notation

(4.72)

Multiplying from the left with (<<Pn I gives

(«PnIA<<Pn) = («Pnlan<<Pn) = an(<<Pnl«Pn) (4.73)

Since A is Hermitian, we can write the left-hand side as

(4.74)

Equating the last terms in the latter two equations gives

Problems 109

(4.75)

and an is real.
The second property of Hermitian operators we wish to establish is that their
eigenfunctionsare orthogonal. Again consider (4.72). Now multiply from the left
with another eigenvector of A, (C{J/I . There results

(4.76)

Since Ais Hermitian, the left-hand side of this equation can be rewritten
(4.77)

The eigenvalue a/ is real because it is an eigenvalue of a Hermitian operator

(i.e., A). Subtracting the two equations above gives

(4.78)

If a/ =1= an (which is the case if a/ and an are nondegenerate: See Section 5.3), this
equation says that

(4.79)

which is the expression of the orthogonality of the set of functions {C{Jn}. If these
functions are all normalized, then (4.79) may be generalized to read

(4.80)

Thus, the eigenvalues of a Hermitian operator are real, and its eigenfunctions are
orthogonal.

\
PROBLEMS

4.16 Show that if an operator iJ has an eigenvalue hi f= hi *, then iJ is not Hermitian .

4.17 Consider the operator C,

Ccp(x) = cp*(x)

(a) Is C Hermitian?
(b) What are the eigenfunctions of C?
(c) What are the eigenvalues of C?

4.18 Given that the operator 0 annihilates the ket vector 1f), that is, 01 f) = 0, what is
the value of the bra vector (f 10 t ? Interpret the meaning of your answer.

4.19 The parallelogram law of geometry states that the sum of the squares of the diagonals
of a parallelogram equals twice the sum of the squares of the sides . Show that this is
110 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

also true in Hilbert space ; that is, if 1/f and I{J are any two elements of a Hilbert space,
then

4.20 Show that the standard properties of cos 8, together with the definition of the inner
product between two vectors I{J and 1/f, in fl, with respective lengths, III{J II and 1I1/f II,
imply the Cauchy-Schwartz inequality

4.21 Use the Cauchy-Schwartz inequality to prove the triangle inequality

4.22 Construct the squared length of (1/f -I{J) to show that

4.23 Let the sequence {l{Jn} be an orthonormal basis in fl . Let the sequence {cos8 n} rep-
resent the angles between the vectors {l{Jn} and an arbitrary element 1/f in fl . Using
Bessel's inequality,
00

L I(I{Jn 11/f)12 ::: lI1/ff

n=1

show that

Under what circumstances does the equality hold ?

4.24 Every convergent sequence is also a Cauchy sequence. A sequence (l{Jn(x)} is a

Cauchy sequence if

}l~ IIl{Jn - I{JIII =0

1-' 00

A function space fl is a complete space if every Cauchy sequence in fl converges

to an element of fl . This is a requirement that a function space must satisfy in order
that it be termed a Hilbert space . [See property 4 after (4.27) .] Show that the space
of functions on the unit interval with the property I{J(O) = 1{J(1) = 0 is not a Hilbert
space.
4.25 In addition to a complete space, one also defines a complete sequence. An orthonor-
mal sequence {l{Jn} is complete in fl if there is no vector 1/f, in fl of nonzero length
(1I1/f II > 0), which is perpendicular to all the elements in the sequence {l{Jn} . Show
that if {l{Jn} is an orthonormal basis of fl , it is complete in fl.

Answer
Let {l{Jn} be an orthonormal basis of fl. Let 1/f be an element of fl with nonzero length.
which is normal to all the elements of {l{Jn} .1f {l{Jn} is a basis, then we may expand 1/f.
Problems 111

But 1fr is normal to alll{Jn. Therefore, (I{Jnl1fr) = 0, which gives 1fr = 0, so the
hypothesis leads to a contradiction, hence the hypothesis is an incorrect statement
and there is no such 1fr in Sj.
4.26 Show that any operator A may be expressed as the linear combination of a Hermitian
and an anti-Hermitian (Bt = -B) operator.

Answer

_
A= (A+At)
- - - +1. ( -
A--At
-)
2 2i

[Note: A+ At and i (A - At) are both Hennitian.]

4.27 Show that the wavefunctions for a particle in a one-dimensional box with walls at
x = 0 and a satisfy the equality

The subscript x denotes differentiation.

4.28 Use the equality proved in Problem 4.27 to establish the following variational
principle . If the expectation f 1fr* H1fr dx is minimum, the normalized wavefunc-
tion 1fr is the ground state. Specifically, establish the theorem for a particle in a
one-dimensional box, assuming real wavefunctions.

Answer
Apart from a constant factor and with the results of Problem 4.27, we may write

Let 1fr minimize (H) . Then infinitesimal variation of 1fr causes no change in (H ).
Let 1fr -+ 1fr + 81fr . The variation 81fr is an arbitrary infinitesimal function of x that
vanishes at x = 0 and a. Then

(H) =f 1frx 2 dx -+ f (1frx + 81frx)2 dx = (H) + 8(H)

8(H) = f 2 1frx 81frx dx =2 f 1frx ~ 81fr dx =0

Integrating the last term by parts and dropping the "surface" terms gives

f 1frxx81fr dx =0
Variation of the normalization statement (both 1fr and 1fr + 81fr are normalized) gives

). f 1fr81frdx =0
112 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

whence 'A is an arbitrary undetermined multiplier. Combining the last two equations
yields

If this equation is to be satisfied for an arbitrary variation of 1{! about the minimizing
value, we may conclude

1{!xx = 'A1{!
where 1{! is an eigenstate of if. It follows that, in any eigenstate, (H ) is stationary
and is minimum in the ground state.

4.29 Let

Show that

Answer

Ani = (If>nIAIf>I)
= (Alf>il lf>n)* = (If>il At If>n )*
Taking the complex conjugate of the last and first terms in this equality gives the
desired result.

4.30 Employing Hermitian propertie s of H, show that in general, extreme values of (H)
yield the eigenfunctions of H , and that any eigenfunction of H makes (H) an ex-
tremum . Your derivation should be independent of specific boundary conditions.

Answer
To incorporate wavefunction normalization in the analysi s we write

(H) = (1{! IH 1{!)

(1{! I1{!)

Taking the variation of this form we obtain

«(1{!I1{! )2 8(H ) = - (1{!IH1{!)[(81{!I1{! ) + (1{! 181{! )]

+ (1{! I1{!) [(81{! IH 1{!) + (1{! IH 81{!) ] + O[(81{!)2]
Neglecting terms of O[(81{!)2], and recalling the Hermiticity of H, the preceding
equation becomes

«(1{!I1{! )2 8(H ) = - (1{!IH1{!)[(81{!11{! ) + (1{! 181{!)]

+ (1{! I1{!)[(81{! IH 1{!) + (H 1{! 181{!) ]
Problems 113

With (1{! IH 1{!) = A, and (1{! I1{!) 1, it follows that any function 1{! which make s
o(H ) = 0, to first order in o1{!, gives

0= 2 Re (01{!I (-A + H)1{!)

=
If we label (o1{! l( -A + H)1{!) z, where z is a complex number, then the preceding
statement remains valid if o1{! is multiplied by z, in which case the preceding gives

(01{! 1( -A + H)1{!) =0
If this relation is to be satisfied for any arbitrary variation, o1{!, it must be the case
that H1{! = A1{! . Furthermore, it is clear from the equation above for o(H ), that any
eigenfunction of H makes (H) an extremum.

4.31 (a) An electron propagates through a periodic potential with period a. What is the
maximum energy, Ee , for which electron dynamics is quantum mechanical?
(b) Repeat the preceding problem for the case of a proton. Call your answer E p '
(c) What can you conclude regarding classical versu s quantum behavior for these
two particles from the ratio E pi E e ? Call electron and proton masses m and M,
respectively.

Answer (partial)
(c) The motion is quantum mechanical if A 2: a, where A is the de Broglie wave-
length. There results E pi Ee = .Jml M « 1. Thus, there is a wide rage of
energies for which electron motion is quantum-mechanical, but proton motion is
classical.

4.32 Our definition of Hermitian adjoint (4.52) is that (At rpn Irpm ) = (rpn IArpm) is satisfied
for all rpn and rpm in the relevant Hilbert space. Show that an equivalent definition is
that (rpn IAt rpm ) = (Arpn Irpm ) for all rpn and rpm in the same Hilbert space.

4.33 (a) A particle in a one-dimensional box is in the superposition state

rp(x) = bleiA1rpl (x) + b3ei A3rp3 (x )

where AI and A3 are arbitrary real pha se factors, bl and b3 are real coefficients,
and

What is the functional dependence of the probability density, rp* tp, on AI , A3?
(b) Repeat the preceding problem for the superposition state

rp(x) = bleiA1rpl(X) +b3eiA3rp3(X) + b7eiA7IfJ7 (x)

where

Answers
(a) The probability density is a function of IAI - A31 .
(b) The probability density is a function of IAI - A71, IAI - A31 , IA7 - A31 .
114 Chapter 4 Preparatory Concepts. Function Spaces and Hermitian Operators

4.34 Given that the inverses A-I and ir 1 are known:

(a) What is the inverse of S = Ain
(b) Does your answer to part (a) depend on the Henniticity of A or in
(c) Employ your answer to part (a) to establish that if A-I A = i then ,,1,,1 -1 = 1.
(An operator A and its inverse, A-I, satisfy the relation A-I A = i. For further
discussion see Section 11.2.)
4.35 An electron in a one-dimensional box with walls at x = (0, a) is in the quantum state

1/f(x) = A, 0 < x < al2

1/f(x) = -A , al2 < x < a

(a) Obtain an expression for the normalization constant, A.

(b) What is the lowest energy of the electron that will be measured in this state?
4.36 (a) Is the following wavefunction an energy eigenstate for a free particle moving in
one dimension ?

1/f(x) = Ae 'ik: x + - A e- Iikx

~
(b) Is this wavefunction an eigenstate of the one-dimensional momentum operator ?
(c) If your answer to part (b) is no, what values of momentum will be found in
measurement and with what probabilities will these values occur?
CHAPTER

5 Superposition and
Compatible Observables

5.1 The Superposition Principle

5.2 Commutator Relations in Quantum Mechanics
5.3 More on the Commutator Theorem
5.4 Commutator Relations and the Uncertainty Principle
5.5 "Complete" Sets of Commuting Observables

In this chapter we encounter the superposition principle. which is considered by

many to be one of the more fundamental concepts of quantum mechanics. This
principle represents one of the basic differences between classical and quantum
mechanics and also provides a deeper understanding ofthe uncertainty principle.
Closely related to the superposition principle are the commutator theorem and
the notions of compatible observables and simultaneous eigenfunctions.

5.1 • THE SUPERPOSITION PRINCIPLE

Ensemble Average
Consider again a particle in a one-dimensional box. Let us imagine a large number
of identical replicas of the system (called an ensemble in statistical mechanics),
such as described in Section 3.3. If each such box is in the same initial state
1/1 (x, 0), after an interval of time t, each box will again be in a common state
1/I(x, t), as shown in Fig. 5.1. Suppose that we ask what the energy of the particle
is in each box, at the time t. The laws of nature are such that the energy measured
in each of the identical boxes, which are all in the identically same state 1/1 (x , t),
are not the same [save for the case that 1/1 (x , 0) is an eigenstate of If].
How does one answer the question above: What will the energy be? Since the
energy measured at the time t in each box of the ensemble will most likely not be
the same, more appropriate questions are: (1) What is the average of the energies
measured in all the boxes of the ensemble? (2) If we measure the energy in one
box, with what probability will the value, say, E3 be found ? To answer these
questions, we first recall that if the probability of finding the value En in a given
measurement of energy is P(En ) , then the average energy over measurements of
all members to the ensemble in the limit as this number become large is given by

115 -
116 Chapter 5 Superposition and Compatible Observables

First member Second member Third member Nth member

of members
of ensemble
.. .. ~ ..
Initial state

0 a x 0 a x 0 a x 0 a x

..o
.. ..
a x o a x o a x o a x

Energy measured
at time t :

.. ..
State
of system
at time
t + e-

O a x o a x

FIGURE 5.1 Measurementof energy of N identical one-dimensionalboxes which com-

prise an "ensemble." All boxes are in the same state at t = O.

the expression

(E) = L P(En)En (5.1)

all En

[Recall (3.34)] . This formula holds for all physical observables. For example, the
average particle position is given by

(x) = la xP(x) dx (5.2)

In this case the integral is a sum over the continuum of values x may assume.
The quantum mechanical prescription for calculating the average of a dynami-
cal observable in the state 1fr is given by the third postulate of quantum mechanics
5.1 The Superposition Principle 117

[Section 3.3, (3.32)]. Specifically, for the energy we have (in Dirac notation)

(5.3)

Let us expand the state 1/! in the eigenstates of H. These eigenstates obey the
equation

(5.4)

For the box problem they are explicitly (4.15)

ifJn =
. (nnx)
(2 Sill
y-;; -a- (5.5)

The expansion of 1/! in these eigenstates appears as

00

1/!(x, t) = Lbn(t)ifJn(X) (5.6)

n=!

The state 1/! is that of the system at the time t , so that it is, in general, a function
of x and t . Since ifJn is a function of x only the coefficients of expansion bn may,
in general, be functions of time.
In Dirac notation, (5.6) appears as
00

I1/!} = L IbnifJn} (5.7)

n=!

Substituting this series into (5.3) gives

(E ) = (~bnifJnl H ~b/ifJ)
= LLbn*b/(ifJnIHifJ/}
n /

= L Lbn*b/E/(ifJnlifJ/}
n /

= L L bn*b/E/8n/
n /

(5.8)

Equating this average to that given by (5.1) gives

2E (5.9)
L Ibnl n= L P(En)En
n n
118 Chapter 5 Superposition and Compatible Observables

This equation dictates the following interpretation of the square of the modulus
of bn . It is the probability that at the time t, measurement of the energy of the
particle which is in the state 1{r(x , t) yields the value En.

(5.10)

These coefficients have the correct normalization, provided that the states 1{r and
CPn are normalized. In this case we have

1 = (1{rI1{r) = (~bncpnl I;b/CP/)

= L L bn*b/(CPnICP/)
n /

(5.11)

When this is the case the coefficient Ibn 12 is an absolute probability. If not, the
correct expression for the probability that measurement finds En is

P E _
( n) -
21C 2
Ibnl nl _ Ibnl n
21C
~lbnl2lCnl2 - (1{r I1{r)
e (5.12)

where

Let us return to the expansion (5.7). The coefficients bn are calculated in the
following manner. Multiply this equation from the left with the bra vector (CPn' I.
Owing to the orthonormality of the set {CPn}, one obtains
(5.13)

The coefficient bn is the projection of 1{r onto the eigenvector CPn. The physical
interpretation of bn is that Ibn 12 is the probability that measuring E finds the value
En when the system is in the state 1{r. This prescription is true for any dynamical
observable. Consider the symbolic operator ft
(5.14)

At a given time t , the system is in the state 1{r(x, t) . What is the probability that
measurement of F at this time finds the value h? The state 1{r is a superposi-
tion state. It is composed of the/eigenstates of ft. Here we are assuming that the
eigenstates of ft are a basis for the Hilbert space that 1{r is in. So we may write

1{r = LbnCPn (5.15)

bn = (CPnl1{r ) (5.16)
 5.1 The Superposition Principle 119

1\
F operator
State
I/J
correspond ing
to observable
1\
F 'Pn f n'Pn • t; 'Pn
F

PUn) probability
When f n is a continou s set,
then :
--.. bI/Jn =Lb<n'P'Pnn II/J that measurement
of F givesfn bn 2

f
I/J L b; 'Pn -> b(n) 'P(n) dn
b(n) 2dn P(j) dn probability
thatflies in intervalf(n),J(n dn)

FIGURE 5.2 Elements of the superposition principle.

This assumption that an arbitrary state 1/f may be represented as a superposition

of the eigenstates of a physical observable is the essence of the superposition
principle. With {lpn} and 1/f normalized to unity, the probabil ity that measurement
finds the value f3 is Ib31 2. This procedure is depicted in Fig . 5.2.

Hilbert-Space Interpretation
When we look in Hilbert space, {lpn} is one set of vectors and 1/f is another vector.
The system is in the state 1/f. Measurement of F causes the state 1/f to fall to one
of the Ipn vectors. Ch ances are that it goes to the Ipn vector to which it is most
inclined (in the geometrical sen se; see Fig. 5.3) .
Consider the following illustrative example. A particle of mass m is in a one -
dimensional box of width a. At t = 0 the particle is in the state

(5.17)

The Ipn functions are the orthonormal eigenstates of iI:

Ipn = V{2~ sm. (nnx)

-a- (5.18)

What will measurements of E yield at t = 0 and what is the probability of finding

this value? First let us see if 1/f is normalized. In Dirac notation we have , for the
120 Chapter 5 Superposition and Compatible Observables

(3) (b)

FIGURE 5.3 (a) State of the system before measurement at t, superimposed on the basis
{lI1n}, which are the eigenvectors of the operator F. The probability that measurement of
F finds the value In is proportional to the projection of 1/1 on II1n . (b) State of the system
immediately after measurement has found the value II . Measurement acts as a "wave
filter." It filters out all components of the superposition 1/1 (x , t) = 'E bn (t)lI1n (x), passing
only the 1111 wave.

state (5.17) ,

(5.19)

so that
1
(y,Iy,) = 25 {(3(qJ21 + 4(qJ91)(3IqJ2) + 4IqJ9))}

1
= 25 {9(qJ2IqJ2) + 12(qJ2IqJ9) + 12(qJ91qJ2) + 16(qJ91qJ9)}

= 1 (5.20)

and y, is normalized, The inner products (qJ2lqJ2) = (qJ9lqJ9) = 1 while the other
two are zero, owing to the orthogonality of the set {qJn}.
The superposition principle stipulates the following. If we want the probability
that measurement finds the value En, we must expand y, in the eigenstates of iI .
The square of the magnitude of the coefficient of qJn is the said probability.

• t, _ '" b __
3qJ_2---;+=4::-,-qJ_9
'I' - L...J nqJn - J25 (5.21)

In this simplified problem, by inspection we find that

5.1 The Superposition Principle 121

4
b9=--
J25
bn =0 (n =1= 2 or 9) (5.22)

Therefore, the probability P(En ) that measurement of Eat t = 0 finds the value
En is

9
P(E2) = 25

16
P(E9) = 25

P(E n ) = 0 (n =1= 2 or 9) (5.23)

In an ensemble of 2500 identical one-dimensional boxes, each containing an

identical particle in the same state 1jJ(x, 0) given by (5.17), measurement of Eat
t = 0 finds about 900 particles to have energy E2 = 4E1 and about 1600 particles
to have energy E9 = 81E1.
Is there a chance that, in an ensemble of 1017 boxes, measurement of E finds
E2 in all 1017 boxes? Yes. This remarkable response carries the philosophical
impact of the superposition principle. Although the state 1/J (x , 0) is a precise su-
perposition of well-defined eigenstates of the observables being measured, one
is not certain what measurement will yield. There is nothing in classical physics
that is similar to this concept. Any uncertainty in classical physics arises from
uncertain initial data. In quantum mechanics, although the initial state 1/J (x, 0)
is prescribed with perfect accuracy, one is never certain in which eigenstate, CPn,
measurement will leave the system.
However, once E is measured and, say, the value E9 is found, then one knows
with absolute certainty that the state of the system immediately after this mea-
surement is CfJ9 .

The Initial Square Wave

As a second illustrative example, we consider the following free-particle problem
in one dimension . Suppose that at t = 0 the system is in the state (Fig. 5.4)

1/J(x ,O)
Iff o
-
a
[x] <-
2
a

elsewhere
(5.24)

If at this same instant, the momentum of the particle is measured, what are the
possible values that will be found, and with what probability will these values
occur?
To answer these questions we must first expand 1/J (x , 0) in a superposition of
the eigenstates of p:
122 Chapter 5 Superposition and Compatible Observables

'" (x, 0) b(k) Ib(k)12

-a12 al2 x k
I-I 21t/a 41t/a
!J.k =41t/a
"""" k

FIGURE 5.4 Square wave packet at t = 0 and corresponding momentum eigenstate

amplitudes b(k) . The interval over which momentum values are most likely to be found is
Sp = n!J.k = 4Jrn/a .

1 e ik x
IfJk = __ ' (5.25)
5
Since these states comprise a continuum, the corresponding superposition of
eigenstates of p is an integral.

1{r(x, 0) = i: b(k)lfJk dk (5.26)

Inverting this equation (see 4.42 et seq.) gives the coefficient b(k) .

b(k) =
100

-00
1{r(x ,O)lfJk*dx= I
v2rr
{;C
1
00

-00
1{r(x , O)e- l kx dx
°

= -1- . : e- l kx dx = - -
I -
2 ° (e ika 2
/ - e- ika 2
/ )
J2rra -a /2 J2rra k u
= (2 sin(kaj2) (5.27)
V~ k

Again, this coefficient is the projection of the state 1{r(x, 0) onto the eigenstate IfJk .
Its square times the differential dk is the probability that measurement of momen-
tum yields p = Iik, in the interval hk, n(k + dk) . The corresponding probability
density (in momentum space) is

2 2 sin2(kaj2)
IbI = -rr-a --k--::2,------ (5.28)

This function has its maximum at k = O. It drops to zero at

ka
- =rr (5.29)
2
5.1 The Superposition Principle 123

or equivalently at

2rrn
p=!ik=- (5.30)
a
It is most probable that measurement of momentum finds the value p = O. The
momentum values (±n2rrnla) with n an integer greater than 0 are never found,
for at these values, b(k) = O.
Referring to Fig. 5.4, we see that the interval of momentum values that mea-
surements are most likely to uncover has the approximate width
4rr
flk=-
a
(5.31)
4rrn
flp=nflk=-
a
On the other hand, from (5.24) , it is uniformly probable that measurement of x
finds the particle anywhere in the interval (-aI2 , +a I2) , of width

flx =a (5.32)

Combining these two latter uncertainties (5.31 and 5.32) gives

(5.33)

The approximation sign is used because of the qualitative manner in which Sp

was calculated. The result (5.33) is another example of the Heisenberg uncertainty
principle at work.

The Chopped Beam

To further exhibit the significance of the probability density Ib(k) 12 , we consider
the following problem . Suppose that the free-particle system above is composed
of N noninteracting electrons. Every electron is in the state 1fr(x, 0) given by
(5.24). The density p (number/length) is related to 1fr through

Number of particles in dx = p dx = N 11fr 12 dx (5.34)

The total number in the whole "beam" is

N =
100
-00
p(x)dx =N .:
-a /2
11fr1
2dx
=N (5.35)

Suppose that we now ask how many electrons have momentum in the interval
(-2rr nla , +2rrnla) , or equivalently, how many have wavenumber in the interval
(- 2rr I a , 2rr I a) . For a single electron, the probability of finding an electron with
momentum in the interval hk: to hk: + hdk is

P(k) dk = Ib(k)1 2 dk (5.36)

124 Chapter 5 Superposition and Compatible Observables

This is a correct statement provided that

i: 2
Ib(k)1 dk = 1 (5.37)

If this is not the case, one must divide Ib(k)1 2 in (5.36) by the last integral.
For a totality of N electrons in the beam , the number of them that have mo-
mentum in the interval hk, hk + lidk is

p(k) dk = Nlb(k)1 2 dk (5.38)

i: i:
The total number in the whole beam is

2 (5.39)
N = p(k) dk = N Ib(k)1 dk

For the example at hand

00 Ib(k)12 dk = ~ /00 Sin2(~a/2) dk

/ -00 Jra-00 k
= .!. /00 sin ~ dTJ = 1
2
(5.40)
Jr -00 TJ

The dummy variable TJ == ka /2. To return to the original question, the number of
electrons t.N in the beam with momentum in the interval (-2Jr Ii/a, +2JrIi/a) is
given by the integral

+2tr/a 2 sin2(ka/2)
t.N = N - 2 dk
/ -2tr/a Jra k
2
N /+tr sin TJ
= - - 2 - dTJ = 0.903N (5.41)
it -tr TJ

Thus, we find a majority of the electrons in this momentum interval.

Superposition and Uncertainty

Let us return to the case of a single electron in the state 1/!(x, 0) given by (5.24).
Suppose at this time , t = 0, we measure the electron's momentum. What value
do we find? The answer is (a) the values p = ±n2JrIi/a are never found ; (b)
any other value may occur with corresponding probability density Ib(k) 12 . Let the
measurement find the electron to have the momentum
Jrli
p=- (5.42)
a

Immediately after this measurement, what is the state of the particle? The answer
is
5.1 The Superposition Principle 125

1
1fr = --exp (iJrX)
- (5.43)
./iii a

The electron is now in the state (5.43). Suppose that we measure the energy of
the particle . What value is found? Since this state is also an eigenstate of if, it is
a certainty that measurement yields

(rr iii a)2

E = -'----...:........:- (5.44)
2m
The system is still left in the eigenstate (5.43). Suppose that we now measure the
position of the particle. What values may occur? The probability density is

P = 11fr12 = _1 (5.45)
2Jr

which is a constant. It is uniformly probable to find the electron anywhere along

the whole x axis. The uncertainty in x is b.x = 00 . For this same state it is certain
that measurement of momentum finds the value n iii a, so that Sp = O. Again we
find corroborating evidence for the Heisenberg uncertainty principle.
Now we place a uniform array of scintillation detectors along the x axis. One
of them scintillates at x = x' . What is the state of the electron immediately after
measurement? The answer is the eigenstate of the position operator corresponding
to the eigenvalue x ' (Fig. 5.5) .

1fr = 8(x - x') (5.46)

Now we measure momentum again. What values can be found ? To answer this
question, we again call on the superposition recipe : expand 1fr in the eigenstates
of p.

8(x - x ') = 1
M:: 1
...;2Jr
00

-00
b(k)e i kx dk

b(k) = -- f
1 8(x - x')e- 1'k: x dx = __
1 e- 'k x
1 I
(5.47)
./iii ./iii
The corresponding momentum probability density is

P(k) = Ib(k)1 =
2
2~ (5.48)

It is uniformly probable to find the electron with any momentum along the whole
k axis. The uncertainty in momentum is Sp =
00 for the state (5.46), for which
b.x = 0, and the uncertainty principle holds firm (Fig. 5.6).
We have been using the phrase superposition principle , but have not given a
concise statement of this principle. P.A.M. Dirac, one of the early investigators
of quantum mechanics, was first to grasp the full significance of this principle. His
 1/1
I
= l) (x - x ')

-- x·

FIGURE 5.5 Measuring x destroys the momentum eigenstate I{!ko'

Problems 127

l'Pk (x)12

!ko k

(a)
1{2rr

1b(k) 12 1'11 12

~
1{2rr

k x . x
(b)

FIGURE 5.6 (a) In the state 1{f = I{Jko = (l/5)e i kOx , ~p = 0 and Sx = 00. (b) In
the state 1{f = 8(x - x') , ~x = 0 and Sp = 00.

description 1 is perhaps the most succinct. The superposition principle "requires

us to assume that between. .. states there exist peculiar relationships such that
whenever the system is definitely in one state we can consider it as being partly in
each of two or more other states. The original state must be regarded as the result
of a kind of superposition of the two or more new states, in a way that cannot be
conceived on classical ideas."
The superposition principle is a cornerstone of quantum mechanics. We have
used it previously in some elementary one-dimensional problems. We will return
to it in the remainder of the text in relation to more extensive one-dimensional
problems as well as more practical problems in two and three dimensions. A
sound understanding of this principle is prerequisite to a working knowledge of
quantum mechanics .

PROBLEMS

5.1 If an arbitrary initial state function for a particle in a one-dimensional box is ex-
panded in the discrete series of eigenstates of the Hamiltonian relevant to the box

1P.A.M. Dirac , The Principles of Quantum Mechanics, 4th ed ., Oxford University Press, New York,
1958.
128 Chapter 5 Superposition and Compatible Observables

configuration, one obtains (5.6)

00
1fr(x, 0) =L bn (O)qJn (x)
n=1

On the other hand, if the particle is free, its Hamiltonian has a continuous spectrum
of eigenenergies and the superposition of an arbitrary initial state in the eigenstates
qJk of iI becomes an integral (5.26):

1fr(x,O) = L: b(k)qJk dk

(a) What are the dimensions of Ibn 12 and Ib(k)1 2 , respectively?

(b) What is the source of the difference to dimensionality ?
(c) What are the dimension s and physical interpretation of the integral

L: 2
Ib(k)1 dk ?

Answer (partial)
(b) The term Ibn12 represents a probability, whereas Ib(k)1 2 represents a probability
density.
5.2 One thousand neutrons are in a one-dimensional box, with walls at x = 0, x = a. At
t = 0, the state of each particle is

1fr(x,O) = Ax(x - a)

(a) Normalize 1fr and find the value of the constant A.

(b) How many particles are in the interval (0, a/2) at t = O?
(c) How many particles have energy E5 at t = O?
(d) What is (E) at t = O?
5.3 Using the expressions for qJk and 1fr given by (5.25) and (5.26), respectively, show
that

(1frI1fr) = 1 ---+ L: 2
Ib(k)1 dk =1
5.4 A pulse 1 m long contains 1000 IX particles. At t = 0, each IX particle is in the state
1 '' ko X
_e [x] :s 50 ern, ko = rr/50
1fr(x,O)= ~O '
1 elsewhere

(a) At t = 0, how many IX particles have momentum in the interval (0 < hk < /iko)?
(b) At which values of momentum will IX particles not be found at t = O?
(c) Describe an experiment to "prepare" such a state.
(d) Construct ~x and ~p for this state, formally. What is ~x~p? [Hint: To calculate
Sp, use Ib(k)1 2 .]
Problems 129

5.5 At t 0 it is known that of 1000 neutrons in a one-dimensional box of width

10-5 em, 100 have energy 4E\. and 900 have energy 225E\ .
(a) Construct a state function that has these properties. (Coefficients may be com-
plex.)
(b) Use the state you have constructed to calculate the density p(x) of neutrons per
unit length. [Note that p(x) is a real function .]
(c) How many neutrons are in the left half of the "box"?
5.6 Over a very long interval of the x axis, a uniform distribution of 10,000 electrons is
moving to the right with velocity 108 cm/s and 10,000 electrons are moving to the left
with velocity 108 cm/s. Assuming that the electrons do not interact with one another,
construct a state function that yields the preceding properties for the combined beam.
Calculate (p) for this state.
5.7 Give an argument in support of the conjecture that one cannot measure the momen-
tum of a particle in a one-dimensional box with absolute accuracy. Support the theo-
retical argument with an argument involving an experiment.
5.8 A one-dimensional box containing an electron suffers an infinitesimal perturbation
and emits a photon of frequency

hv = 3E\
where E\ denotes the ground state of the particle. A student concludes that the elec-
tron was in the state rp2 prior to perturbation. Is the student correct?
Answer
What the student has in mind is that the photon corresponds to the decay

hv = E2 - E\ = 3E\
However, suppose that the electron was in the superposition state (3rp2 + 8r(6)/.Jf3.
Then it is still possible that a photon of frequency hv = 3E\ is emitted . So the
student is incorrect.
5.9 Measurement of the position of a particle in a one-dimensional box with walls at
x = 0 and x = a finds the value x = a12.
(a) Show that in the subsequent measurement , it is equally probable to find the par-
ticle in any odd-energy eigenstate .
(b) Show that the probability of finding the particle in any even eigenstate is zero.
(An eigenstate rpn is even if n is even and odd if n is odd.)
5.10 It is known that at time t = 0, a particle in a box (described in Problem 5.9) is not
in the right half of the box. The particle is in one of an infinite number of states. Six
such states are depicted in Fig. 5.7.
(a) Write down an approximate wavefunction for each of these states.
(b) Calculate (E) for each of these states.
(c) Argue that the state depicted in Fig. 5.7a is the state of minimum (E) (assuming
thatrp = A sin Zn x]«, x < aI2) .
5.11 A particle in the one-dimensional box described in Problem 5.9 is in the ground
state. One of the walls of the box is moved to the position x = 2a, in a time short
130 Chapter 5 Superposition and Compatible Observables

q> (x , 0) q> (x , 0) q> (x , 0)

x x x
o al2 a o a/2 a o al2 a
(a) (b) (c)

q> (x , 0) q> (x , 0) (x, 0)

x x x
o al2 a o a/2 a o a/2 a
(d) (e) (f)

FIGURE 5.7 Six initial states for a particle in a one-dimensional box, with the property
that 11fr1 2 = 0 in the right half of the box. (See Problem 5.10.)

compared to the natural period 2Jl'/ WI, where liwl = E I. If the energy of the particle
is measured soon after this expansion, what value of energy is most likely to be
found? How does this energy compare to the particle's initial energy (EI)?

5.2 • COMMUTATOR RELATIONS IN QUANTUM MECHANICS

An important operation in quantum mechanics is the commutator between two

operators, A and B. It is written [A , B] and is defined as

(5.49)

An immediate property of the commutator is that

[A, B] = -[B, A] (5.50)

If
(5.51)

the two operators are said to commute (A and B are compatible) with each other.
That is,

AB = BA (5.52)
5.2 Commutator Relations in Quantum Mechanics 131

Any operator A commutes with any constant a.

[A, a] = 0 (5.53)
[A ,aB] = [aA, B] = a[A, B] (5.54)

Any operator A commutes with its own square, ,12.

[A, ,12] = (,1,12 - ,12 A) = (,1,1,1 - ,1,1,1) =0 (5.55)

The meaning of this relation is that, no matter what A is, when [A , ,12] operates
on any function 9 (x) , one gets zero,

(5.56)

More generally, A commutes with any function of A, f(A) .

[f(A) , A] = 0 (5.57)

As an example of this rule, consider the following commutator involving the mo-
mentum operator p.

[eP, p] = [f
n=O
p:, p]
n.
,,1 ~n
= L..J ,[p , p]
n.
~

= [I, p] + [p, p] + ;![p2, p] + ... = 0 (5.58)

It follows that

. ~
[eP , p]g(x) = [exp(-a;
ilia) '-a;
iMJ] g(x) = 0 (5.59)

where 9 (x) represents any function of x.

One of the most important commutators in physics is that between the coordi-
nate, X, and the momentum, p. Let us calculate it.

[x~ , p]g(x)
~ (a
= in - x -
ax
a) g(x)
+ -x
ax

. ( - xag
= In - +xag- + 9) .
= lng(x) (5.60)
ax ax

It follows that

I [x,p]=in I (5.61)
132 Chapter 5 Superposition and Compatible Observables

In other words, the operator [x, p] has the sole effect of a simple multiplication
by the constant iii. As an immediate consequence (using Problem 5.12)

[x, p2] = [x , p]p + p[x, p]

= 2ilip (5.62)

so that
ag
A pA2] 9 (x)
[ x, = 2"'2
It- (5.63)
ax
In a similar vein,

[x 2, Pl = x[x, Pl + [x, p] x
= 2i/iX = 2i/ix (5.64)

The operator [x 2 , Pl multiplies by 2i lix,

We now prove an important theorem in quantum mechanics which is related to
the commutator between two operators . It states: if Aand B commute

(5.65)

then A and B have a set of nontrivial (i.e., other than a constant) common eigen-
functions. The proof is as follows.
Let <{Ja be the eigenfunctions of A that corresponds to the eigenvalue a.

(5.66)

Then

(5.67)

Since A and B commute, the left-hand side of this last equation may be rewritten

(5.68)

Inspection of this equation reveals that B<{Ja is also an eigenfunction of A corre-

sponding to the eigenvalue a. If <{Ja is the only linear independent (defined below)
eigenfunction of A that corresponds to the eigenvalue a, the function B<{Ja can
differ from <{Ja by, at most, a multiplicative constant J.L . That is,

(5.69)

(B<{Ja and J.L<{Ja are in the same direction in Hilbert space: see Fig. 5.8) But this is
the eigenvalue equation for the operator B. It follows that <{Ja is also an eigenfunc-
tion of B.
5.2 Commutator Relations in Quantum Mechanics 133

FIGURE 5.8 If BlfJa is an eigenvector of A corresponding to the eigenvalue a, BlfJa and

lfJa are in the same direction in Hilbert space 5),

We have already encountered the implication of this theorem for the problem
of the free particle moving in one dimension. For this case

(5.70)

It follows by the theorem above that p and iI have common eigenfunctions. They
do . We recall that

(5.71)
li,2k 2 'k x
He'ik x = --e'
A

2m
Before pursuing the case when rpa is not the only linearly independent eigenfunc-
tion of Acorresponding to the eigenvalue a, we consider the definition of linearly
independent functions.

Linearly Independent Functions

When is a set of functions a linearly independent set? The N functions of the set
{rpn} are linearly independent if the linear combination

(5.72)

for all x is only satisfied when

(5.73)

For example, the two functions eX and sinx are linearly independent since

(5.74)

for all x is only satisfied by

(5.75)
134 Chapter 5 Superposit ion and Compatible Observables

The two functions eX and 3ex are not linearly independent since

(5.76)

is true for all x if

(5.77)

The concept of linearly independent functions has an interesting geometrical

interpretation in Hilbert space. If two "vectors" cp\ and CP2 in a Hilbert space 55
are linearly independent, they do not lie along the same axis (line) in 55 (Fig. 5.9).
Similarly, if the set of N vectors {CPn} is such that all members are linearly inde-
pendent, no two elements of this set lie on the same axis. If cp\ and CP2 are linearly
independent, one must "rotate" cp\ to align it with <n.
If CPa is the only linearly independent eigenfunction of A corresponding to the
eigenvalue a, all eigenfunctions of A corresponding to a must be of the form
p.,CPa . The functions CPa and p.,CPa are two linearly dependent eigenfunctions of A
corresponding to the eigenvalue a .

(5.78)

How many such vectors are there? Since p., can be any constant, there is a contin-
uum of such linearly dependent eigenfunctions of A corresponding to the eigen-
value a. In any given problem only one of these states is relevant. For bound states
(11/r1 2 ~ 0, [x] ~ 00), 1/r is fixed (and therefore p.,) by normalization. For an un-
bound state (11/r1 2 fr 0, Ixl ~ 00), 1/r is fixed through an appropriate boundary
condition. The latter case is appropriate to beam or scattering problems, where
the boundary conditions usually involve stipulations on particle current or num-
ber density at Ix I = 00 . These concepts are discussed in greater detail in Section
7.6 , which concerns one-dimensional barrier problems.

FIGURE 5.9 If {cp\, CP2 . CP3 . CP4} are a linearly independent set, no two lie along the same
axis in Hilbert space Sj.
Problems 135

PROBLEMS

5.12 If A, B, and C are three distinct operators, show that:

(a) [A + B, C] = [A, C] + [B , C]
(b) [AB , C] = A[B, C] + [A, C]B

5.13 If A and B are both Hermitian, show that AB is Hermitian if [A , B] = O.

5.14 Show that the solution to the time-dependent Schrodinger equation given in Problem
3.18, that is,

1/f(r, t) = exp [ -~ L H(t')]

dt ' 1/f(r, 0)

is correct, provided that

[H(t) , H(t')] = 0 (t =1= t')

Answer
For 1/f(r, t) as given above to be a solution, the expansion

must be valid in order to obtain the Schrodinger equation. For this to be so, eWin the
second term must precede H. Here we have set

w== LH(t')dt
and have absorbed the constant -i/ Ii into H. We must show that

_a ew=e
· w_. aw
at at
In general

-a eW = -a ( 1 + W• + -1'2
W + -1 W
'3 + ... )
at at 2 6

aw 1 ( W-at+-at
=-at+z • aw aw W.) + ...

Thus the equality above holds if we are able to set

In this case
136 Chapter 5 Superposition and Compatible Observables

In terms of the integral definition of W, the commutation criterion above becomes

H(t) It H(t')dt ' = (L H(t')dt') H(t)

which is guaranteed if [H(t) , H(t')l = O.

5.15 Discuss the linear independence of the following sets of functions .
(a) {x,3x,e X }
(b) {eix , sinx, cosx}
(c) {x 2 ,x 3 ,x 5 }
(d) {x,3sin2x,4cos2x ,lnx}

5.16 If J-i is an arbitrary constant, the two vectors ffJ and iup in fl are linearly dependent.
Show that the cosine of the angle between these two vectors has modulus I.

[cose] =I
5.17 From Problem 5.16 we conclude that ffJ and iup lie along the same axis in fl . Show
also that tup is IJ-il times longer than ip, that is, that (see Fig. 5.10)

1IJ-iffJ1I = 1J-illlffJlI

FIGURE 5.10 The vectors ffJ and J-iffJ in Hilbert space fl lie along the same axis and
1IJ-iffJ1I = 1J-illlffJll· (See Problem 5.17.)

5.18 Show that if AffJn = anffJn and BffJn = bnffJn for all eigenvalues {an} and {bn} of
A and B, respectively (i.e., A and B have completely common eigenstates), then
[A, Bl = 0 on the space offunctions spanned by the basis {ffJn} . (Hint: Any element
of this space may be written

and one need merely show that

[A, Bl L CnffJn = 0.)

 5.3 More on the Commutator Theorem 137

Note: In a more general vein one may say the following : let the eigenstates common
to A and B span a subspace g, of a Hilbert space n. Then [A , B]1/J = 0, where 1/J is
any element of g,.

5.3 • MORE ON THE COMMUTATOR THEOREM

The Concept of Degeneracy

Suppose there are two (and only two) linearly independent eigenfunctions of the
operator A which both correspond to the eigenvalue a . Call them !PI and !P2 .

A!PI = a!pI
(5.79)

Under such circumstances one says that the eigenvalue a is doubly degenerate.
The eigenfunctions !PI and !P2 are degenerate . Now we ask, what is the most gen-
eral eigenfunction of A that corresponds to the eigenvalue a? The answer is, any
function of the form

(5.80)

with ct and {3 arbitrary constants . Let us test that this is the case.

A!Pa = A(ct!pI + {3!P2) = aao; + {3a!p2

= a(ct!pI + {3!P2) (5.81)

In Hilbert space the two functions !PI and !P2 span a plane (two-dimensional
subspace) . Equation (5.80) indicates that any vector !Pa in this plane is an eigen-
function of A corresponding to the eigenvalue a (Fig. 5.11).
Le,t us re~um to the commutator theorem discussed in Section 5.2. The opera-
tors A and B commute . If we operate on the first of equations (5.79) with Band

"-
A";2 =a";2

FIGURE 5.11 If «PI and «P2 are two linearly independent degenerate eigenvectors of A,
they span "plane" (two-dimensional subspace) in Hilbert space n. Any vector in this plane
is an eigenvector of A corresponding to the eigenvalue a.
138 Chapter 5 Superposition and Compatible Observables

use the commuting property of A and B, there results

(5.82)

We conclude that BIfJ) is an eigenstate of A that corresponds to the eigenvalue a.

But there is a continuum of such eigenstates, all of the form (5.80) . All we can
say is that there are some ex and f3 such that

(5.83)

Inspection of this equation [compare with (5.69)] reveals that 1fJ) need not be an
eigenfunction of B.
SO we have the following rule: If [A , B] = 0, and a is a degenerate eigenvalue
of A, the corresponding eigenfunctions of A (which all have the same eigenvalue,
a) are not necessarily eigenfunctions of B. Loosely speaking, degenerate opera-
tors have "more" eigenstates than nondegenerate operators. This concept may be
illustrated in tenns of the Venn diagrams depicted in Fig . 5.12 .

'Po and 'Ph

(a) (b)

(c)
(d)

FIGURE 5.12 Various cases pertaining to the sets of eigenfunctions of two compatible
operators, A and E. [A, E] = o. (a) Eigenfunctions of A = all eigenfunctions of E. (b) A
has only nondegenerate eigenfunctions. (c) E has only nondegenerate eigenfunctions. (d)
[A, E] = [E, C] = [A, C] = o. A has only nondegenerate eigenfunctions.
Problems 139

A very simple physical example of this situation is provided by the problem of

the free particle moving in one direction. The eigenvalue

(5.84)

of the Hamiltonian [see (3.14)]

~2

H=L (5.85)
2m

is doubly degenerate. All of the following functions are eigenfunctions of Hcor-

responding to this eigenvalue.

{<PI, <P2, <P3} = (coskx, sinkx, exp(ikx)} (5.86)

This is not a linearly independent set. However, any two are, so that for the free
particle the eigenvalue (5.84) is doubly degenerate. For example, the two linearly
independent functions , say,

{<PI , <P2} = {coskx, sinkx} (5.87)

both have the eigenvalue n2 k 2 / 2m . Although [p , H] = 0, for the free particle,

the set of functions (5.87), being degenerate eigenfunctions of energy, need not
be eigenfunctions of p. In fact, they are not.
Another linearly independent set of degenerate eigenstates corresponding to
the eigenenergy = n2k2/ 2m is {<P2, <P3} . Of these, <P3 is an eigenstate of p and <P2
is not. Of the set {<PI, <P3}, <PI is not an eigenstate of p, and again <P3 is.
When there are n (and only n) linearly independent eigenstates of an operator
A that all correspond to the same eigenvalue, the eigenvalue is n-fold degenerate.
Suppose that [A, B] = O. What can then be said is that from these n degener-
ate eigenstates of A, one can form n linear combinations which are n linearly
independent eigenstates of both A and B.
For instance, from the two degenerate eigenstates (5.87) in the free-particle
problem above, we can form

<P+ = <PI + I.<P2 = s- Ii sm

cos kxx + . k x = e ik x (5.88)
<P_ = <PI - I.<P2 = cos k x - I' sin
. k x = e -ikx (5.89)

These two functions are common eigenstates of Hand p. They remain degenerate
eigenstates of H but are nondegenerate eigenstates of p.

PROBLEMS

5.19 Construct two linearly independent linear combinations of !P2 and !P3 given in (5.86)
which are common eigenfunctions of if and p.
140 Chapter 5 Superpos ition and Compat ible Observables

5.20 Given that i and p operate on functions in 5)2 and the relation [i , p] = i h, show that
if i = x (i.e., multiplication by x), p has the representation

p, = - I'Ii axa + f (x)

where f(x) is an arbitrary function of x . [Note: Dirac 2 has shown that through proper
choice of phase factor (Section 4.1), the arbitrary function f may always be made
to vanish. Thus, the basic commutator relations between i and p are equivalent to
explicit operator forms for these variables . Listing of such commutator relation may
serve in place of postulate I (Section 3.1).]

5.21 The operators A and 8 both have a denumerable number of eigenstates. Of these, the
single eigenstate I{! is known to be common to both. That is,

AI{! = atp, 81{! = btp

(a) What can be said about the commutability of Aand 8?
(b) Suppose that it is further known that all the eigenstates of A and 8 are degener-
ate. Does this additional information in any way change your answer to part (a)?

5.4 • COMMUTATOR RELATIONS AND THE UNCERTAINTY PRINCIPLE

As we have seen above, owing to the fact that for a free particle, p and iI com-
mute, they have a set of simultaneous eigenfunctions. Namely, any function of the
form

(5.90)

is a common eigenstate of both p and iJ. If the system (particle) is in this state,
it is certain that measurement of p gives hk and measurement of energy gives
lih 2 / 2m . Since (5.90) is a common eigenstate of p and iI, measurement of p,
which (absolutely) gives hk, leaves the particle in the state (5.90). Subsequent
measurement of E gives li 2 k 2 / 2m and also leaves the particle in the state (5.90) .3
The operators iI and p are compatible; that is, they commute. Quantum mechan-
ics allows p and E to be simultaneously specified (for a free particle). Further-
more, there is only one (unique) state which gives these two values, the state
(5.90) .

2Dirac, The Principles of Quantum Mechan ics.

3Here we mean an ideal measurement. This is a measurement which least perturbs the system . Any
real measurement causes the system to suffer a greater perturbation. After the energy of the particle in
the state (5.90) is measured , ideal measurement maintains that state. However, it is also possible that
after finding !i2 k 2 / 2m , the particle is in any linear combination of the independent degenerate energy
eigenfunctions of fI which correspond to this eigenvalue (e.g., IX cos kx + f3 = sin kx) . However,
measurement that leaves the particle in this state must have interfered with the momentum, since this
state is a superposition of momentum eigenstates . Measurement that leaves the system in the original
state (5.90) does not perturb the momentum. It is the ideal measurement of energy.
5.4 Commutator Relations and the Uncertainty Principle 141

Although there exists a state in which both energy and momentum may be
specific simultaneously, the same is not true for the observables and p. There x
is no state in which measurement is certain to yield definite values of x and p .
Measurement of p leaves the system in an eigenstate of p (5.90). Subsequent
measurement of x is infinitely uncertain. The state (5.90) is not an eigenstate of x.
Conversely, measurement of x that finds x' leaves the system in the eigenstate
ofX,

ljJ = 8(x - x') (5.91)

When the particle is in this state, measurement of momentum is infinitely uncer-

tain.
For the free particle, there are states in which the uncertainty in energy and
momentum obeys the relation

/lE/lp=O (5.92)

On the other hand, in any state, the uncertainties in observation of p and x are
such that the product /lp Sx is always greater than a fixed magnitude.

/lx/lp> -
n (5.93)
-2
It is quite clear at this point that these uncertainty relations have their origin in
the compatibility properties (5.51) of the operators that correspond to the observ-
abIes being measured.
Suppose that two observables A and i3 are not compatible:

(5.94)

For example, such is the case for displacement and kinetic energy. Then one can
show the following :" If measurement of A , in the state ljJ, is uncertain by the
amount /lA , then measurement of B is uncertain by the amount /lB, such that5

/lA/lB ~ ~I(C}I (5.95)

We recall (Section 3.3) that the uncertainty of an observable A in the state ljJ is
the root mean square of the deviation of A away from the mean (A) .

Expectation values in (5.95) are calculated in the state ljJ. For example,

(/lA)2 = ((A - (A})ljJl(A - (A})ljJ) (5.96)

4See Problem 5.42 .

5This generalization of the uncertainty principle is sometimes called the Robertson-Schrodinger re-
lation [H. P. Robertson, Phys. Rev. 35, 667A (1930) ; E. Schrodinger, Sitzungsber. Preuss. Akad. Wiss.
(1930) , p. 296] .
142 Chapter 5 Superposition and Compatible Observables

The mechani sm at work behind these uncertainty relations is as follows . If A

and B do not commute, then the eigenstate CPa of A which the system goes into on
measurement of A is not necessarily an eigenstate of B. Subsequent measurement
of B will give any of the spectrum of eigenvalues of B with a corresponding
probability distribution P(b) . This probability distribution is obtained from the
coefficients in the expansion of CPa in the eigenstates CPb of B.

(5.97)

(with {CPa} and {CPb} normalized) . Remeasurement of A is then in no way certain

of finding the system in the state CPa.
We note that the commutator-uncertainty relation , (5.94) and (5.95), is among
the more fundamental relations in quantum mechanics. In addition to its important
practical significance, it stands as an immutable barrier separating quantum and
classical physics.

PROBLEMS

5.22 How do the states for a free particle

CPI = Ae i kx
fP2 = Bcoskx

differ with regard to measurements of momentum and energy?

5.23 For a particle in a one-dimensional potential field V(x), show that

5.24 Consider three observables, A, E, and C. If it is known that

[E , Cj = iA
[A, Cj = iE
show that

5.25 Obtain uncertainty relations for the following products

(a) f!.xf!.E
(b) f!.pxf!.E
(c) f!.xf!.T
(d) f!.pxf!.T
relevant to a particle whose kinetic energy is T and whose total energy is E. (A
closely related example is discussed in Problem 2.30.)
 5.5 "Complete" Sets of Commuting Observables 143

5.26 If 9 (x) is an arbitrary function of x, show that

dg
' , 9 ] = - I·ndx
[p»
5.27 If 9 (x) and f (x) are both analytic functions, show that

9 (A)f(rp) = 9 (a)f(rp), where Arp = arp

5.28 The time-dependent Schrodinger equation permits the identification

, a
E=in-
at
Using this identification together with the rule (5.95), give a formal derivation of the
uncertainty relation

1
b..Eb..t > -n
-2
Note that in a stationary state (eigenstate of if), b..E = O. The implication for this
case is that a stationary state may last indefinitely.
5.29 Can the total energy and linear momentum of a particle moving in one dimension in
a constant potential field be measured consecutively with no uncertainty in the values
obtained?
5.30 If

and Aand 8 are both Hermitian , show that t is also Hermitian .

5.31 Prove that if A and 8 are Hermitian, [A, 8] is Hermitian if and only if [A , B] = o.
Answer (partial)
Set [A, 8] = i . Then i = -it. But i t = i , hence i = -i.
5.32 (a) Obtain an uncertainty relation for mass and time from the relativistic mass-
energy equivalency formula .
(b) A free neutron has a mean lifetime of ~ 103 s. Apply the uncertainty relation
found in part (a) to find the uncertainty in the neutron's mass.

Answer (partial)
(b) Sm ~ 10- 27 amu (M p ~ 1 amu)

5.5 • "COMPLETE" SETS OF COMMUTING OBSERVABLES

We have already seen that for the free particle in one dimension, the eigenvalues
of iI are doubly degenerate. The two eigenfunctions of iI corresponding to the
eigenvalue fi 2k 2 / 2m are exp( +ikx) and exp( -ikx) . However, once we specify
what p is (say, +fik), in addition to E, then one can say that the system is in
one and only one state, exp(+ikx) (to within a multiplicative constant). Merely
144 Chapter 5 Superposition and Compatible Observables

prescribing the energy of the particle does not uniquely determine the state of the
particle. Further specifying the momentum removes this ambiguity and the state
of the particle is uniquely determined.
Suppose that an operator A has degenerate eigenvalues. If a is one of these
values, specifying a does not uniquely determine which state the system is in. Let
B be another operator which is compatible with A. Consider all the eigenstates
{«Pab} which are common to A and B. Of the degenerate eigenstates of A, only a
subset of these are also eigenfunctions of B. Under such conditions, if we specify
the eigenvalue b and the eigenvalue a, then the state that the system can be in is a
smaller set than that determined by specification of a alone. Suppose further that
there is only one other operator C which is compatible with both A and B. Then
they all share a set of common eigenstates. Call these states «Pabc. Then

A«pabc = a«pabc
B«Pabc = b«pabc
C«Pabc = c«pabc (5.98)

These functions are still a smaller set than the set {«Pa} or {«Pab} . Indeed, let us
consider that «Pabc is uniquely determined by the values a, b, and c. This means
that having measured a , b, and c: (I) Since «Pabc is a common eigenstate of A, B,
and C, simultaneous measurement (or a succession of three immediately repeated
"ideal" mea surements) of A, B, and C will definitely find the values a, b, and c.
(2) The state «Pabc cannot be further resolved by more measurement. This state
contains a maximum of information which is permitted by th~ I~ws of quantum
mechanics. (3) There are no other operators independent of A, B , and C which
are compatible with these. If there were , the state «Pabc could be further resolved.
An exhaustive set (in the sense that there are no other independent operators com-
patible with A, B, and C) of commuting operators such as A, B, and C above,
whose common eigenstates are uniquely determined by the eigenvalues a, b, and
c and are a basis of Hilbert space, is called a complete set ofcommuting operators.

Maximally Informative States

The values , a, b, and c, which may be so specified in the state «Pabc, are sometimes
referred to as good quantum numbers. These are analogous to the generalized
coordinates whose values determine the state of a system classically. As discussed
in Section 1.1, such classical coordinates are also labeled good variables .
Suppose that there are, in all, five independent operators that specify the prop-
erties of a system: A, B, C, D and fro Of these, A, B, and C are compatible
with one another and D and fr are compatible. However, these two sets are in-
compatible with one another, so that , for example,

(5.99)
Problems 145

One can simultaneously specify either the eigenvalues a, b, and c or the eigenval-
ues d and f . One cannot, for instance, say that the system is in a state for which
measurement of A definitely gives a and measurement of D definitely gives d.
For this case there are two sets of states that are maximally informative: {!Pabcl
and {!Pdf} ,
Suppose that A has degenerate eigenvalue a. What is the state of the system
after one has measured and found a? The state lies in a subspace of Hilbert space
which is spanned by the degenerate eigenfunctions that correspond to a. This
subspace na has dimensionality N a (a is an Na-fold degenerate eigenvalue). Af-
ter measurement of B, the state of the system lies in the space nab, which is a
subspace of na and is spanned by the eigenfunctions common to A and B. This
subspace has dimensionality Nab, which is not greater than N a .

(5.100)

Subsequent measurement of C (mutually compatible with A and B) leaves the

state of the system in a space nabc that is a subspace of nab and whose dimen-
sionality does not exceed that of nab .

Nabc S Nab (5.101)

In this manner we can proceed to measure more and more mutually compatible
observables. At each step of the way the eigenstate is forced into subspaces of
lesser and lesser dimensionality, until finally after the successive measurement of
A , B, C, D, . . . the state of the system is forced into a subspace of dimensionality
N = 1. This is a space spanned by only one function . It is the eigenstate common
to the complete set of observables (A, B, C, b, ...): namely, !Pabcd. . . . This state
cannot be further resolved by additional measurements. Measurement of any of
the observables (A, B, C, D, . .. ) in this state is certain to find the respective
values (a, b, c, d , .. . ).

PROBLEMS

5.33 (a) Show that for a particle in a one-dimensional box, in an arbitrary state 1/1 (x, 0),

(b) Under what conditions does the equality maintain ?

5.34 A free particle at a given instant of time is in the state
A
1/1= -----,,--
(xkO)2 +4

At this same instant, (ideal) measurement of the energy finds that

1i2k02
E=--
2m
146 Chapter 5 Superposition and Compatible Observables

The measurement leaves the momentum uncertain . Under such circumstances, what
is the state {if of the particle immediately after measurement ?

Answer
Since the momentum is uncertain after measurement, we know that the state is not
one of the eigenstates of momentum rp±ko' Instead, one may say that the state vector
lies in a subspace of.lj spanned by the vectors cos kOx and sin kox .

{if = ex cos kOx + fJ sin kox

The coefficients ex and fJ are proportional to the projections of 1/1 on cos kOx and
sin kox , respectively. Since

(1/Ilsin kox) =0
it follows that after measurement, the state of the particle is

{if = excoskox
5.35 Show that

(Hint: Taylor-series expand /(TJ) == e71A Be- 71A about TJ = O. Also note the deriva-
tive property of 1<TJ) : dj /dTJ = [,4, fj .)
5.36 Show that [Baker-Hausdorf lemma]

e A/J = e A+BeO/Z)[A ,B)

given that A and B each commutes with [A , Bj. (Hint: First show that [e71A, Bj =
TJe 71A [A, B]. Then establish that the derivative of

is

dg " ,
dTJ = TJ[A , Bjg
and integrate .) Note that for fJ « I, one may always write

This relation is important in statistical mechanics, where fJ plays the role of inverse
temperature and A+ B is the Hamiltonian.
5.37 The operator Ahas only nondegenerate eigenvectors and eigenvalues, {rpnl and {anI.
What are the eigenvector s and eigenvalues of the inverse operator, A-I ? Is your
answer consistent with the commutator theorem?
Problems 147

5.38 (a) Construct a one-dimensional wave packet that has zero probability density out-
side a domain of length a at time t = 0 and which has average momentum
(p) = MO. That is, it is propagating to the right.
(b) The wave packet collides with a mass m. Estimate the probability that the mass
is deflected to the left with momentum MollO ± liko/lOO. Take koa = IOn.
(Assume that complete momentum exchange occurs simultaneously at t = O.
The mass is located at the origin.)
5.39 Show that the expectation of an observable A of a system that is in the superposition
state

1/J(x, t) = Lbnipneiwnf
n

may be written in the form

(A) =2LLbn*b/(nIAI1)cos(wn -w/)t+ Llbne(nIAln)

n >/ n

for Ibn*bLl and (nIAI1) real. The states ipn exp(iwnt) are eigenstates of the Hamilto-
nian of the system. Here we are writing In) for ipn.
5.40 What is the average (x) and square root of variance ~x for the following probability
densities?
(a) P(x) = A[a 4 + (x - xO)4j-1
(b) P(x) = Ax2e-x2 /2a2

(c) P(x) = A sin2 (x:;;; - 8n) exp {_ [(X~f)2]}

5.41 (a) Show that for a particle in a one-dimensional box with walls at (-aI2, a12)

~Pmin = J (p 2 )min = !!..-

2a

(b) Show for this same configuration that

~xmax = J(x 2)max = ar;:;

2 v3
(c) In which states are ~Pmin and ~xmax realized?
(d) From part (a) obtain the following momentum uncertainty relation for this con-
figuration:

h
a Sp >-
-2
5.42 Given that Aand iJ are Hermitian operators and that

show that

~A~B ::: ~I(C)I

148 Chapter 5 Superposition and Compatible Observables

Answer
The uncertainties in Aand 8, when written in terms of the operators
gA = A- (A)
gB = 8 - (B )

appear as

(~A)2 = (gA1/IIgA 1/1) = IIgA 1/111 2

(~B)2 = IIgB1/I11 2

These expressions may be incorporated into the Cauchy-Schwartz inequality

(Problem 4.20). There results

IIgA1/I11 211gB1/I11 2 ::: l(gA1/IIgB1/I)1 2

(~A)2(~B)2::: I (gA 1/1 19B 1/1) 12 = 1(1/I!gAgB1/I)1 2
The latter equality is due to the Henniticity of gA . We now recall that any operator
can be written as a linear combination of two Hermitian operators :

Here we have used the fact that [g A, gB 1 = [A, 8]. Substituting the expression above
into the preceding inequality gives

Owing to the Henniticity of Gand C, their expectation values are both real. It follows
that

5.43 The linear independence of two functions u(x) and v(x ) may be specified in terms
of their Wronskian,

W(U, v) = I:, ~,I

Thus, if u and v are solutions to a linear, second-order differential equation , and
W(u, v) :f= 0 in some interval, then u and v are independent solutions in this in-
terval. Employing this criterion , establish that the two functions given by (5.87) are
independent over the entire x axis. What value of W do you find for this case?
5.44 Determine an expression for the Bohr radius ao from the following crude approxi-
mation. The electron moves to the nucleus to lower its potential energy,

e2
V(r) = - -
r
Problems 149

If the electron is in the domain 0 :::: r :::: r, then we may write Ap ::::: Ii/r, with corre-
sponding kinetic energy 1i2 /2mr2 . With this information estimate ao by minimizing
the total energy. How does your answer compare with the actual expression for ao
given by (2.14)?
5.45 Show that in three dimensions, the coordinate-momentum commutation relation
(5.61) may be written

[r, p] = mi
In this expression , the dyadic commutator has nine components and may be written
as a 3 x 3 matrix and i is the identity operator.
5.46 Show that in three dimensions , the coordinate-momentum uncertainty relation (5.93)
may be written

where

(M)2 = ((r - (r)2)

(!:::"p2) = ((p _ (p) )2)

Answer
Working in a frame where (p) = (r) = 0, we have (!:::"r)2 = (r 2 ), (!:::"p)2 = (p2) .
The resulting nine Cartesian products separate into two groups as follows:

The first three bracketed terms give

In the second bracketed six terms, with !:::"Py ::: Ii/ Ay, etc., we write

As (z + z-l) ::: 2, for all positive z, we find

Adding both contributions gives the desired result.

5,47 (a) A particle of mass m moves in one dimension (x) . Itis known that the momentum
of the particle is Px = MO' where kO is a known constant. What is the time-
independent (unnormalized) wavefunction of this particle, v« (x)?
(b) The particle interacts with a system. After interaction it is known that the proba-
!
bility of measuring the momentum of the particle is for Px = 2MO and ~ for
150 Chapter 5 Superposition and Compatible Observables

Px = 8Mo· What is the time-independent (unnonnalized) wavefunction of the

particle in this state, o/b(X)?
(c) What is the average momentum for the particle (Px ), in the state o/b(X)?
(d) What is the particle 's average kinetic energy (T) in the state o/b(X)? Express
your answer in terms of the constant Eo = nZkOZ 12m.

5.48 Consider a situation where it is equally likely that an electron has momentum ±po.
Measurement at a given instant of time finds the value +PO. A student concludes
that the electron must have had this value of momentum prior to measurement. Is the
student correct?

Answer
The given information indicates that the electron was in a superposition state prior
to measurement. In quantum mechanics one cannot rely on the premise of inference .
The student is incorrect.
5.49 (a) Show that in one dimension , the energy spectrum of bound states is always non-
degenerate . That is, to an eigenenergy there corresponds only one linearly inde-
pendent eigenstate .
(b) In what step in your derivation does the Wronskian (Problem 5.43) come into
play?
(c) In what manner does your proof depend on the given bound-state property?
(d) What is the nature of the potential you have included in your proof?
[Hint: See Problem 10.68. (The notion of degeneracy is discussed in detail in
Chapter 8 et seq.)]

5.50 Relevant to one-dimensional motion, what is the value of the commutator

[Px , x sin(px)]?

5.51 A free particle of mass m , moving in one-dimensional space, has the following mo-
mentum probability amplitude .

b(k) = bO, Ikal < If

b(k) = 0, Ikal ~ If

(a) Obtain the value of bo from the normalization of the wavefunction, o/(x) . What
are the dimensions of b(k)* b(k) dk ? Does your answer agree with these dimen-
sions?
(b) At which values of x will the particle not be found?

sinZcy

1
00
Data: --z-dy = ctt
00 Y
5.52 A student argues the following . If a particle is in an eigenstate of a one-dimensional
box of width a, then we know its energy exactly. But the energy in the box is purely
kinetic . Therefore we know the particle's momentum as well. This conclusion is a
counter example to the uncertaint y relation as the uncertainty in the particle 's location
is finite (a). Punch a hole in the student's argument.
Problems 151

5.53 A free particle moving in one dimension is in the state

(a) What values of momentum, Px, of the particle will not be found ?
(b) If momentum of the particle in this state is measured , in which momentum state
is the particle most likely to be found?
(c) If a = 2.1 A, and the particle is an electron, what value of energy (in eV) will
measurement find in the state described in part (b)?
5.54 (a) Show that [A , B] = 0 {:> [B- 1, A -I] = O.
(b) Show that [B- 1, C] = 0 {:> [B, C] = O.
(c) Establish Jacobi 's relation : [A , [B, Cll + [B , [C , Al] + [C , [A , B]] = O.
Answer (Partial)
(b) B- 1C = C B- 1, C = BC B- 1, CB = BC . The converse is proved by reversing
the sequence of relations .
CHAPTER

6 Time Development, Conservation

Theorems, and Parity

6.1 Time Development ofState Functions

6.2 Time Development ofExpectation Values
6.3 Conservation ofEnergy, Linear and Angular Momentum
6.4 Conservation ofParity

In this chapter we pursue the study of time development of the state function in
greater generality than we did in our previous discussion in Chapter 3. This de-
scription leads naturally to the concept of constants of the motion in quantum
mechanics and again to the notion of stationary states. The distortion of a wave
packet in time is obtained with the aid of the free -particle propagator. Classical
motion ofthe packet is obtained in the limit Ii -+ O. The significance to physics of
constants of the motion was described in Chapter 1. We now find that such con-
stants stem from related fundamental symmetries in nature. In the two chapters
to follow, the principles and mathematical formalism developed to this point are
applied to some practical one-dimensional problems.

6.1 • TIME DEVELOPMENT OF STATE FUNCTIONS

The Discrete Case

Let us recall the recipe for solution to the initial-value problem in quantum me-
chanics (Section 3.5). The initial -value problem poses the question: Given the
state y,(x , 0), at time t = 0, what is the state at t > 0, y,(x, t)? The answer is
(3.70):

y,(x, t) = exp
(-dlt )
-li- y,(x ,O) (6.1)

We recall that the exponential operation is written for its series representation,

(6.2)

152
6.1 Time Development of State Functions 153

Suppose that this exponential operator operates on an eigenfunction ({In of iI .

Then H as it appears in the exponential is simply replaced by En; that is,

exp - i iIt ) ({In = exp (-iEnt)

-fj,- - f j , - ({In (6.3)
(

As an application of this property we consider the problem of a particle in a one-

dimensional box with walls at (0, a), which is initially in an eigenstate of the
Hamiltonian of this system.

Vrn(x,O) = ((In(x) (6.4)

Then the state at time t is

Vrn(x , t) = exp (-ifft)

-fj,- ((In(x) .
= e-IWnl({Jn(x)
Vrn(x , t) = e-iwnl({Jn(x) (6.5)

lUvn = En = n 2EI
As described in Section 3.5, the time-dependent eigenstates, Vrn (x , t) of iI , are
called stationary states. We recall a very important property of a stationary state
(3.76)-that the expectation of any operator (which does not contain the time
explicitly) is constant in a stationary state . As an example of a stationary state ,
=
consider the n 5 eigenstate of the problem at hand, .

Vrs(x, t) = e-i2SEll lli/f; sin c: X

) (6.6)

The eigenstate Vrs oscillates with the frequency 25 E 1/h. Both real and imaginary
parts of Vrs(x, t) are standing waves. The expectation of energy in this state is
constant and equal to 25EI .
Suppose, on the other hand, that Vr (x , 0) is not an eigenstate of iI. Under
such circumstances, to determine the time development of Vr (x, 0) one calls on
the superposition principle and writes Vr(x,O) as a linear superposition of the
eigenstates of iI.

Vr(x,O) = I)n({Jn(X)
(6.7)
bn = (({JnIVr(x, O)}

If we now invoke (6.1), the calculation of Vr (x, t) becomes tractable.

Vr(x, t) = exp - i iIt )

-fj,- ""
£....Jbn({Jn(x)
(
154 Chapter 6 lime Development, Conservation Theorems, and Parity

"" (-iilt)
= ~bnexp ---h--- ~n(X) (6.8)

= Lbne-iWnt~n(X)

liwn = En = nZE,

This solution indicates that each component amplitude bn~n oscillates with the
corresponding angular eigenfrequency W n .
Consider the specific example in which the initial state is

+ 2 sin(Jrxja)
~
Sin(2JrX ja)
1jr(x,O) = - ~ (6.9)
a v5
This state is depicted in Fig. 6.1 and is simply the superposition of the two eigen-
states ~Z and ~, . That is, in the expansion (6.7), one obtains

2 1
b,=- , bz=-
-/5 -/5 (6.10)
b; = 0 (for all other n)

The state of the system at t > 0 is given by (6.8).

+ 2e- iw\t Sin(JrXja))

~
e- iW2t sin(2Jrxja)
1jr(x, t) = - ( ~ (6.11)
a v5
How are these time-dependent solutions related to experimental observations? Let
us rewrite (6.8) in the form

I/! (x , 0)

a x

FIGURE 6.1 Initial state

f2 (Sin(2JrXja) + 2 Sin(Jrxja))
1/!(X, 0) = Y~ .,f5
6.1 Time Development of State Functions 155

(6.12)

so that bn (t) now includes the exponential time factor

(6.13)

Suppose that the energy is measured at t > O. What values will result, and with
what probabilities will these values occur? As in Section 5.1, calculation of the
expectation of E yields

(6.14)

Again, we find that the square of the coefficient of expansion bn (t) gives the
probability that measurement of E at the time t finds the value En.

(6.15)

For the state (6.9) this probability distribution is

4
P(Ed =5
1
P(Ez) = 5 (6.16)

P(En) = 0 (for all other n)

For the initial state (6.9), at any time t > 0, the probability that measurement of
energy finds the value E 1 is ~ . Similarly, the probability that measurement finds
the value 4EI is ~ .
What is the expectation of Eat t > 0 for the initial state (6.9)?

«e-iW2t<pzl + (2e-iw \t<pII)(Hle-iW2t<pz) + HI2e- iW\t<pI)

(E)t >o = 5 (6.17)

Ez +4EI 8
= 5 = (E)t=o = 5E 1

The "cross terms" vanish due to orthogonality of the eigenstates of H, and one
finds that the expectation of energy is constant in time. More generally, for any
isolated system, in any initial state: (l) the probability of finding a specific energy
En is constant in time; (2) the expectation (E) is constant in time.
These rules follow directly from (6.13) through (6.15).

P(En) = Ibn(t)l z = e+iwnt e- iwnt bnr»,

P(En) = Ibnl z = constant in time (6.18)
(E) = L Ibn (r) I En = L Ibn eEn = constant in time
Z
(6.19)
156 Chapter 6 lime Development, Conservation Theorems, and Parity

!/I(x, 0)

FIGURE 6.2 Initial state for a free particle.

The Continuous Case. Wave Packets

Next, consider the problem of a free particle moving in one dimension. Let the
particle be initially in a localized state y, (x, 0) such as that depicted in Fig. 6.2.
Since the eigenstates of the Hamiltonian for a free particle comprise a contin-
uum, the representation of y,(x, 0) as a superposition of energy eigenstates is an
integral [see (5.26) et seq.].

y,(x,O) = 1
M=
...;2Jr
100

-00
b(k)eikx dk (6.20)

b(k) = 1
M=
...;2Jr
100

-00
y,(x, O)e- ikx dx

The state of the particle at t > 0 follows from (6.1).

y,(x , t) = exp ( -iliHt) ~ i: b(k)eikx dk (6.21)

y,(x, t) = -1-
..j2i(
100

-00
b(k)ei(kx-wt) dk
(6.22)
li2k2
!iw= - =Ek
2m
While the component amplitudes of the state function of a particle in a box os-
cillate as standing waves, the k-component amplitudes of the free-particle state
function propagate . For each value of k, the integrand of (6.22) appears as

b(k)exP[ik(x-~t)] (6.23)

The phase of this component, [x - (wi k)t] , is constant on the propagating "sur-
face,"
co
x =-t (6.24)
k
6.1 lime Development of State Functions 157

This is a surface of constant phase. It propagates with the phase velocity

w lik
V=-=- (6.25)
k 2m

The components with larger wavenumbers (shorter wavelengths) propagate with

larger speeds. The long-wavelength components propagate more slowly.
Suppose that at t = 0, the state 1{t (x, 0) is a tight bundle of eigenstates of if .
When the clocks begin to move, each k component propagates with a distinct
phase velocity. The initial state begins to distort. It may be that the initial state
remains somewhat intact and moves. In this case one speaks of a propagating
wave packet. To have a wave packet propagate, it is necessary that the average
momentum of the particle in the initial state does not vanish.

(p)t=O = (1{t(x, 0) Ip1{t (x , 0» i= 0 (6.26)

Furthermore, since the packet is localized in space,

only over a small domain (6.27)

The velocity with which such a packet moves is called the group velocity.

Vg = awl (6.28)
ak kmax

The meaning of kmax is that the amplitude Ib(k)1 2 is maximum k = k max.

(6.29)

This approximation becomes more accurate the more peaked is! Ib(k)1 2 .
Combining (6.28) and (6.29) gives

(6.30)

likmax (p)
=--=-=VCL
m m
The packet moves with the classical velocity (p)/m .
As an example of these concepts, consider a beam of neutrons each of which
has momentum nko. The beam is "chopped," producing a pulse a em long and
containing N neutrons (Fig. 6.3). The state function for each neutron at the instant

I However, if Ib(k)1 2 becomes too peaked, condition (6.27) is violated ; that is, !/f(x) spreads out too
much.
158 Chapter 6 lime Development, Conservation Theorems, and Parity

a
.

FIGURE 6.3 Chopped wave of length a.

after the pulse is produced is

1 'kox a a
-e' -- < x <
2 - - +-2
1jr(x,O) = : (6.31)

1 elsewhere

If the momentum of anyone of the neutrons is measured at t > 0, what values

may be found and with what probability do these values occur? To answer this
question, we need calculate only the expansion coefficients b(k) of (6.20).

b(k) = -1-
../2rra -a 12
l+ al 2
eikoxe-ikx dx (6.32)

= (2 sin[(k - ko)aj2]
V~ k-ko

The state at time t > 0 is

__1_1
1jr(x, t) -
rrya
r:
-00
00
sin[(k - ko)aj2] i(kx-wt)
k k
- 0
e dk (6.33)

with

(6.34)

The amplitude b(k) is sketched in Fig. 6.4.

The momentum probability density P (k) gives the probability that measure-
ment of momentum of any of the neutrons yields a value in the interval hk to
Ii(k + dk) . It is given by

2
P k = Ib(k)1 = b(k) 2 = 2- sin2[(k - ko)aj2]
J~oo Ib(k)1 2 dk I I
(6.35)
() it a (k - kO)2
6.1 Time Development of State Functions 159

I b(k) 12

FIGURE 6.4 Momentum probability density corresponding to the pulsed wave of

Fig. 6.3.

This probability density is constant in time .2 At any time t > 0, it is most likely
that measurement of momentum of any particle in the pulse finds the value

p = l'ikmax = nko (6.36)

Recall that this was the only momentum the neutrons had before the beam was
chopped.
At any time t > 0, the momentum values

hk = nko + 2mr n (n = 1,2,3, .. . ) (6.37)

a

have zero probability of being found. These momentum eigenstates do not enter
into the superposition construction of 1/!(x, 0) .
How many neutrons will be found with momentum in the interval n(k - ko) -
l'iko to n(k - ko) + l'iko? The answer is

!:i.N = N l k

(k- 2kol
2
Ib(k)1 dk (6.38)

This number is also constant in time .

Consider next the Fourier decomposition of a square wave packet as depicted
in Fig . 5.4. There we see that the largest k component corresponds to k = 0 with

1
({JO =-- (6.39)
grr
This is a "flat" wave. The other k components in the superposition of the square
wave serve to taper the sides of the pulse. Since p = 0 for this packet, it does not
propagate-it only diffuses .

2This property of the free-particle momentum probability density is more fully developed in Sec-
tion 7.4.
160 Chapter 6 lime Development, Conservation Theorems, and Parity

The Gaussian Wave Packet

A more rewarding problem both from the pedagogical and physical points of view
is that of the diffusion and propagation of a Gaussian wave packet, discussed
previously in Section 3.3. The initial state is

1{r(X 0) = 1 e'ik ox e-X 2/4a2 (6.40)

, a 1/2 (21T) 1/4

The corresponding initial probability density

P(x,O) = 1{r*1{r = _1_ e- x 2/2a2 (6.41)

a,J2ii
is properly normalized as

The initial uncertainty in position of a particle in the state (6.40) is the square root
of the variance

tl.x = a (6.42)

The complex modulation exp(ikox) in the state (6.40) serves to give the particle
the average momentum

(p) = liko (6.43)

It follows that the initial Gaussian state function (6.40) represents a particle local-
ized within a spread of a about the origin and moving with an average momentum
Mo.
The momentum amplitude corresponding to this initial state is

(6.44)

The Fourier transform of a Gaussian is itself Gaussian (see Fig. 6.5). The initial
momentum probability density

(6.45)

is normalized, centered about the value k = ko, and has a spread tl.k = (2a)-I. It
follows that in the initial Gaussian state,
6.1 lime Development of State Functions 161

I b(k) 12

a I
2a

o x k

FIGURE 6.5 The momentum probability density Ibl 2 corresponding to a Gaussian po-
sition probability density, 11/11 2 • is Gaussian . In this state /).p /).X has its minimum value at
Fi12.

t.x t.pl
Gauss
= t.xlit.k = ~2 = t.x t.pl .
mm
(6.46)

The product of uncertainties has its minimum value in a Gaussian packet.

Free-Particle Propagator
Next we turn to the construction of 1jr(x, t) from the initial state (6.40) . The value
of this function may be obtained from (6.21 et seq.)

1/1 (x , t) = _1_1 1
21T
00

- 00
00

-00
dx ' dk e- ikx'1jr(x', O)ei(kx-wt) (6.47)

1 (21T)5
= a1/2 1 /4 1 00

-00
(
dx' exp ikox' - 4a2
x'2 )
2 2
x
1 00
-00 dk exp [ i [ k r
k(x - x') - - a- t]!
where the time constant r is defined as

lik 2 2ma 2
w= - r=-- (6.48)
2m' Ii

Let us take advantage of our construction of 1jr (x, t) at this point of the analysis
to introduce the free-particle propagator, K (x ' , x ; t) . This function provides a
formal solution to the free-particle , initial-value problem through the prescription

1jr(x, t) = i: dx' 1jr(x', O)K(x', x; t) (6.49)

162 Chapter 6 Time Development, Conservation Theorems, and Parity

The explicit form of K(x', x; t) is inferred from (6.47) .

K(x ', x; t) = 27r

1 1 00

-00 dk exp I[
i k(x - x') - -
kra-
2 2
t]I (6.50)

With the aid of the integral (see Problem 6.5)

~
1 00

-00
e-uy 2 eVY dy = _e v2 / 4u
U
(Reu > 0) (6.51)

there results'

~
i (X -xl)2r]
K(x' , x ; t) = -'--2- exp 2 (6.52)
147ra t [ 4a t

~tu
i m (X _XI)2]
= --exp
'Iat i [ 2nt

Having found this explicit form for the free-particle propagator, let us return to
(6.49) and see its meaning. The wavefunction 1jr(x, t) gives the probability ampli-
tude related to finding the particle at x at the instant t . If the particle was at x ' at
t = 0, then the probability that it is found at x at t > 0 depends on the probability
that the particle propagated from x ' to x in the interval t. This is what (6.49) says.
The probability amplitude that the particle is at x at time t is equal to the initial
amplitude that the particle is at x ' multiplied by the probability amplitude of prop-
agation from x ' to x in the interval t, summed over all x'. Thus we may interpret
K (x, x ' ; t) as the probability amplitude that a particle initially at x' propagates to
x in the interval t. It should be noted that the explicit fonn (6.52) is appropriate
only for free-particle propagation. For more general problems involving interac-
tion , the form of (6.49) still maintains, although the propagator function is more
complicated (see Problem 6.26) .

Distortion of the Gaussian State in lime

Let us return to the calculation of 1jr(x, t), given the initial Gaussian distribution
(6.40). To complete the calculation one need merely complete the x ' integration
in (6.49).

00
I
1jr(x, t) = 1/2 1 1/41 dx ' [ex p (ikOX - XI:)] K(x ' , x; t)
a (27r) -00 4a

3To obtain a convergent integral, first replace i by ex == i + E, where E is a small real positive number.
After integrating, let E --> O.
6.1 Time Development of State Functions 163

Employing the explicit form (6.52) for K and once again utilizing the integral
formula (6.51) gives the desired result.
1
1Jr(x, t) = a l / 2 (2rr) I/4(1 + it/r:)1/2 exp[iko(x - vgt)]

(x -Tikot/m)2]
x exp - (6.53)
[ 4a 2(l +it/r:)

where vg = 1iko/2m. The corresponding probability density is

2 1 [ (x-1ikot/m)2]
P(x , r) = 11Jr(x, t)1 = exp - 2 2 2 (6.54)
a,J21T(l + t 2/r:2)1 /2 2a (l + t /r: )

If we compare this form with the initial probability density we see that the generic
shape of P (x, 0) (i.e., that of a bell) has remained intact with three modifications.
It has become wider,

Second, the center of symmetry of the packet is now at

x = vot
where we have labeled
1iko
Vo=:-
m
It follows that the probability density of a Gaussian wave packet propagates with
a velocity that is directly related to the expectation of momentum of the particle
in the Gaussian state. Finally, the height of the density function has diminished.

The area under the curve P, at any time, remains unity.

A sequence of packet contours is shown in Fig. 6.6. It is quite clear that the
packet begins to distort significantly after a time interval r . If we represent a piece
of chalk by a wave packet, a c:: 1 em, m c:: 1 g, there results

r c:: 1027 s c:: 1020 yr

But the universe is only r - 1010 yr old. That is why classical objects are never
observed to suffer a quantum mechanical spreading.

Flattening of the 8 Function

There are two limits that can be taken on the probability density P(x, t) related
to the Gaussian wave packet which are very revealing. The first evolves from the
164 Chapter 6 Time Development, Conservation Theorems, and Parity

P(x ,O)
[1 + (t/r)2) 1/2
.................... \ P(X, r)
. . . . . . i.
-- ......
---
P(x,2r)

x=O

FIGURE 6.6 Shrinkage and spreading of the probability distribution corresponding to a

Gaussian wavepacket. At any time t ,

i: P(X, t)dx = 1

initial state

P(x,O) = 11/I(x, 0)1 2 = 8(x) (6.55)

A valid representation of the delta function is given by the limit

8(x) = lim I e- x 2/ 2a2 (6.56)

a--.O a (2Jr) 1/2

This function has the correct delta function properties (3.27 et seq .).
Measurement of the position of a particle which finds the value x = 0 leaves
the particle in the state

1/1 = 8(x) (6.57)

This state is not normalizable. The state given by (6.55) is a little less sharply
peaked than (6.57) and is normalizable.
To obtain the probability density P (x , t) which follows from the initial value
(6.55) , we merely examine (6.54) in the limit, a -+ O. There results

.
hm P(x, t) = lim
. 2ma
;;:c exp -
[2a 2(X2-11k
2
ot/ m)2]
2 (6.58)
t li v 2Jr t Ii / m

2ma 2
= lim ;;:c[l + O(a )]
tliv 2Jr
The notation O(a 2 ) denotes "order" of a 2 . It stands for a group of terms, the sum
of which goes to zero like a 2 , with decreasing a.
From expression (6.58) we see that for all t > 0, P vanishes uniformly for all
x, in the limit a -+ O. This instantaneous flattening of an infinitely peaked state
(6.55) is due to the following circumstance. The momentum probability density
Problems 165

Ib(k) 12 corresponding to such a state, depicted in Fig. 5.6b, is flat. This means that
it is equally probable to find any k value, no matter how large k is, in this state. At
any instant t > 0, at any point x, the components I{Jk with k values which obey the
inequality, (lik/m)t :::: x , have overtaken that point. The initial infinitely peaked
distribution assumes an (almost) instantaneous flattening ."

The Classical Particle

The second limit we wish to consider relative to the probability density P (x , t),
(6.54), changes P(x , t) to the classical probability relating to a point particle of
mass m moving with velocity liko/m . This is accomplished by setting Ii -+ 0 in
P(x , t) (except where Ii appears in Po = !iko) .

lim P(x, t) =
1
r;:c exp -
[(X - pot/m)2]
2 (6.59)
n..... o av 2rr 2a
liko = PO = constant
For this probability to relate to a "point particle" we impose the additional con-
straint, a -+ O. This gives

. P(x , t) = 8 (xp- o
hm t ) = PcLCx , t)
--;;;- (6.60)

The probability of finding the particle at t is zero everywhere except on the clas-
sical trajectory

pot
X=- (6.61)
m
This is another example of the correspondence principle at work. In essence, the
"leading term" (i.e., the term not containing Ii) in the expansion of P (x , t) about
Ii = 0 gives the classical result.

PROBLEMS

6.1 (a) Find 1/!(x , t) and P(En) at t > 0, relevant to a particle in a one-dimensional box
with walls at (0, a), for each of the following initial states.

(1) 1/!(x,O) = AI sin C: X

) cos CaX
)

(2)5 1/!(x,O) = A2X2(X - a)2

40f course, these conclusions become erroneous for x f t 2: c. To obtain a completely physically
valid solution for the infinitely peaked initial state, it is necessary to solve the relativistic form of
the Schrodinger equation. See related discussions on the Dirac equation in A. Messiah, Quantum
Mechanics, Wiley, New York, 1966. See also Chapter 15 of the present work.
5Faulty apparatus. £2 cannot be measured in this state.
166 Chapter 6 lime Development, Conservation Theorems, and Parity

(3) 1{!(x,O) = A3[ei1T(X-a)/a - 1]

(b) If measurement of E finds that E = 4 E) at 6 s, what is 1{! (x , f) at f > 6 s for each
of the initial states in part (a)?

6.2 Consider the following three dispersion relations.

(1) W2=9k
c2 k 2
(2) w2= ~
1 - (wO/w)2
(3) w 2 = wp2 + 3C 2k2
The first relation obtains to deep-water surface waves (9 is the acceleration due to
gravity), the second to electromagnetic waves in a waveguide, and the third to longi-
tudinal waves in a "warm" plasma (w p is the plasma frequency and C is the thermal
speed). For all three cases find (1) the phase velocity and (2) the group velocity of a
wave packet propagating in the respective medium.

6.3 (a) Show that the free-particle propagator (6.52) has the following property and inter-
pret the result physically.

K(x', x ; 0) = 8(x' - x)

(b) Show that K satisfies the integral equation

K(x',x; f - fO) = f K(x',x"; f - f)K(x" ,x ; f) - fo)dx"

and interpret this result physically in terms of the evolution in time of the state
1/f(x, fO), first from fO to f) and then from f) to f .

Answer (partial)
(a) Set if = E 2 and compare with (C6) of Appendix C. The interpretation of this
result is that for infinitesimally short time intervals, the probability amplitude for
propagation away from the initial point x' is zero, except in a small neighborhood
about the initial point.

6.4 At f = 0, 105 noninteracting protons are known to be on a line segment 10 em long. It

is equally probable to find any proton at any point on this segment. How many protons
remain on the segment at f = 10 s? [Hinf: Let the center of the segment be at x = O.
Then the formal answer to the problem with 1{!(x, f) normalized is

6.N = 10515 11{!(x, 10)1 2 dx

-5

To construct 1{!(x, f) , the initial square pulse must first be written as a superposition of
f{Jk states . With b(k) calculated,

11{!(x, t)12 = -I
2n
1100

-00
00

-00
dkdk'b(k)b*(k')ei(w'-w)
t ei(k-k)x
'

where w' is written for w(k') .]

Problems 167

6.5 The integration involved in obtaining (6.51) is of the type

1
00 2
S= e- UY eVY dy
-00

where u and v are constants. Evaluate this integral.

Answer
The aim is to transform the exponent _ uy 2 + vy to a perfect square. First we set

ui - vy == a 2i - 2afJy

which gives

The exponent may now be written

ui - vy = (ay - fJ)2 - fJ2

and the integral S becomes (with TJ = ay - fJ)

6.6 (a) An electron is in a Gaussian wave packet. If the packet is to remain intact for at
least the time it takes light to move across I Bohr diameter, 2ao = 2;,,2/ me 2 , what
is the minimum width, a, that the Gaussian packet may have (in ern)?
(b) What is the diffusion time (r) for an electron in a Gaussian wave packet of width
e2/mc2 (in s)? This is the classical radius of the electron. How far does light travel
in this time (in ern)?

6.7 A free particle of mass m moving in one dimension is known to be in the initial state

1/f(X, 0) = sin(kox)
(a) What is 1/f(x, t)?
(b) What value of p will measurement yield at the time t , and with what probabilities
will these values occur?
(c) Suppose that p is measured at t = 3 s and the value hko is found. What is 1/f(x, t)
at t > 3 s?

6.8 A particle moving in one dimension has the wavefunction

1/f(x, t) = A exp[i(ax - bt)]

where a and b are constants.

(a) What is the potential field Vex) in which the particle is moving?
(b) If the momentum of the particle is measured, what value is found (in terms of a
and b)?
(c) If the energy is measured, what value is found?
168 Chapter 6 lime Development, Conservation Theorems, and Parity

6.2 • TIME DEVELOPMENT OF EXPECTATION VALUES

The law that covers the time development of the expectation of an observable,
(A), follows from the time-dependent Schrodinger equation. We wish to calculate
d(A)jdt . Since (A) has all its spatial dependence integrated out, it is at most a
function of time . We may therefore write

d(A) a(A)
= (6.62)
dt at
In the state lJr(x, t), this expression becomes

(6.63)

The time derivative of the product is

a ~ (a. ',*) ~
-(lJr*AlJr)= _'1'-
~ a,l, aA
AlJr + lJr*A-'I' +1Jr*-lJr (6.64)
at at at at
Employing the time-dependent Schrodinger equation

alJr -iH alJr* i HlJr*

at = -li- lJr, at"=-li- (6.65)

in (6.64) gives

a *AlJr)
-(lJr HlJr * AlJr
A = -i (A A -lJr * AHlJr n * -lJr
A A + -:-lJr aA ) (6.66)
at Ii I at

Substituting this expansion in (6.63) gives

-d(A) = -i ( (HlJrIAlJr)
dt Ii
A A - (lJrIAHlJr)
AA + -:-IiI ( lJr I -lJr
aA ))
at
(6.67)

Since His Hermitian, the first term on the right-hand side of (6.67) may be rewrit-
ten to yield the final result,

-d(A) =
dt
(i
ti
~A aA)
-[H,A]+-
at
(6.68)

If A does not contain the time explicitly, then the last term on the right-hand side
vanishes and

(6.69)
6.2 lime Development of Expectation Values 169

In the event that A commutes with H, the quantity (A) is constant in time and
A is called a constant ofthe motion. For a free particle, p commutes with Hand
(p) is constant in time for any state (wave packet). Since H commutes with itself,
(H), the expectation of the energy, is always constant in time.
Let a particle moving in one dimension be in the presence of the potential
V (x) . The Hamiltonian of the particle is
A2
H = L + Vex) (6.70)
2m
How does (x) vary in time? Equation (6.69) gives
d(x) i A

dt = "h,([H,x]) (6.71)

= ~ ([ t~, x]) = 2~n (p[p, x] + [p, x]p)

i .
= -(-2Inp) = (p)
-
Zmtt m

or, equivalently,

d(x)
m = (p) (6.72)
dt
This equation bears the same relation between expected values of displacement
and momentum as in the classical case. Equation (6.72) cannot hold for the eigen-
x
values of and p, since such an equation implies that x(t) and pet) are simulta-
neously known .

Ehrenfest's Principle
The reduction of quantum mechanical equations to classical forms when averages
are taken, such as demonstrated above, is known as Ehrenfest 's principle. New-
ton's second law follows from the commutator [H , p], which for the Hamiltonian
(6.70) is

[H,
A

p] = in-
av (6.73)
ax
Again using (6.68), one obtains

d(p) _
- --- - (av) (6.74)
dt ax
which is the x component of the vector relation

d(p)
- = -(VV(x, y, z») = (F(x, y, z») (6.75)
dt
170 Chapter 6 Time Development, Conservation Theorems, and Parity

where F is the force at (x , y, z). In any state 1/r(x, t), the time development of the
x
averages of and p follow the laws of classical dynamics, with the force at any
given point replaced by its expectation in the state 1/J(x , t) . (See Problem 6.31.)

PROBLEMS

6.9 Show that if [H ,A] = 0 and aAIat = 0, then ( ~ A ) is constant in time.

6.10 Show that

~ (A)=O
dt

in a stationary state, provided that aAlat = 0, using the commutator relation (6.68).

Answer
d (A ) i A A i A A A A

----;j( = /i(fPnl[H, AlfPn) = /i(fPnl(HA - AH)fPn)

i A A A A

= /i«(HfPnIAfPn) - (fPnIA H fPn))

6.11 Show that for a wave packet propagating in one dimension,

d(x 2 )
m = (xp ) + (px)
dt
6.12 A particle moving in one dimension interacts with a potential V (x) . In a stationary
state of this system show that

~(x~V)=(T)
2 ax
where T = p 2 / 2m is the kinetic energy of the particle.

Answer
In a stationary state,

d i A A A

dt (xp) = /i([H,x p]) = 0

Expanding the right-hand side, we obtain

0= (x[H , p] + [H , x]p)
= (x [ V , p] + [T ,x]p)
= ih(X ~~ - 2T)
 6.3 Conservation of Energy, Linear and Angular Momentum 171

6.13 Consider an operator A whose commutator with the Hamiltonian iJ is the constant c.

Find (A) at t > 0, given that the system is in a normalized eigenstate of A at t = 0,

corresponding to the eigenvalue a.

6.14 A system is in a superposition of the two energy eigenstates fIJI and flJz . Physical
properties of the system characteristically depend on the probability density 1/1*1/1.
Show that resolution of any such property involves measurements over an interval
6.t > ii/lEI - Ezi.
Answer
The superposition state is

1/I(r, t) = fIJI (r) exp ( - i - ) + flJZ (r) exp (-iEzt)

- ilElt -li-

so that

z Z
1/I*1/I=lflJIi +lflJzl +2RefIJl*flJZeXp
[i(EI -Ii EZ)t]

This function oscillates between the two extremes (lflJl 1+ IflJzl)Z and (lflJII - IflJzl)Z
with the period ii/lEI - Ezi . It follows that changes in related properties become
discernible only after an interval greater than or of the same order as this period .
The situation is similar to the process of tuning an oscillator to a frequency Wo
by "listening" for beats . The period between beats varies as the inverse frequency
(w - wO)-I . Thus one is certain that w = wo only after an infinite interval.

6.3 • CONSERVATION OF ENERGY, LINEAR AND ANGULAR MOMENTUM

The principle of conservation of energy in classical physics states that the energy
of an isolated system or a conservative system is constant in time. A conservative
system is one whose dynamics are describable in terms of a potential function. A
particle in a one-dimensional box is a conservative system . Suppose that at t = 0,
the state of the particle is

1/1 (x, 0) = 3cpI + 4cps (6.76)

v'25
What can be said of the energy of the particle at the time t > O? Measurement
of the energy has a ls
probability of finding the value Eland a 1~ probability of
finding the value 25EI. At t > 0 the state (6.76) becomes

(6.77)
172 Chapter 6 lime Development, Conservation Theorems, and Parity

The probability that measurement yields E 1 is

(3e-iElt /fi)*(3e-iElt /fi) 9

P(EJ) = 25 = 25 (6.78)

A similar calculation of P(E3) yields the constant value ~. In other words, in

the state given, one cannot say with certainty what the energy is at t ~ O. In what
sense is energy conserved? The answer is, in the average sense. It follows directly
from (6.69) that

(H) = (E) = constant (6.79)

For the example given, at any instant in time the expectation of the energy is

9E1 + 16E5
(E) = 25 = 16.36E1 = constant (6.80)

For a free particle, p also commutes with Ii; hence we can conclude from
(6.68) that

(p) = constant (6.81)

The energy and total momentum of an isolated system are constants of the motion.
Conservation theorems in physics are closely related to symmetry principles.
Consider, for example, the fact that the laws of physics do not depend on the time
at which they are applied. Newton's second law, Maxwell's equations, and so on,
do not change their structure with time. This symmetry of time (i.e., homogeneity)
gives rise to the conservation of energy. Let H be the Hamiltonian of the whole
universe . Homogeneity of time implies that H is not an explicit function of time.
This together with (6.68) implies that d(E}/dt = O. We may reach the same
conclusion for any isolated system (Fig. 6.7).

Black box far from

any other heavenly
bodies

t» 0, t » t' > °
FIGURE 6.7 The laws of physics are the same at (and t' -* aH/a( = 0 -* (E) =
constant.
6.3 Conservation of Energy, Linear and Angular Momentum 173

AX

t =0 t =0
FIGURE 6.8 The laws of physics stay the same --+ aHlax = 0 --+ (p ) = constant.
Note that in this "thought" experiment, translation occurs in zero time. This is called a
"virtual" displacement. It is demanded by the details of the argument. To physically effect
a virtual displacement one merely imagines two identical, noninterfering boxes a distance
~x apart.

Conservation of momentum for an isolated system depends on the homogene-

ity of space. Go out in space to a point far removed from other objects. Enclose
yourself in a box with no windows and opaque walls. Let the box suffer a "virtual"
displacement (Fig. 6.8). There is no experiment which will reveal that the box is
at a new location. Consequently, for example, the dynamical laws of an isolated
system of particles can only depend on the relative orientation of particles, not
on the distances from these particles to some arbitrarily chosen origin. Equiva-
lently, the Hamiltonian of the system can always be transformed so that it does
not contain these variables (i.e., the coordinates of the center of mass).
To find the basis of the relation between the homogeneity of space and conser-
vation of linear momentum, we tum back to Problem 3.17, where it was shown
that the p operator effects the displacement

D(Of(x) = exp (i{px)

A
-li- f(x) = f(x + {) (6.82)

In this expression f is any differentiable function of x. For infinitesimal displace-

ment ({ --+ 0), the displacement operator becomes

n(o = j + i{px (6.83)

Ii
or, equivalently,
Ii A A

Px = i{ [D(O - I]

where the identity operator is 1. As observed previously, the Hamiltonian of an

isolated system cannot depend on displacement of the system from an origin at an
174 Chapter 6 lime Development, Conservation Theorems, and Parity

arbitrary point in space. Therefore, the displacement operator i> commutes with
if, whence Px does also. Again calling on (6.69), we recapture the constancy of
(Px). However, in the present argument we see how this conservation theorem
finds its origin in the symmetry of the homogeneity of space.
In three dimensions the displacement operator becomes

:D fer) = exp
A ( -fj,-
i~'P) fer) = fer +n (6.84)

Again, for an isolated system, one may conclude that if commutes with i> and
therefore with p, the total linear momentum of the system. It follows that the
vector (p) is conserved.
Let us return to the experimental "black box" described above. The fact that
experiments performed within the box are impervious to the box's location in
space or time implies, respectively, conservation of linear momentum and energy.
Suppose now that the box undergoes a rotation through the angle t1¢ about an
arbitrary fixed axis in space. Owing to the isotropy of space, experiments within
the box cannot detect such rotational displacement. They are impervious to the
box's orientation in space (Fig. 6.9). It follows that the Hamiltonian of the system

(a) (b)

FIGURE 6.9 If fer/»~ is a property of an isolated system that depends on the orientation
r/> of the system about an arbitrary fixed axis, rotation of the system through the angle f:!,.r/>
causes the property to change to fer/> + f:!,.r/» . Isotropy of space precludes the existence of
such a property. This invariance with respect to rotation implies conservation of angular
momentum .
6.3 Conservation of Energy, Linear and Angular Momentum 175

cannot depend on ¢ , the rotational orientation with respect to some fixed axis,
in the same way that it cannot depend on the displacement l; from an arbitrary
point in space. As a consequence of this rotational symmetry, the total angular
momentum of the system is conserved.
Suppose that there is a property of the system which is dependent on the sys-
tem's rotational orientation ¢ about a fixed axis, which we designate the z axis.
Let the measure of this property be j(¢) . After rotation of the system through
the angle 6.¢ , ¢ ~ ¢ + 6.¢ and j(¢) ~ f(¢ + 6.¢). This transformation of
function is effected by the rotation operator R6.l/> :

RD.lj>j(¢) = j(¢ + 6.¢)

(6.85)
R6.l/> = exp
A ( -i6.¢L
li-)
z

Here i; is the z component of the total angular momentum of the system.P Since
the Hamiltonian of the (isolated) system cannot depend on ¢ , it is insensitive to
the rotation operator, RD.lj>; that is, if commutes with RD.lj> . Hence it also com-
mutes with i ; and we may conclude that (L z ) is constant. More generally, rota-
tion through the vector angle 4.«fJ (the direction of 4.«fJ is parallel to the axis of
rotation) is effected by the operator

R6.l/> = exp
(i4.«fJIi • L) (6.86)

The argument demonstrating the constancy of i; carries over to L, the total an-
gular momentum of the system .
In summary, with p and L denoting , respectively, the total linear and angu-
lar momentum of an isolated system whose Hamiltonian is H, the following
symmetry-conservation principles hold .

Homogeneity of Space

d
=0 ~ =0
A

[H , p] - (p) (6.87)
dt

Isotropy of Space

d
=0 ~ =0
A A

[H, L] - (L) (6.88)

dt

6This relation is derived in Problem 9.17.

176 Chapter 6 Time Development, Conservation Theorems, and Parity

Homogeneity of TIme

ail
-at =o~
d
-(E) =0 (6.89)
dt

PROBLEMS

6.15 Under what conditions is the expectation of an operator A (which does not contain
the time explicitly) constant in time?

Answer
Under either of the following conditions:
(a) [A, H] = O.
(b) (A ) is calculated in a stationary state.

6.4 • CONSERVATION OF PARITY

Consider an experiment and its mirror image (Fig. 6.10). Such an experiment
might be the observation of the orbit of a missile fired in a uniform gravity field
or two particles colliding. These phenomena obey certain physical laws. Suppose
that we formulate the laws obeyed by the image orbits in the mirror. They are
the same as the laws that the orbits in the real world obey. This is a symmetry

Mirror

Laboratory Mirror-image
experiment experiment

FIGURE 6.10 The laws of physics are the same for the lab experiment and for the mirror-
image experiment. This symmetry statement gives rise to the conservation of parity.
6.4 Conservation of Parity 177

f(x) =f(-x) fix) =-f(-x)

x x

(a) (b)

FIGURE 6.11 Any even function is an eigenfunction of JP with eigenvalue +1:

JPf(x) = +If(x)
Any odd function is an eigenfunction of JP with eigenvalue -1 :

JPf(x) = -If(x)

principle.' In quantum mechanics this principle is associated with a conservation

law, conservation ofparity.8
Parity is a property of a function . A function f(x) has odd parity if

f( -x) = - f(x)

A function has even parity if (see Fig. 6.11)

f(-x) = f(x)

The parity operator JP is defined as?

JPf(x) = f(-x) (6.90)

What are the eigenvalues of JP? Let 9 be an eigenfunction of JP with eigenvalue ex;
then

JPg(x) = g(-x) = exg(x) (6.91)

To find ex we operate again with JP.

(6.92)

7At the molecular level, a striking example of mirror asymmetry is present in biological amino acids.
All such molecule s are "left-handed," (The plane of polarization of plane-polarized light, when passing
through an aqueous solution of left-handed molecules at a given frequency, rotates in a direction
opposite to that caused by a solution of mirror-image right -handed molecules.)
8"Broken symmetry" refers to a phenomenon in which a given symmetry law is not obeyed . For exam -
ple, parity is not conserved in weak-interaction fi-decay . For further discussion, see H. Frauenfelder
and H. Henley, Subatomic Physics , Prentice-Hall, Englewood Cliffs, N. J., 1974.
r
9In three dimensions f(x , y , z) = f (-x, -y , - z) . See Problem 6.23 .
178 Chapter 6 Time Development, Conservation Theorems, and Parity

Hence

ex = ±1 (6.93)

For ex = 1, from (6.91), we obtain

g(-x) = g(x) (6.94)

Any even function is an eigenfunction of J? with eigenvalue +1. For ex = -1,

9 (-x) = -g (x) (6.95)

Any odd function is an eigenfunction of P with eigenvalue -1 . The order of de-

generacy of ex = 1 is infinite. There are no other eigenvalues of J?
How is this parity property connected with the symmetry principle relating
to mirror images mentioned above? Consider that a particle (m) moving in one
dimension interacts with another stationary particle (M) which is at the position
x = O. The potential of interaction between the particles is V (x) . Suppose that the
(moving) particle is at a position x ' > O. The image of the particle seen in a mirror
which intersects the x axis normally at x = 0 is at x = -x' < 0 (Fig. 6.12). The
temporal behavior of the image particle will be the same as that for the laboratory
particle if V (x) = V (-x) . [The potential "seen" by the image particle is V (-x).]
The Hamiltonian for the particle in the laboratory system is
A2
fI = L + Vex) (6.96)
2m

For Vex) an even function , fI commutes with J? To show that J? commutes with
Vex), let 9 (x) be an arbitrary function of x . Then

J?V(x)g(x) = V(-x)g(-x) = V (x)J?g (x) (6.97)

V(x)

x' x

FIGURE 6.12 A mass m interacting with stationary mass M through potential Vex) . If
image dynamics is to be the same as lab dynamics, Vex) = V( - x) .
6.4 Conservation of Parity 179

The fact that JP commutes with the kinetic-energy part of iI is shown in Prob-
lem 6.16. Since JP commutes with both parts of iI, it commutes with iI itself.

(6.98)

Together with (6.69) this gives the conservation principle

(IP') = constant (6.99)

The parity of the state of a system is a constant of the motion.

As an example of this principle, consider a one-dimensional box centered at
the origin, so that its walls are at x = a12, x = -a12 (see Problem 4.1). The
eigenstates and eigenenergies of the Hamiltonian for this system are

_ Vfi~ Sill. (n7rx)

<Pn = -a- (n = 2,4,6, . . . ,2j, . .. ) ]
En = n2El
<Pn = Vfi~ cos (n7rx)
-a- (n = 1,3,5, .. . ,2j + I, . . . )
(6.100)

The eigenstates ipn are odd while <Pn are even.

JPipn = -ipn (6.101)

JP<Pn = <Pn

Suppose that at t = 0 the particle is in the state

VJ (x, 0) = J4~a [6 sin (2: x) + 3 cos (:X) ] (6.102)

I _
= r::;c(6<p2 + 3<pI>
",45
At! > 0,

VJ(x, t) = J4~a [6 sin (2:X) e-iE2t/1i + 3 cos (7raX)e-iElt /li] (6.103)

The expectation of IP' at t = 0 is

(IP') = (VJ(x, 0) IJPVJ (x, O)} (6.104)

A I
IP'VJ (x, 0) = r::;c ( -6ip2 + 3<pl)
",45
I 27
(IP') = 45 {-36(ip2Iip2} +9(<Pll<pl}} = - 45
180 Chapter 6 lime Development, Conservation Theorems, and Parity

Since [J?, H] = 0, this is the value of (IP') for all time. In that the initial state
(6.102) is a superposition of the eigenstates of J?, the squares of the coefficients
of expansion give the probability that measurement finds the system in a state of
even or odd parity. Measurements on an ensemble of 4500 boxes all of whose
particles are in the initial state (6.102) at the time t = 0 would find approximately
3600 of the particles in the odd state CP2 (t) and approximately 900 of the particles
in the even state !PI (t) at the subsequent time t > O.

PROBLEMS

6.16 (a) Show that IP anticommutes with the momentum operator p. That is, show that
[IP, p]+ == IPp + pIP = 0
(b) Use your answer to part (a) to show that IP commutes with the kinetic-energy
operator t = p2/2m .
6.17 A particle in one dimension is in the energy eigenstate

lfiko = A cos(kox)
Ideal measurement of energy finds the value

nho2
E=--
2m
What is the state of the particle after measurement?
Answer
If we recall postulate II of quantum mechanics (Section 3.2), the system is left in
the eigenstate of ii corresponding to the eigenenergy above. Any state in the two-
dimensional subspace of Hilbert space spanned by sin(kOx) and cos(kOx) [or, equiv-
alently, exp(ikox) and exp(-ikox)] gives the eigenenergy above. However, an ideal
measurement perturbs the system least. In the state before measurement the proba-
bility distribution relating to momentum is 1/2 for p = ±/ikO' If we guess that the
system is left in the state exp(ikox) after measurement, the momentum distribution
of the original state was disturbed . If we guess that the state of the measurement is
sin(kox), then measurement has not disturbed the momentum distribution ; however,
this measurement has disturbed the parity. In the original state the parity is +1 (with
respect to the origin x = 0) while that of the hypothesized state after measurement is
-1. This is still not the ideal measurement. We can find a measurement that perturbs
the system even less. Consider that the particle is left in the state cos(kox) . The cor-
responding measurement did not perturb the momentum distribution or the parity of
the original state. It is the ideal measurement.
6.18 (a) If f(x) is any function, show that
f(x) + f(- x) .
f+ = 2 = even function
f(x) - f (-x ) .
f- = 2 = odd function
Problems 181

(b) Show that

A i+IP
lP'+=--
2
is such that

The identity operator is 1.

(c) Show that

A i-IP
lP'_=--
2
is such that

IP-f(x) = f-(x)
The operator IP+ "projects" f onto f + while IP- projects f onto f - .
(d) Show that the projection operators IP+ and IP_ satisfy the following properties:

IP±2 = IP±
[IP+' IP_] = 0
IP++IP_=i
6.19 What is (lP') for a particle in a one-dimensional box with walls at (-aI2, +aI2) in
the initial state

0/ (x, 0) = ~ (3~2 + 4~4 + 21p3)

",29
6.20 For the same one-dimensional box as described in Problem 6.19, it is known that the
particle is in a state with energy probabilities

I I
P(El) = 3' P(E2) = 3'
P(E n ) = 0, (n:j:.I,2,3)

The parity of the state is measured ideally and -I is found. If some time later E
is measured, what value is found? What is the answer if the original measurement
found the parity to be + I?
6.21 For a free particle moving in one dimension, divide the following set of operators
into subsets of commuting operators.

{IP, x, tt, p}
6.22 A free particle moving in one dimension is in the initial state o/(x, 0). Prove that
(p) is constant in time by direct calculation (i.e., without recourse to the commutator
theorem regarding constants of the motion).
182 Chapter 6 lime Development, Conservation Theorems, and Parity

Answer
With

b(k , t) == b(k)e iwt

nh 2
-2m = tuo

b(k) = I
[;C
v2JT
1-00
00
1/f(X, O)e- ikx dx

the expectation of p appears as

which, given the structure of b(k, t) above, is time-dependent. Alternatively, we

may note that for a free-particle wave packet, fI = p2/2m, so that d(p}/dt =
(l /ili)([H , pJ) = O.

6.23 In three dimensions , JP is defined as

JP1/f(x, y , z) = 1/f(-x , -y, -z)

(a) What does this definition become if 1/f is measured in spherical coordinates: that
e,
is, 1/f = 1/f(r, ¢)?
(b) What does it become if 1/f is measured in cylindrical coordinates : that is, 1/f =
1/f(p , z, ¢) ?
(c) What are the parities of the following functions?

1/f1 = A(x + y + z)e- (x2+y 2+ z2)

2
1/f2 = Bre- r cose
oJ p2 + z2 sin ¢
1/f3 = C 5
Z

6.24 Discuss the consistency or inconsistency of the concept of the trajectory of a particle

r = r(t)

(whose mass is also known) in quantum mechanics.

Answer
The trajectory (through differentiation) implies the momentum, p = p(t) . Thus, at
each instant, rand p are known, which is in violation of the uncertainty principle. On
the other hand, for a wave packet one may construct the equation

d I
-(r)
dt
= -(p)
m

If (p) is a known function of time, this equation may be integrated to obtain the
trajectory, that is, (r), as a function of time. The classical trajectory ensues when
Problems 183

quantum fluctuations about the mean (r) oscillate rapidly, thereby "averaging out."
Equivalently, one may say that this limit is reached when the de Broglie wavelength
is small compared to characteristic distances of the configuration.
6.25 Let a particle move in the potential

where A is constant and n is a finite integer. Show that Ehrenfest's equation (6.74)
gives the classical relation

dp = -Anx n- I
dt
only for n = 2 (the harmonic oscillator).

Answer
From (6.75) we obtain

To obtain the classical fonn we must equate (x n- I) = (x)n-I, which is only valid
forn = 2.

6.26 (a) Consider that a particle is bound to the potential V (x) and is initially in the state
1/f(x , 0) . If eigenfunctions of the Hamiltonian iI = p2 / 2m + V (x) are written
Ipn (x), then

iIIpn = Enlpn
obtain the state function 1/f(x, t) in terms of an integral over the propagator
K (x', x ; t) . That is, construct an explicit form for K as it appears in the inte-
gral

1/f(x,t) = J 1/f(x',O)K(x',x ;t)dx'

(b) Show that the propagator you have constructed satisfies the integral equation

K(x',x ;t-tO)= J K(x' ,x";t-tl)K(x",x ;tl-to)dx"

Answer (partial)
(a) The formal solution to the time-dependent Schrodinger equation appears as [see
(3.70)]

1/f(x, t) itiI) 1/f(x, 0)

= exp ( --,;

Expanding the initial state in eigenstates of iI [see (5.6)] gives

1/f(x,O) = I)nlpn(X)
n
184 Chapter 6 Time Development, Conservation Theorems, and Parity

bn = J 1fr (x ' , O)CPn *(x') dx'

Substituting in the above gives

1fr (x, t) -T CPn(x)

= "~bnexp (iEnt)

=J 1fr(x', 0) [ :~;:>n *(x')CPn(x)e-iEnt / n] dx'

=J 1fr(x' ,O)K(x',x ;t)dx'

which serves to identify the propagator

K(x', x ; t) =L CPn * (x')CPn (x)e-iEnt /n

n

In Dirac notation this equation appears as

K(x',x ; t) = L Icpn(x»)e-iEnt/n (cpn(x')1

n

which allows the solution to be written:

11fr(x, t») = KI1fr(x, 0»)

6.27 The wavefunction of a particle of mass m which moves in one dimension is

\IJ(x, t) = Ae i(kx-wt)

where A , k , and (f) are constants. Determine the potential V (x) in which the particle
moves in terms of m, h; k , and (f).

6.28 Consider a particle of mass m moving in one dimension (x) . Let cp(x), q5(x) be eigen-
states of the particle with corresponding eigenenergies E > E. Let xl , x2 be two
successive zeros of q5(x). [Values of x for which q5(x) = 0 are called "zeros" of
q5(x) .]
(a) Show that cp(x) has at least one zero in the interval (Xl, X2) .
(b) If energies are discrete and E is the next larger energy than E, then show that
cp(x) has only one zero in the interval (Xl, X2) . (Zeros are then said to be inter-
la ced. See, for example, Fig . 10.3.)

Answer (partial)
(a) The Schrodinger equation may be written [compare with (7.2)]

=0
cp"(x) +k 2(x)cp(x)

q5"(x) +p(x)q5(x) = 0

k2 >P
where primes denote differentiation. Let Xl , x2 be two successive zeros of
q5(x). Assume that cp(x) =1= 0 on (xj , X2) and that (without loss of generality)
Problems 185

rp(x), ip(x) > 0 on (Xl, X2). Consider the Wronskian of these two solutions :

W ( ip, ip) = rpip' - iprp'

dW(rp , ip) _/I _ /I
dx = rprp - rprp

= rpip(k 2 - 7(2) > 0

In the second equation we employed the preceding Schrodinger equations . Thus
W(x) has positive slope, and we may conclude

Since ip(x) vanishes at Xl, x2, we obtain (from the first of the preceding equa-
tions)

W(Xl) = rp(X\)ip'(X\) > 0

W (X2) = rp(X2)ip' (X2) < 0

so that

which contradicts our previous result. Thus our assumptions are incorrect and we
conclude that rp(x) has at least one zero in (xj , X2)./(Note: The preceding result
is often referred to as the Sturm comparison theorem in the theory of differential
equations .)

6.29 (a) What is the value of the spread a of the classical probability density, (6.60)?
(b) What is the spreading time r of this distribution? Are your answers compatible
with the picture one has of a classical particle?
6.30 An electron is initially in the Gaussian state, (6.40). If a ~ Ae, the de Broglie wave-
length, what is the spread time r of the subsequent time development of the wave
packet? Is this a rapid or a slow spread?
6.31 To complete Ehrenfest's correspondence principle, one must convert (6.74) to the
classical relation
d(p) av(X)
---
dt a(x)
For what class of potentials would this relation be valid?

Answer
If V(x) is slowly varying. To show this, write
av
- == G(x)
ax
Then under the said condition we may expand

, (x-(x))21/
G(x) = G«(x)) + (x - (x ))G «(x)) + 2! G «(x)) + ...
186 Chapter 6 Time Development, Conservation Theorems, and Parity

Neglecting derivatives of G (second derivatives of V) gives the desired result:

av(x) aV«(x)
~ ~-----aw-

r
6.32 Let represent a symmetry operation such as translation, rotation, etc. Suppose a
given system with Hamiltonian H is invariant under an operation represented by i:
(a) What is the value of [H , h?
(b) If the eigenstates of Hare known, what can you say about the eigenstates of f?
(c) What can you say about the expectation, (I')?
r
(d) From descriptions given in this chapter, state four examples of and respective
r
systems on which operates for which your answers to parts (a), (b), and (c) are
valid.
6.33 In quantum mechanics, a process described by a wavefunction 1/f(x, t) is said to
be reversible if 1/f(x, t) = 1/f*(x , -t) . Show that solutions to the time-dependent
Schrodinger equation are reversible. (Alternatively, one says that the Schrodinger
equation is invariant under time reversal .)
Note: As described in Problem 1.24, time reversibility in classical physics in-
cludes the operation (t -+ t . P -+ - p) . In quantum mechanics, momentum, p , is re-
placed by the operator, -i Ii a/ ax . It follows that the classical prescription, p -+ - p,
may be effected in quantum mechanics through complex conjugation.
6.34 Consider the time-dependent Hamiltonian
, , de
H(x, t) = Ho(x) + I idt-
where

and e is a dimensionless c-number, which depends only on time.

(a) What are the time-dependent eigenfunctions, 1/fn(x, t), of H(x, t)?
(b) What is (E) in the state 1/fn(x, t)?
(c) What is the time dependence of energy for the related classical system?

Answers (partial)
(a) 1/fn(x, r) = exp{-i[wnt + e(t)]}<pn(x)
de
(b) (E) = En + lid(
6.35 Consider a point particle of mass m confined to the interior of a cubical box that
is centered at the origin of a Cartesian coordinate system with edges parallel to the
x, y, z axes, respectively. The edge lengths of the cube are a. Obtain the eigenfunc-
tions for this case and list these in parallel form to those of the one-dimensional box
given in (6.100).
CHAPTER
Additional One-Dimensional

7 Problems. Bound and

Unbound States

7.1 General Properties of the One-D imensional Schriidinger Equation

7.2 The Harmon ic Oscillator
7.3 Eigenfunctions ofthe Harmon ic Oscillator Hamiltonian
7.4 The Harmonic Oscillator in Momentum Space
7.5 Unbound States
7.6 One-Dimensional Barrier Problems
7.7 The Rectangular Barrier. Tunneling
7.8 The Ramsauer Effect
7.9 Kinetic Properties ofa Wave Packet Scattered from a Potential Barrier
7.10 The WKB Approximation
7.11 Principle ofLeast Action and Feynman's Path Integral Formulation

In this and the following chapters we examine some practical and fundamental
problems in one dimension. Included are the very important examples of the har-
mon ic oscillator and scattering configurations in one dimens ion. Creation and
annihilation operators are introduced in algebra ic construction of eigenenergies
of the harmonic oscillator. The purely quantum mechan ical effect of tunneling
is encountered in a study of transmission through a barrier. The chapter contin-
ues with a description of the WKB technique ofsolution appropriate in the near-
classical domain. This method of approximation find s application in still more
realistic configurations, such as cold emission from a metal surface and a de-
cay from a radioactive nucleus. A review of Hamilton's principal of least action
precedes a description ofFeynman 's path integral formalism.

7.1 • GENERAL PROPERTIES OF THE ONE-DIMENSIONAL

SCHRODINGER EQUATION

The time-independent Schrodinger equation for a particle of mass m moving in

one dimension in a potential field V (x) appears as

liZ iP
----z
[ 2m ax
+ Vex) ] <p(x) = E<p(x) (7.1)

187
188 Chapter 7 Additional One-D imensional Problems. Bound and Unbound States

Vex)

~-------~~-----E

II x

FIGURE 7.1 Domains relevant to a particle of energy E moving in a one-dimensional

potential field Vex) . I: E > V. Kinetic energy is positive. II: E < V . Kinetic energy is
negative ("forbidden domain") . III: E = V . This is a turning point of the corresponding
classical motion.

With subscripts denoting differentiation, this equation may be rewritten

I{Jxx = -k 2 (x )1{J
lih 2
-=E-V (7.2)
2m
The partition of energy

E=T+V (7.3)

permits us to identify li,2 k 212m as the kinetic energy of the particle

li,2 k 2
T=- (7.4)
2m
This identification is especially relevant if E > V . More generally, there are three
distinct possibilities (Fig. 7.1). These are E > V, E < V , and E = V . In the
first case the kinetic energy is positive and the corresponding classical motion is
permitted. Classical motion is forbidden in the second domain , where the kinetic
energy is negative. The points where E = V are the classical turning points.
(Recall Section 1.4.)
In the domain where the kinetic energy is negative, the Schri:idinger equation
becomes
2
I{Jxx = K (X)I{J
li,2K2
--=V-E>O (7.5)
2m
7.1 General Properties of the One-Dimensional Schr6dinger Equation 189

/i,2K2
Kinetic energy = - - - = E - V <
2m
°
Recall now that in analytic geometry, I{Jxx is related to the curvature of I{J (at
the point x) . If I{Jxx > 0, then I{J is concave upward. If I{Jxx < 0, then I{J is con-
cave downward. When the kinetic energy is positive, the Schrodinger equation
takes the form (7.2) and I{J has the following properties: I{Jxx is less than zero in

Nature of 'fJ
in upper and lower Oscillatory behavior
half-planes possible

r.
x x

(a)

J Nature Of 'fJ
in upper and
lower half-planes
Oscillatory behavior
not possible

x x

(b)

FIGURE 7.2 (a) Kinetic energy positive:

rpxx = _k 2rp k2 > 0

rpxx < 0 for rp > 0 (upper half-plane)

rpxx > 0 for rp < 0 (lower half-plane)

(b) Kinetic energy negative:

rpxx = K 2rp
rpxx > 0 for rp > 0 (upper half-plane)

rpxx <0 for sp < 0 (lower half-plane)

190 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

1--------+---£

Oscillation Decay
FIGURE 7.3 Characteristic behavior of wavefunction corresponding to the configuration
shown in Fig. 7.1.

the upper half-plane so f{J is concave downward; f{Jxx is greater than zero in the
lower half-plane, so f{J is concave upward. As shown in Fig. 7.2, these conditions
permit oscillating solutions. When the kinetic energy is negative, the Schrodinger
equation takes the form (7.5) and the following properties pertain: f{Jxx is greater
than zero in the upper half-plane so f{J is concave upward; f{Jxx is less than zero
in the lower half-plane and tp is concave downward. Again referring to Fig. 7.2,
these conditions are seen to lead to growing or decaying solutions (as opposed to
oscillating solutions) . At a turning point, f{Jxx = 0 and f{J has a constant slope.
For the potential shown in Fig. 7.1, one might then expect an eigenfunction of
the Hamiltonian to behave as depicted in Fig. 7.3.

PROBLEMS

7.1 (a) Let a particle of mass m move in a one-dimensional potential field with en-
ergy E as sketched in Fig. 1.18. Write down the form of the time-independent
Schrodinger equation [i.e., (7.2) or (7.5)] for the four domains that lie in the
interval 0 ~ x ~ D . In each case identify the wavenumber k or K .
(b) Given cp(O) = CPo > 0, make a rough sketch of cp(x) in the interval 0 ~ x ~ F.

7.2 • THE HARMONIC OSCILLATOR

The configuration of a harmonic oscillator is depicted in Fig. 7.4. The classical

equation of motion of a particle of mass m is given by Hooke's law,

(7.6)
7.2 The Harmonic Oscillator 191

x=o

FIGURE 7.4 The one-dimensional harmonic oscillator. Displacement from equilibrium

(x = 0) is denoted by x.

The spring constant is K. In terms of the natural frequency (vo,

K
(7.7)
m
the above equation appears as

(7.8)

Multiplying this equation by i gives

(7.9)

Integrating, one obtains the constant of motion

(7.10)

The potential energy is

K
V = _x 2 (7.11)
2
When the particle comes to rest, the energy is entirely potential.

K 2
(7.12)
E= -xo
2
Such points (xo) are turning points. For x 2 > xo2 , the kinetic energy T is negative,
so that classically this is a forbidden domain.
192 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

K
V=-x 2 V(x)
2

---~---+-----+---E

FIGURE 7.5 The turning points of the harmonic oscillator are at x = ± xo, where
Kxo2
--=E
2

K 2
T =E - V = -(xo
2
- x
2)
(7.13)
2 2)
T < 0 (for x > xo
See Fig. 7.5.
With these properties of the classical motion established, we tum next to the
quantum mechanical formulation of the harmonic oscillator problem. The Hamil-
tonian for a particle of mass m in the potential (7.11) is
p2 K
H=-+-x2 (7.14)
2m 2
The corresponding Schrodinger equation appears as

li,2 [Pcp K 2
- 2m ax2 + "2 x cp = Etp (7.15)

In the classically accessible domain, E > Kx 2 / 2, and this equation may be writ-
ten
CPxx = _k 2 cp
li,2k 2(x) K 2 (7.16)
---=E--x > 0
2m 2
The wavefunction cp is oscillatory in this domain.
In the classically forbidden domain where x 2 > xo 2, E < Kx 2/ 2 and the
Schrodinger equation becomes
2
CPxx = K cp
li,2K2 K 2
- - = -x -E >0 (7.17)
2m 2
7.2 The Harmonic Oscillator 193

so the wavefunction is nonoscillatory in this domain. In the asymptotic domain

Kx 2 / 2 » E, the Schrodinger equation becomes

mK 2 4 2
I{Jxx = IifX I{J = f3 x I{J (7.18)

where f3 is the characteristic wavenumber

(7.19)

In terms of the nondimensional displacement

~ == f3x (7.20)

(7.18) appears as

In the domain under consideration, ~ » 1 and the solution to the latter equation
appears as

I{J "" Aexp ±2 ( ~2) = [ (f3 X)2]

Aexp ±-2-

The growing solution (+) violates the normalization condition

i: 1{J*l{Jdx < 00 (7.21)

and one is left with the exponentially decaying wavefunction

I{J "" Aexp

(~2)
-2 [
= Aexp --2-
(f3 X)2] (7.22)

The character of the wavefunction changes from oscillatory for x 2 < xo2 to de-
caying for x 2 > xo 2 , so the turning points x = ±xo are also physically relevant
in quantum mechanics. These properties are depicted in Fig . 7.6.

Annihilation and Creation Operators

We tum to a general formulation of the solution to (7.15). The technique of solu-
tion we will develop is known as the algebraic method . It involves the operators

A
a
f3 (A
==.J2 x
iP) + mwo
(7.23)
aAt - f3 (Ax - -iP)
_ - -
.J2 mwo
194 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

FIGURE 7.6 Typical behavior of energy eigenfunction for the simple harmonic oscilla-
tor.

a a
Inasmuch as =1= at, is non-Hermitian. The properties that these operators have
are determined through the fundamental commutator relation

[x ,p] = iii (7.24)

For instance , it is readily shown that (see Problem 7.5)

[a, at] = I
aat = 1 +ata (7.25)

With the aid of the inverse of (7.23),

X= -
a+a t
- ,
A mwo a- at (7.26)
P = - i - .j2f3
.j2f3
the Hamiltonian for the harmonic oscillator becomes

(7.27)

In this manner we see that the problem of finding the eigenvalues of fj has been
transformed to that of finding the eigenvalues of the operator

(7.28)

Let <{In be the eigenfunction of N corresponding to the eigenvalue n, so that

(7.29)

(We do not assume that n is an integer at this point. This property is established
later.) Consider the effect of operating on aePn with N.
7.2 The Harmonic Oscillator 195

Nal{Jn = ataal{Jn = (aa t - l)al{Jn = a(ata - l)l{Jn

(7.30)
Nal{Jn = a(N - l)l{Jn = a(n - l)l{Jn = (n - l)al{Jn
It follows that al{Jn is the eigenfunction of N which corresponds to the eigenvalue
n - 1. That is (apart from normalization factors),

al{Jn = I{Jn-! (7.31)

Similarly,

(7.32)

a
and so forth. Because of this property, is called an annihilation or stepdown or
demotion operator.
In similar manner, if we consider the operation Nat I{Jn, there results

(7.33)

This equation implies that atI{Jn is the eigenfunction of N corresponding to the

eigenvalue n + I.

(7.34)

Similarly,

(7.35)

and so forth. The operator at is called a creation or stepup or promotion operator

(Fig. 7.7).
Since the Hamiltonian for the harmonic oscillator is the sum of the squares of
two Hermitian operators,

(H) 2: 0 (7.36)

--------
(see Problem 4.13). In the eigenstate I{Jn,

·l .
'fin + I

at J --------
(a)
'fin

(b)

FIGURE 7.7 Schematic representation of the raising and lowering operators at and a.
196 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

Hrpn = lUvo ( it + ~) rpn = lUvo (n + ~) rpn

(7.37)
(rpnIHrpn) = lUvo (n +~) : : 0
This implies that the eigenvalues n must obey the condition

I
n > -- (7.38)
- 2

That is, all eigenstates of H, or equivalently it, corresponding to eigenvalues

n < -! must vanish identically. For the harmonic oscillator such states do not
exist. This condition is guaranteed if we set

arpo = 0 (7.39)

With (7.31) we obtain

arpo = rp-I = 0
(7.40)
a(rp-d = rp-2 = 0
As will be shown, (7.39) has a nontrivial (i.e., other than zero) solution for rpo.
Furthermore,

(7.41)

and we may conclude that the eigenvalue of it corresponding to the eigenfunction

rpo is zero. It follows that

itatrpo = ataatrpo = at(ata + l)rpo = atrpo

(7.42)
itatrpo = la trpo = rpl
The eigenvalue of it corresponding to rp\ is the integer 1. This construction [same
as in (7.34) et seq.] allows one to conclude that the index n, which labels the
eigenfunction rpn , is indeed an integer.
Repeating (7.37),

Hrpn = lUvo (n + ~) rpn (7.43)

one finds that the energy eigenvalues of the simple harmonic oscillator are

(n = 0, 1, 2, . . . ) (7.44)
7.2 The Harmonic Oscillator 197

5 11
h
zOOo

4 9
Zhroo

2
5
Zhroo

3
Zhroo

o 1
Zhroo - - - - - Lowestenergy of harmonic
--- E 0 oscillator Eo thOOo
zero-pointenergy

FIGURE 7.8 The energy levels of the simple harmonic oscillator are equally spaced.

The energy levels are equally spaced by the intervallUuo (Fig. 7.8) . If a molecule,
for example Hel, which resembles a dumbbell, has vibrational modes of excita-
tion (the arm of the dumbbell acts as a spring), the Bohr frequencies emitted by
the molecule fall in the scheme

hv = En' - En = lUuo (n' + ~) -lUuo (n + ~) (7.45)

= lUuo(n' - n) = lUuos
v = svo, Wo == 2nvo
In the latter sequence of equations, nand n' are integers, so their difference,
s, is also an integer. It follows that the frequencies emitted by a vibrational di-
atomic molecule are integral multiples of the natural frequency of the molecule,
Vo (Fig. 7.9) .

o
IIII D
-------I.~ V

FIGURE 7.9 Spectrum of a vibrational diatomic molecule. (See Problem 10.59.)

198 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

PROBLEMS

7.2 A harmonic oscillator consists of a mass of 1 g on a spring. Its frequency is 1 Hz and

the mass passes through the equilibrium position with a velocity of 10 cmls. What
is the order of magnitude of the quantum number associated with the energy of the
system ?

7.3 The spacing between vibrational levels of the CO molecule is 2170 cm-1 . Taking the
mass of C to be 12 amu and 0 to be 16 amu, compute the effective spring constant
K, which is a measure of the bond stiffness between the atoms of the molecule . [Hint:
The mass that enters is the reduced mass, mMI(m + M) . The spacing between lines
is given in terms of wavenumber k = 2rr l A, where co = ck (c = the speed of light),
so that 6.w = c 6.k .]

7.4 The derivation in the text of the eigenvalues of N is based on the constraint that there
are no states corresponding to the eigenvalues n < -!.This constraint was guaranteed
by setting alfJO = O. It would appear that it can also be guaranteed by setting alfJl/2 =
0, for in this case

alfJl/2 = lfJ-l /2 = 0
Show that IfJl /2 as defined is an eigenfunction of N with the eigenvalue zero; hence
IfJl /2 is more properly termed IfJO.

7.5 Using the fundamental commutator relation

[i,p]=ih

show that

7.3 • EIGENFUNCTIONS OF THE HARMONIC OSCILLATOR HAMILTONIAN

When written in terms of the nondimensional displacement ~ (7.20),

(7.46)

the operators aand at become

(7.47)
7.3 Eigenfunctions of the Harmonic Oscillator Hamiltonian 199

The time-independent Schrodinger equation becomes

At A 2£ ) ( 2£
( 2il a+1- liwo CP=CPH+ Iiwo -~
2) cP=o (7.48)

The ground-state wavefunction CPO of the simple harmonic oscillator Hamilto-

nian obeys (7.39)

acpO = 0

or, equivalently,

_1
.j2
(~+~)
a~
CPO = 0 (7.49)

This has the solution

(7.50)

The requirement that cpo(~) be normalized implies that

00 100
1-00 CPO d~ -00 e-~ d~
2 2 2 2
1= 1 1 = Ao = .j7i Ao
(7.51)
Ao = ]'[-1 /4

so

(7.52)

In terms of the dimensional displacement x, the normalized ground state is

(7.53)

Normalization (with respect to x) gives

(7.54)

The ground state CPO is a purely exponentially decaying wavefunction. It has no

oscillatory component. The higher-energy eigenstates, on the other hand, will be
found to oscillate in the classically allowed domain and decay exponentially in
the classically forbidden domain .
With CPO given by (7.52), the remaining normalized eigenstates of the harmonic
oscillator Hamiltonian are generated with the aid of the creation operator at, in
200 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

the following manner:

!PI = at rpo
I At I At 2
!P2 = - a !PI = - ( a ) !PO (7.55)
.j2 .j2
!Pn = _I_(a t)n!po
JnT
With at written in terms of~, as in (7.47), the equation for !PI above becomes
!PI = Al (~ - :~) e-~2/2
(7.56)
!PI = AI2~e-~2/2

Al = (2,J]i)-1 /2
where Al is the normalization constant of !PI. The nth eigenstate is given by the
fonnula

(7.57)

The nth-order differential operator (at)n, when acting on the exponential form
exp( _~2 /2) , reproduces the same exponential factor, multiplied by an nth-order
polynomial in ~ .

(7.58)

Thus the nth eigenstate of the simple harmonic oscillator Hamiltonian may be
written together with its eigenvalue as

!Pn = AnJfn(~)e-~2/2
(7.59)
En = lUvo (n + ~)
The nth-order polynomials Jfn(~) are well-known functions in mathematical
physics. They are called Hermite polynomials. From (7.56) we see that Jfl = 2~ .
The first six Hermite polynomials are listed in Table 7.1.
The nth-order Hermite polynomial Jfn enters in the eigenfunctions !Pn of the
quantum mechanical harmonic oscillator as
7.3 Eigenfunctions of the Harmonic Oscillator Hamiltonian 201

TABLE 7.1 The first six eigenenergies and eigenstates of the simple
harmonic oscillator Hamiltonian

o haiO/2 Aoe-~2 /2
1 3haio/2 AI2~e-~2/2
2 5haio /2 A2(4~2 - 2)e-~2 /2
3 7ha>o /2 A3(8~3 - 12~)e-~2/2
4 9haio /2 A4(l6~4 - 48~2 + 12)e-~2 /2
5 llhaiO/2 A5(32~5 - 160~3 + 120~)e-~2 /2
An = (2nn!J1f)-1 /2

Jfn is a solution to Hermite's equation,

The formulas connecting fPn and fPn+l (see Problem 7.9) are very useful in
many problems relating to the simple harmonic oscillator. In Dirac notation they
appear as
1 2IfPn_d
alfPn) = n /
(7.60)
atlfPn) = (n + 1)1/2 IfPn+d
In place of IfPn), let us write the ket vector In) . In this notation the equations above
appear as

aln) = n 1/ 21n - 1)
(7.61)
atln) = (n + 1)1/2In + 1)
Let us check that

Min) = nln) (7.62)

With the aid of (7.61), we obtain

ataJn) = at n 1/ 2Jn - 1) = n 1/ 2n1/ 2In)

(7.63)
ataln) = Min) = nln)

Inasmuch as {fPn} are normalized and are eigenstates of a Hermitian operator,

they comprise an orthonormal sequence.

i: fPn *fPI d~ = (nil) = Onl (7.64)

202 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

To gain familiarity with the manner in which these concepts are used in problems,
we will work out a few illustrative examples.
First, consider the question: What is (x) in the nth eigenstate fPn? Here we must
calculate

(x) = (nl xln) (7.65)

I ~ ~t
= h (nla +a In)
v2f3
I
= _ _ {n 1/ 2(nln - I) + (n + l)1 /2(nln + I)}
./if3
=0
The last step follows from the orthogonality relation (7.64) . The fact that the av-
erage value of x in any eigenstate fPn vanishes is a consequence of the symmetry
of the probability density P = IfPn 12 about the origin (see Fig. 7.10).
The second example we consider is the expectation of momentum p, in the nth
eigenstate fPn .
~ mwo ~ ~t
(p) = (nlpln) = h (nla - a In) (7.66)
v2if3
= mwo {n 1/ 2(nln _ I) _ (n + 1)1/2(nln + I)}
./i if3
=0
In any eigenstate fPn of the Hamiltonian of the simple harmonic oscillator, the
probability of finding the particle with momentum hk is equal to that of finding
the particle with momentum -lik. Were we to express fPn(x) as a superposition of
momentum eigenstates, exp(ikx), we would find the probability amplitude b(k)
to be an even (symmetric) function of k [i.e., b(k) = b( -k») .

Correspondence Principle
Next , we consider the manner in which the solution to the quantum mechan-
ical harmonic oscillator problem obeys the correspondence principle To these
ends let us calculate the classical probability density P , corresponding to a one-
dimensional spring with natural frequency WOo Let the particle be at the origin at
t = 0 with velocity xowo. The displacement at the time t is then given by

x = xo sin(wot)
(7.67)
x= xowo cos(wot)

This gives the correct initial data

x(O) = 0
(7.68)
x(O) = xowo
 0.75 0.5
N

0.60 0.4
~
<Po 0.45 0.3
III
0.30 Q.,0 0.2
0.15 0.1
0.0 0

0.55 0.40
N
- 0.32
0.30 .;'
0.05 III 0.24
<PI
-0.20 a: 0.16
- 0.45 0.08
-0.70 0

0.6
0.4 0.40
'"- 0.32
0.2 s,
N

<P2 0.0 0.24

III
-0.2 r{,' 0.16
-0.4 0.08
-0.6 0

0.5 OAO
0.3
'"- 0.32
.;
0.1 III 0.24
<P4
- 0.1 Q.,'" 0.16
-0.3 0.08
-0.5 0

0.6
0.4 0.40
'"-
0.2
c 0.32
<P5 0.0 III 0.24
-0.2 Q.,'" 0.16
-0.4 0.08
-0.6 0
-6.0 -3 .0 0.0 3.0 6.0 t -6.0 -3 .0 0.0

FIGURE 7.10 The first few eigenstates of the simple harmonic oscillator and corre-
sponding probability densities. Turning points, ~o(n) = ,J I + 2n , are denoted by vertical
marks.
204 Chapter 7 Add itional One-Dimensional Problems. Bound and Unbound States

The product P(x) dx is the probability of finding the particle in the interval dx
about the point x at any time. If To is the period of oscillation

2rr
To=- (7.69)
Wo
then
dt wodt
Pdx=-=-- (7.70)
To 2rr

where

dt = d: (7.71)
x
Using (7.67), one obtains

(7.72)

so that
wo dx
P dx = -dt = ----,===;;==:;c (7.73)
2rr 2rr xo2 - x 2 J
The probability density so found is normalized with respect to the angular dis-
placement dB = Wo dt , 0 S B S 2rr . The interval in displacement x is one-half as
long, so the properly normalized P function , over the interval -xo < x < +xo,
IS

P = -r===;:====
2 2
rrJxo - x

t:
- xo
P(x) dx = I
(7.74)

This function is sketched in Fig. 7.11, where it is superimposed on the quantum

mechanical probability density corresponding to a state with n » 1. The singu-
larities in P at the turning points ±xo are due to the fact that the particle comes
to rest at these points.
The correspondence which the quantum mechanical formulation displays in
the present case is clearly exhibited in Fig. 7.11, where we see that

lim (Pn QM) = peL (7.75)

n-+oo
7.3 Eigenfunctions of the Harmonic Oscillator Hamiltonian 205

FIGURE 7.11 Classical probability density

pCL = 1
rrJ x o2 - x 2
superimposed on the quantum mechanical probability density

For the case n » I,

lim (Pn QM) = pCL

n~ oo

The superscripts QM and CL denote quantum mechanical and classical , respec-

tively, and E is an arbitrarily small interval. The integral above is called a local
average. It represents the average of pQM in a small interval centered at x.
The classical configuration that corresponds to the quantum state in which a set
of commuting observables are specified is the configuration which includes these
same parameters as constant and known . Thus, in the problem of a particle con-
fined to a one-dimensional box, considered in Chapter 4, when one concludes that
the classical probability density is uniform, it should be noted that this is the case
206 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

provided that all one knows about the particle is its energy. The classical state
of this system permits a more elaborate description . Unlike the quantum case,
for the classical particle one may specify both its energy and position in time,
x(t) . Given this maximally informative classical description, the configurational
probability density becomes 8[x (t) - x]. When one speaks of configurational cor-
respondence in the limit of high quantum numbers, what is usually meant is that
in this limit the quantum probability density goes to the classical probability den-
sity in which, consistent with the quantum description, not more than the energy
is specified.
In our consideration of correspondence for the harmonic oscillator, this rule
is again obeyed. The expression (7.74) for the classical probability density is
relevant to the case where only the amplitude xo, or equivalently the energy
E = Kxo2/2, is known. The quantum density sketched in Fig. 7.11 is likewise
connected to the energy eigenstate CPn for which measurement of energy finds
with certainty the value En.

PROBLEMS

7.6 (a) Show that the Hermite polynomials generated in the Taylor series expansion

are the same as generated in (7.58).

(b) Show that Jfn as generated by

are equivalent to those given by (7.58).

(c) Use any of the preceding relations to establish

Jfn' = 2nJfn_l

(d) and the recursion relation

(e) Use the generating formula of part (b) to find JfO(~) , Jfl (~), and Jf2(~) '

7.7 The general formula for the normalization constant of f{Jn is

Show that this gives correct normalization for f{J4 .

Problems 207

7.8 Show directly from the form of f{Jn given by (7.57) that

Pf{Jn = (- )nf{Jn
where P is the parity operator.
7.9 (a) Show that the normalized nth eigenstate f{Jn is generated from the normalized
ground state f{JO through

1 't n
f{Jn = r.:;(a) f{JO
'In!

(b) Show that part (a) implies the following relations.

where f{Jn, f{Jn-l> and f{Jn+! are normalized.

7.10 Show that in the nth eigenstate of the harmonic oscillator, the average kinetic energy
(T) is equal to the average potential energy (V) (the virial theorem) . That is,

(V) K 2) = (T) = _(p2)

= _(x 1 1
= -(E) = -fiwO (n +-1)
2 2m 2 2 2

Answer (partial)

~(X2) = (4~2) (nl(a+a t)2 In)

= (~)
4{32
(nla 2 + at2 + (aa t + ata)ln)

= (4~2){0+0+(nl(l+2N)ln)}
fiwo fiwo ( 1)
(V) = 4(1+2n) =2 n+2"

7.11 In Problem 7.2, what is the average spacing (in em) between zeros of an eigenstate
with such a quantum number?

7.12 A harmonic oscillator is in the initial state

1{!(x,O) = f{Jn(x)
that is, an eigenstate of if. What is 1{!n (x, t)?

7.13 For a harmonic oscillator in the superposition state

1
1{!(x, t) = .j2[1{!O(X, t) + 1{!! (x, t)]
208 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

show that

(x) = C cos(wot)
In the notation above,

7.14 Show that in the nth state of the harmonic oscillator,

(x 2) = (L\x)2
(p2) = (L\p)2
7.15 Find (x) for a harmonic oscillator in the superposition state

1
1/!(x, t) = .j2[1/!o(x, t) + 1/!3(X, t)]
The harmonic oscillator has natural frequency woo
7.16 A large dielectric cube with edge length a is uniformly charged throughout its volume
so that it carries a total charge Q. It fills the space between condenser plates, which
have a potential difference <1>0 across them. An electron is free to move in a small
canal drilled in the dielectric normal to the plates (Fig . 7.12).
The Hamiltonian for the electron is (with x measured from the center of the canal
and e written for -leI)

Unifonnly charged dielectric of

total charge Q fills space
~---=r---I----~ between
condenser
platesP

Electron

cr::-::}c:=
Canal through
center of
dielectric

<1>0

FIGURE 7.12 Configuration described in Problem 7.16 .

Problems 209

~ jJ2 Kx 2 e<l>o
H=-+-+-x
2m 2 a

(a) What is the spring constant K in terms of the total charge Q?

(b) What are eigenenergies and eigenfunctions of if? [Hint: Rewrite the potential
energy of the electron as

K 2 K
V = 2(x + 2yx) = 2[(x + y)2 _ y2]

e<l>o
y=-
aK
then change variables to z =x + y . To evaluate K, use Gauss's law (neglecting
"edge effects") .]

7.17 (a) Show that the time-independent Schrodinger equation for the harmonic oscilla-
tor, with the energy eigenvalues (7.44), may be written

(b) Using the relations of Problem 7.6, show that

is a solution to this equation .

(c) Obtain Hermite's equation

7.18 Use the uncertainty principle between x and p to derive the "zero-point" energy

1
EO = 2/iwo
of a harmonic oscillator with natural frequency W() (see Fig. 7.8).
7.19 Show that

in the nondimensional ~ notation.

7.20 Show that the asymptotic exponential behavior of ipn (n
agrees with that obtained
directly from the Schrodinger equation, in the limit that ~ --+ 00.
7.21 Show that

a
( ~ + a~
)t = ~ - a~
a

in 5)2 [see (4.31)].

210 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

7.22 (a) What is the asymptotic solution rpn to the Schrodinger equation (as given in Prob-
lem 7.17)

in the domain

~2 « I + 2n ~ 2n?
(b) Show that

lim (Pn ) = (IrPnI 2) = constant

n»1

Answer
In this domain the Schrodinger equation above becomes

which has the (even) solution

rPn = C cos(.J2,;~)

It follows that the local average of IrPn 12 in this domain is given by

This result explains the flatness of (pQM) in the central domain ~2 «2n , as seen in
Fig. 7.11.
7.23 Estimate the length of interval about x = 0 which corresponds to the classically
allowed domain for the ground state of the simple harmonic oscillator.

Answer
The turning points occur at

~ =±1 or equivalently at x = ±j Ii
mwO

At this value.lrpol2 is e- 1 times smaller than its value at the origin (Fig. 7.10).
7.24 Show that the nth stationary state In) of a harmonic oscillator with fundamental fre-
quency wo,

6.p 6.x = -En = Ii ( n + -

wo
1)
2
 7.4 The Harmonic Oscillator in Momentum Space 211

7.4 • THE HARMONIC OSCILLATOR IN MOMENTUM SPACE

Representations in Quantum Mechanics

Let us recall (4.42) et seq., which relate the wavefunction rp(x) to the momentum
coefficient b(k).

rp(x) = i: b(k)rpk dk

b(k) = i: rp(X)rpk * dx
(7.76)

The eigenfunction of momentum corresponding to the value p = hk is rpk . The

wavefunction rp(x) gives the probability density in coordinate space through the
Born relation

P(x) = Irp(x)!z (7.77)

The momentum coefficient b(k) gives the probability density [probability of find-
ing the particle to have momentum in the intervallik to liCk + dk)] in momentum
(k) space through the relation

P(k) = Ib(k)l z (7.78)

The integral formulas (7.76) serve to determine rp(x) given b(k), and vice versa. It
follows that any information contained in rp(x) can be obtained from knowledge
of b(k) and vice versa. Given the Hamiltonian of a system, rp(x) is determined. Let
us construct an equation which similarly determines b(~) from the Hamiltonian
[i.e., without first finding rp(x)]. To these ends we first recall the time-independent
Schrodinger equation for the harmonic oscillator.

Z
liZ cj2 Kx )
(
---- z+- rp(x) = Erp(x) (7.79)
2m ox 2

Substituting the Fourier decomposition of rp(x) above and noting the equality,

(7.80)

gives

1 00

-00
dkb(k) (liZkZ
-
2m
K -oZ
- - z ) rpk = E
2 ok
1
00

-00
dkbikyp; (7.81)

Integrating the second term on the left-hand side by parts twice and setting

(7.82)
212 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

gives

[(n?k2 2

1 00
-00
dkrpk - -
2m
K-a2 - E ) b(k) ] =0
-
2 ak
(7.83)

It follows that the term in brackets is the Fourier transform of zero, which is
zero. We conclude that b(k) (appropriate to the harmonic oscillator) satisfies the
k-dependent Schrodinger equation

n?k2- K a )
2

( 2m
- - -2 b(k) = Eb(k)
2 ak
(7.84)

This equation is also called the Schrodinger equation in momentum representa-

tion .
We note that the Hamiltonian in momentum representation includes the
simple multiplicative operator hk in place of p and the differential operator
+iajak in place of x . This rule for obtaining the structure of the Hamiltonian
in momentum representation always holds providing the potential Vex) is an
analytic function I of x (i.e., has a well-defined power-series expansion). For
such cases the Schrodinger equation in either coordinate or momentum space is
obtained through the recipes :

In x-space: A p) ~ HA(x, - ilia)

H(x, ax

In p-space: A
H(x,p)~H
ia )
A(-ak,lik

The time-dependent Schrodinger equation in momentum representation appears

as

a A

ili-b(k, t) = H(k)b(k, t) (7.85)

at
Paralleling the development of (3.70) permits the solution to (7.85) for the initial-
value problem for b(k, t) to be written

b(k, t) = exp ( -n
itf!) b(k,O) (7.86)

For free-particle motion with f! = li 2k 2 j2m, the latter relation gives

I In the more general case the Schrodinger equation in momentum space becomes an integral equation.
These topics are discussed in greater detail in E. Merzbacher, Quantum Mechanics , 2d ed., Wiley, New
York, 1970.
7.4 The Harmonic Oscillator in Momentum Space 213

Ib(k, t) 12 = Ib(k, 0) 12 (7.87)

The momentum probability density for free-particle motion is constant in time.

Geometrically, the function b(k) is the projection of the state qJ(x) onto the
momentum eigenstate qJk [recall (4.44)] .

(7.88)

For any given state qJ(x), the function b(k) represents a distribution of values,
corresponding to the projections of qJ(x) onto the set of basis vectors {qJk} . The
functions b(k) and qJ(x) are equally informative. In momentum representation a
state of the system is represented by its projections onto the basis of Hilbert space
{qJd. (See Fig. 4.6.)
This is analogous to the statement that a vector B in 3-space is represented by
its projections onto the three unit vectors ex, e y, and ez, namely, Bx , By, and B z .
These are not the only basis vectors one can use to represent the vector B. For
instance, one can employ the basis ex', ey', and ez' given by

(7.89)

(See Fig. 7.13.) In this basis B is represented by the three components

FIGURE 7.13 Projections onto two sets of basis vectors (in the x y plane) of the vector
B. The two bases are related through (7.89). The z component is the same in both repre-
sentations.
214 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

There are countless other triads of unit vectors which are valid bases of 3-space
(i.e., they all span 3-space) . The three components of B in anyone of these repre-
sentations completely specify B.
Similarly, one may describe the state of a system in quantum mechanics in
different representations. In each of these, a distinct set of vectors serves as a
basis of Hilbert space. Particularly important in the theory of representations is the
concept of common eigenfunctions of some complete set ofcommuting operators
relevant to a given system . Suppose that the complete set of commuting operators
are A, E, and C. In the state ({Jabc (common eigenstates of A, E, and C) one
may specify the "good" quantum numbers a, b, and c. The state of the system
cannot be further resolved. Such states may serve as a basis of Hilbert space. The
representation in which all states are referred to the basis {({Jabc} is called the abc
representation, just as we call the representation in which states are referred to the
eigenfunctions of momentum, the momentum representation. t One also speaks of
the abc representation as the one in which A, E, and are diagonal. c
PROBLEMS

7.25 Show that b(k) is even for any even potential V (x) = V ( -x) . What can be concluded
about the oddness or evenness of b(k) if V (x) is an odd function, V (x) = - V ( - x) ?
7.26 What are the eigenfunctions bn (k) of the harmonic oscillator Hamiltonian H(k) in
momentum space [as given by (7.84))? [Hint: Note the similarity between H(k) and
H(x), and the boundary conditions of rpn(x) and bn(k) .)

7.27 What is the Schrodinger equation in momentum representation for a free particle
moving in one dimension ? What are the eigenfunctions b(k) of this equation?
7.28 Consider the Gaussian wave packet whose initial momentum probability density is
given by (6.45).
(a) What is Ib(k, t)12 at t > O?
(b) What is ~x Sp at t > O?
7.29 Consider an arbitrary differentiable function of p, rp(p) . Show that with p = p and
x
= +ili%p,

[x. p)rp(p) = i 1irp(p)

x,
7.30 What is the eigenfunction of the operator in the momentum representation, corre-
sponding to the eigenvalue x ? That is, give the solution to the equation

x
7.31 Let Ix') denote an eigenvector of the position operator with eigenvalue x' and let
Ik') denote an eigenvector of the momentum operator p with eigenvalue M ' . Show

2For further discussion of the coordinateand momentum representations. see Appendix A.

Problems 215

that
(a) (klk') = 8(k - k')
(b) (xix') = 8(x - x')
1
(c) (xlk) = r.:c exp(ikx)
v2Jr
For each case, state in which representation you are working.
7.32 Suppose that the operators a and at in

obey the anticommutation relation

(a) What are the values of aln) and atln) that follow from the anticommutation
relation above?
(b) Since (H) ~ 0, for consistency we may again set

alO) =0

Combining this fact with your answer to part (a), which are the only nonvanish-
ing states In)?
(c) If, in addition to the anticommutation property above, a and at also obey the
relations, {a, a} = {at, at} = 0, show that i1 2 = i1.

Answers (partial)
(a) aln) = v'nll - n)
atln) = .JI=n11 - n)
(b) The only nonvanishing states are 10) and 11) . [Note: Anticommutation relations
between a and a t are used to describe particles that obey the Pauli exclusion
principle/' In this context the operator N denotes the number of particles in a
given state so that (b) implies that there is no more than one particle in any
state. The 10) state is called the vacuum state. The formalism is known as second
quantization. 4 ]

7.33 What is the lowest value of kinetic energy (T) a harmonic oscillator with frequency
Wo can have?
Answer
In Problem 7.10 we found that in the nth eigenstate of the oscillator

(V) = (T) = 2hwo (n + 1)

2 ~ 4hwo
Thus, the lowest allowed energy of the oscillator is hwO/2. It is impossible to force
the oscillator to a lower energy. In a solid, for example, whose nuclei are bound

3A formal statement of this principle is given in Chapter 12. See also Appendix B.
4Second quantization is encountered again in Problem 13.37.
216 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

together by harmonic forces, this zero-point energy persists at 0 K. (See Problem

7.18.)
(
7.5 • UNBOUND STATES

If a wavefunction 1/1 represents a bound state (in one dimension), then

Ixl-+ 00 (7.90)

for all t . A wavefunction that does not obey this condition represents an unbound
state. The square modulus of a bound state gives a finite integral over the infinite
interval.

(7.91)

The square modulus of an unbound state gives a finite integral over any finite
interval.

Ib -al < 00 (7.92)

The eigenstate of the momentum operator

(7.93)

represents an unbound state . The eigenfunction of the simple harmonic oscillator

Hamiltonian

(7.94)

(see Section 7.3) represents a bound state . Unbound states are relevant to scatter-
ing problems. Such problems characteristically involve a beam of particles which
is incident on a potential barrier (Fig . 7.14) .
Since f~oo 11/11 2 dx diverges for unbound states, it is convenient to normalize
the wavefunction for scattering problems in terms of the particle density p. For
one-dimensional scattering problems we take

11/I1 2 d x = p dx = dN
= number of particles in the interval dx

l b
2
11/11 dx = N
(7.95)

= number of particles in the interval (b - a)

7.5 Unbound States 217
V(x)

-00 x
-
J trans

+00
V(x) = constant = 0 V(x) = constant = V

FIGURE 7.14 One-dimensional scattering problem. Incident particle current .lfinc initi-
ated at x = -00 is partially transmitted (.lftrans) and partially reflected (.lfrefl) by a potential
barrier Vex) . The potential is constant outside the scattering domain.

For a one-dimensional beam of 103 neutrons/em, all moving with momentum

p = fiko, the wavefunction is written
1f! = 103 j2ei(kox-wt) , 11f!1 2 = 103 cm:'
fi2k02 (7.96)
--=ruv
2m

The sole difference between 1f! so defined and a wavefunction whose square mod-
ulus is probability density is a multiplicative constant. It follows that 11f!1 2 , when
referred to particle density, is proportional to probability density also. For uni-
form beams, 11f!1 2 is constant, which in tum implies that it is uniformly probable
to find particles anywhere along the beam. This is consistent with the uncertainty
principle. For instance, for the wavefunction (7.96), the momentum of any neutron
in the beam is Mo, whence its position is maximally uncertain.

Continuity Equation
One-dimensional barrier problems involve incident, reflected, and transmitted
current densities, .lfinc, .lfrefl, and .lftrans, respectively. In three dimensions the num-
ber density and current density J are related through the continuity equation
ap
-+V·J=O (7.97)
at
To clarify the physical meaning of this equation, we integrate it over a volume V
and obtain

aN = _ {J - d S (7.98)
at 1s
The total number of particles in the volume V is

N= !vPdr (7.99)
218 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

dS

FIGURE 7.15 Geometry relevant to integration of the continuity equation.

(Gauss's theorem was used to transform the divergence term.) The surface S en-
closes the volume V (Fig. 7.15). Equation (7.98) says that the number of particles
in the volume V changes by virtue of a net flux of particles out of (or into) the
volume V. It is a statement of the conservation of matter because it says that this
is the only way the total number of particles in V can change. If particles are
born spontaneously in V with no net flux of particles through the surface S, then
aN jat > 0, while f J · dS == 0 and (7.98) is violated.
If particles are moving only in the x direction,

(7.100)

and the continuity equation becomes

ap + aJJx == 0 (7.101)
at ax
We already have identified p with IVr 12 . To relate JJ x to 1/1, we must construct an
equation that looks identical to (7.101) with 11/1 12 in place of p. Then the functional
of 1/1 which appears after ajax is JJ x .
The wavefunction for particles in the beam obeys the Schrodinger equation
a.. /I'* •
_\f_ == +~ fI 1/1* (7.102)
at n
The time derivative of the particle density 1/1*1/1 is

- - == Vr *-a 1/1 + 1/1-

a1/l*Vr
at at at
(-infI)
a 1/1* == Vr * -1/1 + Vr -1/1 *) (+infI (7.103)

For the typical one-dimensional Hamiltonian

fI == L "'2

2m
+ Vex) (7.104)

the latter equation becomes

a1/1*1/1 in * *
-a-t- == 2m (1/1 1/Ixx - 1/11/1 xx)
(7.105)
a1/1*1/1 + ~ [~ (1/1* eN! _ 1/1 a1/1* )] == 0
at ax 2mz ax ax
7.5 Unbound States 219

(The subscript x denotes differentiation.) Comparison of this equation with

(7.101) permits the identification

(7.106)

Note that the dimensions of .JJ x are number per second. In three dimensions the
current density is written

J= ~(1fr*V1fr -1frV1fr*) (7.107)

2ml

and has dimensions cm- 2 s-I .

Transmission and Reflection Coefficients

For one-dimensional scattering problems , the particles in the beam are in plane-
wave states with definite momentum. Given the wavefunctions relevant to in-
cident, reflected, and transmitted beams, one may calculate the corresponding
current densities according to (7.106). The transmission coefficient T and reflec-
tion coefficient R are defined as

(7.108)

These one-dimensional barrier problems are closely akin to problems on the trans-
mission and reflection of electromagnetic plane waves through media of varying
index of refraction (see Fig. 7.16). In the quantum mechanical case, the scattering
is also of waves.
For one-dimensional barrier problems there are three pertinent beams . Particles
in the incident beam have momentum

Pine = tiki (7.109)

Particles in the reflected beam have the opposite momentum

Prefl = -tiki (7.110)

In the event that the environment (i.e., the potential) in the domain of the trans-
mitted beam (x = +(0) is different from that of the incident beam (x = -(0),
the momenta in these two domains will differ. Particles in the transmitted beam
will have momentum tik2 f= tiki,

Ptrans = tik2 (7.111)

In all cases the potential is constant in the domains of the incident and trans-
mitted beams (see Fig. 7.14), so the wavefunctions in these domains describe free
220 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

n n'

2
a & _
ax 2
(!!)c 2 aa(22 & =0
(a)

Surface
of
constant
phase k o
--f--,~ 1\
C
s
'"
c
o
u
II
::..

v=o
1/1 = Aei(kx - WI)

(b)

FIGURE 7.16 (a) Scattering of plane electromagnetic waves through domains of dif-
ferent index of refraction n . (b) Scattering of plane, free-particle wavefunctions through
domains of different potential.
Problems 221

particles, and we may write

li2kl2
!iwl = Einc = - -
2m
!iwl = Erefl = Einc
(7.112)
li2k22
!iw2 = Etrans = --
2m
+V
= Einc =!iwl
Energy is conserved across the potential hill so that frequency remains constant
(WI = W2). The change in wavenumber k corresponds to changes in momentum
and kinetic energy. Using (7.106) permits calculation of the currents

Ii , 2
.Jfinc = -.2lktlAI
2ml
Ii. 2
.Jftrans = - .2lk21C1 (7.113)
2ml
Ii . 2
.Jfrefl = - ,2lkllBI
2ml
It should be noted that these relations are equivalent to the classical prescription
for particle current, .Jf = pv, with p = 11/11 2 and v = hkfm , These formulas,
together with (7.108), give the T and R coefficients

(7.114)

In the event that the potentials in domains of incident and transmitted beams are
equal, kl = k2 and T = rc/ A 12 . More generally, to calculate C/ A and B/ A as
functions of the parameters of the scattering experiment (namely, incident energy,
structure of potential barrier), one must solve the Schrodinger equation across the
domain of the potential barrier.

PROBLEMS

7.34 Show that the current density J may be written

J= 2~ [1/J*p1/J + (1/J*p1/J)*]
where p is the momentum operator,
7.35 Show that for a one-dimensional wavefunction of the form [where ep(x , r) is real]

1/J(x, t) = Aexp[iep(x, t)],

Ii 2aep
]=-IAI -
m ax
222 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

7.36 Show that for a wave packet 1fr(x, t), one may write

1
1 00 ]
-00
dx = -2
m
«(p) + (p)*) = i!!l
m

7.37 Show that a complex potential function, V*(x) f= Vex), contradicts the continuity
equation (7.97).
7.38 (a) Show that if 1fr(x, t) is real, then

]=0

for all x .
(b) What type of wave structure does a real state function correspond to?

7.6 • ONE-DIMENSIONAL BARRIER PROBLEMS

In a one-dimensional scattering experiment, the intensity and energy of the par-

ticles in the incident beam are known in addition to the structure of the poten-
tial barrier V (x). Three fundamental scattering configurations are depicted in
Fig. 7.17. The energy of the particles in the beam is denoted by E.

The Simple Step

Let us first consider the simple step (Fig. 7.17a) for the case E > V . We wish
to obtain the space-dependent wavefunction ({J for all x . The potential function
is zero for x < 0 and is the constant V, for x :::: O. The incident beam comes
from x = -00. To construct ({J we divide the x axis into two domains : region I
and region II, depicted in Fig. 7.18. In region I, V = 0, and the time-independent
Schrodinger equation appears as

li2
-2m ({Jxx = E({J (7.115)

In this domain the energy is entirely kinetic. If we set

(7.116)

then the former equation becomes

2({J
({Jxx = -k1 in region I (7.117)

In region II the potential is constant V and the time-independent Schrodinger

equation appears as

li2
--({Jxx = (E - V)({J (7.118)
2m
7.6 One-D imensional Barrier Problems 223
Vex)
Vex)

V V
E
x x
e» V E< V
(a)

Vex)
Vex)
£
~ ~V ~~ V
E

x x

£> V £<V
(b )

Vex)

-V

(c)

FIGURE 7.17 (a) The simple step. (b) The rectangular barrier. (c) The rectangular well.

---------+----------£
V .....- - - - - - - -

x =O x
Region I
I Region II

'Pn

FIGURE 7.18 Domains relevant to the simple-step scattering problem for the case
E ~ v.
224 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

The kinetic energy decreases by V and is given by

(7.119)

In terms of ka. (7.118) appears as

in region II (7.120)

Writing fill for the solution to (7.117) and filII for the solution to (7.120), one
obtains

fill = Ae ik 1x + Be- ik 1x
(7.121)
filII = Cei k2X + De- ik2X
Since the term De- ik2X (together with the time-dependent factor e- iW2 t ) repre-
sents a wave emanating from the right (x = +00 in Fig. 7.18), and there is no
such wave, we may conclude that D = O. The interpretation of the remaining A,
B , and C terms is given in (7.112). To repeat , A exp(i kl x) represents the incident
wave; B exp( - iklX), the reflected wave; and C exp(ik2X), the transmitted wave.
It is important at this time to realize that fill and fIlII (with D = 0) represent a
single solution to the Schrodinger equation for all x, for the potential curve de-
picted in Fig. 7.18. Since any wavefunction and its first derivative are continuous
(see Section 3.3), at the point x = 0 where fill and filII join it is required that

fill (0) = filII (0)

a a (7.122)
ax fill (0) = ax filII (0)
These equalities give the relations

A+B=C
k2 (7.123)
A - B =-C
kl

Solving for C/ A and B / A , one obtains

B 1- k2/ kl
(7.124)
A 1 + k2/ kl

Substituting these values into (7.114) gives

(7.125)
7.6 One-Dimensional Barrier Problems 225

T+R= I

E= V E=oo k 2 /k 1 !J="Y
=,,/1-£
FIGURE 7.19 T and R versus kZ/ kl for the simple-step scattering problem for E ::: V.

The ratio ka] kl is obtained from (7.116) and (7.119) .

kZ) Z = 1- V (7.126)
( kl E

In the present case E ::: V, so 0 ~ kz/ kl ~ 1. For E » V, kz/ kl -+ 1 and

T -+ 1, R -+ O. There is total transmission. For E = V, kz/ kl = 0 and T = 0,
R = 1. There is total reflection and zero transmission. The T and R curves for the
simple-step potential are sketched in Fig. 7.19. For all values of (kz/ kl) we note
that

T+R=1 (7.127)

The validity of this relation for all one-dimensional barrier problems is proved in
Problem 7.39 .
In the second configuration for the simple-step barrier, E < V (see Fig. 7.17a).
Again the x domain is divided into two regions, as shown in Fig. 7.20. In region I
the Schrodinger equation becomes

rpxx = -klzrp in region I (7.128)

Ae i k 1X Ce- K X
• •
Be...,iklX
V
E
x
Region I x=O Region II
I
'PI 'PH

FIGURE 7.20 Domains relevant to the simple-step scattering problem for the case
E:::: V.
226 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

where

!i 2k 12
--=E (7.129)
2m
In region II the Schrodinger equation is
_ 2
qJxx - K qJ in region II (7.130)

where

(7.131)

The kinetic energy in this domain is negative (-!i 2K 2 12m). In classical physics
region II is a "forbidden" domain . In quantum mechanics, however, it is possible
for particles to penetrate the barrier.
Again calling the solution to (7.128) qJI and the solution to (7.130) f/J!I, we
obtain

qJI = Ae i k 1X + Be- i k1X

qJI! = Ce- K X (7.132)

Continuity of qJ and qJx at x = 0 gives

B C
1+1\ =1\
(7.133)
B K C
1--=i--
A kl A
Solving for (CIA) and (B I A) one obtains

C 2
-
A
= l+iKl kl
(7.134)
B 1 - iKI kl
-
A
= l+iKlk l
The coefficient B I A is of the form z"I z, where z is a complex number. It follows
that IB I A I = 1, so

T=O (7.135)

There is total reflection; hence the transmission must be zero.

To obtain the latter result analytically from our equations above, we must cal-
culate the transmitted current. The function qJI! is of the form of a complex am-
plitude times a real function of x (7.132). Such wavefunctions do not represent
Problems 227

propagating waves. They are sometimes called evanescent waves . That they carry
no current is most simply seen by constructing Jf trans (7.106).

(7.136)

We conclude that T = O.

PROBLEMS

7.39 Show that

T+R=1
for all one-dimensional banier problems .

Answer
Since the scattering process is assumed to be steady-state, the continuity equation
(7.101) becomes

a]x =0
ax
Integrating this equation , one obtains

1 (a]x)
00
-00ax
dx = ]+00 - ]-00 =0

But

]-00 = ]inc - ]refl

]+00 = ]trans
so that the equation above becomes

Dividing through by ]inc gives the desired result.

7.40 Electrons in a beam of density p = 1015 electrons/m are accelerated through a po-
tential of 100 V. The resulting current then impinges on a potential step of height
50V.
(a) What are the incident, reflected, and transmitted currents ?
(b) Design an electrostatic configuration that gives a simple-step potential.

7.41 Show that the reflection coefficients for the two cases depicted in Fig. 7.21 are equal.
7.42 For the scattering configuration depicted in Fig. 7.20, given that V = 2E, at what
value of x is the density in region II half the density of particles in the incident beam ?
228 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

JI inc

FIGURE 7.21 Reflection coefficients for these two configurations are equal. (See Prob-
lem 7.41.)

7.43 Equation (7.123) may be written in the matrix form

Calling the 2 x 2 matrix 9) , the left column vector V , and the right column vector 'U
permits this equation to be simply written

9)V = 'U
This inhomogeneous matrix equation has the solution

where 9)-1 is the inverse of 9) , that is,

(a) Find 9)-1 and then construct V using the technique above. Check your answer
with (7.124) .
(b) Do the same for (7.133) and (7.134) .

7.7 • THE RECTANGULAR BARRIER. TUNNELING

The scattering configuration we now wish to examine is depicted in Fig. 7.17b.

The energy of the particles in the beam is greater than the height of the poten-
tial barrier E > V . For the case at hand there are three relevant domains (see
Fig. 7.22):

Region I: x < -a, V =0.

Region II: -a ~ x ~ +a , V> 0, and constant. (7.137)
Region III: a < x , V =0

The solutions to the time-independent Schrodinger equation in each of the

three domains are
7.7 The Rectangular Barrier. Tunneling 229

•
Ae i k Jx

Be- i k Jx
• --
Ce i k ZX

De - i k 2X
Fe i k Jx
~

-a I 0 II+a x
I
Region I : Region II: Region III
I I
q>, i q>1I : q>1II

FIGURE 7.22 Domains relevant to the rectangular-barrier scattering problem for the
case E 2: V.

(7.138)

The parameter 9 contains all the barrier (or well) characteristics. The latter equa-
tion (conservation of energy) reveals the simple manner in which aki and ak; are
related . In Cartesian aks , ak; space they lie on a hyperbola (Fig. 7.23). The per-
mitted values of ki (and therefore E) comprise a positive unbounded continuum.
For each such eigen-zj-value, there is a corresponding eigenstate ((j1J, <pn, <pm)
which is determined in terms of the coefficients, (BIA , CIA , DIA , FIA).
Knowledge of these coefficients gives the scattering parameters

The coefficients are determined from the boundary conditions at x = a and

x=-a,
230 Chapter 7 Add itional One-Dimensional Problems. Bound and Unbound States

~ __ 12a I .....

,
" ........ - -+-'"""''-----

FIGURE 7.23 For rectangular-barrier scattering with E ~ V , ak] and ak2 lie on a
hyperbola.

The energy spectrum /i,2k]2 12m comprises an unbounded continuum.

(7.139)

These are four linear, algebraic, inhomogeneous equations for the four unknowns:
(B I A ), (C I A), (D I A ), and ( F I A ). Solving the last two for ( D I A) and (CIA)
as functions of (F I A) and substituting into the first two permits one to solve for
(B I A) and (F I A) . These appear as

- 1
F . i
- = e-2lkla co s(2k2a) - -
kl
2
+ k2 2 sin(2k2a)
A [ 2 ( )
klk2 ]
(7.140)

(B) = (F)
2
k2 - k] 2
2 - i - sin (2k2a )
A A k]k2

The transmis sion coefficient is most simply obtained from the second of these,
together with the relation
7.7 The Rectangular Barrier. Tunneling 231

(7.141)

There results

(7.142)

Rewriting k, and k2 in terms of E and V as given by (7.138), one obtains

E>V (7.143)

The reflection coefficient is 1 - T .

For the case E < V , as depicted in Fig. 7.24a, we find that the structure of the
solutions (7.138) are still appropriate, with the simple modification

• •
Be-i klx
E

-0 0 a x

Region I Region II Region 1Il

"'I
(a)

(b)

FIGURE 7.24 (a) Domains relevant to the rectangular-barrier scattering problem, for the
case E ~ V . (b) Real part of <p for the case above, showing the hyperbolic decay in the
barrier domain and decrease in amplitude of the transmitted wave.
232 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

(7.144)

This latter conservation of energy statement indicates that the variables ak, and
a« lie on a circle of radius 9/2 (Fig. 7.25). The permitted eigen-zj-values now
comprise a positive, bounded continuum, so that the eigenenergies

also compri se a positive, bounded continuum.

The algebra leading to (7.140) remains unaltered so that the transmission coef-
ficient for this case is obtained by making the substitution of (7.144) into (7.142) .
We also recall that sin( iz) = i sinh z. There results

(7.145)

which, with (7.144), gives

1 1 V2
- = 1+ - sinh2 (2Ka) (7.146)
T 4 E(V - E)
Writing this equation in terms of T,

1
T = ------,;0----- E<V (7.147)
2
1 V 2
1 + 4: E(V _ E) sinh (2Ka)

___Rt- V
-
~12al~
- -

I I
\ I aK
\ I
\ I
\ I
\ I
\ /
\ //
............. _-- --,;,""

FIGURE 7.25 For rectangular-barrier scattering with E :::: V , ak) and a« lie on a circle
ak, 2: 0, a« 2: O. The energy spectrum (1i2k) 2 12m ) comprises a bounded continuum.
7.7 The Rectangular Barrier. Tunneling 233

indicates that in the domain E < V , T < 1. The limit that E -+ V deserves
special attention . With

one obtains

1
T = 2 + O(E) < 1
1 + 9 /4
2 2m(2a)2V
(7.148)
9 == ti2

The expression 0 (E) represents a sum of terms whose value goes to zero with E.
We conclude that for scattering from a potential barrier, the transmission is less
than unity at E = V (Fig. 7.26).
Returning to the case E > V, (7.143) indicates that T = 1 when sin2(2k2a) =
0, or equivalently when

(n = 1,2, . . . ) (7.149)

Setting k: = 2n / A, the latter statement is equivalent to

2a=n(~) (7.150)

1.0
0.9
0.8

0.6

0.4

0.2

3 4 5 6 7

E/V=£
FIGURE 7.26 Transmission coefficient T versus E / V for scattering from a rectangular
=
barrier with 2m(2a)2V /!i2 9 2 = 16. The additional lines are in references to Problems
7.50 et seq.
234 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

When the barrier width la is an integral number of half-wavelengths, n (A/2) , the

barrier becomes transparent to the incident beam; that is, T = 1. This is analogous
to the case of total transmission of light through thin refracting layers.
Written in terms of E and V, the requirement for perfect transmission, (7.149),
becomes

(7.151)

where E) is the ground-state energy of a one-dimensional box of width la [see

(4.14)].
Equations (7.143) and (7.146) give the transmission coefficient T , as a func-
tion of E , V , and the width of the well la . The former of these indicates that
T ~ 1 with increasing energy of the incident beam. The transmission is unity
for the values of E given by (7.151). Equation (7.146) gives T for E ~ V . The
transmission is zero for E = 0 and is less than 1 for E = V. A sketch of T versus
E I V = E for the case 9 2 = 16 is given in Fig. 7.26.
The fact that T does not vanish for E < V is a purely quantum mechanical
result. This phenomenon of particles passing through barriers higher than their
own incident energy is known as tunneling. It allows emission of ex particles from
a nucleus and field emission of electrons from a metal surface in the presence of
a strong electric field.

PROBLEMS

7.44 In tenns of the new variables ,

-B= ivn '

Rel'PR
A

(7.140) may be rewritten in the simpler form

. e2iakl
..IT el'PT = - - - - - -
cos f3 - ia+ sin f3
.fR ei'PR = ia_..IT ei'PT sin f3
Use these expressions to show:
(a) T + R =1
(b) cPT = cPR - n(1f /2), n = 1,2,3, ...
(c) tan (cPT - 2k) a) = a+ tan f3
(d) What is t/> R for the infinite potential step : V (x) = 00, x ~ 0; V (x) = 0, x < O?
 7.8 The Ramsauer Effect 235

Answers (partial)
(a) Solving for T + R from (7.140) gives
I + a.: 2 sin2 ,B
T + R = -~-----::----'-;;-­
cos 2 ,B + a+ 2 sin2 B

Substituting the definitions of a± gives the desired result.

(c) From the first of the two given equations above, we obtain

.JTei('PT-2kla) = I
cos,B + ia+ sin,B
e-i'P

Equating the tangents of the phases of both sides gives the desired result.

7.45 An electron beam is sent through a potential banier 4.5 A long. The transmission
coefficient exhibits a third maximum at E = 100 eV. What is the height of the
banier?
7.46 An electron beam is incident on a banier of height 10 eV. At E = 10 eV, T =
3.37 x 10- 3 . What is the width of the banier?

7.47 Use the correspondence principle with (7.147) to show that T = 0 for E < V , for
the classical case of a beam of particles of energy E incident on a potential banier of
height V .

7.8 • THE RAMSAUER EFFECT

The configuration for this case is depicted in Fig. 7.17c. The relevant domains are
shown in Fig. 7.27. Once again eqs. (7.138) et seq. apply with the modification

--
Ae i k l x

Be- i k 1x -
Ce i k 2X
~

D e- i k 2X
__- - - - + - - - - + - - - - - - E
-a a
•x
----;\ I~-IVI
Region I I Region II I Region III

'PI 'Pn 'Pm

FIGURE 7.27 Domains relevant to the rectangular-well scattering problem, E > O.

236 Chapter 7 Additional One-Dimensional Problems. Boundand Unbound States

li,2k2 2
~ =E- V = E+IVI (7.152)

The transmission coefficient (7.143) becomes, for E ~ 0,

(7.153)

Again there is perfect transmission when an integral number of half-wavelengths

fit the barrier width.

2ak2 = nit (n = 1,2, .. . ) (7.154)

This condition may also be cast in terms of the eigenenergies of a one-dimensional

box of width 2a:

(7.155)

From (7.153) we see that T --+ 1 with increasing incident energy. At E = 0,

T = O. Thus we obtain an idea of the shape of T versus E . It is similar to the
curve shown in Fig. 7.26. The transmission is zero for E = 0 and rises to the
first maximum (unity) at E = E I - IV I. It has successive maxima of unity at the
values given by (7.155), and approaches 1 with growing incident energy E .
The preceding theory of scattering of a beam of particles by a potential well has
been used as a model for the scattering of low-energy electrons from atoms. The
attractive well represents the field of the nucleus, whose positive charge becomes
evident when the scattering electrons penetrate the shell structure of the atomic
electrons. The reflection coefficient is a measure of the scattering cross section.l
Experiments in which this cross section is measured (for rare gas atoms) detect
a low-energy minimum which is consistent with the first maximum that T goes
through for typical values of well depth and width according to the model above,
(7.153) . This transparency to low-energy electrons ofrare gas atoms is known as
the Ramsauer effect.
The student should not lose sight of the following fact. For any of the solu-
tions to the scattering problems considered in these last few sections, we have in
essence found the eigenfunctions and eigenenergies for the corresponding Hamil-
tonian . These Hamiltonians are of the form
p2
H = 2m + Vex) (7.156)

with the potential V (x) depicted by any of the configurations of Fig. 7.17. In each
case considered, the spectrum of energies is a continuum, E = li,2k2/2m . For each
value of k, a corresponding set of coefficient ratios (B I A, CIA for the simple
step and B I A, CIA, D I A , F I A for the rectangular potential) are determined.
5The notion of scattering cross section is discussed in Chapter 14.
Problems 237

TABLE 7.2 Transmission coefficients for three elementary potential barriers

-----------£

T =0, R = 1

1 1 y2

vn £ - = 1+
T

!ihl
--=E-Y
-
4 E(E - Y)
sin2(2k2a)

--1 2a f-.- 2m

V 1 1 y2 2

-l
Rf-.-
2a
£ T = 1 + 4 E(Y _
/i,2 K2
E) sinh (2Ka)

--=Y-E
2m

£ I y2
- = 1+ - sin2(2k2a)
T 4E(E+IVD

--l
LJ-2a
1V1
f-.-
/i, 2k22
-2m- = E - Y =E+ IVI

The coefficient A is fixed by the data on the incident beam. These coefficients
then determine the wavefunction, which is an eigenfunction of the Hamiltonian
above. All such scattering eigenstates are unbound states. A continuous spectrum
is characteristic of unbound states, while a discrete spectrum is characteristic of
bound states (e.g., particle in a box, harmonic oscillator).
The transmission coefficients corresponding to the one-dimensional potential
considered above are summarized in Table 7.2.

PROBLEMS

7.48 The scattering cross section for the scattering of electrons by a rare gas of krypton
atoms exhibits a low-energy minimum at E ::::: 0.9 eV. Assuming that the diameter of
the atomic well seen by the electrons is 1 Bohr radius, calculate its depth.
238 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

7.49 Show that the transmission coefficient for the rectangular barrier may be written in
the form

T = T(9 , E)
where

2 _ 2m(2a)2V
9 = 1i2

- E
E=:-
V

Answer (partial)
For E ~ I,

7.50 Using your answer to Problem 7.49, derive an equation for an approximation to the
curves on which minimum values of T fall.

Show that the values of T and E at the first minimum in the sketch of T versus E
depicted in Fig. 7.26 (9 2 = 16) agree with your equation . [Hint: The minima of T
fall at the values ofE where T- 1 is maximum . From Problem 7.49,

T- 1 < I +~ I .J
- 4 E(E - 1)

7.51 For the rectangular barrier:

(a) Write the values of E for which T = 1 as a function of 9 .
(b) Using your answer to part (a) and the two preceding problems , make a sketch
of T versus E in the two limits 9 » 1, 9 « 1. Cite two physical situations to
which these limits pertain .
(c) Show that for an electron , 9 2I V =: 2m(2a)211i2 = O.26(2a)2(eV)-I, where a
isinA.
7.52 For the case depicted in Fig. 7.26, show that the first maximum falls at a value con-
sistent with your answer to part (a) of Problem 7.51.
7.53 Write the transmission coefficient for the rectangular well as a function of 9 and E.

Answer

7.54 In the limit 9 2 » I, show that the minima of T for the rectangular well fall on a
curve which is well approximated by

Tmin =4E
Problems 239

1.0 - ----

0.8

0.6

0.4

0.2
_--- Tmin =4E/1V1

0.015 0.030
E/IVI=£
FIGURE 7.28 Resonances in the transmission coefficient for scattering by a potential
well for 9 2 = 105 . (See Problems 7.54 et seq .)

Use this result together with (7.155) for the values of E where T 1 to obtain a
sketch of T versus E for the case 9 2 = 105 .
Answer
See Fig. 7.28.
7.55 Show that the spaces between resonances in T for the case of scattering from a po-
tential well grow with decreasing 9 .
7.56 (a) Calculate the transmission coefficient T for the double potenti al step shown in
Fig.7.29a.
(b) If we call TI the transmission coefficient appropriate to the single potential step
VI , and T2 that appropriate to the single potential step V2, show that

T ~ T2

Offer a physical explanation for these inequalities .

(c) What are the three sets of conditions under which T is maximized? What do
these conditions correspond to physically?
(d) A student argues that T is the product TI T2 on the following grounds . The par-
ticle current that penetrates the VI barrier is TI.lfinc ' This current is incident on
the V2 barrier so that T2(TI.lfinc) is the current transmitted through the second
barrier. What is the incorrect assumption in his argument?
Answer (partial)
Applying boundary conditions to the wavefunctions

CPI = Aeik 1X + Be- ik 1X (region I)

CPn = Ceik2X + De- ik2X (region II)

CPm = F eik3X (region III)

240 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

----+---+-------£
V2 1-- - - - - - -
Jltrans =T JI inc
a x
Region I Region II Region III

(a)

- ....- - + - - - - + - - - - - - £

VI r----+-------
a x

(b)

- ....--+----+-------£
V2 1--- - - - - - -

a x

(c)

FIGURE 7.29 (a) Double potential step showing three regions discussed in Problem
7.56. (b) and (c) Two related single potential steps: Tl 2: T2 and T2 ~ T .

-a +a x

-aV

FIGURE 7.30 Tunneling configuration for Problem 7.57. The constant ex is real and
greater than zero.
 7.9 Kinetic Properties of a Wave PacketScattered from a Potential Barrier 241

at x = 0 and x = a, respectively, and solving for T = (k3/kI)IF/AI2 gives the

desired result:
2
T = 4kl k3k2
k2 2(kI + k3)2 + (k3 2 - k2 2)(kI 2 - k2 2) sin2 (k2a)
Note that

where /::;.2 is as implied . With (7.125) we see that TI 2: T2.

7.57 Calculate the transmission coefficient for the potential configuration and energy of
incident particles depicted in Fig. 7.30. (Note: T is easily obtained from the answer
given to Problem 7.56.)

7.9 • KINETIC PROPERTIES OF A WAVE PACKET SCATTERED

FROM A POTENTIAL BARRIER

The time-dependent one-dimensional scattering problem addresses itself primar-

ily to the problem of a wave packet incident on a potential barrier. It seeks the
shape of the reflected and transmitted pulses. We will restrict our discussion to
the kinematic properties of these pulses.
To formulate this problem we first construct a wave packet whose center is at
x = - X at t = O. In previous chapters we obtained such wave packets centered
at x = 0 at t = O. They are of the form

1/J(x, t) = -1-
-/2ii
1 -00
00
b(k)ei(kx-wt) dk (7.157)

For this same packet to be centered at x = - X initially, one merely effects a

translation in x so that

1/J(x, t) = _1_1
-/2ii
00

-00
b(k)eik(x+X)e-iwt dk (7.158)

= _1_1
-/2ii
00

-00
b(k)eikXei(kx-wt) dk

For example, for a chopped pulse, L em long, containing particles moving with
momentum Mo, b(k) is given by (6.32):

b (k)= [£ sin(k - ko)Ll2

--~-"-'----'-
nL k-ko
(7.159)

See Fig. 7.31. The group velocity of this packet is vo = Mol m . Let us call the
wave packet (7.158), 1/Jinc. This packet is a superposition of plane-wave states of
242 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

p=lik o
-.- TV

-1 f-
2a

_------x------_
·1
FIGURE 7.31 Wave packet incident on a potential barrier. x » L ::::: 2a.

the form (7.112). Each such incident k-component plane wave is reflected and
transmitted. The corresponding reflected and transmitted waves are constructed
from the amplitude ratios B / A and F / A given by (7.140), which are functions of
k [kl in (7.140)]. Reassembling all of these waves, one obtains

x < - a ,I..
'PIne
= _1_1
IF
v ...n
00

-00
b(k)eikXei(kx-wt) dk

x < - a Vrrefl = _1_ roo -JR ei</>Rb(k)eikXe-i(kx+ (J)I) dk (7.160)

.fIi 1-00

x> +a , I.
'Ptrans
= _1_1
IF
v ...n
00

-00
..IT ei</>T b(k)e ikXei(kx-wt) dk

Here we are using the notation of Problem 7.44.

To uncover the kinetic properties of these packets, we use the method of sta-
tionary phase. This relies on the fact that the major contribution in a Fourier in-
tegral is due to the k component with stationary phase. If we call this component
ko, then the phase of the Fourier integral for Vrrefl vanishes when

o + kX -
-(f/>R kx - wt) = 0 (7.161)
ok
This gives the trajectory of the reflected packet,

1iko
x=--t+X+ (Of/>R)
- x < - a (7.162)
m ok ko

In like manner, for the incident and transmitted packets, one obtains

1iko
x= - t - X , x < -a (7.163)
m
x = 1iko t _ X_ (of/>r) , x> a (7.164)
m ok ko
 7.10 The WKB Approximation 243

The latter three equations illustrate the effect of a potential barrier on the trajec-
tory of an incident wave packet. Were there no barrier, the packet would move
freely in accordance with (7.163). However, there is a delay for both the trans-
mitted and reflected packets. The transmitted pulse arrives at any plane x > a,
(a4>T/akO)Vo-1 seconds after the free pulse. The reflected pulse arrives at any
plane x < -a, (a4>R/akO)Vo-1 seconds after the free pulse would be reflected
from an impenetrable wall at the x = 0 plane.P

PROBLEMS

7.58 For a pulse such as described in (7.158) and (7.159), containing 1.5-keYelectrons,
which scatters from a potential well of width 0.5 x 10- 7 cm and of depth 25 keY,
what is the delay in the transmitted beam (in s) imposed by the well?
7.59 Is there a delay in the scattering of a wave packet from a simple-step potential?
Present an argument in support of your answer.
7.60 In the text we mentioned the method of stationary phase for evaluating Fourier inte-
grals. Use this method to show that

L: !(k)eis(k) dk ::::: ~ !(ko)e i[s(ko)±Jr /41

Is"(kO)1

s'(kO) =0
The phase factor +iJT/4 applies when s"(kO) > 0 and -iJT/4 applies when
s" (kO) < O. Primes denote k differentiation . [Hint: Expand s(k) in a Taylor se-
ries about k kO' keeping O(k 2) terms.]
=
7.61 There is a tacit assumption in the construction of (7.160) that no interaction occurs
between the incident wave packet and the potential barrier in the interval 0 ~ t ~
X/VO . Is this a valid assumption?
Answer
All k components in the distribution (7.159) with k > ko reach the barrier in
a time less than X/VO . The number of such components diminishes in the limit
X»2a:::::L.

7.10 • THE WKB APPROXIMATION 7

Correspondence
In Section 7.3 we found that the quantum probability density goes over to the
classical probability density in the limit of large quantum numbers. Such quan-
tum states have many zeros and suffer rapid spatial oscillation . Equivalently, we
6The time development of a wave packet scattering from a potential barrier is graphically depicted in
D. A. Saxon, Elementary Quantum Mechanics, Holden-Day, San Francisco, 1968.
7Named for G. Wentzel, H. A. Kramers , and L. Brillouin.
244 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

may say that in this classical domain the local quantum (de Broglie) wavelength
is small compared to characteristic distances of the problem. For the harmonic os-
cillator, such a characteristic distance is the maximum displacement or amplitude
xo (7.12) . More generally, this characteristic distance may be taken as the typical
length over which the potential changes. Since the de Broglie wavelength changes
only by virtue of a change in potential, the latter condition may be incorporated
in the criterion (for classical behavior) that the quantum wavelength not change
appreciably over the distance of one wavelength. Now the change in wavelength
over the distance 8x is
d)'"
8)", = dx
-8x

In one wavelength this change is

d)'"
8)..,=-)..,
dx
In the classical domain , 8)", « ). , (Fig. 7.32). This gives the criterion

~~
8
(7.165)
1: I = 1 I « 1

In terms of the momentum p , we find that

or, equivalently,
d): mh dV
dx 7 dx

FIGURE 7.32 In theWKBanalysis, thefractional change 0>"/>" « 1. The potential scale

of length is also largecompared to wavelength.
7.10 The WKB Approximation 245

Thus, the condition (7.165) for near-classical behavior becomes

8A
-
A
I_lm
- -hdVI
- «1
p3 dx
(7.166)
1

The WKB Expansion

We seek solutions to the time-independent Schrodinger equation (7.1) which are
valid in the near-classical domain (7.166).
If the potential V is slowly varying, one expects the wavefunction to closely
approximate the free-particle state

cp(x) = Aei kx = Aeipx / n

Thus we will look for solutions in the form

cp(x) = AeiS(x)/n (7.167)

Substitution of this function into (7.1) gives

a2s + (as)2
- ili-2 - = p2(x) (7.168)
ax ax
p 2 = 2m[E - Vex)]

To further bias the solution (7.167) to the classical domain we examine the solu-
tions to the nonlinear equation (7.168) in the limit Ii -+ O. Recall (Section 6.1)
that it is in this limit that the Gaussian packet reduces to the classical particle . To
these ends we expand Sex) in powers of Ii as follows:

(7.169)

Substituting this expansion into (7.168) gives

(7.170)

Since this equation must be satisfied for small but otherwise arbitrary values of
h, it is necessary that the coefficient of each power of h. vanish separately. In this
manner we obtain the following series of coupled equations for the sequence {Sn}.
246 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

(7.171)

Integrating the first of these equations gives

So(x) = ±
JXQ
r p(x) dx

or, equivalently, in terms of wavenumber k = pi Ti,

/i=±
So lXQ
x
k(x)dx (7.172)

Substituting this solution into the second equation in (7.171) and integrating gives

SI = -2i In (OSo)
-
ox
= -2i In hk
or, equivalently,

. 1
exp(ISI) = TiI /2k l / 2 (7.173)

Substituting (7.172) and (7.173) into the third equation in (7.171) and integrating
gives

_ ~ m(oVlox) _ ~
S2 - 2 p3 4
f m 2(oV lox)2
p5 dx (7.174)

In that SI is the log of the derivative of So, we cannot in general ignore SI

compared to So, and both terms must be retained in the expansion (7.169). How-
ever, comparison of S2 (7.174) with the criterion (7.166) shows that in the near-
classical domain, the contribution of the second-order term Ti2 S2/2 to the phase
of tp is small compared to unity. Higher-order contributions to S (x) are likewise
small. Thus, it is consistent to say that near the classical domain, tp is well de-
scribed by the first two terms in the expansion (7.169). Inserting these solutions
into (7.167) gives

cp(x) = k~2 exp (i f kdX) + k~2 exp (-i f kdX) (7.175)

The Near-Classical Domain

In what sense does the solution (7.175) approximate classical behavior? To answer
this question we consider the probability density cp*cp . Specifically, consider that
7.10 The WKB Approximation 247

the momentum of the particle is specified so that it is known that the particle
is moving to larger values of x . Then the corresponding WKB solution (7.175)
reduces to

qJ(X) = k~2 exp (i f kdX)

The probability density for this state is

2 2
P(X) = qJ*qJ = IAI = IA1 1j,fm
k v
where v is written for the classical velocity, v = p/ m. The probability of finding
the particle in the interval dx about x is

This result, apart from a multiplicative constant, is the same as the classical prob-
ability, P dx ,....., dt [see (7.70)].
To obtain correspondence with the classical current , we renormalize qJ so that
it is relevant to a beam of N particles such as described in (7.95). Calculation of
the current (7.106) gives

N1l.IAI 2
.JJ = = N P(x)v(x)
m
.JJ = p(x)v(x)

This is the classical expression for the current across a plane at the point x of a
beam of particles with number density p(x) moving with velocity v(x).
Thus, the lowest-order WKB solution (7.175) reproduces the classical proba-
bility and current.

Application to Bound States

Consider the potential shown in Fig. 7.33. The WKB solution (7.175) is invalid
at the classical turning points Xl and X2, for at these points E = V and lik = 0,
thereby violating the criterion (7.166). However, the WKB solution becomes valid
in regions far removed from the turning points where IE - V I is sufficiently large.
In region I, far to the left of Xl (X ~ -00), the solution is

qJl = 1
r.; exp IX K dx
"11 K Xl
(7.176)
2 K2
1l.
--=V-E>O
2m
Far to the right of X2, the wavefunction also decays exponentially.
248 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

E-+---~-~-~-+------

Region I Region III

x

FIGURE 7.33 Domains relevant to the WKB approximation of bound states.

(7.177)

In the classically allowed region II, the WKB solution is oscillatory. It is nec-
essary in the WKB construction of rp to separate this component of the solution
into two parts.

rpn «X)= ~sin(l: kdX+O), XI < X (7.l78a)

~ sin (l
X2

ffJII >(X) = kdx + 0), X < X2 (7.l78b)

Iih2
--=E-V>O
2m

Through connection formulas obtained below, ffJII < is matched to rpl and ffJII > is
matched to ffJIn. This connecting process will serve to determine all but one of
the constants A, B , C, and O. The remaining constant is determined in stipulating
that rpII < join smoothly to ffJII > ' This continuity condition will also be found to
generate energy eigenvalues within the WKB approximation .

Connecting Formulas for Bound States

If rpI, rpn <, rpn >, and rpm were valid representations of rp throughout their re-
spective domains, the constants of these functions could be obtained by simply
matching these component solutions as was done in preceding sections of this
chapter. This method clearly cannot be followed in the present analysis since the
WKB solutions are invalid at the turning points.
The technique of matching rpI to rpn < and rpn > to ffJIn in the WKB approxi-
mation is as follows. The Schrodinger equation is solved exactly in the regions
of the turning points for potentials that approximate V (x) in these domains. The
asymptotic forms of these exact solutions are then used to match rpI to rpn < and
7.10 The WKB Approximation 249

to match fPII > to qJm. Following this prescription, we approximate Vex) in the
neighborhood of x I with the linear potential VI (x) .

(7.179)

The constant FI is the slope of V (x) at XI . Similarly, in the neighborhood of X2

we write

(7.180)

(See Fig. 7.34 .) The Schrodinger equation then appears as

x near XI (7.181)

X near X2 (7.182)

Further simplification of these equations is accomplished through the change in

variable

2m F I ) I/3
y=- ( ~ (x-x))

in (7.181) and

2)1/3(x -X2)
2m F
Y = ( -/i-

in (7.182). Both equations then reduce to the same equation (see Problem 7.93),

(7.182a)

l"~
Vex)

FIGURE 7.34 Approximate linear potentials VI (x) and V2(X) valid in the neighbor-
hoods of the turning points XI and X2 , respectively.
250 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

The solutions to this equation are called'' Airy functions and are denoted by the
symbols Ai (y) and Bi (y) (see Table 7.3). For the problem at hand , the wavefunc-
tion q> (x) must approach zero in the domains x « Xl and x » X2. Both these
regions correspond to large positive values of Iyl. The function with this property
is Ai (y ), which has asymptotic forms

Ai(y) ~ 1
2"fii y1/4
exp (_~y3/2)
3
(y > 0)
(7.183)
Ai(y) ~ 1
"fii(_y)1/4
sin [23 7T]
_ (_y)3/2 +-
4
(y < 0)

It is exponentially decaying for y > 0 and oscillatory for y < 0 and strongly
resembles the behavior of a harmonic oscillator wavefunction across a turning
point, such as shown in Fig. 7.6.
In the neighborhood of Xl, from (7.179), we obtain

p2 = 2m(E - Vd ::::= 2mF1 (x - xd =- (2m F11i)2/3y

2mF1 dx = -(2mF11i)2/3 dy

and we may write

l
x

Xl
K dx = - iY
0
2 y3/2
,.jYdy = __
3

To the right of Xl, in the oscillatory well domain, p2 = 1i2k2 and

so y is negative in this domain. The integral of k gives

lXl
x
kdx =- iY
0
~dy = 2_(_ y)3/2,
3
y <O

8Named for an English astron omer, G. B. Airy (180 1-1892).

TABLE 7.3 Properties of Airy functions"
Differential equation

Solutions
(a) Series representation

Ai(x) = af(x) - bg (x)

Bi(x) = v'3 [af(x) + bg(x)]

where
a= r 2/ 3/ f(2/3) = 0.3550, b = 3- 1/ 3 / f(l /3) = 0.2588

f(x) = 1 + .!..x 3 + ~x6 + ~x9 + ...

3! 6! 9!
2 4 2 ·5 7 2 ·5 ·8 10
g(x)=x+4i x +"7!x +~x + ...
(b) Integral representation

1 [00
Ai(x) = -; cos (s3
10 "3 +sx ) d s

Bi(x) = ~ faoo [e SX
-
O / 3)S3 + sin ( ~ + sx) ] ds

Relations to Bessel fun ctions offra ctional order

With Y == jx3/ 2 , the following relationshold.
. I
AI(X) = -/X!3 Kl /3( Y)
n
. I
Al(-X) = 3v'x[lt /3(Y) + LI /3(Y)]
Bi(x) = /X!3 [l-1 /3(Y) + Il /3(Y)]
Bi(-x ) = /X!3[LI /3(Y) - JI /3(Y)]

The I and K functions are modified Bessel functionsof the firstand second kind, respectively.
Asymptotic forms
For large [x] , leading terms in asymptoticseries are as follows:

· ) ~ 2.jii1x / exp ( -3
AI(X 2x3/2), x > 0
l 4

· I . [2
Al (x) ~ .jii(_x)I /4 sin 3 (-x)
3/2 7r] '
+ 4" x < 0

·
BI(X) ~ .jii l / Ix 4 exp (2 3/2) 3x, x >0

Bi(x) ~ I cos [~(_X)3/2 + ::.], x <0

.jii(-x)I /4 3 4

aFor further propertiesof these functions, see Handbook ofMathematical Functions, N. Abramowitz
and 1.A. Stegun,eds., Dover, NewYork, 1964; H. and B. S. Jeffries, Methods ofMathematical Physics ,
3d ed., CambridgeUniversity Press, New York, 1956.

251
252 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

In these same respective doma ins, the WKB functions 'PI (7.176) and 'PH <
(7.178a), when written in terms of the variable y, appear as

1 (2 3/2)
'PI = - - exp --y
yl /4 3
y>O

C . (2
'Pn< = (_y)I /4 sin 3(-y)3/2 + 8 ) y <O

These agree with the asymptotic forms (7.183) for the exact Airy function solu-
tions [corresponding to the approximate linear potential (7.179, 7.180)] provided
that we set C = 2 and 8 = rrj4.
In this manner we find that the WKB approximation in region I,

(x < xI) (7.184)

matches (or "connects") with the WKB approximation

(7.185)

in region II.
In like manner we find that the WKB approximation in region III

(7.186)

matches with the WKB approximation

2A
'PII > = -Jk sin [,
(r Z
k dx
rr)
+ "4 (7.187)

in region II. The remaining constant A is determined in matching 'PII < to 'Pn>.

The Four Connection Formulas

There are in total four connection formulas which serve to relate WKB component
wavefunctions across turning points . In the preceding analysis two of these rela-
tions were uncovered. Namely, these are given by the manner through which 'PI
connects to 'Pn< (7.184 and 7.185) and that by which 'Pn> connects to 'PIlI (7.186
and 7.187). Carrying through a parallel analysis and employing the asymptotic
forms for the Airy functions B i(y) gives the remaining two relations. The com-
plete list of four connecting formulas" is given below with XI ,2 denoting either XI
or X2 .

9Here we are assuming that no other linearly independent components of the wavefunction enter the
analysis.
 7.10 The WKB Approximation 253

x < xl ,2 Xl,2 < x

(7.188a) -cos
2
.jk
(l kdx --Jr)
4 x
X2

~exp (-l>dx)
(7.188b)
x
-sin
I
.jk
(l kdx--Jr)
4 x
X2

- ~ exp (1: KdX)

(7. I89a)
x
~ exp (1: K dX) ~
.jk
cos (l
xl
kdx -~)4
X

(7.189b) __I exp

.jk
(-l X

Xl
KdX) _I
.jk
sin (l kdx _~)
X

XI4

Bohr-Sommerfeld Quantization Rules

The energy levels of the finite well depicted in Fig . 7.33 may be obtained to within
the accuracy of the WKB approximation by joining fPIl < and fPIr > smoothly within
the well . This gives

sin (1: kdx + ~) = A sin (l

X2

k dx + ~)
With
X2 X2
7T:
TJ ==
1 Xl
kdx , a ==
l
X
kdx +-
4

the continuity condition above becomes

. ( TJ
sm 7T: -
+ "2 a ) = A sm
. a

or, equivalently,

. (7T:)
sin TJ +"2 cos a - cos (7T:) . a = A sm
TJ +"2 sm . a

The solution to this equation which gives A, constant and independent of the
parameter a, is obtained by setting!"

7T:
TJ + "2 = (n + 1)7T: (n = 0, I, 2, .. . ) (7.190)

IOWriting n + 1 instead of n ensure s that 1/ is nonnegative.

254 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

FIGURE 7.35 Classical vibrational motion between the turning points (Xl, x2) depicted
in (x, p) "phase space." The enclosed surface has the area :f p dx.

Corresponding values of A are (-l)n. Thus, continuity of <Pn implies the condi-
tion

When written in terms of momentum p == hk; this criterion appears as

l
X2

Xl
pdx (1)
== n + - -h
2 2
(7.191)

In the corresponding classical motion, the particle oscillates between the turn-
ing points x I and X2. In Cartesian x, p space, this "orbit" is a closed loop, as
depicted in Fig. 7.35, with area 1 p dx. With (7.191) we find that

(7.192)

This equation is nearly the same as the Bohr-Sommerfeld quantization rule l l

(2.6). As discussed in Section 2.4, this rule prescribes that an integral number
of wavelengths fit the orbit perimeter (see Fig. 2.11). When cast in terms of wave-
length A == hlp, (7.192) becomes

+~
r1.. A ==
dx n
2
(7.193)

The integral represents the number of wavelengths in the orbit perimeter. The dis-
tinction between the result and the Bohr-Sommerfeld prescription is that in the
present WKB analysis, the wavefunction may extend into the classically forbid-

lIThe distinction between the loop integral in (2.6) and that in (7.192) is that f pe de is relevant to
rotational motion while f P dx is relevant to vibrational motion. In either case such integrals play a
major role in the study of periodic motion and are called action integrals. The Bohr-Sommerfeld quan-
tization rule stipulates that these action integrals have only discrete values, nh. That is, in quantization
of periodic systems, one quantizes the action variables f p dx.
7.10 The WKB Approximation 255

den region, or, equivalently, that the wavefunction need not vanish at the turning
points. If the wavefunction must vanish at the turning points, such as is the case for
a very sharply rising potential, a half-integer number of wavelengths are allowed
between turning points, thereby returning the Bohr formula. On the other hand,
leakage of the wavefunction into the classically forbidden domain in the WKB
analysis is evident from the wavefunction (7.184) through (7.187) and matching
value, A = (_l)n.

WKB Eigenenergies
The continuity result (7.191) serves to determine eigenenergies within the WKB
approximation. Since this analysis becomes more accurate for large energies, val-
ues so found will generally give better estimates for large quantum number n. In
that this number is also a measure of the number of zeros of the wavefunction
between turning points, we see that in this limit the wavelength becomes small
compared to the distance between the turning points. As described in the first
paragraph of this section, such is the domain of the classical WKB analysis.
As an example of the application of (7.191), let us consider calculation of the
energies of the harmonic oscillator. For this configuration the momentum is given
by

with turning points given by (7.12). Introducing the variable

permits the condition (7.191) to be written

4E
-
Wo
i
0
7C

sin 2 ede =(1)+- h

n
2
2rrE
- -=
n ( n+-1) h
WO 2
These are seen to be, somewhat fortuitously, the exact eigenenergies of the har-
monic oscillator (7.44), valid for all n.
An example in the opposite extreme is given by the one-dimensional box po-
tential (4.1). There is no penetration of the particle wavefunction through the
sharply rising potential wall, and the validity of the WKB formula (7.191) be-
comes very questionable, especially at low energies. Using this formula, we read-
ily obtain the eigenenergies

= (n + 21)2 8ma
2
h
= En (4n+l)
1+ ~
KB
E: 2
256 Chapter 7 Additional One -Dimensional Problems. Bound and Unbound States

Here we have written En for the exact eigenenergies (4.14), En = n 2 EI. As

expected , the estimate gives a large fractional error for small quantum numbers.
However, in the high-quantum-number domain, n » I, where the walls of the
potential are many wavelengths apart, we find that the WKB estimate agrees with
the exact result

E nW KB -~ E n (n » 1)

Note, however, that for this singular case where wavefunction penetration does
not occur, the simpler Bohr-Sommerfeld rule (2.6) gives exact results.

WKB Wavefunctions
Wavefunctions in the WKB approximation incorporate Airy functions together
with matching conditions (7.184) et seq. Results of this analysis to the fourth
state of the harmonic oscillator are shown in Fig. 7.36. Here we may observe the
dramatic disparity between the exact wavefunction and the WKB approximation
in the vicinity of the turning points.

Application to Transmission Problems

In concluding this section we will obtain a very important formula for the trans-
mission coefficient relevant to a potential barrier such as depicted in Fig. 7.37. As
described previously in Section 7.5, the transmitted wave in region III has only
one momentum component, so that to within the WKB approximation, we may
write

~II1 = ~ exp [i (1: k dx - ~) ] (7.194)

The procedure we will follow to obtain the incident component of ~I is as follows.

Rewriting (7.194) as a combination of trigonometric functions permits application
of the connection formulas (7.189), which allows calculation of ~I1 . With ~I so
found, we again rewrite it in a manner that permits application of (7.188) to con-
nect ~I1 to ~I . Finally, ~ is decomposed into incident and reflected components.
Comparison of the incident component with ~II1 permits calculation of the trans-
mission coefficient.
Rewriting (7.194) in the form

~II1 = ~ [cos (1: k dx - ~) + i sin (1: k dx - ~) ] (7.195)

permits application of (7.189) and we obtain

~I1 = 2.jK
A exp (l
- x
X2

K i A exp
dx ) - .jK (l x
X2

K dx ) (7.196)
 7.10 The WKB Approxi mation 257

-8 -6 -4 -2 0 2 4
x_
The Airy function Ai(x) = (1 /71") So cos (s3/ 3 + sx)ds
(see Table 7.3)

o1--=1=f-+---f--+----+----j~:=!!!Io-4_~
2

~ =-,,-
2 _ mwox2

Turning points for harmonic oscillator: ~~ = (2n + I)

FIGURE 7.36 WKB approximation for the fourth state of the harmonic oscillator, to-
gether with a graph of the Airy function . Also shown are the potential V (~) and the fourth
eigenenergy. Note the divergence of the WKB approximation at the turning point ~4 . This
calculation was performed previously by J. D. Powell and B. Craseman (Quantum Me-
chanics , Addison-Wesley , Reading , Mass ., 1965). For extensive discussion of numerical
techniques in the WKB analysis, see C. M. Bender and S. A. Orszag , Advanced Mathe-
matical Methods for Scientists and Engineers, McGraw-Hill , New York, 1978.
258 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

V(x)

E-+----+----+-----

FIGURE 7.37 Domains relevant to the WKB approximation of the transmission through
a potential barrier.

Let r denote the integral (compare with 1], p. 253)

(7.196a)

Appropriate division of the interval of integration gives the relations

exp (-l \ X

dx) = r- I
exp (1~ KdX)
(l \ dX) (-1~ KdX)
X

exp = rexp

Substituting these expressions into (7.196) gives

~II = 2r~ exp (1~ KdX) - ~ exp ( -1~ KdX) (7.197)

which allows application of the connection formulas (7.188). There results

A sin
~I = - 2rJk (lx X1

k dx - 4"Jr) - i2Ar
Jk cos (lx k dx - 4"Jr)
X1
(7.198)

We are now at the point where we must extract the incident component of ~I . If
we label the argument 1 k 1: dx - x / 4 == z and express both trigonometric terms
as exponentials, (7.198) may be rewritten

A ei z - e- i z i2Ar ei z + e- i z
(7.199)
~ = - 2rJk 2i - Jk 2

A
= i Jk (1 ) iz
4r - r e - i Jk
A (1 + )
4r r e"
iz

Far removed from the boundary, k ~ constant and z ~ -k(x - xr) - n /4. From
this we may infer that the second term in the last equation for ~I represents the in-
cident component wavefunction . Employing the expression (7.108) for the trans-
Problems 259

mission coefficient T, with the second term in (7.199) representing the incident
wavefunction and (7.194) the transmitted wavefunction, we obtain
1 1
T= =-.."-------,,. (7.200)
(r + 1/4r)2 r 2 + 1/2 + 1/16r2
It is consistent with the WKB criterion (7.166) to neglect all but the term r 2 in the
denominator of (7.200) , thereby obtaining (see Problem 7.89)

2
T = r- = exp (-21: KdX)2
(7.201)

The simplest application of this formula is in calculation of the transmiss ion

through a square potential barrier. Exact analysis gives the result (7.147). We
should find that this expression reduces to the WKB formula in the limit

«a = J2ma 2(V - E)lli2 » 1.

In this limit (7.147) gives the transmission coefficient

T:::: 16E e-4Ka

V
whereas (7.201) gives

which is seen to be in good order-of-magnitude agreement with the limiting form

of the exact result given above. Further application of the exceedingly important
result (7.201) is left to the problems . A discussion on the Feynman path integral ,
closely allied to the WKB analysis, is given in Section 7.11.

PROBLEMS

7.62 In the phenomenon of cold emission, electron s are drawn from a metal (at room
temperature) by an externally supported electric field. The potential well that the
metal presents to the free electrons before the electric field is turned on is depicted in
Fig. 2.5. After application of the constant electric field 8, the potential at the surface
slopes down as shown in Fig. 7.38, thereby allowing electrons in the Fermi sea to
"tunnel" through the potential barrier. If the surface of the metal is taken as the x = 0
plane, the new potential outside the surface is

V(x) = <I> + EF - e8x

where E F is the Fermi level and <I> is the work function of the metal.
(a) Use the WKB approximation to calculate the transmission coefficient for cold
emission .
260 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

v = <I> + E F - e&x

FIGURE 7.38 Potential configuration for the phenomenon of "cold emission." (See
Problem 7.62.)

(b) Estimate the field strength 8, in V/cm, necessary to draw current density of the
order of mAlcm 2 from a potassium surface. For]" inc [see (7.108)] use the expres-
sion ]"inc == en v, where n is electron density and v is the speed of electrons at the
top of the Fermi sea. The relevant expression for E F may be found in Problem
2.42. Data for potassium is given in Section 2.3.

Answer (partial)
(a) Using (7.201), and the form of the potential exterior to the metal given in the
statement of the problem, we obtain, for transmission at the Fermi level (V -
EF == <I> - e8x),

2 [<PIeS ]
T = exp [ -Ii 10 J2m(<I> - e8x) dx

== exp (_~ V2m <1>3 /2)

3 Ii «e
This equation is referred to as the Fowler-Nordheim equation.

7.63 An ex particle is the nucleus of a helium atom. It is a tightly bound entity comprised of
two protons and two neutrons, for which the approximate binding energy is 7 MeV.
A primary mode of decay for radioactive nuclei is through the process of ex decay.
A consistent model for this process envisions the ex particle bound to the nucleus
by a spherical well potential.if Outside the well the ex particle is repelled from the
residual nucleus by the potential barrier

2(2 - 2)e 2 A
V== - - - -
r r
12The rigid spherical well potential is described in Section 10.3. The effective one-dimensional Hamil-
tonian for the configuration at hand is given by (l 0.107) with angular momentum L set equal to zero.
This corresponds to assuming in part (b) that the bounce motion of the a-particle is through the origin.
Problems 261
V(r)
v = 2(Z - 2)e 2 ; r

FIGURE 7.39 Nuclear ex particle potential model for the process of ex decay. (See Prob-
lem 7.63.)

The original radioactive nucleus has charge Ze , while the ex particle has charge 2e
(Fig. 7.39).
(a) Use the WKB approximation to calculate the transmissivity T of the nuclear
barrier to ex decay in terms of the velocity v = .J2E/ m and the dimensionless
ratio .j(ro/rl) == cos W. What form does T assume in the limit ro --+ O?
(b) Assuming that the ex particle "bounces" freely between the walls presented by
the spherical well potential with a speed ~ 109 cmls and that the radius of the
heavy radioactive nucleus (e.g., uranium) is ~ 10- 12 em, one obtains that the
ex particle strikes the nucleus wall at the rate ~ 102 1 s-I . In each collision
the probability that the ex particle penetrates the nuclear Coulomb barrier is equal
to the transmissivity of the barrier T . It follows that the probability of tunneling
through this barrier, per second, is

and that the mean lifetime of the nucleus is

1 10- 2 1
r=-=--
P T
Use your answer to part (a) for T and the following expression for the nuclear
radius

ro=2x 1O- 13Z 1/ 3 cm

to estimate the mean lifetime for uranium ex decay.

Answer (partial)

(a)

A
q="E
262 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

Integrating, one obtains Gamow 's formula,

T = exp [ - 2A .
liv (2W - sm2W) ]

As rO --+ 0, W --+ nl2 and T ~ exp(-2n Alliv) .

7.64 Use the WKB relation (7.191) to estimate the eigenenergies of the displaced spring
potential

How do these values compare with the exact values obtained in Problem 7.16?
7.65 An electron with charge -e and mass m, constrained to move in the x direction,
interacts with the uniform electric field 8, which points in the positive x direction.
(a) Show that energy eigenstates may be written as Airy functions.
(b) Do eigenenergies comprise a continuous or a discrete spectrum? [Hint for part
(a): Set e8x - E = Kx' and then find the value of K that gives Airy's equation.
(See Table 7.3.)]
7.66 Use the WKB approximation to determine the bound-state energies of the potential
well
Vo
V(x) = -lxi , Ixl:::: a
a

V(x) = Vo =;;;1 (h)2

~ , Ixl > a

Answer
Eigenenergies appear as

En = Eo(n + ~)2/3
-3/2 3Voh
E ---
o - 8a-J2m

With Vo as given, we obtain Eo = 0.41 Vo, so there are four bound states: Eo =
0.63Eo , E1 = 1.3lEO, E2 = 1.85Eo , and E3 = 2.3IEo·
7.67 Show that for the singular potential

V =aV08(x)

boundary conditions on rp(x) become

(a) rp(O)- = rp(O)+
li2
(b) -[rp'(O)+ - rp'(O)_] = aVOrp(O)
2m
7.68 Use the result of Problem 7.67 to construct the bound state of the potential well,
V = -aV08(x) . (Hint: Set E = -lEI and look for the solution for x i= 0.)
Problems 263

7.69 Find T and R for the potential barrier V = a V08(x) .

7.70 The initial state for the harmonic oscillator
CJn
11/r(0» ) = a- g
2
/2 L
00
-In)
n=O JilT

represent s a minimum uncertaint y wave packet. The parameter CJ is real.

(a) Show that

where 10) is the n = 0 eigenstate.

(b) Show that for this state

(p ) =0

so that

and
Ii
tlx tlp = "2
This property establi shes the fact that the given superposition state represents a
wave packet of minimum uncertainty.
(c) Show that for this state

which gives a physical interpretation of the parameter CJ .

(d) Show that 1/r(t) is

11/r(t» ) = e-iwot /2e-g2 /2 exp(CJe-iwota t )10)

[Hint:

Also,

Here we have recalled that j(a t a)ln) = j(n)ln) .]

264 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

(e) Show that in the state 'I/J(f)

n
(x ) = lJ
/l£- - cos WOf
worn
(p) = -lJJ2/iworn sinwOf

so that
n
~x ~p ="2
at any time f . The packet remains a packet of minimum uncertainty for all time
and oscillates in classical simple harmonic motion. Note also that the probability
density ('I/J (x , f)I'I/J(x , f)) is a Gaussian form, centered at (x) .
7.71 In closed form the wave packet of minimum uncertainty for the harmonic oscillator
appears (at time f = 0) as

'I/J(x,O) = J ,BIrr 1/2 exp [iX:

O
- ~,B2(X - xO)2]

(a) Show that

I
(x ) =xO ~x=--
,B.,fi
n,B
(p) =PO ~p=-
.,fi
Hence Sx ~P = n12, and we are justified in calling 'I/J a packet of minimum
uncertainty.
(b) Show that in the initial state above,

and

(xp + px) = 2xopo

(c) In order to establish that 'I/J(x) remains a packet of minimum uncertainty for all
time, one must show that ~x ~p is constant. Recalling the equation of motion
for the average of an operator (6.68),

d 1"
df (A) = ih ([A , H])
8A)
+ /\ at
and introducing the operator

ij = xp + px - 2(x) (p)

show that
Problems 265

!!:...-(6.X)2 = ~
dt m
d
dt (6.p)2 = -mw2 (1j)
d 2(6.p)2
- (1j) = - - - 2mw 2(6.x)2
dt m

Using these results , show that for the initial state above, (6.x)2 and (6.p)2 are
both constants in time.
(d) Show that 1{f(x, 0) is an eigenfunction of the annihilation operator a. What is the
a
eigenvalue of in this state? [Hint : Employ the representation of 1{f(x, t) given
in Problem 7.70.]
Note: In quantum optics the radiation field is viewed as a collection of har-
monic oscillators. In this representation, the appropriate generalization of the
state of minimum uncertainty is called the coherent state.
7.72 The Hamiltonian of a particle is

where A and B are constants. What are the energy eigenvalues of the particle? (Hint:
Introduce the operator iJ = aa
+ {3 ; iJt = t + {3 .)aa
7.73 What is the form of the potential that gives the Gaussian probability density with
variance a 2 in the ground state?

7.74 The reflection coefficient for the smooth potential step

Vo
V (x) - -:-I-+-e-=--:-:Y"'-
x

for E > Vo is l3

R = (Sinh[Jr(kl - k 2)/ y ]) 2
sinh[Jr(kl + k2)/Y]

The energy of incident particles at x = -00 is

while the kinetic energy of transmitted particles at x = +00 is

(a) Make a sketch of the potential V(x) and indicate roughly the length scale of
potential and its relation to the wavenumber y .

13The coefficients Rand T given in Problems 7.74 and 7.75, respectively, are calculated in L. Land au
and E. Lifshitz, Quantum Mechanics, 2d ed., Addison-Wesley, Reading , Mass., 1965.
266 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

(b) Show that in the limit that V(x) approaches the simple step (Fig. 7.18), R goes
to the value given by (7.125).
(c) Show that the classical value of R emerges for wavelengths small compared to
the potential scale of length.
7.75 The transmission coefficient for the symmetric potential hill

V(x) = Vo
cosh 2 (y x)

for E < V is

where

Incident and transmitted particles at x = -00 and x +00, respectively, have

energy

(a) Sketch the potential and indicate roughly the length scale of potential and its
relation to the wavenumber y.
(b) Show that the classical value of T emerges for values of y appropriate to the
classical domain .
(c) Formulate an expression for the next-order approximation to the entirely classical
result (b) for the transmission coefficient using the WKB analysis .
(d) Obtain an explicit expression for the transmission coefficient in the near-classical
domain that you have formulated in part (c) by expanding the exact formula for
T given in the statement of this problem .

Answer (partial)
(d) The classical limit is attained in the limit y -+ O. From the given expression for
T , we obtain

eJrk / y I
T ~ = ----------
- eJrk/Y+eJrp /2 l+exp(rr/yTi)(Tipy/2-TiK)

~ exp [- ;Ti (J2m Vo- J2mE)]

7.76 A uniform homogeneous beam of electrons is incident on a rectangular potential

barrier of height V . Each electron in the beam has energy E > V and unit amplitude
wavefunction
Problems 267

If the transmitted electrons have wavefunction

!Ptrans = !Pm = 0.97 e ikt X

(a) What is the total wavefunction !PI, of electron s in region I?

(b) If E = 10 eV and V = 5 eV, what is the minimum barrier width compatible with
the information given above?

Answer
(a) In general for unit amplitude incident waves,

!PI = eik 1X ± i-Jii e- ik t X

!Pm = .ITeik 1X
where the ± signs refer to the sign of sin(2k2a) . (See Problem 7.44.) It follows
that

T = (0.97)2 = 0.94, R =I- T = 0.06, and -Jii = 0.24,

so that

(b) We then find that sin2(k22a) = 8R IT = 0.51, k22a = 0.80 < rr/2, and

/2m(E - V) 8 I
k2 =V ti2 = 1.14 x 10 em

Therefore,

2a = 0.70 A
7.77 (a) What are the values of k and K at the "turning points" of a potential hill or po-
tential barrier?
(b) What are the values ofWKB wavefunctions, I!pIl, l!pn I, l!pm I, at these points (for
either bound or unbound states)?
(c) A student argues the following : We see from part (b) of this problem that WKB
wavefunctions blow up at the turning points and are therefore invalid. Such wave-
function s cannot be of any use. Is the student correct? Explain .

Answers
(a) k = K = 0 at turning points.
(b) I!PI I = I<pJII = l!pm 1= 00 at turning points .
(c) The WKB wavefunctions are valid in domain s removed from the turning points
(where they were derived from the Schrodinger equation) . For bound-state prob-
lems these solutions give an estimate for eigenenergies. For unbound-state prob-
lems they give an estimate for transmission and reflection coefficient s.
268 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

7.11 • PRINCIPLE OF LEAST ACTION AND FEYNMAN'S PATH

INTEGRAL FORMULATION

Action Integral and the Lagrangian

Classical dynamics may be formulated in terms of Hamilton's principle of least
action. This principle states the following : the classical trajectory between two
fixed points Xl (td and X2(t2) renders the integral

12
S=
1 I)
L[x(t) , x(t)] dt (7.202)

a minimum. Here we have written

L(x ,x) = T(x,x) - Vex) (7.203)

for the Lagrangian, where T represents kinetic energy and V potential energy and
a dot represents differentiation with respect to time. (Whereas these expressions
are written in one dimension, they are easily generalized to three dimensions.)
As noted above, Hamilton's principle states that of all possible paths between
the fixed points Xl and X2, the path which minimizes the integral (7.202) is the
actual physical path between these two points . Thus Hamilton's principle may be
restated as follows : the physical path between the points Xl and X2 renders the
integral (7.202) stationary. That is,

12
8
1
II
L(x, x) dt = 0 (7.204)

where 8 represents an arbitrary, infinitesimal variation about the true motion of

the system. The variable S in (7.202) is called the action, and the rule (7.204) is
alternatively called Hamilton's principle or the principle ofleast action.

Relation to the Hamiltonian

The relation of the Lagrangian (7.203) to the classical Hamiltonian (1.13) (for the
present one-dimensional configuration) is given by

. aLex, x) .
H(x,p)=x . -L(x,x) (7.205)
ax

Here we have written

aLex, x)
p= ax (7.205a)

for the momentum conjugate to x. Employing the relation (7.205) in (7.202) and
following a variational calculation similar to that described in Problem 4.28 leads
7.11 Principle of LeastAction and Feynman's Path Integral Formulation 269

to Hamilton 's equations (1.35).14Thus Hamilton's principle (7.204) is an alterna-

tive description of classical mechanics .

Minimum Action
Let us examine the meaning of Hamilton's principle by way of a specific case.
Thus, for example, consider the harmonic oscillator described previously in Sec-
tion 7.2. Recalling (7.203) the corresponding Lagrangian is given by

I .2 I 2
L = -mx - -Kx (7.206)
2 2
The integral (7.202) then becomes

(7.207)

Consider the specific case that at t = 0 the particle is at x = Xo with energy E.

The motion for this problem is then given by

x(t) = Xo cos cot

I 2 2 K
E = -Kxo w =- (7.208)
2 ' m
Substituting these values into (7.207) with t1 = 0 and t: related to time t, we find

E .
S(t) = - - sm2wt (7.209)
2w
Now in what sense is this value of the action minimum ? To answer this question,
we consider the varied motion
. t»t
x(t) = xocoswt + E2xosm- (7.210)
E

where E « 1. The preceding function is seen to represent a high-frequency,

small-amplitude oscillation, which follows the unperturbed cos tot motion. See
Fig. 7.40.
Constructing the Lagrangian for the motion (7.210), we find [keeping terms of
O(E 2)]

wt]
L = Lo - 2E E [ sin cot cos -tot + E cos tot sin -totE E 2
- - cos -
E 2 E

where Lo is the Lagrangian corresponding to the unperturbed motion (7.208).

Integrating the preceding over time and passing to the limit E --+ 0, we see that
contributions from the first two of the bracketed terms "wash out," whereas the
14See, for example, H. Goldstein, Classical Mechanics, 2d ed., Addison-Wesley, Reading, Mass .,
1980. Sec . 8.5.
270 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

x(t)

1.0

O.O-+---------'k------.,-----....",.----_

-1.0

FIGURE 7.40 High-frequency, small-amplitude motion of (7.210).

third term yields

S = So + -EE (wt- + sin -cot cos -wt) = So + -1E2 Et + 0 (E 3 )

3
(7.211)
2w E E E 2

where So is the action (7.209). Thus to O(E 2 ) we may write

S > So (7.212)

We may conclude that the action S, (7.211), corresponding to the varied mo-
tion (7.210), is larger than So, (7.209), corresponding to the true motion (7.208),
thereby corroborating Hamilton's principle.

Feynman Path Integral

In 1948, R. P. Feynman 15 presented a new formulation of quantum mechanics
based in large part on the preceding classical concepts.l"
Our discussion of this formalism begins with the solution of the initial-value
problem (3.70) written in Dirac notation.

(7.213)

In coordinate representation (see Appendix A) this equation becomes

(xl1Jr(t)) = J
(xle-itH /filx /) (x /I1Jr(O)) dx ' (7.214)

Note that in this expression H = H(x , p), whereas in (3.70) H = tu», -ilia/ax) .
(See Problem 7.86.) We may write (7.214) in its equivalent form

15R. P. Feynman, Rev. Mod. Phys. 20,367 (1948).

16For further discuss ion, see R. P. Feynm an and A. R. Hibbs, Quantum Mechanics and PathIntegrals,
McGraw-Hill , New York, 1965.
7.11 Principle of Least Action and Feynman's Path Integral Formulation 271

ljI(x, t) = f (xle- itH /nl x'}ljI(x', 0) dx' (7.215)

In the Feynman description one assumes the following form for the propagation
term in (7.215) :

(xle-itH /nlx ') = f exp[~[,t L(X,X)dt].D[X' ,X] (7.216)

all
paths

The integration on the right represents a sum of the exponential over all paths
x = x(t) between x' and x and .D[x', x] is a differential-like measure related to
the sum over paths . Furthermore, L(x, x) is the classical Lagrangian. Over each
path (7.216), L(x , x) is evaluated from the given x = x(t). Substituting (7.216)
in (7.215) gives

ljI(x,t) = ff eiS/nljl(x ',O)dx' .D[x',x] (7.217)

all
pa ths

which is Feynman's solution to the initial-value problem, where S is the action

(7.202) .
Consider that it is known that the particle is at Xo at t = O. Then

ljI(x',O) = 8(x' - xo)

and (7.217) reduces to

ljI(x ,t) = f [exp~ltL(X,X)dt].D[X' X]

all
(7.218)

paths

This expression represents an extension of the WKB approximation (7.167) to the

full quantum domain.

Domains of Large and Small Contribution to the Wavefunction

In the integration over all paths in (7.218), most domains of integration do not
contribute for the following reason . Away from the region where action is sta-
tionary, a small change in path causes a large change in action compared with h,
The corresponding rapid fluctuation in the exponential in (7.218) causes cancel-
lation, thereby diminishing contribution to the wavefunction. On the other hand,
for paths near the classical orbit , the action is stationary and the exponential term
likewise does not vary, resulting in a net contribution to the wavefunction.
We may conclude that the main contribution to the wavefunction (7.218) stems
from paths lying near the classical orbit. Thus, to lowest approximation, (7.218)
272 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

returns the quasi-classical WKB expression (7.167):

1/!(x) = Ae(i /Ii)S(x) (7.219)

In this expression the constant Xo in (7.218) was absorbed in Sex) . Furthermore,

the action S (x) in (7.219) is evaluated on the classical orbit.
The reader will note that there is a difference between the action integral in
(7.204) and that given by the result (7.172). The latter expression may be rewritten

So(x) = ± 1 1
2T dt

Now note that

L = T- V =2T- E

Thus, with the constraint 8E = 0 in the variation (7.204), the principle of least
action becomes

8 f 2Tdt = 0

This relation contains the action relevant to (7.172).

PROBLEMS

7.78 The scattering amplitude 5 , in one dimension, may be defined by the equation

Vtran s = 5 Vine

For transmission above a potential well, one finds that the bound-state energies of the
well are given by the negative real zeros of 5- 1 (E) .
(a) What is the transmission coefficient T in terms of 5 corresponding to the config-
uration shown in Fig. 7.14? [See (7.112).]
(b) Consider the transmission past the rectangular potential described in Section 7.8.
(See Fig. 7.27.) Construct an expression for 5 for this configuration.
(c) Using your answer to part (b) and the rule stated above, obtain the energy eigen-
values of the well (as discussed in Section 8.1).

Answers
(a) T = k21512
kl
In the event that V (+00) = V( -00), then

T = 151 2
(b)
Problems 273

(c) Setting k)2 ~ _K 2 = -2m IEI //i,2, and k22 ~ k 2, we find that the poles of S
occur at

2Kk 2 2 2ma2V
tan2ka = k2 _ K2' (ak) + (aK) = ~
which are in the desired form (8.64).

7.79 (a) Do the two integral expressions given in the connecting formula (7.188a) join
smoothly at x = X) ?
(b) Explain your answer.

Answers
(a) They do not join smoothly.
(b) These two expressions represent asymptotic forms of the Airy function on either
side of the turning point x = x) . The (approximate) wavefunction at the turning
point is the continuation of the Airy function from these asymptotic values.

7.80 Electrons in a beam have the wavefunction

The beam passes through a potential barrier and electrons emerge with the wavefunc-
tion

!Ptrans = 9 (x )eik2X
In these expressions f(x) and g(x) are real functions and k) and k2 are real con-
stants . Show that

7.81 A particle of mass mo confined to a potential well of width ao and depth Vo is known
to have NO » I bound states . How many bound states N does the well have if:
(a) a = 2aO and V = 25VO?
(b) a =aO/2and V =4VO?
(c) m = O.lOmO. V = 30VO, and a = O.577aO?
7.82 In physics. one requires that fundamental equations of motion be reversible in time .
In quantum mechanics, the operation of time reversibility is given by 1fr(r, t) ~
1fr*(r, -t) . Show that the quantum continuity equation (7.97) [with p given by (7.95)
and J given by (7.107)] remains invariant under this operation.

7.83 Consider the plane wave

1fr(x , t) = A exp i(kx - wt)

In the operation of time reversibility, what is the effect of the successive operations:
(a) t ~ -t?
(b) 1fr ~ 1fr*?
274 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

Answers
(a) This component of the transformat ion causes the wave to move in the reverse
direction .
(b) This component of the transformation causes momentum hk: to reverse direction .
Both of these properties come into play in classical dynamic reversibility.

7.84 (a) Consider a particle of mass m in the complex potential field

iii
<f> = V(r) + "2w(r)
where V(r) and w(r) are real functions . What form does the continuity equation
(7.97) assume for this potential ?
(b) Offer an interpretation for the field w(r) .
(c) Is the new continuity equation found in part (a) time-reversible ?
(d) Is the related Hamiltonian Hermitian ?

Answers (partial)
(a) Repeating steps leading to (7.105), we obtain
ap
-+V ·J=wp
at
(b) The complex potential implies a source of particles (in violation of conservation
of matter) .
(c) The equation does not obey time reversibility.

7.85 (a) Show that the Gaussian wave packet given in Problem 7.71 satisfies the eigen-
value equation

a1fr (x ) = c1fr(x)
(b) Under what conditions will the eigenvalue c be zero?
(c) If conditions of part (b) are satisfied, what function does the Gaussian wave
packet reduce to?
7.86 Given that [see (A.7)]

(x lp lx ' ) = -ili~o(x - x')

ax
show that for free-particle motion

(xle-itH(x ,p)/nl x' ) = [ex p -~H (x, -iii a:)] o(x - x')

Answer
For free-particle motion we label
it _ _ -2
h,H(x, p) == ia P
We may then write
Problems 275

For the second term in the sum we obtain

/
(xlp 2Ix ) =J dx" (xlplx")(x"lplx
/)

=
J ." a < - x
d x ,,[ -1,,-o(X
ax
II)] [-1,,-8(x
."aax" " - x) I ]

." a
= -1"-
h
J" "[ ."
a " - x),]
dx 8(x - x ) -1,,-8(x
h"

= (-ilia:r 8(x -x')

Following in this manner we obtain

(xle-iap2Ix/) = exp [ -La (-iii a:) 2] 8(x - x')

which was to be shown .

7.87 (a) What is the value of the classical action So(xo. x) corresponding to the free-
particle motion of a particle of mass m which is at Xo at time t = 0 and x at
t > O?
(b) Introduce an infinitesimal perturbation about the motion found in part (a) and find
the new action S(xo. x). Following the steps (7.210 et seq.), show that S > So.
(c) What is the lowest-order wavefunction, 1fr(x. t), corresponding to the classical
limit for this problem?
7.88 (a) Do the action functions So. SI, . . . in the expansion (7.169) have the same di-
mensions ?
(b) If not, what are their dimensions?

Answers
(a) They are not the same.
(b) From (7.167) we see that the dimensions of S(x) are those of Ii. Thus, from the
series (7.169) we conclude that [linSn] = Ii or, equivalently, [Sn] = Ii 1-n.

7.89 In developing the WKB expression for the transmission coefficient (7.201), it was
assumed that r » 1, where r is given by (7.196a). Show that this assumption is valid
within this approximation.

Answer
For slowly varying V(x) of a potential barrier, one obtains r :::::: (2a jli).j2m(V - E),
which , in the limit V » E, grows large. In this same limit T ~ 0 [see (7.201)],
which corresponds to the classical value. This, we recall, is the domain of validity of
the WKB expansion.
7.90 An electron beam is incident on a rectangular barrier of width 5 Aand height 34.8 eV.
The electrons have energy 2.9 eV. If the incident electron current is 10- 2 cm- 2 s-l,
what is the reflected current, within the WKB approximation? Is this a good approx-
imation for the problem at hand? Why?
276 Chapter 7 Additional One-Dimensional Problems. Bound and Unbound States

7.91 Again consider the configuration of Problem 7.62. However, now let the external
potential be given by the following parabolic form:

EF + <1» 2
V(x) =(~ (xO -x) ,

V(x) = 0, xo:s: x

(a) Draw the appropriate figure for this problem. Identify the Fermi energy.
(b) Calculate the transmission coefficient for emission through this potential barrier
in the WKB approximation. Call your answer Ta . How does Ta compare to the
transmission coefficient corresponding to the linear potential of Problem 7.26?

7.92 A particle of mass m is confined in a finite potential well of depth IV I and width 2a,
as depicted in Fig. 7.27. (See also Fig. 8.1.)
(a) Employ the WKB method to obtain an expression for the bound-state eigenener-
gies of the particle.
(b) If

13 ) h2
IVI = ( 2 8m(2a)2

how many bound states (En < IVI) does the system have?
(c) In what extreme do you expect your answer to agree with exact quantum me-
chanical values? (See Section 8.1.)

7.93 Derive Airy's equation (7.182a) from either of the given transformations preceding
(7.I82a).

Answer
Let Q' =
2m F /!i2, in which case the said transformation appears as

Furthermore,

Substituting these forms into (7.181) gives Airy's equation (7.182a).

7.94 Show that the transmission coefficient, T = I, for the potential I?

where a is a scale length of the potential.

7.95 (a) Obtain an expression for the commutator [p, H], for a one-dimensional har-
monic oscillator.

17R. E. Crandall, J. Phys. A16, 3005 (1983).

Problems 277

(b) The oscillator is in a state with uncertainty in momentum, c flp 10- 2 eV,
where c is the speed of light. Given that the spring constant K = 3.2 x
108 eV I cm2 and (x) = 100 A, what is the related uncertainty in energy,
flE (in eV), of the oscillator in this state?

7.96 (a) A one-dimen sional current of electrons , each of energy 2lR, is incident on a rect-
angular barrier of width (Jrao/2) and height R, where aO is the Bohr radius and
lR is the Rydberg. What is the value of the transmission coefficient, T , for this
system? State units of your answer.
(b) If particle current ]inc = 1.2 x 104 s-I is incident on the barrier described above,
what is the reflected current , ]refl?
Hint: In part (a), recall n2 / 2m = lRao2
CHAPTER
Finite Potential Well, Periodic

8 Lattice, and Some Simple

Problems with Two Degrees
of Freedom

8.1 The Finite Potential Well

8.2 Periodi c Lattice. Energy Gaps
8.3 Standing Waves at the Band Edges
8.4 Br ief Qualitative Description ofthe Theory of Condu ction in Solids
8.5 Two Beads on a Wire and a Particle in a Two-Dimensional Box
8.6 Two-Dimensional Harmonic Oscillator
8.7 Linear Combination ofAtomic Orbitals LCAO) Approximation
8.8 Density ofStates in Various Dimensions

In this chap ter we meet perhaps the most eminently successful application of
quantum mechanics to a one-dimensional configuration. This is the problem ofa
charged particle in a periodic potential. When coupled with the exclusion princi-
ple for electrons, the analysis ofthis configuration provides a deep understanding
of the process of conduction in solids. Some elementary problems in two dimen-
sions are given, together with a discu ssion of degeneracy in quantum me chanics.
The chapter continues with an approximation technique important to molecular
and solid-state physics which ca rries the acronym LCAO. A concluding section
describes density ofstates in various dimensions.

8.1 • THE FINITE POTENTIAL WELL

Eigenstates
Scattering from a rectangular potential well was discussed previously in Sec-
tion 7.8. The configuration is depicted again in Fig. 8.1. The scattering, unbound
states correspond to a continuum of eigenenergies [e.g., (7.129)]:

If we seek solutions to the Schrodinger equation for negative energies, E < 0,

only a finite, discrete number of eigenstates are found. For the three regions de-
picted in Fig. 8.1, the Schrodinger equation and corresponding solutions are (for
lEI < lVI, E < 0, V < 0):
278
8.1 The Finite Potential Well 279

Continuum
unbound-state
energies
e, >0

Discrete Vex)
bound-state
energies
En < 0 E 2 t----~----I
E 1 t----~----I
.....- ......- ..... -IV\
(a)
Region I Region II Region III
~(x)

(b)

FIGURE 8.1 Finite rectangular potential well. (a) The potential function V (x) and en-
ergy spectrum. (b) Typical structure of a bound eigenstate. Function oscillates in region II,
where kinetic energy is positive, and decays in regions I and III, where kinetic energy is
negative.

Region I: x < -a

ti2
- 2m CPxx = -IElcp, CPxx = K 2 cp
(8.1)
KX
ti2 K2
CPI == Ae , 2m == lEI> 0

Region II: -a ~ x ~ a

ti2
--CPXX == (IVI - IEl)cp,
2m
(8.2)
CPu == B e i kx + C e- i kx ,
280 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Region III: x > a

li2 2
- 2m rpxx = -IElrp, rpxx = K rp
(8.3)
KX
li 2 K2
rpm = De- , 2m = lEI> 0

First we note that k and K obey the constraint

(8.4)

The coefficients A, B , C, and D determine the eigenstate corresponding to the

eigenenergy li 2 K 2 / 2m . These coefficients are determined by the continuity condi-
tions at x = a, x = -a . Equating rp and its first derivative at these points gives

Ae- Ka = Be- ika + Ceika

Be ika + Ce- i ka = De- Ka
(8.5)
KAe- Ka = ik(Be- i ka _ Ce i ka )
ik(Be ika - Ce- ika ) = -K De- Ka
These are four linear, homogeneous equations for the four unknowns A, B, C,
and D . They may be cast in the matrix form (where the right-hand side denotes
the null column vector)

DV ==
( ~Ka
e
-,.
Ke
_e- i ka
eika
_ike- ika
_e ika
e- i ka
ike ika
_e- Ka
o )C) B
C =0 (8.6)
0 ike ika _ike- ika Ke- Ka D

which serves to define the coefficient matrix D and the column vector V .
Cramer's rule tells us that this system has nontrivial solutions (i.e., other than
A = B = C = D = 0) only if the determinant of the coefficient matrix vanishes.

det D = 0 (8.7)

After a little manipulation (8.6) is rewritten

G o
G* o
_e ika K -Ka (8.8)
--e
ik
o
8.1 The Finite Potential Well 281

where

G == (K + ik)e
ika (8.9)

(Note the rearrangement of the column vector V.) Expanding about the fourth
column, one obtains

l!
G

det D ~ d"
(G*
G*
o
o
K -Ka
-e
ik
o

o
o
_"-:_"
o
1 ~ [G' - (G')']
2
C:l ,-2<.
ik
(8.10)

This is zero when

(8.11)

or, equivalently, when

G=±G* (8.12)

Rewriting (8.9) as

G == (K + ik)e ika == IGlei(ka H)

k
tan e = - (8.13)
K

allows the conditions (8.12) to be recast in the form

ei(ka+<f» = ±e-i(ka+<f» (8.14)

The positive root gives ka + IjJ = 0, or, equivalently,

k
tan IjJ = -K = - tan ka (8.15)

This may be put in the more normal form

G
kcotka = -K , -=1 (8.16)
G*
The negative root gives ka + IjJ = TC /2 or, equivalently,

tan IjJ = ~= tan (~ - ka ) = cot ka (8.17)

This may also be put in the more normal form

G
k tanka = K, -=-1 (8.18)
G*
282 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

The values of k that make det:D = 0 fall into two categories. These are the
solutions to (8.16) and (8.18), respectively. From our starting matrix equation
(8.8) we see that these values of k imply the relations

B G
- = - - =±l (8.19)
C G*
The minus sign corresponds to the roots (8.16). Substituting this value (B = -C)
into the last two equations of the set (8.5) gives [with (8.16)]

CAD «a
- = -1, - = - - = -2i sin(ka)e (8.20)
B B\ B
Substituting these values into (8.1) et seq . gives the eigenstate

rpI = - 2i B sin(ka )eK(x+a) )

rpn = 2i B sin kx kcotka =-K (8.21)
rpm = 2i B sin(ka)e-K(x-a)

This state has odd parity; that is,

rp(x) = -rp(-x) (8.22)

The second class of solutions corresponds to the plus sign in (8.19) and stems
from the roots (8.18) . Substituting this value (B = +C) into the last two equations
of the set (8.5) gives

C A D
- =+1 , - = - = 2cos(ka)e Ka (8.23)
B B B
The corresponding eigenstate is [with (8.18)]

rpI = 2B cos(ka)eK(x+a) )
rpn = 2B cos kx ktanka = K (8.24)
rpm = 2B cos(ka)e-K(x-a )

This state has even parity .

Since both eigenstates (8.21) and (8.24) are bound states , we may impose the
normalization condition

(8.25)

This determines the remaining constant B. (See Problem 8.53.)

Next we tum to construction of the eigenenergies corresponding to the eigen-
states (8.21) and (8.24). The energy is directly determined from K . For the even
eigenstates, the eigenenergies are determined from (8.4) and (8.24). Written in
8.1 The Finite Potential Well 283

terms of nondimensional wavenumbers,

~ = ka , 7) = «a (8.26)

these equations appear as

~tan~=7) even eigenstates

2 (8.27)
2 22ma 1V1
~ + 7) = Ti2 (1 + tan 2 ~ = 1/ cos2~)

For a given potential width 2a, depth 1V I, and particle mass m , (8.27) describes a
circle of radius p , in Carte sian ~ 7) space. The intersections of this circle (in the first
quadrant) with the graph of the first equation of (8.27) determine the eigenenergies
corresponding to the even eigenstates (8.24) . This graphical technique is sketched
in Fig. 8.2 for the case p slightly less than it . The sketch tells us that for this value
of p, the finite potential well has only one bound even eigenstate.
The energies of the odd eigenstates (8.22) are the intersections of the two
curves

~ cot s = -7) odd eigenstates

~2 + 7)2 = p2 (l + cor' ~ = 1/ sin2 ~) (8.28)

7T/2 7T 37T12 27T 57T/2 g

~I
IEII 1-1
2a

FIGURE 8.2 The curves TJ = ~ tan ~ and the circle ~ 2 + TJ2 = p2 for the case p slightly
less than n . Intersections in the first quadrant give bound-state eigenenergies for the po-
tential well Hamiltonian which correspond to even eigenstates .
284 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

~I
I---l
2a

FIGURE 8.3 The curves TJ = -~ cot~ and the circle ~ 2 + TJ2 = p2 again for the case
p slightly less than n . Intersections in the first quadrant give bound-state eigenenergies
for the potential well Hamiltonian which correspond to odd eigenstates. Note that E2 lies
higher than the ground state E I .

These curves are sketched in Fig. 8.3 for the case p slightly less than tt . For this
choice of data, we see that there is only one bound odd eigenstate . These two
lowest-energy eigenstates , (8.21) and (8.24), are sketched in Fig. 8.4.
At this point we wish to consider again the difference between the unbound
scattering states of Chapter 7 and the bound states just encountered. The continu-
ity conditions on the wavefunction tp and its derivative, together with the statement
of conservation of energy, determine eigenenergies and eigenstates . For scatter-
ing states, the continuity conditions (7.139) are in the form of an inhomog eneous
matrix equation,

(8.29)

where, for example, the column vector 'V is

'V=
BIA)
CIA (8.30)
( DIA
FIA
The solution to (8.29) is

(8.31)
8.1 The Finite Potential Well 285

x x

Potential
Potential well
well

'----+----' -IVI L.---+----'-IVI

(a) (b)

FIGURE 8.4 First two bound eigenstates for the potential well problem. For
2ma 2 1V1/ n2 < ]f2, these are the only bound states.

For these unbound scattering states, conservation of energy serves only to relate
wavenumbers connected to distinct potential domains . The eigen-zj-values com-
prise a continuum. For each such ki value, there corresponds an eigenstate of the
form (7.138).
For bound states the continuity conditions (8.5) are in the form of a homoge-
neous matrix equation (8.6),

,v(K)V = 0

which has nontrivial solutions (V i= 0) only if

det,v(K) = 0 (8.32)

This dispersion relation restricts the eigen-K-values to values that obey certain
transcendental relations [the first equation in (8.27) and in (8.28)]. In addition , K
is further restricted by the conservation of energy statement-namely, the second
equation in (8.27). The intersections of this circle (depicted in Figs. 8.2 and 8.3)
with the said transcendental curves generate a discrete spectrum of eigenenergies

Let us consider the time dependence of the eigenstates corresponding to the

finite potential well. The bound time-dependent eigenstates appear as

En < 0 (8.33)
286 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

with ({In (x) given by (8.1) et seq. For positive energy, the unbound time-dependent
eigenstates form a continuum,

(8.34)

where ({Jkl (x), for example, is of the form (7.138) with the modification V -+
-IVI·
To employ the superposition principle in problems relating to the finite poten-
tial well, one must call on the finite number of bound states, and infinite contin-
uum of unbound states.'

The E = 0 line
As stated above, energies relevant to the finite potential well are directly obtained
from K or, equivalently, 1]:

These energies are measured with respect to the top of the well taken as the E = 0
line. It is sometimes convenient to measure energies with respect to the bottom of
the well as the zero energy line. The energies, E' , measured with respect to the
bottom of the wall are directly obtained from k or, equivalently, ~ :

See (8.2) and Fig. 8.5.

In either the modeling of one-electron atoms or the analysis of the present finite
potential well, energy measured from the top of the well, IE I ex 1]2, represents
binding energy. The energy E' ex ~2 measures energy from the bottom of the well
and, in accord with the infinite potential well (Section 4.1), is representative of the
eigenenergy of the particle. This representation is also relevant to the modeling of
quantum wells in superlattice structures where the E' = 0 line might, for instance ,
be set equal to the minimum energy of the conduction band. (See Sections 8.8 and
12.9.)

PROBLEMS

8.1 A deuteron, which is a neutron and a proton bound together, has only one bound state.
Assume that the potential of interaction between the two particles may be described

I Note the continuum of unbound states developed in this chapter excludes states with negative k in
region III. The states discussed are appropriate to the superpo sition of a wave packet incident on a
potential barrier from the left. For the superposition of a state with zero average momentum , one must
include the negative k waves in region III.
Problems 287

~=IVI> o
E=O

} IEI-.'.' /
IVI

} E' = IVI-
E'= 0
-v
I I

11

IE,I= (11\/p)2IVI E; = (~I /p)2 V

(a) (b)

FIGURE 8.5 Relative orientations of bound-state energies for the finite one-dimen sional
well.

as a square well. The effective mass of the system is 0.84 x 10- 24 g. The range of
nuclear force is approxim ately 2.3 x 10- 13 em, while the ground state of the deuteron
is 2.23 MeV below the zero-energy free-particle state. Assuming that only the odd-
parity solutions are permitted for this case, estimate the depth of the potential well,
IV I, which you may take to be large compared to the binding energy of the system .
8.2 An electron is confined to a potential well of finite depth and width, 10- 9 em. The
eigenstate of highest energy of this system corresponds to the value ~ = 3.2.
(a) How many bound states does this system have?
(b) Estimate the energy of the ground state with respect to the zero energy line at the
bottom of the well. Express your answer in eV.

8.3 Show that the graphical solutions of Figs. 8.2 and 8.3 give the eigenenergies of a one-
dimensional box, in the limit that the well becomes infinitely deep.

Answer
In the said limit p -+ 00, the circles of constant p intersect the tan and cot curves on
the vertical asymptotes
288 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

nit
~=2 (n = 1,2, . .. )
[Compare (4.12). See also Problem 8.58.]
8.4 Given that

for an electron in a potential well of depth IV I and width 2a = 10- 7 em, if a 1OO-keV
neutron is scattered by such a system, calculate the possible decrements in energy that
the neutron may suffer.
8.5 For the potential well described in Problem 8.4, what is the parity of the eigenstate of
maximum energy? How many zeros does this state have?
8.6 Consider a rectangular potential well of depth IV I and width 2a, such that
2ma 2 1V1/1i2 = (8Jr/18)2 . The lowest-energy normalized bound state, rpl, has
wavenumber k ::::: tt / 4a. Let ip be a wavefunction that is a square wave of height
1/ J4ll and width 4a. The centers of ip and the rectangular potential well are coinci-
dent. At t = 0 a particle of mass m is in the state

1fr(x ,O) = 3rpl + 4ip

5
At time t = 0:
(a) What is the expectation of momentum of the particle?
(b) What is the expectation of energy?
(c) What is the parity of the state?
(d), (e), (f) Repeat parts (a), (b), and (c) for t > O.
8.7 Consider the semi-infinite potential well

l
OO' x < 0
Vex) = -lVI, O ::::x ::::a
0, a < x

(see Fig. 8.6).

(a) Using the solutions to the finite potential well (width 2a) developed in the text,
sketch the first three eigenfunct ions of lowest energy for a particle in this well.
(b) Which ground-st ate energy is lower-that of the finite potential well (width 2a)
or that of the semi-infinite well (width a) ?
(c) Are the eigenfunctions you have sketched eigenstates of the Hamiltonian appro-
priate to the finite potential well?
8.8 An electron is trapped in a rectangular potential well of width 3 A and depth 1 eV.
What are the possible frequencies of emission of this system (in hertz) ?
8.9 Establish the following criteria for the number of bound states in a finite potential well:
(a) (nJr)2 < p2 < (n + 1)2Jr2 (n + 1 symmetric states).
(b) (n - !)2Jr2 < p2 < (n + !)2Jr2 (n antisymmetric states) .
 8.2 Periodic Lattice. Energy Gaps 289

FIGURE 8.6 Semi-infinite potential well and its companion finite potential well. (See
Problem 8.7.)

(c) Total number of bound states = maximum integer (p in) .

(d) Show that p2 = a2 1VI, where a == alao, V == VIR (We recall that ao represents
the Bohr radius and IR. represents the Rydberg.)
(e) Show that £' = 1i2~2 12ma 2 = ~2 l a2 .

8.2 • PERIODIC LATTICE. ENERGY GAPS

In this section we consider the problem of a particle in a periodic potential. This

is of extreme practical importance in the theory of conduction and insulation in
solids.
Consider the simple model of a solid (more precisely, a metal) in which the
positive ions comprise a uniform array of fixed sites. The valence electrons are
assumed to be free. They are the conduction electrons. For sodium, for instance,
there is one free electron per ion. Each such electron finds itself in a periodic
potential supported by the ions. Such a one-dimensional potential configuration
is depicted in Fig. 8.7.
If the distance between sites is d, then inside the metal the potential is periodic
in the distance d.

Vex) = vex + d) (8.35)

290 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

:" 1O~3 POsitive} ....

Ion sues '-....-'
d

I-
FIGURE 8.7 Periodic potential that an electron sees in a one-dimension al crystalline
solid.

A simple potential function that maintains this periodic quality and all the salient
properties of the more realistic potential sketched in Fig. 8.7 is the Kronig-Penney
potential, depicted in Fig. 8.8. The periodic property of Vex) as given by (8.35)
fails at the ends of the lattice. To remove this difficulty the model is further sim-
plified. This simplification derives from the fact that there are an overwhelm-
ingly large number of ion sites in the length of the sample. The change in the
character of the potential at the ends of the sample is therefore relatively unim-
portant to the transport properties of an interior electron. For this reason we
change the ends of the sample to best facilitate analysis. It is assumed that when
an electron leaves the end of the sample, it reenters the front of the sample.
This idea is best realized if the one-dimensional potential function is assumed
to lie on a circle of radius r which is very large compared to the distance be-

I. b .1

FIGURE 8.8 The Kronig-Penney model for a potential due to fixed ion sites separated
by the distance d. The dashed curve represents a hypothetical periodic u component of the
Bloch function cp = u(x) exp(ikx) . The eigenfunction cp (8.48) et seq. and corresponding
dispersion relations, (8.53) and (8.55), are obtained by matching u and u' at x = 0 + e to
their respective values at x = d - e and matching cp and cp' across the potential barrier at
x =a .
8.2 Periodic Lattice. Energy Gaps 291

FIGURE 8.9 Ring model of a one-dimensional periodic potential. Black dots represent
positive ion sites. For N sites in all, and N » 1, Nd :::: 27fT .

tween ion sites, d (see Fig. 8.9). The Hamiltonian for an electron in this potential
is
p2
H = 2m + V(x) (8.36)
V(x) = V(x + d)
Bloch Wavefunctions
To find the eigenfunctions of this Hamiltonian, we first recall the displacement
operator iJ, introduced in Problem 3.4:

iJ f(x) = f(x + d) (8.37)

The eigenfunctions of this operator are

rp = eikxu(x) (8.38)
u(x) = u( x + d)
with k arbitrary. The eigenvalue of iJ corresponding to rp is exp(ikd) . Although
both factors of tp, namely, exp(ikx) and u(x), are periodic, rp need not be. The
eigenfunction rp(x) is periodic if d, the period of u, is commensurate with 27f/ k,
the period of exp(ikx)-that is, if 2rr/ kd is a rational number.
Since iJ commutes with tt
(8.39)
292 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

these two operators have common eigenfunctions. We conclude that the eigen-
functions of the Hamiltonian (8.36) are of the form (8.38) . These functions are
called Bloch wavefunctions. The related theorem that the eigenstates of a peri-
odic Hamiltonian such as (8.36) are in the product form (8.38) is called Bloch's
theorem.t We have obtained these functions using the displacement operator iJ.
More simply, one may argue that, on the average, the density of an electron beam
propagating through a crystal with a periodic potential should exhibit the same
periodicity as the crystal. That is, one expects that

This equation admits the solutions

qJ(X) = u(x) exp[ia(x)]

where, again, u(x) is periodic with period d and a(x) is any real function indepen-
dent of d . In the limit that the periodic potential becomes constant, V = constant,
d ---+ 00, and the wavefunction qJ(x) becomes the free-particle wavefunction
exp(ikx), with k arbitrary but real. Since a(x) is independent of the period length
(or lattice constant) d, this value of a (i.e., kx) is its value for all d and we again
obtain the Bloch wavefunction

The shape of this wavefunction suggests the manner in which the crystal struc-
ture influences the wavefunctions of particles propagating through the crystal.
This structure is primarily contained in the periodic factor u(x) , which in tum in-
cludes the lattice constant d and which modulates the free-particle form, exp(ikx).
Another way of writing (8.38) is

qJ(x + d) = ei kd qJ(x)
(8.40)
qJ(x) = ei kd qJ(x - d)

If the eigenstate qJ is known over any cell in the periodic lattice (more generally
over any interval of length d), equations (8.40) generate the values of tp in all
other cells.
For any value of k, the corresponding function tp; given by (8.38), is an eigen-
state of iJ. When tp is also an eigenstate of fJ, the values that k may assume
become restricted. For example, the eigenstates of fJ, with V defined over a ring,
have the property

qJ(X) = qJ(x + Nd) (8.41)

Substitution into (8.38) gives

i kN d
e = 1, kNd =2mr (n=0 ,±I,±2, ... ) (8.42)

2F. Bloch, Z. Physik 52 (1928).

8.2 Periodic Lattice. Energy Gaps 293

kd/27T

N=IO N=50 N= 1000

n=0,1,2, . . , , 10 n = 0, 1, 2, .. . , 50 n = 0, 1, 2, .. . , 1000

FIGURE 8.10 Permitted values of k for the periodic ring model depicted in Fig. 8.9. For
N » 1 the spectrum of permitted k values approximates a continuum.

This implies that the allowed values of k form a discrete spectrum [kn =
n(2JT:/ L)] . However, since N is very large (e.g., N :: 108) , the difference between
successive values of k is very small and the spectrum of the permitted values of
k may be taken to comprise a continuum (see Fig. 8.10). With k restricted to the
values given by (8.42), the ratio 2JT: / kd = N / n, a rational number. It follows
that for the closed-ring periodic potential, the eigenfunctions of iJ in the Bloch
waveform (8.38) are periodic .

The Quasi-momentum
The variable hk is called the quasi-momentum of the particle. We list four of the
properties of hk which motivate this name.

1. The eigenstates given in (8.38) resemble the form

i kx
I{Jk = e x constant (8.43)

This is the momentum eigenfunction of a free particle with momentum hk,

The momentum of an electron in a periodic lattice is, of course, not constant
due to the lattice's space-dependent potential field. Nevertheless , there is a
constant value of hk associated with every eigenenergy of the Hamiltonian
(8.36).
2. The group velocity of an electron wave packet in a given band is given by

aE(k)
vg -
-
-
a/ik- (8.44)
294 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

where E(k) represents energy values in the band.' The relation above fol-
lows the classic al recipe for obtaining the velocity of a free particle with
energy E, provided that we associate hk with its momentum.
3. If a particle in a lattice is acted upon by an outside force F, its acceleration
is not F [ m , but F I m" : The "effective mass" m * may be less than m, greater
than m , negative, and even infinite. In one dimension m" is given by

li 2
m * = -;:-----:-~ (8.45)
a2E lak 2

which is suggestive of the classical relation for a free particle E = p212m*,

again with p = hk, (See Problem 8.57.)
4. Eigenenergies E(k) are periodic in k with period 2JTld, so

E(k) = E(k + 2JTn ld) ,

when n is a positive or negative integer. The "central" E (k) curve lies near
the parabola E = li 2 k 2 / 2m , which again suggests a free particle with mo-
mentum lik .

Eigenstates
Next we tum to construction of the eigenstates and eigenenergies of the Kronig-
Penney Hamiltonian. We know that eigenstates are in the Bloch form (8.38). The
continuity conditions that apply to qJ(x) clearly apply also to the periodic compo-
nent u(x) . It follows that u(x) and u'(x) must vary continuously from the right
side of the point x = 0 to the left side of the point x = d, which is one periodic
length displaced from x = 0 (see Fig. 8.8). With u >(O) denoting u(x) evaluated
at x = 0 + E, where E is an infinitesimal, this condition on the periodic continuous
quality of u and u' gives the two equations

u>(O) = u «d) (8.46a)

u > '(O) = u /(d) (8.46b)

Now
u = e-ikxqJ(x)

so that
u' = qJ' e- ikx - iku

and the continuity of u' (8.46b) across a periodic length becomes

tp-; (0) = qJdd)e -ikd (8.47a)
qJ/(O) = qJ /(d)e- ikd (8.47b)
3These properties of the quas i-momentum hk are derived in L. D. Landau and E. M. Lifshitz, Quantum
Mechanics, 2d ed ., Addison-Wesley, Reading , Mass ., 1965.
8.2 Periodic Lattice. Energy Gaps 295

In the well domain of the potential array

qJI(X) = Ae ik1X + Be- ik1X (0 .s x :s a)

lih l 2 (8.48)
--=E
2m
In the barrier domain (with E > V)

qJII(X) = Ce ik2X + De- ik2X (a :s x :s a + b = d)

(8.49)
!i2k 2 2
--=E-V
2m
The periodicity conditions [(8.46), (8.47)] on u(x) then become

A +B = e-ikd(Ceik2d + De- ik2d)

(8.50)
kl(A - B) = k2e-ikd(Ceik2d - De- ik2d)

The remaining two equations for the four coefficients (A, B, C, D) are obtained
from continuity of qJ(x) and tp' (x) across the potential barrier at x = a . This gives

(8.51)
kl (Ae ik1a - Be- ik1a) = k 2(Ce ik2a _ De- ik2a)

The latter four equations may be rewritten in the matrix notation

( kl
1
e ik1a
1
-kl
e- ik1a
_eid(k2- k)

_k2e id(k2-k)
_e ik2a
_'_id(k,+")
k2e-id:k2+k)
_e- lk2a
n
B
C
= 0

kle ik1a _kle- ik1a -k2eik2a k2e- ik2a D

With :D representing the above 4 x 4 coefficient matrix and V the four-column

vector, the preceding equation may be written

This homogeneous equation has nontrivial solutions only if

det:D = 0 (8.52)

This is the desired dispersion relation which is seen to involve the propagation
constant k and the wavenumbers kl and ka . The latter two variables contain the
energy [(8.48), (8.49)], so for a given value of k, the dispersion relation (8.52)
determines the eigenenergy E. As will be shown, this dispersion relation also ex-
hibits the band-gap quality of the energy spectrum attendant to all periodic poten-
tials. The dispersion relation (8.52) is similar to (8.7), which gives the eigenener-
gies for the bound states of the potential well problem. As the domain of existence
296 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

of the wavefunctions of fI is over the finite interval 0 ~ x ~ Nd, these wave-

functions are labeled "extended states." Our main goal is to obtain the energies of
these states.
From (8.52) one obtains the dispersion relation (after a bit of algebra)

E>V
(8.53a)

(8.53b)

The related formula for the case E < V is simply obtained from the latter relation
through the substitution

(8.54)

There results

E<V
k)2 _ K 2
cos k) a cosh K b - k sin k) a sinh K b = cos kd (8.55a)
2 )K

2 2 2mV
kl +K = -2- (8.55b)
!i
Equations (8.53) and (8.55) are implicit equations for the eigenenergies E as a
function of the propagation constant k, valid for all energies. Owing to the tran-
scendental nature of these equations, one turns to a numerical technique for ob-
taining E(k). For example, consider that n = 2, N = 1000. Then the right-hand
side of (8.55a) is cos( 4n / 103 ) ::::= 1. One then plots the left side of the same equa-
tion as a function of the dimensionless energy E / V . Superimposed on this same
curve is the line RHS = 1 (Fig. 8.11). The values where these curves cross give
the eigenenergies E(k) .

Energy Gaps
The fact that values of the right-hand sides of both (8.53a) and (8.55a) lie between
+1 and -1 (lcos kdl ~ 1) implies that the only solutions to these equations are
values of E for which the left-hand sides of these respective equations fall in the
same interval, that is, values of E for which

-1 ~ [left-hand sides of (8.53a) and (8.55a)] ~ +1 (8.56)

Values of E that violate this condition are excluded from the energy spectrum.
The condition (8.56) gives rise to a "band" structure for the spectrum of
eigenenergies. This is again well exhibited with a diagram. In Fig. 8.12, the left-
hand sides of (8.53a) and (8.55a) are plotted versus E / V. On the same graph we
8.2 Periodic Lattice. Energy Gaps 297

RHS of (8.53a)
LHS of (8 .53a) or (8 .55a) for
for E/V> 1 kd= 41T/I03
1.0 t-~----......,~+----.p~I.-.-

t--+-\----I-I---+-\---f--+-f-l--E/V

FIGURE 8.11 Graphical evaluation of eigenenergies of the Kronig-Penney Hamiltonian

corresponding to kd = 4Jr/ 103 . Eigenenergies are given by intersections of the horizontal
line and the oscillating curve.

draw the lines that represent the constant ordinates, + I and -I . The values of
E that qualify as eigenenergies are values for which the oscillating curve falls
between the two horizontal lines, + I and -I .
This construction illustrates the band property of the energy spectrum of a
particle in a periodic potential. This band feature is also illustrated in a plot of
E versus k which may be inferred from the graph of Fig. 8.12. At the left of
Fig. 8.12, values of cos kd are marked off. If a horizontal line is drawn from one
of these values (e.g., cos kd = 1/,J2, kd = ±Jr/ 4), the intersections of this line
with the oscillating curve give all the energies that correspond to the propagation-
constant values, k = ±rr/ 4d . There are infinitely many of them. Continuing this
process for all values of kd gives the curve of E versus kd sketched in Fig. 8.13a.

LHS of (8 .53) and (8 .55)

Forbidden band s
cos kd

E/V

T
l Energy bands

FIGURE 8.12 Band structure of the energy spectrum of the Kronig-Penney Hamilto-
nian. The only eigenenergies are those values for which the left-hand side of (8.53a) or
(8.55a) falls between ±1.
298 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

o 1r 31T 41r kd

(a)

} 1l£3

}1l£2

}OE]
kd
-1r 1T

(b)

FIGURE 8.13 (a) Typical E versus k curves for the Kronig-Penney potential. The grain-
iness of the curves stems from the fact that the k values in each band are discrete [i.e.,
kd = tn] N)2n, N » 1 or, equivalently, (~k)min = n / L]. (b) The first four bands in
the reduced-zone scheme. Also shown are the first three energy gaps, 8E1' 8E2, and 8E3.
Consider that the second and third bands shown in (b) are identified as the valence and
conduction bands, respectively. At k = 0, one finds that me * > °
and mt; * < O. This
effective-mass property is typical in semiconductors. One notes further that effective mass
is consistently defined in the neighborhood of extrema points in E versus k curves (i.e.,
points where vg = dE [dk = 0). For further discussion see K. Hess, Advanced Theory of
Semiconductor Devices, Prentice-Hall, Englewood Cliffs, N.J., 1988.
8.2 Periodic Lattice. Energy Gaps 299

If we look at any single band, the curve E versus k is periodic in k. This

results from the fact that the right-hand sides of (8.53a) and (8.55a) maintain the
same values if kd is replaced by (kd + n2Jr), where n is an integer. The value
of E that satisfies this equation [i.e., either (8.53) or (8.55)] for a given value
of kd satisfies it for (kd + n2Jr) . It suffices then to draw all bands in the single
interval -Jr ~ kd ~ Jr. This gives the reduced-zone description (Fig. 8.13b) of
eigenstates. These bands consist of very closely packed discrete energies (recall
Fig. 8.10) and constitute all the eigenenergies of the Hamiltonian (8.36). This
discrete nature of the energy spectrum is a consequence of the boundedness of
the system. The quasi-continuous quality (bands of closely packed levels) of the
spectrum reflects the propagating nature of the eigenstates.
Superimposed on the E versus k curves in Fig. 8.13a are the free-particle en-
ergy curves

(n = ±O, 1, 2, . . . ) (8.57)

This corresponds to a free-particle momentum

(8.58)

From Fig. 8.13 we see that (1) much of the locus of the E versus k curves falls
near the free-particle energy curves, and (2) energy gaps occur at the values

kd =qJr (8.59)

where q is a positive or negative integer. At these values of k an integral number

of half-wavelengths span the distance d between ions.

Bragg Reflection
To understand the physical origin of energy gaps at these values of k, it is best to
recall that the one-dimensional solution we have found is appropriate to propaga-
tion of plane waves through slabs of constant potential V, thickness b, and spaced
a normal distance d from one another. This situation is depicted in Fig. 8.14, in
which two typical wavefronts are also sketched. Suppose that these plane waves
are incident at the angle e, such as is drawn in Fig. 8.15. In the Bragg model of
reflection, wavefronts scatter from ion-site lattice planes. We recall that the con-
dition that reflected waves from adjacent planes add constructively is given by
Bragg's formula,

kd case = qn (8.60)

where e is the angle that the incident k vector makes with the normal to the lattice
e
plane. In the limit of normal incidence, -+ 0, and the one-dimensional model
of the analysis above becomes relevant. Equation (8.60) reduces to the condi-
tion (8.59), which k satisfies at an energy gap. At these values of k, an integral
300 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Surface of constant
phase for "incident" Reflected wave:
wave: A exp [i(kx - wt)] B exp [i(kx + wt)]

t--....-_....._ - _......- ......._ - _....._-........... x

FIGURE 8.14 Shaded regions depict domains of constant potential. They are slabs that
extend out of the paper. Surfaces of constant phase are also normal to the plane of the
paper.

number of wavelengths fit the distance 2d and the reflected waves constructively
superpose. They are Bragg-reflected. Consider that a wave is moving in one di-
rection with a critical k value (8.59). It is soon Bragg-reflected and propagates in
the opposite direction. There is a similar reversal of direction of propagation on
each reflection until finally the only steady-state solution is that which contains an
equal number of waves traveling in either direction. As will be shown in the next
section, at these critical values of k, the eigenstates of fI are composed equally
of waves moving to the right and left, so that, for example, in the well domain
of the periodic potential, cp r-v exp(iklX) + exp( -iklX) r-v COS(klX), which is
the spatial component of a standing wave. A similar standing-wave structure pre-
vails across the barrier domain. When these solutions are matched at the potential
steps, a standing wave ensues over the whole periodic potential. In such states
(p) == O. Electrons are trapped and lose their free-particle quality. Energy curves
appropriate to a one-dimensional periodic potential are shown in Fig. 8.13b.

Spreading of the Bound States

Let us now demonstrate that the band-energy spectrum relevant to an electron in
a periodic potential collapses to the discrete bound-state spectrum appropriate to
a particle in a single finite potential well in the limit that the wells of the periodic
potential grow far apart (b ---+ (0). Toward these ends it suffices to demonstrate
that the dispersion relation (8.55), for the states E < V for a periodic potential,
in the limit of b ---+ 00, gives the relations (8.27) and (8.28):
n -;
tan s = ~' tan; == -
n
(8.61)
8.2 Periodic Lattice. Energy Gaps 301

Reflected
plane waves

-+--+
t
d Planes of
t . lag = 2A positive ion sites
-+--+--+---Cf--I +--+--+-

.+- - +- - +- - - + +---+---+--+-

A=dcos8

FIGURE 8.15 Constructive interference between reflected waves from different planes
occurs when
2nl
2dcos(} = LA = k
which gives

kd cos () = In

For normal incidence, () = 0, and this condition becomes

kd = In

These are the relations for the even and odd states, respectively, of a particle in a
finite potential well.
Let us recall that the nondimensional parameters ~, 1], and p introduced in
(8.26) and (8.27) contain half the well width, which for the Kronig-Penney po-
302 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

tential is a /2, so for the present case

kia «a
~= 2' 11 = 2"
(8.62)
2 22m(a/2)2y2
~ + 11 = P = li2

In terms of these variables, (8.55a) becomes

2bl1 )
cos 2~ cosh ( - + 112 - ~2 sin
. 2~ sinh
. (2b l1)
- = cos kd (8.63)
a 211~ a

This equation must reduce to the two equations (8.61) in the limit b -+ 00, a =
constant, Y = constant. First we note that in this limit

2b l1 ) ~ sinh
. (2b l1) I exp (2b l1)
cosh ( ---;- ---;- ~ 2 ---;-

Dividing through by this exponential factor and allowing b to grow infinitely large
reduces (8.63) to the form

2~11
tan2~= 2 2 (8.64)
~ -11

The double-angle formula for tangents permits this equation to be rewritten as

which in turn may be rewritten as

This is a quadratic equation for tan ~ ' Solving for the two roots gives

These are the two relations (8.61) that give the discrete bound states of a single
isolated finite potential well.
Thus we find that the band structure of the energy spectrum of a particle in a
periodic potential collapses to the discrete energy spectrum of a particle in a finite
potential well in the limit that the wells of the periodic array became far removed
from one another. Consider, for instance, a finite potential well that has two bound
states. Such, for example, in the case if p = Sr:/4 (see Fig. 8.2). For a periodic
array of such potentials, the relation (8.55) applies in the domain E < Y. If the
left-hand side of this equation is plotted versus E such as in Fig. 8.12, two bands
Problems 303

(a) (b)

FIGURE 8.16 (a) Single isolated finite potential well with two bound states. (b) Cor-
responding periodic potential with two energy bands. For N wells each band contains N
states.

will be found to fall in the domain E < V. This transition" from the discrete states
of an isolated well to the band structure of a lattice is illustrated in Fig. 8.16.

PROBLEMS

8.10 (a) What is the expectation of momentum for an electron propagating in a Bloch
wavefunction with spatial component

(b) Show that if the periodic function u (x) is real, (p) = hk.

Answer (partial)
(a) (p) = hk. + (ulfilu)

8.11 What is the period of the Bloch wavefunction under the following conditions?
(a) kd = 2In
(b) kd = (2I + l)n
(c) kd = nst jq

Here I, n, and q are integers.

8.12 (a) Use the dispersion relation (8.55) to obtain the dispersion relation for the prop-
agation of electrons through an infinite array of equally spaced delta-function
potentials' separated by d em (see Fig. 8.17). Note that the delta-function poten-
tial may be affected by constructing a potential barrier whose height is infinite
and whose width is infinitesimal such that the area under the potential curve is
fixed. This limit is easily constructed with the model at hand by setting
2
lim
K-+OQ
K bd = 2F = constant
b-+O

4Details of a numerical analysis for this transition for the case of a well with four bound states may be
found in V. Rojansky, Introductory Quantum Mechanics, Prentice-Hall, Englewood Cliffs, N.J., 1938.
5This limiting case was, in fact, the one treated by R. de L. Kronig and W. G. Penney in their original
paper, Proc. Roy. Soc. A130, 499 (1931).
304 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

v
-f-oo

-Sd/2 -3d/2 -d/2 an 3d /2 Sd/2 x

FIGURE 8.17 Periodic delta-function potential. The explicit fonn of the symmetric pe-
riodic delta-function potential is given by (see Problem 3.6e)

V(x) = Vod {~ 8 [x - (2n + 1) (~)] + ~ 8 [x + (2n + 1) (~) J}

= Vod
2
~(2n + 1)8 [ x 2 - (2n + 1)2 (~r]

(See Problem 8.12.)

(b) Make a plot of your dispersion function for the value F = 3n12 and thereby
illustrate the persistence of the band structure of the energy spectrum in this
delta-function limit.
(c) How is it that electrons are able to propagate through the infinitely high potentials
presented by the delta functions?
(d) Write down a formal expression for the potential you have considered. '
(e) What are the eigenstates at the band edges kd = ntt'l Show that one of the
energies at a band edge is the free-particle energy fi2 k 2 / 2m , while the other
energy is larger. In this manner obtain an expression for the width of the energy
gap at kd = nit .

Answer (partial)
(a) In a limit given above it follows that

2F
«b -+ - -+ 0
Kd
Hence sinh ee -+ «b , coshxb -+ I, and d -+ a. The resulting dispersion
relation appears as

Fsink,d
---=-- + cos k, d = cos kd
kid

8.13 Show that the E(k) spectrum for the arbitrary finite Kronig-Penney array draws close
to the free-particle parabola E = fi2k 2 12m in the limit E V.»
Problems 305

Answer
With V I E = 13 « 1, one obtains

Substitution into the dispersion relation (8.53) gives

Neglecting terms of 0(13) gives the spectrum k, = k or, equivalently, E = 1i2k2/2m .

Recalling further that k is discrete (8.42), one obtains En = n 2(h 212m L 2) .

8.14 The E (k) spectrum for an electron in a periodic lattice, such as illustrated in Fig. 8.13,
does not fall to zero at k = O. Estimate this zero-point energy using results appropri-
ate to a particle confined to a one-dimensional domain of length L. What value of k
is implied by your answer? How does this value compare to the minimum value of k
for a crystal of length L ?

8.15 (a) Show that the eigenenergies of the one-dimensional box of width a (k, a = mf)
lie in the energy gaps of the Kronig-Penney potential of well width a (in the
domain E < V) . [Hint: Use (8.55).]
(b) Show that these box energies become the lower energies of the band gaps for the
periodic delta-function potential described in Problem 8.12.
8.16 (a) Show that in the limit that the atomic sites of the Kronig-Penney model become
far removed from each other (b -+ 00) , energies of the more strongly bound
electrons (E « V) become the eigenenergies k, a = nit of a one-dimensional
box of width a.
(b) In this limit what do the lower-band E(k) curves-shown in Fig. 8.13 become ?
What is the functional form of E(k) for these bands? (Note: The approximation
in which one begins with electronic states of isolated atoms is called the tight-
binding approximation.)

8.17 (a) Construct an equation for the periodic component u(k) of the Bloch wavefunc-
tion <p = U exp(ikx), from the Schrodinger equation with a periodic potential
V(x) .
(b) The periodic potential V(x) may be expanded in a Fourier series as follows :

Expand the periodic component u(x) in a similar series, substitute in the equation
obtained in part (a), and derive coupled equations for the coefficients of expansion Un .

Answer
1i2
(a) -(u" + 2iku' - k 2u) + [E - V(x)]u =0
2m
(b) [E - Bq(k)]uq = L~-oo Vq_IU/ , 2mBq(k) /1i2 == (2rrq ld)2+
2k(2rr q I d) + k 2
306 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

8.18 Show that in the limit V --+ 0, the equation for the periodic component u obtained in
Problem 8.17 gives the free-particle eigenenergy

with u = constant.
8.19 What is the number of J{E of discrete energies in any of the energy bands depicted
in Fig. 8.13?

Answer
The k values that enter the lowest energy band are given by the sequence

2n 2n N 2n
k=O ,±1 x L ,±2 r r: '±2" L
L=Nd

This series is cut off at Ikdl = n inasmuch as energy values begin to repeat beyond
this value. Furthermore, as these k-values are at the zone boundary (see Fig. 8.13)
they are counted only once. There is a distinct energy corresponding to each value of
Ikl in the sequence above. This gives

8.20 What is the number J{k of independent eigenstates in a band for a one-dimensional
crystal comprised of N uniformly spaced ions?

Answer
There is a distinct eigenstate (8.46 et seq.) corresponding to each value of k in the
series in the example above. It follows that J{k = 2J{E = N . There are as many
eigenstates in a sample as there are ion sites. There are approximately half as many
eigenenergies.
This result may also be obtained geometrically. Referring to the reduced-zone en-
ergy diagram (Fig. 8.13b), each energy band has width (/).kd)b = 2n . The minimum
interval in each band is (/)'kd)min = 2n /N or, equivalently, (/).k)min = 2n/Nd =
2n/ L. Thus the number of points (states) in each band is

(Note: With the two spin orientations taken into account, one obtains 2N independent
states in each band .6 The concept of spin is described in Chapter II.)

6This result maintains in three dimensions, where N represents the number of primitive cells in the
crystal. For further discussion, see C. Kittel, Introduction to Solid State Physics, 5th ed., Wiley, New
York, 1976.
 8.3 Standing Waves at the Band Edges 307

8.3 • STANDING WAVES AT THE BAND EDGES

Let us return to the nature of the eigenstates of if at the band edges, that is, at
kd = nn . We will demonstrate that these eigenstates are standing waves and
illustrate the relation between the eigenenergies of these states and the energy
gaps at the band edges.
The eigenstates of the Kronig-Penney Hamiltonian established above have
components [see (8.48) and (8.49)]

qJI = Aeik,x + Be-ik,x (8.65)

qJn = Ce ik2X + De- ik2X

In order for these to be components of a standing wave, the magnitude of the

amplitudes of the waves moving to right and left must be equal. That is, at the
critical values kd = nit , one must have

IAI
jBj = 1,

We will establish the first equality and leave the second as a problem. At the values
kd = nit, exp(ikd) = (_l)n . Consider that n is even so that exp(ikd) = +1.
With this value substituted into the equations of continuity, (8.50) and (8.51) , one
quickly obtains the following two equations for the expression 2Ck2 exp(ik2a):

2Ck2eik2a = e- ik2b[A(kl + k2) + B(k2 - kd]

2Ck2eik2a = eik,aA(kl + k2) + e-ik,aB(k2 - kl)

Setting these two expressions equal to each other and solving for AI B gives

(8.66)

Forming the square of the modulus IAI B 12 = (AI B)(AI B)* gives

1
~ 12 = Ik 2 - kl12 2 - 2cos(kla - k2b) (8.67)
B k2 + kl 2 - 2cos(kla + k2b)
2
= I k2 - kl 1 1 - cos kla COSk2 b - sin kla sink2b
k2+kl l-cosklacosk2b+sinklasink2b

We must show that this expression is unity for the allowable values of kl and k2,
that is, those values which are obtained from the dispersion relation (8.53). Again
with exp(ikd) = +1, this relation reads
308 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

When this formula for the cos product is substituted into (8.67), the desired result,
IAIBI = 1, follows.
Let us proceed to construct such a standing-wave state. If the potential is placed
in a symmetric position about the origin, such as in Fig. 8.18, then the Hamilto-
nian commutes with the parity operator JP> and these two operators share a set of
common eigenfunctions; that is, iJ has even and odd eigenstates. It will be shown
that eigenstates at the band edges exist in pairs, with each pair containing an even
and an odd eigenstate . Very simply, one expects that this is the case since, in the
steady-state situation, electron density 1<p12 should enjoy the same symmetry as
the periodic potential Vex) ; that is, 1<p12 is even, so <p is either even or odd. We
will find that if kd I T[ is an even integer, eigenstates have period d . If kd I T[ is an
odd integer, eigenstates have period 2d. Let us consider the construction of the
pair of eigenstates at a band edge corresponding to kdf n an even integer. Con-
sider first the symmetric eigenstate. In the well domain about the origin

(-aI2 :s x :s a12) (8.68)

To obtain ({JIll in the second well domain, one uses Bloch's theorem (8.40) together
with the value exp(ikd) = +1.

(a12 + b :s x :s al2 + d) (8.69)

The standing wave in the barrier region II which joins these waves is symmetric
about the midpoint dl2 (see Fig. 8.18).

<PII(X) = DcOSk2 (
X -d)
-2- (a12 :s x :s al2 + b) (8.70)

The coefficient D and energy !i2k , 212m are obtained by matching the components
of <P and <p' at the potential interface at x = a12 . There results for the even

x
-d -a/2 a/'2 d
~
'PI 'PII 'PIlI

FIGURE 8.18 Standing-wave eigenfunction referred to in the construction <p(x) as given

by (8.68) to (8.70). Owing to the symmetric form of <p, it suffices to match <PI to <PII at
x = a/2 .
8.3 Standing Waves at the Band Edges 309

eigenstates,

2b)
cos ( 2ak' ) =Dcos (k2
(8.71)
. (ak')
k, sm 2 = - Dk2 sm . (k22b)
Identical equations are obtained by matching <pn to <pm at a 12 + b (see Fig. 8.18).
Equations (8.71), together with the energy statement
2 2 2mV
k, r ka =7
are three equations for the three unknowns k" kz, and D. They are reminiscent
of (8.27) and (8.28) and appropriate to the bound states of a particle in a finite
potential well. Here, as there, solution may be effected through a numerical pro-
cedure (see Problem 8.24). The companion odd eigenstate ip may similarly be
constructed with the modification that the standing wave in the barrier region II is
odd about the midpoint d 12. One obtains

ip,(x) = sink, x (-aI2 ::; x ::; a12)

- - (d2 - x )
ipn(x) = D sink2 (a12 ::; x ::; al2 + b) (8.72)

ipm(x) = sin t. (x - d) (aI2+b ::;x ::;aI2+d)

Matching conditions at the interface position al2 gives the following relations for
the odd eigenstates:

2b)
k,- cos (k'a)
2 = -k2Dcos
- - (k2 (8.73)

-2 -2 2mV
k, -k2 =7
Again , num~rical procedure yields values for the energy li 2k,2/2m and eigenstate
parameter D.
The most significant result of such calculation is the width of the energy gap
8En at the band edge kd = ntt . This is the difference in energy between the even
and odd standing-wave eigenstates:
2
li 2 - 2
(8E)n = -(k, - k, )
2m
An analytic evaluation of this energy jump may be obtained in the "nearly free
electron" model. This model is described in Section 13.4.
310 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

Parity Properties
Next we tum to a discussion of the parity properties of these standing-wave eigen-
states at the band edges. These states are either even or odd in x . Again consider
the case that kdf tt is an even integer. Then the relation

~(x + d) = eikd~(x) (8.74)

which is true for any eigenstate of the Kronig-Penney Hamiltonian , gives

~(x + d) = ~(x) (8.75)

It follows that for kd an even multiple of n, the period of ~ is d. Setting x = -d/2

in (8.75) gives

(8.76)

From this equation one concludes that ~ can be an odd eigenfunction provided
that

(8.77)

This property, taken together with the fact that ~ has period d, gives

(~is odd, kd = 2qrr) (8.78)

with n an integer. The only stipulation on the even eigenfunctions is that they are
of period d .
In this manner we find that the eigenfunctions of the Kronig-Penney Hamilto-
nian at the band edges kd = Zqtt exist in pairs. Each pair contains an even eigen-
function and an odd eigenfunction. A typical pair of these functions is sketched
in Fig. 8.19. The eigenenerg ies that accompany these eigenstates are the close-
spaced pairs of values depicted in Fig. 8.13, where the vertical lines kd = Zqr;
intersect the oscillating curves.
Having treated the case where kd is an even multiple of n , we next consider
the case kd = (2q + l)rr, again with q an integer. From (8.74) one obtains

~(x + d) = -~(x)
(8.79)
~(x + 2d) = -~(x + d) = +~(x)

It follows that for kd an odd multiple of n , the period of ~ is 2d . Setting x = -d/2

in the equation above gives

(8.80)
8.3 Standing Waves at the Band Edges 311

Only odd eigenfunctions are

such that 'P(±nd/2) = 0, n = 0, 1.2, . . .

<P odd

- - 'P.v• n

I 'Podd 12 is maximum at barrier faces

2 is maximum at well and
I <Peven 1
barrier centers

FIGURE 8.19 Typical pair of eigenfunctions for Kronig-Penney Hamiltonian at the

band edges: kd = Zqn . Periodicity of rp is d .

while x = -d gives

q;(d) = q;(-d) (8.81)

Equation (8.80) indicates that q; can be an even eigenfunction provided that

q; [±(l + 2n)~] = 0 [kd = (2q + I)Tl] (8.82)

with n an integer.
Equation (8.81) indicates that q; can be an odd eigenstate provided that

q;(±nd) = ° [kd = (2q + 1)7t] (8.83)

again with n an integer.

A typical pair of eigenfunctions is sketched in Fig. 8.20, together with accom-
panying plots of electron density 1q;1 2 . From this sketch one notes that each pair
of eigenstates, corresponding to kd an odd multiple of x , contains one eigenstate
with density 1q;1 2 , maximum at the ion sites and minimum at the barrier centers,
while the other eigenstate has its extremum values of 1q;1 2 reversed.
312 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

V(x)

Only odd eigenfunctions are such that

~'1W:';~ 'Podd ",(±nd) = 0,11 = 0, 1,2, . ..
Only even eigenfunctions are such that
- ""even
",[±(1+ 2n)d/2) = 0

I "'odd 12 is maximum at barrier centers

I"'<v<n 12 is maximum at well centers
(ion sites)

FIGURE 8.20 Typical pair of eigenfunctions for the Kronig-Penney Hamiltonian at the
band edges : kd = (2q + l)JT. Periodicity of rp is 2d .

Thus we conclude that at the band edges [kd = Zqtt or kd = (2q + l)rr]
eigenfunctions appropriate to the Kronig-Penney Hamiltonian are standing waves
and that there are two such functions with opposite parity at each edge.

PROBLEMS

8.21 The standing-wave quality of the eigenstate rp(x) at the band edges was demonstrated
for the component of rp in the valley regions of the potential (IA I BI = 1) for the case
exp(ikd) = + 1. Following this analysis, demonstrate that the component of rp in
the barrier domain is also a standing wave (i.e., show that lei D I = 1) for the case
exp(ikd) = -1. [See (8.65) et seq.]

8.22 Show that the expectation of momentum (p ) vanishes for a particle in a standing-
wave eigenstate.

8.23 Consider a typical pair of eigenstates appropriate to kd = 2qJT and the adjacent
pair appropriate to kd = (2q + l)JT. By inspection only, conclude which pair of
corresponding eigenenergies is of higher value .
 8.4 Brief Qualitative Description of the Theory of Conduction in Solids 313

8.24 (a) Introducing nondimensional variables

k2 b 2V
; == k)a, 2 _ ma
2 1]=2' P = 21i2

show that (8.71) and (8.73) relevant to even and odd standing-wave band-edge
solutions, respectively, give the dispersion relations

; tan; = _};2 - p2 tan (~};2 _ p2) (even)

~ cot~ = -}p - p2 cot (~}~2 _p2) (odd)

(b) Numerical solution of either equation may be effected by plotting the right-hand
side and the left-hand side of the equation as functions of; (or #)
on the same
graph . Intersections then give the eigenenergies E = (21i2Ima 2);2 . Use this
procedure to estimate the lowest even and odd eigenstate energy corresponding
to the barrier parameters, p = n 12, (a lb)2 = 15.
(c) Use your answer to part (b) to obtain the width of the energy gap SE at this band
edge.
8.25 It was shown in Problem 8.13 that in the high-energy domain E » V , the E(k)
spectrum approaches the free-particle curve E = 1i2 k 2 /2m . Show that the disper-
sion relation appropriate to the band edge kd = 'lntt , (8.73), yields a free-particle
standing wave in this limit.

Answer
With k) ::::: k2, (8.73) gives (dropping the tilde notation)

sin(k)a I2) cos(k)a I2)

D= =
sin(k)bI2) cos(k)bI2)

The second equality gives

sm
. [k) (a + b)] = 0
2
so

kId = 2nJr = kd

When substituted back into the expres sion above, one obtains D = I, which is nec-
essary in order that the eigenstate (8.72) with k ) = k2 be a free-particle standing
wave.

8.4 • BRIEF QUALITATIVE DESCRIPTION OF THE THEORY

OF CONDUCTION IN SOLIDS

The spectrum of eigenenergies of electrons in an actual three-dimensional crys-

talline solid closely parallels that of the Kronig-Penney model described pre-
viously. In the three-dimensional case one also obtains a band structure for the
314 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

Conduction
band
DJ ~t
6 eV

valenc e / -
band

FIGURE 8.21 Energy bands of diamond, a good insulator.

allowed eigenenergies. The electrons in a solid occupy these bands. The proper-
ties of the two bands of highest energy for most practical cases determine whether
the solid is an insulator, a conductor, or a semiconductor.
Suppose that the band structure of a solid is such that the band of highest
energy is full (see Fig. 8.21). Furthermore, the gap between this filled band and the
next completely unoccupied band is reasonably large. For example, for diamond
this gap has a width of 6 e V. When an electric field is applied, electrons in the filled
bands have no nearby unoccupied states to accelerate to. The sample remains
nonconductive. It is an insulator. Furthermore, photons that comprise the visible
spectrum do not have sufficient energy (hv) to raise electrons from the valence
band (last filled band) to the conduction band , so that diamond is transparent to
light.
These statements are precisely true at absolute zero (-273°C) . The student will
recall that at absolute zero a system of particles falls to its lowest energy state,
called the ground state of the system. When the temperature is raised, thermal
agitation excites electrons to states of higher energy. For instance, for diamond
at room temperature the characteristic energy of thermal agitation is :::::: 0.03 eV.
The concentration of electrons which are raised to the conduction band is c; 1.1 x
10-34 electrons/cm''. This gives rise to a conductivity which is lower than can be
measured with present-day equipment, and diamond remains an insulator at room
temperature.
In some crystalline solids, the conduction band is empty and the energy gap
to the valence band is not prohibitively large. For instance , in silicon, this gap is
1.11 eV wide. In germanium it is 0.72 eV wide. At room temperature the concen-
tration of electrons in the conduction band in silicon is 7 x 1010 electrons/em" , In
germanium it is 2.5 x 1013 electrons/cm' (see Fig. 8.22). These densities give
measurable conductivities. Such materials are called intrinsic semiconductors.
The conduction of an extrinsic semiconductor is due to the presence of impurities
in the sample .
A semiconductor acts as an insulator at sufficiently low temperatures. It begins
to conduct at higher temperatures. In a semiconductor, charge transfer in the va-
lence band may also contribute to conduction. In this case it is simpler for, say,
8.4 Brief Qualitative Description of the Theory of Conduction in Solids 315

T=OK Room temperature

FIGURE 8.22 Valence and conduction bands for germanium, a typical semiconductor,
at absolute zero and room temperature .

calculation of conductivity, to speak of hole conduction. A hole is an unfilled

state, found usually in the valence band.
In a metal the band of highest energy is only partially filled and electrons are
readily accelerated by an electric field, to states of higher energy (see Fig. 8.23).
Photons also fall prey to these electrons, which explains the opacity of metals to
light. Note that the Fermi energy, described in Chapter 2, appears in the conduc-
tion band.
The description we have presented for the band structure of energy levels in
periodic structures is a vast simplification of that which occurs in actual solids.
A more accurate description of the formation of such bands with shrinking in-
teratomic distance is shown in Fig. 8.24 for the metal sodium. Note in particular
the strong band overlap at very small interatomic distance . This property suggests

f
Conduction
band
---------
---------

kd
FIGURE 8.23 In a conductor the states in the conduction band are partially filled. The
diagram to the right indicates the manner in which electrons fill the corresponding bands in
the reduced-zone description for the idealized one-dimensional model. The Fermi energy
E F is also shown.
316 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

o
]4J3d
~=.zZZZlll---l3p

~=-_-----l3s

-10

-20

~----------~2p

3.67 5 10 15
Internuclear distance (A)

FIGURE 8.24 Energy-level diagram for sodium showing the development of a band
structure as internucleon distance decreases. Labeling of states, 2p, 3s, etc., is fully de-
scribed in Section 12.4. In the ground state of sodium, electrons fill up to the 3s level,
which is half-empty. This property accounts for the conductivity of the metal.

why materials become electrically conductive under extreme compression. The

theory of semiconductors is returned to in Section 12.9.
The problem of a particle propagating in a periodic potential is revised in Sec-
tion 11.14 where it is reformulated in terms of symmetry properties of the poten-
tial and the "transfer matrix" method.
 8.5 Two Beads on a Wire and a Particle in a Two-Dimensional Box 317

PROBLEMS

8.26 What is the minimum frequency of radiation to which diamond is opaque? What kind
of radiation is this (e.g., x rays, etc.)?

8.27 The mobility /-L of an electron in an electric field E is defined by

v=/-LE

where v is the drift velocity of the electron. In a given semiconductor the mobil-
ity of electrons is /-Ln. while the mobility of holes is /-Lp . If at a given temperature,
the density of conduction electrons is n electrons /crrr' and the density of holes is
p holes/cm-', obtain an expression for the current flow in the semiconductor if an
electric field E is applied across it.

8.5 • TWO BEADS ON A WIRE AND A PARTICLE

IN A TWO-DIMENSIONAL BOX

Exchange Degeneracy
In this and the following section we discuss some simple examples of quantum
mechanical systems with two degrees of freedom (see Section 1.2). The first such
example is that of two beads constrained to move on a straight frictionless wire
that is tightly stretched between two perfectly reflecting, rigid walls. The space
between walls is a (see Fig. 8.25). We still assume that the particles do not interact
with each other (they are "invisible" to each other). The Hamiltonian for this
system is

(8.84)

The two particles have the same mass, m . The potential functions VeX)) and
V (X2) are relevant to a one-dimensional box. Their properties are given in Sec-
tion 4.1.

o a x

FIGURE 8.25 Coordinates of two beads on a wire stretched between two perfectly re-
flecting walls separated by the distance a.
318 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

This Hamiltonian may be partitioned into two independent terms,

H(xI ,xz) = HI(xI) + Hz(xz)

A PI Z (8.85)
HI(xI) = 2m +V(xd

A jJzZ
Hz(xz) = 2m + V (xz)
Under such circumstance, solution of the Schrodinger equation

(8.86)

is greatly simplified . It is given by the product

(8.87)

where

Hlrpnl(xd = Enlrpnl(xd
(8.88)
Hzrpn2 (xz) = En2rpn2 (xz)

The function rpnl is the eigenfunction of HI corresponding to the energy Enl,

while rpn2 is the eigenfunction of Hz corresponding to the eigenenergy E n2.

rpn = yf2
~
.
Sill
(nlrx)
--;;- , (8.89)

where n denotes either n I or nz .

Let us test to see if rpnln2' as given by (8.87), is an eigenstate of H (XI, xz).

Hrpnln2(XI, xz) = (HI + HZ)rpnl (Xdrp n2(XZ)

= rpn2Hlrpnl + rpnlHZrpn2
(8.90)
= E nl rpnl rpn2 + En2rpnl rpn2
Hrpnln2 = (E nl + E n2)rpnlrpn2

Thus we find that rpnl rpn2 is an eigenstate of H (XI , xz), and furthermore that the
eigenenergy corresponding to this state is

(8.91)

For example, the eigenstate

2 Sill
rpZ,3 = ~ . (2lrXI) . (3lrXz)
-a- Sill -a- (8.92)
8.5 Two Beads on a Wire and a Particle in a Two-Dimensional Box 319

has corresponding eigenenergy

(8.93)

This energy is doubly degenerate since the eigenstate

2 Sin
CP3 ,Z = ~ . (3nXI) . (2nXZ)
-a- Sin -a- (8.94)

also corresponds to the eigenenergy £Z,3, One may look upon the difference be-
tween CPZ,3 and CP3,Z as being due to the exchange in the positions of particle 1 and
particle 2. Such degeneracy is called exchange degeneracy.

Symmetric and Antisymmetric States

If two eigenstates correspond to the same eigenenergy, any linear combination
of these eigenstates also corresponds to the same eigenenergy. Of all such linear
combinations, two are of particular physical significance. These are of the form

1
CPs = J2[CPn 1(XdCPn2(Xz) +CPnl(xZ)CPn2(XI)]
(8.95)
1
CPA = J2 [CPn I (XdCPn2 (Xz) - CPn I (XZ)CPn2 (XI)]

The symmetric state CPs has the property that

(8,96)

It is symmetric under the exchange of the particles. The antisymmetric state CPA
has the property that

(8.97)

It is antisymmetric under the exchange of particles.

When referred to systems with two degrees of freedom, such as that of two
particles in a one-dimensional box, the probability amplitude related to the system
is given by (see Problem 3.20)

(8,98)

P12 dXI dxz is the probability of finding particle 1 in the interval dXI about the
point XI and particle 2 in the interval dxz about the point Xz, in any given mea-
surement.
When the two particles in the one-dimensional box are identical (m I = mz),
such as in the case considered, we note that for both classes of wavefunctions
(symmetric and antisymmetric)

(8.99)
320 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Physical properties of a system are not affected by an exchange of the position

of the two particles. This is a manifestation of a quantum mechanical property at-
tached to identical particles; that is, in quantum mechanics identical particles are
also indistinguishable (they cannot be labeled). In the scattering of electrons off
electrons, for example, the scattered beam contains both incident and target elec-
trons. The indistinguishability of these particles must be taken into account in any
consistent formulation of the theory of such scattering. It is the indistinguishabil-
ity of identical particles which selects CPA or CPs (8.95) to be the physically relevant
linear combination of eigenstates for the two-particle problem.
If the masses of the two particles in our one-dimensional box are different (m I
and m2) , the Hamiltonian (8.84) becomes

(8.100)

The particles are now distinguishable and the states of the system do not suffer
exchange degeneracy. The eigenstate

(8.101)

corresponds to the eigenenergy

(8.102)

The exchange state CPn 2n I corresponds to the eigenenergy

(8.103)

Thus the exchange degeneracy associated with systems containing identical par-
ticles is removed.
We now turn to the time-dependent Schri:idinger equation for systems with two
degrees of freedom

a,l, A

iii-'/' = H1{t (8.104)

at
The solution of this equation is (see Section 3.5)

(8.105)

Given the arbitrary initial state 1{t(XI , X2. 0), the state at time t > 0 is
8.5 Two Beads on a Wire and a Particle in a Two-Dimensional Box 321

Examples of the use of this equation are given in the problems that follow the next
subsection.

Symmetry and Accidental Degeneracy

Much of the preceding analysis may be carried over to the problem of a single
particle moving in a two-dimensional box (see Fig. 8.26). This is another case of
a system with two degrees of freedom. In the example of two beads on a wire,
cited above, the good coordinates are (Xl, X2). For the single particle in a two-
dimensional box, good coordinates are (x, y). The Hamiltonian for this system
appears as

A 2 A 2
Px
A

H(x, y) == -
2m
+ -Py
2m
+ Vex) + V(y) (8.106)

The potential V (x) is the same as that of a one-dimensional box which lies be-
tween x x a x
== 0 and == on the axis, whereas V (y) is the same as that of a
one-dimensional box that lies between y == 0 and y == a on the y axis. Eigenfunc-
tions and eigenenergies are

. (nIJrx) . (n 2JrY)
CPn l n2 ( x , y) == -;;2 SIn -a- sm -a- (8.107)

2
E n 1n2 == EI (n1 + n2 2)

FIGURE 8.26 Particle in a two-dimensional box.

322 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

This eigenenergy also corresponds to the eigenstate

({Jnznl (x, y) = ~ sin C2: X

) sin Cj:y)
The probability density l({Je may be plotted as a height above the xy plane. The
distinction between l({Jnln zl 2 and l({Jnzn l 12 is then as follows . The surface l({Jnln z l 2
is obtained from the surface I({Jnzn 1 12 by reflecting this surface through the plane
x - y = O. The energy corresponding to both these distributions is the same. The
degeneracy of these states is sometimes called symmetry degeneracy (as opposed
to exchange degeneracy). Degeneracy that is neither symmetric nor exchange is
often referred to as accidental degeneracy (see Problem 8.34).

PROBLEMS

8.28 At time t = 0, two particles of mass m j and m2 , respectively, in a one-dimensional

box of width a are known to be in the state
,1,( 0) 3ip5(Xj)IP4(X2) + 7fP9(Xj)ipS(X2)
v Xj ,X2 , = J58
(a) If the energy of the system is measured, what values will be found and with what
probability will these values occur?
(b) Suppose the measurement finds the value £5,4' What is the time-dependent state
of the system subsequent to measurement?
(c) What is the probability of finding particle I (with mass mj) in the interval
(0, a /2) at t=O?
Answer (partial)
(c) If the state of the two-particle system is ip(Xj , x2), the probability of finding
particle 1 in the interval dx; (independent of where particle 2 is) is

P(xj)dxj =dxj i a
lip(Xj ,X2)12dx2

8.29 Show that ipA (Xj , X2) in (8.95) may be written as a 2 x 2 determinant.
8.30 Show that ipA and ipS in (8.95) correspond to the same eigenenergy relevant to the
Hamiltonian (8.84).
8.31 In the event that two particles in a one-dimensional box are identical one must ask:
What is the probability of finding a particle in the interval dx about x ? In this vein,
show that for either ipS(Xj, X2) or ipA (xj , X2) , the two integrations

give the same functional form.

8.32 Consider two identical particles in a one-dimensional box of length a . Calculate the
expected value for the square of the interparticle displacement
Problems 323

in the two states vs and rpA . Show that

thus establishing that (in a statistical sense) particles in a symmetric state attract one
another while particles in an antisymmetric state repel one another. Such attractions
and repulsions are classified as exchange phenomena. They are discussed in further
detail in Chapter 12.

Answer
With rpnl(XI) represented by InIl, rpnl (X2) by liil ), and rpnl (XI)rp n2(X2) by Inlii2) ,
the symmetric and antisymmetri c states appear as

Thus
2)S
2(d ,A = «nlii21 ± (ii l n2I)d 2(ln l ii 2) ± liiln2»
= (n l ii 2Id2In l ii 2) + (ii l n2Id 2 Iii l n2) ± (n l ii 2Id 2 In l n2)
± (ii l n2Id 2 In l ii 2)
In the last two ± contributions with d 2 = xl xi -
+ 2xIX2 , only the -2XIX2 term
is found to survive. Consider the term (n l ii 2Id 2 Iii l n2) = -2 (nlii2Ixlx2Iiiln2 ) =
-2 (nllxlln2) (ii2Ix2 Iii l) ' In that (n l lx l ln2) = (ii Ilx2lii2) == XI2 (write out the inte-
grals and change variables) , one obtains

There results

8.33 For a single particle in a two-dimen sional box such as described in the text, one may
also construct symmetric and antisymmetric states. The symmetric state rps has the
property

rps(X, y) = vst» . x)
while the antisymmetric state has the property

What are the eigenstates, rpsand rp A , that corre spond to the energy 29E I ? The sym-
metry of these eigen states reflects the fact that there is no intrin sic distinction between
the diagonal halves of the box depicted in Fig. 8.26.
8.34 Construct the eigenstates and eigenenergies of a particle in a two-dimensional rectan-
gular box of edge lengths a and 2a. Take the origin to be at a comer of the rectangle .
324 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Account geometrically for the removal of most of the degeneracy present in the case
of the square, two-dimensional box described previously. The degeneracy present
for this configuration (e.g., the energy 5£ is doubly degenerate) is sometimes called
accidental degeneracy, in that it is neither exchange- nor symmetry-degenerate.

8.6 • TWO-DIMENSIONAL HARMONIC OSCILLATOR

The two-dimensional problem we consider now is that of a point particle of mass

m, constrained by a set of four coplanar, orthogonal springs, all with the same
spring constant K (see Fig. 8.27).

A2 A2 K K
H(x, y) = !!!...- + !!..L +
2m 2m
_x 2
2
+-l
2 (8.108)
H(x, y) = H(x) + H(y)
Again, we find that the total Hamiltonian partitions into two independent parts,
H(x) and H(y) . These are the Hamiltonians relevant to one-dimensional har-
monic oscillation in the x and y directions, respectively (see Sections 7.2 through

Particle
displaced
to tx, y)

Tracks

FIGURE 8.27 Two-dimensional harmonic oscillator. Springs are free to move on tracks
but are otherwise constrained to displacements parallel to the coordinate axes. All springs
have the same spring constant K.
8.6 Two-Dimensional Harmonic Oscillator 325

7.4). The eigenstates and eigenenergies of these Hamiltonians are

_~2 /2
f{Jnl (~) = Ani J(nl (~)e

E n l = lUvo (n l +~)
(8.109)
f{Jn2 (T]) = A n2 J(n2 (T])e_7/2 / 2

E n2 = 1Uvo (n2 +~)

The nondimensional displacements ~ and T] are defined by (7.46)

(8.110)

while J(n (n is the nth-order Hermite polynomial(7.58) and An is a normalization

constant (Problem 7.7).
Owing to the separability of fj (x, y), it follows that its eigenstates are the
product forms

(8.111)

while the eigenenergies of fj (x, y) are the sums

(8.112)

For example, the ground state of the two-dimensional harmonic oscillator is

(8.113)

This is the only nondegenerate eigenstate of the two-dimensional harmonic oscil-

lator. All the remaining states are degenerate. The order of the degeneracy of the
eigenenergy E n l n2 is obtained from (8.112), from which we see that any eigen-
326 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

function ({In, '({Jnz' whose indices nl', n2' sum to the value (nl + n2) corresponds
to the same eigenenergy, En,n z-

eigenfunctions corresponding to] = [({J~"({Jnz',' such that]

[ (8.114)
En,nz nl + n2 = nl + n2

For example, to find the eigenstates that correspond to the eigenenergy

E = 51iwo = (4 + l)nwo (8.115)

one must find all pairs of integers n I' and n2' that sum to 4:

(8.116)
(nl' , n2') = (0,4), (4,0), (1,3), (3, 1), (2,2)

It follows that E = 51iwo is a fivefold-degenerate eigenenergy. The five degener-

ate eigenstates are

({JO(~)({J4(1/) = ({J04
({J4(~)({JO(1/) = ({J40
({JI (~)({J3(1/) = ({J13 = eigenstates corresponding to E = 51iwo (8.117)

({J3(~)({JI (1/) = ({J31

({J2(~)({J2(1/) = ({J22

Of these five states, ({J04 suffers symmetry degeneracy with ({J40, as does ({J13 with
({J31 . On the other hand, the three states ({J04, ({J13, and ({J22 are accidentally degen-
erate with each other.

PROBLEMS

8.35 What is he order of degeneracy of the eigenstate

Es = lUuo(s + 1)

of the two-dimensional harmonic oscillator?

Answer
The degeneracy equals the number of ways of writing an integer s as the ordered sum
of two numbers . There are (s + 1) ways to do this.

8.36 (a) Write down the Hamiltonians, eigenenergies, and eigenstates for a two-dimen-
sional harmonic oscillator with distinct spring constants K x and Ky.
Problems 327

(b) If K, = 4K x , show that the eigenenergies may be written

En j n 2 = luvO (n 1 + 2n2 + ~)
where n I corresponds to x motion and n2 to y motion.
(c) For part (b), what is the order of degeneracy of E 2,3? List the correspond-
ing eigenstates. Account for the absence of symmetry degeneracy among these
states.
8.37 A right circular cylinder of infinite height and large, but finite radius is uniformly,
positively charged throughout its volume. The charge density is Po esu/cm 3. An elec-
tron moves in a plane normal to the cylinder. Its position is close to the central axis
of the cylinder (see Fig. 8.28).
(a) What is the electrostatic potential <I> near the central axis of the cylinder?
(b) What are the eigenenergies of the electron? [Hints: For part (a), use Poisson's
equation, V 2 <1> = -4np = -4npO. The radial operator in V 2, in cylindrical
coordinates, is r-Iajar(rajar). From symmetry you may assume <I> = <I> (r).
For part (b), note that the potential energy of the electron VCr) = -Iel<l>(r),
where r 2 = x 2 + y2.]

Motion of
electron
is restricted
to this domain

Uniformly positively
charged cylinder

FIGURE 8.28 Configuration for Problem 8.37.

8.38 A particle moves in the xy plane in the potential field

v= Vex) + V(y)
Vex) = VI = constant

V(y) = V2 = constant

at constant energy E. Give the time-dependent wavefunction 1/1 (x, y, t) of the particle
corresponding to the initial data

ljJ(x, 0, 0) = ljJ(O, y, 0) = 0
328 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

8.39 A particle of mass m is confined to move on the two-dimensional strip

-a < x < a , -00 < y < 00

by two impenetrable parallel walls at x = ±a.

(a) What is the minimum energy of the particle that measurement can find?
(b) Suppose that two additional walls are inserted at y = ±a. Can measurement of
n
the particle' s energy find the value 3rr 2 2 /8ma 2? Explain your answer.
8.40 Consider that three-dimen sional space is divided into two semi-infinite domains of
constant potential VI and V2.

Z> O

z ~ o

A beam of particle s carrying the current nkJiAI 2 1m particles /cme-s is incident on

the z = 0 interface and is in part reflected and transmitted. Particles in the reflected
beam have momentum nkl ' , while those in the transmitted beam have momentum
nk2 . The vectors kj , kl ' , k2 are all parallel to the xz plane. The configuration is
shown in Fig. 8.29.
(a) What is the wavefunction 0/1 appropriate to a particle in the upper half-space
z > O? What is the wavefunction 0/2 of a particle in the lower half-space (i.e.,
that of a particle in the transmitted beam) ?
(b) Determine the relation between the angles a , a', and a" through matching 0/1 to
0/2 and their derivatives across the z = 0 plane .
(c) Using the matching equations obtained in part (b) determine the transmission
coefficient T and reflection coefficientR. Show that T + R = 1.

z
z

k~ h
,/ 1
""- ,/
,/

" "

x x

(a) (b)

FIGURE 8.29 Orientation of k vectors for a beam of particles incident on the z = 0

plane at the angle a . (See Problem 8.40.)
Problems 329

Answers (partial)
(a) The wavefunctions of particles in the upper and lower half-spaces are

0/1 = Aei<Pl + Be i ¢>2

0/2 = C e i <P3

The phases 4>1, 4>2, and 4>3 are

4>1 =kl·r-wt=klxsina-klzcosa-wt
4>2 = kl ' . r - wt = klx sin a' + klzcosa' - tot
4>3 = k2 . r - cot = k2X sina" - k2Z cos a" - tot

(b) Matching 0/1 to 0/2 on z = 0 gives

For this equation to be satisfied for all x, it is necessary that the phases be equal.

kl sina = kl sina' = k2 sina"

sina k2 JE - V2 1- V2/E
a = a' and
sina" = kl = E - VI = ---=--
1- VI / E
=n

where n is the relative index of refraction. (Compare with Snell's laws of op-
tical refraction .) These results, together with the matching equation 80/II8z =
80/2/8Z on z = 0, give
A+B=C
k cos a"
A-B= 2 C
kl cosa
which serve to determine T and R : namely,
2
T = k2cos a" I£A 1
klcosa

The transmission coefficient is seen to involve only the normal components of

incident and transmitted fluxes.

8.41 Calculate the reflection coefficient of sodium metal for low-energy electrons as a
function of electron energy and angle of incidence. For electrons of sufficiently long
wavelength, the potential barrier at the metal surface can be treated as discontinuous .
Assume that the potential energy of an electron in the metal is -5 eV. Calculate the
"index of refraction" of the metal for electrons . (See Problem 8.40.)
8.42 A beam of electrons of energy E in a potential-free region is incident on a potential
step of 5 V at an incident angle of 45°. Is there a threshold value of E below which
all the electrons will be reflected? If so, what is this value?
330 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

Answer
Call the threshold (or "critical") incident energy Ec . Then at E c , the angle of refrac-
tion a" = n / 2. That is, at Ec the transmitted ray runs along the interface between the
two media . If E is increased above Ec (at the same angle of incidence a), electrons
penetrate the potential step and R < 1. On the other hand , if E is decreased below
Ec , there is no transmitted ray at all. The analytic manife station of this observation
is that a" becomes imaginary for E < Ec , while R maintains its value of unity for
all such values of E . From Snell's law (for a = tt / 4),

. I
Sill a = ..ti = n Sill. a "
where n is the index of refraction .

n = )1- i
At E c , sin a" = I and one obtains E c = 2 V = 10 V. The reflection coefficient is
given by

cos o - n cosa" I
R= Icosa + n cosa"
For the problem under discussion

cos a" = J1- sin2 a" = )1 _2n 1

2

The critical value of 2n 2 is 2n~ = I. If E < E c , then 2n 2 < I and cos a" becomes
imaginary so that for these values of incident energy, R assumes the form R = IZ/ zI,
z
where z is a complex number and is its conjugate. It follows that R = 1 for E < Ec .
8.43 This problem addresses the notion of supersymmetry' relevant to eigenenergies of a
given Hamiltonian.
(a) Working in one dimension , and in units !i 2 / 2m = I , and deleting hats over
operators, show that, apart from the ground state, the two Hamiltonians
2
H± = L + [W 2 ± W']
2m
W(x) = _(1f!o'No)
H_1f!o =0
have the same eigenenergies, where a prime denotes differentiation.
(b) Obtain H± relevant to the harmonic oscillator (with V = x 2). Show explicitly
that H-1f!o = o.

7For further discussion, see W. Keung , E. Kovac s and U. P. Sukhatme, Phys. Rev. Letts. 60, 41 (1988).
 8.7 Linear Combination of Atomic Orbitals (LCAO) Approximation 331

o
x

-vo
- - - - - - - - - - - 2 d - - - - - - - - - -.. -I
FIGURE 8.30 Configuration of two identical wells (l and 2) separated by the dis-
tance 2d.

8.7 • LINEAR COMBINATION OF ATOMIC ORBITALS

(lCAO) APPROXIMATION

The LCAO approximation is employed to estimate the states of a molecule in

terms of a linear combination s of quantum states ("orbitals") of isolated con-
stituent atoms. This formalism finds wide application in solid-state physics in the
tight-binding approximation."
In the present analysis, the LCAO method is employed to estimate the ground-
state energy and wavefunction of a given molecule.

The Molecular Ion

We will apply the LCAO method to the potential shown in Fig . 8.30, which , in
the present discussion, is representative of a one-dimensional model of a homonu-
clear, diatomic molecular ion, such as H2 +. For fixed intemucleon spacing this
example permits our analysis to remain purely one-dimensional, with the dis-
placement of the valence electron as the only free variable.
The Hamiltonian for our system is given by
A 2
if = Px + VI (x) + V2(X) (8.118)
2m
where VI (x) and V2(X) are identical square wells separated by the distance 2d, as
illustrated in Fig. 8.30 .
Let us introduce the normalized ground-state wavefunctions /PI (x) and /P2 (x)
relevant to well 1 and well 2, respectively, given by the symmetric even wavefunc-
tion (8.24). See Fig. 8.31. For wells sufficiently separated, d » a, it is evident
that

(8.119)

8For further discussion , see C. W. N. Cumper, Wave Mechanic s for Chemists , Heineman , London,
1966; W. A. Harrison, Electronic Structure and the Properties ofSolids, Dover, New York, 1989.
332 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

-d d
x

FIGURE 8.31 Ground states of isolated atoms corresponding to the configuration de-
picted in Fig. 8.30.

With these equalities at hand, we may write

HC{JI = EOC{JI
(8.120)
iJ IfJ2 = EoC{Jz

where Eo is the ground state of an isolated well. It follows that the ground state
of our composite decoupled molecule is doubly degenerate. Furthermore, as is
evident from (8.120), this ground-state energy is the same as that of either one of
the isolated atoms . See Fig. 8.32.
When atoms are brought together to form a molecular ion, it is assumed that
the Hamiltonian maintains the summational form (8.118) but that C{JI and IfJ2 are
no longer eigenstates of iJ .

Ground-State Energy and Wavefunction

To estimate the ground state of our molecular ion in the LeAO approximation,
one writes

C{J(x) = CI C{J I (x) + cZC{Jz(x) (8.121)

where C{JI (x) and C{Jz(x) remain centered about respective atomic sites.
To evaluate the coefficients CI and Cz , we recall Problems 4.28 and 4.30 accord-
ing to which the ground state may be obtained through the variational statement

8(E) = 0 (8.122)

x
-+-----+--------+--------+----t--Eo

FIGURE 8.32 In the limit that atoms grow uncoupled, the ground-state energy of the
composite molecule is the same as that of one of its isolated atoms.
8.7 Linear Combination of Atomic Orbitals (LCAO) Approximation 333

where

(E) = (rpIHlrp) (8.123)

(rplrp)

In developing the variation al statement (8.122), we assume CI, Cz, rpl, and rpz to
be real. There results

(8.124)

where

(8. I24a)

represents the overlap integral between rpl and rpz. Due to the symmetry of our
configuration,

As CI and Cz are the only free variables in (8.124) , to minimize (E) it suffices to
set
a(E} a(E}
-=-=0 (8.125)
aCI acz

There results

CI (Hll - (E)) + cz(Hl2 - (E}S) = 0

(8.126)
cI(Hl2 - (E}S) +cz(Hzz - (E}) = 0

With C representing a column vector with components (CI, cz), the preceding
equations may be written in the matrix form

Rc= 0 (8.126a)

where R is the implied coefficient matrix. A nontrivial solution to (8.126a) occurs ,

providing det R = O. There results

H ll - (E) Hl2 - (E}SI = 0

(8.127)
IHl2 - (E}S Hu - (E)

which gives

(E ) = Hll ± Hl2 (8.128)

± I ±S
334 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Substituting this finding into (8.126a) reveals that (E)_ corresponds to the odd
wavefunction (CI = -cz) and (E)+ to the even wavefunction (CI = cz). Conse-
quently, we conclude that the even wavefunction, C{J+, corresponds to (E)+ and
that the odd wavefunction, C{J- , corresponds to (E)_. These states, with corre-
sponding energies, are listed below (with (E)± replaced by E±).

HII - H12
E_=-----,----
I-S
(8.129)
+ H12
E+-----
HII
- 1 +S '

where the normalization constants are given by

I
A± == -~=i=±=S (8.129a)

For the configuration at hand (see Fig. 8.30) we may set

HII = -IHIlI
(8.130)
Hlz = -IHlzl

Substituting these values into (8.129) reveals that for sufficiently small wavefunc-
tion overlap, the ground-state wavefunction and energy for the molecule at hand
are given by

C{JG = C{J+

(IHIlI + IH121) (8.131)

I+S
The difference between new energies is

(8.132)

For small wavefunction overlap, S « I, we find

liE = 21H121 (8.133)

Thus we find that the interaction incurred by bringing the atoms in close proxim-
ity removes the twofold degeneracy of the uncoupled molecule. [Recall (8.120).]
This situation is depicted in Fig. 8.33.
The lower energy of the symmetric state, C{J+, may be understood on the ba-
sis of the following. In this state the valence electron has a finite probability of
occupying the mid-domain between positive ions, as opposed to the situation in
the antisymmetric state, C{J- , for which the electron has zero probability of being
found in this domain. Thus, the charge density associated with C{J+ tends to bind
the positive ions, thereby lowering the energy of the system. A more realistic de-
Problems 335
E_ (anl ibond ing state)

(::::: 1-- E; = Eo (bonding state)

FIGURE 8.33 The unperturbed twofold degeneracy in energy of the uncoupled molecule
is removed when atoms interact. For small wavefunction overlap, within the LCAO approx-
imation, the spread in new energies is 2IH12I. In the bonding state the valence electron
partially occupies the midregion between ions, thereby tending to bind them and lower the
energy.

scription of this situation is given in Section 12.7, which addresses the actual H2
molecule .
Additional problems in this category of estimating properties of a system are
encountered in Chapter 13 under the general topic of perturbation theory. In this
context the configuration d » a in the problem above is viewed as the unper-
turbed system. Perturbation is incurred due to the interaction between atoms,
which comes into play with decrease in the separation d . Note in particular that
the ground state in the present configuration is doubly degenerate and that the
perturbation removes this degeneracy. See Fig. 8.33. That is, with the interac-
tion between atoms "turned on," the twofold degenerate energy Eo is split into
two separate values, E±. Degenerate perturbation theory is described in detail in
Section 13.2.

PROBLEMS

8.44 Consider the following Gaussian form for the wavefunction of atom 1 corresponding
to the configuration shown in Fig. 8.30.

(a) What is the value of the normalization constant A ?

(b) What is the form of the companion wavefunction !P2(X)?
(c) Obtain an expression for the overlap integral S (8.124a). In what manner does
your expression vary with the parameter A == d l a'!
8.45 Show that E±, as given by (8.128), corresponds to (cl / c2) = ±1 in (8.121).
8.46 Establish the equality HII = H22. [See equation above (8.125).]
8.47 A linear molecule comprised of three identical , singly ionized atoms share a single
electron. This molecular ion is modeled by three identical quantum wells, each of
width a and depth Vo. The wells are separated from each other by the distance d and
are symmetrically displayed about x = o.
336 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

Let rpl , rp2, rp3 be the ground-state wavefunctions of the three separate wells.
(a) What is the net charge (in units of e) of the molecular ion?
(b) Write down the Hamiltonian for the ionized triatomic molecule . What are the
criteria for an uncoupled molecule ?
(c) What are the possible states rp of the uncoupled molecule in terms of rpl, rp2, rp3
for the case of equal probabilities of occupation among the three wells?
(d) What is the LCAO form of the ground state for the triatomic molecule ? Call your
constants CI , C2, C3'
(e) What is the form of (E ) for this molecule?
(f) Estimate the eigenenerg ies of this molecule corresponding to the wavefunction
you have constructed, through the null variation of (E ). Assume HII = H22 =
H33 == A, H12 = H21 = B , and H13 = H31 = S12 = S23 = S13 = o.
(g) What are the relations among the constants CI, C2, C3 corresponding to the en-
ergies you have found in part (f)? Sketch the corresponding wavefunctions and
compare with corresponding sketches for part (c).
(h) What is the ground-state wavefunction rpG and ground-state energy EG for this
molecule, within the LCAO approximation ?

Answers (partial)
(a) +21el
(b) rp = rpl ± (rp2 ± rp3)
(c) rp = CIrpl ± C2rp2 + C3rp3
8.48 A broad frequency band of photons is incident on a rare gas of homonuclear, singly
ionized, diatomic molecules . It is noted that photons offrequency v = 0.75 x 1015 Hz
are absorbed by the gas. Values of constants in the one-dimensional LCAO model of
this molecule are IHIlI = 10 eV, IH121 = 3 eV. What is the value of the overlap
integral S? State units of your answer.

8.8 • DENSITY OF STATES IN VARIOUS DIMENSIONS

We recall (see Problem 2.37) that the density of states 9 (E) is defined so that
9 (E) dE gives the number of energy states in the interval E , E + dE .
In this section we wish to obtain expressions for 9 (E), relevant to a particle
confined to boxes in one, two, and three dimensions. These results are then ap-
plied to a quantum-well defined in "slab geometry" important to electron-device
physics ."

One Dimension
For a one-dimensional box of width L eigenenergies are given by (4.14):

h2
EI--- (8.134)
- 8mL2
9For further discussion, see G. Bums, Solid State Physics, Academic Press, New York, 1985.
8.8 Density of States in Various Dimensions 337

It follows that there is an energy state for each value of the quantum number n.
As is evident from (8.134), in the limit of large L, the separation between energy
states diminishes and the energy spectrum grows quasi-continuous. 10 In this limit
we may write

Sn dn
g(E)=-~­ (8.135)
tlE dE

which, with (8.134), gives

1
gl(E) = 2JEIE (8.136)

where the subscript 1 on g denotes one dimension. Thus , for a particle confined to
a one-dimensional box , the density of states vanishes in the limit of large energy
or, equivalently, large quantum number.

Two Dimensions
The two-dimensional square box was discussed in detail in Section 8.5. With the
energy eigenvalue equation (8.107) we write

(8.137)

There is an energy eigenvalue for every pair of quantum numbers (nl, n2). It
follows that in Cartesian n 1n2 space, there is an energy eigenstate at every point
(nl , n2) of this space. (Compare with Fig. 2.17.) Let

denote the radius vector in this space. Thus, all points that lie in the annular region
(n, n + dn) have the same energy (E oc n 2). Again , in the extreme of large L,
eigenenergies form a quasi-continuum and the number of such points is given by
the area of the annular region . There results

1 rr
g(E)dE = - x Zn n dn = - d E (8.138)
4 4EI

The factor! insures that only positive quantum numbers are included. It follows
that in two dimensions, the density of states is given by

(8.139)

which is constant.

IOSince energy increments grow with n, this quasi-continuous spectrum is realized for unbounded L.
338 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

Three Dimensions
The energy eigenvalue equation for a particle confined to a three-dimensional
cubical box is given by (see Table 10.2)

(8.140)

In this case there is an energy eigenvalue corresponding to each point in Cartesian

nln 2n3 space. Again we define the radius vector

and note once more that in the limit of large L all points within the spherical shell,
n, n + dn, have the same energy. There results

and

(8.141)

where the factor of kin the preceding equation insures that only positive quantum
numbers are included.
Here is a recapitulation of preceding results.

(8.142a)
91 = 2JEIE
n
92 = 4EI (8.142b)

Jr,,[E
93 = "4 E13 / 2
(8.142c)

If the particles we are considering have spin !,

then each of the 9-values above is
multiplied by the factor 2. (The concept of spin is discussed in Chapter 11.)

Density of States per Unit Volume

With (8.134) we write

(8.143)

where Q' is the inferred constant, which is seen to be independent of the box di-
mension L. Inserting this value into (8. 142) gives the following expression for the
8.8 Density of States in Various Dimensions 339

density of states per unit volume for a particle confined to boxes in one-, two-,
and three-dimensional Cartesian space, respectively.

_ 91 1
91 = L = 2.;aE (8.l44a)

_ 92 tt
92 = L2 = 4a (8.l44b)

_ 93 7r.JE (8.l44c)
93 = L3 = "4 a3/2
(See Problem 8.49.) Note that in d dimensions, 9d oc E(d-2) /2.

Slab Geometry
We wish to apply the preceding results to calculate the density of states of a par-
ticle of mass m confined to a well in slab geometry defined by the following
potential:

v=o (8.145a)

v= 00 , elsewhere (8.145b)
a «L (8.l45c)

See Fig. 8.34.

The density of states for the particle so confined is an appropriate combination
of 91(0) (8.142a) relevant to the one-dimensional z confinement and 92(L) (8.142b)
relevant to the two-dimensional xy confinement.

z
y

FIGURE 8.34 Sketch of the quantum well corresponding to the potential (8.145), with
L » a.
340 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

We first note that the energy eigenfunctions of this particle may be written in
the product form [compare with (8.111)] :

rp(x, y, z) = rpa(z)rpLCx, y) (8.146)

where

rpa(Z) =
A.-
a
SIn--
nlJrZ
a
(8.l46a)

is relevant to the one-dimensional Z box and

2 . n2Jr x . n3JrY
rpLCx, y) = - SIn - - S I n - - (8.146b)
L L L
is relevant to the two-dimensional xy box.
Eigenenergies are given by the sum

(8.147)

As L » a, EI(L) « EI(a) [recall (8.134)] . For example, if a = L/1O, then

E I (a) = 50E I (L). See Fig. 8.35. However, as the wavefunction of the confined
particle is the product (8.146), no motion exists for energy less than the ground-
state energy of the Z motion . Thus, 9 (E < E I (a» = O. Furthermore, the density
of states for the xy motion is constant. [See (8.142) .]
Thus, for the given configuration we have the following picture. The density
of states 9 (E) is zero for E < E I (a) . At E = E I (a), two-dimensional motion is
allowed and 9 (EI (a» = g2(L) = x /4EI (L). For larger energy, 9 (E) maintains

FIGURE 8.35 Sketch of the eigenenergies corresponding to the z and xy motions in the
quantum well (8.145) for the case a = L / 10. States with energy E < E I (a) do not exist.
8.8 Density of States in Various Dimensions 341

this constant value until the second excited state of the z motion is reached . At
this value, 9 (E2(a») = 292(L) as now there are two allowed (z) modes allowing
for an additional gz<L ) component.
This sequential development of 9 ( E) leads to the formul a

rr 00
gee ) = - - " O(E - E (a» ) (8.148)
4Et (L) ~ n

where O(x ) is the step function

O(x) = I , x >O
(8.148a)
O(x) = 0, x <O

See Fig. 8.36.

As a check on our calculation we should find that

9 (E) ---+ 93(E)

in the limit

as is evident from Fig. 8.36.

FIGURE 8.36 At each step of the graph another quantum state (8.146) comes into play.
The step function shown above applies to all values of L » a. In the limit a --+ L , and
L large, 9 (E ) --+ 93(E ), appropriate to a large cubical box. In this same limit, the two
dashed curves coalesce.
342 Chapter 8 Finite Potential Well, Periodic Lattice, and Some Simple Problems

Envelope of [f:
n=I
B(E - En (Q»)] = J ~Q) EI

In the limit that a = L, which is large, the increments of the B function grow
closer and B is well approximated by the envelope function. Thus

]fiE
9(E) -* 4EI(L)yE;m = 93(E)

In device physics, the configuration in slab geometry considered above is

called a "quantum well" and is realized by epitaxial growth of layers of semicon-
ductors with different band gaps (see Section 12.9). Interfaces between different
semiconductors are called heterojunctions. A multilayered structure of alternate
semiconductors effects an array of quantum wells and is called a superlattice.

PROBLEMS

8.49 (a) Obtain the density of states per unit volume for a free particle of mass m moving
in 3-space. Call your answer 90(E) .
(b) Compare your expression for 90(E) with 93(E) as given by (8.l44c). Explain
the difference (or equality) between these results.

Answers
(a) In six-dimensional Cartesian x - p space there is an energy state at each point
(x, p) . Due to the uncertainty relation , the minimum volume for a state in phase
spacel ' is h 3 . Thus,
_ dxdp
90(E) dE = - 3-
h
For a free particle

p2
E=-
2m

so that states in momentum space in the spherical shell (p, p + dp) all have the
same energy. As the volume of this shell is 471" p2 dp, we obtain

With the preceding expression there results

_ (2m)3 /2 71" JE _
90(E) = 271"--3
h
-.fE = -
4 a
3/2 = 93(E)
!,
11 If the particle we are considering has, say, spin then this minimum volume is decreased by !.
The concept of spin is discussed in detail in Chapter 11.
Problems 343

(b) We conclude that 90(E) and 93(E) are equal. However, it should be borne in
mind that in summing states, 93 (E) is summed over the discrete energies of the
box , whereas 90(E) is integrated over the continuum of free-particle E-values.
We may conclude that in any finite volume and energy interval, there are more
free-particle energy states than box energy states.

8.50 Describe qualitatively how Fig . 8.36 changes if the well in the z direction of the
quantum well is of finite potential height, Va.

8.51 A particle of mass m is confined to a one-dimensional box of edge length L. Let

the edge length be increased to L + oL. If t.oE represents the difference between
adjacent energies for the first box and t.l E the difference of adjacent energies for
the expanded box , show that

That is, show that energies grow closer with increase in L.

8.52 A particle of mass m is confined to move in two dimensions within an impenetrable

rectangular barrier with edge lengths (a, b) . Let the rectangle be situated in the first
quadrant of Cartesian space with one corner at the origin and adjoining edges aligned
with the (x, y) axes .
(a) Working in this configuration, obtain the eigenstates, CPnn'(x, y), and eigenener-
gies , Enn" of the particle. Write eigenenergies in terms of Eo, the ground state
of a particle in a one-dimensional box of edge length a.
(b) Show that if a / b is a rational number, then states of the system are degenerate.

8.53 Derive expressions for the normalization constants B+ = 2B, B_ = 2i B , appropri-

ate to the even and odd wavefunctions, (8.24) and (8.21), respectively.

Answer
2kaJ
1/(2B±)2 1
= -[ka ± sinka coska] + -I [cos
.2
k K sm ka

8.54 Consider the Schrodinger equation with a periodic potential of period a. Are the
eigenfunctions of this equation necessarily periodic? Justify your answer.
8.55 What is the quantum mechanical current density (7.107) for a wavefunction that is
(a) purely real , (b) purely imaginary, or (c) a complex constant times a real function?

8.56 (a) Obtain an expression for the characteristic wavelength, A, of emission due to an
electron in an enclosure of diameter a.
(b) At what value of a is A :::::: a? What is the energy of photons at this value of A
(in eV)?

Answers
(a) Working in one dimension we find,

j c: El /h = h /8ma 2
e 8mea 2
A=-=--
f h
344 Chapter 8 Finite Potential Well , Periodic Lattice, and Some Simple Problems

(b) With AC == hf mc, we obtain,

8a 2
A= -
AC
At A = a, we find A = AC18. The Compton wavelength, AC, corresponds to the
frequency, f :::::: 1021 Hz. Photons at this wavelength have energy E :::::: 1 MeV
(y-radiation).

8,57 (a) Starting with the relation for group velocity (8.44), derive the scalar expression
for effective mass, m *, given by (8.45).
(b) In the preceding it was assumed that the matrix Vk Vk E is diagonal so that ap-
plied force is collinear with acceleration . More generally, this is not the case. In
this event the effective mass becomes a matrix of nine components (correspond-
ing to components of effective mass in the directions xx , xy , . . . ). In "vector
notation" this matrix is written as the dyad, iil *, or, more precisely,

Derive an expression for this dyad. (Note that the inner product of a dyad and a
vector is the product of a matrix and a column vector.) What is the expression for
the xz component of the (inverse) density matrix?
Answer (partial)
(a) In vector notation, the starting equation becomes

The force equation is written

dk
I i - =F
dt
Inserting this expression in the preceding one obtains

which gives the desired expression

8.58 A system composed of a particle of mass m confined to a quantum well of width 2a

and depth V has (N + 1) bound states, where N » 1. The eigenstate of largest energy
of this system (with respect to E = 0 at the bottom of the well) corresponds to the
Problems 345

value ~ = N (sr/ 2) + E , where E « 1. Show that the ground-state energy measured

from the bottom of the well corresponds to the value

1l' 1
~= 2- N

Answer
We must solve the equation

or, equivalently,

Consider the solution ~ = (1l' / 2) - E', where E' « 1. As cos[ (1l' / 2) - E'l = sin E',
we obtain
l!. -E' n n
_2 ~--N(-)
sin e' - 2E' - 2
which gives the desired result

, 1
E=-
N
8.59 Show that a particle of mass m confined to the potential well

V(x) = -aVoS(x)
has one bound state. In the preceding expression , a is a length, Vo is a constant
potential , and S(x) is the Dirac delta function . (Compare with Problem 7.68.) In the
present version , this problem should be examined in the quantum-well formalism of
the present chapter.
 PART
II

Further Development of the Theory

and Applications to Problems
in Three Dimensions
CHAPTER

9 Angular Momentum

9.1 Basic Properties

9.2 Eigenvalues ofthe Angular Momentum Operators
9.3 Eigenfunctions ofthe Orbital Angular Momentum Operators L2 and i;
9.4 Addition ofAngular Momentum
9.5 Total Angular Momentum for Two or More Electrons

Our study of the applications of quantum mechanics to three-dimensional prob-

lems begins with a description of the properties of angular momentum. Wefirst
consider orbital angular momentum, which is closely akin to angular momen-
tum encountered in classical physics. Angular momentum in quantum mechanics,
however, is a more general concept than its classical counterpart. In quantum
mechanics, in addition to orbital angular momentum, one also encounters spin
angular momentum. Spin angular momentum is an intrinsic, or internal, prop-
erty of elementary particles such as electrons and photons, and has no classical
counterpart. The operators corresponding to the Cartesian components of angu-
lar momentum in quantum mechanics obey a set offixed, fundamental commu-
tator relations. These relations are first derived for orbital angular momentum
and then employed as the defining relations for angular momentum in general.
Eigenvalues of angular momentum stemming from these commutator relations
are obtained, and it is at this point that a distinction between orbital and spin
angular momentum first emerges.

9.1 • BASIC PROPERTIES

The significance of angular momentum in classical physics is that it is one of the

fundamental constants of motion (together with linear momentum and energy) of
an isolated system. As we will find, the counterpart of this statement also holds
for isolated quantum mechanical systems. This conservation principle for angular
momentum stems from the isotropy of space. That is, as described previously
in Section 6.3, the physical laws relating to an isolated system are in no way
dependent on the orientation of that system with respect to some fixed set of axes
in space.

349
350 Chapter 9 Angular Momentum

Classically, angular momentum of a particle is a property that depends on the

particle's linear momentum p and its displacement r from some prescribed origin.
It is given by (see Fig. 1.9)

L= r x p (9.1)

One may also speak of the angular momentum of a system of particles or of a

rigid body. For such extended aggregates, one must add the angular momentum
of all particles in the system to obtain the total angular momentum of the system.

Cartesian Components
The classical Cartesian components of the orbital angular momentum L for a
particle with momentum p = (Px, Pv- pz) at the displacement r = (x, y , z) are

L, = yp z - ZPy
L y = zpx - xpz (9.2)
L z = XP y - YPx

The quantum mechanical operators i.; i.; and i.; corresponding to these
observables, derive their definitions directly from the classical expressions above,
with p replaced by its corresponding gradient operator. There follows

A
Lx = ypzA A

- ZPy =
AA 'f«
-1/£
a a)
y- - Z-
az ay

L y = zpx - xpz =
A A A A A ' f<
-1/£
(a
z- - xa)
- (9.3)
ax az
A A
L z = XPy - YPx =
A A A ' f«
-1/£
a a)
x- - y-
ay ax

In terms of the three-dimensional vector linear momentum operator

A A A

p = (Px, Ps- pz) =

A

-In (a
•

ax'
a a) .
ay ' az = -lnV (9.4)

the equations above may be written as the single vector equation

L = -inr x V (9.5)

Commutator Relations
Let us examine the commutation properties of these operators . If, for example, i;
does not commute with i.; then these components of angular momentum cannot
be simultaneously specified in a single state, that is, these operators do not have
common eigenfunctions .
9.1 Basic Properties 351

To examine this specific question, we employ the basic commutator relation

[x, Px] = in (9.6)

There follows

[Lx, Ly] = i.L, - i.L, (9.7)

= (y pz - Zpy)(zPx - xpz) - (zPx - xpz)(y pz - ZPy)

= Xpy(zPz - pzz) - YPx(zPz - pzz)
= in(xpy - YPx)

= tni;
In similar fashion we obtain

[L y , Lzl = tni.,
[L z , Lx] = mi; (9.8)

[Lx, Ly] = tni,

These commutator relations are sometimes combined in the single vector equation

(9.9)
which in determinantal form appears as

ex ey e,
in(exLx + eyLy+ ezL z) = L, L, i; (9.10)
i., L, i;
As illustrated in Problem 9.1, only one of the three Cartesian components of
angular momentum may be specified in a quantum mechanical state. Suppose, for
example, that tp is an eigenstate of i.; What will measurement of i; find? To
answer this question we must bring the superposition principle into play. Expand
tp in the eigenstates of i.; The squares of the coefficients of expansion give the
distribution of probabilities of finding different values of Lx .
Although no two values of the Cartesian components of angular momentum
can be simultaneously specified in a quantum mechanical state, if one component,
say the value of L z, is specified, it is still possible to specify an additional property
of angular momentum in that state. This additional property is the value of the
square of the total angular momentum, L 2 , or, equivalently, the magnitude of L
(L=~=ffi).
The total angular momentum operator is the vector operator

(9.11)
352 Chapter 9 Angular Momentum

from which we may form L2 .

(9.12)

To show that there are states of a system in which L z and L 2 are simultaneously
specified, one need merely show that i , and L2 commute. Then we know that
these operators have simultaneous eigenfunctions. That is, there are states that are
eigenfunctions of both i; and L2 . Let us prove the commutability of i , and L2 .
~ ~ 2 ~ ~ 2 ~ 2 ~ 2
[L z, L ] = [L z, t., + t., + t., ]
= [L z , t., 2] + [L z , c, 2] + 0
= Lx[L z , Lx] + [Lz , Lx]L x + L y[L z , L y] + [Lz , L y]L y
= ili[LxL y + L yL x - L yL x - L xL y]
=0

In similar manner we find that and i., also commute with L2 . This must
i;
be the case because we have in no way given any special significance to the z
direction. In general
~ ~ 2 ~ ~2 ~ ~2

[Lx, L ] = [L y, L ] = [L z , L ] = 0
(9.13)
~

[L, L ]
~2

=0
It follows that the Cartesian components of L have simultaneous eigenfunc-
tions with L2 . However, the individual components of L do not have common
eigenstates with one another (except for the special case of zero angular momen-
tum) . These properties are depicted in a Venn diagram in Fig . 9.1.
The preceding discussion tells us that L2 and i.; say, have common eigenfunc-
tions. Let us call these eigenfunctions ({JIm. The integer indices I and m are related
to the eigenvalues of L2 and i, as in the following eigenfunction equations.

L2({Jlm = li 21(l + l)({Jlm (/=0,1 ,2, .. .)

(9.14)
Lz({Jlm = nm({Jlm (m = -I, ... , +1 in integral steps)
(These equations are derived in the next section.) The form of the first equation
indicates the following. Suppose that a system (e.g., a wheel) is rotating some-
where in space, far removed from other objects. We measure the magnitude of its
angular momentum. What possible values can be found? The values that exper-
iment finds are only of the form L = Ii.JI(l + 1), where I is some integer. For
example, one would never measure the value L = 1i,J7, since it is not of the form
L 2 = li 21(l + I). There is no integer for which I (I + I) = 7. This is similar to
the fact that a particle in a one-dimensional box is never found to have the energy
E = 7 E I . This value does not fit the energy eigenvalue recipe E = n 2 E I.
9.1 Basic Properties 353

Eigenstate
/\ A')
of Ly and L~

Eigenstate
of 22 but not
of Lx, ~y, or Lz

FIGURE 9.1 Venn diagram for the eigenstates of i ; i y , i.; and i 2 . Every point rep-
resents an eigenfunction of i.2 . Depending on which sector the point is in, it is also an
eigenfunction of i ; i. y, or i.z- The state at the center is the null eigenvector of Land
i.2 . It corresponds to the eigenvalues Lx == L y == L z == O. Peripheral points depict states
that are eigenstates of i.2 only. Can you think of one such function? Note that the space of
eigenstates of i.2 is "bigger" than the space containing all the eigenstates of i-, i.v- and
i.; Compare with Fig. 5.12.

Suppose that we measure the magnitude of angular momentum of the wheel

and find the value L 2 == 30lt2 . This corresponds to the I value I == 5. Having
measured L 2 , the system is left in an eigenstate of L2 . What value does subsequent
measurement of L; yield? The answer is given by the form of the eigenvalues of
L, given in (9.14). For the case in point, since I == 5, L, can only be found to have
one of the eleven values

t., == 51t, 41t, 31t, 21t, It, 0, -L; -21t, -31t, -41t, -51t
Suppose that measurement finds L: == 31t. Then the wheel is left in the state C{JS,3.
The form of equations (9.14) indicates that the eigenvalues of L2 are (21 + 1)-
fold degenerate. For the problem considered, all the eleven states C{Js,s; C{JS,4; ... ;
2 2 2
C{Js,-s correspond to the same value of L (i.e., L == 30lt ) . (See Fig. 9.2.)

Uncertainty Relations

Angular momentum is a vector. The square magnitude of this vector is given

by L 2 . Having measured L 2 , it is possible to measure any of the three Cartesian
components of L and leave the system (such as a wheel, a particle, an atom, a rigid
rod, etc.) with the same value of L 2 that it had before measurement. Specifically,
suppose that we measure L 2 and L, and find the values 56lt2 and 31t, respectively
(I == 7, m == 3). We know that the system is left in a simultaneous eigenstate of
L 2 and L z , namely, C{J7,3.
It is impossible further to resolve the state of the system. We cannot obtain
more information on the vector L without destroying part of the information al-
354 Chapter 9 Angular Momentum

511
411
311
211
~---III

i:::::::------O
L 2=1I 2/(/+I )
----III
/ =5
-211
-311
-411
-511
2
FIGURE 9.2 The eigenvalue L = ;,,21(1 + I) is (21 + I)-fold degenerate . For a fixed
magnitude, L = ;"../1(1 + 1), there are only 21 + 1 possible projections of L onto a given
axis. (See Fig. 9.3c.)

ready known . Suppose that Lx is mea sured and the value 5!i is found. In measur-
ing Lx, the information about L , previously determined is destroyed. I The system
is left in a simultaneous eigen state of i.2 and i.; Since this is not an eigen state
of i.; subsequent measurement of L z is not certain to yield any specific value.
Similarly for measurement of L y . Thi s conclusion is contained in the uncertainty
relation

(9.15)

Con sider the case of a wheel whose center is fixed in space. L 2 and L , are mea-
sured. What motion of the wheel will preserve these values but not preserve L;
and L y ? A very worthwhile model for such motion is given by a cla ssical solution
in which the angular momentum vector of constant magnitude prece sses about the
z axis at a constant inclination to that axis (see Fig. 9.3), thereby maintaining L z .
(Such motion is realized by a spinning top, with fixed vertex, in a gravity field.)
In the classical problem L is precisely determined as a function of time. At
any instant L may be observed and completely specified. Not so for the quantum
mechanical motion. Ifthe wheel is in an eigenstate of L2 and i.; it is in a super-
position state (i.e., a linear combination of the eigenstates) of i., or i.; At best
one can only speak of the probability of finding a certain value of L x or L y . If a
system is in such a state with definite 1 and m values, it is therefore more consis-
tent to view the related configuration as one in which the L vector is uniformly
e
spread over a cone about the z axis with half apex angle = cos"! m] JIU + 1)
(see Fig . 9.4).

ITh at is. the outcome of subsequent measurement of L z is rendered more uncertain.

 9.1 Basic Properties 355

direction of
precession

successive directions
of the angular
momentum vector L
fixed center

(a) (b)

z direction

(c)

FIGURE 9.3 (a) The angular momentum vector L of a rotating wheel whose center is
fixed in space. (b) Classical precession of L about the z axis with the constant projection
L z- (c) For 1 == 2, L 2 == 6!i2 . The only possible orientations of L onto the z axis are the five
values shown. The precessional motion depicted preserves L 2 and L z- () == cos -1 2/-/6 is
the smallest possible angle between L and the z axis.
356 Chapter 9 Angular Momentum

FIGURE 9.4 For the quantum mechanical state in which L 2 and L z are specified, L may
be pictured as being uniformly distributed over the surface of a cone with half-apex-angle
f) == cos- 1 m/,Jl(T+T).

For a given value of L [i.e., liJl(l + 1)] the maximum value of L; is hl, But
1 < Jl (1 + 1). It follows that the angular momentum vector is never aligned with
a given axis. Furthermore, there are only a discrete, finite (21 + 1) number of incli-
nations that L makes with any given axis. This extraordinary property (classical
physics permits a continuum of inclinations) is sometimes called the quantiza-
tion of space. For reasons that will become clear in the following sections, 1 is
often referred to as the orbital quantum number while m is often referred to as the
azimuthal or magnetic quantum number.

Orbital Versus Spin Angular Momentum

The commutator relations (9.8) are the trademark of angular momentum in quan-
tum mechanics. Although they are consistent with the differential and coordinate-
momentum operator relations (9.3 et seq.), they may be taken independent of
these and assumed to be the defining relations for quantum mechanical angular
momentum. When such is the case, angular momentum need not refer to the space
coordinates or linear momentum components of a particle, since the relations (9.8)
by themselves do not. The first example that incorporates this concept is given in
Section 9.2, where the eigenvalues of angular momentum are obtained using only
the commutator relations (9.8). As will be shown in Section 9.3, only a subset
of these eigenvalues are relevant to orbital angular momentum. Orbital angular
momentum derives from the space and momentum coordinates of a particle and
is akin to classical (r x p) angular momentum. In contrast, spin angular momen-
tum does not relate to a particle's coordinates or momenta, nor are the eigenstates
Problems 357

of spin dependent on boundary conditions imposed in coordinate space. Spin, as

mentioned previously, is an internal property of a particle, like mass or charge .
It is an extra degree of freedom attached to a quantum mechanical particle and
must be prescribed together with the values of all other compatible properties of
a particle in order to designate the state of the particle. In particular, note that spin
wavefunctions do not have a coordinate representation. The properties of spin are
developed in detail in Chapter 11.

PROBLEMS

9.1 Show that, if a state exists which is a simultaneous eigenstate of i, and i.; this state
has the eigenvalues Lx = L y = L z = O.

Answer
Let I(i be the said state. Then

It follows that I(i is an eigenstate of i, corresponding to the eigenvalue L z = O. (I(i is

a "null eigenfunction" of Lz .) From the uncertainty principle, (5.94) and (5.95), and
the fact that I(i is an eigenstate of i, and L, we find that

Since I(i is an eigenstate of i.; there is no spread in the values obtained on measure-
ment of L y in this state. This fact, combined with the preceding equation, gives

(L y ) = Ly = 0
Similarly, Lx = O.
It follows that a state of a system corresponding to finite angular momentum cannot
be a simultaneous eigenstate of any two of the Cartesian components of L Further-
more, from the defining equations for i.; i.; and i, (9.3), it follows that any constant
is a simultaneous, null eigenfunction of Lx, Ly , and i.;

9.2 Show that i., and f2 are Hermitian.

Answer (partial)
To prove the Hermiticity of i.; we must show that
, 't
Lx = Lx
or, equivalently, that

Look at the ypz term.

The last two equalities follow from (a) pz and y are Hermitian and (b) [y, Pzl = o.
358 Chapter 9 Angular Momentum

9.3 Measurements are made of the angle f) that L makes with the x axis of a collection
of noninteracting rotators, all of which are known to have angular momentum L =
1ivts6. What is the minimum f) that will be measured?
, , , A , A . A2 A2
9.4 If [A, Lx] = [A, L y ] = [A, Lzl = 0, what IS [A • L ]?

9.2 • EIGENVALUES OF THE ANGULAR MOMENTUM OPERATORS

In this section we derive the eigenvalues of angular momentum that follow from
the commutator relations (9.8). Eigenvalues relevant to two classes of angular
momentum emerge: orbital and spin. In the remainder of the text j will be used
to denote angular momentum in general while t will be reserved for orbital an-
gular momentum and S for spin . The operator j may represent t, or S, or the
combination t + S. The defining relations for the components of j are

[Jx~ J y ] = mi,
[Jy , J~l = iliJx (9.16)

[Jz, Jxl = iliJ~

J2 = Jx2 + Jl + J~2 (9.17)

The components of j obey all rules obtained above from the commutator relations
(9.8). These include:
A A2 AA2 AA2
[Jx , J 1= [Jy , J 1= [Jz, J ] = 0

!1Jx !1J y ~
n
2"1 {Jz}l
(9.18)

Ladder Operators
We seek the eigenvalues of J2 and i;
To facilitate the derivat ion we introduce the
"ladder operators" J+ and L. The reader will find these similar to the annihilation
and creation operators (a, at) introduced in Section 7.2. The ladder operators are
defined according to

J"-+ = i , + t i,
A_A .A_At
(9.19)
1- - Jx - I Jy - J+
Some immediate properties of these operators are

[iz, J+l = t.i;

[Jz , i-l = -Iii-
(9.20)
[p, J+l = 0
[P,Ll=O
9.2 Eigenvalues of the Angular Momentum Operators 359

The latter two equations follow from (9.8). To establish the first two relations one
merely inserts the definitions of i; and L. For example,

[iz , i+l = [iz, i x + iiyl = [J~, ixl + i[iz, iyl (9.21)

= mi, - i • inix = n(ix + iiy) = tii;

Other relations that J~ and L satisfy are

Consider the relation

.. 2
J = (Jx - IJy)(Jx + IJy) + Jz2 + nJ. .z
A ..... .... • A A

(9.23)
A2 A2 A2 . A A AA A
= Jx + Jy + Jz + I(JxJy - JyJx) + nJz
With these relations between i 2, i; i+, and L established we tum to construc-
tion of the eigenvalues of i, and P,
Let

(9.24)

We wish to show that m is either an integer or an odd multiple of one-half.

Consider the operation

J~i+({Jm = ini; + i+iz)({Jm = (ni+ + i+nm)({Jm

(9.25)
iz(i+({Jm) = him + 1)(i+({Jm)
where we have employed (9.21). The latter equation (9.25) implies that i+({Jm is
an (unnormalized) eigenfunction of J~ corresponding to the eigenvalue n(m + 1).
That is,

(9.26)

Applying J~ again gives

(9.27)

In a similar manner, we obtain

J-({Jm-l = ({Jm-2 (9.28)

360 Chapter 9 Angular Momentum

Thus we have found a scheme of generating a sequence of (unnormalized)

eigenfunctions of i, from a single eigenfunction rpm, with successive values of m
in the sequence differing by unity.

(. .. , rpm-2, rpm-I, rpm , rpm+l , rpm+2 , ...) (9.29)

Since }2 commutes with i; these operators have common eigenfunctions. Let

rpm be a common eigenfunction with the eigenvalue li2K 2, that is,

(9.30)

Operating on this equation with i; gives [using the third equation in.(9.20)]
(9.31)

The last equality asserts that i.»; = rpm+1 is also an eigenfunction of }2 corre-
sponding to the eigenvalue li2K 2. It follows that the sequence of eigenfunctions
of i, found previously (9.29) are all eigenfunctions of }2 corresponding to the
same eigenvalue li2 K 2 . How many such eigenfunctions are there? From (9.30)
one obtains

(1 2) = li2K 2 = (J/) + (J/) + (1/)

(9.32)
li2K 2 = (1/) + (J/) + li2m 2
where the average has been taken in the rpm state . It follows that

li2K 2 ~ li2m 2 (9.33)

(recall (lx 2 ) ~ 0; see Problem 4.13) or, equivalently,

IKI~lml (9.34)

For a given value of K · > 0, the possible values of m in the sequence (9.29) fall
between + K and - K . If mmax is the maximum value that m can assume for a
given magnitude of angular momentum, liK , then

(9.35)

Similarly,

(9.36)

From (9.22) and the last two equations, one obtains

A2 2 2 A2 A
1 rpmmax = Ii K rpmmax = l z rpmmax + lilzrpmmax
li2K 2 = li 2 m max (mmax + 1) (9.37)
9.2 Eigenvalues of the Angular Momentum Operators 361
A2 2 2 A2 A
J V?mmin = Ii K V?mmin = JZ V?mmin - IiJZV?mmin
1i2 K 2 = 1i2mmin(mmin - 1)

It follows that

mmax(mmax + 1) = mmin(mmin - 1) (9.38)

which is satisfied if

mmax = -mmin (9.39)

The possible values that m may assume for a given value of J2 form a sym-
metric sequence about m = 0 (see Fig. 9.5).
Let us call

mmax == j (9.40)

Since m runs from - j to +j in unit steps, one obtains

j = an integer if m =
0 is included in the
sequence of m values
(9.41)
j = 21 x an odd integer if m =
0 is not included in the
sequence of m values

Furthermore, if j is an integer, the related m values are integers. If j is an odd

multiple of one-half, the related m values are odd multiples of one-half (Fig. 9.6).

----------+K

successive values
m =0 - - - - - - - - - - ofm differ
by unity

mm in +1
---------- - K
" \ "
Jz'Pm =tlm'Pm J 2'Pm =11 2 K2'Pm
FIGURE 9.S The possible values that m may assume , for a given value of J2 = ti,2 K 2,
form a symmetric sequence about m = O.
362 Chapter 9 Angular Momentum

, . . - - - - m=2
, - - - - m=3 /2
,_---m=1
. _ - - - m = I /2
; = 2 ----f------ m = 0 ; =3/2 - - - -«:
- - - - - - m=-1 /2
' - - - - - m=-(
' - - - - - m = -3/2
' - - - - - m =-2

FIGURE 9.6 The angular momentum quantum number i. which enters in the eigenvalue
expression J2 = li 2 j (j + 1), may be either integral or an odd multiple of one-half. In either
case, for a given value of j, the azimuthal quantum number, m , runs from - j to +j in unit
steps .

In either case, inserting j = mmax = -mmin into (9.37) gives the form of the
eigenvalues of P;

(9.42)

Angular Momentum Eigenstates

In this manner we find that the eigenvalues of j2 and i, take the form

J2 = n?j(j + I)
(9.43)
Jz = timj (m j = - j , ... , +j)

with j an integer or half an odd integer. The structure of these eigenvalue equa-
tions is very significant and is another trademark of quantum mechanical angu-
lar momentum. In that they stem directly from the commutation relations (9.16),
which in tum are obeyed by all quantum mechanical angular momenta, it follows
that such eigenvalue relations are also appropriate to orbital angular momentum,
L; spin angular momentum, 8; or their sum, L + 8. Such, for example, are the
eigenvalues of £2, i , as given in (9.14).
As will be shown in the following section, boundary conditions imposed on the
common eigenstates of L2 , i; suggest that the related eigenvalues (l, ml) be inte-
gral. Thus, of the entire spectrum of quantum angular momentum j values, only a
subset (I = j = integer) corresponds to orbital angular momentum. The complete
j spectrum (integral and half-odd-integral values) will be found to correspond to
either spin angular momentum or the combination of spin plus orbital angular mo-
mentum. An example of the latter case is given by atomic electrons which have
both orbital and spin angular momentum and for which one must write j = L+8.
For the present we will concentrate on orbital angular momentum.
The eigenvalue equations for the orbital angular momentum operators L2 and
i, (with m written for ml), together with the equations for L±, appear as
9.2 Eigenvalues of the Angular Momentum Operators 363
A2 2
L f{Jlm = Ii l(l + l)f{Jlm
(m = -l, ... , +l)
(9.44)
L+f{Jlm = f{Jl,m+! (L+ = i; + iL y )
L-f{Jlm = f{Jl,m-! (L_ = L, - iL y )
Since m = l is the maximum value of m and m = -l is the minimum value of m ,

L+f{Jll =0 (9.45)
L-f{Jl ,-l = 0

These equations will be used in the next section for the derivation of the f{Jlm
eigenfunctions.

The Rigid Rotator/Dumbbell Molecule

As an application of the preceding results relevant to the eigenvalues of L2 and t.;
let us consider the problem of the energy spectrum of a rigid rotator, or, equiv-
alently, a dumbbell molecule (at sufficiently low temperature/). The rotator has
two particles each of mass M separated by a weightless rigid rod of length 2a.
The midpoint of the rotator is fixed in space (Fig. 9.7). The moment of inertia of
the rotator, about an axis of rotation through this point, is

Let the rotator be far removed from any force fields so that its energy is purely
kinetic.

L2
E=- (9.46)
21

midpoint fixed in
space

FIGURE 9.7 Rigid rotator with fixed midpoint. Moment of inertia about an axis of rota-
tion through the midpoint is L = 2M a 2.

2See footnote, p. 626.

364 Chapter 9 Angular Momentum

The quantum mechanical Hamiltonian operator is

~ 2
~ L
H=- (9.47)
21
and the time-independent Schrodin ger equation for this system appears as

(9.48)

The eigenvalues of Ii are the same as those of the square angular momentum
operator i.2 . With the results obtained we may rewrite the equation above with
the I, m indices.

Tenn
Energy m
notation

3
2
All these
xigenvalues of
F 121'1 2 /21 3 0 t., correspond
- I to the same
eigenenergy, 1211 2 / 21
-2
-3

2
1
D 6"h 2 /21 2 0
-I
-2

1
P 2"h 2/2I 0
- I

S 0 0 0

FIGURE 9.8 Term diagr am for the rigid rotator of moment of inertia, I . The Ith eigenen-
ergy,/i?I (1 + 1)/2 /, is (21 + I)-fold degenerate.
Problems 365

( ~;) ~lm = El~lm

(9.49)
fj,2 l (l + 1)
e, = --2-/--

This energy is (2l + lj-fold degenerate. For any value of l, there are (2l + 1)
eigenfunctions

~l,l , · · · , ~l ,-l = {~lm} (9.50)

all corresponding to the same eigenenergy, (9.49). The energy of the rotator does
not depend on the projection of L into the Z axis or onto any other prescribed
direction. The energy-level diagram for this system is sketched in Fig. 9.8, to-
gether with the "term notation" of levels. This notation is common to atomic
spectroscopy and will be used in the next three chapters. When a particle is
in a state of definite orbital angular momentum, characterized by the quantum
number l = 0, 1, 2, . . . , one speaks of the particle being, respectively, in an
S, P , D, F, . .. state.

PROBLEMS

9.5 Show that the frequencies of photons due to energy decays between successive levels
of a rotator with moment of inertia I are given by

9.6 An HCl molecule may rotate as well as vibrate . Discuss the difference in emission
frequencies associated with these two modes of excitation . Assume that only I --+ I ±
1 transitions between rotational states are allowed . Assume the same for vibrational
levels. For rotational levels assume I .:s
50. Spring constant and moment of inertia
may be inferred from the equivalent temperature values for HC1: IiwO / kB = 4150 K;
1i2 / 2I kB = 15.2 K.
9.7 Show that
(a) [i.x , x] = 0 (I')[Px.Lxl=O
(b) [Lx.Y]=iliz (g) [Px , Lyl = ilipz
(c) [L y , z] = i fiX (h) [Py, Lzl = ilipx
(d) [Lz.x] = iliy (i) [Pz , i., 1= i lipy
(e) [Ly. z] = [y, Lzl (j) [Ly, pzl = [Py, Lzl
9.8 Calculate
(a) Lzkr
(b) i., sin kr
(c) Lzf(kr)
366 Chapter 9 Angular Momentum

explicitly in Cartesian coordinates, with r 2 = x 2 + y2 + z2 . The function f is an

arbitrary function of r , and k is a constant wavenumber.
9.9 (a) Prove that

e == L x f' - ilif' = i lif' - f' x L

(b) Show that this operator is Hermitian .
(c) Show that

9.10 Show that

' 2' 2
[Lx , Ly ] = [L' y 2, t.;
' 2 ' 2' 2
] = [L z .t., ]

9.11 Evaluate
(a) [£.2, ill (c) [L, fi2]
(b) [L, ill (d) [L, Lx L]
Note that parts (b) and (d) have nine components . They are called dyadic operators.

9.12 Show that

9.13 Show that the expression

is implied directly by the two assumptions :

(a) The only possible values that the components of angular momentum can have on
any axis are Ii( - i . ... , +j ).
(b) All these components are equally probable.

Answer
Because all axes are equivalent,

(J 2 ) = u, 2 + J y 2 + t, 2) = u, 2) + ( J y 2 ) + (i,2 ) = 3( Jx 2 )

Since all values of Jx 2 are equally probable,

2li2L~=! m 2
2j +1
Substituting the relation

i: m 2 = j(j + 1~(2j + 1)
m=!
 9.3 Eigenfunctions of the Orb ital Angular Momentum Operators [2 and i, 367

into the above gives

9.3 • EIGENFUNCTIONS OF THE ORBITAL ANGULAR MOMENTUM

OPERATORS [2 AND L

Spherical Harmonics
There are two techniques for obtaining the common eigenfunctions ({JIm of the
orbital angular momentum operators L2 and i.; First, one may directly solve the
eigenvalue equations

U({Jlm = /i21(1 + 1)({Jlm

(9.51)
L z({Jlm = 1im({Jlm

Second, one may seek solution to the equation

(9.52)

Once having found ({Jll, the remaining eigenfunctions of L 2 and i.; corresponding
to the orbital quantum number 1,

{({JIm} = «({Jll, ({Jl,l-l , .. . , ({JI , -I) (9.53)

are obtained by applying L_ to ({Jll. That is,

({JI ,I-l = L-({Jll

(9.54)
({JI,I-2 = L_({JI ,I-l

In either technique for obtaining the eigenfunctions ({JIm , it proves both conve-
e,
nient and practical to work in spherical coordinates (r , ¢) (see Fig. 1.6). These
coordinates are related to the Cartesian coordinates (x , y , z) through the transfor-
mation equations

x = r sine cos¢
y = r sine sin e (9.55)

z = rcose
With these equations, the Cartesian components of t , (9.3) , are transformed to
(see Problem 9.14)
368 Chapter 9 Angular Momentum

Lx =
A
in (a a)
sin¢- +cotecos¢-
ae a¢

Ly =
A
in ( a +cotesin¢-a)
-cos¢- (9.56)
eo a¢
~

L =
a
-in-
z a¢

Using expressions (9.56) we obtain first the ladder operators

A A

L+=Lx+iL y=!ie'
A 'q, (
icote
a + aea )
a¢
(9.57)
L_ = L - iL = !ie-loy
A A A ' A. ( a --
icote- a)
x y a¢ ae

and second, the operator L2

2]
LA2 =-n2 [ - 1- - a ( sme-
sin e ae
. 1 a
a ) +----
ae sin2 e a¢2
(9.58)

We are now prepared to seek solutions to (9.51). This is the first technique , as
mentioned above, for finding the eigenstates ({JIm. These solutions are quite com-
mon to many branches of physics. They are called spherical harmonics and are
universally denoted by the symbol Yl m • Following this protocol we change nota-
tion: ({JIm ---* Yl m.

Angular Momentum and Rotation

Before discussing these solutions we note two points. First, all the angular mo-
mentum operators, when expressed in spherical coordinates as listed above, are
independent of r . They are functions only of the angular variables (e, ¢) . This
means that the eigenfunctions of L2 and L, may be chosen independent of r , that
is, Ylm = Ylm(e, ¢) . This property stems from the fact that angular momentum
operators are related to rotation . For instance , the operator
A L
Roq, = 1 +i84> 'Ii (9.59)

(described previously in Section 6.3) when acting on !(r) rotates r through the
azimuthal displacement 84>, so that

Roq,!(r) = !(r + 8r)

(9.60)
8r = 84> x r

So the effect of the operation 84> ' L on a function r is to cause a rotational

displacement of r. If 84> is parallel to the Z axis, 84> . L = 8¢L z . This operator
9.3 Eigenfunctions of the Orbital Angular Momentum Operators [2 and i, 369

induces a rotation of r about the z axis, without changing the magnitude of r. If we

write fer) = fer, (), ¢), then L, when operating on f affects only the variable ¢.
When i.2 operates on this function, () and ¢ are both affected, but not r. So here
we have the reason that the eigenstates of i.2 and i, may be chosen independent
ofr.

Normalization
The second point we wish to note relates to the normalization of the Yi'" functions.
This normalization is taken over the surface of a unit sphere. The differential ele-
ment of area dS, on the surface of a sphere of radius a, is conveniently expressed
in terms of the element of solid angle dQ.

(9.61)

(see Fig. 9.9). The solid angle subtended by dS about the origin is dSja 2 = dQ.
The solid angle subtended by a sphere (more generally any closed surface) about

dS=a 2 sinO dO d</)=a 2 dn

adO

Jdn =411' steradians

all directions

a sin () d¢

(a)

f dn = 211' steradians

(b)

FIGURE 9.9 (a) Element of the solid angle dQ = d S/ a 2 . (b) Solid angle subtended by
the hemisphere about the origin 0 is 2n .
370 Chapter 9 Angular Momentum

the origin is

2Jr d¢ l Jr 8in8 d8
1 all directions
dQ =
10 0

= {2Jr d¢jl dcos8 (9.62)

10 -I

= 4n steradians

which is the same as the area of a unit sphere (a sphere of unit radius) .
To normalize the eigenfunctions y/m, set

(9.63)

which we see is the same as requiring that 1y/m 12 integrate to unity over the surface
of a unit sphere .
We are now prepared to discuss the solutions to (9.51). The eigenfunction
equation for i ; gives

(9.64)

This equation determines only the ¢ dependence of y/m. If we set

(9.65)

the equation above gives

1 . A.
cfJ m(¢) = __ e,m", (9.66)
v'2i
which satisfies the normalization

(9.67)

The index m can be determined from the single valuedness' of the wavefunction
cfJ . That is

cfJ(¢) = cfJ(¢ + 2n)

eimtP = e im(tP+2Jr) (9.68)
i m2Jr
e = 1
30n physical grounds it is more appropriate to require that 1<1>1 2 be single-valued. However, this can
be shown to be equivalent to the single valuedness of <I> for the case in point. For further discussion,
see K. Gottfried , Quantum Mechanics, W. A. Benjamin, New York, 1966.
9.3 Eigenfunctions of the Orb ital Angular Momentum Operators f.2 and Lz 371

which is only satisfied for integral values of m:

m=±0,1 ,2, .. . (9.69)

As demonstrated in Section 9.2 the values of m run from -1 to +1, whence 1is also
an integer. Thus we obtain the result stated previously that the 1, m orbital angular
momentum quantum numbers are integers only. We also see how this property
follows directly from boundary conditions imposed on the wavefunctions Ylm •
Spin , being an intrin sic property of a particle, is not so constrained, and the related
quantum s numbers may assume half-odd -integral as well as integral values .

Legendre Polynomials
We have found that the eigenfunction Ylm has the structure

(9.70)

Substituting this function into (9.51) , together with the explicit expression for f2
as given by (9.58), gives the following equation for elm (deleting 1 and m indices,
for the moment):

2
- . 1- -d ( sin e-
de) + [ 1(1 + 1) - - m- ] e= 0 (9.71)
sin e de de sin 2 e

or equivalently, in terms of the variable,

JL == cos e
2
- d [ (1- JL)-
dJL
2 de]
dJL
+ [ 1(1 + 1) - - m- ]
1 - JL2
e= 0 (9.72)

-1~JL~+1

Let us outline the method by which this equation is solved ." Setting m = 0
and 1(1 + 1) = A in (9.72) gives Legendre 's equation,

- d [ (1 - JL 2 ) -del] + Ael = 0 (9.73)

dJL dJL

where we have set elo == el. Referring to (9.58) , we see that (9.73) is an eigen -
value equation for f2 /1i2 (corresponding to L; = 0), with eigenvalue A. A solu-
tion to (9.73) may be sought as a series'' in powers of JL . The requirement that this
4For a more detailed description of this method of solution , see E. Merzbacher, Quantum Mechanics ,
2d ed., Wiley, New York, 1970. A closely related but more conci se technique of solution is described
in P. Stehle, Quantum Mechanics, Holden-Day, San Francisco, 1966.
5This method of series solution is explicitly demonstrated in Chapter lOin the generation of Laguerre
polynomials, which are components of the wavefunctions for the hydrogen atom .
372 Chapter 9 Angular Momentum

series solution remain bounded in the interval -1 :s JL :s +1 implies that (l) the
eigenvalue A must be of the form 1(1 + 1), where I 2: 0 and is an integer, and
(2) the series solution for 8l contains, at most, a finite number of terms. The first
conclusion returns the form of the eigenvalues of L 2 given previously by (9.14),
namely, L 2 = !i21(l + I). The second conclusion indicates that 8l is a polynomial
of order I. These polynomials, called Legendre polynomials, are commonly de-
noted as Pl(JL), so that apart from a multiplicative constant, 8l(JL) = Pi (JL) . The
series summation for this solution may be expressed in the concise form, called
the formula ofRodrigues.

(9.74)

With this solution to (9.73) at hand, the solution to (9.72) is obtained by first
constructing the associated Legendrefunctions . These are defined by" the follow-
ing differential operation on Pl(JL):
dm p ( )
Plm(JL) = (_l)m(l- JL2)m /2 1 JL (9.75)
dJLm

for positive integers m :s l. Differentiating Legendre's equation (9.73) m times

with A = 1(1 + I), and 8l set equal to Pi. and employing the definition (9.75) , one
readily deduces the equation

(9.76)

Comparison with (9.72) indicates that Pi" (JL) is a solution to this same equation.
Furthermore, (9.72) remains unchanged if m is replaced by -m , and we may
conclude that p l - m (JL) is also a solution to this equation, so that apart from a
multiplicative constant, Pi" is equal to Pl- m .
In summary, we have found that the solutions 8l m(JL) to (9.72) are given by
the associated Legendre functions Pi" (JL) . In addition, we see from the foregoing
construction how the quantum conditions (9.14) emerge from the requirements
that Ylm (8, if» remain nonsingular and single-valued in the intervals -I :s JL :s
+1,0 :s if> :s 2n .
The precise relation between 8l m(JL) and Pi" (JL) as defined by (9.75) follows
from the normalization condition (9.63) .

6Another popular notation for these polynomials includes the (_l)m factor explicitly in the Ylm func-
tions.
9.3 Eigenfunctions of the Orbital Angular Momentum Operators F and Lz 373

TABLE 9.1 The first few normal ized spherical harmonics and corresponding associated
Legendre functions"

Yr(O , t/» = [21 + I (11-

4rr (+m) .
m)., ]
1/2 pr(cosO)eimq, y- I _ I
I -211!
J+ (21 I)! . 1 -ilq,
~sm 0 e

1 1 dcosO [21C dt/>yr(Yl'm ')* =Omm'0ll'

-I Jo
YP = fFlI I+ I
- - PI (cos 0)
4JT

21+ I
Po = I
pi = - sinO LI l
yr (0 , t/» 12 = ~
m=-I
pp = cosO y - m = (_l)m(Yr)*
l
- I I . II
I )1 /2
P1 = 2 sm u yO _ _
o- ( 4JT
pi = 3sin20
pi = -3sinOcosO y l = --
1
I ( 3 )1 /2
-
2 2rr
.
sinOe'q,
I
P~ = 2(3cos 20 - I) yO = _1(3)1/2
_ cos e
I 2 tt
Pi = ~ sinOcosO
l

y- I = -I ( - 3 )1 /2 sin Oe-'q,
.
Pi2= 8 sin20
I 1 2 2JT
I ( _15) 1/2 sin 2 0 e2,q,
.
pi = -15 sin30 y2 = _
2 4 2rr
pf = 15 sin2 Ocos O
yl = __1(15)1/2
_ .
sinO cosO e'q,
pI = -~ sin O(5 cos2 0 - I)
2 2 2JT

I yO = -I (5)1 /2 (3 cos 2 0 -I)

pf = 2(5cos 3 0 - 3co sO) 2
-
4 JT

P3- 1 = ~ sinO(5cos20-1) Yi l = -I (15

- )1 /2 sinO cosOe-'q,
.
2 2JT
P3-
2
= ~ sin20cosO 1(15)1
y-2 = _ _ /2 sin 2 0 e- 2,q,
.
2 4 2JT
p-3
3
=...!-.
48
sin30
y3 = __I (35)1
_ / 2 sin3 0 e 3,q,
.
3 8 JT
1(105)1
y2 = _ _ /2 sin20cosOe2,q,
.
3 4 2JT

yl=_-I (21)1
- /2 sinO(5cos20-l)e'q,
.
3 8 JT
I(7 )1 /2 (5cos 30 -
Yf = 4;; 3cosO)

I (21)1 /2 .
y3- 1 = 8 -;- sinO(5cos 20 - I)e-'q,

y-2 = _I (105)1
_ /2 sin20cosOe-2'q,
.
3 4 2JT

y- 3 = _I (35)1
_ /2 sin 3 0 e- 3,q,
.
3 8 JT

aDefining relations for PI(J1.) and pr(J1.) are given in Table 9.3. Comparison with other notations
for the spherical harmonics and their related functions may be found in D. Park, Introduction to the
Quantum Theory , 2d ed., McGraw-Hill, New York, 1974.
374 Chapter 9 Angular Momentum

There results

m [21+1(I-m) !]1 /2 m
81 (11-) = - 2 - (l + m)! PI (11-) (9.77)

The first few spherical harmonics , Ylm , are listed in Table 9.1. Important proper-
ties of the Legendre polynomials, PI, are listed in Table 9.2, while properties of
the associated Legendre functions, Pi'"; are listed in Table 9.3.

TABLE 9.2 Properties of the Legendre polynomials

Generating formulas
00
(1 - 2f.Ls +s2) -1 /2 = L PI (f.L)sl
1=0

PI(f.L) = 211!
1 (d l
df.LI
) 2
(f.L - 1)
I {-I s W::o 1
1= 0,1,2,3 , . . .

Legendre's Equation

Recurrence Relations

(I + 1)Pl+I(f.L) = (21 + 1)f.LPI(f.L) -IPI-I(f.L)

2 d
(1 - u. ) df.L PI(f.L) = -1f.LPI(f.L) + IPI_I (f.L)

Normalization and Orthogonality

(l = m)
(1:1 m)

The First Few Polynomials

1
Po =1 P4 = g(35f.L4 - 30f.L2 + 3)
1
Ps = g(63f.LS - 70f.L3 + 15f.L)

Special Values
9.3 Eigenfunct ions of the Orbital Angular Momentum Op erators [2 and Lz 375

TABLE 9.~ Properties of the associated Legendre functions

Definition
d m
pr(/1-) = (_l)m(l- /1-2) m/2 d/1-m PI(/1-) ; pp = PI
r;»; )_(_I)m(l-m)!pm() . p-I 1 . If)
I /1- - (l + m)! I /1- , I = 21/ ! sm
For these equations, m is taken as ~ O. In the formulas below,
however, m may be < 0 also; 1= 0,1 ,2, . .. , Iml ~ I.

Differential Equation

Recurrence Relations

(21 + 1)/1-pr(/1-) = (/- m + I)P[~1 (/1-) + (I + m)P[':_1 (/1-)

(21 + 1)(1- /1-2)1 / 2pr(/1-) = p/~-tI(/1-) - P/~tl(/1-)
dp m(/1-)
(1 - /1-2) I = (l + 1)/1-pr(/1-) - (/- m + I)P[~1 (/1-)
d/1-
= -1/1-pr(/1-) + (I +m)P[':_I(/1-)
(1 - /1-2)1 /2 pr+ I(/1-) = (/- m)/1-pr(/1-) - (I + m)P[':.-1 (/1-)

= -(I + m + l)pr(/1-) + (/- m + I)P[~1 (/1-)

Normalization and Orthogonality

I pm pm d _ _2_ (I +m) !
J_I I (/1-) k (/1-) /1- - 21 + 1 (l- m)! (I = k)

=0 (1:1 k)

Polar Plots of yJm and Spherical Harmonic Expansions

When a system such as a rigid rotator is in an eigenstate of L2 and i.; the z axis
is said to be preferred. Namely, measurement of L ; is certain to find a specific
value. However, in this state, it is still true that the x direction is in no way pre-
ferred over the y direction. Thus the probability density, 1Ylm 12 , is rotationally
symmetric about the z axis or, equivalently (from 9.70), IYlml is independent of
¢ . The function 1Ylm 1 is a surface of revolution about the z axis .

I [21+1 (l-m)!]1 /2
IYlml = j;L18Im(cose)! = - - IPlm(cose)1 (9.78)
v2rr 4rr (l+m)!
376 Chapter 9 Angular Momentum

Polar plots of these functions for I = 0, l , 2, and all accompanying m values, in

any plane through the Z axis, are sketched in Fig. 9.10. These diagrams give the
probability amplitude of rotational states for, say, a dumbbell molecule . Thus, in
the diagram for IYIII, one notes that it is most probable that the molecule rotates
in a plane normal to the z axis, corresponding to maximum projection of L onto
the z axis.
The functions y/m(e, ¢) are a basis of the Hilbert space of square-integrable
functions <pee , ¢) defined on the unit sphere. Such functions may be normalized
as follows.

1 11
21T
1I<p(e, ¢)1I 2 = (<pl<p) = d¢ d cos e <p*<p = I (9.79)
o -I

The expansion of <p in spherical harmonics is given by

00

<pee, ¢) = LL a/my/m(e, ¢) (9.80)

/=Olml :o/

z z z z z

z z

I Yi I

FIGURE 9.10 Polar plots of Iy/m I versus (J in any plane through the z axis for I = 0, I , 2.
The equality Iy/m I = Iy/-m I is exhibited .
9.3 Eigenfunctions of the Orbital AngularMomentum Operators f.2 and i, 377

The coefficient expansion aim is given by the inner product,

(9.81)

Suppose that at a given instant a system (e.g., a rigid rotator) is in the state cp(f}, ¢) .
Then the probability that measurement of L 2 finds the value /i21 (l + 1) is

+1
P[nh(l + 1)] = L lalm 1
2
(9.82)
m=-I

while the probability of finding L; with the value lim is

00

P(/im) = L lalml
2
(9.83)
1=lml

Domain of (I,m) values

For example, consider that a rotator is in the state

What values of L 2 and L; will measurement find? To answer this question, in prin-
ciple we should first evaluate the coefficients aim given by the operation (9.81).
However, for the case at hand, reference to Table 9.1 reveals that cp is the simple
superposition

where A and A' are constants . So the only coefficients that enter the expansion
(9.80) are a22 and a2-2 . We may conclude that measurement will find the value
of L 2 = 6/i2 with probability 1 and the values L , = ±2/i with equal probabilities
4.
of No other values of L; and L 2 would be found for a rotator in the state given
above.

Second Construction of the Spherical Harmonics

Let us now tum to the second procedure for finding the Ylm eigenfunctions, ini-
tiated by (9.52). Consider that we have already solved for the eigenfunction of
i.; so that CPlm is known to be in the form given by (9.70). Equation (9.52) then
378 Chapter 9 Angular Momentum

becomes

(9.84)

Bringing the exp(iI</» factor through the differential operator gives (deleting the
I-scripts for the moment)

o
- 8 = IcotO 8 (9.85)
00
Substituting the relation

o .I
I cot 0 = 00 In sin 0 (9.86)

and then dividing through by 8 gives

I 0 0 O .
- - 8 = -ln8 = -lnsmlO (9.87)
800 00 00
This is simply integrated to yield

8/ = All sin' 0 (9.88)

where All is a normalization constant. It follows that fl l is

(9.89)

which agrees with the values given in Table 9.1. The eigenfunction fl l- I is ob-
tained from fl l through the operator L_.

(9.90)

In this manner we obtain

(9.91)

which is also in agreement with the values given in Table 9.1. The relations be-
tween L+,L_, and the fl m functions with correct normalization factors are given
in Table 9.4.
We conclude this section with the following example. Suppose that a rigid
rotator is in the eigenstate of L2 and i , corresponding to I = 1 and m = I (i.e.,
fl l). What is the probability that measurement of L, finds the respective values
m = 0, ±l ? To answer this question we must expand fl l in the eigenfunctions of
i.; These eigenfunctions are solutions to the equation
LxX(O, </» = liaX(O , </» (9.92)
9.3 Eigenfunctions of the Orbital Angular Momentum Operators [2 and t, 379

TABLE 9.4 Normalized relations between L+ , L_, Lx, i., and the states Ilm)a
= mhllm)
L zllm)
L+llm) = h[(l - m)(l + m + 1)]1 /211, m + 1)
L_llm) = h[(l + m)(l - m + 1)]1 /211, m - 1)

Lxllm) = ~h[(l- m)(l + m + 1)]1 /2 11, m + 1) + ~h[(l + m)(l- m + 1)]1 /2 11, m - 1)

Lyllm) = -~ih[(l - m)(l + m + 1)]1 /2 11, m + 1) + ~ih[(l + m)(l - m + 1)]1 /2 11, m - 1)

L±llm) = h[(l(l + 1) - m(m ± 1)]1/211, m ± 1)

aThese normalization relations also apply to the total angular momentum operators, J±,
, ,A '2
Jx, i;
Jz, and J ,where

J2ljmj) = h 2 j(j + 1)ljmj)

Jzljmj) = hmjlimj)

The student may question why these functions are not simply the spherical har-
monics Ylm. After all, there is no intrinsic difference between i , and i.; The
answer is that the eigenfunctions of i., are the Ylm functions if we define the
x axis as the polar axis, so that () is angular displacement from the x axis. How-
ever, for the problem at hand the z axis is the polar axis and the X functions are a
bit more complicated.
Writing i, as

(9.93)

it is clear that LxYl m gives a combination of spherical harmonics with the same
I value. Also, since all Ylm functions with 1m I s I are eigenfunctions of f2 with
eigenvalue li 21(l + 1), any combination of such functions is an eigenfunction of
f2 with eigenvalue li 21(l + 1).
With these properties in mind we seek a solution to (9.92) in the form

(9.94)

The problem of finding the eigenfunctions of i, (corresponding to I = 1) is then

reduced to finding the coefficients a, b, and c in the expression above .
From the properties of L+ and L_ listed in Table 9.4, we have
, 0 r: I
L+YI = v2liYI
, I r: 0
L+YI- = v2liYI
(9.95)
L_Yl o = ./2liYI- I
, I r: 0
L_YI =v2liYI
380 Chapter 9 Angular Momentum

Substituting the expansion (9.94) into the eigenvalue equation (9.92) and using
the relations above gives the equation

(9.96)

Since the Yim functions form a linearly independent sequence, it follows that the
only way to guarantee equality for all values of () and ¢ in the equation above
is to set the coefficients of individual Yim functions equal to zero. This gives the
following set of three homogeneous algebraic equations:

(9.97)

A nontrivial solution to these equations occurs only for values of a that make
the determinant of the coefficient matrix vanish. Setting the determinant equal to
zero, one obtains

a(a 2 - 1) = 0 (9.98)

which gives the eigenvalues

a =0, a = 1, a =-1 (9.99)

Substituting these values back into (9.97) gives the (normalized) eigenvectors

a = +1 (9.100)

a =-1

With these eigenfunctions of i; at hand it becomes a matter of inspection to

construct the linear combination that gives YI I . It is given by

(9.101)

It follows that if the rotator is in the eigenstate of £2 and i , corresponding to

I = 1, m = 1, then the probability that measurement of L, finds the value +n
is i, n i,
the probability of finding - is and the probability of finding 0 is ~
(Fig. 9.11) .
Problems 381

L =fiti

FIGURE 9.11 Given the state L 2 == 2n 2 , L z == ti, what is the probability that measure-
ment of Lx finds the values ±n,O?Geometrical construction shows that two projections of
L give L., == 0, while one projection gives Lx == +n and one projection gives Lx == -no

PROBLEMS

9.14 Use transformation equations (9.55) to obtain the expression

"
Lz == -in-
a
a¢

Answer
From (9.55) we obtain the following useful relations.

z y
r 2 == x 2 + y2 + z2, cose == -, tan e == -
r x
ae cos ¢ cos e a¢ y 2
- == --cos ¢
ax r ax x2
ae sin e cos e a¢ cos 2 ¢
-
ay r ay x
ae sine a¢ == 0
-
az r 3z
ar x ar y ar z
- -
ax r ay r az r

For example, from cos e == z/ r, one obtains

. ae a 1 zx (z/r)(x/r) cos e sin e cos ¢
- slne- == z- - == - -3 ==
ax ax r r r r

Substituting these expressions in the expansion

»;
Lz
(a
== -in x ay - y ax a)
382 Chapter 9 Angular Momentum

gives the desired result.

9.15 (a) What is [~, L zl?

(b) Calculate the root-mean-square deviation !:!ilfJ for a particle in the uniform state
rp = 1l ./fir. (Hint: Perform your integrals over the interval -n, rr.)
(c) Write down an uncertainty relation seemingly implied by your answer to part (a)
and argue the physical inconsistency of this relation in view of your answer to
part (b).

Answers
(a) [~ , Lzl = iii
(b) !:!ilfJlmax = n l../3
(c) One is tempted to write!:!ilfJ !:!iLz ::: li12; however, by virtue of the result in
part (b), uncertainty in lfJ greater than tt 1../3 has little physical meaning. In
the extreme that the system is in an eigenstate (e.g., Ylm) of L z, !:!iLz = 0
and the uncertainty relation gives !:!ilfJ = 00. Thus we may conclude that the
assumed uncertainty relation is erroneous . [Note: Consider the space of func-
tions SJ¢ whose elements have finite norm on the finite interval (0,2n) (i.e.,
J~1r rp*rp dlfJ < 00) . It has been pointed out by D. judge? that L, is not Hermi-
tian on this space. As a consequence, the derivation of the uncertainty relation
between lfJ and L, from their commutator relation fails. The non-Hermiticity of
L, on SJ¢ may be seen as follows. It is evident that the Herrniticity condition
(L zrp l lf{!2 ) = (rp I IL zrp2) is valid only on the subspace SJ¢' C SJ¢ whose ele-
ments are periodic: rp(O) = rp(2n). Hence i , is non-Hermitian on SJ¢ . Specifi-
cally, note that even though rp(lfJ) is periodic, the product lfJrp(lfJ) is not periodic
and one may not invoke Herrniticity of i , with respect to functions of this type.
This is the crux of the breakdown in the proof of the uncertainty relation. See
Problem 5.42.]

9.16 In regard to inconsistencies presented by the azimuthal angle lfJ, as discussed in Prob-
lem 9.15, it has been pointed out by W. Louisell 8 that more consistent angle variables
are sin lfJ and cos lfJ.
(a) Show that

[sin lfJ, L zl = i Ii cos lfJ

[cos e, Lzl = -ilisinlfJ

(b) Use these commutator formulas to obtain uncertainty relations between sin lfJ, L z
and coslfJ, L z·

7D. Judge, Nuovo Cimento 31, 332 (1964). For further discussion and reference, see P. Carruthers and
N. Nieto, Rev. Mod . Phys. 40,411 (1968) .
8w. Louisell, Phys. Lett. 7, 60 (1963) .
Problems 383

Answer (partial)
. h (cos eM
si.; /:),. SIncfi ~ --2-
(b)
h (sin cfi)
st., /:),.coscfi ~ --2-

9.17 (a) Show that the operator

when acting on the function f(cfi) changes f by a rotation of coordinates about

the z axis so that the radius through cfi is rotated to the radius through cfi + /:),.cfi .
That is, show that

(b) Show that the operator

RI:>¢ = exp (i/:),.c/>h .L)

when acting on f(r) changes f by rotating r to a new value on the surface of
the sphere of radius r , but rotated away from r through the azimuth /:),.cfi , so that
r(O, cfi) ~ r' = r(O, cfi+/:),.cfi) . Forinfinitesimal displacements oc/> , we may write

Rot/,!(r) = f(r + or)

or = oc/> x r

See Fig. 9.12.

lir = lirll x r

"
R6~(r) = ftr + lir)

"
R " "
6 • =J + i li~' L/1l

FIGURE 9.12 The rotation operator Rot/> changes f(r) by rotating r through the az-
imuthal increment oc/> . (See Problem 9.17.)
384 Chapter 9 Angular Momentum

Answers
(a) Rt>¢f = [exp ( ~¢> a:)] f

af (~¢»2 a2 f
= f(¢» + ~¢> a¢> + - 2 - a¢>2 + ... = f(¢> + ~¢»
(b) Let o¢> be an infinitesimal angle so that ~¢> = no¢> in the limit that n » 1. For
the infinitesimal rotation

r' = r + or = r + ocjJ xr

so that

f(r + or) = f(r) + ocjJ x r· V f(r)

= f(r) + ocjJ • r x V f(r)
i ,
= f(r) + h0cjJ • r x pf(r)
i '
= f(r) + h0cjJ • Lf(r)

In the Taylor series expansion of f(r + or) above we have only kept terms of
O(O¢» . [The expression or = ocjJ x r is valid only to terms of O(O¢» .] In this
manner we obtain

f(r + or) = (i + ~ocjJ i)• f(r) = Ro</>f(r)

For a finite rotational displacement through the angle AcjJ = nocjJ, we apply
the operator Ro</>, n times :

and pass to the limit n -+ 00 or, equivalently, ~¢> / o¢> -+ 00.

,)t>¢lo¢ .... Lin

,
Rt>¢ = lim
( , i
1+ -ocjJ • L = e'.t> 0/'
t>¢lo¢-H>o h.

(Note: The operator Ro</> rotates r to r + ocjJ x r with respect to a fixed coor-
dinate frame . If, on the other hand, the coordinate frame is rotated through ocjJ
with r fixed in space, then in the new coordinate frame this vector has the value
r-ocjJ x r . Thus, rotation of coordinates through ocjJ is generated by the operator
L o</> ')
9.18 Show that f2 may be written as
2 2)
L' 2 = -Ii2 ( -a 2 +cote-
a + - I- -a-
ae so 2
sin o a¢>2
Problems 385

9.19 Show by direct operation that

Uyl = 6h2 Yl
2 2
= 2 h Y2
A

L ZY2

9.20 First calculate P2(P,) using the generating function (1 - 2p,s + s2)-1 /2 . Then obtain
P2 1(p,) using the relation between PI and PI m given in Table 9.3 . Having found P2 1,
form e2 1 and then Y2 1. Check your answers with the values given in Table 9.1.
9.21 Using the explicit form of Ylm , show that

m 1m'
9.22 Operate on YI I- I with L_ to obtain the angular dependent factor of Yj-2.
9.23 Assume that a particle has an orbital angular momentum with z component hm and
square magnitude h21(1 + I) .
(a) Show that in this state

(L x ) = (L y ) = 0
(b) Show that

[Hints: For part (a), use L+ and L_ . For part (b), use U = £X2 + Li + Lz2.1

9.24 The same conditions hold as in Problem 9.23. What is the expectation of the operator
!(LxLy + LyL x) in the Ylm state?
9.25 A D2 molecule at 30 K, at t = 0, is known to be in the state

(a) What values of Land L z will measurement find and with what probabilities will
these values occur?
(b) What is 1/r(O, q" t) ?
(c) What is (E) for the molecule (in eV) at t > O?
(Note: For the purely rotational states of D2, assume that h/4rr Ie = 30.4 em-I .)
9.26 At a given instant of time, a rigid rotator is in the state

<p(O ,q,) = (3 sin e sin e

'1"4;
(a) What possible values of L z will measurement find and with what probability will
these values occur?
(b) What is (Lx) for this state?
(c) What is (L2) for this state?
386 Chapter 9 Angular Momentum

FIGURE 9.13 Configuration relevant to Problem 9.28.

9.27 Suppose that a rotator is in the state Y1 -1 . What values will measurement of i., find
and with what probability will these values occur? (Hint: Most of the analysis in the
text [(9.92) et seq.] involving the expansion of the state Y1 1 may be used here.)
9.28 A one-particle system is in the angular state Yi'", Measurement is made of the com-
ponent of L along the z' axis. The z' axis makes an angle A with the z axis. What
is the expectation of this component? What is the expectation of the square of this
component? (See Fig. 9.13.)

Answer
For the first problem we must calculate (ez! • L), where ez! is the unit vector in the
direction of the z' axis. For the second problem, we must calculate ((ez! • L)2). The
components of e z! are

ez! = (sin A cos 13, sin A sin 13, cos A)

where 13 is the azimuthal coordinate of ez! with respect to the original axes.

(ez! • L) = sin Acos 13 (Lx) + sin A sin 13 (L y ) + cos A(L z) = hm cos A

((ez! • L)2) = sin 2 A(L x 2 ) + cos 2 A(L z 2 )

9.29 With Gz(JL) replaced by Pz(JL) in (9.73), show that the single differentiation of this
equation gives (9.72) with G(JL) = Pz 1 (JL) and m = 1.

9.4 • ADDITION OF ANGULAR MOMENTUM

Two Electrons
In this section we examine the relation between the angular momentum of a total
system and that of its constituents. This problem is of practical importance in
atomic and nuclear physics where one encounters systems of many particles (e.g.,
9.4 Addition of Angular Momentum 387

FIGURE 9.14 Classical addition of angular momentum. The two angular momentum
vectors LI and L2 add to give the resultant L.

electrons, neutrons, protons, etc.). In many cases one is chiefly concerned with
the resultant angular momentum of the atom or nucleus.
Consider two systems that are rotating about a common origin. They could be
two rotators or two electrons in an atom (Fig. 9.14). We will speak in terms of
an atom. If the angular momentum (neglecting spin) of the first electron is LI
and that of the second electron is L2, the magnitude and z component of the total
angular momentum of the composite system of the two electrons is
~ 2 ~ ~ 2 ~ 2 ~ 2 ~ ~

L = (LI + L2) = LI + L2 + 2LI • L2

(9.102)
i; = Li z + L 2z
Suppose that the total system is in a state with definite values of Li z, L2z (e.g.,
1m1m2}) . How much further may this state be resolved? Since there are only two
good quantum numbers associated with each electron (i.e., mill and m2/2) , one
suspects that the composite system will have no more than four good quantum
numbers . As it turns out, the eigenstate 1m1m2} may further be resolved to the
state 1/1/2mlm2} . This state cannot be further resolved. For instance, one might
wish to measure L 2 • If the atom is in the state I/lhmlm2} before measurement,
we are not assured that it will be in that state after measurement of L 2 • That this
is so follows from the fact that f2 does not commute with, say, Liz '
~ ~ 2 ~ ~ 2 ~ 2 ~ ~

[Li z, L ] = [Li z, LI + L2 + 2LI • L2] (9.103)

= 2[L lz, LI . L2] = 2in(L lyL 2x - L x L2Y)

In order to establish that the set of eigenvalues (/1 ,/2, m I , m2) are good
quantum numbers (i.e., that these values may be simultaneously specified in an
eigenstate I/lhmlm2}), one must show that the set of four operators (Li z, L 2z ,
388 Chapter 9 Angular Momentum

i 12, i 22) are a set of mutually commuting operators . The fact that no other com-
muting operators (restricting the discussion to the angular momentum properties
of the system) can be attached to this set indicates that (ilz, i 2z, i 12, i 22) is a
complete set of commuting operators.
We wish to show that

(9.104)

The fact that [i I 2, il z] = 0 was shown in Section 9.1. The commutators

[il z, i 2z] vanish because the coordinates of system 1 are independent of the
coordinates of system 2, so that, for example,

All other terms in (9.104) vanish for similar reasons.

Suppose that we measure L 2 and L, and establish the state 11m} . Can this state
be further resolved? Yes. One may subsequently measure L 12 and L2 2 and not
destroy the eigenvalues of L 2 and L z already established. After measurement, the
system is left in the state IlmII/2}. To show that I, m, II, and h are good quantum
numbers , we must establish that the set (i I 2, i 22, f2, i z) is a set of commuting
operators. The only questionable pairs are of the form [i I 2, f2] and [i I 2, izl.
Expanding these, we obtain

[i I 2, i l 2 + i 22 + 2LI • L2] = 2[i I 2, LI . L2] = 2[i I 2, LIl ' L2 = 0

A2 A A A A A2 A
[LI , L z] = [LI 2, L Iz + L2z] = [LI ,Llzl = 0
(9.105)

Coupled and Uncoupled Representations

Thus we find, in quantum mechanics , that the angular momentum states for a
composite system consisting of two subsystems are characterized by either of two
sets of good quantum numbers. These correspond, respectively, to the eigenstates
1/1/2mlm2} and IlmII/2}. The latter states pertain to problems where the total an-
gular momentum of the composite system is important. We will call this repre-
sentation where L 2 and L z (together with LI 2 and Ll) are specified the coupled
representation. The representation where the Z component and magnitude of an-
gular momentum are specified for all subcomponents (i.e., L1 2, LI z, L2 2, L2z)
will be called the uncoupled representation (Fig. 9.15).
The eigenstates in either representation are constructed from products of the
eigenstates Iml/l} and Im2h} . In the uncoupled representation the simultaneous
eigenstates of (i I 2, il z, i 22, i 2z) are given by the products

(9.106)
9.4 Addition of Angular Momentum 389

tI
I

(a)

FIGURE 9.15 (a) In the uncoupled representation LI and L2 move independently. Good
quantum numbers are II, m1,/2, m2. (b) In the coupled representation, LI and L2 couple to
give L , which then exhibits discrete orientations along any prescribed axis. In this vector-
model sketch of the state I/mII/2}, the z components of LI and L2 are not conserved. This
corresponds to the fact that most generally, I/m11/2 } is a superposition state involving all
ml , m2 values with fixed ml + m2 = m.

or, equivalently,

The spherical coordinates of electron I are 01 , ¢I while 02 , ¢2 are the coordinates

of electron 2 (Fig. 9.16). For given values of II and 12 there are (2/1 + 1)(2/2 + 1)
linearly independent eigenstates of the composite system of the form (9.106) and
each with specified values of (II, 12, m I, m2) .
Eigenstates I/m11/2} of the coupled representation are simultaneous eigenstates
of the commuting operators

2+L22+ 2LI
£.2=LI ·L2

i, = LI z + L2z (9.107)
2
LI , L l

Any such state may be written as a superposition of the eigenstates of the uncou-
pled representation (9.106) . In both representations II and iz are good quantum
390 Chapter 9 Angular Momentum

/'
/
/

x
\
\
\
\

FIGURE 9.16 Angularcoordinates for particle ml and particle mz. Two important ad-
dition theorems involving the angle {3 between Om I and Omz are
(a) cos {3 = cos 01 cos Oz + sin 01 sin Oz cos((p! - <l>z)
(1- m)! .
(b) p/(cos{3) = L/ --- x p/m(cosOdP/m(cosOz)e,m(¢>I-</>z)
m=-I (I + m)!
/
= 214rr
+I x
"m
c: *
[Y/ (01. <1>1)] Y/ (Oz. <l>z)
m
m=-I

numbers. It follows that in the expansion

11m11/z) = LL 11I/zmlmz)(111zmlmzllmll/z) (9.108)

ml+m2=m
summation can only run over the quantum numbers m I and mz. The constraint
ml + mz = m stems from the middle equation (9.107) and the orthogonality of
the states 11llzmlmz) . Equation (9.108) may be rewritten

11m11/z) = LL Cmlm2111/zmlmz)
ml+m2=m (9.109)
Cm1m2 = (lllzmlmzI1m11/z)
The expansion coefficients Cm1m2 are called Clebsch-Gordan coefficients and
their significance is as follows. Let the composite system be two electrons. In the
state 11m1llz) it is known that these electrons have respective angular momentum
quantum numbers II and ii. and total angular momentum and z component quan-
9.4 Addition of Angular Momentum 391

numbers I and m. The question then arises as to what measurement of Liz and
L2z will find in the state Ilmllh). The answer to this question is that

ICmlm212 = probability that at fixed Land L z (9.110)

measurement finds one electron with Liz = min
and the other electron with L2z = m2n

As an elementary illustration of the technique employed to construct these

coefficients, let us consider expansion of the state

11m1lh) = 11, -1, 1, 1)

With ml + m2 = -1, the expansion (9.108) becomes

11 , -1, 1, 1) = Co-til, O}t 11 , -Ih + C-IOII, -I}t 11, 0h

The coefficients CO-I and C-1O are determined by normalization and propi-
tious application of the L+ and L_ operators. For the case at hand, we operate on
the last equation with

There results (recall Table 9.4)

L_II, -1,1 ,1) = °

= (LI_ + L2-)(Co-III, O}t 11, -Ih + C-IOII, -I}t 11, 0h)
= J2(CO-I + C-IO)II , -I}tII, -Ih

We may conclude that

CO-I = -C-IO
Normalization of the state 11 , -1, 1, 1) gives CO-I = 1/./2. So it is equally
probable that measurement finds (m I, m2) = (0, -1) or (-1,0).

Coordinate Representation
We have been writing 11m) for the eigenvectors of f2, i.; The coordinate repre-
sentation of these states is given by the projection

Likewise, the coordinate representation of the composite state 1/I/2m1m2) is given

by the projection
392 Chapter 9 Angular Momentum

In this manner, the expansion (9.109) gives the coordinate representation

(O,<P,(h<P21 /mll12) = LL Cm,m2(OI<PI 02<P21 /112mlm2)

m.+m2=m

Thus we see that the Clebsch-Gordan expansion affords a means of obtaining the
coordinate representation of the composite state I/mll12).
The theory of representations is discussed further in Section 11.1 and Ap-
pendix A.

Values of I for Two Electrons

Next we consider the problem of finding the allowed values of (I , m) , given
(11,/2)' This problem is directly related to "two-electron atoms," such as He and
Ca, whose energy levels are I-dependent. Suppose that one electron is a p elec-
tron (i.e., it is in a P state) and the other electron is a d electron. What values can
result for I and m (still neglecting spin)?
Let us consider the general case where the two electrons have respective I
values II and 12. Since

it follows that the maximum value m can have is

mmax = m I max + m2 max

or, equivalently,

(9.111)

It is clear that of the various values the total angular momentum quantum number
I may assume, the maximum value is equal to mmax. With (9.111) we then obtain

I max = II +12 (9.112)

In that I is an angular momentum quantum number, successive values of I differ

from Imax in unit steps down to some minimum value. What is this minimum
value? To obtain it, we note the following .
As noted previously, in the uncoupled representation there are (2/1+ 1) (2/2+ 1)
independent, common eigenstates of i 12, i 22, il z, and i 2z relevant to the two-
electron system . These states span a (2/1 + 1)(2/2 + I)-dimensional space. A
change in representation? from this basis to the common eigenstates of i 2 , i.;

9The notion of changes in representation was discussed in Section 7.4 and will be developed further
in Chapter II .
9.4 Addition of Angular Momentum 393

LIZ, and Lz z maintains the dimensionality of this space. This observation affords
a method of obtaining lmin . That is, keep decreasing I m ax in unit steps until the
total number of independent states equals (2/1 + 1)(2Iz + 1).
Now the number of independent eigenstates with a given I number is 21 + 1.
Then the value of lmin we seek satisfies the equation

ll+lz
L (21 + 1) = (2/1 + 1)(21z + 1)
l=lmjn

This relation is satisfied if we set

(9.113)

In this manner we find that the values of I corresponding to a system comprised

of two electrons with respective I values II and Iz are

(9.114)

For the problem cited above with one p electron (II = 1) and one d electron
(Iz = 2), the total angular momentum may be any of the values

1=1 ,2,3
L = liv2, L = li../6, L=liM

There are a totality of

N = (2 x 1 + 1) + (2 x 2 + 1) + (2 x 3 + 1)
= (2 x 1 + 1)(2 x 2 + 1)
= 15

states, corresponding to these three values of l. For the case of two p electrons
(II = =
Iz 1) there are nine eigenstates. These are listed in Table 9.5.
The distinction between the coupled and uncoupled representations is brought
out in the following two sets of eigenstate equations.

Coupled Representation
394 Chapter 9 Angular Momentum

TABLE 9.5 Nine common eigenstates 11m1112} of the operators f2, i.; i 2
1 , il for
two p electrons

Diagrammatic
Representation" 11m11 12} = LCmlmzlllmlhl12m2h
tt 12211} =lllhl11h

t ·+ ·t 12III} = Ify 1) II 10h + !fI1Ohl 11h

H+ ··+H 120 II} = !f111h 11, -Ih + flllOh l10h + !fll' -Ihllih

12, -Ill} = !fIlOh 11, -Ih + !fII, -Ih l10h

12, -2ll} = 11, -IhII, -Ih

t· - ·t IIIll} = !flllhllOh - !fI1Ohllih

H-H 110 II} = !flllhI I, -Ih - !fII, -Ihlllh

11 , -Ill} = !fI1OhI I, -Ih - !fII, -IhllOh

H- ··+H 10011) = !fllih 11, -Ih - !fIlO}I11Oh + !fII, -Ih IIIh

"The diagrammatic representation of states is such that an up-arrow, a down-arrow, and a dot represent,
respectively, m = I , -I, and 0 of individual electrons.

Uncoupled Representation

These equations are relevant, for example , to the case of two electrons, given that
one is an II electron and the other, an [2 electron.
For three electrons , in the uncoupled representation the six operators
A2 A A2 A A2 A
(LI ,Llz,L2 ,L2z,L3 , L 3z)

form a complete commuting set. Good quantum numbers associated with these
states are (II , m 1,12, m2, 13, m3). In the more relevant coupled representation, the
six operators
Problems 395

form a complete commuting set. The operator A12 is given by

A2 A A 2 A A 2 A A 2
Al == a12(LI + L2) + al3(LI + L3) + a23(L2 + L3) (9.115)

where the a coefficients are arbitrary. If l' is the eigen-I-value related to A12, six
good quantum numbers for the case at hand are (l , m , /1, /2, /3, I').

PROBLEMS

9.30 What are the eigenvalues of the set of operators ii. 12, Li z, L 22, L 2z) corresponding
to the product eigenstate 1m III )lm2/2)?
9.31 Let J I and J2 be the respective angular momenta of the individual components of a
two-component system. The total system has angular momentum J = J I + J2 . Show
that
AA lAA AA AA

(a) JI • J2 = "2(Jl+lz- + h-lz+) + h zlzz

(b) 12 = 112 + 12 2 + 211z12z + (11+ 12- + 11_12+ )
9.32 (a) Using the expansions developed in Problem 9.31, operate on the coupled angular
momentum eigenstates for two p electrons as listed in Table 9.5 with L2 and L;
respectively, to verify the lm entries in each of the nine I/m/1/2) eigenstates.
(b) What are the Clebsch-Gordan coefficients involved in the expansion of the state
10011)?
(c) What is the inner product (20 1110011)?
9.33 (a) With respect to the diagrammatic representation of states depicted in Table 9.5,
what are the states corresponding to the diagrams

0/0 = ",
(b) Expand each of these functions in terms of the nine diagrams listed in Table 9.5.
(c) Are any of these three states eigenstates of £2 ? [Hint: Use the expansions ob-
tained in part (b).]
(d) Two electrons are known to be in the coupled state 0/1 , What values of total
angular momentum L will measurement find and with what probabilities will
these values occur?
9.34 Two p-electrons are in the state Ilm/1/2) = 11 , -111) . If measurement is made of
Li z in this state, what values may be found and with what probability will these
values occur?
9.35 Two p electrons are in the coupled angular momentum state I/m/1/2) = 12, -2, 11).
What is the joint probability of finding the two electrons with Li z = L2z = -Ii?
9.36 How many independent eigenstates are there in the coupled representation for a two-
component system, given that II = 5 and 12 = I ? Make a table listing the ml values
for all these states.
9.37 Show that AI 2 as given by (9.115) commutes with L2 and i.;
396 Chapter 9 Angular Momentum

9.38 The eigenstate corresponding to maximum I for the three-electron case is

(a) What are the eigenvalues of P , i , correspond ing to this state?

(b) What is the eigenvalue of the operator A1Z given by (9. 115) corresponding to
this state?

9.5 • TOTAL ANGULAR MOMENTUM FOR TWO OR MORE ELECTRONS

We are now concerned with the possible values the total angular momentum
1 numbers may assume for a system of N electrons with respective 1; val-
ues 11, lz, . . . , lN, in the coupled representation. A totality of (211 + 1)(21z +
I) .. . (21N + I) product states may be formed which are simultaneous eigenstates
of the set of operators

We must make sure that our procedure for calculating these 1 values preserves this
number of states. This affords a check that we have found all 1 values.
The possible values that 1 can assume may be obtained by one of two tech-
niques . The first technique follows from the rule (9.114) for the addition of the
angular momenta of two electron s with respecti ve 1 values 11 and l z. In this case
the combined angular momentum

has eigenvalues, !i zl (1 + I), where

Consider the case of three electrons. Their total angular momentum is given by

This may be rewritten in the form

f2 = (i / + ( 3 )z
L' =L1 +Lz
Suppose that one of the 1 values corresponding to L'z is 1' . Then the 1 values
corresponding to the total angular momentum LZ are

(9. 116)
9.5 Total Angular Momentum for Two or More Electrons 397

This again follows the rule of (9.114). For example , consider the case of three p
electrons (II = 12 = 13 = 1). Then for the first two electrons we have

I' = III + hi, . . . , III -121 = 0, 1,2

Adding the third electron gives [using (9.116)] the I values

1'=0,1,2 I=Il'+I31, . . . ,Il'-I31

I' = °
-------+ I = 1
I' = 1 -------+ I = 0, 1,2
I' = 2 -------+ I = 1,2,3

Thus we obtain the result

\1=0,1,2,3 for three p electrons I

There is a distinct eigenstate for each distinct manner in which I may be formed .
This gives a total number of

(2 x 0+ 1) + 3(2 x 1 + 1) + 2(2 x 2 + 1) + (2 x 3 + 1)
= 1 + 9 + 10 + 7 = 27

states, which agrees with the product

(211 + 1)(2h + 1)(213 + 1) = 3 x 3 x 3 = 27

For the case of N electrons with respective I values 11,12, ... , IN, we follow
a similar procedure. First, we add the angular momenta of the first two electrons.
This gives

l' = III + 121 , .. . , III - hi

To these values we add the angular momentum of the third electron to obtain

I" = II' + 131, . . . , II' - 131

there is a separate sequence of I" values for each value of I' . Adding the angular
momentum of the fourth electron gives

We continue in this manner until all individual angular momentum I values are
accounted for. The final sequence gives all possible values of I. For three electrons
the sequence of I" gives all the values of l. For four electrons the sequence for 1111
gives the values of I .
398 Chapter 9 Angular Momentum

Addition Rules
The values of total I obtained by sequential addition as described above may more
simply be arrived at by the following rule. Consider N electrons with respective
angular momentum values 11,12, . . . , IN. These values may always be ordered so
that

Let
N-I
A= LI;
;=1

Then we have the following :

(a) If IN - A > 0,

Imin = IN - A (9.117)

(b) If IN - A :::: 0,

Imin = 0 (9.118)

(c) In all cases

N
Imax = LI; (9.119)
;=1

(d) The possible values of I that give the values of total L,

L 2 = (LI +L2 + . .. +LN)2 = nh(l + 1)

are given by

1= Ilmaxl , Ilmax - 11, · ··, Ilmini (9.120)

As a simple example of this technique, consider the case of two p electrons and
one f electron (II = 12 = 1,13 = 3). Then

A=I+I=2
13 - A = 3 - 2 = 1 = Imin
Imax = 1 + 1 + 3 = 5
Therefore,

1= 1,2,3,4,5 for two p electrons and

one f electron
Problems 399

The electron orbital angular momentum notation s, p, d, I, .. . stems from

atomic physics. The correspondence between these letters and I values of indi-
vidual electrons follows the scheme

Symbol s p d I 9 h
I value 0 1 2 3 4 5

This notation will be used again in Chapter 12.

In Chapter 10 we will see how i.2 enters the Hamiltonian for one- and two-
particle systems . the Ylm functions will take on further significance . They will
emerge as the angular dependent factors of the energy eigenfunctions for these
systems.
The topic of the addition of angular momentum is returned to in Chapter 11,
where the rules developed above are applied to the addition of spin angular mo-
mentum. In Chapter 12 these rules are again applied to the addition of orbital and
spin angular momentum as related to one- and two-electron atoms . In general, the
rules developed above for the addition of angular momentum are valid for orbital,
L, spin, S, and total angular momentum, J.

PROBLEMS

9.39 What are the possible values of I for

(a) Four p electrons?
(b) Three p and one f (/4 = 3) electrons?
9.40 What is the wavefunction (in Dirac notation) for three p electrons in the state with
1= m = 3?
9.41 Show that the two schemes for obtaining the total I value for three electrons with
respective I values 11,/2, and 13. as described in the text, are equivalent.
9.42 (a) Show that the technique of sequential addition for obtaining total I values in the
coupled representation gives

(2/1 + 1)(212 + I) · .. (21N + 1)

eigenstates. (Hint: Assume that 11 < 12 < .. . < IN .)
(b) How many eigenstates are there for three f electrons?
9.43 In the uncoupled representation, N electrons are described by the simultaneous
eigenstates of the 2N operators

In the coupled representation. the N + 2 commuting operators

400 Chapter 9 Angular Momentum

are relevant, and there are N + 2 good quantum numbers corresponding to these
operators . One suspects that 2N - (N + 2) = N - 2 operators may be added to this
sequence , yielding a set of 2N commuting operators .
(a) Construct such a set of N - 2 operators, {Ai 2 }.
(b) Show explicitly that the terms in the sum ..1 2 2 commute with the sequence of
N + 2 operators given above.

Answer (partial)
(a) The first operator is

The second operator is

The (N - 2)nd operator is

9.44 The spherical harmonics Yl m (8, ¢) are simultaneous eigenstates of L, and L2 . How
must the Cartesian x, y, z axes be aligned with the spherical r, 8, ¢ frame in order
for this to be true, or is the validity of this statement independent of the relative
orientation of these two frames?
9.45 Suppose that L 2 is measured for a free particle and the value 6h 2 is found . If L y is
then measured, what possible values can result?
9.46 The parity operator, JP, in three dimensions is defined by the equation JP f ir, 8, ¢) =
f(r,Jr - 8, x + ¢). Show that JPYl m = (-iYl m . That is, the parity of Ylm (odd or
even) is the same as that of 1. (Compare with Problem 6.23.)
9.47 Establish the following equalities .

x = -rf!f-(YI 1 - YI-
1),
y = -Tf!f-(YI 1 + YI- 1), z= r~YIO
2 2
r fIn 2 -2 r fIn 1 1
xy = TV 15(Y2 - Y2), yz = -TV 15(Y2 + Y2- ),

2 fIn 1 -I
ZX = -r V 15(Y2 - Y2 )

x2 _ i = r2 ~(Y22 + Y2-2), 2z 2 - x 2 - y 2 = r 2 Jl~Jr Y20 ,

i - z2 = _r 2 fIn(Y2 2 +./6 Y20 + Y2-2)

'115
Problems 401

9.48 Using the preceding relations, argue that x ir is an eigenfunction of i., and y/r
is an eigenfunction of i.; What are the respective eigenvalues of these functions?
(Note: Such functions are widely employed in quantum chemistry and are commonly
called orbitals. These topics are returned to in our discussion of hybridization in the
following chapter.)

9.49 The Clebsch-Gordan expansion (9.109) affords a coordinate representation of an-

gular momentum states in the coupled scheme . What is the explicit 0" cf>" 02, cf>2
representation of the coupled state I/ml,/2} = II, -I, I, I}, relevant to two p elec-
trons?
9.50 (a) Show that the energy eigenvalues relevant to a rigid rotator (9.49) follow from
the Bohr-Sommerfeld quantization rule (7.192) in the limit of large quantum
numbers .
(b) What conclusion may be inferred from your answer to part (a) concerning the
domain ofrelevance of the Bohr-Sommerfeld rules?

Answers
(a) We find

Thus, for I » I (but not neglecting I compared with 12 ) ,

E ~
l(l + 1)1i2
-'----....,....,...'--
21
(b) If we associate large quantum numbers with the classical domain , the preceding
example indicates that the Bohr-Sommerfeld rules are relevant to this same re-
gion. Note also that the first-order solution which enters the near-classical WKB
analysis (7.172) is the action integral, f p dx .

9.51 (a) Show that in the state 11m}

2 1--
(Lx) = (L y 2 2
) = Ii [/(1 + I) - m
2
1- 4([L+ , Ll+)

Here [, 1+ represents the anticommutator

(b) Evaluate ([L+, L-l+) in tenus of a function of I and m .

9.52 Establish the following relations relevant to a system in a state with definite angular
momentum 1i.;7(l+T5. The degeneracy factor is 9/ = 21 + I .
402 Chapter 9 Angular Momentum

(a) n-2 (L 2 ) = L Iml

m=-/
/
(b) n-4 (L 2 )2 = 2 L Iml3
m=-/
/
(c) n-2 9/ (L 2 ) = 3 L m
2

m=-/
In addition, show the following:
n-I
(d) L 9/ = n 2
/=0
/1+/2
(e) L 9/ = 9/(9/2
/=1/1-/21

9.53 A student argues that the rotational kinetic energy of a classical rigid sphere, spinning
about a fixed origin with angular frequency w, may also be associated with quantum
mechanical spin . Is the student correct? Explain your answer.

Answer
The student is incorrect. The kinetic energy of the sphere obeys the relation

wS
E=-
2
where S is the "spin" angular momentum of the sphere. When evaluating S, one in-
tegrates differential elements of orbital angular momentum over the volume of the
sphere. So, as with all classical angular momentum, classical "spin" is orbital an-
gular momentum. With the preceding relation, the kinetic energy of the sphere, E ,
is likewise associated with orbital angular momentum. (The concept of spin is fully
developed in Chapter II.)

9.54 Assuming the form

and the integral relanonl?

1£ 2m+lm'
sin 2m+ 1 ede = ...,.,-----..,.,..,-, m > 0
10o (2m + I)!!'

and the normalized expression for y/m (e, 4» given in Table 9.1, obtain an expression
for Cm '

Answer
C2 = _(2_lm.,...--;-1-_1_)'_'(_lm_I_+_m_)!
m 2lmliml! (lml - m)!

Thus, CI = I, C2 = 3, in agreement with values in Table 9.1.

101. S. Gradshteyn and I. M. Ryzhik, Tab/esof Integrals Series and Products, Academic Press, New
York, 1965, equation [3.63(5»).
Problems 403

9.55 The energy of a rigid molecule, free to rotate about its center of mass, is given by

where moments of inertia (/1 , h!J) are evaluated in principal axis l l with It = h
and the origin at the center of mass.
(a) Is the preceding expression a valid Hamilton ian form?
(b) Write down the proper Hamiltonian of the system together with eigenfunctions
and eigenenergies .
(c) What are the energy degeneracie s of this system? In what manner is your answer
to this question related to symmetrie s of the molecule ?
(d) The molecule obeys the following relation.

I
8mao2
If the molecule is in its ground state, at what energy (in eV) will an incident
photon raise the molecule to the second excited state (above the ground state)?

Answers (partial)
(a) As L x, L y , L z are not canonical momenta, the given form is not a valid Hamil-
tonian. (See Problem 1.25.)
(b) The appropriate Hamiltonian is given by (dropping hats on operators)

L2 _ L 2 L 2
H= z +_z_
21t 2/3

9.56 Show that the angular momentum commutator relations (9.8), as well as anyone of
the equations (9.13), are satisfied by corresponding Poisson-bracket relations [i.e., in
(9.8), set in = I). Poisson brackets are defined in Problem 1.15.
9.57 For a rotating system with angular momentum L , show that the classical domain is
described by the criterion, L » h:
9.58 With J1- == cos 0, show that (see also Problem 9.18)
2
L-2 =-n2 [ -(I-J1-
a
aJ1-
2)a- + 1- -a- ]
aJ1- 1 - J1-2 acjJ2

11H. Goldstein, Classical Mechanics, 2d ed., Addison-Wesley, Reading, Mass ., 1980 .

CHAPTER

10 Problems in Three Dimensions

10.1 The Free Particle in Cartesian Coordinates

10.2 The Free Particle in Spherical Coordinates
10.3 The Free-Particle Radial Wavefunction
10.4 A Charged Particle in a Magnetic Field
10.5 The Two-Particle Problem
10.6 The Hydrogen Atom
10.7 Elementary Theory ofRadiation
10.8 Thomas-Fermi Model

In this chapter we discuss the structure of the Schrodinger equation for a par-
ticle moving in three dimensions. General properties are developed through
examination of the free-particle problem in Cartesian and spherical coordinates.
Separation of variables in spherical coordinates yields product solutions for the
free-particle problem comprised of spherical harmonics and spherical Bessel
fun ctions. Solution to the corresponding radial wave equation for the hydrogen
atom gives Laguerre polynomials. Application is also directed toward the motion
ofa charged particle in a magnetic field. An elementary description of the theory
of radiation from atoms and the formulation of selection rules are given. The
chapter concludes with a description of the Thomas-Fermi model important to
atomic physics.

10.1 • THE FREE PARTICLE IN CARTESIAN COORDINATES

We again recall that the linear momentum operator p is given by

p= -iii"\' (10.1)

Inserting this form into the Hamiltonian for a free particle of mass m moving in
three dimensions gives
~2 ",2
~ P n 2
H=-=--V' (10.2)
2m 2m
404
10.1 The Free Particle in Cartesian Coordinates 405

It follows that the time-independent Schrodinger equation for this same particle
appears as

(l0.3)

or, alternatively,

v 2 cp = _k 2 cp
(l0.4)
Iih 2
E=-
2m
Separating variables

cp = X(x)Y(y)Z(z) (l0.5)

permits (l0.4) to be rewritten as

Xxx + Yyy + Z zz = _k 2
X Y Z
(l0.6)
_ Xxx = k2 + (Y yy + Zzz) = k 2
X Y Z x

In the last equation, the left-hand side is a function only of x, while the middle
term is a function only of y and z. The only way for the equality to hold for all
(x , y, z) is for both terms to be equal to the same constant. Labeling this con stant
k, 2 gives the equation

Xxx +kx 2X = 0 (l0.7)

which has a solution 1

(l0.8)

In similar manner we obtain

Y= B 'eikyy, (l0.9)

where

(l0.1O)

Combining all three factor s X, Y, and Z gives the solution

tp = A' B' C' exp[i (kxx + k y y + kzz)] = CPk = Ae i k ' r (l0.11)

I Here we consider only the forwardpropagating wave.

406 Chapter 10 Problems in Three Dimensions

The wave vector k and position vector r have components

k = (kx , k y , kz )
(10.12)
r=(x,y,z)

The function Ipk so obtained is an eigenfunction of iI (10.2) with the eigen-

value

(10.13)

Plane Waves
The corresponding solution to the time-dependent Schrodinger equation (3.52)
appears as

1/tk(r, t) = Aei(k·r-wt)
(10.14)
luo = Ek

This solution represents a propagating plane wave. At any instant of time, 1/tk (r, t)
is constant on the surfaces k . r = constant. These are surfaces normal to k.
Consider one such surface. The projection of r onto k

k ·r
rll=- (10.15)
k

from any point on this surface is constant. This is the normal displacement be-
tween the origin and the surface. See Fig. 10.1. Rewriting (10.14) in the form

1/tk(r, t) = Aeik[rn-(w/k)t) (10.16)

reveals that the rate of increase of rll with respect to a surface of constant 1/tk is
the wave speed
(J)
v= - (10.17)
k

The normalization constant A may be chosen so that

The three-dimensional delta function is defined as the product

8(r - r ') = 8(x - x')8(y - y' )8(z - z') (10.19)

10.1 The Free Particle in Cartesian Coordinates 407

x
FIGURE 10.1 At any instant of time, the plane wave

1/!k(r, t) == A exp[i(k. r - cut)]

is constant on the surface k · r == constant. These are surfaces normal to k. At every point
r, on such a surface, the projection '11 == k · r / k is constant.

and has the representation

8(r - r') == _1_

(2n)3
If! eik.(r-r')dk
(10.20)
dk == dk; dk; dk:
Comparison of this representation with (10.18) yields the normalized wavefunc-
tion

(10.21)

Superposition of Free-Particle States

A free-particle wave packet may be represented by the superposition

1jJ(r, t) = (2n\3/2 fff b(k, t)ei(kor-wt) dk (10.22)

with corresponding inverse

b(k, t) =
1
(2n )3/ 2 fff 1jJ(r, t)e-i(kor-wt) dr
(10.23)
dr == dxdydz
408 Chapter 10 Problems in Three Dimensions

lik x

FIGURE 10.2 In the plane-wave decompos ition of the free wave packet 1/!(r, t), as given
by (10.22), the Fourier amplitude b(k, t) is such that

Ib(k , t)12 dk

is the probability that measurement finds the particle with momentum in the volume ele-
ment n3 dk about the value nk.

As in the one-dimensional case, the coefficient b gives the probability

P(k) dk = Ib(k, 01 2 dk (10.24)

that measurement at the instant t finds the particle with momentum in the volume
element 1i3 dk about the value hk (Fig. 10.2).
If the probability amplitude b(k, t) is peaked about a value of k, say ko, then
the three-dimensional wave packet (10.22) propagates with the group velocity

(10.25)

where Vk is written for the gradient with respect to k. Inasmuch as (10.22) depicts
the state of a free particle, for each k-wave component one has

(10.26)

This gives w(k), which with (10.25) yields

liko
vg = - = VeL (10.27)
m
This is the classical velocity of a particle of mass m, moving with momentum
liko.
For a free particle

(10.28)
Problems 409

so that p and fj have simultaneous eigenstates . These are the functions (flk (r) . In
the eigenstate (flk, the linear momentum Iik and energy 1i2 k 2 j 2m are specified.
The state cannot be further resolved . For instance, suppose that we measure the
z component of the angular momentum of the particle L z . This measurement
destroys the information in the state before measurement, relating to the linear
momentum p. The components of p and L, in general, do not commute.
What, then, are the states for the free particle, which include specification of L 2
and L z ? To find these states it proves most convenient to express fj in spherical
coordinates. This is discussed in the next section.

PROBLEMS

10.1 If 1{t(r, r) is a free-particle state and b(k, t) the momentum probability amplitude for
this same state, show that

fff 1{t*1{tdr= fff b*bdk

10.2 At time t = 0, a free particle is in the superposition state

Jr-3/2
1{t(r, 0) = - 2 - sin 3x exp[i(5y + z)]
(a) If the energy of the particle is measured at t = 0, what value is found?
(b) What possible values of momentum (Px , Py, pz) will measurement find at t = 0,
and with what probability will these values occur ?
(c) Given the above state 1{t(r, 0), what is 1{t(r, t)?
(d) If p is measured at t =
1{t(r, t) ?
° and the value p = h(3ex + 5e y + ez) is found, what is

10.3 (a) What is the Hamilton ian for N free, noninteracting particles of mass m ?
(b) What is the eigenstate of this Hamiltonian corresponding to the eigenvalue

h2 N
E=-L:>?
2m j=l

(c) Show that the eigenstate found in part (b) is also an eigenstate of the momentum
of the center of mass . What is the velocity of the center of mass in this state?

Answers (partial)
2
_ N
(a) H - - " - V ·
2 h V·- (a
- - a - a)
- ~ 2m
J=l
J ' J - ax"
J
ayJ' ' az J.

(b) Vtk! .k 2,... .k N = A exp «f, J=l

kj . rj )

(c) VCM = Lhkj /M

410 Chapter 10 Problems in Three Dimensions

10.2 • THE FREE PARTICLE IN SPHERICAL COORDINATES

Hamiltonian
We wish to express the free-particle Hamiltonian (10.2) in spherical coordinates,
(r, e,¢) (see Fig. 1.6). We have already found the classical expression for H in
spherical coordinates in Chapter 1 [see (1.20)] . Let us again construct this clas-
sical form. However, in the present instance we wish H to include the angular
momentum term

(10.29)

The linear momentum is written p. It follows that

(10.30)

where Pr is written for the radial component of the particle 's momentum.

1
Pr = -(r·
r
p) (10.31)

If we wish to carry (10.29) over to quantum mechanics, we must make sure that
all terms in fi are Hermitian. The two operators in (10.30) are r- 2 £'2 and Pr 2. To
examine the Hermiticity of the first operator, we note the following.

Rotation and Angular Momentum

In Section 9.3 we found that the effect of the rotation operator Ro.p when operat-
ing on a function f(r) is to change f by rotating r to r + o~ x r . Suppose that
a function f is isotropic/ in r; that is, f is independent of the direction of r . It
depends only on the magnitude of r . Any function of the form f(r 2 ) is isotropic
in r. For example, f = ar 2 + br", where a and b are constants, is isotropic in r .
What is the value of f (r 2 ) on the surface of a sphere of radius ro? The answer is,
the constant f(ro 2 ) . An isotropic function is constant on the surface of any sphere
about the origin. Now suppose that we operate on an isotropic function with Ro.p .
This causes f to change by rotating r to the value r + o~ x r . The new vector
lies on the same sphere on which r lies.

Terms of O(8¢2) are neglected while the middle term vanishes because r is nor-
mal to o~ x r . It follows that the operator Ro.p has no effect on f(r 2 ) .

A 2 2
Ro.pf(r ) = f(r ) (10.32)

20ne may also say that f is spherically symmetric.

10.2 The Free Particle in Spherical Coordinates 411

Since ROt/> = (1 + i oc/> • Lin), we may conclude that

ioc/> • L 2 0
(10.33)
n f (r ) =

In that this statement is true for all axes of rotation about the origin, or equiva-
lently, for all directions of the vector oc/>, it follows that any isotropic function is
a null eigenstate of the three components of angular momentum as well as of L2 .
A 2 A 2 A 2 A2 2
Lxf(r ) = Ly/(r ) = Lzf(r ) = L f(r ) = 0 (10.34)

As noted previously in Section 9.2, these spherically symmetric states are called
S states.
If 9 (r) is any function of r (for example , 9 = x I r) and f (r 2 ) is any isotropic
function , then owing to the conclusion immediately above,

f2 f(r 2)g(r) = f(r 2)f2 g(r)

(f2 f(r 2 ) - f(r 2)f2)g(r) = [f2, f(r 2)]g(r) = 0
Since this latter equality holds for all differentiable function s g, we obtain

Similarly,
A 2
[Lx . f(r )] = [LAy , f(r 2)] = [LAz, f(r 2)] = 0 (10.35)

We are now prepared to investigate the Hermiticity of the term r- 2 L2 in the

Hamiltonian (10.30). With p-2 denoting multiplication by r- 2 , we write

(10.36)

so that p-2 L2 is Hermitian. In the last equality we used the fact that L2 commute s
with the isotropic function r- 2 .

Radial Momentum
Next we consider the operator

A
Pr=r -l( r pA) =r -l(x p;
v
A + YPy
A + zpz
A) (10.37)

Forming the Hermitian adjoint of Pr gives

A )t [_I(A A AA AA)]t
(Pr = r x Px + YPy + ZPz (10.38)
= (xpx)t(p-l)t + (Yp y)t(p-l)t + (zpz)t(p-l)t
412 Chapter 10 Problems in Three Dimensions

x
The operators 1'-1 and cannot be brought through Px . Similarly for the other
two terms . We conclude that Pr is not Hermitian.
The more appropriate operator corresponding to radial momentum is given by
the symmetric form (see Problems 10.5 and 10.6)
A I(A At)
Pr = 2 Pr + Pr (10.39)

or, equivalently,

A = -1 (1-r· pA + p.
Pr A r-1) (10.40)
2 r r

The component of p in the direction of r is given by

1 A 1 0
-r · p = -ili-r · V = -ili- (10.41)
r r or
while the second term in Pr is given by
A 1
P . r- = -i IiV • er (10.42)
r
where e, is written for the unit radius vector, r / r. Let f (r) be a differentiable
function of the radius vector r. Consider the operation

Pr f (r) = --i Ii ( -0 + V • er ) f (10.43)

2 or

= --iii (Of
- + er • Vf + fV . er )
2 or

= -iii (Of
2 or
+ of + 2f)
or r
= -iii (Of
or
+ L)
r

Equivalently, we may write

A 1 0 •

Prf = -lli- -rf

r or
(10.44)
A 1 0 •

Pr = -lli--r
r or

With the above definition of fir, we may write the following for the Hamilto-
nian operator:
A 2 f.2
iJ = !!.!.- + - -2 (10.45)
2m 2mr
Problems 413

The student may well ask the following question at this point. We know that the
Hamiltonian fI has the correct representation

A p2 22
!i V
H=-=--- (10.46)
2m 2m

How are we assured that fI, as given by (10.45), with the definition of Pr obtained
by symmetrization of fir, is equivalent to this correct form (10.46) ? This question
is answered by demonstration . The representation of the Laplacian operator V2 ,
in spherical coordinates, is

2 2
V2 =1- 0 1- -Os m
1 (- . 0e - +1 - -( - )
-r+-
r 2 or2 r 2 sin eoe oe sin eo</>2 (10.47)

Noting the equality

(~:rrr = ~ : / (~:rr) = ~ ::2 r (10.48)

and recalling the expression for f2, as given by (9.58), permits the equation

(10.49)

This is the correct form of fI, in spherical coordinates. In the next section we will
examine its eigenfunctions and eigenvalues.

PROBLEMS

10.4 What is the time-independent wavefunction in spherical coordinates of a free particle

of mass m, zero angular momentum , and energy E which satisfies the property
Irrpl = 0 at r = O? (Hint: Introduce the function u == rrp.)

10.5 (a) Show that

[r, Prl = i ii
(b) What properties of rp and 1{t insure that

(rp IPr , 1{t ) = (Pr rp !1{t )

[Note: (rp l1{t) =f dQJ drr 2 rp* 1{t .l

10.6 The current vector J associated with a wavefunction 1{t(r, t) is given by (7.107)

J= ~(1{t*V1{t - 1{tV1{t*)
2ml

The wavefunction 1{t(r, r) may be termed source-free, if V • J = 0 for all values of r .

414 Chapter 10 Problems in Three Dimensions

(a) What is the eigenfunction of Pr , corresponding to the eigenvalue lik?

(b) Calculate V • J for this eigenfunction of Pr .

Answers
(a) Integration of the eigenvalue equation

"I -rf{Jk
-i,,- lJ _
= IUlf{Jk
u t. >:
r lJr
gives

e ikr
sD=A-
r
The corresponding time-dependent solution is

_ A i(kr-wt)
f{Jk = -e
r

This "outgoing wave" is a solution to the time-dependent Schrodinger equation

for a free particle with no angular momentum . It is important in the construction
of scattering states, which will be discussed in Chapter 12.
(b) The current vector corresponding to sDk only has an r component.

]r = IAI 2 (Iik / m ) = r(O)

r2 4Jrr 2

The divergence of this current is

V •J = -rI2 -lJrlJ r 2 ]r = 0 (for r ::J= 0)

Since J is radial and a function only of r, we may write

r(o)
f V
V • J dr =
1r=R
J . dS =
1 - - 2 r 2 dQ
4rr 4Jrr
= I'(O)
The spherical volume V has radius R and is centered at the origin, while dQ is
an element of solid angle about this same origin. Given these two properties of
V • J. it follows that

V • J = F(O)8(r)

The three-dimensional Dirac delta function is 8(r) [see (10.19)].

[Note : Thus we see that the eigenstates of Pr have the unreasonable property of
implying that a constant flux of particles, I'(O), emanates from the origin. We
may infer from this that the operator Pr , in spite of its symmetric form (10.39)
and proper commutation property with r (Problem 10.5), is not a good observable
equivalent. Nevertheless, for problems involving a central potential, in quantum
mechanics the operator Pr proves to be a valuable tool. It is interesting to note
that inconsistencies that accompany Pr are also found in classical mechanics .
If a free point particle crosses the origin, Pr changes sign instantaneously. This
 10.3 The Free-Particle Radial Wavefunction 415

jump in Pr stems from a choice of coordinate frame. It is in no way associated

with a force (the particle is free) .3,4]

10.7 Show that the kinetic-energy operator

A 1i2 2
T = --'V
2m
is Hermitian for function s in 5), (the space of square integrable function s-see Sec-
tion 4.4) . [Hint: Use Green 's theorem

The volume V is enclosed by the surface S.]

10.8 Show that

for the superposition state

provided that

v . J(1{r\) = V . J(1{r2) = 0
and

10.3 • THE FREE-PARTICLE RADIAL WAVEFUNCTION

The time-independent Schrodinger equation for a free particle in spherical coor-

dinates appears as

I 2A LA2)
2m (p- + 7f CPklm = EklmCPklm (10.50)

The quantum number k is defined below. The radial kinetic-energy operator

Pr 2 12m is inferred from (10.48), while the angular momentum operator i.2 is
given by (9.58). Insofar as Pr 2 is a function only of r, and i 2 is a function only
of the angle variables (0, ¢), one may seek solution to (10.50) by separation of

3For further discussion of this problem, see R. L. Liboff, 1. Nebenzahl, and H. A. Fleishmann, Am. J.
Phys. 41,976 (1973).
4Related ambiguities of the radial momentum operator in cylind rical coordinates are discussed in
Problems 10.73 and 11.90.
416 Chapter 10 Problems in Three Dimensions

variables. Substituting the product form

(l0.51)

into (10.50) gives

(10.52)

In obtaining (10.52) we have recalled the eigenvalue equation for f2 (9.51) . With
the substitution
h,2k2
E=:-
2m (l0.53)
x=: kr

(l0.52) becomes the "spherical Bessel differential equation"

d
2
-R(x)
2
dR(x)
+- - - + [ 1- 1(1 + 1)] R(x) = 0 (10.54)
dx 2 X dx x2

Spherical Bessel Functions

This ordinary linear equation for the radial function R has two linearly indepen-
dent solutions.P They are called spherical Bessel and Neumann functions and are
denoted conventionally by the symbols jl (x) and ni (x), respectively. The first few
values of these functions are
sinx cosx
jo(x) = - no(x) = - - - (10.55)
x x
sinx cosx cosx sinx
h(x)=~--x- n1(x) = - - - - - -
2
x x

h(x) = (3 -
x3 x
~)sinx - 3 cos x
x2
n2(x) = - (2 - ~)cosx
X3 x
2x - 2
sin x

These functions are sketched in Fig. 10.3, from which it is evident that of the
two classes of functions , only the spherical Bessel functions Ul} are regular at the
origin. These are the solutions appropriate to the Schrodinger equation (10.50)
inasmuch as they are not singular anywhere. Some additional properties of these
spherical Bessel and Neumann functions are listed in Table 10.1.
In this manner we find that the eigenstates and eigenenergies of the free-
particle Hamiltonian in spherical coordinates are

50ne obtains these solutions by the method of series substitution. Detail s may be found in most book s
on mathematical physics, e.g., G. Goertzel and N. Tralli , Some Mathematical Methods in Physics,
McGraw-Hili, New York, 1960.
10.3 The Free-Particle Radial Wavefunction 417

10 15 x

1.0

0.5

0
10 15 x

-0.5

-1.0

FIGURE 10.3 Spherical Bessel functions hex) and spherical Neumann functions n/(x)
for I = 0, I , 2. Note that only j/ (x) are regular at the origin .

qJk/m(r, e, ¢) = j/(kr)Yt(e, ¢)
h,2k2 (10.56)
Ek=--
2m

The orthonorrnality of this sequence {qJk/m} is given by the relation

(lmkll'm'k') = {
J4rr
dQ[Yt(e, ¢)]*y/,m' (e, ¢) roo j/(kr)j['(k'r)r
Jo
2
dr

rr ,
= 8/1' 8mm, 2k 2 8(k - k ) (10.57)

The vector r has the spherical coordinates (r, e, ¢) . This orthonorrnality condition
is similar to that corresponding to the free-particle states expressed in Cartesian
coordinates (10.18), as well as that corresponding to free-particle motion in one
418 Chapter 10 Problems in Three Dimensions

TABLE 10.1 Properties of the spherical Bessel and Neumann functions

Spherical Bessel Functions Spherical Neumann Functions

jl(kr) = ( -"kr)I(1-;: dr
d)1 jO(kr) r)I(1-;: dr
nl(kr) = ( -"k d)1 nO(kr)

sinkr coskr
jO(kr) = - - nO(kr) = ---
kr kr
Equat ion

I" + ~ f +
x
[I - 1(1 x+ I)J f = 0
2

Asymptotic Values
x~O
xl 1 • 3 • 5 .. . (21 - 1)
h (x) ~ -,-----,---------,-,------ nl(x) ~ - x l+ 1
1 ·3·5 · .. (21 + 1)
x~ 00

jl(x) ~ ~ cos [x - I(I + 1)] 1 .

nl(x)~~Sill [7T
x- (l
2 + I )]

Recurrence Relations (f is written for j or n)

d l+ 1
ft-I (x) + fl+1 (x) = (21 + l)x- 1ft(x) dx [x jl(X)] = x l+ 1jl-I (x)
d d I
Ifl-I (x) - (l + l)fl+1 (x) = (21 + 1) dx ft(x) dx [x- hex)] = -x- I jl+1 (x)
Generating Functions
1 sl 1 . (-d
= L,. h-I(x) + 2xs = L --nl_1 (x)
00 00
- cos../x 2 - 2xs - Sill";x 2
x 0 I. x 0 I!
Orthogonality

roo jl(kr)jl (k'r)r 2 dr = !!.-2

10 2k
8(k - k')

f 2
fl (x)x dx
2 x
= lUI2
(x) - ft-I (X)fl+1 (x)]
Connection to Bessel and Neumann Functions of Integral Order, JI and NI

h(kr) = 2~rJ JI+I /2(kr) nl(kr) = 2~r NI+I /2(kr)

J

dimension (4.41). In all these cases, the allowed values of momentum, hk, com-
prise a continuum.
Once again we note that the projection
10.3 The Free-Particle Radial Wavefunction 419

gives the coordinate representation of the ket vector Ilmk}. In a similar manner,
the coordinate representation of the free-particle ket vector [k) is given by the
projection

I Ok
(rlk) = el or
(2rr)3 /2

(See Problem 7.31.)

Measurements on a Free Particle

Given that a particle is in the eigenstate qJklm, measurement of

E gives n2k 2 12m

L2 gives n2l (l + 1) (10.58)

Lz gives hm

How do we know that these values may be measured simultaneously? The answer
is that qJklm is a simultaneous eigenfunction of iI, f2, and i.; The existence
of such common eigenfunctions follows from the fact that iI, L2 , and i; are a
commuting set of operators. We have already discussed the commutability of L2
and i,
in Chapter 9. The fact that these operators commute with iI follows if they
commute with Pr 2 • But Pr 2 is an isotropic operator; Pr 2 fer) is constant on the
surface of any given sphere. It follows that Pr 2 is unaffected by rotations about
the origin; hence

(10 .59)

and
~ ~ ~ ~ 2

[H, Lzl = [H, L ] = 0 (10 .60)

e,
The solution qJklm(r , ¢) should be compared to the eigenstate of the free-
particle Hamiltonian in Cartesian coordinates (10 .11) ,

(10 .61)

Given that a particle is in this state, measurement of

E gives
n2 k2
--
2m
Px gives lik x (10 .62)
Py gives lik y

pz gives lik z
420 Chapter 10 Problems in Three Dimensions

In the spherical representation, (L z , L 2 , E) are specified. In the Cartesian rep-

resentation, (p, E) are specified. In the latter representation, E is redundant (E =
p2 12m), but in the former representation it is not. It is not determined by L 2 and
L z- Thus we find that in either Cartesian or spherical representations, there are
three good quantum numbers [recall (1.41)].

Free-Particle 5 States
The special case L o is of interest. For this case the Schrodinger equation
(10.50) becomes

(10.63)

The radial kinetic-energy operator Pr 2 12m commutes with the radial momentum
operator Pr and they have common eigenstates. Due to degeneracy, however (the
eigenstates Ek are doubly degenerate), eigenfunctions of Pr 2 are not necessarily
eigenfunctions of Pr (see Fig. 10.4). Owing to the inadmissibility of the eigen-
functions of Pr, it is the eigenstates of Pr 2 alone which are physically relevant.
Namely, these functions are

sinkr
C{Jk = jo(kr) = -- (10.64)
kr
Rewriting C{Jk in the form

C{Jk
1
= iP+k + iP-k = 2i b - ~
(e ikr
e- ikr
) (10.65)

reveals that it is a superposition of the outgoing wave iP+k, and the ingoing wave
iP-k, which gives zero flux at the origin.

FIGURE 10.4 Central domain represents the eigenstates common to Pr 2 and Pre Pe-
ripheral domain represents only eigenstates of Pr 2 , which alone are the physically relevant
ones.
 10.3 The Free-Particle Radial Wavefunction 421

Measurement of Lz and L2 for a Plane Wave

Next, we consider the following important problem . Suppose that a particle of

mass m is "prepared" so that it has momentum lik. Then we know that it is in the
plane-wave state (10.11).

(10.66)

Measurement of E is certain to find li 2 k 2 / 2m . Measurement of momentum is

certain to find lik. What will measurement of L; or L 2 find? And in what states
do such measurements leave the particle ? To answer this question we must expand
the given plane wave in the simultaneous eigenstates of iI, i.2, and i.; that is,
lfiklm, as given by (10.56). This expansion appears as6

00 I
eik ' r = LL aim (k)lfiklm (10.67)
I=Om=-1

where the coefficients of expansion, aim , are

(10.68)

and (k , ex, fJ) are the spherical coordinates of k (Fig. 10.5).

z z

r(r, 6, 1/1)
k = (0, 0, k)

y y

(a) (b)

FIGURE 10.5 Coordinate s relevant to the expansion of a plane wave in the eigenstate s
of f2 and t.;
(a) Direction of k is arbitrary. (b) k in the direction of the polar axis.

6See Goertze l and Tralli, Some Mathematical Methods in Physics. This expression is also discussed
in Problem 10.11.
422 Chapter 10 Problems in Three Dimensions

The relative probability that measurement of L 2 finds the value 1i2[(l + I) is

the partial sum [see (9.82) and (9.83)]

I I
p[1i2[ (l + I)] = L lalm l
2
= (4rr)2 L Iyt(a, ,8)1
2
(10.69)
m=-l m=-l

The relative probability that measurement of L 2 finds the value lim is

00

P[Iim] = (4rr)2 L Iyt(a, ,8)1

2
(10.70)
1=lml

The Spherical Well

We consider a particle of mass M confined to the interior of a spherical well with
impenetrable walls. In the domain r ~ a, the wavefunction vanishes . In the do-
main r < a, the time-independent Schrodinger equation is given by (10.50)
with general solutions given by (10.56). To impose the boundary condition
CfJ (r = a) = 0 we set

Jl(ka) =0 (1O.7Ia)

With Fig. 10.3 we see that (10.7Ia) has an infinite number of solutions . To
delineate these values we return to the notation x = kr in terms of which (10.7Ia)
becomes

(1O.7Ib)

where Xln is the nth zero of JI(X) .

Eigenfunctions and eigenenergies for the spherical well are then given by

CfJnlm(r, e, ¢) = n. (Xlnr)
---;;- Ylm(e, ¢) (1O.72a)

1i2Xln2 (a o)2 2
Enl = --2 = EIR - xt« (1O.72b)
2Ma a

where E = melM. Orthogonality of spherical Bessel functions is given by

(1O.72c)

which indicates the nature of normalization of these functions. Note that the con-
tinuous spectrum of k values for the free particle in spherical coordinates trans-
lates to the discrete spectrum of the quantum number n for the finite spherical
well problem .
10.3 The Free-Particle Radial Wavefunction 423

The following is a table of the first number of zeros of j[(x).

(t , n) (0, 1) (1, 1) (2, 1) (0,2) (3, 1) (1,2) (4, 1)

X[n tt 4.49 5.76 2n 6.99 7.73 8.18

The ground-state wavefunction and eigenenergy for the spherical well are
given by

(1O.73a)

(1O.73b)

Note that this ground state is nondegenerate.

The Cylindrical Well

We next consider the case of a particle of mass M confined to a cylindrical box of
radius a and length L. Expressed in cylindrical coordinates (Fig. 1.7), the potential
of this configuration is given by (with r written for p as depicted in Fig. 1.7)

V(r, z, ¢) = 0, r < a, 0 < z< L (1O.74a)

= 00 elsewhere (1O.74b)

Employing the Hamiltonian given in Table 10.2 and the expression for the Lapla-
cian in cylindrical coordinates given in Appendix D, the Schrodinger equation for
the confined particle is given by

a2cp I acp I a2cp a2cp 2

(1O.75a)
ar2 + -;: ar + r2 a¢2 + az2 +k cp = 0

(1O.75b)

With the separation of coordinates

cp(r, Z, ¢) = R(r)Z(z)et>(¢) (l0.76a)

(l0.75) becomes

~(a2R+~aR)+ 1 a2et>+~a2z+k2=0 (1O.76b)

R ar 2 r ar r 2et>(¢) a¢2 Z az 2

It follows that
1 d 2Z
- -
Z dz2
= constant == -kz2 (lO.77a)

2
-et>1 -dd¢2et> = constant == _m 2 (lO.77b)
424 Chapter 10 Problems in Three Dimensions

TABLE 10.2 Solutions to the three fundamental box problems in quantum mechanics

The Rectangular Box The Cylindrical Box The Spherical Box

Radius a. Height b Radius a
z z

b
P
M
j-....:..:..-+--+-~---- y

Hamiltonian

Eigenfunction

<Pqst = Aqst sinkqx sinksy sinktz <Pqmn = Aqmn Jm (Kmnp) sin kqze im¢ <Pnlm = Anlmjl(klnr)Yt«(} . 4»

(A qst)2= 8/ala2 a3 2
(A qmn) = 2/lTb[aJm (Kmna)]
2 I
(A nlm)2 = 2/a 3 [j/ (klna)]2
sinkqal = sinksa2 = sin kta3 = 0 sinkqb = Jm(Kmna) = 0 jl(kl na) = 0
Wave Equation Bessel's Equation Spherical Bessel Equation

(::2 + k
2)
sin kx =0 x1 ( x d)2 +1- m
[ 2 dx
x
2]
2 Jm(x)=O
I d)2
[( -x -
dx
x +1- 1(1+1)]
- 2-
x
n(x) = 0
Eigenenergy

Eqst = fj,2(k q2 + ks 2 + kt 2) /2M

(10.77c)

where a prime denotes differentiation with respect to r. With conditions (10.74)

we find

Z(z) = AsinkzZ. n z = 1, 2, ... (10.78a)

10.3 The Free-Particle Radial Wavefunction 425

Furthermore, as <P(</» = <P(</> + 2n), we obtain

<P(</» = Be im ¢ , m=0,±I,±2,.. . (l0.78b)

where A and B are constants. Returning to (lO.77c) and labeling

k2 _ kz2 == K 2
(lO.78c)
P == Kr

there results

(l0.79)

which is known as Bessel 's equation. General solutions to this equation are given
by

(l0.80)

where CI and C2 are constants. The functions Jm(p) and Nm(p) are called Bessel
and Neumann functions of the first kind , respectively. As Nm(O) = -00, it is
omitted in (l0.80). The functions Jm (p) oscillate," Graphs of these functions are
similar to those for the spherical Bessel functions shown in Fig . 10.3. The remain-
ing boundary conditions [first of (l0.74a)] gives

R(r = a) = 0 = CIJm(aK) (1O.81a)

Let us call the sth finite zero of Jm (p), X ms so that

(l0.81b)

The following is a table of the first few zeros of I n (p) .

s =1 s=2 s=3
XO s 2.40 5.52 8.65

Xis 3.83 7.02 10.17

X2s 5.14 8.42 11.62

X3s 6.30 9.76 13.02

Eigenenergies are given by

2 2 2
E = -I1h
2M
ti
= 2M
_(K 2 ti
+ kz2) = -2M [ mit 2]
K 2 + (-)
L

7 For integer m , J-m(p) = (_l)m Jm(p) , so that with no loss in generality, we consider only m 2: O.
426 Chapter 10 Problems in Three Dimensions

With X ms == aKms , the preceding becomes

(l0.82a)

Corresponding eigenfunctions are given by

flJmsn z(r , Z, ¢) -_ Alm (rx

ms) .
-a- SIll (nLz7TZ) eimd: (l0.82b)

m ~ 0, s > 0, nz ~ I

All three parameters are integers.

The ground state of this system is given by

2]
E = ~ [(2.40)2 7T (1O.83a)
G 2M a2 + L2
2.40r ) sin
flJG(r, Z, ¢) = Ala ( -a- (7TZ)
T (1O.83b)

and is positive within the cylindrical box. Orthogonality of the Bessel functions
in (1O.82b) parallels that of (lO.72c) and is given by

(l0.84)

which, again, indicates the nature of normalization of these functions. A compi-

lation of properties of the rectangular, spherical , and cylindrical quantum wells is
listed in Table 10.2.
Note that for spherical Bessel functions (lO.72a), Xtn represents the nth zero
of jt(x) and I is orbital quantum number. For Bessel functions of the first kind
(1O.82b), X ms represents the sth zero of lm(x) and m is azimuthal quantum num-
ber [see (9.51)]. Thus , for both spherical and cylindrical quantum wells, the or-
ders of respective Bessel functions that enter the analysis are angular-momentum
quantum numbers . Note that for Hcyl(r, Z, ¢), [i,z, Hcyll = 0, whereas for
A2 A A A

Hsph(r, fJ, ¢), [L , Hsph] = [L z , Hsph] = O. For the latter case, I and m are good
quantum numbers , corresponding to invariance of the Hamiltonian to rotations
that leave r 2 invariant. For the cylindrical case, only m is a good (angular mo-
mentum) quantum number, corresponding to invariance of the Hamiltonian to
rotations about the z axis.
Problems 427

PROBLEMS

10.9 Calculate the divergence of particle current, V • J, for a collection of particles that
are all in the state

1/!(r, t) = h(kr)e- iwt

li2k2
nw=-
2m
10.10 A spherically propagating shell contains N neutrons , which are all in the state

at t = 0.
(a) What is 1/!(r, t)?
(b) What is the expectation of the energy for this "beam"?
(c) What possible values of L 2 and L z will measurement find and how many neu-
trons will have these values?
(d) If at t = 0, measurement of L 2 finds the value 2li 2 , what is 1/!(r, t) ?
(e) If at t = 0, measurement of L z finds the value -Ii, what is 1/!(r, t) ?
10.11 Use the expansion of a plane wave in spherical harmonics ,

00 1
ei k ' r = 4JT L L l
i i!(kr)[Yt(a, /3)]*Yt(O , r/J)
I=Om=-!

together with the spherical coordinate representation of 8(r - r') ,

(the spherical coordinates of k are k , a , /3 ; see Fig. 1O.5a) to obtain the orthonor-
mality condition

The spherical coordinates ofr are (r, 0, r/J) , and those of r' are (r', 0' , r/J') . [Compare
with (C.14) in Appendix C.]
10.12 Use the addition theorem for spherical harmonics (see Fig. 9.16) to reduce the first
equation of Problem 10.11 to the expansion
00
ei kz = eikrcos B = L(21 + l)i l jl(kr)PI(cosO)
1=0
428 Chapter 10 Problems in Three Dimensions

Note that in this description, k is aligned with the polar (z) axis, so that k · r = kz
(see Fig. 10.5b). This expansion is important to the theory of partial wave scattering
and will be called upon in Chapter 14.
10.13 The expansion in Problem 10.12 of the plane wave ei kz indicates that the probability
of measuring L 2 = fi 2Z(l + 1) is

P[fi 2Z (l + 1)] ~ (2Z + Z)2

Give a semiclassical heuristic argument in support of this conclusion (i.e., that
P r-v Z2).

Answer
Consider a surface S, of constant phase of the plane wave, exp(ikz). (See Fig. 10.6.)
All points in the annular region d S = 2n r ...L d r ...L = n d (r ...L 2 ) correspond to angular
momentum L = r ...LPz = r ...Lfik. It follows that

dS = n d ( n~;2 )
The probability of finding such "points" is proportional to the annular surface d S,
so (k 2 is constant)

In the classical (correspondence) limit, L 2 r-v fi 2Z2 and P r-v Z2.

k = (0,0, k)

x
FIGURE 10.6 The probability of finding a particle in a plane-wave state with angular
momentum <hl increases as Z2. (See Problem 10.13.)
Problems 429

10.14 How many independent eigenstates are there corresponding to a free particle mov-
ing with energy Ek = Iik2 /2m in
(a) the Cartesian coordinate representation ?
(b) the spherical coordinate representation ?
Give a classical description of the different orbits corresponding to these degenerate
states.

Answers (partial)
(a) In the Cartesian representation , any state

is an eigenstate corresponding to the given value of E . These k vectors describe

a sphere of radius J2m E I 1i2 . This continuum of states correspond s to aiming
the particle in different directions , while holding its speed, hk l m, fixed.
(b) In the spherical representation , any state

2 2mE
f{!klm = jl(kr)Yt(8, 4», k =-
1i2

is an eigenstate corresponding to the given value of E . Different states are ob-

tained by choosing different values of I and m. This countable infinity of states
corresponds to propitious choice of straight-line trajectories about the origin,
all at constant speed, hk[ m ,

10.15 An electron is confined to a spherical well of radius a = 10 A. Calculate the emis-

sion frequency, !J (in Hz), of radiation for this system due to transition from the first
excited state to the ground state. State the class of radiation to which this frequency
belongs (IR, visible, UV, soft x rays, etc.).
10.16 A "quantum dot," relevant to semiconductor devices, may be modeled as an electron
in a spherical well with perfectly reflecting walls. Design a quantum dot whose
characteristic frequency of emission is 10 GHz, where "characteristic frequency"
corresponds to decay from the first excited state to the ground state. That is, obtain
the radius, a, of the spherical cavity that has this property. Repeat this problem for
the quantum dot that is a cube of edge length 2b. In either case assume an effective
mass of m = 0.067me . Give your answers in /Lm.
10.17 (a) What are the eigenfunctions and eigenenergies (in eV) of the ground and first
excited states of an electron trapped in a cylindrical box of radius a = 1.5 A
and length L = 3.0 A? What is the degeneracy of these two eigenenergies ?
(b) An electron propagates in a "quantum wire" of circular cross-sectional radius
5.2 A and z component of kinetic energy 1.03 eV. What is the lowest energy
wavefunction and energy (in eV) corresponding to this condition ?
(Note: A quantum wire is a rectilinear wirelike domain of narrow band-gap
material immersed in a medium of wide band-gap material. In this problem
we model the quantum wire as being of infinite length and assume that the
propagating electron acts as a confined free particle.)
430 Chapter 10 Problems in Three Dimensions

10.4 • A CHARGED PARTICLE IN A MAGNETIC FIELD

A closely allied motion to that of a free particle is the motion of a charged particle
(e.g., an electron) in a uniform, constant magnetic field B. The Hamiltonian for
the electron is given by

H = - 1 ( p--A e)2 (10.85)

2m c

The magnetic field is related to the vector potential A through the relation

B=VxA

The Cartesian components of A,

A = (-yB , 0, 0)

generate a uniform magnetic field which point s in the z direction .

B = (0,0, B)

Substituting this value of A into the Hamiltonian above gives the time-independent
Schrodinger equation

(10.86)

Since the coordinates x and z are missing from the Hamiltonian, it follow s that

and we may conclude that Px, Pz, and fI have simultaneous eigenstates. The
eigenstates of Px and pz appear as

so that we may write the common eigenstates of fI, Px, and pz in the form
(10.87)

Substituting this product into (10 .86) gives

-py-2 + -(y
[ 2m 2
K - YO)2 ] f 2) f
= ( E - _rh z_ (10.88)
2m
10.4 A Charged Particle in a Magnetic Field 431

where we have set

chk;
YO= - - -
e!B

K = (e!B)2 = Q2
m me

The frequency Q is called the cyclotron frequency . This is the frequency of ro-
tation corresponding to the classical motion of a charged particle in a uniform
magnetic field (see Problem 10.18).
The Schrodinger equation (10.88) is the same as that for a simple harmonic
oscillator constrained to move along the y axis, about the point YO, with natural
frequency Q . From Section 7.2 we recall that the eigenenergies of this equation
are

which gives the desired result

(10.89)

The kinetic-energy term 1i2 kz2 12m corresponds to free, linear motion parallel to
the z axis. Classically, such motion is unaffected by a magnetic field in the z
direction . The first term in En corresponds to the rotational motion normal to
the :B field. In the corresponding classical motion the charged particle moves
in a helix of constant radius, constant energy, constant rotational frequency, and
constant z velocity. The projection of the motion onto the xy plane is a circle with
a fixed center (Fig. 10.7). The energy levels (10.89) are commonly referred to as
Landau levels.
The eigenfunction corresponding to the eigenenergy (10.89) is

[recall (7.59)]. The nth-order Hermite polynomial is written Jfn , while An is a

normalization constant. Together with (10.87), this form for [« gives the wave-
function

(10.90)
432 Chapter 10 Problems in Three Dimensions

Projection of orbit
onto the z axis is that
of a free particle
Z = vzt + Zo
y
- , - - - - - - projected motion in
xy plane is circular

x
FIGURE 10.7 Helical motion of a positive charge in a uniform, constant magnetic field
that points in the z direction.

for a charged particle moving in a uniform magnetic field which points in the z
direction.

Degeneracy of Landau Levels

In this section we wish to discover the manner in which the continuous energy
spectrum of a free electron changes when the electron is in the presence of a
magnetic field. In the course of this discussion we will obtain an expression for
the degeneracy of Landau levels.
To examine this problem, we consider the electron to be enclosed in a large
cubical box of edge length L. Free-particle wavefunctions and energies are given
by (10.14). To account for the finite enclosure, for L sufficiently large, say L 2 »
n/mQ, we impose periodic boundary conditions:

f{J(X, y, z) = f{J(X + L , y + L, z + L)
As the Landau levels given by (10.89) are evidently degenerate in kx , we will fo-
cus on this wavenumber for the free-particle motion . Substituting the free-particle
state (10.14) into the relation above gives

2rrn x
kx = - -
L
n x =0,±1,±2, .. .
10.4 A Charged Particle in a Magnetic Field 433

Thus we discover the following important fact. The continuous energy spectrum
of a free, unconfined particle changes to a nearly continuous discrete spectrum
when the particle is confined in a large enclosure .
In the presence of a magnetic field, the Schrodinger equation for the electron
is given by (10.88). As noted in Problem 10.18, YO is associated with the center
of the corresponding classical circular motion. Thus, for the present case we may
take YO to lie between 0 and L.
It follows that the maximum value of this parameter YO is given by

_ cli],x chii x
yo=L= 14~ = 14~L
We may conclude that n x values are positive numbers and have the maximum
value"

n- x =- 9
- -- (~)L2 (10.91)
c
This is the desired expression for the degeneracy 9 of a Landau level. The pres-
ence of L 2 in (10.91) corresponds to the property that a given helical orbit can be
displaced anywhere in the xy plane without changing the energy of the electron
(see Fig. 10.7).
Thus, we find that the energy spectrum of a confined electron changes from
a nearly continuous one for 93 = 0 to a discrete spectrum for 93 > 0, with
degeneracy given by (10.91).
From (10.89) we see that the spacing between Landau levels, at fixed kz, is the
constant value

~E = IiQ

The degeneracy 9 given by (10.91) gives the number of free-particle states that
contribute to the increment ~E . See Fig. 10.8. Note, in particular, the resem-
blance between the equally spaced Landau levels and the equally spaced levels of
the harmonic oscillator shown in Fig. 7.8. This congruence of spectra stems from
the previously described parallel structure of the two respective Hamiltonians . In
either event, the density of states, 9 (E) (Section 8.8), when plotted as a function
of energy, is a series of equally spaced delta functions.
Note further that in the classical limit Ii ~ 0, the degeneracy 9 grows infinite
and the spacing between levels ~ E goes to zero.

Fermi Energy and Landau Levels

Application of the preceding results may be made to the conduction electrons in
a two-dimensional metal. In this event, the degenerate states shown in Fig. 1O.8a
become filled with electrons . The highest energy state occupied (at 0 K) is that of
8We employ the barred variable 9 to distinguish it from the unbarred variable 9 used to denote density
of states elsewhere in the text.
434 Chapter 10 Problems in Three Dimensions

>------£5

>------£4

>--...----E2
n 0: increment
>---------E I

(a) $= 0 (bL~> 0

TC onstant
number
occupied

3BO

(c)
5BO
1
FIGURE 10.8 (a) Nearly continuous discrete spectrum for a particle confined to a large
box with 93 = O. (b) Equally spaced Landau levels corresponding to 93 > O. Each incre-
ment of energy, liQ, corresponds to 9 free-particle states, which, in tum, is the degeneracy
of each Landau level. (c) Variation of the Fermi energy with change in 93.

the Fermi energy, EF [Sections 2.3 (Fig. 2.3), 8.4, 12.9]. With a magnetic field
/B normal to the plane of the sample, we may identify E F as the highest Landau
level occupied. How does E F change with change in /B? To answer this question,
we first note that the maximum number of electrons at the Fermi level is given
by the degeneracy factor "9 (10.91). Let nOc :s "9 denote the number of occupied
Problems 435

states in this level. The value of E F remains constant as long as n oc > O. When 93
is increased, E F changes to the value of the new partially occupied Landau level.
See Fig . 1O.8c. Such variation of EF exhibits a periodicity with respect to 93- 1 .
This phenomenon is a component of the de Haas-van Alphen effect.'

PROBLEMS

10.18 The following is a problem in classical physics. The force on a charged particle in
a uniform magnetic field B is

d e
F = -(mv)
dt
= -v
c
x B

with B = (0,0,93) .
(a) Show that

I
2mv2 = constant
(b) Show that

Pz = mv z = constant
(c) Show that the motion of the particle is that of a helix whose axis is parallel
to B and whose projection onto the x y plane is circular with constant angular
frequency n.
(d) Show that the center of this circle in the xy plane has coordinates

-cpx -cmvx
Yo= --+y= - - + y
e93 e93
cmvy CPy
Xo = - -
e93
+x = -e93 + x
Note that Px- canonical momentum, is not equal to mvx for A = (Ax, 0, 0). The
correct relation follows from (1.l4) and (10.85).
10.19 Show that the operator

A A cpy
xO=x+-
e93

commutes with fI as given in (10.86) but does not commute with

A -cpx
Yo=--
e93
These operators correspond to the coordinate s of the center of the related projected
classical motion in the xy plane. In quantum mechanics we see that although xo and

9For further discussion, see C. Kittel, Introduction to Solid State Physics, 6th ed., Wiley, New York,
1986, Chap. 9.
436 Chapter 10 Problems in Three Dimensions

E, or YO and E, may, respectively, be specified simultaneously, -o and YO may not

be simultaneously specified.
10.20 (a) What is the vector potential A which gives the uniform 13 field (0,0, /11) which
includes Ax = A z = o?
(b) What is the form of the wavefunctions f{!n corresponding to this choice of vector
potential ? How do they compare to the wavefunctions corresponding to A y =
A z = 0 found in the text?
(c) How do the eigenenergies compare to those found in the representation A y =
A z =o?
10.21 What is the nature of the frequency spectrum emitted by a charged particle moving
in a uniform magnetic field? (Assume that the kinetic energy parallel to 13 does not
change .) For an electron moving in a 13 field of 104 gauss, what type of radiation is
this (x rays, microwaves , etc.)?

10.5 • THE TWO-PARTICLE PROBLEM

Coordinates Relative to the Center of Mass

When dealing with systems containing more than one particle (e.g., an atom), it is
convenient to separate the motion into that of the center of mass of the system and
motion relative to the center of mass. This separation is effected through a parti-
tioning of the Hamiltonian into a part HCM, involving center of mass coordinates,
and a part, Hrel , containing coordinates relative to the center of mass .
For example, consider the two-particle Hamiltonian

pl2 p22
H = -
2ml
+ -2m2 + V(lrl - r21) (10.92)

The potential of interaction V (Irl - r21) is a function only of the radial distance
between the particles. For instance, for the hydrogen atom, the interaction V is
the Coulomb potential

e2
V=-- (10.92a)
r
where we have written r for the distance between particles, Irl - r21. Such poten-
tials , which are only a function of the scalar distance r, are called central poten-
tials.
In the Hamiltonian above, PI and P2 are the linear momenta of particle I
and particle 2, respectively, while ml and mz are the respective masses of these
particles.
A two-particle system has six degrees of freedom which are the number of
coordinates in the state vector (rj , PI ; r2, P2) . The partitioning of the Hamiltonian
into HCM + Hrel is generated through the transformation of variables

(rj , PI ; r2, P2) --+ (r, P; s; 9» (10.93)

10.5 The Two-Particle Problem 437

where

3' = p, +P2
m,P2 - m2P, :R = m,f, + m2f2
P= , (10.94)
m,+m2 m, +m2
Using these equations the Hamiltonian (10.92) is transformed to the sum

H = -
9'2 + [p2
- + V (r)
] == HCM + Hrel (10.95)
2M 2/-L

where the reduced mass /-L and the total mass M are

M =m, +m2 (10.96)

Equation (10.95) represents the desired separation of H into the Hamiltonian of

the center of mass, HCM, and the Hamiltonian of the coordinates relative to the
center of mass, Hre, (Fig. 10.9). Since :R is absent in H (i.e., :R is a cyclic co-
ordinate; see Section 1.2), the momentum of the center of mass, 3' , is constant.
The center of mass moves in straight rectilinear motion , characteristic of a free
particle of mass M. The motion relative to the center of mass is that of a particle
of mass /-L moving in the central potential VCr).

FIGURE 10.9 The relative vector r and the center-of-mass vector :R . In the classical
motion, :P = M IR = constant = :P(O) = the initial value of :P(t) . At any time t

:R(t) = IR(O) + :P~)t

Solving the dynamical equations (viz., Hamilton 's equations) using Hrel gives ret), which
when affixed to :R(t) gives the motion in the "lab frame."
438 Chapter 10 Problems in Three Dimensions

The Transformation of H
For the quantum mechanical case, the transformation of H is again effected with
the equations (10.94), which are now interpreted as operator relations . Cartesian
components of "old" coordinate and momentum operators (rl , PI ; h, pz) obey
the commutation relations

[rlj , Plj] = in
j = 1,2,3 (10.97)
[rzj, Pzj] = in

These are the only nonvanishing commutators. With these relations and (10.94),
one obtains that the only nonvanishing commutator relations for components of
the "new" operators (r, P; i?, 9» are

[rj , Pj] = in
j = 1,2,3 (10.98)
[~j , ~j] = in

Thus, in obtaining

H = HCM + Hrel
A ~Z
HCM=- (10.99)
2M
AZ
Hrel = L + V(r)
2J.L

the Hamiltonian is separated into two parts involving components that are inde-
pendent of one another. For such cases, the Schrodinger equation has product
eigenfunctions

(10.100)

and summational eigenvalues

E= ECM + Erel (10.101)

where

Hrp = Erp
HCMIfJCM = ECMIfJCM (10.102)

HrellfJrel = ErellfJrel

The Schrodinger equation for the center of mass appears explicitly as

(10.103)
10.5 The Two-Particle Problem 439

y y

(a) (b)

FIGURE 10.10 (a) Spherical angle variables for the center-of-mass radius vector R.
(b) Coordinates relative to the center of mass.

This is the Schrodinger equation for a free particle of mass M . Its solution was
obtained in the previous section . With the linear momentum P specified, the states
are

(10.104)
P = IiK,

In the representation where LCM 2 and LCM z are specified, the eigenstates are

(10.105)

The spherical coordinates of Rare (R , Be, ¢>c) (see Fig. 1O.lOa).

Radial Equation for a Central Potential

The Schrodinger equation for ~rel appears as (dropping the "rel" subscript)

(10.106)

For central potential functions V (r), it proves most convenient to express the
above Hamiltonian in spherical coordinates. The interparticle radius r has co-
ordinates (r , B, ¢» with the polar axis depicted as lying in the z direction (see
Fig. 1O.10b).
440 Chapter 10 Problems in Three Dimensions

In these coordinates the Schrodinger equation above becomes

(10.107)

First, we note that L2 and i , both commute with fl . The remaining components
in fl are all isotropic forms and are therefore unaffected by angular momentum
operators . It follows that fl, L2 , and i; have simultaneous eigenstates. These are
given by the product form

({! = R(r)Yt(e, ¢) (10.108)

Substituting this solution into the Schrodinger equation above gives the "radial"
equation

!i2 / (1 + 1) ]
!!!-2 +
A

2
+ VCr) R(r) = ER(r) (10.109)
[
2J-L 2W

This is an ordinary, second-order, linear differential equation for the radial depen-
dent component of the wavefunction R(r). Since only one variable is involved
in (10.109), it is suggestive of one-dimensional motion with the effective poten-
tial

!i21(l + 1)
Veff = V Cr) + 2 (10.110)
2W

The second term in this expression is called the "angular momentum barrier." It
becomes infinitely high as r ---* 0 and acts as a repulsive core, which for I > 0
prevents collapse of the system (see Fig. 10.11).
The normalization of the eigenstates (10.108) is given by the integral

roo dr r 2 r dQ IR(r)Yt(e, ¢)1

2
(RYtIRYt) = = 1 (10.111)
Jo J4rr
= 1 00

r
2IR(r)1 2
dr = 1

The radial displacement r separates the two particles m 1 and ma - If we envision

particle m 1 at the origin, then

(10.112)

is the probability of finding m-; in the volume element r 2 dr dQ about ml (an

equally valid statement is obtained with ml and ma reversed). What is the prob-
ability of finding m2 in a spherical shell of radius between rand r + dr , about
10.5 The Two-Particle Problem 441

Angular momentum
barrier
11 2/(1 +-!)
2j1r 2

Discrete, {
bound
states

(b)

(a)

1=2
1=3

-}
(c)

(d)

FIGURE 10.11 (a) The effective potential in relation to the angular momentum barrier

n2 Z(l + 1)
Veff = VCr) + 2
2J-tr

(b) Nature of the quantum mechanical spectrum for central potential problems. (c) The
classical motion corresponding to the energy E. Shaded regions define classically forbid-
den domains. (d) The effective potential energy Veff for hydrogen for several values of the
orbital quantum number Z. Units of r are angstroms. Veff is in units of 10- 11 erg.
442 Chapter 10 Problems in Three Dimensions

fJ. particle

Center of
force
Volume of shell = 41rr 2dr

FIGURE 10.12 Probability of finding the fictitious I-t particle in a spherical shell between
rand r + dr is
2
P; dr == IR(r)rI dr

This is also the probability of finding m2 in a spherical shell about m 1 in the configuration
shown (or ml in a shell about m2).

ms'! The answer is (Fig. 10.12)

r. dr = (Lr 2
Pr dQ) dr = IR(r)1 2r 2 dr == 2
lu(r)1 dr (10.113)

so that

Et---......~---..,.------

FIGURE 10.13 The radial probability amplitude UE corresponding to the energy E de-
cays to zero in the classically forbidden domain.
Problems 443

The classically forbidden domains (see Chapter 1) correspond to values of r

for which E < V. The related property for a spherical quantum mechanical sys-
tem is that the probability density lu(r)1 2 becomes small in these domains (see
Figs. 1O.11c and 10.13).
Having found the radial functionR(r), in a specific two-body problem, the
wavefunction for the system relative to the laboratory frame (as opposed to the
center-of-mass frame) is either of the forms

if = Ae i :P. 3{/ li R(r)Yt(B , ¢)

(10.114)
if = jle(K :R)Yle me (Be, ¢c)R(r)Yt (B, ¢)

In the first representation, the six parameters UP, L 2 , L z , E) are specified. In

the second representation, the six parameters (ECM, LCM 2, LCM z ; L 2 , L z- E) are
specified.

Continuity and Boundary Conditions

Some general properties of the radial wavefunction are as follows . With R (r)
everywhere bounded, we note first that u(r) == r R(r) must vanish at the origin. 10
For r > 0, with energy E and potential energy V (r) bounded , the radial equation
(10.109) indicates that

p,2R(r) = _n2~r ~ [~u(r)]

or or

is likewise bounded. It follows that ou/or is continuous. The existence of this

derivative implies that u(r) is continuous. The latter two conditions infer continu-
ity of the logarithmic derivative

1 du dlnu
u dr dr
These conditions on the wavefunction in spherical coordinates are employed in
obtaining the ground state of the deuteron (Problem 10.31) and in construction of
the bound states of the hydrogen atom as described in the following section . They
will also come into play in construction of the states for low-energy scattering
from a spherical well (Section 14.2).

PROBLEMS

10.22 Consider a two-particle system. The momenta of the particles are PI and P2, re-
spectively.

IODirac obtains this boundary condition from the stipul ation that solut ions to the Schrodinger equation
in spherical coordinates agree with those obtained in Carte sian coordinates. For furth er discu ssion , see
P. A. M . Dirac, The Principles of Quantum Mechanics, 4th ed ., Oxford University Pre ss, New York,
1958.
444 Chapter 10 Problems in Three Dimensions

(a) What is [PI, P2]? Explain your answer.

(b) Use the answer to part (a) to show that [p, 3>] = 0, where P is the "relative"
momentum [defined in (10.94)] and tp is the momentum of the center of mass .
(c) The particles interact under a potential that is a function of the distance between
them . As shown in the text, transformation to coordinates relative to the center
of mass effects a partitioning of the Hamiltonian, H = HCM + Hrel' What is
[HCM, Hred?
10.23 Prove that the following equations are compatible with the transformation equations
(10.94).
PI 2 P22 p2 9'2
(a) - + - = - + -
2ml 2m2 2/L 2M
(b) mlrI2+m2r22=w2+M:R2
(c) PI • rl + P2 • r2 = p' r + tp • tR
(d) LI +L2 =L+LCM
In part (d),

LI = rl x PI L=r x P

L2 = r2 x P2 LCM=tRXtp

10.24 At a particular time, the wavefunctions of a mass m moving in a three-dimensional

potential field is

cp = A(x + y + z)e- kor

(a) Calculate the normalization constant A.
(b) What is the probability that measurement of L 2 and L z finds 2h2 and 0, respec-
tively? (See Table 9.1.)
(c) What is the probability of finding the particle in the sphere r ~ k I ?
O
10.25 For a two-particle system (m I, m2) , what is the fractional distance to the center of
mass from ml and m2 , respectively ? What are these numbers for hydrogen ?
10.26 Let e be a unit vector in an arbitrary but fixed direction . Show that the commutators
between the components of rand P, respectively , with the component e . L obey the
relations

[p, e· L] = ihe x P
[r, e . L] = i he x r
10.27 Use the results of Problem 10.26 to show that jJ2 , ;:2, and r • P all commute with
every component of L Then show that every component of L commutes with any
isotropic function f(r 2 ).

Answer (partial)
If the statement is true for arbitrary e, it is true for all components of L

[jJ2, e . L] = p[p, e . L] + [p, e . L] . p

= ih(p . ex p + ex p . p) = 0
Problems 445

10.28 Two free particles of mass ml and m i - respectively, move in 3-space . They do not
interact. Write the eigenfunctions and eigenenergies of this system in as many repre-
sentations as you can . Indicate the number of parameters specified in the eigenstates
associated with these representations.

10.29 Write down the time-dependent wavefunction corresponding to eigenstates for a

two-particle system in the two representations (10.114).
10.30 Consider two particles that attract each other through the potential

The displacement between particles is r, K is a constant, and I-t is reduced mass. In

states of definite angular momentum, what are the values of the angular momentum
quantum number I for which the effective force between particles is repulsive ?
10.31 In Problem 8.1 the depth of the potential well appropriate to a deuteron was evalu-
ated using a one-dimensional approximation. A more refined estimate may be ob-
tained using a three-dimensional spherical well with characteristics

V(r) = -lVI, r < a region I

V(r) = 0 r :::: a region II

(a) Construct components of the ground-state wavefunction in regions I and II,

respectively.
(b) Show that matching conditions at r = a give the dispersion relation
7J = -~cot ~
p2 = ~2 + 7J2
where

~ ss ka; 7J ss « a

(c) To within the same approximation suggested in Problem 8.1, obtain a numerical
value for the depth IV I of the three-dimensional deuteron well. From the ratio
lEI/IVI for this bound state, would you say that the deuteron is a strongly or a
weakly bound nucleus?

Answer (partial)
(a) Component wavefunctions in the well domain , region I, are the spherical Bessel
functions . The ground-state component is therefore

sinkr
fPI = - -
kr
In region II the component ground-state wavefunction is exponentially damped .
446 Chapter 10 Problems in Three Dimensions

e- K r
<PI! = AKr
--

(b) Here one must invoke continuity of din u j dr,

(c) You should obtain the answer IEI/IVI = 0.08. The binding energy of a
deuteron is 2.2 MeV which, with the preceding estimate, indicates a relatively
large well depth. A sketch of the normalized wavefunction further reveals that
there is approximately only one chance in three that the nucleons are closer
together than the well radius a.

10.32 Show by explicit calculation that the eigenfunctions and eigenenergies as given in
Table 10.2 are correct for each of the three respective "box" configurations shown.
(Primes denote differentiation .)

10.6 • THE HYDROGEN ATOM

Hamiltonian and Eigenenergies

The (relative) Hamiltonian for the hydrogen atom (more accurately, for a "hydro-
genic" atom 11 of atomic number Z) appears as

(10.115)

The corresponding Schrodinger equation is

(10.116)

We are seeking the bound states of hydrogen . These correspond to the negative
eigenenergies, E = -lEI. Setting q; = R(r)Yt(B, ¢) in the latter equation gives
the radial equation (10.109)

(10.117)

Changing the dependent variable to

u =rR

introduced previously in (10.113), gives

(10.118)

11 Hydrogenic atoms are atoms that are ionized with all but one electron bound to the nucleus which
carries the charge +Ze (e.g., He+, Li++ , etc .).
10.6 The Hydrogen Atom 447

Introducing the notation

li2K 2
p == 2Kr, -=1£1
2jJ.,

Z)2 Z2~
).,2 = ( KaO = lET (10.119)

li2 li2
~=--, aO=-
2jJ.,a02 I-U 2

where ~ is the Rydberg constant'? (2.13) and ao is the Bohr radius (2.14), the
radial equation may be further simplified to the form

+ (~ _ ~) u =
2
d u _ 1(1 + 1) u 0 (10.120)
d p2 p2 P 4

For large values of p this equation reduces to

d 2u u
---=0
dp 2 4
so that
u ~ Ae- p / 2 + Be P/ 2

In order that u vanish as p ~ 00, we set B = 0, so

(p ~ 00)

In the neighborhood of the origin, (10.120) reduces to

d 2u l(l + 1)
d p2 - p2 u = 0

Substitution of the trial solution u = pq gives

u ~ Ap-l + Bpl+1
In order for u to vanish at the origin, we must set A = O. This gives
u ~ /+1 (p ~ 0)

With these two asymptotic forms at hand, we are prepared to solve (10.120)
through a polynomial expansion . Solution in the form
u(p) = e- p / 2/ +1F(p)
00 (10.121)
F(p) = LCipi
i=O
12The Rydberg constant written with m in place of p., (i.e., assuming infinite proton mass) is sometimes
written !Roo .
448 Chapter 10 Problems in Three Dimensions

with F finite everywhere, gives the proper behavior at p ~ 0 and p '" 00 . Substi-
tuting (10.121) for u into (10.120), we obtain

P-2~ + (21 + 2 - d - (l
p)- + 1- A)] F(p) = 0 (10.122)
[ dp dp

Note that, for a given value of the orbital quantum number I, this is an eigenvalue
equation with eigenvalue A. The values of A (or, equivalently, the eigenenergies,
lEI) are those values which ensure that F(p) is finite for all p. Substituting the
series (10.121) into the latter equation and equating coefficients of equal powers
in p gives the recurrence relation

C + 1+ 1) - A C. =
(i r 'IC , (10.123)
i+ 1 = (i + l)(i + 21 + 2) I - I I

In the limit that i -+ 00 , this relation becomes

Ci
CHI'" -:-
I

which is the same ratio of coefficients obtained in the expansion

. pi
e'' = L Cip' = L ~ l.

CHI i! 1
---=-- ~-
C, (i + 1)! i +1
It follows that the form of u(p) generated by the series (10.121) behaves as
u(p) ~ e- p / 2pl+l eP = eP/ 2pl+1

which diverges for large p . To obtain a finite wavefunction, the expansion

(10.121) for any given value of I must terminate at some finite value of i , which
we will call i max . At this value of i, ri/ = O. Since all parameters in (10.123) are
positive, ri/ can only vanish if

i max + 1+ 1 = A

The function u so generated is a polynomial and, due to the exponential term in the
form (10.121) , we see that, as demanded , the wavefunction is finite everywhere.
Since i and I are integers, it follows that A is also an integer, which is called
the principal quantum number, n ,

n = i max + 1+ 1
Thus the above cutoff condition on the series (10.121), which ensures that u (p) is
finite for all p , also serves to determine the eigenenergies A.

2 2 Z2~
An =n = - -
IEnl
10.6 The Hydrogen Atom 449

(10.124)

These are the same values found previously in the simpler Bohr model (Sec-
tion 2.4).

Laguerre Polynomials
The hydrogen eigenfunction corresponding to the eigenvalue En is given by
(10.121) with the series over i cut off at the value

i max = n -I - 1 (10.125)

and the recurrence relation for the coefficients {C;} given by (10.123).
n-l-l
Unl(p) = e- p/ 2pl+ 1Fnl(P) = Anle-p/2pl+l L C;/
;=0 (10.126)
Z
Kn = -
aon
where Anl is a normalization constant. The polynomials Fnl(p) (of order n-I-l)
so obtained are better known as the associated Laguerre polynomials, L~l~~1
(see Table 10.3). The reader should take note of the fact that the scale length of
p changes with different values of n . This is because the radial displacement r is
nondimensionalized through the wavenumber K n , which is dependent on n. Note
also that eigenstates in Table 10.3 are in terms of R(r) = u(r) jr.

Degeneracy
Since i max 2: 0, with (10.125) we obtain

ISn-l

So for a given value of the principal quantum number n, the orbital quantum
number I cannot exceed the value

Imax = n - 1 (10.127)

This corresponds to the values I = 0,1 ,2, . . . , (n - 1). Each of these I val-
ues corresponds to different values of i max and therefore different wavefunctions.
Inasmuch as the eigenenergy En depends only on the principal quantum number
n, these n distinct orbital states are degenerate . For instance, there are three dis-
tinct radial functions that correspond to the eigenenergy E3. These are U3.0 , U3.1 ,
and U3,2.
The complete eigenstate of the Hamiltonian (10.115) contains the factor
Yr(e , 4» [see (10.108)]. For each value of I, there are 21 + 1 values of mi :
m; = -I , ... ,+1, which correspond to distinct Y]" functions that give the same
450 Chapter 10 Problems in Three Dimensions

TABLE 10.3 Eigenfunctions of hydrogen in tenns of associated Laguerre polynomials

The Normalized Eigenfunctions of Hydrogen (2 = 1)

If!nlm(r, e, 1/» = (2K)3/2 Ani pi e- p/ 2 Fnl(P)Yt (e, 1/» = Rnl(r)Yt (e, 1/»
22
P = 2Kr = - r fooo IRnl(r)1 2 r 2 dr = 1
aOn

(n -I - I) !
If!IOO = -1- (22)3/2
- e-(Z/ao)r
2n[(n + 1)!]3 -/Sir aO

n-I-l ( i 2 i
F: ( _L21+l ( _L21+l _" -1) [(n+/)!] p
nl p) - n-I-l p) - i max (p) - :So
i! (n -I - 1 - i)!(21 + 1 + i )!

Associated Laguerre Polynomials L~(p) and Laguerre Polynomials Lp(p)

Differential equation

Generating function
e-ps/(l- s) 00 sP
(l-s)q+1 = L (p+q)!L~(P) Lb(O) = p!
p=O
Orthonormality

00
e
-p qLq Lq d - [(p
p p , p -
+ q)!]3 8
pp'
10o p p!
Rodrigues 's formula

dP
Lp(p) == L~(p) = e P dp P (pPe- P ) ,

q dq
Lp(p) == (-I)q dpq [Lq+p(p)]

Recurrence relations

pL~(p) = (2p + q + I)L~(p) - [(p + 1)/(p + q + 1)]L~+1 (p) - (p + q)2 L~_l (p)

(p :p +q - p) L~(p) = (p + I)L~~~ (p)

~Lq
dp P
(p) = _L q+ l (p)
p-l

Relationto Other Notations

Alternative notations for the polynomial L ~ may be found in other texts. The relation between this notation (B) and our own (A) is given by
the following table."
 10 .6 The Hydrogen Atom 451

TABLE 10.3 (Continued)

Notation A Notation B
(e.g., found in Merzbacher, Messiah , and here) (e.g., found in Pauling and Wilson, Schiff, and Tomonaga)
(-)q L~ L~+q
p(L~)1I + (q + I - p)(L~)' + pL~ =0 p(L~+q)1I + (q + I - p)(L~+q)1 + pL~+q =0
Rnl = A(2K)3/2 e-p /2 pl L~~~I (p) Rnl = _A(2K)3 /2 e-p /2 pl L~~I(p)
L ~ is a polynomial of order p L ~+q is a polynomial of order p or, equivalently,

b
L is a polynomial of order (b - a)

The first row in this table tells us that L ~ is written L ~+q in notation B. The second row indicates that both L functions satisfy the same dif-
ferenti al equation. The third row gives the form s of the radial solution Rnl in both notations. Still another notation appears in I. S. Gradshteyn
and I. M. Ryzhik .l' where

L~(p)[here) = (p + q) !L~(p)[G and R)

"E. Merzbacher, Quantum Mechani cs, 2d ed., Wiley, New York, 1970.
A. Messiah, Quantum Mechanics, Wiley, New York, 1966.
L. Pauling and E. B. Wilson, Introduction to Quantum Mechan ics, McGraw-Hili, New York, 1935.
L. Schiff, Quantum Mechanics, 3d ed., McGraw -Hili, New York, 1968.
S. Tomonaga , Quantum Mechanics, North-Holland, Amsterdam , 1966.
bI. S. Gradshteyn and I. M. Ryzhik , Tables ofIntegrals, Series and Products, Academi c Press, New York, 1965.

TABLE 10.4 Allowed values of 1 and mi for n = I, 2, 3

n I 2 3
1 0 0 I 0 I 2
Spectroscopic IS 2S 2P 3S 3P 3D
notation of state

ml 0 0 -1 ,0, +1 0 -1,0, +1 -2,-1 ,0,+1 ,+2

Degeneracy I 4 9
of state (n 2 )

eigenvalues of i.2 [i.e., 1i21(l + 1)]. All these 21 + 1 functions when substituted into
(10.116) give the same radial equation, (10.117), which contains only the orbital
number l. It follows that for each solution Unl of (10.120) , there are (21 + 1) solu-
tions to the Schrodinger equation (10.116) corresponding to the same eigenenergy
En (see Table lOA). In this manner we obtairr' '
n-I
Degeneracy of En = 2)21 + 1) = n2 (10.128)
l=O

13Induding spin, degeneracy of states is 2n 2 . This topic is more fully discu ssed in Section 12.4.
452 Chapter 10 Problems in Three Dimensions

~---m=+/

n- - - --.,......
/ =n - 1

m =-/

~---m=2
1
/ =2
o
-1
-2

m= 1
/=1
o
-(

/=0
m =0
FIGURE 10.14 Term diagram for a hydrogenic atom illustrating all n 2 degener ate states
corresponding to the principal quantum number n .

To recapitulate, the allowed values of n, l , and m are (see Fig. 10.14)

n=I ,2,3, . ..
l = 0, 1,2, . . . , (n - 1) (10.129)
m = -l, -l + 1, . . . ,0, 1, 2, .. . , +l
Additional Properties of the Eigenstates
The eigenfunctions and eigenenergies of the hydrogenic Hamiltonian (10.115) are

f{Jnlm(r, e, <!» = Rnl(r)Yt(e, <!»

Rnl = AnlUnl (10.130)

r
Z2JR Jl,,(Ze 2 ) 2
En = - - - = - '----"-;;--;0-
n 2 21i2 n 2

A term diagram of these energies is given in Fig. 10.15 (compare Fig. 2.8). The
normalization constant Ani (see Table 10.3) is determined by the condition
10.6 The Hydrogen Atom 453

-0.25 2-

-13 .58 eV -0.5 Bound states

-0.75

-1.00 I-
• S p D F G H
n
[--0 2 3 4 5
FIGURE 10.15 Energy-level diagram for hydrogen , including the I ::: 5 terms . Energy
is measured in units of III

Note also the orthogonality of these functions

The Ground State

To construct the ground state <PIOO (n = I , I = 0, m = 0) we must first find ulO.

From (10.126), with Co = I, and inserting the normalization constant!" AIO, one
obtains

14Altematively, we may take Co = AIO '

454 Chapter 10 Problems in Three Dimensions

Normalization gives

This gives the normalized ground-state wavefunction (with Z = I)

1
UIO = --pe- p/2 = -2r
- e - ria 0
~ ao3/ 2

_ UIO _ _ 2_ -r/ao
R 10 - - 3/2 e
r ao

_ R ~ 0 _ 2 -r/ao
CPIOO- 10 0 - (4Jr)1/2ao3/2 e

Additional Properties
The first few normalized eigenstates of hydrogen, with corresponding eigenener-
gies obtained as outlined above, are listed in Table 10.5.
In Fig. 10.16 nondimensionalized radial functions , Rnl = ao3/ 2Rnl , are plot-
ted together with the corresponding nondimensionalized probability density func-
tions, Pnl = 41WOUnl2 = 4JraOPr. These sketches reveal the shell structure of
hydrogen found earlier in the Bohr theory.
The time development ofthe states of hydrogen follows from (3.70). Consider
that

1/!(r,O) = CPnlm = Rnlyt (10.131)

The state at time t 2: 0 is then

1/!(r, t) = e- d f t / fi1/!(r , 0) = e-iEnt /ficpnlm (10.132)

The charge density associated with this state is

q(r, t) = el1/!nlm(r, t)12 = q(r) = eICPnlm(r)1 2 (10.133)

which is independent of time. The electronic charge is e. Thus the atom suffers
no radiation in these states. This topic will be returned to in the next section.
The density configurations, ICPnim 12 , corresponding to some of the eigenstates
of hydrogen are sketched in Fig. 10.17. Since the angular dependence of ICPnim 12
2
is entirely contained in the factor 1 yt 1 , it follows that ICPnim 12 is independent of
the azimuthal angle 4> [see (9.78)]. It is rotationally symmetric about the Z axis.
Thus we need only present a representation of Icpl2 in any plane which includes
the Z axis, such as is depicted in Fig. 10.17. The value of Icpl2 is proportional to
the density of whiteness in each of the states depicted.
10.6 The Hydrogen Atom 455

TABLE 10.5 Several normalized time-independent eigenstates of hydrogen

Spectroscopic
Notation Normalized Time-Independent Eigenstates

IS

2S

2P

3S

3P

<fJ322 )
111321 2v7. r 2

~~~l CJ
3D _ _ e-r/ 3ao
( - 27../5 (3aO)3/ 2
11132-2

Hybridization
Having discovered the wavefunctions of hydrogen, a significant consequence of
quantum mechanics emerges important to the formation of molecules. This prop-
erty concerns the manner in which atomic wavefunctions exhibit geometric ori-
entation in atom-atom binding.
Consider the 2P states of hydrogen listed in Table 10.5. Owing to their degen-
eracy, any linear combination of these states is an eigenstate corresponding to the
same eigenenergy, -lR/4. Thus, for example, the following set of three orthog-
onal wavefunctions spans the same subdimensional Hilbert space as the original
three 2P states:
1 r/ 2ao (10.134a)
CP2Px = J2(CP211 - CP21-}) = Axe-

l (CP211 + CP21-1 )=' A -r/ 2ao

CP2Py = J2 - I ye (lO.l34b)

CP2Pz = CP2lO = Aze-r / 2ao (l0.134c)

 5
4
3
PIO 3
2
2

OL----""'---''------'--........- _ OL---~ ........- - - " - - - - - ' - -....

1OJ 15
r/ao
0.8

l~---,---~
0.6
0.4
P" 0.2
0.0 L--~_'--_::::::;~~----:-'=__-
5 10 15 20
r/ao

l~--.L------'-_ o 0.6
0.4

P" 0.2
I 0.0 L -_ _ ~ = _ _ _ . l . _ _.....1-_ _

5 10 15 20 10 15

:L
r/ao r/ao
0.3
P" (0 0.2
0.1
O-~'.
5 10 15 20 25 10 15

:L
r/ao r/ao

0.3
P" (0 0,2
0.1
O-~II
5 10 15 20 25 15
r/ao

P" . . @ :::0.3L
~"
_~I 0.0 ~5===i;:::::::::::::"~- I
o 5 10 15 20 10 15
r/ao r/ao

FIGURE10.16 Nondimensional radial probability density P and nondimensional radial

wavefunction R, versus nondimensional radius r/ ao. for hydrogen . Note that the probabil-
ity density P exhibits the shell structure of the atom .
456
10.6 The Hydrogen Atom 457

IS 2P 3D 4F

(a)

25 3P 4D SF

(b)
FIGURE 10.17 Probability density, I Cf>nlm 12 , for various states of hydrogen. The plane
of the paper contains the polar axis, which points from the bottom to the top of the figure.
From Principles ofModern Physics, by R. B. Leighton. Copyright 1959 by McGraw-Hill.
Used with permission of the McGraw-Hill Company.
458 Chapter 10 Problems in Three Dimensions

2
Angular plots (at fixed r) of the corresponding probability densities 1({I2P 1
x ,
2 2
1({I2Py 1 , I ({I2P 1 , reveal that they are figure-eight surfaces of revolution about the
z
x, y, and z axes, respectively (similar to those shown in Fig. 9.10).
Linear combinations of the wavefunctions (10.134) come into play in the for-
mation of the methane molecule CH4. The four outer-shell electrons of carbon
are in the 2s 22 p 2 configuration (see Table 12.2). Wavefunctions appropriate to
these four electrons are formed from linear combinations of the three 2P states
(10.134) together with the 2S state of hydrogen and are given by

1
1/II = ,J4«{I2S + ({I2Px + ({I2Py + ({I2Pz ) (l0.135a)

1
1ft2 = ,J4«{I2S + ({I2Px - ({I2Py + ({I2Pz ) (l0.135b)

1
1ft3 = ,J4«{I2S + ({I2Px - ({I2Py - ({I2Pz ) (1O.135c)

1
1ft4 = ,J4«{I2S - ({I2Px + ({I2Py - ({I2Pz ) (l0.135d)

With the four outer electrons of carbon in these respective four 1ft orbitals,
the following picture emerges. Angular plots of the probability densities of
these wavefunctions about a common origin reveal maxima along the (I, 1, 1),

H
[1, I , I]

I
H--C--H

I
H

.J----+-----~~H

H~----------:-:--V

FIGURE 10.18 Orientation of C and H atoms in the CH4 molecule. Here we have
employed crystallographic notation for direction, where a bar over a direction num-
ber indicates negative direction. Shown also is the two-dimensional representation of
this molecule. Generalizations of this diagram for CnH2n+2 hydrocarbons are evident.
(Straight-chain alkanes.)
Problems 459

(-1 , -1 , 1), (1, -1, -1), and (-1, 1, -1) direction s, respectively. With the car-
bon atom at the center of a cube, these orbitals reach out to four tetrahedral
comers and covalently bond with hydrogen atoms at these sites to form the Cf4
molecule . See Fig. 10.18.
These orbitals come into play in the chaining of CH3 molecules. Thus, for ex-
ample, in the formati on of C2H6, an uncoupled electronic 1fr orbital of CH3 bonds
with a like orbital of an adjacent CH3 molecule to form the ethane molecule. The
concept of atomic bonding in the formation of molecule s is returned to in Chap-
ter 12.

PROBLEMS

10.33 With Co = 1 in the recurrence relation (10.123). obtain Cl . Then use (10.126) to
show that

U20 = A20re-r/2ao (1 - ~)
2ao
Calculate A20 and CiJ200. Check your answer with the value given in Table 10.5.
10.34 Show that U10 2 has its maximum at r = aO, the Bohr radius.
10.35 Solve the equation Ix + I = 0 by expansion technique and check with the solution
1= Ae-x .
Answer
Assume that
00

1= LCi Xi
o
to obtain
00 00
LCiixi-1 + LCiXi = 0
o 0
With s = i-I in the first series, we get
00 00

L Cs+l(S + l)x s + L Ci x i = 0
s=-l i= O

Since the first term in the first series is zero, we may write this equation in the form
00
L[Ci+l (i + 1) + C;Jxi = 0
i=O

which is satisfied if and only if

460 Chapter 10 Problems in Three Dimensions

10.36 The average energy of the hydrogen atom in an arbitrary bound state X(r) is given
by the integral

(E) = (x(r)IHlx(r»)

Show that

Answer
Since X(r) lies in the Hilbert space spanned by the basis {fPnlm}, we may expand
00 n-I I
Ix) = L L L bnlmlnlm)
n=II=Om=-1

so that

(The ket vector Inlm) represents the state fPnlm ') Owing to the fact that all eigenener-
gies are negative, En ::::: 0, the statement to be proven is equivalent to the inequality

I(E)I ::::: lEI!

I(E)I = LLLl bl 21Eni

.s IElmax LLLl bl 2 = IElmax = IEII
10.37 (a) What is the effective Bohr radius and ground-state energy for each of the fol-
lowing two-particle systems?
(1) H2 , a deuteron and an electron (heavy hydrogen).
(2) He+ , a singly ionized helium atom.
(3) Positronium, a bound positron and electron .
(4) Mesonium, a proton and negative f-t meson. The f-t meson has mass 207me
and lasts ~ 10-6 s.
(5) Two neutrons bound together by their gravitational field.
(b) Calculate the frequencies of the (n = 2) -+ (n = I) transition for each of the
systems above.
10.38 At time t = 0, a hydrogen atom is in the superposition state

(a) Calculate the value of the normalization constant A.

(b) What is the probability that measurement of L 2 finds the value 1i2t(l + I)?
(c) What is the probability density Pr(r) [see (10.113)] that the electron is found
in the shell of thickness dr about the proton at the radius r?
Problems 461

(d) At what value of r is Pr(r) maximum ?

(e) Given the initial state 1/!(r, 0), what is 1/!(r, t) ?
(f) What is 1/!(r, t) if at t = 0, measurement of L z finds the value Ii?
(g) What is 1/!(r, t) if at t = 0, measurement of L z finds the value zero?
(h) What is the expectation of the "spherical energy operator," (Hs ), where

, 2
, , Pr
Hs=H--
2Jt

at t = O?
(i) What is the lowest value of energy that measurement will find at t = O? (Lowest
means the negative value farthest removed from zero.)

10.39 Find the lowest energy and the smallest value for the classical turning radius of the
H -atom electron in the state I = 6 (see Fig . 10.11).

10.40 In what sense does the Bohr analysi s of the hydrogen atom give erroneous results
for the magnitude of angular momentum, L ?

Answer
The Bohr analysi s that yields the eigenenergies -lR/n 2 assumes circular orbits.
Circular orbits do not exist in the Schrodinger theory. Quantization of the action .
f P9 dO, in the Bohr theory gives L = nh, In the Schrodinger theory, the maximum
value of L is Ii.jn(n + I) , which is less than the value that L assumes (nli) in the
Bohr theory.

10.41 What is the ionization energy of a hydrogen atom in the 3P state?

10.42 Show that Rn/(r) has (n - I - I) zeros (not counting zeros at r = 0 and r = 00).
10.43 (a) Show that the expectation of the interaction potential V (r) for hydrogenic
atoms is

(b) Calculate (nl m IT lnl m) , where the kinetic-energy operator t is given by

What relation do (T) and (V) satisfy (the virial theorem)?

10.44 Obtain an explicit expression for the probability density Pr(r) corresponding to the
state whose energy is £2 , for a hydrogenic atom [see (10.113)].

Answer
There are four degenerate eigenstates correspond ing to the energy £2 . Since no
direction is preferred for a Hamiltonian whose only interaction term is the central
potential V (r) , all these degenerate states carry the same "weight" (all 1m states are
equally probable). There results
462 Chapter 10 Problems in Three D imension s

= i. 12~n (~r e-
Zr a
/ [(2- ~~r + (~~r
x (~ sin2 e + ~ sin 2 e + cos 2 e) ] 2
x r dr dQ

P(r) = -- -
128n
1 (2)3
aO
e- Zr / ao [(2- - 2r)2
aO
+ (2r)2]
- ,
«o
Pr(r) = 4n r 2 P(r)
10.45 Give a physical argument in support of the conjecture that the sum

n-l +1
Ln == L L [Yt]*[Ytn Rntl
I=Om=-1
2

is independent of e or cP o
10.46 Show that for a hydrogen atom in the state corresponding to maximum orbital an-
gular momentum (I = n - 1),

(n, n - l lr ln , n -1 ) =aon (n+~)

(n, n - llr 2ln , n - 1) = ao2n2(n + (n +~)

1)

10.47 Use the result of Problem 10.46 to show that for large values of n and I,

f(;2) -+ aOn 2
t!>.r -+ 0
(r)
1 e2
En -+ - - - -
2 n2ao
That is, show that for large values of n , the electron is located near the surface of a
sphere of radius aon2 and has energy which is the same as that of a classical electron
in a circular orbit of the same radius. Recall that (t!>. r)2 = (r 2 ) - (r) 2.

10.48 Calculate (r) in the state f{inlm of hydrogen.

Answer

(r ) = ex (x) + ey (Y) + ez(Z)

(x ) = fff I2 2 2
r cos o sin e Ytl Rnl r dr d cos e dcP

=0
 10.7 ElementaryTheory of Radiation 463

since

l27r cos¢d¢=O

2
and I yt 1 is independent of ¢. Similarly, (y) = O. For (z) we must calculate

(z) = [11 Iptl2cosBdcosB f ...

Using the recurrence relations listed in Table 9.3, we find that (z) = O. It follows
that (r) = O.
10.49 Establish the following properties for hydrogen in the stationary state If!nlm'
1
(a) 2 +2 (r s ) _ (2s + l)ao(r s- I )+ :'[(21 + 1)2 _s2]ao 2 (r s-2 ) = 0, s > -2/-1
n 4
(b) (r) = n 2 [1 +~ (1 _ 10 ~ 1») ] aO

(c) (~) = 2
2
r (21+ l)n3ao2
1
(d) (r 2) = 2[5n 2 + 1 - 3/(1 + 1)]n2ao2

(e) (~) =_1

r n2ao

(f) (~) _ 2
3
r - ao 3n31(1 + 1)(21 + 1)
[H int : Multiply (10.120) by {ps+I u' +[(s+ 1)/2]pS u} and integrate by parts several
times . Note that for hydrogenic atoms, ao is replaced by ao /Z .]
10.50 Show that the most probable values of r for the 1 = n - 1 states of hydrogen are

r = n2ao
These are values that satisfy the equation
d 2
dr (Unl) = 0

10.7 • ELEMENTARY THEORY OF RADIATION

In the last section we found that the hydrogen atom does not radiate in its eigen
(stationary) states. The charge density (10 .133) is fixed in space with configura-
tions such as depicted in Fig. 10.17 . In these states the hydrogen atom is stable
against radiation. This is opposed to the classical description in which the elec-
tron loses kinetic energy to the radiation field and collapses to the nucleus (see
Section 2.1).
The student may be perplexed about the absence of radiation from the state
1/Jnlm . He/she may well ask: Doesn't the electron have a well-defined angular mo-
mentum in such a state, and doesn't this correspond to accelerated motion which
gives rise to radiation? His/her friend answers: Maybe the orbit of the electron is
464 Chapter 10 Problems in Three Dimensions

so peculiar that, on the average, the radiation field washes out. After all, we know
that if L 2 is specified, two of the three components of L remain uncertain.
The best way to see what the electron is doing in quantum mechanics is to
calculate (r) . Specifically, we must calculate this expectation in the state 1/Jnlm.
Suppose that we find (r) ,...., e z cos on, Then the electron is suffering linear, sim-
ple harmonic oscillation. Such oscillation gives rise to dipole radiation. But we
have already calculated (r) in Problem 10.48, where we found that (r) = O. Not
only is (r) time-independent in the eigenstates of hydrogen, but it is also centered
at the origin. Note that we may also reach this conclusion by the much simpler
argument: Calculate

The average of r is independent of time; hence it must also be zero since the
Hamiltonian , (10.115), is isotropic. It contains no vectors. It in no way implies a
"preferred" direction, so (r) cannot be a finite constant vector.
Thus while the stability of the hydrogen atom to radiative collapse is totally
inexplicable on classical grounds , our quantum mechanical model renders a de-
numerably infinite set of states {1/Jnlm} in which the atom suffers no radiation .
How, then, does the atom radiate? In the Bohr theory of radiation, we recall
that a photon is emitted when there is a transition from one eigenenergy state to
a lower one. Such a decay might be induced by the collision of the atom with
another atom in a gas. It might also be induced by collision with an electron in a
discharge tube. It might also be induced by collision with a photon in the interior
of a star. IS
Suppose that at time t = 0, the atom is in an excited (stationary) state 1/Jn (n
denotes the sequence nlm). The atom is perturbed, emits radiation, and decays to
the state 1/Jn' (Fig. 10.19). We may conclude that in the interim, the atom is in the
superposition state

(10.136)

En -------.~----

En' -----.....1-----
FIGURE 10.19 Atom emits a photon in decaying from state o/n to o/n"

I5Fundamentally, all these collision processes involve the exchange of photons. For further discu ssion,
see E. G. Harris, A Pedestrian Approach to Quantum Field Theory, Wiley, New York, 1972 .
10.7 ElementaryTheory of Radiation 465

At any time t , in this interim (by the superposition principle) la 12 represents the
probability that the atom is in the state 1/Jn and Ibl2 represents the probability that
it is in the state 1/Jn' > These coefficients are therefore time-dependent. At t = 0,
(Ial = 1, Ibl = 0). At t = 00, (Ial = 0, Ibl = 1). Let us calculate the expected
value of the position of the electron during this collapse.

(r) = (a1/Jn + b1/Jn' Irla1/Jn + b1/Jn')

2(1/Jn
= laI2(1/Jnlrl1/Jn} + IbI 'lrl1/Jn'}
+ a*b(1/Jnl rl1/Jn'} + b*a(1/Jn'lrl1/Jn}
The first two terms are time-independent and do not contribute to radiation. The
last two terms combine to yield

(r(t)} = a*bei(En-En,)t/Ii(CPnlrICPn'} + b*aei(En,-En)t/Ii(CPn'lrICPn} (10.137)

= 2 Re [a*b(CPn IrlCPn' }ei(En-En,)t/Ii]

= 21 a *b(CPn IrICPn'} Icos(wnn,t + 8) = 2lrnn,I cos(wnn,t + 8)

where W nn' is the Bohr frequency

8 is a phase factor and la*bl is assumed to be slowly varying and of order unity.
Atomic transitions typically occur in an interval of the order of 10-9 s. The
frequency of emitted radiation , on the other hand, is typically of the order of
lOIS s-I , so the radiative oscillatory behavior of (r(t)} is due almost exclusively
to the cos term, with accompanying Bohr frequency W nn' (Fig. 10.20). When the
atom is undergoing a transition between the states 1/Jnand 1/Jn' » the average posi-
tion of the electron oscillates with the Bohr frequency corresponding to the energy
difference between these states. At the beginning and conclusion of the transition,
the atom is in stationary states in which it does not radiate. (These topics are
returned to in Section 13.9.)

Selection Rules
Harmonic oscillation of an electron about a proton gives rise to what is commonly
referred to as dipole radiation'" (Fig. 10.21). The average radiated power from
such an oscillating dipole is 17

(10.138)

16An atom also radiates in higher multipole channels (e.g., quadrupole). For the most part, dipole
radiation is predominant.
17See J. D. Jackson, Classical Electrodynamics, 2d ed., Wiley, New York, 1975.
466 Chapter 10 Problems in Three Dimensions

laI 2 + l bI 2 = 1
(a)

COS w nn ' !

'V 10- 8 S (

I..

'V I0 7 oscillatio ns

(b)
FIGURE 10.20 Radiation decay from the 1/!n state to the 1/!n' state involves the super-
position state 1/! = a1/!n + b1/!n' with time-varying coefficients a and b. The "beat" fre-
quency W nn' between the 1/!n and 1/!n' states is much greater than the switchover frequency
of1/!.

where d is the dipole moment

d = era, (r) = ra cos cot (10.138a)

We may apply this formula to calculate the power radiated when the hydrogen
atom decays from the nth state to the (n')th state. From (10.137) we obtain

d = 2er nn, (10.138b)

so that, with (10.138),

(10.139)
Problems 467
+e

r= ro cos wt

Radiation intensity
from oscillating
dipole "v sin 2 8
d =ero
Average power
-e radiated :
4
P =!~ldI2
3 c3
( a) (b)

FIGURE 10.21 Energy characteristics of an oscillating dipole . (a) Dipole configuration.

(b) Radiation profile of a dipole .

Calculation (see Problems 10.52 to 10.54) of the squared matrix element Irll ll '1 2
with n standing for nlm and n' for n'l'm ' gives the following selection rules: The
only conditions under which Irllll '1 2 (and therefore P) is not zero are

ill = l' -I = ±l and Sm; =m' -m =O,±l (10.140)

For example, the transition 3S ~ IS (ill = 0) is forbidden, as is the transition

3D ~ 2S (ill = 2). Such transitions are not accompanied by any (dipole) radi-
ation and therefore are excluded by conservation of energy. The exclusion of the
transitions between S states finds analogy with the classical theorem that spheri-
cally symmetric oscillatory charge distributions do not radiate .
The rule ill = ± 1, together with the law of conservation of angular momen-
tum, indicates that for ill = -1 the electromagnetic field (i.e., the photon) carries
away angular momentum. As it turns out, photons have angular momentum quan-
tum number equal to 1 and are therefore called bosons.
There are no restrictions on an atomic transition corresponding to change in the
principal quantum number n. This is in agreement with the pre-Schrodinger spec-
tral notation for emission from hydrogen: namely, the Lyman series correspond s
to transitions from all n states to the ground state, the Balmer series corresponds
to transitions to the n = 2 states, etc. (see Fig. 2.8).

PROBLEMS

10.51 What are the allowed transitions from the 5D states of hydrogen to lower states?
Accompany your answer with a sketch repre senting these transitions.
468 Chapter 10 Problems in Three Dimensions

10.52 With n representing the triplet nlm and n' the triplet n'l'm', show that the matrix
elements of r have the following complex representation :

[Hint: Call (n lx ln' ) == X nn' , etc. Note also that l(r)1 denotes the magnitude of the
vector (r ), while I(x + iy)1 denotes the modulus of the complex variable (x + iy).l
10.53 Show that the matrix elements of x ± iy have the following integral representation :

i:r [ytl*Yl'm'sinBe±i<f:>dQJoroo Rn[Rnl[lr

(a) (n lx ± iYln' ) = 3dr

(b) (n lz ln' ) = r [ytl*y[ ,m' cosBdQ Joroo Rn[Rn,[,r3 d r

J4rr
10.54 Using the results of the last two problems and Tables 9.1 and 9.3, establish the
selection rules for dipole radiation (10.140).
10.55 At the start of Section 10.7 it was noted that the energy eigenstates of an atom are
stable against radiative decay, owing to their stationarity in time. However, these
states do contain angular momentum, and one may argue that they therefore also
contain rotating charge, which does radiate. Although the premise of this argument
is correct, why is there still no radiation from the stationary states (from a classical
point of view)?

Answer
An aggregate of N uniformly spaced point charges confined to move with fixed
speed in a closed loop will radiate as a result of the acceleration of individual
charges . However, in the limit that the charges approach a uniformly continuous
distribution, N ~ 00, f:!. ~ 0, q ~ 0 (with total charge Nq and line charge
density q / f:!. constant), the radiation may be shown 18 to vanish. This limiting case
closely resembles the state of affairs for the stationary states of an atom. Although
there is rotating charge, such charge is continuously distributed and, in accord with
the classical prescription, does not radiate .

10.56 The interaction potential of an electron moving in the far field of a dipole d is

ed
V = --cosB
2r
The dipole is at the origin and points in the z direction. The spherical coordinates
of the electron are (r, B, 1».
(a) Write down the time-independent Schrodinger equation for this system corre-
sponding to zero total energy. What condition of the "atom" does this state
describe ?
(b) Show that solutions to this equation are of the form rp = r S I(B , 1». Obtain an
equation for I(B, 1» .
10.57 Two particles that are isolated from all other objects interact with each other through
a central potential. As was established in Problem 1O.23(d), the total angular mo-

18See Jackson. Classical Electrodynamics. Chapter 14.

Problems 469

mentum of the system may be written

Show quantum mechanically that this total angular momentum is conserved.

10.58 (a) Prove that the Runge-Len; vector

,I ' r
K = --2 [L x
2p,e
P- p x L] + -r
commutes with the Hamiltonian of the hydrogen atom (10.115) .
(b) Show that the operator

satisfies the commutation relations

(c) Use the result above to show that the operators

, I "
B+ = Z(L+A)
, 1"
B_ = -(L-A)
2

obey the angular momentum commutation relations and the equality

, 2
B+ 2
B_ .
(d) Derive the relation

B 2 + B_2 = ~ (tt2+ p,e4)

+ 2 2£

and use it to obtain the Bohr formula for the energy levels of hydrogen.

10.59 The classical harmonic oscillator with spring constant K and mass m oscillates at
the single frequency (independent of energy)

The quantum mechanical oscillator, on the other hand, gives frequencies at all inte-
gral multiples of wo, as follows directly from the eigenenergies

(n = 0, 1,2, .. . )

If one end of the oscillator is charged, dipole radiation is emitted. In the classi-
cal domain, this radiation has frequency wo . Show that selection rules that follow
from calculation of the dipole matrix elements x nn' reduce the quantum mechanical
spectrum to the classical one. (The concept of matrix elements of an observable is
developed formally in Chapter 11.)
470 Chapter 10 Problems in Three Dimensions

10.60 Consider a gas of noninteracting rigid dumbbell molecules with speeds small com-
pared to the speed of light. The moment of inertia of each molecule is l .
(a) What is the Hamiltonian of a molecule in the gas?
(b) What are the eigenenergies of this Hamiltonian?
(c) Let a molecule undergo spontaneous decay between two rotational states. Ow-
ing to the recoil of the center of mass, there is a change in momentum of the
center of mass of the molecule as well. Show that the frequency of the photon
emitted in this process is

1i.k'D)
V = vI ( 1- Me

The initial momentum of the center of mass is 1i.k, D is a unit vector in the
direction of the momentum of the emitted photon, and {VI} is the rotational line
spectrum .
(d) What is the nature of the frequency spectrum emitted by the gas?

Answers
(a) Let :P and M denote the momentum and mass, respectively, of the center of
mass. Then

1i.2 k 2 1i.2[([ + 1)
(b) Ek,l = 2M + 2/
(c) The frequency of photons emitted by a molecule is given by the change in
energy

where hVI is written for the change in rotational energy. Since the molecule is a
free particle, the momentum of the center of mass 1i.k can change only by virtue
of the momentum carried away by the photon emitted in the transition . As a first
approximation we will assume that the momentum carried away is hvi]c. If 1i.k'
is the momentum of the center of mass after emission, then by conservation of
momentum we have (see Fig. 10.22)

FIGURE 10.22 Change in linear momentum of rigid rotator due to recoil in the emission
of a photon (see Problem 10.60).
Problems 471

nk = nk' + hVI n
e
so that
, hVI ,
n~k= n(k -k) = - - n
e
Substituting this value in the expression above for h v gives

v = VI ( 1-
nk.n)
Me

Since hk « Me , the assumption that momentum carried away by this radiation

field is approximately hvi ]«: is justified.
(d) The rotational spectrum (vil remains a line spectrum with an infinitesimal
broadening of lines.

10.61 Consider two identical rigid spheres of diameter a, which move in 3-space.
(a) What is the Hamiltonian of this system?
(b) Separate out the center-of-mass motion to obtain an equation for the wavefunc-
tion for relative motion, rp(r) .
(c) What boundary conditions must rp(r) satisfy?
(d) What are the eigenstates and eigenenergies for this system?
(e) What are the parities of these eigenstates ?
(1) What is the radial probability, P(r), for this system?

Answers (partial)
, Pl 2 fil2
(a) H(I ,2) = 2m + 2m + V(lrl - r21)

V(lrt - r21) == V(r) = 00, r ~a

V(r) =0, r >a

(b)
' 2
!!!..-
[ 2p.,
L2 + V(r) -
+ --2
2W
E
] rp(r) = °
(c) rpklm(r) = 0, r ~a
(d) rpklm=rplm(kr)Yt(8,t/J), r >a

rplm(kr) = A[nl(ka)h(kr) - h(ka)nl(kr)]

h2k2
Ek=--
2p.,
(e) Referring to Problem 9.46, we obtain

JPrplm (kr) = (_)1 rplm (kr)

472 Chapter 10 Problems in Three Dimensions

10.8 • THOMAS-FERMI MODEL

The Thomas-Fermi model is appropriate to atoms with sufficient number of elec-

trons which permits the system to be treated in a statistical sense. The model
views an atom as a spherically symmetric gas of Z electrons which surrounds
a nucleus of charge Ze . It is further assumed that the potential in the medium,
<I>(r) , varies sufficiently slowly so that it may be taken as constant in small vol-
umes of the electron gas. The description of the Fermi energy, such as given in
Section 2.3, then applies. (See Fig. 10.23.) There, we recall the Fermi energy was
equated to the maximum kinetic energy of an electron in an aggregate of electrons
°
at K. This description maintains in the present model since at nonnal tempera-
ture k s T « <I>(r) . (The Fermi energy is discussed further in Section 12.9.)
The relation between EF and electron density nCr) is given in Problem 2.42,
and with the preceding description we write

li 2
Kin - E - _[3x 2n (r ) ]2/3
E max- F- (10.141)
2m

The energy of an electron in the medium at a distance r from the nucleus is

p2
E(r) = - - e<l>(r) (10.142)
2m

where we have written e = lei. The preceding equation indicates that for
p 2 / 2m > e<l>(r), E > 0, which implies that the electron escapes the atom.
As this does not occur in the present model, we may conclude that maximum

Small cube of electron

gas has 4> = constant
withE F given by (10 .141)

Nucleus

Charge Ze

FIGURE 10.23 The Thomas-Fermi statistical atom.

10.8 Thomas-Fermi Model 473

kinetic energy may also be set equal to e<l>. With (10.141) we may then write

An additional relation between the potential <I> (r) and density n (r) is given by
Poisson's equation (cgs),

Working in the given spherically symmetric geometry, and eliminating nCr) from
the latter two equations, gives

2 I d2 4ne (2me)3/2 3/2

V <I> = -; dr 2 (r<l» = 3n 2 li3 <I> (10.143)

This is the desired single equation for <I> (r) .

The following boundary conditions apply to the potential. Near the origin <I> (r)
is due predominantly to the nucleus, and we write

Ze
<I> ~­
r
or, equivalently,

lim r<l>(r) = Ze (10.144a)

r ...... O

At large r we obtain

lim r<l>(r) = 0 (10.144b)

r...... oo

The resulting equation (10.143) may be written in a more concise form in terms
of new dimensionless variables (x , X) defined by

r =ax (10.145a)

a = ~ (3n )2/3 li2jme 2 0.885ao

(10.145b)
- 2 4 ZI /3 ZI /3
Ze
<l>(r) == -X(x) (10.145c)
r
When written in terms of these new variables (see Problem 10.68), (10.143) re-
duces to

(10.146)
474 Chapter 10 Problems in Three Dimensions

log x

16
x
FIGURE 10.24 Semi logarithmic plot of solution to the Thomas-Fermi equation
(10.146) subject to the boundary conditions (10.147).

This is the standard form of the (highly nonlinear) Thomas-Fermi equation.

Boundary conditions (10.144), when written in terms of the new potential X (x),
appear as

lim X = I , lim X = 0 (10.147)

x->O x->oo

The semilogarithmic plot shown in Fig. 10.24 reveals the rapid decay of the
potential X with increase in x . Thus , for example, for x > 4, X < 0.1. 19

Variation of Atomic Size with Z

An important conclusion of the preceding model is as follows : It is evident from
(1O.145b) that the parameter a may be viewed as an effective radius of the atom
(i.e., the radius of a sphere which contains a fixed fraction of atomic electrons).
Thus there is an implied decrease in effective atomic size with increase in Z .
However, it should be borne in mind that the shell structure of atoms is lost in the
Thomas-Fermi model. Thus one might expect the implied decrease in atomic size
to be roughly valid within a given atomic shell of fixed principal quantum number.
Observations find this property to be approximately obeyed . (See Fig. 12.11b.)
The closely allied "Hartree-Fock" approximation is described in Section 13.10.

19Numerical integration of the Thom as-Fermi equat ion (10.146) was originally performed by V. Bush
and S. H. Caldwell , Phys. Rev. 38 ,1898 (1931) .
Problems 475

PROBLEMS

10.62 A hydrogen atom suffers a decay from the 3D to the 2P state and emits a photon.
(a) What is the frequency of this photon (in Hz)?
(b) Given that the decay time of this transition is ~ 10- 10 s, what is the power (in
W) emitted in the transition ?
10.63 Show that the three IfJ2p states (10.134) comprise an orthonormal set. What is the
value of the constant A? Leave your answer in terms of square roots and QO .
10.64 Of what angular momentum operators are the respective three 1fJ2P states (10.134)
eigenfunctions?
Answer
As described in Chapter 9, when operating a wavefunction j(r), the operators
Lx, i.; L, effect a rotation of r about the x, y, Z axes, respectively. Thus 1fJ2Px
is an eigenstate of Lx; vz», is an eigenstate of i.; and IfJ2pz is an eigenstate of i.;
All three functions are eigenstates of L2 .
10.65 Show that (¥rl l¥r2) = 0 for the orbitals (10.135).
10.66 (a) Show that the function IfJ2px as given by (10.134a) is maximum along the x
axis.
(b) With this property at hand, argue that ¥rl as given by (10. 135a) exhibits maxima
along the (l , I, I) directions .
Answers (partial)
(a) We may write

1fJ2Px = A sin II cos cf>re-rj2ao

A polar plot of sin II reveals a circle tangent to the Z axis at the origin. Thus
sin II generates a circular torus about the z axis. A polar plot of cos cf> reveals
a circle in the xy plane tangent to the y axis at the origin and with maximum
at cf> = 0 (positive x axis). The intersection of these two factors, sin II cos cf>,
is evidently maximum along the positive x axis. The remaining r-dependent
component merely modulates this result.

10.67 This problem concerns annihilation and creation operators for spherical Bessel
functions .20 Introducing the operators (for 1 > 0)

± i d .1
bI =---x±/-
x dx x
(a) Show that

, +' __ ' _ + 1 d2 1(1 + 1)

bI bI -bI+1bI+ 1 =-~ dx 2 +~

(b) Show that Bessel's equation (10.54) may be written

(bI+bI- - I)RI(X) =0
20L. Infeld, Phys. Rev. 59,737 (1941).
476 Chapter 10 Problems in Three Dimensions

(c) Show that

(d) Show that

(e) Apart from normalization, employ the preceding formalism to obtain:

i, h (x) from jO(x) .
ii. nl (x) from nO(x)
10.68 The radial wavefunction satisfies the second-order differential equation (10.109).
The student may recall that any such equation has two independent solutions . Show
that for bound states there is only one linearly independent solution for given I and
E values.
Answer
The equation may be written

b2 _ l(l + 1) + 2J.t (E _ V)] U = 0

[ r2 1i2
, d
D==.-
dr
U ==. rR

Let solutions to the preceding equation be written Ul and U2 . Multiplying the equa-
tion for U2 by uland vice versa and subtracting the resulting equations gives

ulb2u2 - u2b2ul = 0

or, equivalently,

Integrating, we find

where C is a constant. For bound systems one may assume that either U 1 , U2 -+ 0
or b«1, bU2 -+ 0 as r -+ 00, which, in either case, gives C = O. There results

dlnU2 dlnUl
dr dr
which gives

U2 = aUl
where a is a constant. Thus, U2 and U1 are not independent.
10.69 Employing the transformation equations (10.145), show that (10.143) reduces to the
Thomas-Fermi equation (10.146) .
Problems 477

10.70 This problem addresses motion of an electron in a steady magnetic field as described
in Section lOA .
(a) Find the mean square radius a 2 of an electron moving in a plane normal to a
steady magnetic field.
(b) Obtain your answer to part (a) once again, but now through use of the corre-
spondence principle .

Answers
(a) Referring to (10.86) et seq. indicates that

Due to the harmonic oscillator structure of (10.88), we may write

2 2
((y - YO) ) = K (V)

where V is potential energy. With Problem 7.10 we may then write

2 2 (E ) I
a = - x - = - (E )
K 2 K

a= ~ (n+~)
2
mQ 2

(b) The classical kinetic energy of the electron is

I I I
-mv 2 = _ma 2 Q 2 = - E
222
Thus

2 E
a =--
mQ2

Inserting the quantum value for E gives the desired result.

10.71 You are asked to design an absorbing sink for rotationally polarized radiation of
frequency 2.5 x 10- 10 Hz. You propose to use a gas of dipole molecules . What is
the moment of inertia, I (cgs), of molecules that will perform this task in I = 0 --+
I = I transitions ? What is the effective radius, ro (in A), of these molecules if their
reduced mass is 30 amu ? (See Problem 1.23.)

10.72 (a) Show that iplm (r, 0 , ¢J) = r l yt

(0 , ¢J) are eigenfunctions of the free-particle
Hamiltonian. What are the corresponding eigenenergies of these states?
(b) Show that this finding is consistent with the free-particle eigenproperties
(10.56).
(c) What is the order of degeneracy of these eigenenergies? [Hint: For part (b),
consult Table 10.1.]

10.73 (In the following, hats over operators are deleted .) The formulation for evaluating
the momentum, Pn, conjugate to a given coordinate, qn, for a system described
478 Chapter 10 Problems in Three Dimensions

by N curvilinear coordinates is based on the following three propernes r'! (i) the
differential volume element: dV = D dq, .. . dqN ; (ii) the commutator : [qn, Pnl =
in; (iii) Henniticity: (cp IPnl/J ) = (Pncpll/J ). The momentum operator conjugate to
qn is then given by

. I dyD
Pn = -In-- ---
yD dq.,

(a) Obtain an expression for the radial momentum operator, Pp, appropriate to
cylindrical coordinates (Fig. 1.7).
(b) Show that for the operator you have found, [p, Pp 1= in.
(c) Show that the operator Po is Hermitian (with respect to square integrable func-
tions in a plane of constant z).
(d) Show that this operator obeys the following relation. (An expression for \1p 2 is
given in Appendix D.)

2=-n2[ 2 I]
PP \1p - 4
p2

(e) With your answer to part (d), write down a "corrected" Hamiltonian in cylin-
drical coordinates for a particle moving in the presence of a potential field,
V(p , lj» .
(I) Show that your answer to part (e) gives the correct Schrodinger equation for a
free particle in cylindrical coordinates (lO.75a).
(g) Show that your answer to part (e) reduces to the classical Hamiltonian in the
classical limit.

Answers (partial)
. I - - = -In (I- + -
(a) PP = -In--
pl /2
l
d p /2
dp
.
2p
d )
dp

(b) H = 2~ [ + ;:(1/2)2] +
p p2 Ptp2 - V(plj»

(c) Setting n= 0 returns the classical Hamiltonian .

10.74 What are the eigenfunctions, good quantum numbers, and eigenenergy degeneracies
for the following configurations?
(a) A free particle moving in three dimensions in (i) Cartesian, (ii) spherical, and
(iii) cylindrical coordinates.
(b) A particle confined to a (i) cubical box, (ii) spherical box, and (iii) cylindrical
box.
(c) In each case, state which type of degeneracy is described. (See Section 8.5.)
For infinite degeneracies discuss the nature of these infinities (i.e., continuous
or discrete).

21W. Pauli, "Die Allgeme inen Prinzipien der Wellenmechanik ," Hand. d. Physik XXIV, 83-272
(1933).
Problems 479

10.75 An electron is free to move in the domain exterior to a rigid , impenetrable, smooth-
walled, charge-neutral sphere of radius a.
(a) What are the eigenfunctions and eigenenergies of this system ?
(b) Write down the normalization condition for these wavefunctions. (Hint: See
Problem 10.60.)
10.76 An electron is confined to the interior domain of a cylindrical cavity of radius a and
length L. Photons at 3.0 eV are incident on the cavity and raise the electron from
the ground state to the first excited state. What is the radius a (in A) of the cylinder,
given that L = 4.7 A?

10.77 What is the value of {exp i2rr [Sin (t(7{~) ~~)]} Yt.m ?
CHAPTER

11 Elements of Matrix Mechanics.

Spin Wavefunctions

11.1 Basis and Representations

11.2 Elementary Matrix Properties
11.3 Unitary and Similarity Transformations in Quantum Mechanics
11.4 The Energy Representation
11.5 Angular Momentum Matrices
11.6 The Pauli Spin Matrices
11.7 Free-Particle Wavefunctions, Including Spin
11.8 The Magnetic Moment ofan Electron
11.9 Precession ofan Electron in a Magnetic Field
11.10 The Addition of Two Spins
11.11 The Density Matrix
11.12 Other "Pictures" in Quantum Mechanics
11.13 Polarization States. EPR Revisited
11.14 The Transfer Matrix

In this chapter some mathematical formalism is developed which is necessary for

a more complete description ofspin angular momentum. This formalism involves
the theory ofrepresentations described briefly in Chapter 7 and matrix mechanics.
Spin angular momentum operators are cast in the form ofthe Pauli spin matrices .
The spinor wavefunction of a propagating spinning electron is constructed and
used in problems involving the Stern-Gerlach apparatus. Examples involving the
precession ofan electron in a magnetic field and magnetic resonance are included
as well as a prescription for adding spins. The coupled spin states so obtained are
used extensively in the following chapter in conjunction with the Pauli principle in
some basic atomic and molecular physics problems. The density matrix relevant
to mixed states is introduced. Descriptions are included of the Heisenberg and
interaction pictures in quantum mechanics. The chapter continues with a review of
polarization states and a description of Bell 's theorem in relation to present-day
experiments concerning the notion of hidden variables. The chapter concludes
with analysis of the transfer matrix method important to the study of particle
transport through a periodic potential.

480
 11 .1 Basis and Representations 481

11.1 • BASIS AND REPRESENTATIONS

Matrix Mechanics
At very nearly the same time that Schrodinger introduced his wavefunction de-
velopment of quantum mechanics, an alternative but equivalent description of
the same physics was formulated. It is known as matrix mechanics and is due
to W. Heisenberg, M. Born, and P. Jordan .
We have already encountered the concept of representations in quantum me-
chanics in Section 7.4. There we noted that in the" A representation," states are
referred to a basis comprised of the eigenfunctions of Ii. What we will now find
is that within any such representation it is always possible to express operators as
matrices and wavefunctions as column vectors. Operator e9uations become ma-
trix equations . For example, an equation of the form 1/1 = F1/I' may be rewritten
as a matrix equation with the wavefunctions 1/1 and 1/1' written as column vectors
and the operator F written as a square matrix.

Basis
Previously we have found that wavefunctions related to a given quantum mechan-
ical problem must satisfy certain criteria. Examples include

Configuration Wavefunction Criteria

(a) Particle in a box 1{r(0) = 1{r(a) = 0, f 11{r 12 dx < 00

(b) One-dimensional harmonic

oscillator

(c) Particle in a central potential

v = VCr)

For a given problem each such set of conditions implies a related space of
functions . Consider a specific problem. Let the space of functions relevant to that
configuration be called Sj. Let the set of functions

~= {<PI, <p2 , . • . } (11.1)

be a basis of Sj. For instance, for a particle in a one-dimensional box, these func-
tions are
482 Chapter 11 Elementsof Matrix Mechanics. Spin Wavefunctions

For the hydrogen atom, they are

whereas for a free particle (in spherical coordinates), they are

Inasmuch as s.B is a basis of Sj, any function 1fr in Sj may be expanded in terms
of the basis functions lfJn :

(11.2)

or, equivalently,

11fr} = L IlfJn}(lfJnl1fr} (11.3)

n

The coefficients of expansion, an, represent 1fr in the representation where s.Bis the
basis. These coefficients are projections of 1fr onto the basis vectors (see Fig. 4.6).
The equivalence of {an} to the state function 1fr is akin to the equivalence between
a three-dimensional vector A and its components (Ax, A y , A z) .
If {an} is equivalent to 1fr, one should be able to rewrite equations involving
1fr as equations involving only {an}. Consider the typical quantum mechanical
equation, where F is an arbitrary operator:

1fr = F1fr'
(11.4)
11fr} = FI1fr'}
Expanding the right-hand side of the latter equation in accordance with (11.3) and
multiplying from the left with (lfJq I gives

(lfJql1fr) = L(lfJqIFllfJn)(lfJnl1fr'} (11.5)

n

or, equivalently,

(11.6)

where

(11.7)

is the matrix representation ofthe operator F in the basis s.B. The term Fqn is also
called the matrix element connecting lfJq to lfJn. Equation (11.6), involving only the
expansion coefficients {aq}, {an'}, and the matrix elements {Fqn }, is equivalent to
11.1 Basis and Representations 483

(11.4) involving the wavefunctions 1{r, 1{r' and the operator F. Equation (11.6) is
called a matrix equation. It may be written in the form

a2 ) = (Fll
a1 F21 F12
F22 ) (a<)
a2 (11.8)
( ... ..
. ...
In this equation the wavefunction 1{r is represented by the column vector on the
left, and 1{r' is represented by the column vector on the right. 1

(11.9)

The operator F is represented by the matrix Fqn •

(11.10)

The infinite dimensionality of these matrix equations is a consequence of the

infinite dimensionality of Hilbert space. Finite matrix equations are relevant to
vector spaces of finite dimension .

Diagonalization of an Operator
Let the orthogonal basis 23 be comprised of the eigenfunctions of a Hermitian
operator G:

GCPn = 9nCPn (11.11)

The matrix elements of Gare

(CPqIGICPn) = 9n(CPqICPn} = 9n8qn
Gqn = 9n8qn

G=
o 920
91
0 0
0
0
93
"'.J . (11.12)
(

IThe arrow in Ihese identifications denotes "is represented by."

484 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Thus the matrix of an operator in a basis of the eigenfunctions of that operator

is diagonal. The column vector representations of the eigenfunctions f{Jn are the
coefficients {Q q (n)} in the expansion

If{Jn} = l::>q (n) If{Jq}

q

Multiplying from the left by (f{Jpl gives

"(n)8 (n)
8pn = L.,.-Q q pq = Qp
q (11.13)
Qp
(n)
= 8pn
Thus the matrix representation of the eigenvector f{Jn is a column vector with a
single nonzero unit entry in the nth slot

1 0
0 1
("'(I») =
f{J\ -+ Q2:
(I) 0
o '
0
f{J2 -+
(""")
(2)
Q2; =
0
o '
0

0 0 (11.14)
0 0
1 0
f{J3 -+ 0 f{J4 -+
1
0 0
0 0

The eigenvalue equation (11.11) can be written

L(f{JpIGIf{Jq}(f{Jqlf{Jn} = 9n(f{J plf{Jn}

q

"
L.,.- G pqq
Q (n)
-- 9np
Q (n)

For Q(3) it appears as

9\ 0 0 0 o 0
o 92 0 0 o 0
o 0 93 0 1 1
o = 93 0 (11.15)
o 0
11.1 Basis and Representations 485

The "length" (squared) of a vector 1/1 is given by [recall (4.30)]

111/111 2 = {1/I11/I} = L{1/IIcpq} {cpq 11/I} (11.16)

q

The lengths of the orthonormal basis vectors {CPn} are

2
IICPnl1 = L la/
q
n)1 2
= 1 (11.17)

In matrix representation,

IIcp411 2 = {CP4Icp4 } -+ 900100 < 0o =I

o
1
o
o

Suppose that Gis known to be diagonal in the basis {CPn}. Then

(11.18)

or, equivalently,

Multiplying from the left with the sum L q IfI!q} gives

L Icpq} {CPqIGICPn} = 9n L IfI!q}{CPqICPn } (11.19)
q q

Recognizing the sum over CPq products to be the unity operator i (Problem I 1.1)
allows this latter equation to be rewritten as

which in tum implies that

(11.20)

Thus we find that if Gis diagonal in a basis ~ then 23 is compri sed of the eigen-
vectors of G. One then notes the following important observation. The problem
of finding the eigenvalues of an operator is equivalent to finding a basis which
diagonalizes that operator.
486 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Complete Sets of Commuting Operators

t
Suppose that .4, B, and are a "complete" set of three commuting operators. Let
~ = {<PI, <P2, •• • } be as set of simultaneous eigenstates of these three operators .
t
Then with respect to this basis, .4, B, and are all diagonal and one speaks of
t
"working in a representation in which .4, B, and are diagonal." For example,
for a free particle moving in 3-space, the representation in which if, £2, and i;
are diagonal contains the basis (10.56), while the representation in which Px, py,
and pz are diagonal contains the basis (10.11).

The Continuous Case

In some cases the indices of a matrix range over a continuum of values. Such, for
example, is the Hamiltonian matrix for a free particle in the basis (10.11). In one
dimension this basis is comprised of the states (3.24), and the Hamiltonian matrix
assumes the continuous form

Summing over the index of a continuous matrix is equivalent to integration . For

example,

The matrix representation of a quantum mechan ical equation involving a wave-

function 1/1 is the corresponding equation for the projection coefficients of 1/1 into
the basis ~. If this set of coefficients forms a discrete set, then equations are
of the form (11.6), involving summations over a discrete index. For the contin-
uous case, these sums become integrals . In Section 7.4 we considered the case
of the simple harmonic oscillator in "momentum space." Since the eigenstates of
the momentum operator form a continuous set, the Schrodinger equation for the
projection coefficients {b(k)} becomes the integral equation (7.81). Owing to the
simple form of the harmonic oscillator potential, this in tum is reducible to the
differential form

(7.84)

Thus the "matrix" form of the Schrodinger equation in the momentum representa-
tion remains a simple differential equation . Its argument is the single component
b(k) of the "column vector" {b(k)} .2

2For further remark s on the xand p representation s, see Appendix A.

Problems 487

PROBLEMS

11.1 Let (ipn} be a complete orthonormal basis of a Hilbert space, n. Show that the identity
operation i has the representation

in n. (This is sometimes calIed the spectral resolution of unity) .

Answer
Forming the matrix elements of i , as given above, gives
lpq = (ipplilcpq) = ~)ipplipn)(ipnlipq)
n

These are the matrix elements of the identity operator. This is a square matrix with
unit entries along the diagonal.

(Note : As described in Chapter 4, in order that an operator be a valid quantum me-

chanical representation of an observable, it must be Hermitian. To ensure further
consistency of the theory, one also demands that the eigenstates of the operator com-
prise a complete set.3)

11.2 Show that if 1/f = 0, then an = 0, for alI n , where 1/f = Ln anipn and (ipn} is an
orthogonal sequence.

11.3 What is the matrix representation of the operator fix in the momentum representa-
tion?

11.4 Show that the diagonal elements of b == a/ax , in n\ (4.30), in any basis are purely
imaginary.

11.5 Determine the wavefunctions b(k) in the momentum representation for a particle of
mass m in a homogeneous force field F = (FO, 0, 0). (Compare with Problem 7.65.)

Answer
The Hamiltonian is

, p2 . a
H=--IFo-
2m ak
3While completeness of eigenvectors is ensured for Hermitian operators in finite-dimensional spaces,
this association is not guaranteed in infinite-dimensional spaces. For further discussion, see P. T.
Matthews, Introduction to QuantumMechanics, 2d ed., McGraw-Hill , New York, 1968.
488 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

and the time-independent Schrodinger equation appears as

2 2
8b
- iFO-+ (1i- -k - E ) b=O
8k 2m

which has the solutions

2
bE(k) = - -Ie x p - [ik (E - -1i2k- ) ]
J2:rrFo FO 6m

These solutions obey the normalization

L: bE*(k)bE'(k)dk = o(E' - E)

11.2 • ELEMENTARY MATRIX PROPERTIES

The following are a series of definitions and properties of matrices and operators
relevant to the theory of matrix mechanics.

The Product of Two Matrices

(AB)nq = L AnpB pq (11.21)

p

As an example of matrix multiplication, consider the product of the two (finite)

2 x 2 matrices:

The Product of Two Wavefunctions

(11.22)
n n

The Inverse of A The inverse of A is labeled A-I . It has the property that
A-I A = AA- I =j (11.23)

If Ac is a matrix composed of the cofactors of the elements of A, then

~ I 1;::
A- = - - A (11.23a)
D(A) c
11.2 Elementary Matrix Properties 489

where D(A) is the determinant of A. For example,

G~) -1 = 4 ~ 2i (_~ -~)

The Transpose of A The transpose of A is written A.
The matrix elements of A
are obtained by "reflecting" the elements A nq through the major diagonal of the
matrix of A.

(A)nq = A qn (11.24)

A Is Symmetric or Antisymmetric If A is symmetric, then

A=A (11.25)

If Ais antisymmetric, then

A A

A =-A (11.26)

The Trace of A The trace of Ais the sum over its diagonal elements. It is written

TrA = LA
q
qq (11.27)

The Hermitian Adjoint of A This operator is written At. To construct At, one
first forms the complex conjugate of A and then transposes.

At =A* (11.28)

Matrix elements of At are given by

(11.29)

or, more explicitly,

A Is Hermitian If At = A, then A is Hermitian or, equivalently, if

(A t)nq = A nq

With (11.29), this definition becomes

(A qn )* = A nq (11.30)

or
490 Chapter 11 Elements of Matr ix Mechanics. Spin Wavefunctions

TABLE 11.1 Matrix properties

Matrix Definition Matrix Elements

Symmetric A=A A pq = A qp
Antisymmetric A=-A A pp = 0; A pq = -A qp
Orthogonal A =A- I (AA)pq = 8pq
Real A =A* A pq = A pq*
Pure imag inary A =-A* A pq = iB pq ; B pq real
Hermitian A = At A pq = A qp*
Anti-Hermitian A = -At A pq = -A qp*
Unitary A = (At)-I (AtA)pq =8 pq
Singular detA = 0

fJ Is Unitary Ifthe Hermitian adjoint [;t of an operator [; is equal to [;-1 , the

inverse of [;, i.e.,
[;t = [;-1 (11.31)

then [; is said to be unitary . The matrix elements of [; satisfy the relations

(u t)nq = (U-I)nq

(Uqn)* = (U-I)nq (11.32)

{j* = [;-1

If [; is unitary, then
[;[;t = j
~t ~

(UU )nq = Onq (11.33)

L Unp(Uqp)* = Onq
p

These matrix properties are summarized in Table 11.1.

PROBLEMS

11.6 Rotation of the xy axes about a fixed z axis through the angle ¢I changes the com -
ponents (x, y) of the vector r to (x' , y') (see Fig . 11.1). These new components are
related to the original components through the rotation matrix R(¢I) in the follow-
ing way:

,_(x') _(
r -
y
, - cos¢1
.
-sm¢1
sin¢l)
COS¢I
(x) =
y
R(¢ )r
I
Problems 491

FIGURE 11.1 The rotation operator

sin cP)
coscP

is unitary,

Rt R = I
and obeys the group property

(See Problem 11.6.)

A second rotation through <!>2 gives

r" = R(cP2)r' = R(<!>2)R(cPl)r

(a) Show that the rotation matrix Rhas the "group property "
R(cPl)R(<!>2) = R(cPl + cP2)

(b) Show that

[R(cPl), R(cP2)] =0
(c) Show that R is an orthogonal matrix (see Table 11.1).
(d) In 3-space, rotation about the z axis is effected through the matrix

R= ( -
COS cP
sin cP
sin cP
cos cP
0)
0
o 0 I

Show that the eigenvalues of R have unit magnitude. (The angle A in Fig. 11.1
remains fixed under rotation.)
492 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.7 Fill in the missing components of the matrix t: which make t Hermitian .

A (~~~~)
c= _ _ 4 9
- 6

11.8 Construct a 2 x 2 unitary matrix which has at least two imaginary elements .

11.9 What is the inverse of the matrix

Answer
Calculation shows that A has no inverse. Under such circumstances A is said to be
singular.

11.10 Show that if U is unitary, then the eigenvalues an of U are of unit magnitude .
Answer

U ifin = anifin

= (ifinl/ ifin) =
(U ifin IU ifin) (ifinlifin) = an*an (ifinlifin)
2
an*an = lanl = I

11 .3 • UNITARY AND SIMILARITY TRANSFORMATIONS

IN QUANTUM MECHANICS

The significance of unitary operators in quantum mechanics is due to the follow-

ing. We are already aware of the fact that a given Hilbert space has many bases.
This is similar to the fact that 3-space is spanned by one of a continuum of triad
basis vectors. One can obtain a new orthogonal triad basis in 3-space through a
rotation of axes about the origin . In the new basis a vector V has components Vx' ,
v, ', and Vz'. The length of V remains the same (V . V = V' • V'). Furthermore,
the angle between any two vectors remains fixed (V . F = V' . F/). The final
orientation of the new Cartesian frame with respect to the old may be obtained by
a single rotation about a fixed axis (through the origin). The eigenvectors of the
rotation matrix lie along this axis. Any vector along this axis remains fixed during
the rotation. It follows that the eigenvalues of the rotation matrix are all of unit
magnitude.
The related transformation from one basis to another basis in Hilbert space is
a unitary transformation. It has all the properties listed above that a rigid rotation
in 3-space has. These properties are as follows (for the most part, proofs are left
to the problems) .
11.3 Unitary and Similarity Transformations in Quantum Mechanics 493

Transformation of Basis
Let the sequence Un} denote a new basis. These are related to the old basis {q:>n}
through the unitary transformation U,

lin) = L Iq:>p) (q:>p lin) = L(Unp)*Iq:>p) (11.34)

p p

(11.35)

These matrix elements are the projections of old basis vectors into the new ones .
(The fact that U is unitary is established in Problem lUI.)

Transformation of the State Vector

Let us consider how the components of an arbitrary state vector 1/J transform to
components in the basis {In} . In the basis Un}, these components (elements of a
column vector) are

1/Jn' = (fn 11/J) (11.36)

Taking the complex conjugate of (11.34) gives

(11.37)

Substituting into (11.36) we obtain

1/Jn' =L Unp(q:>pl1/J ) =L Unp1/Jp (11.38)

p p

This is the matrix representation of the equation

11/J/) = UI1/J) (11.39)

This equation tells us how an arbitrary state vector transforms under a change of
basis .
If Iq:» and 11/J) are two arbitrary vectors in Hilbert space, then under a transfor-
mation of basis (U), these vectors transform to Iq:>/) and 11/J/) according to (11.39) .

Iq:>/) = Ulq:»
11/J/) = UI1/J)

Under such transformation the inner product (1/JIq:» is preserved.

Setting 11/J) = Iq:» gives (q:>/Iq:>/) = (q:>Iq:» . Thus a unitary transformation preserves
the length of vectors and the angle between vectors .
:1-94 Chapter 11 Elements of Matr ix Mechanics. Spin Wavefunctions

The Unitary-SimilarityTransformation
Next we consider the manner in which operators transform under a change of
basis. A typical quantum mechanical equation appears as

ftl<p} = It}
In the new basis, the two state vectors transform according to (11.39). Multiplying
these equations from the left with (;-1 gives
l<p) = [;-ll<p/}
It} = (;-llt/)

Substituting these forms into our typical equation above gives

ft(;-II<p/} = V-Ilt/}

Multiplying from the left with (;, we obtain the result

ft/I<p/) = It/}
where
ft/ = (; ft[;-l (11.40)
This transformation preserves the form of our typical equation . As a special case
(<p = t) we see thatthe eigenvalue operation for ft is preserved under such a
transformation. Equation (11.40), which describes how an operator transforms
under a change of basis, is called a unitary-similarity transformation . The more
general class of transformations, A ---+ At = SAS- I , where S is not necessar-
ily unitary, are called similarity transformations. However, of these, the unitary-
similarity transformations are more relevant to quantum mechanics .

Invarianceof Eigenvalues
Since the eigenvalues of an operator corresponding to an observable are physi-
cally measurable quantities , one does not expect these values to be affected by a
transformation of basis in Hilbert space. The eigenenergies of a harmonic oscilla-
i)
tor are lUvo(n + in all representations. In a similar vein, the eigenvalues of such
Hermitian operators must be real. It follows that (1) the eigenvalues of a Hermi-
tian operator are preserved under a unitary-similarity transformation, and (2) the
Hermiticity of an operator is preserved under a unitary-similarity transformation.

PROBLEMS

11.11 Show that D, with matrix elements

Unp = (in l<pp )

is unitary. The sequences Un I and (II'p I are complete and orthonormal.

Problems 495

Answer
We must establish the property (11.32) for U.
A_ I At
(U )np = (U )np = (U pn)
*

Equivalently, we must show that

Iqp = 8qp = L uqn(u-I)np =L Uqn(Upn)*

n n

= L {fq Irpn )(rpn lip }

n

11.12 Show that the inner product, (1{r lrp) , is preserved under a unitary transformation.

Answer

(1{r' lrp') = (U 1{r !U rp) = (1{r lu t U rp)

= (1{r IU - I U rp) = (1{r lrp)

11.13 What is the matrix representation of the equation

in the basis Un I? Write this equation explicitly, depicting elements of column and
square matrices.

11.14 The matrix elements of F in the basis {rpn I are

In the basis Un} they are

Show that
, AAA_I AAA_ I
Fnq =(UFU )nq= (rpnIUFU Irpq)

Answer

Fn/ = {fnIFllq } = L L { fnlrpr}(rprlfrlrpp}(rppllq }

r p

= L L UnrFrp(Uqp)* =L L Unr Frp (u t)pq

r p r p

= LLUnrFrp(U-I)pq = (UfrU-I) nq
r p
496 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.15 Let A' = VAV- I , where V is unitary. Show that this transformation may be
rewritten A' = t-IAt, where i is unitary. (Note: It follows that both VAV-I
and V-I AV represent unitary-similarity transformat ions. For demonstrating cer-
tain properties of the unitary-similarity transformation , it may prove more conve-
nient to work with the form V-I AV.)
11.16 (a) Show that A and V-I AV have the same eigenvalues. Must V be unitary for
this to be true?
(b) If the eigenvectors of A are (CPn) , what are the eigenvectors of V-I AV?
11.17 If V is unitary and A is Hermitian, then show that VAV- I is also Hermitian. That
is, show that the Hermitian quality of an operator is preserved under a unitary-
similarity transformation .
11.18 Show that the form of the operator equation

is preserved under a similarity transformation .

11.19 Consider the following decomposition of an arbitrary unitary operator V:
, V + vt V - vt , ,
U = - - 2 - + i -2-i- == A + iB

(a) Show that A and B are Hermitian .

(b) Show that [A, BJ = [A, VJ = [B, VJ = o.
(c) From part (b) we may conclude that A, B, and V have common eigenfunctions.
Call them lab). Use these eigenstates to show that the eigenvalues of V have
unit magnitude.
11.20 (a) Show that diagonal matrices commute .
(b) Let Ajk = ajOjb Bjl = blojl , and C nm = cnonm be three matrices . What are
the components of ABC?
(c) Again consider the diagonal matrix Ajk = aAk . What is the matrix represen-
tation of exp A?What is the matrix representation of sin A?
11.21 If A, B, and C are three n x n square matrices, show that
Tr(ABC) = Tr(CAB) = Tr(BCA)
11.22 Let A and Bbe two n x n square matrices. Employ the following property of deter-
minants,"

det AB = det Adet B

to show that

det AB = det BA
4It is assumed that the student is familiar with the concept of a determinant. However, a definition of
determin ants may be found in Section 12.5. For further discuss ion, see G. Birkoff and S. Macl.ane, A
Surv ey ofModern Algebra, Macmillan, New York, 1953.
Problems 497

11.23 A property of a matrix A which remains the same under a unitary transformation
A --+ U AU- I is called an invariant. Show that the trace of A is an invariant. That
is, show that

TrA = TrU AU- I

In your proof, establish that

TrA = Lan
n

where an are the eigenvalues of A.

Answer (partial)
Let Udiagonalize Aso that the diagonal matrix
U AU- I = A'
is comprised of the eigenvalues of A. With Problem 11.21, we have
'"
Lan = TrA" = TrUAU
A" _ I ' A_l A '
= TrAU U = TrA

11.24 Let A be an n x n square matrix with eigenvalues ai, az , .. . . o«. Show that

detA = alaZ " · an

(Hint: Let U diagonalize Aand refer to Problem 11.22.)

Note: This problem establishes that det A is another invariant property of A.
Together with the trace, these two fundamental properties appear as

TrA = La;
i

detA = na;
i

In general , an n x n matrix has n invariants, two of which are the trace and de-
terminant. These n invariants are the coefficients of the characteristic equation for
the eigenvalues of A [i.e., det(A - i a) = OJ, which itself is invariant. Namely ,
det(A - i a) = det[U(A - i a)U-Ij = det(U AU- I - i a) .
11.25 Let A be a Hermitian n x n matrix . Let the column vectors of the n x n matrix S be
comprised of the orthonormalized eigenvectors of A.
(a) Show that Sis unitary.
(b) Show that S-IAS is a diagonal matrix comprised of the eigenvalues of A.
(Note: This establishes that a Hermitian matrix is always diagonalizable by a
unitary-similarity transformation.)

11.26 Again, let Abe a Hermitian n x n matrix . However, now let the column vectors of the
n x n matrix i be comprised of the unnormalized, but still orthogonal eigenvectors
of A.
(a) Is i unitary ?
498 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

(b) Is i :' At = A' diagonal ? If so, what are the elements of A'?
(c) Is the inner product between two n-dimen sional column vectors preserved un-
der this transformation ?
Answers
(a) We note that although t tt is diagonal , it is not the unit operator, so that i is
not unitary.
(b) Let the eigenvector equation for A be written

Aln ) =anln )
It follows that the column vectors of At are aIll ), azI2), .. . ,an In ). This ma-
trix may be rewritten

where the diagonal matrix A' is comprised of the eigenvalues of A. We then

have

t-IAt = A'
(Note: Although the similarity transformation described in this problem diago-
nalizes A and yields a diagonal matrix comprised of the eigenvalues of A, it is
not a unitary-similarity transformation and therefore is not relevant to quantum
mechanics. Changes in representations in quantum mechanics must preserve
the inner product between state vectors, which in tum ensures preservation of
the Hermiticity of operators . These invariances are maintained in a unitary-
similarity transformation .)

11.27 In the Schrodinger description of quantum mechanics, the wavefunction evolves in

time according to the equation (3.70)

(a) Show that the operator

U =exp
A (-ifit)
-fi-

is unitary. (Hint: Use the property fit = H.)

(b) Having shown that 1!Jr(t» ) = O'I!Jr(O» ), show that the normalization of !Jr,
(!Jr (t )I!Jr (t») , is constant.
[Note: In this description the state of the system is represented by a vector 1!Jr(r, t»),
which migrates in Hilbert space according to the unitary transformat ion above. This
behavior is opposed to that of eigenvectors correspond ing to observables (e.g., i.z,
iI, Px , etc.). These are fixed in the Hilbert space.]
11.28 Show that if A is Hermitian , then
0' = (A + ii)(A + ;/)-1
is unitary. [Hint: Multiply from the right with (A + i h]
 11.4 The Energy Representation 499

11.29 Show that if the unitary operator (; does not have the eigenvalue I , then

is Hermitian.

11.30 Consider two Hermitian operators Aand 8, which satisfy the commutation relation ,
[,.1,8] = iii. Suppose a system is in an eigenstate la) of ,.1. What can be said of the
probability distribution relating to 8 (i.e., l(alb )1 2 )? Does your argument apply to
the observables rf> and L z? (Recall Problem 9.15.)

11.4 • THE ENERGY REPRESENTATION

One-Dimensional Box
In the energy representation, the Hamiltonian is diagonal. This representation in-
cludes a basis comprised of the eigenfunctions of the Hamiltonian. For a particle
in a one-dimensional box, the basis in which iI is diagonal is

~-a
. n x . 2n x . 3n x }
~=
{ sm-,sm--,sm--,
a a a
... (11.41)

The Hamiltonian matrix in this representation is

4 0
9
16
H=El (11.42)

0 n2

Simple Harmonic Oscillator

For the one-dimensional harmonic oscillator, the basis that diagonalizes iI is [re-
call (7.59)]

~= e-~2/2{AoJ(o(~), AIJ(I(~), A2J(2(~) , ...}

== {I0}, II} , 12} , . . . } (11.43)

~2 == f32 x2, f32 == mwo
It
The nth-order Hermite polynomial is written J(n(~) . The Hamiltonian matrix in
this representation is
500 Chapter 11 Elements of Matrix Mechanics . Spin Wavefunctions

1/2
3/2 o
(11.44)
o (2n + 1)/2

Position and Momentum Operators

Let us calculate the matrix representation of the position operator for the har- x
monic oscillator in the energy representation. Recalling (7.61) and with k and n
representing nonnegative integers, we have

(n lx lk) = ; (n [a + atlk ) (11.45)

v2f3

= ~ f3 [k I/ 28n,k _ 1 + (k + 1)1 /2 8n.k+ Il

This gives the matrix

o v'l 0 0 0
v'lO v'20 0
A I o v'2 0 J3 0
X=--
o 0 J3 0 A (11.46)
v'2f3
o 0 0 A 0

For the momentum operator p, we find that

mW() 1/2 1/2

= (k + 1) (11.47)
A

(n lp lk) M [k 8n k-I - 8n k+ Il
v2if3 ' ,

which gives the matrix

o v'l o 0 o
-v'l o v'2 0 o
A mWQ
o -v'2 o J3 o (11.48)
p = v'2if3 o
o -J3 0 A

Creation and Annihilation Operators

For the creation and annihilation operators we have [recall (7.61)]
11.4 The Energy Representation 501

ank = (n la lk) = k l / 2(nlk - I) = k l / 28 n,k_1

(11.49)
ank t = (n lat lk) = (k + 1)1 /2(nlk + I) = (k + 1)1/28n .k+1
which gives the matrices

0 0 0 0 0
~ 0

~ (~
0 0
~ 0 0 0 0

a
0 .j2
0 0
0
J3
0
0
) at = 0
0
.j2
0
0
J3
0
0
0
0

(11.50)

Let us check that these matrix operators promote and demote according to
(7.61) . The eigenfunctions {In )} for the harmoni c oscillator Hamiltonian are col-
umn vector s with the only nonzero entry in the (n + I)st slot.

I 0 0
0 I 0
0 0 I
0 0 0
10) = 0 II) = 0 12) = 0 (11.51)
0 0 0
0 0 0

The time-dependent eigenst ates of iI appear as

I
o
o
1/Jo(x, t ) = e-iHt / IiIO) = e- iwot / 210) = e- iwo t / 2 0
o
o

o o
I o
o I
1/J1(x ,t)= e-i3wot /2 0 1/J2(X, t) = e-i5wot/2 0 (1 1.52)
o o
o o

a
Consider the operators a [2) and t I2).
These equations very simply illustrate the promotion and demotion properties of
the at and a operators.

The Number Operator

In addition to the Hamiltonian (7.27)

the number operator (7.28)

is also diagonal in the energy representation.

0 0 0 0 0 v'l 0 0 0
v'l 0 0 0 0 0 .J2 0 0
N= 0 .J2 0 0 0 0 0 ./3 0
0 0 ./3 0 0 0 0 0 J4

0 (1 1.54)

2
0
3

0 n
Problems 503

The reader may readily check that the column vectors (In)}, as given by (11.51),
are eigenvectors of both fj as given by (11.44) and IV as given by (11.54), with
respective eigenvalues {hwo(n + !)} and {n}.

PROBLEMS

11.31 In Section 5.5 the importance of complete sets of commuting observables was dis-
cussed. The number of such variables (good quantum numbers, see Section 1.3) are
analogous to the number of canonical variables relevant to the description of a clas-
sical system. It is important in classical physics that the number of such variables
be preserved under a canonical transformation. In quantum mechanics it is equally
significant that the number of operators comprising complete sets of commuting
observables be preserved under a unitary transformation.
(a) Let such a set of compatible operators be ,.1,8, C, and b. Show that this com-
patibility is preserved under a unitary transformation.
(b) Let F not commute with any element in the set A, 8, C, b. Is this property
preserved under a unitary transformation?
11.32 Show that

where A is an n x n matrix and i is the identity matrix in n dimensions.

11.33 Show that

det(exp A) = exp(TrA)
where A is an n x n matrix .

Answer
This equality may be established in two independent ways. In the first method we
let D()") == det[exp()..A)]. Then with Problem 11.22 we obtain

dD
-=Dhm
d);
. {det[eXP(EA)]
E--+O E
-I}
In the limit that E goes to zero,

det[exp(EA)] = det(l + EA) = I + ETrA

where we have used the results of Problem 11.32 in establishing the second equality.
It follows that
dD '
d)" = DTrA

which has the solution

D()") = D(O) exp(ATrA)

But D(O) = I, hence

D(l) = det(exp A) = exp(TrA)

504 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

In the second way we first note that if {; diagonalizes A, it also diagonalizes eA .

Furthermore, det[{;(exp A){;-I] = det exp A (see Problem 11.23).
Now the diagonal matrix U(exp A)U- I has the explicit form

where (ai) are the eigenvalues of A . The determinant of this matrix is

det[U(expA){;-I] =e Q'eQ 2
••• = exp (~ai)

This is the value (in all representations) of the left-hand side of the equality to
be established. For the right-hand side we recall that the trace is independent of
representations (Problem 11.23) so that

exp(TrA) = exp(TrA') = exp ( ~ ai ) Q.E.D.

The matrix A' is the diagonal representation of A.

x
11.34 Use the matrix representations (11.46) and (11.48) for and p to obtain the matrix
representation for the commutator [x , p] for the harmonic oscillator in the energy
representation.
x
11.35 Calculate the matrix representations for 2 and p2 for the harmonic oscillator in the
energy representation.
11.36 Using the fact that any Hermitian matrix can be diagonalized by a unitary matrix,
show that two Hermitian matrices, Aand 8, can be diagonalized by the same unitary
transformation {; if and only if [A, 8] = O.
11.37 Consider the following equations :

(a) Show that 82 = B.

(b) Obtain explicit (2 x 2) matrices for A and 8.
Answer (partial)
(b) A=~(1
2 i -I
i)
11.5 • ANGULAR MOMENTUM MATRICES

The i Matrices
It was shown above that the matrix representation of an operator A in the basis
consisting of the eigenvectors of A, is diagonal. In Chapter 9 we found that the
11.5 Angular Momentum Matrices 505

eigenfunctions of the angular momentum operators L2 and i , are the spherical

harmonics yt(O, ¢). It follows that in the basis s.8 = {ytl , the matrices LTm .l'm'
and the matrices (L z)lm,l'm' are diagonal. That is,

LTm I'm' = (lmIL 21I'm ') =

1 12rr d cos e d¢(yt)* f2Yl' m' (11.55)
, / -I 0

= nh(l + 1)811,8mm,
(L z)lm,l'm' = (Im ILz II'm' ) = nm811'8mm' (11.56)

The manner in which these elements are displayed is as follows . The rows and
columns of a given matrix are ordered so that for every value of I, mt runs from
(-I , . .. ,+1). For each of these mi values, I is fixed. The diagonal matrix for f2
appears as

1/ -+ 01 I 122222
m' -+ o I 0 -I 2 I 0 -1 -2
o
__ - __ - - -
0 0
- - - _ ..
0 ---I-
0 _

1 1 lz 0 0 i
I ol o l o 2 ol 0 (11.57)
____~ ______~ !__ _ l_~ ~ ~ l _
22 160000
2 1 i O 6 0 0 0
200 0 i O 0 600
2 -I ! 0 0 0 6 0
2 -2
______ __ _
i 0 0 0 0 6
L _

In this same scheme, we obtain the following diagonal matrix for L;

o
o o
-I
2
(11.58)
o o
-1
-2
506 Chapter 11 Elements of Matrix Mechan ics. Spin Wavefunctions

To obtain the matrices for i., and L, in the representation in which [2 and i,
are diagonal, we first construct the matrices for the ladder operators L+ and L_
(see Section 9.2). Since L_ = L+t, one merel y needs to calculate the L+ matrix
and then , from its Hermitian adjoint, find L_. Once these matrices are known, i.,
and i., are obtained from

(11.59)

Using the relation (see Table 9.4 )

L_yt = [( I + m )(1 - m + 1)]1/2/i yt - 1 (11.60)

with a similar expression for L+ yt . One obtains

(11.61)

and the matrices [exhibiting only the I ~ 2 terms )

0 0 0
0 0 0
0 h 0 0 0
0 .n 0
0 0 0 0 0
2 0 0 0 0
0 0 0 ./6 0 0 0
0 0 ./6 0 0
0 0 0 2 0
(11.62)
0 0 0
0 h 0
0 0 0 h 0
0 0 0
0 2 0 0 0
0 0 ./6 0 0
0 0 0 0 0 ./6 0
0 0 0 0 2
0 0 0 0 0

Adding and subtracting these two matri ces according to (11.59) give s (again ex-
hibiting only the I ~ 2 terms)
11.5 Angular Momentum Matri ces 507

0 0 0
0 -/2 0
0 -/2 0 -/2 0
Ii.
0 -/2 0
L, =-
2 0 2 0 0 0
2 0 ./6 0 0
0 0 0 ./6 0 ./6 0
0 0 ./6 0 2
0 0 0 2 0
(11.63 )
0 0 0
0 -/2 0
0 --/2 0 -/2 0
Ii.
0 --/2 0
L y =-
2i 0 2 0 0 0
-2 0 ./6 0 0
0 0 0 -./6 0 ./6 0
0 0 -./6 0 2
0 0 0 -2 0

Next we consider the matrix representation of the eigenvectors of f2 and i.;

These are column vectors whose elements are the coefficients of expansion of Yt
in the basis {YI' m' }.

l'
YIm = 'Z::
" ' 'Z::
" ' alm ,I'm ' Yl' m'
I' m'=- I' (11.64)

alm ,I'm ' = Oll'Omm '

The elements of these column vectors have zero entries for all values of l , m
except at l = l', m = m', where the entry is unity. For example, the representations
of the l = 1 eigenstates are [entries in these vectors follow the scheme of (11.57)]

0 0 0
1 0 0
0 1 0
1 O Y1-I ---+ (11.65)
YI ---+ 0 , YI ---+ 0 , 1
0 0 0
508 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

When the matrix for L2 operates on any of these three column vectors it gives
2li2 times the vector. When i; operates on them, it gives the respective values
(Ii , 0, -Ii) times the vector.

Sub L Matrices
One often speaks of the submatrices of L2 , i.; ... corresponding to a given value
of I. For example, the L2 matrix for I = 1 is

20 0)
L2 = li2
(
0 2 0
002

while the i.; i.; and i; matrices corresponding to I = I are

(0 1
~ JJ
Ii
i; = M 1 0
...;2 0 1

(11.66)

The i.; i.; and i; matrices corresponding to I = 2 are

0 0
' Ii (0 22 0 ./6 0
t., ="2 0 ./6 0 ./6
o 0 ./6 0
o 0 0 2 !J.
-n 0 0
'
i,
(0
~2 ~
Ii i2 0
i./6
-i./6
0
0
-i./6 o
0
° 1 , (11.67)
0 i./6 0 -~2
0 0 i2

~ (~
0 0 0
I 0 0
i; n
0
0 0
0 0
0 0
-1
0
]J
We may consider the eigenvectors corresponding to these matrices. These are
also subcomponents of the infinitely dimensional column vectors (11.64) . For
example, the eigenvectors of i; for the case I = 1 appear as
11.5 Angular Momentum Matrices 509

" -. = ~ ( -f) , (11.68)

,,' = ~ (f)'
In the representation where i; is the differential operator [recall (9.56)]

. ( Slll¢ae+cotecoS¢a¢
Lx=lh . a a)

the eigenvector ;x I corresponds to the linear combination of spherical harmonics

which was previously labeled X+ in (9.100) .

The j Matrices
In the preceding construction of the matrices for the angular momentum operators
Land L2 , it proved convenient to work from the yt (e, ¢) eigenstates relevant to
the coordinate representation. As we recall from Chapter 9, the more inclusive
angular momentum, j (which may represent L, S, or L + S), is defined in tenns
of the commutation relations (9.16) and the rule of length appropriate to vectors
(9.17).
The erocedur~ for obtaining the matrices for j and J2 (in a representation
where 1 2 and 1z are diagonal) parallels the construction above. In place of
(11.60), one writes the operationally identical equations (see Table 9.4)

J+ljm) = h[(j - m)(j + m + 1)]1 /21j, m + 1)

Lljm) = h[(j + m)(j - m + 1)]1 /21j, m - 1)

Thus the matrices found above for Land L2 are also valid for j and J2 . Such ma-
trices have the correct commutation properties and obey the Pythagorean length
rule (see Problem 11.41). However, while L matrices are restricted to integral I
values and are therefore of odd (21 + 1) dimension , j matrices also incorporate j
values that are half-odd integral. Such matrices are of even dimension .
Since there are 2j + 1 values of 1z for each value of i . the matrix of i, has
2j + 1 diagonal elements . For a given j value, the operators j and J2 are (2j +
1) x (2j + 1) square matrices and operate on column vectors 2j + 1 elements
long. That is, j and J2 operate on a (2j + I)-dimensional space.
The diagonal matrices for J2 and i, (for a given value of j) are simple to
construct. The first four (j = i, 1, ~ , 2) such pairs appear as
510 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

j -
- -
2''
1 ~23li
J =4 0
2
C ~) , ~
t, =
Ii
2 C ~l)
0

j2~2h2 G
~J
0 0
j = 1: 1 0
0
D' i '=hG
0

~ 15."' (~ ~). (~0

0 0 0 0
3 1 0

I)
, _ h 1 0
j --' j2 Jz - 2
- 2' 0 1 0 -1
0 0 0 0 0

~ (~
~J. ~ (~
0 0 0 0 0 0
1 0 0 1 0 0
j = 2: 2 0 1 0 0 0 0
j2 6h
0
0
0
0
1
0
i, h
0
0
0
0
-1
0 11
When j is an integer and J represents orbital angular momentum L, one
may transform from the Ijm) column vector representation to the coordinate
yt (e, ¢) representation. In this representation the ladder operators j ± appear as
[see (9.57)]

and eigenstates of J2 and Jz are the spherical harmonics. When j represents spin,
S, spherical harmonic eigenstates become inappropriate.
The Rotation Operator
A distinction between angular momenta corresponding to j integral or half-odd
integral is found in the rotation operator.l described in Section 10.2.

~ =exp
Rc/J .J)
(icf>
-li-

When R operates on a function !(r), it rotates r through the angle cf>. If rotation
is solely about the Z axis, cf> = e z¢ , then R becomes

~
Rq, =exp
(i¢jz)
T

5A more fundamental distinction involves the theory of group representations. For a concise, self-
contained discussion of this topic, see L. 1. Schiff, Quantum Mechanics, 3d ed., McGraw-Hill, New
York, 1968, Chapter 7.
Problems 511

Let Ijm} denote a common eigenstate of j2 and i; Then, in particular,

and

For the case that j, and therefore m, is half-odd integral, ei 2rrm = -1, and one
obtains the somewhat surprising result

(2j +1= even no.) R2rrljm} = -lljm}

That is, the eigenstates of j2 and i, corresponding to half-odd integral j values

change sign under complete rotation of axes. If, on the other hand, j is integral,
one obtains

(2j +1= odd no.)

In the coordinate representation, eigenstates for this case become spherical har-
monics . These functions return to their original values under complete rotation .
The smallest finite value j may assume is j = !.
This spin quantum value is
a profoundly important case and is developed in detail in the next section. Eigen-
states corresponding to j = ~ are called spinors . We will find (Problem 11.76),
in accord with the discussion above, that spinors change sign under complete ro-
tation of axes.

PROBLEMS

11.38 The state column vectors ~ , corresponding to the case I = 1, exist in a three-
dimensional vector space. Any element ~ of this space is a set of three numbers
of the form

Write the vector

as a linear combination of the vectors (~x 0 , ~x -1 , ~x 1), as given by (11.68).

Answer
512 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.39 Use the results of Problem (11.38) to answer the following question . A rigid rotator
with moment of inertia I is in the state

1/!(t) = _1_ (~) e- i Et / Ii , E=-

Ii?
v143 I

(a) What is the probability that measurement of Lx finds the value -Ii?
(b) What is the column vector representation of the time-dependent state of the
rotator after measurement finds the value Lx = -Ii?
11.40 What is the column vector representation of the angular momentum state

in the representation in which i; is diagonal?

11.41 Show that the I = 2 angular momentum matrices satisfy the relation

11.42 Show directly, by matrix multiplication, that (i'x2 ) in the state ~x -1 is 1i2 .
11.43 What are the matrix representations of f2, i.; L+, and L_ for the case I = 3?
11.44 What are the column eigenvectors of i , corresponding to I = 3? To what combina-
tions of yt functions do these correspond?

11.6 • THE PAULI SPIN MATRICES

Spin Operators
In Section 9.2 it was concluded that there were two classes of angular momentum.
These, we recall , stem from the fact that orbital angular momentum l values cover
only a subset of the spectrum of j values appropriate to j2 and i; The second
class includes angular momentum called spin . Spin, as described in Section 9.1,
is not related to the spatial coordinates of a particle as is orbital angular momen-
tum. A coordinate representation of spin wavefunctions does not exist. Spin is an
intrinsic or internal property. Other intrinsic properties of a particle are charge,
mass, dipole moment, and so forth . Values of such parameters comprise internal
degrees offreedom for a particle.
Spin angular momentum is denoted by the symbol S. The Cartesian compo-
nents of S, being angular momentum components, obey the commutation rules

(11.69)
11.6 The Pauli Spin Matrices 513

These are the fundamental relations from which all other properties of spin follow.
Similar to the introduction of J± and L± in Chapter 9, one may also introduce
ladder operators for spin.

(11.70)

Furthermore, since the eigenvalue equations (9.43) for J2 and i, were derived
from the fundamental angular momentum commutator relations (9.16), we may
conclude that a similar structure exists for the eigenvalues of S2 and s;
(11.71)

For a given value of s, the azimuth al quantum number ni, runs in integral steps
from -s to +s. The lowest value s can have is zero. Mesons are particles that
have zero spin." Photons have unit spin. For s = 1 there are three values of m s :
-1,0,1.

Spin Eigenstates
Electrons, protons, and neutrons have a spin of one-half. There are two values of
m, for s i.
= These are m, = +i,-i. Let us call the eigenstate corresponding
to (s = i, m, = i)a (also fl z) and the eigenstate corresponding to (s = m, = i,
-i) {3 (also (3z)· These eigenstates obey the eigenvalue equations

(11.72)

The raising and lowering operators have the property that?

S+ls , m s) = nJs(s + 1) - m s(m s + 1)ls, m, + 1)

(11.73)
S_ls, m s) = nJs(s + 1) - ms(m s - 1)ls, ni, - 1)

It follows that

(11.74)

Matrix Representation
We now wish to construct the matrix elements of s..s;
and Sz for s = in the i
fl , {3 basis. In this basis S2 and s,
are diagonal. The first two equations on the left
6More precisely, vector mesons have spin one; pseudoscala r mesons have spin zero.
7See Table 9.4.
514 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

in (11.74) appear explicitly as

(Sx + iSy)a = 0
(Sx - iSy)a = n/3
Adding and subtracting these two equations, respectively, gives
~ 1
s,« = "2 n/3
(11.75)
A i
s,« = "2n/3
In similar manner, addition and subtraction of the remaining two equations in
(11.74) gives

(11.76)

Combining these with the following equations from (11.72) ,

A n
s,« ="2a
(11.77)
~ n
Sz/3 = -"2/3

establishes all the six equations needed to calculate the matrix elements of Sx, s;
and s; For example, the elements (alSx la) is [using the first equation in (11.75)]
~ 1
(alSx la) = "2n(al/3) = 0

The vectors a and /3 are eigenvectors of a Hermitian operator (i.e., Sz) and are
therefore orthogonal. We also take them individually to be normalized. Continu-
ing in this way we find that

S
x
= ((al~x la) (al~x 1/3))
(/3I Sxla) (/3I Sxl/3)
(0 1)
= ~
2 1 0
(11.78)
~ _ in
Sy - 2
(0 -1)
1 0'
~ _~ (1 0)
s, - 2 0 -1

The matrix representations of the eigenvectors a and /3 are the two-dimensional

column vectors

/3 = G) (11.79)
Problems 515

In this representation the orthonormal relations between a and f3 appear as

(alf3) =O(n =0+0=0

(f3la) =[] G) = 0 + 0 = 0
(ala) = ° G) = 1+0 = I
(11.80)

(f3If3) = [ ] (~) = 0+ I = I

The operator
2~
u= -8
n (11.81)

is called the Pauli spin operator. The matrix representation of its Cartesian com-
ponents (in the basis that diagonalizes S2 and Sz) is

~
ax =
(0 °I) '
I
~
ay =
.(0I -I)
I 0 ' ~
az =
(I 0)
0 -I (11.82)

These are called the Pauli spin matrices. They will be brought into play shortly
when we consider the quantum mechanical motion of a spinning electron in a
magnetic field.

PROBLEMS

11.45 (a) For spin corresponding to s = ! ' show that the eigenvectors of Sx and s, are

ax = ~G), o, = ~ (_~)
ay=~G), Py = ~ (-~)
(b) What are the eigenvalues corresponding to these eigenvectors ?
(c) Show that these eigenvectors comprise two sets of orthonormal vectors.

11.46 Spin-! state vectors (~) are called spinors. Spinors exist in a two-dimensional ,
complex space. Any element ~ of this space is a set of two complex numbers. Any
two linearly independent spinors span this space. In particular, show that this space
is spanned by any of the three pairs of eigenstate s (ax, Px; a y, Py; a z, pz). That is,
show that any spinor (~) may be expressed as a linear combination of anyone of
these three pairs of eigenstates.
516 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Answer

(:) =acxz +b~z

(:) = ~[(a + b)cxx + (a - b)~xl

(:) = ~[(a - ib)cxy + (a + ib)~yl

11.47 Measurement of the z component of the spin of a neutron finds the value Sz = !i12.
(a) What spin state is the particle in after measurement ?
(b) Show that in this state
2
is,2 ) = is,2 ) = :21 (S2 - s,2 ) = 4!i = ts, 2 )
by direct calculation .
11.48 An electron is known to be in the spin state a- , Show that in this state

(Sx) = (Sy) = 0
Explain this result geometrically (Fig. 11.2).

x <S(x,y»=O

FIGURE 11.2 Dynamical conception of the spin-! state. The angular momentum vector
precesses maintaining a constant component about an axis. The projection of S onto a plane
normal to the precession axis averages to zero. (See Problem 11.48.)

11.49 Show that it is impossible for a spin-! particle to be in a state ~ = (:) such that
(Sx ) = (Sy) = (Sz) = 0
 11.7 Free-Particle Wavefunctions, Including Spin 517

Answer
Since iT and S are related through a constant (11.81), we may work with iT instead
of S. First we find the relation between a and b which gives (az ) = o.

Setting lal = I , with no loss of generality, this condition implies that ~ is of the
form

where tf> and 1/1 are the phases of a and b, respectively. Substituting this vector into
(ax) = 0 gives

("!:Iax l!:)
"
= (e-ii/J e-
i lfr)
(0 I)
I 0
(eiI/J)
eilfr = 2cosa = 0
a=1/I-tf>
while (ay ) = 0 gives

Since there is no value of a for which sin a = cos a = 0, we conclude that the
stated hypothesis is correct.

11.50 (a) Show that the components of a anticommute. For example, show that

a
(b) Show that y is Hermitian .

11.7 • FREE-PARTICLE WAVEFUNCTIONS, INCLUDING SPIN

The coordinates of a particle include the spin variables (s, m s) and the position
variables (x, y, z). The operators corresponding to these variables (S2 , x, y, s; z)
are assumed to commute. Their eigenvalues may therefore be prescribed simulta-
neously (one may locate a particle without destroying its spin state) . Another set
of commuting operators for a free particle is (S2 , s; Px , py, pz).
The Hamiltonian of a free particle is
~2
~ P
H=-
2m

The reason that S does not enter in the Hamiltonian of a free particle is that the
spin manifests itself only in the presence of an electromagnetic field." It follows
518 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

that iI also commutes with S2 and s;

These operators have simultaneous eigen-
states. Let f/lk be an eigenstate of iI corresponding to the eigenvalue li,zk 2/2m
and (a, fJ) be the eigenstates of S2, Sz corresponding to the respective eigenval-
ues 3fi2/4 and ±fiI2. Then

f/l+ == f/lk(r)a = Ae
i k or
G) (11.83)

gives

The eigenstate

(11.84)

gives

Consider that k is unidirectional. Then the state tp.; gives the same energy
fi2 k 2 / 2m for the two vectors ±k. The same is true for f/l- . Thus we find that
eigenenergies of the spinning free particle propagating in one dimension are four-
fold degenerate. This is illustrated below.

The time-dependent wavefunction for an electron with momentum fik and with
z component of spin -fi12 is the column vector

1f!k(r, t) = Aei(kor-wt) G)
fi2k2
liw=-
2m
With A = (2n)-3 /2, one obtains

(k'[k) = fff 1f!k,*1f!k dr = 8(k' - k)

thereby regaining the normalization (10.18) relevant to free-particle motion .

8For relativistic electrons this is not the case. In this event, the free-particle Hamiltonian includes a
spin-dependent term. See Section 15.3.
 11.8 The Magnetic Moment of an Electron 519

PROBLEMS

11.51 A beam propagating in the x direction is comprised of 1.5-keV electrons, 20% of

which have spin polarized in the +z direction and 80% of which have spin polarized
in the -z direction.
(a) What is the wavefunction of an electron in the beam?
(b) What are the values of the wavenumber k and frequency w of these electrons ?

11.52 A free electron is known to have the following properties :

1. Its orbital angular momentum about a prescribed origin is

L = !iJl(l + I)
2. Its z component of orbital angular momentum is hm],
3. Its z component of spin is -!i12.
4. It has kinetic energy

(a) What is the time-dependent wavefunction, 11jJ} = Ik,l,m/,s ,ms ,t}, for this
electron ?
(b) If in an ideal measurement , the x component of linear momentum is measured
and the value

px = !ikx
is found, what time-dependent state is the electron left in?
(c) What is the degeneracy of the eigenenergy corresponding to part (b)?

11.8 • THE MAGNETIC MOMENT OF AN ELECTRON

Bohr Magneton
The student will recall that a circular loop of wire carrying a current l , which is
of cross-sectional area A, produces a magnetic moment?

J.L=/A (11.85)

The magnetic field due to this current loop is sketched in Fig. 11.3. In the limit that
A -+ 0 and / -+ 00 such that the product / A = J.L remains finite, the magnetic
field generated by the loop becomes a dipole field, whose components are also
given in Fig. 11.3.

9In cgs, A is in cm 2 and I is in emu/so The units of J.L are erg/gauss . Also recall that 1 emu = 1 esu /c.
520 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

<B
~---

FIGURE 11.3 Magnetic field lines of a current loop. The magnetic dipole of the cur-
rent loop is /-t. The magnitude of /-t is I A, where A is the area of the loop. At distances
far removed from the origin, the B field is that due to a magnetic dipole at the origin.
In spherical coordinates, with /-t parallel to the polar axis, the components of the dipole
magnetic field are
2p,cosO p,sinO
s; = - -r3 - ' /B()=--
r3
These values are computed from the Biot-Savart law

B(r) = -1
c
f J(r/) x
(r - r')
[r - r']
3 dr'

which gives the magnetic field at r, due to the current density J distributed over the source
points, r' .

For an electron, one finds that the magnetic moment is directly proportional to
its spin angular momentum. It is given bylO

I IL = -mee8 = -2me
en- u = -IIMT (11.86)

10A more detailed analysis, including radiative corrections, finds a slightly larger value of electron
magnetic moment and (11.86) is more accurately written fJ. = (1.001J.ib)U. Thus to within 0.1%
accuracy, the electron magnetic moment has the value of I Bohr magneton. See also related discussion
in Chapter IS.
 11.8 The Magnetic Moment of an Electron 521

The quantity en/2mc is called a Bohr magneton . It has the value

s
leln 20
lIb = - = 0.927 x 10- erg/gauss (11.87)
2mc

Since the charge of the electron is negative, (11.86) may be written

That is, the spin and magnetic moment of an electron are antialigned (Fig. 11.4).
If a magnetic moment is placed in a uniform, constant magnetic field, a torque
FIGURE 11.4 For an elec- is exerted on it about its origin, given by
tron, the spin and magnetic
moment are antialigned .
N=/LxB (11.88)

(Fig. 11.5). It follows that a magnetic moment tends to align itself with a magnetic
field in which it is immersed. We may use (11.88) to calculate the work done in
rotating the moment from the parallel orientation (B = 0) to the inclination B > 0
(Fig. 11.6):

FIGURE 11.5 Torque on a magnetic dipole in a uniform B field. The force on a point
charge moving in a B field, with velocity v, is

F=:'vxB
c
Four components of this force along the ring current are shown. These forces tend to align
1L with B so that 1L and B are in the same direction .
522 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

/I. (H

6=0

FIGURE 11.6 The energy of interaction between a magnetic dipole I.t and a magnetic
field B is a function of the inclination angle ().

v = -I.t. B

FIGURE 11.7 V versus () for a magnetic dipole in a uniform, constant B field.

v= 1 e
N de = - /I:B cos e + constant
(11.89)
V = -/L' B
e
If we plot this potential versus (Fig. 11.7), it is evident that = 0 is the stablee
e
orientation of the dipole. While the torque vanishes at = Jr (/L antiparallel to
B), any fluctuation about this orientation will cause the moment to "flip" to the
stable position'! e
= O. Although there is a torque on a magnetic moment in
a uniform B field, there is no net force on the dipole . However, the expression
(11.89) for the interaction energy'? between /L and B suggests that there is a net
force on a dipole in a nonuniform B field. This force causes the dipole, at a given
l1This is so if we neglect the angular momentum of the dipole (due to the rotating charge); if the
angular momentum of the moment is brought into play, precession results.
12This expression for the energy does not take into account the energy supplied by the source that
maintains the current in the dipole. It gives correct forces, however, if current and :B field are constant
in time. For further discussion on this topic. see R. Feynman, Lectures on Physics. Vol. II, Addison-
Wesley, Reading, Mass., 1964.
11.8 The Magnetic Moment of an Electron 523
z

V= -jJ" (8

FIGURE 11.8 Force on a magnetic dipole in an inhomogeneous magnetic field. For

the configuration shown, it is the r component of magnetic field, B r , which causes the
downward force on the ring. This may be seen by calculating the force (elc)v x B on
the ring, due to the r and z components of B, respectively. In terms of interaction energy,
V = -p, . B, V is larger at ZI than at Z2 . This causes the dipole to "fall" from ZI to Z2.

orientation with 13, to "fall" to a neighboring point in space where the interaction
energy, - p.. 13, is smaller (Fig. 11.8). The net force on the dipole in a nonuniform
13 field is given by

F = -VV = -V(-p.. 13) = V(p.. 13) (11.90)

Stern-Gerlach Experiment
Equation (11.90) reveals the nature ofthe force which occurs in the Stem-Gerlach
(S-G) experiment originally performed in 1922 using a beam of silver atoms . The
spin of a silver atom is due to its outer 5s electron.P (Thus in the following
we simply refer to electron spins in the beam .) A prototype of the experiment is
shown in Fig. 11.9. The predominant component of 13 is 93 z . Furthermore, 93z
varies most strongly with changes in z so that V{3z ~ ez 893z/8z. It follows that
the force on electrons as they pass through the pole pieces is

(11.91)

The predominant force on the electrons is in the z direction. In addition, the sign
of this force is solely determined by the sign of I-tz .

13See Problem 11.91.

524 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

.. x

Beam emerges
as two polarized
components
Cross section of
incident beam of electrons
depicting isotropic distribution
of S or 11. vectors

""~, ~_Cross section

of beam

End-on view of
pole pieces and
inhomogeneous
(ll-field

FIGURE 11.9 Elements of the Stem-Gerlach experiment.

This z component of force causes electrons in the beam to be deflected. They

strike the detection screen "off-axis ." If we know where an electron strikes the
screen, we can 14 calculate Fz. Since a:Bz/az is known (it is a property of the
apparatus) (11.91) then allows one to determine the z component of magnetic
moment. Thus the S-G apparatus is a device that measures the z component of
magnetic moment. More significantly, since I.t is directly proportional to the spin
S, the S-G apparatus becomes an instrument that measures the z component of
spin. Written in terms of spin of the electron, the force (11.90) becomes

(11.92)

14Thekinetic energy of the incident electron is assumed known.

11.8 The Magnetic Moment of an Electron 525

(a) (b)

FIGURE 11.10 Traces of electron beam on detection screen in the Stem-Gerlach exper-
iment. (a) The trace predicted from classical physics,

Continuous distribution of cos () values implies a continuous trace. (b) The trace observed
experimentally. If s = ~ for an electron, then quantum mechanics predicts that two discrete
beams emerge from the Stem-Gerlach apparatus.

At the start of the description of this experiment, we remarked that the in-
cident beam carries an isotropic distribution of directions of magnetic moments
or, equivalently, spin vectors (see Fig. 11.9). This means that at any point in the
beam it is equally likely to find f-Lz with any value from - f-L to +f-L. Thus accord-
ing to (11.91) one expects a uniform distribution deflections. The pattern that the
deflected particles in the beam make on the screen should be a continuous one.
However, experiment finds that the beam divides into two discrete components
(Fig. 11.10). Thus experiment indicates that f-Lz or, equivalently, S, has only two
components. But this is precisely what one expects if s == !'
for in this case Sz
has only the two values ±!ij2. So according to (11.92), an electron going through
the Stern-Gerlach :B field is acted on by only one of two possible values of force:

These two oppositely directed components of force divide the beam into two sep-
arate components.

Superposition Spin State

Instead of the beam containing an isotropic distribution of spin orientations, let
electrons in the incident beam all be polarized with spins in the +x direction. That
526 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

is, each electron is in the eigenstate (Xx of s, and has the wavefunction
(11.93)

The beam propagates in the +y direction. Let the beam be incident on an S-G
apparatus whose B field is aligned with the z axis. What portion of electrons in
the incident beam emerges with spins in the +z and - z directions, respectively?
Equivalently, we may ask, what is the probability that measurement of the z com-
ponent of spin of an electron in the beam finds the respective values of +!i12 or
-!i12? To answer this question, we call on the superposition principle. Expanding
the column vector (X x in terms of the eigenvectors of Sz gives

(11.94)

The two coefficients of expansion are equal. It follows that the probability of
measuring Sz = +!i12

is equal to that of measuring Sz = -!i12

Thus if a beam of polarized electrons all in the state a, enters an S-G appara-
tus, two equally populated beams emerge. These beams are also polarized, with
electrons in spin states (xz and {3z, respectively.

PROBLEMS

11.53 Show that the magnetic moment due to the orbital motion (as opposed to the spin)
of the electron in the Bohr-model hydrogen atom is given by

The orbital angular momentum of the electron is L. [Hint: If the electron moves in
a circle with velocity v and radius r, the related current is
ev
1= - - (emu/s)]
'In r c

11.54 Consider that a polarized beam containing electrons in the a z state is sent through
an S-G analyzer which measures Sx . What values will be found, and with what
probabilities will these values occur?
 11 .9 Precession of an Electron in a Magnetic Field 527

Answer
We write az as a linear combination of ax and f3x.

az=(~)= ~(ax+f3x)=~{G)+(_~)}
The coefficients of expansion are equal so that it is equally likely to find the values
Sx = + fi12 or Sx = -fiI2.
11.55 A proton is in the spin state ay . What is the probability that measurement finds each
of the following?
(a) Sy = -fi12
(b) Sx = + fi12
(c) Sx = -fi12
(d) Sz = +fi12
(e) Sz = -fi12
Answers
(a) 0
(b) ~
(c) ~
(d) ~
(e) ~

11.56 A collection of electrons has an isotropic distribution of spin values. For an elec-
tron chosen at random, what is the probability of finding it with the following spin
components?
(a) Sx = + fiI2 , -fi12
(b) Sy = +fiI2, -fi12
(c) Sz = +fiI2, -fi12
Answers
If the x component of spin is measured, only two values can be found. If spin is
isotropic, these two values are equally likely; hence there is a probability of 2: that
Sx = fil2 and a probability of ~ that Sx = -fiI2 ; similarly for Sy and Sz .

11.9 • PRECESSION OF AN ELECTRON IN A MAGNETIC FIELD

In this section we consider the motion of a spinning but otherwise fixed elec-
tron which is in a constant uniform magnetic field that points in the Z direction.
Suppose that the electron is initially in the ax state

(11.95)

What is ~(t)? To answer this question , we write down the time-dependent

Schrodinger equation for the state Ht) .
528 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

a A

i!i-~ = H~ (11.96)
at
The Hamiltonian is the interaction energy

fI = -A' B = +f.LbU • B = f.Lb93az

In the matrix representation with Sz diagonal, fI becomes

(11.97)

We seek the solution to (11.96) with this Hamiltonian, for the state vector

Substitution of this form into (11.96) with (11.97) substituted for fI gives the
column vector equation

where 0./2 is the Lannorfrequency and

0. = lel93 (11.98)
mc
is commonly referred to as the cyclotron frequency (previously encountered in
Section 10.4). The column vector equation above is equivalent to the two inde-
pendent equations
. in
a=--a
2
. in
b=+-b
2
which has the solution

~(t) = ( b
a)
=.fi
1 (e- (n I) i
e+ i (n j 2JI
j J
2
(11.99)

This solution is compatible with the initial conditions (11.95). At later times (in-
cluding t = 0) one obtains

~(O) = ax

~(T/4) = e- irr j4ay

~(T/2) = e-
i rrj 2
f3x
i3rrj 4
H3T/4) = e- f3y
11.9 Precession of an Electron in a Magnetic Field 529

z, rzJ3

t = 3T/4
FIGURE 11.11 Precession of a spinning but otherwise fixed electron in a constant, uni-
form magnetic field. Electron is initially in the ax state.

where we have written T for the period 2n / Q. Apart from constant phase factors,
we may conclude the following. At t == 0, the electron is in an eigenstate of
Sx corresponding to the eigenvalue +1i/2. At T /4 it is in an eigenstate of Sy
corresponding to the eigenvalue +1i/2. Proceeding in this manner one finds that
the spin of the electron precesses about the z axis with angular frequency Q /2
(see Fig. 11.11).

Eigenenergies
We now consider the problem of calculating the eigenstates and eigenenergies
of this same system, i.e., a spinning but otherwise fixed electron in a constant
uniform magnetic field that points in the z direction. To solve this problem we use
the time-independent Schrodinger equation. For the case at hand, it appears as

iJ~ == E~
(11.100)

Setting ~ = (~) gives

or, equivalently,
530 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

An
2

FIGURE 11.12 Energy eigenvalues of a spinning but otherwise fixed electron in a uni-
form, constant magnetic field. The orientation of p, with respect to B is also shown for
these two states.

(JLbB - E)a = 0

(JLbB + E)b = 0

If a i= 0, E = +JLbB and b = O. If b i= 0, E = -JLbB , and a = O. Thus we

obtain the (normalized) eigenstates, and eigenenergies

a= G),
(11.101)
(~) ,
/in
fi = E = -JLbB = - -
2

In the state of higher energy, the spin of the electron is parallel to B, so the mag-
netic moment is antiparallel to B and the interaction energy -It· B is maximum.
In the state of lower energy, the spin is antiparallel to B, so the magnetic moment
is parallel to Band - It . B is minimum (Fig . 11.12) .

Magnetic Resonance
As found above (11.86), the relation between the magnetic moment of an electron
and its spin is given by

This expression may be written in terms of the Lande 9 factor,

11.9 Precessionof an Electron in a Magnetic Field 531

with 9 = -2. For nuclear particles such as a proton or a neutron, the relevant unit
of magnetic moment is the nuclear magneton

fJ-N = ~
2M c
= 0.505 x 10- 23 erg/gauss
p

where M p is the mass of the proton. The magnetic moment of a nuclear particle
is written

For a proton 9 = 2(2.79), while for a neutron 9 = 2(-1.91). We wish now to

describe a technique of measuring 9 or, equivalently, the magnetic moment.
As found in the first part of this section, if a magnetic moment is immersed
in a steady B field in the z direction , the moment will precess about the z axis
with the "Larmer" frequency 9 0./2. Measurement of the magnetic moment is
made through inducing a spin flip of the magnetic moment between its two en-
ergy states (for a spin-! particle) parallel and antiparallel to B (Fig. 11.l2). As
in the corresponding classical configuration, in order to change the angle that IL
makes with the z axis, it is necessary to impose an additional transverse magnetic
field normal to the plane through the z axis and IL (Fig. 11.l3). Since this plane
rotates with the Larmor frequency, in the corresponding quantum mechanical mo-
tion one may expect a spin flip of the magnetic moment to be induced when the
frequency of an imposed rotating transverse magnetic field is equal to the Larmor
precessional frequency. Let us examine quantitatively the manner in which this
resonance occurs for a spin-! nuclear particle with magnetic moment 9 (fJ- N / n)S.

"..,--------- ...............
(
..... - \
,/

.....
y

x
FIGURE 11.13 Transverse field 2.1. imposes a torque N that causes p, (or, equivalently,
S) to change its orientation with respect to the z axis.
532 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Since 1:1 has three components, the Schrodinger equation (11.96) takes the
form

. 1i~
I at (a)
b
= _9J-LN
2
[(0 1):B
1 0 x
+ I. (0
1
-1):B
0 y + (10
Let the transverse magnetic field (:Bx , :By) rotate with frequency wand let :Bz
maintain at the constant value :BII.

:Bx = :BJ.. cos tot , :By = -:B J.. sin on,

Substituting these values into the preceding equation gives the coupled equations

-oo = I·( .OJ..

rt beion rt
+ .olla)
at (11.102)
-ab = I. (rt
.0J..ae -iwI - "b)
.011
at
The frequencies r.lJ.. and r.lll are defined through the relations

21ir.lJ.. = 9J-LN:BJ.. ,
21ir.l1l = 9 J-LN:BII

We seek solution to equations (11.102) corresponding to the initial conditions

a = 1, b=O at t = 0

Let us look for solutions in the form

(11.103)

The coefficients ii and b are assumed independent of time. Substituting these

forms into (11.102) gives the homogeneous matrix equation

Wa - ~II) -r.lJ..e-iif>l) (~) _ 0

( -r.lJ..e1if>1 (Wb + r.lll) u : (11.104)
¢ =Wa -Wb - W

In order that 7i and b be independent of time, we must set ¢ = 0, which gives

Wa = Wb +W

Setting the determinant of the coefficient matrix in (11.104) equal to zero gives
the frequencies
11.9 Precession of an Electron in a Magnetic Field 533
W W
Wb=-2"±W, Wa=2"±w
(11.105)
-2
W == (
QII- 2"W)2 +Q..L 2
It is only for these values of Wa and Wb that the proposed forms (11.103) are solu-
tions to (11.102). The fact that two frequencies emerge for Wa and Wb corresponds
to the property that (11.102) represents two independent, second-order differen-
tial equations in time. The general solution for, say, b(t) is a linear combination
of the two frequency components.

bet) = bl exp [-i (~ - w) t] + b: exp [-i (~ + w) t]

To match this solution to the initial condition b(O) = 0, we choose bl = -b: =
C ru. There results

bet) = C sin wt e- i (wj 2)f

To fix the coefficient C, we insert this solution into the second of equations
(11.102) and set t = 0, a(O) = 1, to find

The solution aCt) corresponding to the specific form b(t) obtained above is simply
constructed from the second equation of (11.102) .
Combining our results we obtain

I~) = (a(t)) = sinwt (ei (Wj 2)f [i ( Q II .: w/2) + wcotwtl) (11.106)

b(t) w ie- /(w j 2)f Q ..L

Since normalization of the spinor I~) is obeyed at t = 0,

(~(O)I~(O)) = la(0)12 + Ib(0)1 2 = 1

we should find that it is maintained for all time. From (11.106) we obtain

Recalling the definition of w(11.105), we find that

so that normalization is maintained for all time.

Let us see how the solution (11.106) implies the resonant spin-flip behavior
when the driving frequency W of the transverse field 93..L is equal to the frequency
534 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

2nll ' From the fonn of Ib1 2 ,

(11.107)

we may infer the following . In an ensemble of spins that are all pointing in +z
direction at t = 0, the fraction Ib(t)12 will be found pointing in the -z direction
at the time t. From the preceding expression for I b 12 , it is clear that the number of
such spin flips is maximized at the resonant frequency

(11.108)

which , as expected, corresponds to the precessional frequency. The structure of

the amplitude of Ib(t)12 further indicates that this resonant phenomenon may be
made more pronounced by choosing the transverse field small in comparison to
the steady parallel field $11'

Experimental Description
The molecules in a sample of water have zero magnetic moment save for that
carried by the protons of the hydrogen nuclei. When placed in a steady magnetic
field these protons align in one of two possible orientations. In thermal equilib-
rium there are slightly more protons in the lower-energy orientation with magnetic
moment parallel to B (Fig. 11.12). One may utilize the presence of this net excess
of magnetic moment to measure the 9 factor of protons. As described above, spin
flips of the aligned protons may be induced by an additional rotating transverse
magnetic field. This effect is maximized at the resonant frequency (11.108)
e
w= 2nll =g--$II
2M p c
With the value of this frequency observed, the relation above may be solved for
g , since all other quantities in the equation are known .
In inducing a transition from the lower to the upper energy state, energy in the
amount

e
lin = lig--$II
2M p C

is absorbed by the transition . This energy comes from the supporting transverse
magnetic field coils. Spin flips to the lower parallel orientation expel energy lin.
Since there are slightly more protons in the lower energy state than in the upper
Problems 535

energy state in thermal equilibrium, there will be a net detectable absorption of

energy from the transverse coils at the resonant frequency. Measurement of this
frequency yields the value 9 = 5.58 for the proton.
In concluding this discussion we note the following convention regarding mag-
netic moment. Consider a particle with intrinsic spin S. Its magnetic moment is
given by

The experimentally quoted value of magnetic moment refers to the expectation of

the z component of /L with S inclined maximally toward the z axis, that is, in the
state m, = s. With the wavefunction of the particle written Is, m s) this value is
given by

For a proton

/Lp = 2(2.79) (/L;) S

which gives the value

For a neutron

/Ln = -2(1.91) (/L;) S

to which corresponds the value

/Ln = -1. 91/LN

Similarly, for an electron we find that

PROBLEMS

11.57 What frequencies are emitted by an electron gas oflow density (~108 cm- 3) which
is immersed in a uniform magnetic field of strength 104 gauss due to spin flips?
What type of radiation is this (x ray, microwaves, etc.)? How does this emission
frequency compare to the classical frequency emitted when an electron is executing
Larmor rotation? (See Section 10.4.)
536 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.58 In a nuclear magnetic resonance (NMR) experiment with $11 set at 5000 gauss,
resonant energy absorption by a sample of water is detected when the frequency of
the transverse components of magnetic field passes through the value 21.2 MHz.
What value of 9 for a proton does this data imply?

11.10 • THE ADDITION OF TWO SPINS

In Section 9.4 it was concluded that when adding the angular momenta of two
components of a system , one may work in one of two representations: the cou-
pled representation or the uncoupled representation. This also holds, of course,
for the addition of spin angular momenta. As in the orbital case, the construction
of wavefunctions in the uncoupled representation proves simpler. In this represen-
tation, wavefunctions for the two-electron system are simultaneous eigenstates of
the four commuting operators 05 12,0522 , Slz, and Szz. They may be written

(11.109)

where, for example,

These eigenstates (~I, . . . ,~4) are simple products of the eigenstates a(I) , ,B(I)
of 05 12 , Si z and a(2) , ,B(2) of Sz2, Szz . They are listed in Table 11.2. The column
on the left of Table 11.2 contains diagrammatic representations of these states in
which the relative orientation of the two spins is suggested.

TABLE 11.2 Spin wavefunction for two electrons in the

uncoupled representation

Spin Wavefunction
Alignment ~ = 15152mSlmS2) m St mS2 51 5Z

1 1 1 1
t t ~1 = a(l)a(2) +-2 +-2 -
2
-
2
I 1 1 I
.j, .j, ~z = {3(1){3(2) 2 2
-
2 2
1 1 1 1
t .j, ~3 = a(1)f3(2) +-2 --
2 2
-
2
1 1 I 1
.j, t ~4 = f3(1)a(2) --
2
+-2 -
2
-
2
11.10 The Addition ofTwo Spins 537

In the coupled representation, one constructs simultaneous eigenstates of the

four commuting operators
A2 A A 2 A2 A2 A A
S = (81 + 82) = SI + S2 + 281 · 82

s, = 51z + 52z
51 2,522

The simultaneous eigenstates of these four operators may be written Ismssls2) .

Since s = (0, 1), again there are four such independent eigenstates. In construct-
ing these states we show first that two of the states appropriate to the uncoupled
representation are also eigenstates of the coupled representation. Toward these
ends note that all four uncoupled states are eigenstates of the total z component
of spin .

5z; 1 = !l41
5z;2 = -!142 (11.110)

5z;3= 5z;4= °
The eigenstate sj is an eigenvector of s;
5 1z, and 52z' If it is also an eigenstate
of 52 , it is one of the eigenstates appropriate to the coupled representation. To
see if this is indeed the case, we employ the relation [see (9 .22), Table 9.4, and
Problem 9.31]

81 .82 = (51x 52x + 51y52y) + 51z52 z

lA A A A A A (11.111)
= 2(SI+S2- + SI-S2+) + SI zS2z
in the cross term of 52 to obtain the expansion
A2 A2 A2 A A A A A A
S = SI + S2 + 2SI zS2z + (SI+S2- + SI-S2+) (11.111a)

The raising operator S I+ is defined by l 5

51+ = 51x + i51y

with similar definitions carrying over to 52+, 51-, and 52- . Using the above rep-
resentation of 81 . 8 2, we find that

15See Table 9.4.

538 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

so that

We conclude that;1 is also an eigenstate of 32, hence it is one of the eigenstates

of 32 and 3z for two spin-! particles in the coupled representation. We relabel
this eigenstate ;s(l), for reasons that will become apparent immediately.

;1 = a(l)a(2) == ; s (l)
Similarly, we find that;2 is also a common eigenstate of 32 and Sz. This eigen-
state is relabeled ;S(-I).

;2 == ,8(1),8(2) == ;S(-I)
However, ;3 and ;4 are not eigenstates of 32.

The Exchange Operator

To find the remaining two eigenstates (which will be called ;s (O) and ;A) of the
coupled representation we introduce the exchange operator, i. When i operates
on a function of coordinates (spin or space) of two particles , it exchanges these
coordinates. If cP (l , 2) is a function of the coordinates of two particles (the spin
coordinates of particle 1 are labeled" 1," those of particle 2 are labeled "2"), then

icp(l, 2) = cp(2, 1) (11.112)

From the definition of X, one obtains

so that the ei~envalues of X are ± 1. One may construct the corresponding eigen-
functions of X from any state cp(l, 2) as follows :

CPS = cp(l, 2) + cp(2, 1) ("symmetric")

clearly has eigenvalue + 1, whereas

CPA = cp(l, 2) - cp(2, 1) ("anti symmetric")

has eigenvalue -1 .
Since S2 and s,
commute with i, it follows that S2, s;
and i have common
eigenstates. We already know two of them, ; s (l) and ;S(-I).

S2;S(l) = 2/i2;S(I) , Sz;s(l) = n;s(l) , i;s(l) = +1 ;s(1)

S2;S(-I) = 2/i2;S(-I) , Sz;S(-I) = -n;s(-I), i;S(-I) = +1;S(-I)
11.10 The Addition of Two Spins 539

These equations serve to explain our notation. The superscript on;s (I) (i.e., + I) is
the eigenvalue of s; while the subscript S denotes "symmetric." The eigenstates
;s(1) and ;s(-\ ) are symmetric with respect to particle exchange.
Of the remaining two simultaneous eigenstates of S2, s;
and i , one is sym-
metric, ;~o) , and one is antisymmetric, ;A.Their construction follows simply from
that of Ips and IpA given above, using the two degenerate eigenstates of Sz (i.e.,;3
and ;4):
1 I
;s(O) = .)2(;3 + ;4) = .)2[a(I),B(2) + ,B(I)a(2)]
(11.113)
1 1
;A = .)2(;3 - ;4) = .)2[a(1),B(2) - ,B(1)a(2)]

Using the expansion (11.111) for S2, one finds that

S2;S(0) = 21i2; s(0)
A2
S ;A = Oli2;A
while

Sz;/O) = On;s, i;s(O) = +1;s(O)

Sz;A = On;A, i;A = -1;A

In this manner we obtain that the four independenteigenstates of (S?, Sl , S2, Sz)
are (;s(l) , ;/0), ;s(-I) , ;A) . Properties of these eigenstates are listed in
Table 11.3. The three ;s states correspond to s = 1, whereas the ;A state corre-
sponds to s = O. That is,

TABLE 11.3 Spin wavefunctions for two electrons in the coupled representation
Spin Wavefunction
Alignment ~= IS\S2 sm s ) S ms S\ S2

t t ~s(1) = 0'(1)0'(2) \
2:
\
2:

t,j,+,j,t ~s(O) = ~[O'(l),8(2) + ,8(1)0'(2)] 0 \

2:
\
2:

,j, ,j, ~s(-\) = ,8(1),8(2) -1 \

2:
\
2:

1 \ \
t,j, -,j,t ~A = J2[O'(I),8(2) - ,8(1)0'(2)] 0 0 2: 2:
540 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Spin Values for Two and Three Electrons

In the coupled representation one may speak of the total angular momentum of
the two-electron system. In Section 9.4 we concluded that the total angular mo-
mentum quantum numbers for a two-component system vary in unit steps from
Il] + hi to Il] -hi [see (9.114)] . In similar manner the total spin angular momen-
tum of a two-electron system has spin quantum numbers varying from I! !I + to
I!- !I s= s= s=
in unit intervals . This gives the two values 0 and 1. For
there are three m, values : m, = 1, 0, and -1. For s = 0 there is one m, value,
1

m, = O. This partitioning of states according to total spin number s is depicted in

Table 11.3 in the (s, m s ) column.
If we use the angular momentum addition rules of Section 9.4 to calculate the
spin quantum number corresponding to the possible total spin values for three
electrons, we obtain the series

s=I~+~+~I·
2 2 2 ' ·· , I~+~-~I
2 2 2

which gives

3 1
s= - -
2'2

There are four m, values for s = ~ and two for s = !.

3
m, = 2
1
3 m, =2
s =-
2 1
m s =--
2
3
m s =--
2

These values of spin quantum number are appropriate to the coupled representa-
tion. The corresponding eigenstates are of the form

which are simultaneous eigenstates of the five commuting operators

(11.114)

These concepts of spin angular momentum are very relevant to topics in atomic
physics, as will be discussed further in Chapter 12.
Problems 541

PROBLEMS

11.59 Show that the coupled spin states ~s(O) and ~A given in (11.113) are eigenvectors
of S2 with eigenvalues 21i2 and 0, respectively.

11.60 Obtain the four spin states Ismssls2) , listed in Table 11.3. relevant to two electrons
in the coupled representation, through a Clebsch-Gordan expansion of the form

Isms sls2) = L Cmlm2Islml)ls2m2)

ml+m2=ms

11.61 Consider a spin-I particle. For integral angular momentum quantum number. the
matrices developed in Section 11.5 for orbital angular momentum also apply to
spin angular momentum. Using the three-component column eigenvectors of the
Cartesian components of 8 [e.g., (11.68)]. determine if it is possible for a spin-I
particle to be in a state

such that

11.62 What is the fonn of a spin eigenstate. in Dirac notation. in the uncoupled represen-
tation for the three-electron case? How many "good" quantum numbers are there?

Answer

There are six good quantum numbers.

11.63 Using the results of Problem 9.43 (and the discussion preceding). construct an op-
erator, Ii 12• which commutes with the five commuting operators (11.114) relevant
to the addition of three electron spins in the coupled representation.
11.64 For the case of four electrons. in the coupled representation.
(a) What are the s eigenvalues?
(b) Write down the form of six commuting operators explicitly in terms of the
vector operators 8I , 82. 83, their inner products. and their z components .
(c) What is the form of an eigenstate in Dirac notation? How many independent
states of this form are there?

Answer (partia/)
I I I I I
( a) s= I 2+2+2+2 . .. ·• II2+2-2-2
I I I I

= 2. 1.0
11.65 The expression for the component of spin along an axis z', which makes respective
angles (8. fJ. y) with the (x. y , z) axes. is
542 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

z
z

z'

x
x
(a) (b)

FIGURE 11.14 (a) Direction angles (0 , fJ , y) locate the z' axis with respect to the origi-
nal Cartesian frame . (b) The angles (0, fJ, y) for the case described in Problem 11.65. The
z' axis lies in the yz plane.

az' =axcosO+aycosfJ+azcosy
cos 2 0 + cos 2 fJ + cos 2 Y = I
See Fig . 11.14 . This relation follows from the vector quality of ir.
(a) Write the matrix for az in terms of 0, fJ, y and show that the eigenvalues of az'
are the same as those of az.
(b) An electron is known to be in the state a z . What is the probability that mea-
surement of S ZI finds the respective values ±n/2 for the angular displacements
o = 1T/2, fJ = 1T/2 - y ? See Fig . 11.l4b.
(c) Describe a double S-G experiment in which such measurement may be effected.

Answer (partial)
(b) To obtain the answer, one must find the eigenvectors of az' corresponding to
the eigenvalues ±1 and then expand a z as a superposition of these two states.
The squares of the coefficients of expansion give the respective probabilities

P ( +~) = 1 +~os y

p(_~)= I-~OSY

As y --+ 0, the z' and z axes merge and P(+n/2) --+ 1, P(-n/2) --+ O. For
y = 1T/ 2, the z! axis is normal to the z axis and P (+n/2) = P(-n /2) = 1.
11.66 Show, employing explicit matrix representations, that

S+a = S-fJ = 0
 11 .11 The Density Matrix 543

11.67 Show that the Pauli spin operators obey the relation

axayaZ = if
11.68 Establish the following properties of the Pauli spin matrices:
(a) ax Z = ai = azZ = i (a Z = 31).
(b) (ax, ay} = 0, etc., where {,} denotes the anticommutator.
(c) axay = iaz , etc.
(d) The preceding relations are included in the equality

(o-.a)(o-. b) =a· bi +io-· (a x b)

where a and b are arbitrary vectors.
(e) [ax, ay] = i2az , etc.
(f) If e is a unit vector in an arbitrary direction, what is (0-. e)Z?
(g) Show that an arbitrary but traceless 2 x 2 matrix may be written as a linear
combination of the three Pauli spin matrices. [Hint: Write your test matrix with
diagonal elements (z, -z) and arbitrary off-diagonal elements and solve for the
coefficients of combination of Pauli matrices.]
11.69 Measurement is made of the sum of the x and y components of the spin of an elec-
tron. What are the possible results of this experiment? After this measurement, the
z component of spin is measured. What are the respective probabilities of obtaining
the values ±n/2?

11.11 • THE DENSITY MATRIX

Pure and Mixed States

In our description of quantum mechanics to this point, we have considered sys-
tems which by and large have satisfied the idealization of being isolated and to-
tally uncoupled to any external environment. The free particle, the particle in a
box, the harmonic oscillator, and the hydrogen atom are cases in point. Any such
isolated system possesses a wavefunction of the coordinates only of the system.
This wavefunction determines the state of the system.
Consider, on the other hand, a system that is coupled to an external envi-
ronment. Such, for example, is the case of a gas of N particles maintained at
a constant temperature through contact with a "temperature bath." Very simply,
if x denotes coordinates of a system and y coordinates of its environment, then
whereas the closed composite of system plus environment has a self-contained
Hamiltonian and wavefunction 1/1 (x, y), this wavefunction does not, in general,
fall into a product 1/11 (x)1/Iz(y). Under such circumstances, we may say that the
system does not have a wavefunction. A system that does not have a wavefunction
is said to be in a mixed state. A system that does have a wavefunction is said to
be in a pure state.
It may also be the case that, owing to certain complexities of the system, less
than complete knowledge of the state of the system is available. In quantum me-
chanics, such maximum information is contained in a wavefunction that simulta-
544 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

neously diagonalizes a complete set of commuting operators relevant to the sys-

tem, such as that described in Section 5.5. Let a set of such operators be A, iJ, C
with common wavefunction 1/Jabc. Now suppose that the system is such that it is
virtually impossible to measure A, B, and C in an appropriately small interval
of time. Then the wavefunction 1/Jabc cannot be determined and under such cir-
cumstances one also speaks of a system being in a mixed state. As in classical
statistical mechanics, this situation arises for systems with a very large number
of degrees of freedom , such as, for example, a mole of gas. The quantum state of
such a system involves specification of "-'1023 momenta . The study of such com-
plex systems is called quantum statistical mechanics and it is in this discipline
that the density operator finds its greatest use.

The Density Operator

In dealing with situations where less than maximum information on the state of
the system is available, one takes the point of view that a wavefunction for the
system exists but that it is not completely determined . In place of the wavefunc-
tion, one introduces the density operator p. If A is some property of the system,
the density operator determines the expectation of A through the relation

(A) = TrpA (lLlI5)

and

Trp = 1 (lLlI6)

The trace operation, written Tr, denotes summation over diagonal elements
[see (1l.27)]. From the density operator we may calculate expectation values of
all relevant properties of the system.
Let us calculate the matrix elements of p for the case of a system whose wave-
function 1/J is known. Then

(A) = (1/JIA 1/J)

Let the basis {In}} span the Hilbert space containing 1/J. One may expand 1/J in this
basis to obtain

11/J} =L In}(nl1/J}
n

Substituting this expansion into the preceding equation gives

(A) = L L(1/Jlq}(qIAln}(nl1/J}
q n

= LLPnqAqn = TrpA
q n
11 .11 The Density Matrix 545

Here, we have made the identification

The coefficient an represents the projection of the state ljr onto the basis vector
In} . The nth diagonal element of pis

(1Ll17)

which we recognize to be the probability Pn of finding the system in the state In} .
Thus , the diagonal elements of p are probabilities and must sum to 1. This is the
rationalefor the property (lLlI6), Trp = 1.

The Mixed State

Now consider a system that is in a mixed state. Let IV be a measurable property of
the system such as its energy. Let {In}} be the eigenstates of IV. Since the system
is in a mixed state, we may assume that the projections an are not determined
quantities. In this case we define the elements Pnq to be the ensemble averages
(see Section 5.1),

Pnq = aq*an (11.118)

The diagonal element

Pnn = an*an (1Ll19)

represents the probability that a system chosen at random from the ensemble is
found in the nth state.

Equation of Motion
Suppose again that a system is in a pure state and has the wavefunction ljr. Again,
let N be of a measurable property of the system with eigenstates {In} }. Expanding
ljr in terms of the projections an gives

Iljr} = :~::>n(t)ln}
n

From the Schrodinger equations for ljr, we obtain

in "aa
n
LJ -In} = " .
LJanHln}
n at n

Operating on this equation from the left with (11 gives

-In--
at
"H
.~ aal* = LJ
n
In *an* (1 Ll20)
546 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

We may use these relations to obtain an equation of motion for the matrix ele-
ments of p.

. f,J )Pnl . ~ a al *an . ~ ( aal * *aan)

I n - - = I n - - =In --an+al -
at at at at

Substituting the expressions (11.120) for the time derivatives of the projections
an and setting Hlk* = Hkl gives

(11.121)

This may be written in operator form as

r
in a~ = [H, p] (11.122)

Random Phases
Consider the matrix elements of p (11.118),
Pnm = am*an

The indeterminacy of the state of the system may be manifest in a corresponding

indeterminacy of the phases {¢n} of the projections {an} . These phases are defined
by the relation

where en and ¢n are real. What is the consequence of assuming that the phases
{¢n} are random over the sample systems in the ensemble? Consider the matrix
element

If phases are random, then in averaging over the ensemble, cos(¢n - ¢m) will
enter with positive value equally often as with negative value. The same is true
for sin(¢n - ¢m), so that

cos(¢n - ¢m) = sin(¢n - ¢m) = 0

except when n = m . In this case

It follows that for the case of random phases, p is diagonal.

Pnm = Pnnonm (11.123)
11 .11 The Density Matrix 547

Evolution in lime of Diagonal Density Matrix

Suppose that p is diagonal. How does p then evolve in time? SpecificaJly, does
it remain diagonal ? From (11.121) we conclude that the evolution in time of the
diagonal elements of p depends on the off-diagonal elements of p. If these off-
diagonal elements begin to grow away from zero, {Pll} will change . The equation
for the off-diagonal elements is obtained from (11.121) . If at t = 0, Pis diagonal,
then

. apnl
l l i - = Hnl(Pll - Pnn) =1=
at ° (t = 0) (11.124)

Thus the diagonal distribution is not, in general, constant in time. However, it is

quite clear that the uniform distribution

(aJl states equaJly populated) is stationary in time.

Density Matrix for a Beam of Electrons

An electron beam generated at a cathode is known to have an isotropic distribution
of spins, so that

Let us ~alculate a density matrix that &ive~ this p~operty in a representation in

which S, is diagonal. The matrices for S» , Sy, and S, are given by (11.78). Since
the spins are isotropicaJly oriented, the probabilities of finding Sz with values
±1i/2 are both 1/2. In the said representation, and with the property (11.117), we
i i.
conclude that the diagonal elements of p are and It foJlows that the matrix

P
= ~
2
(1° 0)
1
(11.125)

gives

Ii
(Sz}=TrpSz=Tr 4
A A (1° 0) (1° 0)
1 -1 =0

From (11.78) we quickly conclude that this choice of p also renders (Sx) =
(Sy) = 0.

Projection Representation
As noted above, if the wavefunction of a system is indeterminate, one may de-
scribe properties of the system through the use of an ensemble of replica systems .
Consider that states 1/1 of the ensemble systems are distributed with the probabil-
548 . Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

ity Pt/t . An alternative form of the density operator is given by the projection sum
over states of the ensemble.

(l1.I26)

In the preceding example of a beam of isotropic spins, we found the den sity oper-
ator to be given by ( 11.125). Thi s operator is written in a repre sentation in which
Sz is diagonal, for which the states corresponding to Sz = ±ti/2 are la ) and Ill)
given by (11.79). Since the ensemble of systems contains only two values of PSz '
the summation (11.126) for this example may be written

(l1.I27)

Let us use this form to calculate (Sx).

A A A 1 A 1 A

(Sx) = TrpSx = (alpSxl a ) + (,BlpSxl,B) = 2(alSx1a) + 2(,BISxl,B)

=~ [(I O):e ~) G) +:(0 1):(~ ~) (~) ] =0

In like manner, we find that (11.I27) gives (Sy) = O.
Let a system be in a mixed state. Although the wavefunction is not determined,
it is known that the probability that mea surement of energ y finds the value En is
Pn . In this case (l1.I 26) becomes

p= L 11/In) P {1// 1
n
n n (l1.I28)

Since {1 1/1r) } is an orthonormal sequence, it follows that p is diagonal in this repre-

sentation with diagonal elements equal to Pn . If the system is a gas of N particles,
the number of particles with energy En is N Pn, which for nondegenerate states is
the same as the number of particles in the state 11/1n) . Thus the diagonal elements
of p in the case at hand give the occupation numbers for the state s of a system. 16

PROBLEMS

11.70 Is a system that is in a superposition state in a pure or a mixed state?

A nswer
A system in a superposition state is in a pure state. The wavefunction of the system is
known, and all properties of the system may be determined to the maximum degree
that quantum mechanics allows . Let A, iJ be a complete set of compatible operators
for the system. Let the common eigen states {1{!} of A, iJ span the Hilbert space 5).

16For further discussion and problem s on the density matrix, see R. H. Dicke and J. P. Wittke, Intro-
duction to Quantum Mechani cs, Addison-Wesley, Reading , Mass., 1960.
Problems 549

Then 1fr', which is a superposition state with respect to the observables A and B ,
does not lie along any of the basis vectors {1fr}. As described earlier in the chapter,
1fr', which exists in Sj, is related to 1fr through a unitary transformation , 1fr' = U1fr.
In this manner, one may obtain a new set of states {1fr'} which also span Sj. In this
new basis, the operator A has the value A' = UAU-I. SO A'1fr' = UAU-IU1fr =
Ua1fr = av', Furthermore, [A', B'l = 0 if[A, Bl = O. Thus we may conclude that
1fr' is a common eigenstate of A' , B'.
11.71 What is the spin polarization of a beam of electrons described by the density oper-
ator

p= i G I)? 1 .

11.72 (a) What is the density operator corresponding to an isotropic distribution of

deuterons (spin 1) in the representation in which Sz is diagonal?
(b) What is the value of (Sy)?
(Hint: Your answer should appear as a 3 x 3 matrix.)

11.73 Consider a particle in a one-dimensional box with walls at x = 0 and x = L.

Eigenenergies are En = n 2 E I . It is known that the probability of finding the particle
t
with energy E I is and that of finding it with energy E5 is t.
(a) What is the density matrix for this system in the energy representation ?
(b) Construct two normalized wavefunctions that give the same probabilities and,
therefore, the same density matrix.
11.74 The canonical formr' of the density operator is given by

p = A exp (-H)
kBT

where kB is Boltzmann's constant and T denotes temperature . Consider that H is

the Hamiltonian of a one-dimensional harmonic oscillator with fundamental fre-
quency WO. Working in the energy representation :
(a) Find the diagonal elements of p.
(b) Determine the normalization constant A.
(c) Calculate the expectation (E ) of the oscillator.
(d) Construct the projection representation of p (11.126).
(Hint: For summation of series, see Problem 2.36.)

11.75 Show that

(O-Z)2n = i.
(o-z)2n+1 = o-z
where n is an integer. (This result also holds for the operator e • ir, where e is any
fixed unit vector.)
17This density matrix is relevant to a system with Hamiltonian fI, which is maintained in equilibrium
at the temperature T through contact with a heat reservoir. For a further discus sion, see K. Huang ,
Statistical Mechanics, Wiley, New York, 1963.
550 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.76 (a) Using the results of Problem 11.75, show that

_
Rc/J T z ) = exp (1/>
= exp (iI/>5 _)
i "2O"z

= ( cos "21/»/- + I. ( SIn"2

. 1/» o"z
_

(Recall Problem 9.17.) This rotation operator tells us how spinors transform
under rotation. The transformed spinor is given by ~ / = Rc/J~ .
(b) What is the matrix form of Rc/J ?
(c) Show that Rc/J preserves the length of ~ . That is, show that WI~ /) = (~ I ~) .
(d) Show that Rc/J, at most, changes the phases of the eigenvectors of O-x , O-y , and
o-z·
(e) Show that under a complete rotation (I/> = 2Jr) about the z axis, ~ _ ~/ = -~ .
11.77 Prove the general expansion

expue- iTl/» = (cosl/»i + i(sinl/»e· iT

where, again, e is an arbitrarily oriented unit vector.
11.78 (a) Show that the spin-exchange operator j: for two electrons has the representation
2_ _ _ /t2
/t 1: = 28 1 .82 + 2"

(b) Show that j: may also be written

/t2j: = 52 _ h 2

[Hint: For part (a), let j: operate on a(I),B(2) and a(2),B(l) , respectively.]
11.79 A beam of neutrons with isotropically distributed spins has the density matrix

in a representation where 52 and 5z are diagonal. From the condition (8) = 0, show
that a = o.
11.80 (a) If p. is the magnetic moment of the electron, in what state will ](J1-z ) I = J1-b?
(b) What is the value of (J1-2) in this state?
11.81 Consider a process in which an electron and a positron are emitted collinearly in the
+y and -y directions, respectively. Spins are polarized to lie in the ±z directions.
The pair is emitted with zero linear and spin-angular momentum and with total
energy tuo. With the electron labeled 1 and the positron labeled 2:
(a) Write down a spin-coordinate, time-dependent product wavefunction for the
electron positron pair which contains these properties .
(b) What is the probability that measurement finds the electron's z component of
spin equal to +1i/2 ?
Problems 551

(c) Suppose that measurement of the positron's z component of spin finds the value
-nI2. What is the wavefunction for the pair immediately after this measure-
ment?
(d) What will measurement of the electron's z component of spin now find?
Answers
(a) The appropriate zero spin factor of the wavefunction is found in Table 11.3.

1
o/(l ,2) = .j2 (a(l),s(2) - a(2),s(l» exp[i(kYI - kY2 - wt))
2 (2n)

(b) It is equally likely to find the values ±n12 for the electron's z component of
spin. The same is true of the positron's z component of spin .
1
(c) o/after(l, 2) = 2n a(l),s(2) exp[i (k(YI - Y2) - wt)]

Here we are assuming that measurement preserves Sz and energy.

(d) +nI2.
(Note: This problem contains a key tool of the Einstein-Podolsky-Rosen para-
dox. 18 Consider two observers 01 and 02 positioned along the Y axis equipped,
respectively, with detectors S-GI and S-G2 oriented for measurement of Sz. Up un-
til the time that 01 makes his measurement, 02 is equally likely to measure ±nI2.
Once 01 makes the measurement and measures, say, +nI2, 02 is certain to find the
value -n12 upon measurement. This situation, presumably, maintains for 01 and
02 sufficiently far from each other with electron and positron beyond each other's
range of interaction, thereby violating the principle of locality .)

11.82 Consider the antisymmetric spin zero state

relevant to two spin-1 particles labeled I and 2, respectively. This state is an eigen -
state of s, and S2 with eigenvalue zero.
(a) Show that ~A (x) is also an eigenstate of Sz, thereby establishing that it is a
common eigenstate of the two noncommuting operators, Sx and s;
(b) What property renders the result in part (a) compatible with the commutator
theorem (Section 5.2)?

(8A. Einstein, B. Podolsky, and N. Rosen, Phys. Rev. 47, 777 (1935). For further discussion and
references on this topic, see M. O. Scully, R. Shea, and J. D. McCullen, Phys. Repts. 43,501 (1978).
See also B. d'Espagnat, "The Quantum Theoryand Reality," Scientific American (Nov. 1979), p. 158.
[Note: The configuration described in Problem 11.81 is due to D. Bohm (1952).) These topics are
returned to in Section 11.13.
552 Chapter 11 Elements of Matr ix Mechanics. Spin Wavefunctions

Answers
(a) From the forms of the spinors ax, f3x as given in Problem 11.45, we obtain

I
ax = ,j2(az + f3z)
I
f3x = ,j2(az - f3z)

Substituting into ~A (x) gives ~A (x) = a~A (z) , where a is a constant.

(b) As pointed out previously in Fig. 9.1 and Section 10.2, the null eigenstates of
angular momentum are common eigenstates of the individual Cartesian com-
ponents of angular momentum .

11.83 Employing the rotation operator obtained in Problem 11.76 relevant to spinors,
show that the null eigenstate ~A (x) of 5x and 52, introduced in Problem 11.82,
is invariant under rotation about the z axis.

Answer
The rotation operator is given by

R,p = cos 2I/> i + I. sm. 20"z

I/> A

When applied to the product state ax (l)f3x (2), we obtain

R,p[a x (l)f3x (2)] = [R,pax (l)][R,pf3x (2)]

= [(cos~)ax(l)+i (sin~)f3x(l)]
x [(cos~) f3x(2) +i (sin~) ax (2)]
Applying a similar rotation to the product ax (2)f3x (l ) and carrying out the multi-
plication gives

R,p[ax(l)f3x(2) - ax(2)f3x(l)] = (sin 2 ~ + cos 2 ~ ) [ax (l)f3x (2) - ax (2)f3x(l)]

R,p~A (x) = ~A (x)

[Note: This problem establishes that zero spin states are invariant under rotation of
coordinates and thus transform as a scalar. A like quality is shared by the null orbital
angular momentum states which, we recall, are given by any spherically symmetric
function f (r 2 ) . See discussion preceding (10.34). For both null spin and null orbital
angular momentum states, (lx) = (J y ) = (iz ) = O. There is no preferred direction
for a system in any case of these states.]
11.84 One of the puzzles of the early theory of neutron decay, n -+ p + e, was the
fact that such a process could not conserve spin angular momentum . The neutron
n, proton p, and electron e each have spin 1.
To answer this objection , Pauli l9

19W. Pauli , Rapports du Septieme Conseilde Physique, Solvay, Brussels, 1933, Gauthier-Villars, Paris ,
1934.
 11 .12 Other "Pictures" in Quantum Mechanics 553

proposed that together with the proton and electron, a massless, chargeless, spin-!
particle was emitted, which he called a neutrino.
(a) Explain how the original process cannot conserve angular momentum .
(b) Explain how the corrected process, n -+ p + e + V, can conserve angular
momentum.

11.12 • OTHER "PICTURES" IN QUANTUM MECHANICS

Schrodinger, Heisenberg, and Interaction Pictures

In addition to representations in quantum mechanics stemming from transforma-
tion of bases in Hilbert space, one also speaks of different "pictures" of the theory.
These alternative formulations stem from the fact that wavefunctions and opera-
tors are not objects of measurement. As we have found previously, the superposi-
tion theorem specifies the possible outcome of a measurement in terms of certain
projections in Hilbert space . Thus, for example, if Ala) = ala) , and the system is
in the state lfr, the probability that measurement of A finds the value a is given by
the absolute square , I(a Ilfr) 12 .
The so-called Schrodinger picture refers to the formulation which is based on
the Schrodinger equation (3.45) . With the preceding description one requires that
any alternative picture satisfy two basic requirements: (1) Eigenvalues of opera-
tors corresponding to observables in the new picture must be the same as in the
Schrodinger picture ; (2) inner products of wavefunctions must maintain their val-
ues as well.
In Section 11.3 these properties were found to be obeyed by any unitary trans-
formation. In particular, the so-called Heisenberg picture stems from the unitary
operator (3.66) . If initial time t = 0 is labeled t = to, then this picture becomes

,
Vet, to) = exp [ -net
i - to)H A]
(11.129)
tr«, to) = exp [~(t - to) Ii] = V-I(t, to)

where we have assumed that Ii is not an explicit function of time. With (3.70) we
may write

lfr(t) = 0 (r , to)lfr(to)
(11.130)
lfr(to) = 0- 1(z, to)lfr(t)
Let lfr' be a wavefunction in the Heisenberg picture. Then

lfr' = O-Ilfr
(11.131)
olfr' = lfr
554 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

It follows that

v/(t) = U- I1{!(t) = U- I1{!(to)

(11.132)
1{!'(t) = 1{!(to)

That is, in the Heisenberg picture the wavefunction remains constant. On the other
hand, operators, which in the Schrodinger picture are constant, vary in time in the
Heisenberg picture. This follows from the transformation (11.40) .

A'(t) = U-I (t)AU(t) (11.133)

An important exception to this conclusion is the Hamiltonian. If iI is constant in

the Schrodinger picture , it remains constant in the Heisenberg picture.
An equation of motion for an operator in the Schrodinger picture is given by
(6.68). To obtain an equation of motion for an operator in the Heisenberg picture,
first with (11.133), we write
A A 1 A A

dA' = dV- AU + ir:' A dV + O:' oA U (11.134)

dt dt dt ot
In obtaining the last term in this equation, we have noted the following. Consider,
for example, that A = A(q, p, t) . Then in the Schrodinger picture, as q and pare
stationary in time,

oA(q , p, t)
(11.134a)
ot

To find a relation for the time derivation of U, we consider the following sequence
of events . Consider a particle which at the time to is in the state 1{!(to). At time
tl > to it evolves to the state

At time t: > tl it is in the state

1{!(tz) = U(tz, tl)1{!(tr) = U(tz , tr)U(tI , to)1{!(to)

= V (tz , to)1{!(to)

Thus we may write ZO

(11.135)

Recalling the definition of differentiation, we write

dUCt , to) = lim V(t + e, to) - V(t, to)

(11.136)
dt 0--->0 e
20Steps leading to (I 1.135) remain valid for iI an explicit function oftime. See Problem 3.18 .
11.12 Other "Pictures" in Quantum Mechanics 555

which, with (11.135), gives

c« + B, to) = V(t + B, t)V(t, to)

Substitution into (11.136) gives

dUCt, to) = lim [U(t + B, t) - l]U(t, to)

(11.137)
dt 8~0 B

For sufficiently small B we may write

V(t+B,t) =exp
(-iBH)
-n- = 1- TiBN
When substituted into (11.137), this relation gives

dU A A

i n - = HV (11.138a)
dt

Taking the Hermitian adjoint gives

At
dV At A
-in-=V H (11.138b)
dt

Substitution of these latter two equations into (11.134) gives the desired result:

dli' ali'
+ in-
A A

i n - = [A' , H '] (11.139a)

dt at

where we have set

ali' A_1aAA
-=V
at
-v
at
(11.139b)

Note in particular the similarity between (11.139a) and the equation of motion
in the Schrodinger picture (6.68). Properties of wavefunctions and operators in
these two representations are illustrated in Table 11.4. The right-hand column in
this table refers to operators that are time-independent in the Schrodinger pic-
ture. It should be noted, however, that operators may be time-dependent in the

TABLE 11.4 Elements of Schrodinger and Heisenberg pictures

Wavefunction Operator
Schrodinger picture Varying Constant
Heisenberg picture Constant Varying
556 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Schrodinger picture. Such cases are discussed in Section 13.5 as well as in the
description below of the interaction representation.
The Heisenberg equation of motion (11.139) often comes into play in discus-
sions concerning the correspondence principle. If one makes the identification
I A A

i/j,[A, HlQM -+ (A, H}cL (11.140)

then the equation of motion for an operator in the Heisenberg picture (11.139a) is
seen to reduce to the corresponding classical equation of motion for a dynamical
function as given in Problem 1.15.21

Interaction Picture
Finally, we tum to an approximation scheme important to time-dependent pertur-
bation theory known as the interaction picture . Consider the Hamiltonian

H = Ho + AV (t)
where the "unperturbed" Hamiltonian Ho is assumed to be time-independent and
A is a nondimensional parameter of smallness.
In analogy to the Heisenberg picture (11.131), wavefunctions in the new pic-
ture are given by
A _I
1/11 = Vo 1/1 (11.141)

where

A i - to)Ho
Vo = exp -net[ A ]
(I 1.141a)

and, in analogy with (11.133),

(11.142)

Taking the time derivative of (11.141), and employing the Schrodinger equa-
tion (3.45) and noting that

[Vo, Hol = 0

gives the desired equation of motion,

(11.143)

which is Schrodinger-like in form but only involves the interaction potential VI.
Integrating the preceding equation, we obtain
21For further discussion, see R. L. Liboff, Found. Phys. 17, 981 (1987) .
Problems 557

A
1/11 (t) = 1/11 (to) + :-
In
1 10
1
A
VI (t / )1/11 (t / ) dt / (11.144)

This equation leads naturally to a series solution. Namely, substituting 1/IICt) as

given by the RHS of (11.144) for 1/11 «,
in the integrand and continuing this
iteration gives the series

1/11 (t) = 1/11 (to) A

+ :-;
In
1 1

10
dt / VI (t )1/IICto)
A /
(11.145)

A
+ ( :-
In)
2
1 1
10
I
dt
/
I'

10
/l A / A/I
dt VI (t )VI (t )1/IIC to) + ...

Other techniques in time-dependent perturbation theory are discussed in Sec-

tion 13.5 et seq. Application of the preceding formalism is given in Problem
11.102.22

PROBLEMS

11.85 With regard to the interaction picture, derive (11.143) starting with (11.141).
11.86 Show that the matrix representation for p, (11.48), is Hermitian .
11.87 (a) Integrate Heisenberg's equation of motion (11.139a) for a free particle to obtain
q(t) and p(t) as functions of q(O) and p(O), respectively.
(b) Show that in this case

in
= --t
A A

[q(O) , p(O)]
m

(c) How does your answer to part (b) relate to a localized free-particle wave packet
propagating in one dimension?

Answers (partial)
(a) We find

p = p(O), q(t) = q(O) + p(O)t

m
(c) We obtain ~q(O) ~q(t) 2: tu12m, indicating that the wave packet spreads in
time.

11.88 (In the following three problems, hats over operators have been deleted.) Show that

(a) [Xi , F(p)] = in-

aF
api
aG
(b) [Pi, G(x)] = -in-
aXi

22The interaction representation is encountered again in Chapter 13 (Problem 13.57) and in Section
14.6 in derivation of the Lippmann-Schwinger equation.
558 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

where xi and Pi denote Cartesian components of respective operators and the func-
tions F(p) and G(x) have power series expansions.
(c) Evaluate the operator [Px, e- K X ] in the Heisenberg picture.

11.89 Assume a free-particle Hamiltonian , with unprimed and primed variables denoting
operators in the Schrodinger and Heisenberg pictures, respectively.
(a) How are the operators P and pi related?
(b) How are the operators x and x' related ?
[Hint: You should find (a) p = pi and (b) x' = x + (pt lm) , which, with Problem
11.87, indicates that x(t) = x' (0). This result is consistent with the property that
operators in the Schrodinger picture are constant in time. See Table 11.4.]

11.90 In Problem 10.73 it was found that the radial momentum operator in cylindrical
coordinates is ambiguous.
(a) Working in cylindrical coordinates, what is the Hamiltonian , H(p , z. cjJ), of a
particle of mass m moving in the potential field, V (p, z, cjJ) ?
(b) Employing this Hamiltonian , write down the Heisenberg equations of motion
for momenta, Pp, Pz» PI/>, respectively. Do ambiguities arise in this representa-
tion?
[Hint: In part (b), for evaluation of PI/> , recall discussion of Problem 9.16(a). In the
evaluation of PP and pz, recall Problem 11.88(b).]

11.13 • POLARIZATION STATES. EPR REVISITED

In this section we introduce eigenstates of photon polarization. This formalism is

then applied to the Einstein-Podolsky-Rosen (EPR) paradox (see Problem 11.81)
and closely allied notion of hidden variables briefly discussed in Chapter 2.

Polarization States
We recall that photons have zero rest mass and are spin-I bosons with spin either
aligned or antialigned with photon linear momentum.
The states of polarization of a photon may be written in terms of the two basis
eigenstates IH) and IV), representing horizontal and vertical polarization, respec-
tively. With x and y axes taken as horizontal and vertical directions, respectively,
related electric fields at fixed z are given by

8H = 80[coswt , 0, 0] ::::} IH)

(11.146)
8v = 80[0, cos on, 0] ::::} IV)

Let us consider the field of the superposition state

IR) = ~(IH) + ilV)) (11.147)

11.13 Polarization States. EPR Revisited 559

IR> IL> IH> Iv>

FIGURE11.15 Diagrammatic representation of four photon states of polarization (view-
ing the photon head-on).

Since i!V) = exp(in j2)!V) corresponds to !V) shifted by n j2 radians, the field
corresponding to IR) is

8R = 8o[coswt, sinwt, 0] :::} IR) (11.148)

which we recognize to be right-handed, circularly polarized radiation-viewing

the photon head-on . The wave (11.148) propagates in the +z direction . The
closely allied field

8L = 8o[coswt, - sinwt , 0] :::} IL) (11.149)

represents left-handed , circularly polarized radiation corresponding to the super-

position state

1
IL) = J2(IH) - i!V)) (11.150)

Note that these polarization states have the orthonormal properties

(HIH) = (V!V) = (RIR) = (LIL) = 1

(11.151)
(H!V) = (RIL) = 0
The preceding states of polarization are depicted graphically in Fig. 11.15.

Experimental Setup
With these preliminaries at hand, the question we examine relevant to the EPR
paradox is whether quantum theory is consistent or whether it is necessary to
bolster the theory with hidden variables. An operational means of answering this
question is given by Bell's theorem .P
Two processes relevant to this theorem are considered. The first of these is de-
scribed by the experimental setup depicted in Fig. 11.16 in which an atom under-
goes a two-photon decay with no net change in angular momentum. The notation
is such that IR I) is the polarization state of photon 1, etc.
23J. S. Bell, Physics 1, 195 (1964). This paper and other early works of Bell are reprinted in Quantum
Theory of Measurement , J. A. Wheeler and W. H. Zurek, eds., Princeton University Press, Princeton,
N.J., 1983.
560 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

: :::~:::::---- -® M= 0
.............
I R» ,I R2>•

IL» ,orIL2>

FIGURE 11.16 Schematic of two-photon atomic decay and possible circular polariza-
tion states of emitted photons .

The zero net change in angular momentum of the atomic decay indicates that
the photons must be in the symmetric state [analogous to the antisymmetric spin
state of Problem 11.81(a)]:

(11.152)

In this expression IRl) and IRz) are antialigned states so that their total angular
momentum is zero. The same is true of ILl) and ILz).
Now consider a polarizer which may be set to pass only IR), IL), IH) , IV),
or IHe). Here we have written IHe) for a linearly polarized state at an angle (J
from the horizontal. Two such polarizers are placed in front of photon detectors
as shown in Fig. 11.17. Note in particular that the photons emanating from A in
Fig. 11.17 are in the superposition state (11.152).
The polarizers are set so that P, only passes IHel) and Pz only passes IH o) .
(The motivation for the special notation 8z is given below.) The counter C re-
sponds positively when photons are detected in D, and Dz simultaneously.
We wish to obtain the number of coincidence counts N as a function of the
angles (J, and 8z. To obtain this relation , first we recall the superpositions (11.147)

~ D2
~-0 . .~ D, f--

FIGURE 11.17 Photons from source A pass through the polarizers P to detectors D.
Signals are then transmitted to coincidence counter C.
11 .13 Polarization States. EPR Revisited 561

and (11.150):

1
IR) = J2(IH) + ilV)
1
IL) = J2(IH) - ilV)

Substituting these transformations into (11.152) gives

(11.153)

The two-body polarization state so written indicates, for example, that if the po-
larization of photon 1 is measured to be HI , then measurement of the polarization
of photon 2 is certain to find H2.
The function N (01, 02) we seek is given by

(11.154)

We note that photon polarization states rotate like a spin-l vector.i" Thus

(11.155)

where IVo) represents a polarization state at 0 radians from the vertical. There
results

IHo) = cosOIH) - sinOIV) (11.156)

These expressions together with (11.153) give

2 - -
I(Ho,Ho) PI2 )1 = l(cosOt{HtI- sinOI(VII)(cos02(H21- sin02(V21)
2
x (IHtlIH2) - 1V1)1V2)1
= [cos 01 COS02 - sinOI sin 02]2 = COS 2(01 + 02) (11.157)

Polarizer Angles
An important point is now made concerning the appropriate angle for polarizer 2
(in the configuration of Fig . 11.17). We have been working in the convention
where the field description corresponding to the polarization state of a photon is
written with respect to viewing the photon head-on. The Cartesian axes at detec-
tors DI and D2 are shown in Fig. 11.18.
With reference to this diagram we see that the angle polarizer 2 makes with the
common axis defined by polarizer 1 is 02. From the diagram we also note

24See Section 11.5 concerning the rotation operator.

562 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

IH e2 >
y

D1 I

~
~
x
\ Common tox
kEB axis of (b)
k@

(a) (b)

FIGURE 11.18 (a) Coordinate frame at Vz. The propagation vector k is into the page .
(b) Coordinate frame at VI . The vector k is out of the page. Note that looking at (a) from
the back of the page gives a right-handed frame with k pointed toward the reader. It is with
respect to this frame that H02 is defined in (Il.157).

Thus (11.157) gives

N(OI, OZ) ex: COSZ(OI - Oz + JT) = COSZ(OI - Oz) (11.158)

This is our desired relation. It is plotted in Fig. 11.19. The functional dependence
of N on (01 - Oz) reflects the fact that the experimental arrangement of Fig. 11.17
is invariant with respect to rotation about the axis DzADI.
For X == 01 - Oz = 0, polarizers PI and Pz are aligned and there is maximum
coincidence. As is evident from (11.153), if measurement of photon 1 finds HI,
then measurement of photon 2 is certain to find Hz. At X = JT/2 polarizers are
set at right angles and there is no coincidence.

NI2 x
FIGURE 11.19 Coincidence counts N as a function of the angle X == 91 - 9z.
11.13 Polarization States. EPR Revisited 563

For further reference we set

1
N(X) = - cos 2 X (11.159)
2
See Fig. 11.19. Application of this result is made below.

Bell's Theorem
For discussion of Bell's theorem, first we return to the configuration of Problem
11.81 relevant to electron-positron decay. The appropriate antisymmetric spin-
zero state is given by

~A(1, 2) = ~[a(1),B(2) - a(2),B(1)] (11.160)

Consider that two Stem-Gerlach devices are positioned at equal but opposite
distances from the source A, as illustrated in Fig. 11.20. The setup is such that
S-O(1) detects the projection of particle spin on the unit vector a. Similarly,
S-0(2) detects the projection of particle spin on the unit vector b. We consider
the expectation

(11.161)

where ir is the Pauli spin matrix (11.82) . Inserting the spin state (11.160) into the
preceding gives (see Problem 11.103)

PQM(a, b) = -a· b (11.162)

Here we have inserted the subscript QM to denote that the expectation is quantum-
mechanical.
A formulation of this average stemming from hidden variables proceeds as fol-
lows. We introduce a hidden variable Asuch that to each state of the two-particle
system there corresponds a definite value of A. Let p(A) denote the probability
density of A. Then with A and B denoting the results of measurements of the
spin projections onto the directions a and b, respectively, the expectation P (a, b)
becomes

Ph (a, b) f
= o. p(A)A(a, A)B(b, A) (11.163)

FIGURE 11.20 TwoS-Gdevices situated at equalbutopposite distances fromthe decay

process at A .
564 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

where h denotes "hidden variables ," and

f p(A) d): = 1 (11.164a)

Furthermore, with the above interpretation of A and B we set

IA(a' , A)I S 1, IB(b' , A)I s 1, for all a', b' (11.164b)

Note that in writing (11.163), it is tacitly assumed that measurements of the two
spin projections do not influence one another. With apparatuses for these mea-
surements spatially removed from one another, such as in the present situation,
the preceding assumption is an example of the principle of locality, which is a
basic premise in physics.
Stemming from (11.163) and the conditions (11.164b), we obtain Bell's in-
equality (see Problem 11.105).

S = IP(a, b) - P(a, b')1 + IP(a' , b') + Pea' , b)] s 2 (11.165)

where a', b' are arbitrary unit vectors . Note in particular that this result is inde-
pendent of the probability density peA).
We may also fonn the term (11.165) employing PQM(a, b) as given by
(11.162). Thus , the cos- l (a' . a) = if>, cos- l (a . b) = 0, and cos-l(b. b') = y ,
and assuming that these angles are coplanar, the difference S becomes

SQM = [cos e - cos(O + Y)I + Icos(O - if> + y) + cos(O - if»1 s 2 (11.166)

In what sense is the expression (11.166) quantum mechanical? The answer is

that it stems from the quantum mechanical expectation (11.161), which in tum
involves the superposition state ~A given by (11.160). This state, as described in
Section 5.1, is purely quantum-mechanical. It describes each particle as being
partly in the ex ("spin up") state and the f3 ("spin down") state .
With specific choice of unit vectors : b = a', a . b = b . b' = cos 0, the pre-
ceding relation becomes (see Problem 11.106)

SQM = 2cosO - cos 20 S 1 (11.167)

This inequality is violated over an interval of 0 with maximum violation at

o = lr / 3, where SQM = ~ . We may conclude that quantum mechanics is in con-
flict with assumptions employed in the derivation of Bell's inequality (11.165)
(e.g ., the principle of locality) . The proof of the existence of such conflicts with
quantum mechanics is the essence of Bell's theorem.
We return now to the two-photon decay process, discussed at the start of this
section. In applying Bell's analysis to this process, the unit vectors , a, b, intro-
duced above, are associated with respective polarization axes of the two polariz-
ers (see Fig. 11.17). In this representation relevant to Bell's analysis, the particle
decay process depicted in Fig. 11.17 and the two-photon decay process depicted
Problems 565

in Fig. 11.20 are effectively equivalent. Thus, one expects the two-photon decay
process also to carry an inequality violation paralleling (11.167) .
A generalization of Bell's inequality (11.165) was obtained by Clauser and
Hom .25 When applied to the two-photon decay process one finds

3N(X) - N(3X) :s 1 (11.168)

where N(X) , as given by (11.159), is interpreted as probability of coincidence.

Substituting this value for N (X) into the preceding equation indicates that the
inequality is violated over angular intervals at X 2: 0 and X :s T(. Again, we are
confronted with conflicts between assumptions leading to the inequality (11.168)
and quantum mechanics.
However, it should be borne in mind that the preceding statements of inequality
are idealized in the sense that they do not incorporate limitations of the apparatus
of an actual experiment. Results to date, stemming from experiments performed
in the laboratory, indicate that quantum mechanics is consisrent.I"

PROBLEMS

11.91 In actual S-G experiments, measurement of particle spin through direct observa-
tion of deflection of such particles runs into difficulty with the uncertainty relation
if particle mass is too small.27 Thus, in the original S-G experiment, electron spin
was measured through observation of a beam of silver atoms. Such atoms contain
an uncoupled electron in the 5s shell.
Consider a beam of particles of mass M and spin !,
propagating in the +x
direction . The beam has cross section d. It interacts with an S-G apparatus whose
field is in the z direction . Employing relevant uncertainty relations, show that the
smallest uncertainty in normal displacements (Az) grows large with decreasing
mass M .

Answer
The uncertainty relation indicates that the minimum perpendicular component of
momentum of a particle in a beam of cross section d is

n
pz ~ d
Let the beam pass through the S-G apparatus in time t . Then the related spread in
normal displacements of particles is given by

25J. F. Clauser and M. A. Horn. D 10, 526 (1974). See also, L. E. Ballentine, Quantum Mechanics,
Prentice-Hall, Englewood Cliffs, N.J., 1990, Chapter 20.
26A compilation of experimental results in this context is given by B. d'EspagnatScientijic American
241, 158 (Nov. 1979). See also, J. F. Clauser and A. Shimony, Repts. Prog. Phys. 41, 1881 (1978) ;
M. Redhead, Incompleteness, Nonlocality, and Realism, Clarendon Press, Oxford , 1987, Chapter 4.
271 am indebted to Norman Ramsey for bringing this problem to my attention and for reference to
Pauli's discussion on this topic; see Hand. der Physik XXIV, 1st part, 83-272 (1933) .
566 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

il :c: pz t = ~t
Z M Md

We may view t as an interval over which energy states of the particles are mea-
sured. For the case at hand, particles have two energy states and ilE = E+ - E_ .
With Problem 5.28 we then write

Ii
t>-
'" ilE
Substituting in the preceding equation gives

lit 1i2
ilz~-> ---
- Md '" MdilE

Thus we find that the smallest transverse uncertainty spread in the beam grows
large with decreasing mass.
11.92 (a) How many spin states are there in the uncoupled representation for three elec-
trons?
(b) Calling individual particle states a(1), ,B(1), a(2), etc., write down these un-
coupled spin states.
11.93 With the states found in Problem 11.92 used as basis states, employ a Clebsch-
Gordan expansion to find the normalized coupled spin states for three electrons.
In listing these states, in addition to spin quantum numbers (s, ms), designate
also sl.2, the quantum number corresponding to the subcomponent spin, sI,2 =
sl + s2 ·
Answer
With the notation

(a,B,B) == a(I),B(2),B(3)
the eight coupled spin states are given by

I S I 2, s , ms ) = II,~,~) = (aaa)
3 I) I
II, 2' 2 = .J3[(aa,B) + (,Baa) + (a,Ba)]
3 I) I
1, 2' -2 = .J3[(,B,Ba) + (a,B,B) + (,Ba,B)]
I
11,~, -~) = (,B,B,B)

II, ~, ~) = !f(aa,B) - h[(,Baa) + (a,Ba)]

II,~, -~) = h[(a,B,B) - (,Ba,B)]- !f(,B,Ba)

Problems 567

1 1) 1
I0, 2' 2 = "j2[(a,Ba) - (,Baa)]

1
0, 2' -21) = "j2[(a,B,B)
1
- (,Ba,B)]
I
11.94 An important theorem in quantum mechanics addresses coupled spin states for
three electrons. The theorem states that a coupled spin state which is antisymmetric
with respect to all three particles vanishes identically.28 Another theorem states
that if u (1, 2, 3) is antisymmetric with respect to interchange of particles 1 and 2,
then

w(l, 2, 3) = u(l, 2, 3) + u(2, 3,1) + u(3, 1,2)

is totally antisymmetric.
Consider the spin state 10, i,i)
given in Problem 11.93, which is antisymmet-
ric with respect to particles 1 and 2 but not 3. Show that construction of the totally
antisymmetric function w(1, 2, 3) for this case gives w = O.

11.95 (a) Repeat Problem 11.92 forfour electrons.

(b) What are the possible total s values for this situation?
(c) Working in the notation of Problem 11.93, in the coupled representation, write
down an antisymmetric spin state for four electrons. For which particles is
your state antisymmetric?

11.96 In relativistic formulations one works with observables called four-vectors . For
example, an event which occurs in a frame S at the coordinates x, y, z at the time t
is represented by the four-vector X = (x, y, z, ict). Let the frame S' be in relative
motion with respect to S with velocity v parallel to the z axis . If at t = 0 the two
frames are coincident, then the event X as observed in S' has components X' =
(x', v', z', icr') which are given by the Lorentz transformation [where ,B == vic
and y2 == 1/(1 - ,B2)]

o o
X'=x, -- 0
L- (~ o
1 o
Y
o o -i,By

(The Lorentz transformation is discussed in detail in Section 15.1.)

(a) Show that i. is orthogonal and that i. preserves the inner product (i.e., XI'
X2 = X'1 . X'2)·
(b) Show that the Lorentz transformation reduces to the Galilian transformation,
z' = z- vt, t' = t, in the limit,B « 1.

11.97 With reference to (11.140) it is generally the case that the commutator [A, B] is
equal to in times the Poisson bracket considered as an operator. 29 For this prop-

28Por further discussion, see R. L. Liboff, Am. J. Phys. 52,561 (1984). This topic is returned to in
Section 12.4.
29Por further discussion, see D. Bohm, Quantum Theory, Prentice-Hall, New York, 1951.
568 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

erty to be satisfied it is required that the Poisson bracket be symmetrized in its

arguments . Show that

Answer
First, we write

where the symmetrized term on the right is considered as an operator. For the
commutator we find

which is seen to agree with the given rule.

11.98 Establish the following two vector-operator equalities:
(a) [j . A, j] = j x A
where A is a constant operator.
(b) [ix , Vyj = iliVz
where Vis a vector operator.
11.99 Write down the Clebsch-Gordan expansion for two coupled electrons with j
2, mj = 0, I = I , s = I, in terms of the product states Ilm/)Ism s) . Evaluate
coefficients .
11.100 Show that the total angular momentum of two fennions is always an integer. (Note:
This problem establishes that two coupled fermions constitute a composite boson.)

Answer
The total angular momentum operator of two fennions is

where L is the orbital angular momentum of the two-particle system . With s\

n \ / 2 and S2 = n2/2 , where n \ and n2 are odd integers, we find the total spin
numbers
_ (~_ n 2) (~n2)
s- 2 2 '· · · ' 2 + 2

which are all integers . Total j is then given by

j=l- s, ... ,l+s

which , for all s given above, are integers .

11.101 Employing the vector operator formula given in Problem 11.68, show that

{u. [p - (e/c)A]}2 [p - (e/c)Aj2 _ ~u. B

2m 2m 2mc

where Ais vector potential.

Problems 569

11.102 A one-dimensional harmonic oscillator is initially in the state 11Jr(O») II) [in
the notation of (7.61)]. The system is acted upon by an exponentially decaying
potential in time. The total Hamiltonian is given by

where Vo « IiwO and r is a constant relaxation time.

(a) Working in the interaction picture, calculate 11Jr/ (t») to first order in A -
VO/IiwO ·
(b) From your answer to part (a), obtain the wavefunction in the Schrodinger pic-
ture, 11Jr(t») .
(c) What is the probability Pl--+ n that a transition to the state In) has occurred
after time t due to the perturbation?

Answers
(a) First note that

11Jr/(O») = 11Jr(O») = 11)

and

Vj(t)I1Jr/(O») = eitHo/IiVe-itHo /lill)

= eitHo/li(a +at)Voe-t/re-i3wot /211)
= eitHo/Ii(IO ) + J212)Voe-t /re-i3wot /2
= Vo[e-(t /r)-iwotIO) + J2e-(t /rHiwotI2)]
Substituting this expression into (11.145) and integrating, we obtain
0\1,
11Jr/(t») = 11) + - i i0i [A(t)IO) + A*(t)J212)]
wo

where

e-t,p - 1
A (t) == ---,--,-----
¢ /wo
1
¢== -r
0

+IWO

(b) This answer is obtained by substituting the preceding expression in the inverse
of (11.141). There results

11Jr(t») = 11) + i Ae-itwo / 210) + A *e- itwo / 2J212)

(c) We may view the preceding result as an expansion of 11Jr(t») in the basis {In)}
and further consider the state 11Jr(t») as given in part (b) to be normalized to
O(A), where A == IiwO/ VO . There results
570 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

PI-->I = I + O(),h
2 2 2
PI-->O = 1(01J/t(t»)1 = J.. 1A 1 + O(J..3)
2 2 2
PI-->2 = 1(21J/t(t»)1 = n 1A I + O(J..3)

where

I - e- 2t / T - 2e- t / T coswot
IA 12 = ---------;:----=--
1+ (1/wor)2

Thus we find it is most probable for the oscillator to remain in the first excited
state . The probability for a transition to the second excited state is twice as
probable as that for a transition to the ground state.

11.103 Employing the spin-zero antisymmetric state (11.160), evaluate (11.161) to obtain
(11.162). Hint: See Problem 11.68.

11.104 Show that IPJ2} as given by (11.153) follows from (11.152) through the transfor-
mation (11.147-150).

11.105 Employing the conditions (11.164b), establish Bell's inequality (11.165).

Answer
Introduce the notation A(J.. , a) = A(a) . First we note

IP(a, b) - P(a, b/)1 = If A(a)B(b)p(J..)dJ.. - f A(a)B(b/)p(J..)dJ..

± [f A(a)B(b)A(a/)B(b/)p(J..) d):
-f A(a)B(b/)A(a/)B(b)P(J..)dJ..] 1

= If A(a)B(b)[l ± A(a/)B(b/»)p(J..)dJ..

-f A(a)B(b/)[1 ± A(a/)B(b)]P(J..)dJ..1

.s f IA(a)B(b)I[1 ± A (a/)B(b/»)p(J..) dJ..

+f IA(a)B(b/)I[1 ± A(a/)B(b»)p(J..)dJ..

:::: 2 ± [P(a/, b') + n«, b)]

where we have used IABI :::: 1. Thus

IP(a, b) - P(a, b/)1 ± [P(a/ , b/) + ri« . b») :::: 2

which implies (11.165) .
 11.14 The Transfer Matrix 571

11.106 Derive (11.167) from (11.166). Hint: Recall the rules:

(i) If IAI + IBI ::: n, then IA + BI ::: n .
(ii) If IAI ::: n, then -n ::: A.

11.14 • THE TRANSFER MATRIX

With matrix properties developed in this chapter, we return to Section 7.7 address-
ing the problem of transmission of particles through a potential barrier. Matrices
are introduced which connect coefficients of incident and reflected waves. Em-
ploying reflection symmetry (Section 6.4), time reversibility (Problem 6.33), and
conservation of current, properties of these matrices are developed which allow
wavefunctions to be constructed. In the following section these rules are applied to
the Kronig-Penny model (Section 8.2), and the general dispersion relation imply-
ing band structure is regained. This technique of solution is commonly referred
to as the transfer matrix method and is important in the study of quantum-well
microdevices" as well as in basic science.'!
We recall (7.138), which are generalized to read

(x < -a)
(-a < x < a) (l1.l69)
(x > a)

(with k; written as k and k2 written as K). These relations pertain to a potential

barrier of width 2a centered at the origin, x = O. Furthermore, (A exp i kx) is an
incident wave from the left, (B exp -ikx) is the reflected wave, and (F exp ikx) is
the transmitted wave. In Section 7.7, the transmission coefficient was obtained by
matching wavefunction components at x = -a and x = a. The present analysis
takes the point of view that F and G coefficients are given and that a matrix S
exists that determines the coefficients A and B in terms of F and G. This relation
is written

(BA) = (SII SI2)

S22
S21
(F)
G
=. S (F)
G
(l1.l70)

which defines the "transfer matrix" S. We wish to determine general properties of

the S matrix from fundamental properties of the system .

Reflection Symmetry
We note that the potential barrier of Fig. 7.22 is an even function of x . It follows
that if tp(x) is a solution of the wave equation, then so is tp( -x) . Replacing x by
-x in (l1.l69) gives
30R. C. Leavens and R. Taylor, eds., Interfaces, Quantum Wells and Superlattices , Plenum , New York,
1988.
31M. L. Mehta , Random Matri ces , Academic Press. Boston , 1991.
572 Chapter 11 Elements of Matri x Mechanics. Spin Wavefunctions

(x > a)
(-a < x < a) (11.171)
(x < -a)

In this relation (G exp ikx) is the incident wave from the left, (F exp -ikx) is the
reflected wave, and (B exp i kx) is the transmitted wave. It follows that the relevant
matrix equation connecting these coefficients may be obtained from (11.170) by
the repl acements A ++ G, B ++ F . There results

(G
F
) (SII
S21
S12)
S22 (B)A (l1.1na)

or, equivalently,

(GF) = (S22
SI2
S21)
SII
(A)
B
(11.1nb)

Substituting this relation into (11.170) gives

G~: ~~~) G~~ ~~:) = G n == i (11.173a)

where i is the identity operator. It follows that

05- 1 = (S22
SI2
S21)
SII
(11.173b)

lime Reversibility
With Problem 7.82 we note that the time-reversed solution corresponding to
(11.169) is given by

(x < -a)
(-a < x < a) (11.174)
(a < x)

In this picture (B* exp i kx) is the incident wave from the left, (A * exp -i kx) is
the reflected wave, and (G* exp i kx) is the transmitted wave. With reference to
(11.170) we write

(B:)
A
= (Sll
S21 S22 (G:)
S12)
F (11.175a)

Taking the complex conjugate of this equation and rewriting the relation in normal
form gives

(BA) (S22*
S12*
S21 *)
Sll*
(F)
G
(11.175b)
11.14 The Transfer Matrix 573

Comparison with (11 .170) gives

(l1.175c)

so that

(l1.175d)

With (11.173) these latter relations give

Sll
( SI2*
S\2) (Sll*
Sll * SI2
SI2*)
Sll =
(1
0
0)
1
(l1.176a)

Multiplying matrices we find

ISlll2+ IS121 2 = 1 (l1.176b)

S\\(S\2 + SI2*) = 0 (11.176c)

With these conditions on the components of the transfer matrix at hand, we tum
to the third invariance property of the system.

Current Continuity
For time-independent configurations in one dimension, the continuity equation
(7.97) gives constant current which, when applied to the wavefunction (11.171),
gives

(l1.177a)

or, equivalently,

(A (l1.177b)

Substituting the transpose of (11.170) and its complex conjugate into the left side
of the preceding equation indicates that

(11.177c)

This relation gives the additional condition

(l1.177d)

This latter equation together with (11.175d) and (l1.176c) indicates that SI2 is
purely imaginary.
574 Chapter 11 Elements of Matr ix Mechanics. Spin Wavefunctions

Writing

(11.178)

we may conclude that

(11.179a)

When combined with (11.1nd) the preceding relations give

2 2 2 (11.179b)
al + /31 - /32 = 1

We may then finally write

i/32 ) (11.180a)
al - i/31

With (11.179b) we find

detS = 1 (11.180b)

and with (11.173b) we find

S-I = S* (11.180c)

The transfer matrix and related properties given by (11.180) are derived from
fundamental principles and are independent of wavefunction boundary condi-
tions. The transfer matrix is relevant to a wave packet incident on a barrier of
arbitrary but smooth form, which vanishes at large displacement from the barrier
center. Wavefunctions far removed from the barrier have the structure given by
the first and third equations in (11.169). For general solution to a given problem
one must express the parameters of S in terms of energy, for which boundary
conditions come into play.
Let us apply the preceding formalism to construct expressions for the trans-
mission and reflection coefficients T and R, respectively, in terms of parameters
of a and /3. With reference to (11.170) and matrix values (11.180a), and in the
notation of Problem 7.44, we write
;;;; 'A, F 1
vTe''¥T = - = - - - : -
A al + i/31
(11.181)
.JR ei<PR = !!.- = - i/32
A al + i/31
These expressions, with (11.179b), return the equality
2
T+ R= 1 + /32 = 1 (11.182)
at 2+/31 2

[This equality follows also from (11. Ina) with G = 0.]

11.14 The Transfer Matrix 575

Kronig-Penny Model Revisited

We wish to reformulate the Kronig-Penny model in terms of the transfer matrix
formalism. Our aim is to rederive the basic dispersion relation (8.53) which, as
previously noted, implies a band structure of allowed energies.
For consistency, we choose the periodic potential in accord with (11.169) ,
namely, there is a potential barrier in the interval (-a , a) and a potential val-
ley in the interval (a , a + 2b) . The period of the periodic array is d = 2(a + b) .
In the valley domains, (a - d) < (x - nd) < -a, the potenti al V = O. Solutions
in these domains are given by the free-particle solutions

(11.183)

With barrier centers at x = nd, successive values of coefficients may be ob-

tained employing the transfer matrix S, as given by (11.180a) . There results

(11.184)

or, equivalently,

(l1.185a)

where the reduced transfer matrix, 8, is given by

,_(a
Q=
l -
'fJ
I 2e
ifJ))e ikd
-ikd (l1.185b)

With the constraint (11.179b), we find

det Q = 1 (l1.185c)

Note that consistent with the symmetry of the periodic array, 8 is independent of
8
the barrier number, n. Note further that may be rewritten,

(l1.185d)

In accord with Problem 11.22 we regain (11.185c). Furthermore, with reference

to Problem 4.34, we may write

8-1 = sb:'
which gives

(l1.185e)
576 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

Criteria for Bounded Wavefunction

Iterating (11.185a) we obtain

(11.186)

for wavefunction amplitudes in the nth cell of the lattice.

In the present formulation it is important to ensure that wavefunctions remain
finite in the limit of large n. This condition may be examined in terms of the
eigenvalues of Q, which in tum are roots of the characteristic equation

(11.187a)

or, equivalently

q2 _ 2q(al coskd + fJl sinkd) + 1= 0 (11.187b)

This quadratic expression has the roots

(11.187c)

We may choose constants of the initial state (Ao, Bo) so that these states are eigen-
states of Qcorresponding to the eigenvalues, q±. That is,

A [Ao(±)] [Ao(±)] (l1.188a)

Q Bo(±) = q± Bo(±)

For these solutions, (11.186) gives

(11.188b)

With (11.187c) we note that q± eigenvalues are real if ITrQI > 2. In this case
Iq±In goes to infinity in either of the two limits, n -+ ±oo, in which case wave-
functions grow unbounded. A finite solution is guaranteed, providing

(11.189a)

which is satisfied if, for real y,

I '
'2TrQ = cosy (11.189b)

which gives

(11.189c)
Problems 577

Recalling the expression for TrQ [see (11.187b)], the preceding identifications
give the following condition for allowed energy eigenvalues.

cosy = 0(1 coskld + 131 sinkld

01.190)
[cos y I .s: 1

To identify k with the valley wave vector, we have relabeled it k,. To proceed
further, boundary conditions are brought into play. With reference to (7.140) rel-
evant to propagation through a square barrier of width 2a, and (11.185c) relating
fJi parameters, one may express a, and fJi in terms of k-propagation parameters.
Substituting these values into (11.190) and setting y = kd, returns the dispersion
relation (8.53).32

PROBLEMS

11.107 Show explicitly that

(a) detS = I
(b) S-I = S*
where S is the transfer matrix (11.170).
11.108 Show explicitly that the reduced transfer matrix Q, (l1.l85b), has the following
properties :
(a) TrQ = 2[IXI cos(kd) + f3J sin(kd)]
(b) Det Q = I
(c) Q-I Q = j [where Q-I is given by (11.l85e)]

11.109 Employing (7.140) and (l1.l85c), express the coefficients IXI and f32 in terms of
k 1 and k2 wave vectors.
11.110 List three properties on which the derivation of the transfer matrix (l1.l80a) is
based.
11.111 An electron propagates through a semi-infinite periodic potential : V(x) = V(x +
a) for x ::: 0, V(x) = Vo for x < 0, where Vm < Vo and Vm represents the
amplitude of the periodic component. Does the transfer matrix (11.180a) apply to
this situation? Explain your answer.
11.112 Show that the expectation of an operator in the Schrodinger picture is equal to
expectation in the Heisenberg picture.

Answer
First note that if (1/J' 1= (1/J IU, then 11/J') = U- I I1/J ). There follows

32For further discussion, see E. Merzbacher, Quantum Mechanics, 2d ed., Wiley, New York, 1970.
Note: Merzbacher uses M, P where we use S. Q.
578 Chapter 11 Elements of Matrix Mechanics. Spin Wavefunctions

11.113 Consider a homogeneous rigid spherical particle of finite radius that is free to
rotate about a fixed center.
(a) What is the Hamiltonian of this system?
(b) What are the eigenfunctions of this system ?
(c) Discuss the correspondence between quantum mechanical and classical mo-
tion of this system .

Answers
(a) H = L 2 / 21 (This equation follows from the equality of the three components
of moments of inertia of a sphere .)
(b) Any wavefunction describing this system must be invariant under arbitrary
rotations about the origin. This leaves only the spherically symmetric S state,
corresponding to I = O. It should be noted that the same Hamiltonian as that
given above applies to any rigid body with three equal moments of inertia .
However, it is only for the rigid sphere that the said angular invariance applies .
(c) As a classical sphere may rotate at any angular momentum about a fixed cen-
ter, there is no correspondence with the quantum situation in which the S state
is the only permitted state.

11.114 A uniform non-interacting beam of electrons with an isotropic distribution of

spins , propagates in the z direction in a medium which includes a uniform mag-
netic field,

B = (Bo, 0, 0)

(a) Write down the Hamiltonian of an electron in this beam .

(b) What are the eigenvalues, E(k , s), and time-dependent eigenstates lj;(k , s, r)
of this Hamiltonian?
(c) If the translational motion of the beam is constant and Bo = 1.4 kG, at what
frequency would incident electromagnetic radiation on the beam be absorbed
(in Hz)?
CHAPTER Application to Atomic,
Molecular, Solid-State, and
12 Nuclear Physics. Elements of
Quantum Statistics

12.1 The Total Angular Momentum , J

12.2 One-Electron Atoms
12.3 The Pauli Principle
12.4 The Periodic Table
12.5 The Slater Determinant
12.6 Application ofSymmetrization Rules to the Helium Atom
12.7 The Hydrogen and Deuterium Molecules
12.8 BriefDescription of Quantum Models for Superconductivity
and Superjluidity
12.9 Impurity Semiconductors and the p-n Junction
12.10 Elements ofNuclear Physics. The Deuteron and Isospin

Having developed methods for addition ofspin angular momentum in Chapter 11

and properties of the three-dimensional Hamiltonian in Chapter 10, these for-
malisms, together with the Pauli principle, are now applied to some practical
problems. Symmetry requirements on the wavefunction of the helium atom im-
posed by the Pauli principle serve to couple electron spins. This coupling re-
sults in a separation of the spectra into singlet and triplet series for helium as
well as other two-electron atoms. Symmetrization requirements stemming from
the Pauli principle are also maintained in calculation ofthe binding ofthe hydro-
gen molecule. The relevance of Bose-Einstein condensation to superconductivity
and superjluidity is described. Application of the Pauli principle is further noted
in two examples. In the first of these, semiconductor theory, previously encoun -
tered in Chapter 8, is extended to extrinsic (impurity) conductivity, in which the
Fermi-Dirac distribution comes into play. The second example addresses nuclear
physics, and with concepts of the nucleon and isotopic spin, the totally antisym-
metric ground state ofthe deuteron is constructed.

12.1 • THE TOTAL ANGULAR MOMENTUM, J

In this section we consider the addition of spin and orbital angular momentum
for atomic systems. As previously noted, the total angular momentum of a system
579
580 Chapter 12 Application to Atom ic, Molecular, Solid-State, and Nuclear Physics

(e.g., an atom or a single electron) which has both orbital angular momentum L
and spin angular momentum S is called J.

(12.1)

It has components

A A A

Jy = L y + Sy

Its square appears as

j2 = i.2 + 1;2 + 2L . S (12.2)

In obtaining this expression we have used the fact that Land S commute.

L-S Coupling
Individual electrons in an atom have both orbital and spin angular momentum.
Among the lighter atoms, individual electrons' L vectors couple to give a resultant
L and individual S vectors couple to give a resultant S. These two vectors then
join to give a total angular momentum J (Fig. 12.1). This is called the L-S or
Russell-Saunders coupling scheme . I

FIGURE 12.1 Schematic vector representation of the L-S scheme of angular momentum
addition. J2 and J z are fixed, as are L 2 and S2.

I Thi s scheme is also relevant to "one- or two-electron atoms." More generally, in heavy element s with
large Z, the spin-orbit coupling (Section 12.2) becomes large and serves to couple L, and Sj vectors
of individual electrons. giving resultant J I . jz • . . . • values. These individual electron Jj values then
combine to give a resultant J. Thi s coupling scheme is known as i -i coupling .
12.1 The Total Angular Momentum, J 581

Eigenstates in this representation are simultaneous eigenstates of the four com-

muting operators

(12.3)

There are six pairs of operators in this set which must be checked for commutabil-
ity:
A2 A A2 A A
(i) [£,2 + 52 + 2L . S, £,2] = 0 (iv) [L ,Jzl = [L .t., + Szl = 0
A2 A A2 A A
(ii) [£,2 + 52 + 2L . S, 52] = 0 (v) [S , Jzl = [S . t., + Szl = 0
A2 A
(iii) [J ,Jzl = 0 (vi) [£,2,52] = 0

In (i), £,2 commutes with all its components. In (ii), 52 commutes with all its
components. The remaining relations are self-evident.
Eigenvalue equations related to the commuting operators (12.3) appear as

J2ljmjls} = fi2j(j + 1)ljmjls}

Jzljmjls} = fimjljm jls}

f2ljmjls} = fi 21(l + 1)ljmjls}

52lj m j l s} = fi
2s(s
+ 1)ljmjls}
For a given value of J, m j runs in integral steps from - j to + j .
A very important operator that commutes with all four operators (12.3) is L.S.
A A 1 A2 A2 A2
L· Sljm jls} = 2"(1 - L - S )Ijmjls}

= (fi2j2)[j(j + 1) -/(1 + 1) - s(s + 1)]ljmjls}

Eigen-j-values and Term Notation
In the L-S representation, I and s are known . What are the possible j values
corresponding to these values of I and s ? Since J is the resultant of two angular
momentum vectors, the rules of Section 9.4 apply. These rules indicate that j
values run from a maximum of

jmax = I +s
to a minimum of

jmin = 1/- s]
in integral steps.

(I + s) ~ j ~ II - sl
j = 1+ s, 1+ s - 1, 1+ s - 2, . . . , II - s] + 1, II - sl
582 Chapter 12 Application to Atom ic, Molecular, Solid-State, and Nuclear Physics

For s < I, there are a total of (2s + 1) j values . The number (2s + 1) is called
the multiplicity. In the section to follow, these different j values are shown to
correspond to distinct energy values for the atom. Thus for a one-electron atom
(s = i), a state of given I splits into the doublet corresponding to the two values

. 1
J = 1--
2
(Fig. 12.2). The term notation for these states is given by the following symbol
2s+1 J >J
where J: denotes the letter corresponding to the orbital angular momentum I value
according to the following scheme :

I (£.)
Letter
1-0--------
S
1
P
2
D F
3 4
G H
5 6
I K
7 8
L
9
M
10
N

The doublet P states of one-electron atoms are denoted by the terms

2pl /2 ,2P3/2

The doublet F states (l = 3) are denoted by

2F7/2,2 FS/2

In two-electron atoms , the resultant spin quantum number is either 0 or 1.

These, we recall (Section 11.10), are the resultant s values corresponding to the
i
addition of two spins . These two values of s give rise to two types of spectra
(this is the case, for example, for He):

s = 0 -+ singlet series : IS,lp,ID, .

s = I -+ triplet series : 3S, 3p, 3D, .

(3) (b)

FIGURE12.2 Diagrams depicting coupling of the Land S vectors of a single p electron,

in the L-S scheme. The doublet contains two values of j .
Problems 583

J.

3t1 311 311

211 211

11

o 0

(a) (b) (c)

FIGURE 12.3 Diagrams depicting coupling of L and S vectors for two electrons in an
orbital D state and a spin-I state. The resultant triplet of j values is

j=I .2,3

The j values of the 3D states are j = 1,2,3. These correspond to the states

3Dl, 3D2 , 3D3

In general, any state with l > 1 becomes the triplet

j=/+I,/,/-1

(Fig. 12.3).
The multiplicity corresponding to the case s > l is 2/ + 1. For example, if
s = ~ and / = 1, there are three j values: j = ~, ~ , However, the notation!.
for this state remains 4PS/2,3 /2 ,1/2 , with 4 written for 2s + 1, although in fact the
multiplicity is 2/ + 1.

PROBLEMS

12.1 (a) For given values of Land S, show that the four operators

{L' 2. i.' ; S' 2, Sz

'
}

form a commuting set of observables. This set is an alternative to (12.30) in the

L-S coupling scheme .
(b) Which of these operators are incompatible with those of (l2.3)?
584 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

12.2 Show that

(a) [j2, iz] = 2ine z • (1, x 8)
(b) [j2, Sz] = ut«, . (8 xL)
12.3 What are the multiplicities of the G(I = 4) and H (I = 5) states for the two spectral
series related to a three-electron atom? What is the complete term notation for all
these states ?

12.4 Wh at kind of terms can result from the following values of I and 5?
(a) 1= 2, 5 = '72
(b) 1= 5, 5 = ~
(c) 1= 3, 5 =3
12.5 What are the I , 5, j values and multiplicities of the following terms?
(a) 3D2
(b) 4PS/2
(c) 2F7/2
(d) 3G3

12.2 • ONE-ELECTRON ATOMS

In this section we consider the manner in which the spin of the valence electron in
one-electron atoms interacts with the shielded Coulomb field due to the nucleus
and remaining electrons of the atom. One-electron atoms are better known as the
alkali-metal atoms.? In such atoms all but one electron are in closed "shells" (to
be discussed below) . These "core" electrons, together with the nucleus, present a
radial electric field to the outer valence electron (Fig. 12.4). Furthermore, the total
orbital and spin angular momentum of a closed shell is zero, so that the angular
momentum of the atom is determined by the valence electron.

Spin-Orbit Coupling
The interaction between the spin of the valence electron and the shielded Coulomb
field arises from the orbital motion of this electron through the Coulomb field.
When an observer moves with velocity v across the lines of a static electric field
£$, special relativity reveals that in the frame of the observer, a magnetic field

i:J = -yfl x £$
v
fl =-,
c
is detected (Fig. 12.5). Keeping terms to first order in f3 gives
v
i:J = - - x £$
c
2This analysis is also relevant to hydrogeni c atoms (Section 10.6).
 8
Electric field 8, stemming
from the core is radial

FIGURE 12.4 Properties of "one-electron" atoms.

& (out of the plane

• • •
• • •
• • c.• •
• • • • • • •
of the paper)

•
•
•
•
• • • • • • • •
• • • • • • • •
• • • • .v • • •
• • • • • • • •
• • • • • • • • • • • •
• • • • • • • • • • • •
FIGURE 12.5 An observer in the (perfectly transparent) sphere which is moving with
velocity v across the electric field 8 detects a magnetic field
v
93 = - - x 8
c
586 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

FIGURE 12.6 The magnetic moment I.t of the satellite electron sees a magnetic field due
to its orbital motion across the radial, static Coulomb lines of force which emanate from
the nucleus. The resulting torque on the moment produces a precession of the spin axis of
the electron as shown.

It follows that (to this order) , if an electron moves with momentum p across a
field e, the electron will feel a magnetic field-'

:B = P
-- x & (12.4)
me
This is the nature of the magnetic field with which the magnetic moment of the
orbiting valence electron interacts (Fig. 12.6). The interaction energy between IL
and :B is given by (11.89), modified by the Thomas factor, !.
This correction
factor represents an additional relativistic effect due to the acceleration of the
electron.' Thus the interaction energy between the spin of the orbiting electron
and the magnetic field (12.4) appears as

H
I 1
= -'2IL.:B = 1
'21L'
(V-;; x e) (12.5)

1 IL
= -'2 me • (& x p)

In spherical coordinates, & has only a radial component

d
8 r = --<I>(r)
dr
where <I>(r) is the static Coulomb potential (ergs/esu) seen by the valence elec-
tron. Substituting this expression for & = (8 r , 0, 0) into (12.5) gives

H I = -1 - 1 [1- -
d<l>(r)]
- (r x p) • IL
2 me r dr

Recalling the linear relation (11.86) between IL and S,

3Relativistic momentum p = ymv, so that to terms of O(f32) , p = my . In these formulas m is the rest
mass of an electron : mc2 = 0.511 MeV. For further discussion see Chapter 15.
4L. H. Thomas, Nature 117, 514 (1926) .
12.2 One-Electron Atoms 587

1L = (:JS
and rewriting L for r x p permits H' to be rewritten in terms of its operational
equivalent

A e ,
H = - 2c2
2m
- - -
del>]
r dr
[1
L· S == !(r)L • S
A A A A
(12.6)

which serves to define the scalar function !(r).

Approximate Wavefunction
The Hamiltonian of a typical one-electron atom, neglecting the L-S coupling term
just discovered, appears as [recall (10.107)]

A p2 Pr 2 f2
Ho = 2m + V(r) = 2m + 2mr 2 + V(r)
where V = eel>, L is the orbital angular momentum of the valence electron , and
Pr is its radial momentum. The eigenstates of Ho are hydrogenlike in structure.
They are comprised of the eigenstates of i.2 (spherical harmonics) and solutions
to the radial equation [recall (10.109)] . Incorporating the spin-orbit interaction
(12.6) gives the total Hamiltonian
A 2 A2
A I Pr A L A A A

H=Ho+H = - + - - 2 +V(r)+!(r)L·S
2m 2mr

Rewriting L·S in terms of J 2 , i 2, S2 (12.2) permits this Hamiltonian to be rewrit-

ten

(12.7)

In the preceding section we showed that o:

jz, i.2 , S2) comprise a set of com-
muting operators. Since these operators also commute with Ho, approximate
eigenstates of H may be taken to be of the form'

(12.8)

where Inl} represents the radial component of the eigenstates of Ho:

Holnl} = En Inl}

For hydrogen, for example, Inl} are solutions to the radial equation (10.117), that
is, weighted Laguerre polynomials (Table 10.3).
5The states Ijmj Is ) may be constructed from the product states Ilmtllsms) in a Clebsch-Gordan
expansion (with m, + mi = m j ) '
588 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

Substituting the product form (12.8) into the Schrodinger equation

Iflcp} = Elcp}

with If given by (12.7) gives

! En + ~2 fer) [j(j + 1) -I(l + 1) - ~] !ICP} = Elcp} (12.9)

It follows that the product solutions (12.8) are not eigenstates of If (i.e., Iflcp) i=
constant x Icp}). But due to the fact that the spin-orbit correction to En is small
compared to En, they do serve as approximate solutions . Approximate eigenval-
ues of If may then be obtained by constructing the expectation of If in these
states.

Enlj = (cpIHlcp) (12.10)

= En + 2
2
li [ j(j + 1) -I(l + 1) - 4: 3] (f(r)}nl

Since j can have two values (l ±!) for a given value of I, it follows that an energy
state of given I separates into a doublet when the spin-orbit interaction is "turned
on." The two corresponding values of energy are

li2
Enl/+) == Ej=I+I/2 = En + 2I(f}nl
(12.11)
li2
Enl/-) == Ej=I-I /2 = En - 2(I + 1)(f}nl
An estimate of (f) nl may be obtained using hydrogen wavefunctions and assum-
ing the Coulomb potential for V,

Ze 2
V=--
r
where Z is an effective atomic number. (See Problem 12.13.) Substituting this
potential into f (r) as given in (12.6) gives

1 1 dV Ze 2 1
fer) = 2m2c2 -;: d; = 2m2c2 r3
2 2
(f}nl = Ze ('0 IRnl(r)1 r2 dr (12.12)
2m 2c2 10 r3
li2 (me 4 Z2/2li2n2)2
2n (f}nl = mc 2(l + !)(l + I)I
(where we have used the results of Problem 10.49).
12.2 One-Electron Atoms 589

Fine Structure of Hydrogen

Recalling the energy eigenvalues of hydrogen (10 .124)

permit s the correction factor (12.12) to be written

2n IEnl (Za)2 1
2
1(1 + !)(I + 1) mc n 1(1+ !)(l+ 1)

where a is the fine-structure constant (see Problems 2.20 and 2.29) ,

e2 1
a = lie = -13-7-.0-3-7 ' a 2 = 5.33 x 10- 5

Substituting these forms into the doublet energies (12.11) gives the values

"spin up"
j = 1 +! E . (+ ) _ _
nil -
E
I nl
[1 _(2/+1)(1+1)
1 (Za) 2]
n

'---.---Enlj(-) = -IEnl
down"
"SPill
[1 + 1(21 1+ 1) _(Z_a_)2]
n
j =I-!
(12 .13)

(where we have written En = -lEnD. Thus we see that the spin-orb it correc-
tions to the "unperturbed" energies En are about 1 part in 105 • The fact that these
corrections are indeed small lends consistency to our original assumption that the
product eigenstates (12 .8) closely approximate the eigenstates of the total Hamil-
tonian f! (12.7).
The two energies (12 .13) correspond to the two possible orientations of 8 with
respect to L (see Fig. 12.2). When the spin is "down," the magnetic moment of
the electron (#1- ~ -8) is aligned with the magnetic field i3 (12.4) of the rela-
tive electron-nucleus motion . This is the configuration of minimum energy. The
correction increment corresponding to "spin down" diminishes En in agreement
with the expressions (12 .13) .
The spin-orbit interaction serves to remove the 1 degeneracy of the eigenener-
gies of hydrogenic atoms . If the spin-orbit interaction is neglected, energies are
dependent only on the principal quantum number n and are independent of 1 (and
m/)' In the L-S representation, nljm j (and s = !)
are good quantum numbers,
and the 1 degeneracy is removed . Degeneracy with respect to m i- however, re-
mains. Eigenenergies are dependent only on (n , I, j), as indicated by expression
590 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

----r----- 2 p3/2
-----,f----,..---__ 2Pl/2

A=hc/E

FIGURE 12.7 Wavelengths, in A, corresponding to the transition from the lowest 2p

states to the ground state, for potassium.

(12.10) . For a given principal quantum number n, the orbital quantum number is
restricted to the values 1 = 0, 1, .. . , (n - 1) [recall (10.127)] , while for a given l,
the total angular momentum quantum number j can take the two values j = 1± i.
The partial energy-level diagram" for potassium, depicting the transition from
the doublet 2p states to the ground state, is shown in Fig. 12.7. The corresponding
radiation lies in the near infrared.
The selection rules for dipole radiation developed in Section 10.7 are general-
ized to the following, for one-electron atoms ."

tll = ±1

tlj = ±1 , 0 (but j = 0 --+ 0 is forbidden)

tsm ] = ±1 , 0
tln is unrestricted

Photons are emitted only for transitions between states which obey these condi-
tions.

Relativistic Corrections
The spin-orbit corrections to the energies of hydrogen are the same order of mag-
nitude as the corrections due to the relativistic speed of the electron in its orbit.
This small correction to the "unperturbed" energies En may also be obtained us-
ing the technique developed above to find the spin-orbit correction. The relativis-
tic Hamiltonian for a particle of mass m moving in a potential field V is

6It was in explanation of such doublet spectra that G. E. Uhlenbeck and S. Goudsmit first postulated
the existence of electron spin. Naturw iss. 13,953 (1925), and Nature 117, 264 (1926).
7Derivation of these selection rules may be found in R. H. Dicke and J. P. Wittke, Introduction to
Quantum Mechanics, Addison-Wesley, Reading, Mass., 1960.
12.2 One-Electron Atoms 591

H = (p2 e2 + m 2e 4 ) 1/ 2 - me 2 + V
(12.14)
Ze 2
V=--
r
If P « me, then the radical may be expanded to obtain

p2 p4
H=----+ ···+V
3e 2 2m 8m
(12.15)
~ P~2 ) P~ 4 ~ ~,
H= -+V ---::Ho+H
3 2
( 2m 8m e

Again, as was done above, the correction to the eigenenergies En of Ho due to H '
may be obtained by calculating (H') using the hydrogen wavefunctions (10.126) .
There results

(H') = (nlIH'lnl)
1 ~4
= --8
3 2(P }nl
me
To evaluate this expectation value, we recall that the eigenstates Inl} satisfy the
equation

Holnl} = Enlnl}

(:~ + V) Inl} = Enlnl}

so that (writing Inl) :: ICfJnt})

p2 lCfJnl } = 2m(En - V)ICfJnl}

Owing to the Hermiticity of p,

(CfJnl IP~2 P~ 2 1 CfJnl ) = ( P~2 CfJnl IP~2 CfJnl )

It follows that

4
(p }nl = 1 00

[2m (En - V)CfJ~tl[2m(En - 2

V)CfJntlr dr

Hence

, 1 2
(H ) = - - - 2 (En - 2En(V}nl + (V 2 }nl)
2me

The integrals (r- I }nb (r- 2}nl may be obtained using the results of Problem 10.49.
592 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

There results

(H')=-IE
n
a4n2 (8n
2 2
Z
1-- ---3
2l + 1
)
As with the case of spin-orbit correction, we again find that (H') is smaller than
En by a factor of the order of a 2 . Combining this correction with the expressions
for the spin-orbit correction (12.13) gives the result

Enlj(spin-orbit+ reI.) = -IEnl [1 + (~; rC:! -3)] (j = l± ~)

(12.16)

This expression for the energies of hydrogen include the "fine-structure" cor-
rections, due to spin-orbit and relativistic effects. The energies so found are in
quite good agreement with observed hydrogen emission spectra.f

PROBLEMS

12.6 The magnetic field (12.4) due to the relative nucleus-electron motion may also be
thought of as arising in the following way. If one "sits" on the electron, the nucleus
is seen to move in orbital motion about this position. This nuclear orbit constitutes
a current loop, which in tum generates a magnetic field. Calculate the value of this
magnetic field for a given value of L and compare it to the value obtained from
(12.4).

12.7 In quantum mechanics, when one says that the vector J is conserved, one means that
for any state the system is in, the expectations of the three components of J are con-
stant. This follows if these three operators all commute with the Hamiltonian. Show
that for a one-electron atom with spin-orbit coupling, Land S are not conserved.

12.8 There is no spin-orbit interaction if an electron is in an S state (I = 0). Why ?

12.9 What is the difference in energy between the two states of a doublet for a typical
one-electron atom as a function of nand l?

12.10 What is the wavelength of a photon emitted by a typical one-electron atom when
the valence electron undergoes a spin flip from the 22 P3/2 to the 2 2 Pl /2 state? In
this notation , 2 is the value of the principal quantum number n . Give your answer
in terms of Z. According to the selection rules cited for dipole radiation, is such a
transition allowed?
12.11 Make an estimate of the rotational kinetic energy, L 2 /2mr 2, of an electron in a 2P
state of hydrogen. What is the ratio of this energy to the rest-mass energy mc 2 =
0.511 MeV ?

8Calculation of higher-order effects may be found in L. 1. Schiff, Quantum Mechanics, 3d ed.,

McGraw-Hill, New York, 1968.
Problems 593

12.12 (a) What are the respective frequencies (Hz) emitted in the transitions (1) 2PI/2 --+
ISI /2, (2) 2P3/2 --+ ISI /2 , for lithium ?
(b) What is the percentage change between these frequencies ?
12.13 In the theory developed in Section 12.2, the effect of the inner core electrons of
an alkali-metal atom on the energy spectrum was described by an effective atomic
number. Owing to penetration of the valence electron wavefunction into the core,
such a simple model proves insufficient. A more quantitative model which includes
the effects of this quantum defect is described in the following example .
One assumes that the potential seen by the outer valence electron is of the form

-Z'e
2( b)
V= -- 1+-
r r

This modified potential has the effect of making the force of attraction between
the valence electron and the nucleus (of charge Ze) grow with penetration of the
valence electron into the core. The deeper this penetration , the larger is the net
positive charge "seen" by the valence electron . The effective nuclear charge Z' e
and displacement b may be chosen so as to give the best fit with observed spectral
data.
(a) Show that the method used to solve for the energy levels of the hydrogen atom
can be applied to this problem with only slight modifications to give energy
levels of the form

-Ze 2
En / = - - - - - --=-
2aO[n + D(/)]2

Here ao is the hydrogen Bohr radius and

represents the I-dependent quantum defect.

(b) How do these En energy states vary with increasing I? Give a physical expla-
nation of this variation in terms of core penetration of the valence electron .
12.14 At sufficiently high temperatures, a diatomic dumbbell molecule may suffer vibra-
tional modes of excitation above the normal rotational modes. Consider that the two
atomic nuclei are bound through a central potential V (r) which has a strong mini-
mum at the separation r = a. At low temperatures the nuclei stay at this interparticle
spacing and the effective Hamiltonian is

LA2
+ yea)
A

H = --2
2J1-a

where J1- is reduced mass. At higher temperatures, the particles separate and the
Hamiltonian becomes
594 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

where ~ is the (small) radial deviation from the equilibrium separation , a. For the
case that (L 2/ 2/La 2 ) « (~/a)V(a) :
(a) Obtain a form of Hthat is valid to the order of (~ /a)2 .
(b) By appropriate choice of product wavefunctions, obtain the eigenenergies Eln
of the Hamiltonian you have constructed .
(c) Argue the consistency of omitting the radial kinetic energy Pr2 /2/L in the
Hamiltonians written above.
Hint [for part (a)]: Since V Cr) is minimum at r = a, it follows that V' (a) = O.
12.15 In Chapter 11 [equation (11.86)] it was noted that the magnetic moment of an elec-
tron due to its spin is

e
ILs =-S
me

Consider that the electron moves in a circle with angular momentum L. Show clas-
sically that the magnetic moment due to such orbital motion is

e
ILL = - L
2me

[Note : Thus we see that the total magnetic moment of an atomic electron

IL = 2:e (J + S) = /L; (J + S)
may not in general be assumed to be parallel to its total angular momentum J.]

12.16 Consider an atom whose electrons are L-S coupled so that good quantum numbers
are jlsm j and eigenstates of the Hamiltonian Ho may be written Ijlsm j )' In the
presence of a uniform magnetic field 13, the Hamiltonian becomes

H = Ho + H'
~, e
+ S) . 13
A A

H = -jL . 13 = - (J
2me

where J and S are total and spin angular momenta, respectively, and e has been
written for lei. Before the magnetic field is turned on, Land S precess about J as
depicted in Fig. 12.1. Consequently, IL = -(/Lb / n)(L+ 2S) also precesses about J.
After the magnetic field is turned on, if it is sufficiently weak compared to the
coupling between Land S, the ensuing precession of J about 13 is slow compared
to that of IL about J, as depicted in Fig. 12.8a.
(a) In the same limit show that time averages obey the relation

(b) Assuming, as in the text, that eigenstates Ijlsm j ) are still appropriate to the
perturbed Hamiltonian HO + H', show that an eigenenergy E j Is0 of Ho splits
into 2j + 1 equally spaced levels
Problems 595

Slow cos(j
rotation A == cos a cos "I
B ==sin a sin "I
A + B ...... , . . . . . . - - - - -__...,....-

At---\-----I-----

A - B t------'-<-------

(a) (b)

FIGURE 12.8 (a) In the presence of a weak iJ field, the precession of J about iJ is slow
compared to that of p- about J. (See Problem 12.16.) (b) Variation of cos f3.

Ejlsmj = Ejls
o + LlEmj
li n
LlE mj = T9(jl s)m j (m j = - j , ... , j)
where n /2 is the Larmor frequency introduced in Section 11.9 and 9 is the
Lande 9 factor

'ls ) = 1 j (j + l ) + s (s + I ) - l (l + l )
(
9 J + 2j(j+l)

also briefly discussed in Section 11.9. (Note: This splitting of lines due to the
presence of the magnetic field is called the Zeeman effect. Note the inferred
relation ({t z) = -9 ({tb lli)(Jz) ·)

Answers (partial)
(a) From the orientation of vectors shown in Fig. 12.8a we see that the relation to
be established is correct provided that time averages satisfy

cos f3 ::::: cos y cos a

Again in reference to the figure one finds that

f3max=a+y
f3min = a - y

The variation of cos f3 between these extremum values is very nearly harmonic
(Fig. 12.8b), and forming the time average of cos f3 gives the desired result.
(b) With the given approximation, one obtains (show this)

.,= (n)
H - j2 + j . s .z
2 J2
J
596 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

Expanding the vector (J - 8)2 and forming the expectation (j lsm j IiJ'lj sm j)
gives the desired result.

12.17 Use the results of Problem 12.16 to obtain the Zeeman pattern of spectral lines
which stem from the transition 4F3/2 - 2DS/2 .
12.18 In the classical formulation of the Zeeman effect, one views the orbital motion of an
atomic electron as being perturbed by the imposed magnetic field, thereby altering
frequencies of rotation . Assuming circular motion of unperturbed frequency WO'
show classically that in the presence of a sufficiently weak magnetic field ($1 «
WO) , this frequency divides into the two lines

while radiation polarized parallel to 13 remains with frequency WO0

Answer
The atomic component of force maintaining circular motion before the magnetic
field is applied may be written mwo2r . lethe magnetic field is normal to the plane
of the orbit, the magnetic force is also along the radius and, provided that the field is
sufficiently weak, we may assume the perturbed motion to be slightly altered with
small variation in frequency. Let the new frequency of rotation be co. The total force
may then be written

Solving for ca gives the roots

2w = $1 ± J$12 +4W0 2

The assumption of a weak magnetic field ($1 « 2WO) allows the radical to be ex-
panded , giving the roots
$1
w=WO±-
2
Components of motion parallel to 13 are unaffected by 13 so that frequency WO
maintains for polarization parallel to 13.
12.19 In Problem 12.16 it was discovered that a magnetic field will split an eigenenergy
corresponding to given jls values into 2j + 1 levels

t::.Emj = C~) sm,

Show that this proliferation of levels leads to the classical Zeeman splitting of a
single frequency into three new lines (as demonstrated in Problem 12.18) in the
event that the Lande 9 factors of both levels of the transition are the same. (Hint:
Frequency displacements t::.v from the original unperturbed frequency are given by

h S» = t::.Em j' - t::.Emj

This , together with the selection rules Sm j = 0, ±l , gives the desired result.)
 12.3 The Pauli Principle 597

12.3 • THE PAULI PRINCIPLE

Indistinguishable Particles
The concept of symmetric and antisymmetric wavefunctions was encountered in
Section 8.5. These wavefunctions are appropriate to systems containing identical
particles . There is a very fundamental distinction between the quantum and clas-
sical descriptions of such systems. At the quantum mechanical level of descrip-
tion, identical particles are also indistinguishable . In the classical description of
a system of identical particles, one may conceptually label such particles and fol-
low their respective motion . This is impossible at the quantum level. There is no
experimental result that distinguishes between two states obtained by exchange
(interchange) of identical particles.
Consider a system that consists of two identical particles (e.g., electrons) mov-
ing in one dimension (x) . Let XI be the coordinate of the first particle and X2 be
the coordinate of the second particle. Then
2
PI2 dXI dX2 = Icp(XI, X2) 1 dXI dX2

denotes the probability of finding particle 1 in the volume element dXI about
the point XI and particle 2 in the volume element dX2 about the point X2 [recall
(8.98)]. In this notation the first slot in the wavefunction cp( , ) is reserved for
the position of particle 1, while the second slot is reserved for the position of par-
ticle 2. Now if these two particles are truly indistinguishable, then it is impossible
to discern between the two states :

No. 1 at XI) and (NO' 2 at XI)

( No.2 at X2 No. I at X2
It follows that the probability of finding these two configurations is the same
(Fig. 12.9).

(a) (b)

FIGURE 12.9 Two classically distinct configurations of two identical particles on a wire .
The probability density Irp(XI, x2)12 pertains to configuration (a) and Irp(X2, xl)1 2 to con-
figuration (b). In quantum mechanics, the identical particles 1 and 2 are also indistinguish-
able, so the probability densities associated with these configurations must be the same.
598 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

(12.17)

Only wavefunctions with this exchange-symmetry property are valid wavefunc-

tions for a system of identical particles. Of these, it turns out experimentally that
wavefunctions relevant to quantum mechanics fall into two categories: symmetric
(({Js) and antisymmetric (({JA). These functions have the respective properties

({JS(XI , X2) = ({JS(X2, x»

({JA (XI, X2) = -({JA (X2, x»

which both obey (12 .17). For particles free to move in three dimensions, we write

({Js(rl , r2) = +({Js(r2, rj )

(12.18)
({JA(rl, r2) = -({JA(r2 , r j )

Let the Hamiltonian iJ describe a system that contains two identical particles:
1 and 2. If these two particles are truly indistinguishable, the Hamiltonian iJ must
be symmetric with respect to the positions of these particles, that is,

Exchange Including Spin

In addition to space coordinates r, a particle also has spin coordinates 8. This state
of a free particle, for example'Amay be given in terms of the eigenvalues of these
commuting observables (e.g., S2, Sz, x,.9, z). Thus, more generally, the Hamilto-
nian iJ must be symmetric with respect to spin as well as position coordinates of
particles. This symmetry property for iJ appears as

(12 .19)

The properties (12.18) for ({JS and ({JA become

({JS(rl, 81; r2, 82) = +({Js(r2, 82; ri , 81)

(12.20)
({JA(rl , 81; r2, 82) = -({JA(r2 , 82; rr , 8»

Again , as in the one-dimensional case (12 .17) , the probability densities associated
with these wavefunctions are totally symmetric. Writing "I" for (rj , 8», and "2"
for (r2, 82) , this symmetry property appears as
2 2
l({Js(1 , 2)1 = l({Js(2, 1)1
2
I({JA(I, 2)1 = I({JA(2, 1)1 2
These symmetry concepts are conveniently expressed in terms of the proper-
ties of the exchange operator i [recall (11.112) et seq.], which is defined by the
12.3 The Pauli Principle 599

equation

Xcp(1 , 2) = cp(2, 1) (12.21)

The operator X has two eigenvalues :

Xcps(1,2) = cps (2 , 1) = +cps(l , 2)

XCPA(1, 2) = CPA (2, 1) = -CPA (1, 2)

The eigenfunction CPS corresponds to the eigenvalue + 1. It is even under par-

ticle exchange . The state CPA corresponds to the eigenvalue -1. It is odd under
particle exchange .
Owing to the exchange symmetry of the Hamiltonian (12.19), if commutes
with X.

(12.22)

It follows that Xis a constant of the motion.

d ~
dt (X) =0
If at time t = 0, cp(O) is such that

Xcp (O ) = +cp(O )

then the two-particle system remains with this property for all time. At time t > 0,

Xcp (t) = +cp(t)

Since X commutes with if, it is possible to find simultaneous eigenstates of both

these operators (Fig. 12.10). These common wavefunctions are the eigenstates ap-

An eigenstate of i and if

An eigenstate of .l
FIGURE 12.10 Venn diagram exhibiting the simultaneous eigenstates of the exchange
operator i: and the Hamiltonian if.
600 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

propriate to systems of identical particles. Furthermore, such eigenstates may be

classified in terms of their symmetric or anti symmetric properties under particle
exchange.

Bosons and Fermions

The constancy of (X) is an immutable property of a system of identical particles.
Owing to the permanence of this property, one may assume that it is a property of
the particles themselves (as opposed to a property of the wavefunction). Particles
characterized by the eigenvalue + 1 of i are called bosons . The wavefunction for
a system of bosons is symmetric (rps ). Particles characterized by the eigenvalue
-1 of i are called fermions. The wavefunction for fermions is anti symmetric
(rpA).
The characteristic of a particle that determines to which of these categories
it belongs is given by the spin of the particle." Bosons have integral spin,
while fermions have half-integral spin. Electrons and neutrons are examples
of fermions. Photons, n , and K mesons are examples of bosons.l?

Antisymmetric Wavefunctions
The Pauli principle is obeyed by fermions . It states , as described above, that the
wavefunction for a system of identical fermions is antisymmetric. Consider a sys-
tem of two fermions. They are in the state rpA (l , 2) . This state has the property
that

Consider that both particles have the coordinates rl and 81. That is, the system is
in the state , rp A (1, 1). This state has the property

The only value of rpA which has this property is

There is zero probability of finding the particles at the same point in space, with
the same value of spin . This is the essence of the Pauli principle: two fermions
cannot exist in the same quantum state .I I
Let us consider a problem: What is the wavefunction of two free electrons
moving in 3-space? The Hamiltonian for this system is
~ 2 ~ 2
if (1, 2) = ~ + !!l:...- (12.23)
2m 2m
9See Appendix B.
IOWriting eigenvalues of i as exp(iarr ), for fenni ons a = I and for bosons a = 2. Particles, called
"anyons," for which a is fractional. are hypothesized to exist.
11 Thi s property of fenn ions is also called the Pauli exclusion principle.
12.3 The Pauli Principle 601

The space component wavefunctions of this Hamiltonian are the product states

This same value of energy characterizes the exchange state ,

From these two (degenerate) states one may form the symmetric and antisymmet-
ric eigenstates

1
cps(rl, r2) = ./2 [cpk, (rl)CPk2(r2) + cpk] (r2)CPk2(rl)]
(12.24)
1
CPA(rl, r2) = ./2 [CPkl (rl)qik2(r2) - CPkl (r2)CP k2(rl)]

Since the two particles in this problem are fermions , wavefunctions for the sys-
tem must be antisymmetric with respect to exchange of particle spin and position.
Inasmuch as the Hamiltonian (12.23) does not contain the spin, it commutes with
all spin functions . Thus, if ~ denotes a spin state for the two-electron system, then

are possible eigenstates of iI. In Section 11.10 we found that in the coupled rep-
resentation, two spin-! particles combine to give three symmetric (s = 1) states,
~s(l) , ~s(O), and ~s(-1), and one antisymmetric (s = 0) state, ~A (0) . Combining
these with the space state (12.24), one obtains the four antisymmetric states

, 81; r2, 82) = cps(rl , r2)~A (1,2)

j
l X A (rj

h(I)(I , 2)
3x A(rl , 81; r2 , 82) = CPA(rl , r2) ~s(O)(1, 2) (12.25)

1 ~s(-I)(I, 2)

n,2 2 2
Ek]k2 = -(kl
2m
+ k2 )

Here we are using the simple rule that the product of a symmetric and antisym-
metric state is antisymmetric.
This technique of incorporating spin states to ensure antisymmetry of a given
state also applies to Hamiltonians that include interaction between particles, or
interaction between particles and a central force field, but which are otherwise
602 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

spin-independent. For such cases, the structure (12.25) of antisymmetric eigen -

states is maintained. This concept finds direct application below in discussion of
the helium atom . Symmetrization of the wavefunction for bosons will be applied
in construction of the nuclear component eigenstates of the deuterium molecule.

PROBLEMS

12.20 Show that the two operators

have the projection property

= tps(1, 2)
1:+ tp(1 , 2)
1:-tp(1 ,2) = tpA (1, 2)

That is, 1:+ projects tp onto tpsand 1:- projects tp onto tpA.
12.21 (a) Consider a two-particle system with relative radius vector r = r2 - rl
(Figs. 10.9 and 10.10). Show that 1:1/t(r) = JP>1/t(r), where JP> is the parity
operator introduced in Section 6.4 (see Problem 9.46).
(b) Is the parity of a two-particle system a "good" quantum number in the energy
representation ?
(c) Two particles interact under a central potential V (r). It is known that the system
is in the state R43(r)Y30(9 , q,). What is the symmetry of the state?

Answer (partial)
(a) 1:1/t(r) = 1:1/t(r2 - rj ) = 1/t(rl - r2)
JP>1/t(r2 - rj ) = 1/t[-r2 - (-rl)] = 1/t(rl - r2)

12.4 • THE PERIODIC TABLE

Central Field Approximation

In this model it is assumed that each electron "sees" only the electrostatic field due
to the nucleus and remaining electrons and that this combined field is spherically
symmetric. Owing to this spherical symmetry, the operators

relevant to a single electron form a set of commuting operators so that the state of
each such electron is specified in terms of the eigenvalues
12.4 The Periodic Table 603

These quantum numbers correspond to a wavefunction for the ith electron of the
form

'Pnlm/ms(ri, S i) = Rnl(r;)Yt((Ji, epi)~/(i) (12.26)

~/ == a(i) or f3(i)

The Pauli principle precludes any two electrons being in the same state [i.e.,
having the same set of (n .L, mi , m s) values]. In our discussion of the hydrogen
atom, we found that there are n 2 distinct states corresponding to a given value of
n (10.128). When spin dependence is included in these states [e.g., (12.26)], this
degeneracy is doubled. There are two values that m, may assume for any set of
values nlmi . Thus, corresponding to any value of n, there are 2n 2 functions of the
form (12.26) which give the same energy for the ith electron .
This degeneracy rule, taken together with the Pauli principle, serve to ex-
plain the "shell structure" of the electronic configurations of the elements. As the
atomic number Z increases, electrons fill the one-electron states of lowest energy
first. For the lighter elements these are shells of lower n values. Within an n shell,
for any given value of l there is an l subshell with 2(2/ + 1) states, corresponding
to two m s values and 2/ + 1 values of mi . Atoms with filled n shells have a total
angular momentum and total spin of zero (see Table 12.1). Electrons exterior to
these closed shells ("valence" electrons) determine the chemical properties of the
atom. The "periodicity" of these properties owes to the fact that valence numbers
repeat after shells become closed .
For n = 1, l can only be 0, but m, may take on the two values, ±!.
Therefore,
there can be, at most, only two electrons in the n = I "shell." (This is called
the " K shell" in x ray notation .) In the ground state of helium (Z = 2), the
n = 1 shell is filled. The electronic configuration for this state is described by the

TABLE 12.1 Diagranunatic enumeration of states available in the first three

atomic shells"

Number of Available
States in
ml -2 -1 0 +1 +2 q Each Shell
-->

M shell,
n=3
H H
H
H
H
H
H
H
H 2
1
0
'0~ l
18 = 2 x 3
2

L shell,
n=2
H H
H
H 1
0 ~} 8 = 2 x 22
K shell, H 0 2 = 2 x 12
n= 1
aYertical arrows represent Sz values. The total orbital and spin angular momentum is zero for a closed
shell.
604 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

notation I s2 . This is read: there are two (the exponent) electrons in the state with
n = 1 and I = 0 (denoted by the letter s). The term notation for this state is ISO.
The electronic configuration of the ground state of lithium (Z = 3) is

ls 22s 1

There are two electrons with n = 1, I = 0 and one electron with n = 2, I = O. The
!
uncoupled spin gives a j value of with corresponding term notation, 2S I/2. In
beryllium (Z = 4) the ground -state configuration is ls22s 2. All spins are paired
and the ground state is given by the term ISO.
The electronic configurations and corresponding ground states for the first 36
elements are given in Table 12.2. Note that ground states follow the L-S coupling
scheme. For example, for nitrogen (Z = 7), whose electronic configuration is

Is 22s 22 p3

the ground state is 4S 3/ 2 . The three p electrons can have total spin values

1 3
s= - -
2 '2

and total I values (see Section 9.5)

1=0,1 ,2,3

The corresponding j values are given by the sequence

j = II + s], ... , II - s ]
for all pairs of I , s values. The doublet (s = !) states thus obtained are
2SI/2, 2PI /2 ,3/ 2, 2D 3/2 ,S/2, 2Fs /2, 7/2

The quartet (s = ~ ) states are

4S 3/ 2, 4P 1/ 2,3/2 ,S/2 , 4D 1/ 2,3/ 2,S/ 2,7/ 2, 4F3/ 2,S/ 2,7/ 2,9/2

Of these 19 possible states, the exclusion principle permits only the 2D, -r, and
4S states.
To understand this reduction in terms , consider the simpler case of the carbon
atom which has two p electrons in its outer shell (with n = 2; see Table 12.2).
Following the preceding construction, we find that the total I values, I = 0, 1, 2,
and total s values , s = 0, 1, give the following terms :

IS, 'r , ID, 3S, 3p , 3D

The allowed states for these two electrons must be antisymmetric under particle
exchange. Wavefunctions have the product structure
TABLE 12.2 Distribution of electrons in the atoms from Z = I to Z = 36
X ray Notation K L M N
Values of n, I 1,0 2,0 2,1 3,0 3,1 3,2 4,0 4,1 4,2 4.3
Spectral Notation Is 2s 2p 3s 3p 3d 4s 4p 4d 4/
N
... '"--.
o1:: >
ll)
.0
i: E .~
ll)
'-"
ll)

E ;:l ca-g
ll)

Ul
I::
o
.§
·2·
o ~
0
.... '0
0..
.(

H 1 13.595 I 2S I /2
HE 2 24.481 2 ISO
Li 3 5.39 2 I 2SI / 2
Be 4 9.32 2 2 ISO
B 5 8.296 2 2 I 2PI /2
C 6 11.256 2 2 2 3
PO
N 7 14.53 2 2 3 4S3/ 2
0 8 13.614 2 2 4 3P2
F 9 17.418 2 2 5 2P3/ 2
Ne 10 21.559 2 2 6 ISO
Na 11 5.138 1 2SI / 2
Mg 12 7.644 Neon 2 ISO
Al 13 5.984 configurati on 2 1 2P I/2
Si 14 8.149 2 2 3Po
P 15 10.484 2 3 4S3/ 2
S 16 10.357 10-electron 2 4 3P2
CI 17 13.01 core 2 5 2P3/2
Ar 18 15.755 2 6 ISO
K 19 4.339 I 2S I/ 2
Ca 20 6.111 2 ISO
Sc 21 6.54 I 2 2D3 / 2
Ti 22 6.82 2 2 3F2
V 23 6.74 3 2 4F3/ 2
Cr 24 6.764 5 I 7S3
Mn 25 7.432 Argon 5 2 6SS / 2
Fe 26 7.87 configuration 6 2 SD4
Co 27 7.86 7 2 4F9 / 2
Ni 28 7.633 8 2 3F4
Cu 29 7.724 18-electron 10 I 2S I /2
Zn 30 9.391 core 10 2 ISO
Ga 31 6.00 10 2 I 2P I/ 2
Ge 32 7.88 10 2 2 3Po
As 33 9.81 10 2 3 4S3/ 2
Se 34 9.75 10 2 4 3P2
Br 35 11.84 10 2 5 2P3/2
Kr 36 13.996 10 2 6 ISO
605
606 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

TABLE 12.2 (Continued)

Electronic configurations for the alkali metals

Shell K L M N 0 P
Li Is 2 2s
Na Is 2 2s22 p 6 3s
K Is 2 2s22 p 6 3s23 p 6 4s
Rb Is 2 2s22 p 6 3s23p63d lO 4s 24 p6 5s
Cs Is 2 2s22 p6 3s23p63d lO 4s 24p64d lO 5s 25 p6 6s
aData obtained from Handbook of Chemistry and Physics, 56th ed., CRC Press, Cleveland,
Ohio, 1976.

(12.27)

We assume for the moment that the two electrons are not "equivalent," i.e., that
they have different principal quantum numbers, nl =f. n2. The radial functions
then have the form

(12.28)

where the R functions are hydrogenlike wavefunctions. Angular momentum com-

ponents I} m j Is) are constructed as follows. In the Russell-Saunders coupling
scheme total L adds to total S to give the atomic angular momentum J. Wave-
functions are then given by the Clebsch-Gordan expansion,

l}mjIs) = L Cm/mslim/)Ism s) (12.29)

mj=m/+ms

The Ilm/) and Isms) components as well are given by Clebsch-Gordan expan-
sions, which for the case at hand are displayed in Tables 9.5 and 11.3, respec-
tively. We see that orbital states of odd I are antisymmetric and those of even I are
symmetric. All three s = I states are symmetric, whereas the singlet s = 0 state
is antisymmetric. So we may construct the following list:

3D: 1= 2(sym), s = 1(sym) -+ I}mj)s -+ RA

'e. 1= 1(anti), s=O(anti) -+ I}m j)s -+ RA
3S: 1= O(sym), s = 1(sym) -+ I}mj)s -+ RA
(12.30)
ID: 1= 2(sym), s =O(anti) -+ l}mj)A -+ Rs
3p: 1= l(anti), s = 1(sym) -+ l}mj)A -+ Rs
IS: 1= O(sym), s = O(anti) -+ l}mj)A -+ Rs
12.4 The Periodic Table 607

TABLE 12.3 States allowed by the exclusion principle of the L-S coupling scheme

npO
npl 2p
np2 3p
np3 2p,2D 4S
np4 3p
np5 2p
np6
ndO
nd l
nd 2
nd 3 2D,2p,2D,2F,2G ,2H 4p,4F
nd4 IS, io , lG, IS, ID , IG,IF, II 3p, 3F, 3p , 3D, 3F, 3G,3H 5D
nd 5 2D,2p,2D,2F,2G,2H ,2S,2D,2F,2G,21 4p ,4F,4D ,4G 6S
nd 6 IS,ID ,IG, IS, ID, IG, IF , II 3p,3F, 3p , 3D ,3F, 3G,3H 5D
nd7 2D,2p,2D,2F,2G,2H 4p ,4F
nd 8
nd 9
nd lO

For the np2 configuration under consideration, n I = n2 = n = 2, so that RA = O.

Only the terms which include the Rs factor survive. These are the ID, 3p, and IS
terms, which in all comprise 15 independent states corresponding to five energy
levels. Of these, Hund's rules (Problem 12.25) determine the ground state. The
first rule indicates that the ground state is among the 3Po, 1,2 states . Since there are
only two electrons in the 2p shell of carbon, and this shell can accommodate six
electrons, the third rule indicates that 3Po is the ground state of carbon. The same
is true for the two-electron (np 2) atoms, Si and Ge. Remaining allowed atomic
states for configurations up to the nd lO shell are shown in Table 12.3.
A note of caution is in order for three (or more)-electron atoms. For such cases
spin states which are antisymmetric with respect to exchange of any two of the
three particles do not exist (see Problems 11.92 et seq.). In this case, generaliza-
tions of the product form (12.27) comes into play.12
The symmetry of the states shown in Table 12.3 about the midvalue of occu-
pation in a given I shell is due to the following . Consider, for example, the p shell
which can be occupied by six electrons. In what manner does the p2 configura-
tion resemble the p4 configuration? The p4 configuration may be viewed as one
in which two holes occupy the p shell. This atomic configuration yields the same

12For further discussion, see M. Weissbluth , Atoms and Molecul es, Academic Press, New York, 1978.
Recall also Problem 11.88.
608 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

allowed states as does the p2 configuration (see Problem 12.49). The same holds
true for the d 3 and d 7 configurations, and so on.
A complete description of an atomic state involves the total L, S, and J of
the atom in addition to the quantum numbers of individual electrons. The first of
these is given by the term notation of the state (e.g., SF7j2), whereas the second is
given by the electronic configuration (e.g., Is22s22p 3).
The periodic chart is shown in Table 12.4. The ground states of elements and
outer electron shell configurations'< 14 are listed, as well as atomic numbers .
As stated above, chemical properties of elements are determined by the elec-
tron configuration in the unfilled shell. An important related chemical reaction is
that of the oxidation -reduction process. For example, in the oxidation of iron, two
4s electrons of an iron atom are transferred to an oxygen atom to complete its 2p
shell, thereby forming the strong ionic bond of ferrous oxide . (The oxygen atom
is reduced in this process.)
Atoms with similar valence electron configuration have nearly the same chem-
ical properties. The properties of atoms in some of these groupings are described
below.

The Alkali Metals. Group I

The alkali metals are the atoms with one valence electron: Li 3 , Na ll , K 19 , Rb 37 ,
Csss, and Fr87 . The ground state of these elements is 2S l j2 . Ionization energy
is low. The spectra of these "one-electron" elements resemble that of hydrogen .
Valence is + 1.

The Alkaline Earths. Group II

All the alkaline earths have two s electrons outside a closed p subshell. They are
Be4 , Mg 12 , Ca20 , Sr 38 , BaS6 , and Ra88 . The ground state is ISO. Ionization remains
relatively small. Their valence is +2. When singly ionized, these atoms are known
as "hydrogenic" ions; their spectra resemble that of hydrogen .

The Halogens. Group VII

These are the elements: p9, C1 17 , Br 3S, IS3, and At 8S • They are all missing one
electron in the outermost p subshell and therefore have a valence of -I. Halo-
gens form stable molecules with the one-electron (alkali-metal) atoms (e.g., NaCl)
through ionic bonding.

The Noble Elements. Group VIII

The noble elements are also called the "rare gases" or the "inert elements." They
are He 2 , Ne lO, Ar 18 , Kr36 , Xe S4 , and Rn 86 . Except for He, all these atoms have a

13These data were obtained from G. Baym , Lectures on Quantum Mechanics, W. A. Benjam in, New
York, 1969; S. Fraga, J. Karwowski, and K. Saxena, Handbook ofAtomi c Data, Elsevier, New York,
1976; and H. Gray, Electrons and Chemical Bond ing, W. A. Benjamin , New York, 1964.
14A periodic chart with additional atomic and material data appears on the inside of the front cover.
 TABLE 12.4 The periodic table
Group-o
Period
!
I II III IV V VI VII VIII
HI He '
1 lsI ls'
'S I/1 ISo

Li' Be' B' C· N' O· F' Ne lo

2 ls ' 2s' ls ' 2s' 2s' 2p l 2s' 2p ' 2p' 2p' 2p' 2p'
'SI/' 'So 'PI/, 'Po 'SS/1 'P, 'P'/1 'So
Nail Mg l1 AI13 Sil' pI> SI. CI" Ar'·
3 3s' 3s' 3s'3p' 3s'3 p' 3p' 3p' 3p' 3s'3p'
Transition elements 1P3I ,
'SI/' ISO 'PI/' 'Po 'S'/1 'P , 'So

KU Ca ' O SC" n" V13 Cr 14 Mn 1S Fe Co 17 NI'· Cu 29 Zn'o GaB Ge 31 AsH Se" Br" Kr'6
" 4s lJdlO
4 4s' 4s' 4s'3d' 4s'3d' 4s '3d ' 4.•'Jd' 4s'Jd' 4s'3<I" 4s'3d' 4s'Jd· 4s'3d 1O 4s'Jd I 04pl 3d,04p' Jd,04p' Jd,04p' 3dl o4p' 4s' 4p'
'SI/' 'So 'D3/1 'F, 'F'/1 's, 'Sm 'D, 4F9I 2 'F,. 'SI/' 'So 'P", 'Po 'S,/1 'P, 'P'/1 'So

Rb 37 Sr'· y" Zr'o Nb" Mo" Tc" Ru" Rh" Pd" AgO? Cd'· lo In" Sn'o Sb" Te" I" Xe"
S 5s' 5>' Ss'4d 1 5s'4d ' 5s14d" 5s'4d' Ss'4d' Ss'4d' 5s14d. 4d'o Ssl4d'O 5s'4d 5s'4d lo5pl 4dloSp' 4d lo5p' 4d,05p' 4d'oSp' 5s'Sp'
'SI/, ISo 'D'/1 'F, 'D1/1 'S, -s; 'F, 'F'/1 ISo 'SI/' ISO 'PI/' 'Po 'S'/1 'P, 'p'I' 'So

Cs" Ba" La" Hr" Ta 73 W" Re7S Os" Ir" Pl 7I Au" Hg· O nil Pb lZ Biu Po·' At·' Rn··
6 6s1 6s' 6s'Sd I 6s'Sd ' 6s'Sd' 6s' Sd" 6s'Sd' 6s'Sd' 6s'5d' 6sl Sd" 6s'5d'O 6s'Sd lo 6s 06 6
'SdI pl p1 6p' 6p' 6p' 6p'
'S I/1 'So 'D'/1 'F, 'F'11 'Do ·S", 'D, 'F'/1 'D, 'SI/1 'So 'pl/' 'Po 'S'I' 'P, 'p'/1 ISO

Frl? Ra·· Ac··

7 7s1 7s' 7s'6d'
'SI/' ISo 'D'11

ee'" I4f' Pr" Nd oO Pm° l Sm" Eu" Gd" Th os Dy" Ho o, Er·· Tm" Yb'o Lu 71
Rare earths' 6s'Sd 6s'4f' 6s'4f' 6s'4f' 6s'4f' 6s'4f' 6s'Sd'4f' 6s'Sd'4f· 6s'4f'o 6s'4f" 6s'4f11 6s 6s'SdI4f1'
'4fl3 6s'4f"
'H, '1,/1 'I, 'H", 'F o ·S'/1 'D , 'F", 'So 'D3/2

Th' o Pa 91 tJ9' Np" Pu" Am" Cm" Bit" cr'· E99 Fm'oo Md lol
Heavy elements" 7s'6d' 6d' 6d15f' Sf' Sf' Sf' 6d15f' Sf' Sflo Sf" Sfl1 Sf 13
'I, ·S'/1

aWith La 57 included, this group is also called the lathanides.

bWith Ac 89 included, this group is also called the actinides .
a'I
Q
1.0
610 Chapter 12 Application to Atomic , Molecular, Solid-State , and Nuclear Physics

completed outermost p subshell. The ground state of these elements is ISO. Total
spin and orbital angular momentum are zero, so the atom has no magnetic mo-
ment. Ionization energy is large (see Table 12.2); electrical conductivity is low.
Noble elements are chemically inert and have low boiling points.

The Transition Group

In the transition-group elements the incomplete 3d subshell is filled while 2 (or 1)
electrons remain in the outer 4s subshell . The incomplete 3d subshell gives rise
to magnetic properties. For example, the ground state of Cr 24 is 7S 3 . The spin of
the atom is s = 3, which implies that the five 3d electrons and one 4s electron
all have their spins aligned. These parallel spins contribute to the large magnetic
moment of Cr. The chemical properties of elements in the transition group are due
primarily to the 4s electron(s).
The grouping of atoms according to their shell structure is further motivated
by two very significant sets of observational and inferred data. These are, respec-
tively, first ionization energies and radii of atoms. Thus, in Fig. 12.11a we see
that binding in a given n shell grows roughly as the shell is completed, with one-
electron atoms most weakly bound and noble elements most strongly bound . In
Fig. 12.11b we note that the radii of atoms in a given n shell decrease roughly
with increase in Z, with one-electron atoms having the largest radii and halogens
the smallest radii. These properties become less valid at higher Z where outer
electron shells of atoms in a given row of the periodic chart (Table 12.3) are not
characterized by constant principal quantum number n.

PROBLEMS

12.22 Show that the ground-state electronic configuration of C~4 does not violate the
Pauli principle. Hint: Make a list of the (n, l, mj, ms) values for the 3d 54s 1 elec-
trons.

12.23 What are the possible states for the ground configuration of 0 16 which includes
four p electrons in its outmost shell? Check that the 3P2 ground state is included in
your list.

12.24 Describe the energy band structure of the metal lithium . Specifically, indicate which
electrons fill the valence band and which electrons contribute to the conduction
band. How full is the conduction band? (The band theory of conduction was dis-
cussed previously in Section 8.4.)

12.25 Show that the ground states for the first three elements in the "neon configuration"
(2 = 11 to 18) are consistent with Hund 's rules :
(i) The lowest energy state is the LS multiplet with largest s value.
(ii) When more than one value of l is associated with this maximum s value, the
lowest energy state (among those satisfying the exclusion principle) is the one
with largest l value.
Problems 611

(iii) For a given l subshell containing ne electrons, in the lowest energy state the
total angular momentum number j has the value Ii - s I for ne < N / 2 and
Ii + s I for ne > N / 2, where N is the number of electrons in the completed
subshell.

25
He

Ne
20
>-
~
>.
Ar
~ 15
...c
c

.~~
.s;;
.... 10
"."s .' \ - .R"

Na
K Rb Cs

I I I ..
40 50 60 70 80 90
Z
(a)

l
Cs

~
K Rb
t
N\L\/"
2

<$
" '"
:.0
e U U
u

~
"1§
<:
C
CI .<
F

·H

00 10 20 30 40 50 60 70 80 90
Z
(b)

FIGURE 12.11 (a) First ionization energies versus atomic number Z . (b) Atomic radii
versus atomic number. Recall conclusions of the Thoma s-Fermi model concerning varia-
tion of atomic radius with increase in Z discussed in Section 10.8.
612 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

12.5 • THE SLATER DETERMINANT

In the central field approximation, the Hamiltonian for an N -electron atom is

written

(12.31)

In this notation the number 2 denotes the coordinates of the second electron. The
eigenstates of the individual Hamiltonians are of the form (12.26). Calling the
set of eigenvalues (n , l, mi. m s ) of the ith electron Vi , these eigenstates obey the
equations

H(I)CPVl (I) = EV\CPVl (1)

H (2)CP V2 (2) = E V2CP V2 (2)

(12.32)

H (N)CPVN(N) = EVNCPVN(N)

In this notation the product eigenstates of H (1, ... . N) appear as

(12.33)

However. this function is not properly antisymmetric. If Xu denotes the ex-

change operation of the coordinates of electrons 1 and 3, the correct antisym-
metric wavefunctions of H(1, . . . , N) have the property

Xl .3CP(1 . 2, 3, .. . , N ) = cp(3 . 2,1 , . .. ,N) = -cP(1 , 2. 3• . . . . N) (12.34)

The normalized wavefunction that obeys this rule (for all pairs of particles) and is
an eigenstate of H(1, ... , N) is given by the Slater determinant,

CPVl (1) CPV2 (1) CPVN(1)

1 CPV\ (2) CP V2 (2) CPVN(2)
CPA (1. 2, . . . , N ) = .IN! (12.35)
N!
CPVl (N) CPV2(N) CPVN(N)
This determinant has four outstanding properties, which we discuss next.

Eigenvalues
It is an eigenstate of (12.31) with eigenvalue

(12.36)

The explicit form of CPA appears as

CPA = Jm L
. P(V\ .V2.·.· .VN)
(_I)IPlcp Vl(1)CPV2(2)" 'CPVN(N) (12.37)
12.5 The Slater Determinant 613

The sum is over all permutations P of the quantum indices (VI, V2, . •. , vN ). The
symbol IP I is 0 or 1. It is zero if the permutation P (VI, . . . , vN) can be obtained
from (VI, . . . , VN) through an even number of exchanges of two indices. It is 1 if
P(VI , . . . , VN) involves an odd number of exchanges. For example, the term

corresponds to the permutation

This sequence only involves permutation of the first four indices. To obtain IP I
we must rearrange the sequence (4, 1, 2, 3) in the form of the original sequence
(1,2,3,4) through exchanges of two integers only and count the minimum num-
ber of such exchanges which do the job.

(4,1,2,3) --+ (1,4,2,3) --+ (1,2,4,3) --+ (1,2,3 ,4)

Three exchanges suffice so that (-I)I PI = -1. We conclude that the preceding
product wavefunction carries a minus sign, as written.
Each such N -particle product function is an eigenstate of iI corresponding
to the degenerate eigenenergy (12.36), whence the determinantal form (12.37),
which is merely a linear combination of these product states, is also an eigenstate
of iI corresponding to the eigenenergy (12.36).

Orthonormality
The second property that the deterrninantal states (12.35) have is that they form
an orthonormal sequence . That is,
1
(lfJAllfJA) = N' I>rpVl(I)· · ·rpVN(N)lrpVl(1) .. ·rpVN(N))
. P
= 1

since there are N! terms of unit value in the sum. Furthermore, owing to the
orthogonality of single-particle eigenstates ,

(rpA(Vl, ,VN)llfJA(Vl'•... .VN')) = 0

(VI, , VN) =1= (VI' , . . . , VN')

This establishes the orthogonality of these states.

Antisymmetry
The third property concerns the symmetry of rp A . If Xu denotes the exchange of
the ith-particle coordinates with those of the jth particle, then
614 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

Exchanging particle coordinate numbers in tpA, as given by (12.35), is effected

by an exchange of two rows of the determinant. But a determinant changes sign
under interchange of two rows. It follows that Cf!A (i, j) = -Cf!A (j, i) . For example,
for the exchange particles 1 and 2,

Cf!Vl (1) Cf!v2 (1) Cf!vN (1) Cf!vl (2) Cf! v2 (2)
Cf!vl (2) Cf! v2 (2) Cf!vN (2) Cf!vl (1) Cf!v2 (1)
Xl,2

Cf!vl (N) Cf! v2 (N) Cf!vN (N) Cf!vl (N) Cf! v2 (N) Cf!VN (N)

Cf!vl (l) Cf!v2 (1) Cf!vN (1)

Cf!vl (2) Cf! v2 (2) Cf!vN (2)
=

This property establishes the antisymmetry of the state Cf!A.

Exclusion Principle
Finally, we note that if particle 2 has the same quantum numbers as particle 1,
then Cf! A = O. This property also follows from the determinantal structure of tpA:
namely, if two particles are in the same eigenstate, then two columns of the de-
terminant (12.31) are equal and Cf! A vanishes. Thus Cf!A written in the Slater de-
tenninant form is consistent with the Pauli exclusion principle. The Hartree-Fock
approximation (Section 13.10) offers a refinement of the calculation of wavefunc-
tions in the central field approximation.

PROBLEMS

12.26 Which purely detenninantal properties related to the Slater determinant are involved
in (a) the antisymmetry of CPA? (b) the Pauli exclusion principle?
12.27 Two spin- ~ neutrons move in a two-dimensional box of edge length L and impen-
etrable walls. If the neutrons do not interact with each other, construct the anti-
symmetric detenninantal spin and coordinate-dependent energy eigenstates for the
system .

12 .6 • APPLICATION OF SYMMETRIZATION RULES TO THE HELIUM ATOM

We have already found the Pauli principle to be an important rule in forming the
periodic chart of the elements and in construction of properly antisymmetrized
wavefunctions for atomic electrons in the central field approximation.
In this and the next section we will further demonstrate the important role
played by symmetrization principles in analysis of two elemental systems in na-
12.6 Application of Symmetrization Rules to the Helium Atom 615

ture. The first of these is the helium atom, which has two outer electrons . Among
other properties we will find how the Pauli principle influences the coupling be-
tween these electrons in the construction of properly antisymmetrized wavefunc-
tions for the atom. The second example is the deuterium molecule, whose two
deuteron nuclei are bosons. Consequently, the nuclear component of the molecu-
lar wavefunction must be symmetrized with respect to exchange of space and spin
coordinates. Construction of such properly symmetrized states leads very simply
to intensity rules for emission .

The Helium Atom

The Hamiltonian of helium, in a frame where the nucleus is at rest, is

H= (P1
A
- - 2e
2m
2
- ) + - - 2e e 2
- ) +-+Hso
rl
(pi
2m ri
2 2

r12
A
(12.38)

The last term, Hso, is written for the spin-orbit interaction between the electrons
and the nucleus, while HES is written for the electrostatic interaction , e 2/ r12 ,
between the two electrons. The interelectron displacement is

(Fig. 12.12). If the electrostatic as well as spin-orbit terms are neglected, R re-
duces to the sum of two hydrogenic Hamiltonians (each with Z = 2).

RoO , 2) = RoO) + Ro(2) 02.39)

This Hamiltonian (as well as the total Hamiltonian) is symmetric with respect to
the interchange of the two electrons

(12.40)

This merely reflects the indistinguish ability of the two electrons . This property
must be maintained in the eigenstates that we construct for RoO , 2). With the
abbreviations

these symmetrized eigenstates of Ho(l, 2) appear as l5

02.41)

15In the event that VI = vz, then /PS(rl. rz) = /PVI (I)/PVI (2) .
616 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics
z

The nucleus ---~E-==::::::::----t------1r---~

x

1/1 1

FIGURE 12.12 The coordinates of the two electrons in helium . The six-dimensional
volume element (drJ drz) is given by

drJ drZ = TJ ZdQJ drJ r zZ dQZ drZ = TJ Zrz Z dTJ drZ d cos(h d cos Oz d<pJ diPz
The potential of interaction between electrons is given by

eZ eZ
V = -rJ-Z = -,jrr=OJZ:;==+=r=z""Z=-=2==r=Jr=z=c=o=s=='f3

(See Fig. 9.16 for addition formulas connecting f3 to OJ , 0Z, <PJ' iPz .)

The plus sign gives a symmetric state, rps, while the minus sign gives an antisym-
metric state, rp A . The energy eigenvalue corresponding to either of these states
is

(12.42)

Separation of Multiplets Due to Spin Symmetry

Although the spin-orbit correction to the Hamiltonian of helium is small
[L\Eso/ E ~ a Z; see (12.13)] and may well be neglected in a first approxi-
mation, the spin of the electrons still has an important influence on the proper-
ties of helium. This occurs through a combination of Pauli antisymmetrization
requirements on wavefunctions with respect to exchange of space and spin co-
ordinates and the relatively large electrostatic interaction between electrons (see
Problem 12.28). In what follows, first we will construct the properly antisym-
12.6 Application of Symmetrization Rulesto the Helium Atom 617

metrized space-spin dependent eigenstates of Ho(1 , 2). This immediately implies

a coupling between the electron spins. Three states emerge with s = I (the triplet
series) and one state emerges with s = 0 (the singlet series) . When the electro-
static interaction e Z/ r)z is brought into play, it is found that the triplet states all lie
lower in energy than the singlet states. In this manner we will find that symmetry
requirements couple the electron spins and electrostatic interaction separates the
resulting singlet and triplet states.
Insofar as Ho(1, 2) does not contain the spin, spin-dependent eigenstates of
Ho( 1, 2) are quite simple to construct. If rp(rl , rz) is a space-dependent eigenstate
(12.41) of Ho(1 , 2), then following the procedure described in Section 12.3 for
two free electrons , we find that the properly antisymmetrized wavefunctions are
given by

(s = 0)

(12.43)
(s = 1)

The ~ -spin functions are listed in Table 11.3. We see how symmetrization require-
ments, together with the Pauli principle , effect a coupling between the spins of the
two electrons in helium. In the triplet state (s = 1) the spins are aligned , whereas
in the singlet state (s = 0) the spins are antialigned.

Electrostatic Interaction
To understand how the coupling augments the energies of helium , we recall the
following property of symmetrized states (see Problem 8.32): namely, two par-
ticles in a symmetric state attract one another (in a statistical sense). It follows
that the two electrons in the singlet IX state, which contains the symmetric rps
state, are closer to each other than they are in the triplet 3X state, which contains
the antisymmetric rpA state. Thus, owing to the positive repulsive energy of the
electrostatic interaction ; e Z/ r12, the triplet states lie lower in energy than do the
singlet states (Fig. 12.13).
This is the mechanism behind Hund's first rule (see Problem l2.25)-that the
total spin assumes the maximum value consistent with the Pauli principle . In this
symmetric spin state, the space component of the wavefunction must be antisym-
metric so that electrons are further removed from each other than in the corre-
sponding symmetric space state.

Exchange and Coulomb Interaction Energies

Let us consider the space component eigenstates (12.41) of helium in more detail.
As it turns out , the only states of helium that are of practical significance are
those for which one of the two electrons is in its own ground state, with n = 1,
618 Chapter 12 App lication to Atomic , Molecular, Solid-State , and Nuclear Physics

... <r12 > ~.

H----r---4 IP IX
S=o

4226.73 A 6572.78 A

- - - " " ' - - - 4 IS

FIGURE 12.13 The 4 P states of calcium, illustrating the fact that, in two-electron atoms,
triplet states lie lower than the corresponding singlet state. This is due to the fact that
the average interelectron distance is smaller in the symmetric space state, rps , than in the
antisymmetric space state, rp A, thereby increasing the singlet electrostatic contribution to
the total energy compared to the corresponding triplet contribution .

eXIHESI1x ) > eXIHESI\)

e2
HES=+-
r 12

l = mt = O. The reason for this is that it takes less energy to ionize a helium
atom from the ground state than it does to raise both electrons to excited levels
(Fig. 12.14). This means that one is more likely to find an He+ ion (hydrogenic
ion with Z = 2) than a helium atom with both electrons in excited states. It
follows that the space-dependent states of helium atoms that exist under natural
conditions are mostly of the form

(12.44)

One may use these eigenstates to calculate the corrections to the eigenenergies
of Ho due to the electrostatic interaction, e 2 / n2. As described above, we expect
the triplet states to lie lower in energy than the singlet states.
One obtains

( ~)
rl2 .
singlet
= 61(~)IIX}
rl2
(12.45)

=A+B
12.6 Application of Symmetrization Rules to the Helium Atom 619

-++-
T-+-
Ionization
r~~~~~~:- ---- - --
one excited
electron
T
Excitation energy
to the 2s 2
energy of configuration
He

1 Ground
~ls2
Ground
~IS2
FIGURE 12.14 The ionization energy of He is smaller than the energy of the lowest
1
energy state of He with both electrons excited. It is possible for He in this excited state to
decay to He+ and a free electron.

where

(12.46)

The energy A is called the Coulomb interaction energy. It is akin to the classi-
cal interaction potential of two electron clouds with respective charge densities
eICl'lOo(1) 12 and eICl'nlm(2)1 2 . The second term, B , has no counterpart in classical
physics. It is called the exchange interaction energy.
In the triplet state the Coulomb interaction energy becomes

( ~)
'12 .
tnplet
= eXI(~)13X)
'12
(12.47)

=A-B

For A and B positive this correction energy is smaller than the corresponding
singlet correction energy, A + B (see Problem 12.31).16

16The positivity of these integrals is demonstrated in J. L. Slater, Quantum Theory ofAtomic Structure,
Vol. 1, McGr aw-Hili , New York, 1960, Appendix 19.
620 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

Thus we find that the electrostatic interaction separates the singlet and triplet
states of helium . Furthermore, when spin-orbit interaction is brought into play,
different j values have slightly different energies in the triplet states. For a given
value of I, the total angular momentum j number has the three values

j=l-l,l,l+l

which in tum give three distinct values for the spin-orbit coefficient (L·S), thereby
splitting I levels into triplets .
Helium in antisymmetric spin states (singlet series) is called parahelium. He-
lium in symmetric spin states is called orthohelium. The distinct spectra associ-
ated with these different atomic configurations are shown in Fig. 12.15.
Similar descriptions apply to the heavier two-electron atoms . Their spectra also
are observed to separate into singlet and triplet series. The corresponding energy-
level diagram for Ca20 is shown in Fig. 12.16.

Parahelium iI Orthohelium Energy

Hydrogen (eV)
levels
IS
Singlet
Ip ID IF I
I 3S
Triplet
3p 3D 3F
! 0.00
n=4- 4 s - 4 p - 4 d - 4[-14s- 4p- 4 d - 4 [ -
n=3- 3 s - 3 p - 3 d - 13s-3p-3d-
I
2p- I
n=2- I 2p-
2s-
12s-
I
I ~-6.l2

~ -12.2
n=l-

- -18.4

Is- ~ -24.5

FIGURE 12.15 Energy levels of helium, illustrating the singlet and triplet series. The
fine structure of the triplet levels is not shown. The energies of hydrogen appear at the left.
 12.6 Applicati on of Symmetrizatio n Rules to the Helium Atom 621

Energy Singlet states Triplet states

(eV) IP ID IF 3S »r, 3p
ISO I 2 3 1 ,3P2 oI I
3D
3
3D
2 3D\
, L3F-4,3,2
-.J
0.00

7 7 7

6 6 6 6 5 5

-1.53

5
to-
."
~
cei
s»..,. to-
..,.."
'0

-3.05
4
3

-4.59

-6.11 4
Selection iii = ± I
rules: lij = 0 , ± I

FIGURE 12.16 Energy levels of calcium, exhibiting the fine structure of the triplet
states. A few typical transitions are also shown. Transition wavelengths are in A. Principal
quantum numbers appear at the left of levels.
622 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

PROBLEMS

12.28 (a) What is the ground-state wavefunction for helium in the approximation RES =
Rso = 0 in (12.38)?
(b) What is the ground-state energy in this approximation?
(c) What is the correction to this ground-state energy due to the electrostatic inter-
action, e2 / r\2 ? What is the total ground-state energy obtained in this manner?
(d) In view of your answer to part (c), is the ground state of Ro(1 , 2) a good guess
for the ground state of Ro(1 , 2) + RES?

Answers
(a) The ground state is Ips(rl, r2), with nl = n2 = 1 and /1 = /2 = O. This gives
(Z = 2)

IpS = -Z3- exp ('1+

- ---
r 2)
Jra03 ao/Z

(b) EII(O) = -4R(1 + 1) = -4mc2a2 = -108.8 eV

e2
(c) f).E = (lpsl-Ilps)
rl2
e2 5e 2
=
If Ilpsl2- drl dr2
'12
=-
4ao
= 34 eV
drl = '1 2 dr i dQI, 2
dr2 = r2 dr2 dQ2
2 2 2
'12 = '1 + r2 - 2'1 r 2 cos 13

(See Figs. 9.9 and 12.12, and recall the generating function for Legendre poly-
nomials given in Table 9.2.) This gives the corrected ground-state energy

E = EII(O) + f).E = - 74.8 eV

(d) The fact that f).E is not small compared to E 11 (0) means that IpS is not a good
guess for the ground state of Ro + RES. (Note : Eobs = -78.98 eV.)

12.29 Which of the following operators are diagonalized by the states IX and 2X (12.43)
relevant to helium?

12.30 The spin-orbit interaction in two-electron atoms gives three distinct energies in the
triplet series. What are the values of (L . S) if one electron is an s electron and the
other is a d electron .
12.31 Show that the integrals A and B in (12.45) are both positive. [Hint: Recall the
Fourier transform of the Coulomb potential,

-1 =
r12
-12
2Jr
I dk]
expuk- rt2)Z '
k
 12.7 The Hydrogen and Deuterium Molecules 623

12.32 A positron is the antiparticle of an electron . When in the presence of an electron , it

may bind to the electron, forming a positronium atom that is unstable to positron-
electron annihilation. Prior to annihilation , the energies and wavefunctions of the
atom may be approximated by those of hydrogen with the Bohr radius replaced by
2ao = 2n2 / me 2, owing to the change in reduced mass. What are the spin-dependent
components of the wavefunctions of positronium? (Note : The annihilation time of
the IS state of positronium for decay into two photons is :::: 1.2 x 10- 10 s. The 3S
state decays into three photons and lasts c; 1.4 x 10- 7 s. The fact that positron-
ium in the 3S state must annihilate through the emission of three photons is due to
the principle of charge conjugation. This principle states that electromagnetic in-
teractions are invariant under change of all particles to their antiparticles . 17) The
positron is further discussed in Section 15.3.

12.7 • THE HYDROGEN AND DEUTERIUM MOLECULES

Exchange Binding
Another important area in which symmetrization requirements imposed by the
spin of constituent particles plays a significant role is that of the theory of diatomic
molecules . The simplest of these is the hydrogen molecule, H2. The fact that
the proton nuclei are extremely more massive than the electrons permits analysis
of the molecule to be divided into two parts. 18 The first of these concerns the
chemical binding between the two atoms due to electron coupling . The second
addresses the motion of the nuclei within the bound configuration .
The Hamiltonian of the molecule, neglecting all but electrostatic interaction, is
(Fig. 12.17)

H =Hatom a + Hatom b + Vee + V pp + Vep + Tnuc

A PI 2 e2 e2
Hatoma=--- V = +- ee
2m rl rl2
(12.48)
h 2
e2 e2
= +-
A

Hatomb = -2 - - Vpp
m r: Fab

FIGURE 12.17 Radial dis-

A 1 2
Tnuc = 2M (Pa )
A
Vep = _e 2 (_1 + _1_)
rib rz«
tances appropriate to the hy-
drogen molecule . The mass of an electron is m, nuclear mass is M, and rl is the distance from
electron 1 to nucleus a. The repulsion between the two electrons is given by the
positive potential e 2 / r12 , while the repulsion between the nuclei is e 2 / Tab - The
potential Vep represents the cross attraction between electrons and protons. The
kinetic energy of the two nuclei is written Tnuc . Since m / M « 1, it is consistent
to view the electrons as moving in the field of two fixed protons (Pa 2 = Pb 2 =
0, Lab = constant) .
17For additional discus sion. see S. Gasiorowicz, Quantum Physics. Wiley. New York, 1974.
18This approximation is due to M. Born and J. Oppenheimer. Ann. Physik 84. 457 (1927) .
624 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

To uncover the binding between the two atoms, one constructs the wavefunc-
tions of H, further neglecting the interaction Vee + V ep. The coupling between
atoms which follows is then due primarily to antisymmetrization requirements
imposed on these wavefunctions in accord with the Pauli principle. The residual
Hamiltonian appears as

H = Hatom a + Hatom b (12.49)

with eigenstates
1
([JS ,A = ,J2[([JVa (rl)([JVb (r2) ± ([JVa (r2)([JVb (rl)) (12.50)

Following previous notation [e.g., (12.41)] atomic eigenvalues have been written
Va,b. Spin-dependent states parallel those constructed for the helium atom (12.43)
in that both problems address the antisymmetric states for two electrons . There
results
IX = ([Js(rl, r2)~A
(12.51)
2X =([JA(rl,r2)~s
Using these state functions, it is possible to calculate the expectation of the
total potential of the hydrogen molecule contained in H given by (12.48) .

(V) = e 2( - I + -1 - -1 - -1 - -1 - -1 ) (12.52)
fl2 Tab h ri fib ru
In calculating this average, ri and ri dependence is lost to integration, leaving
only dependence on the internuclear distance Tab - Thus we may write

The resulting two curves for the triplet and singlet states are shown in Fig. 12.18.
The potential of interaction is seen to have a minimum for the singlet state cor-
responding to antiparallel spins and symmetric space dependence as given by ([JS .
Thus, binding of the atoms is possible in the singlet state. As discussed in the pre-
vious section, electrons in the state ([JS tend to occupy the same region of space.
This common domain lies between the nuclei. At this location the electrons serve
to attract each of the protons and bind the molecule. The same mechanism, we
recall, is responsible for the triplet states of two-electron atoms lying lower in
energy than the singlet states (see Fig. 12.13). However, in the present case the
positive energy of repulsion between electrons is overbalanced by the negative
energy of attraction of the protons toward the overlap domain . If Fab is decreased
beyond the minimum in V(r ab), the nuclear repulsion begins to overcome the
binding afforded by the intermediary electrons and the atoms repel.
In the triplet state the antisymmetric wavefunction ([JA (rj , r2) is appropriate,
for which case ([J A (rj , rj) = 0, so that electrons do not tend to occupy the com-
mon domain between nuclei and there is no binding . This repulsion in the triplet
12.7 The Hydrogen and Deuterium Molecules 625

<v>
8

4 Antibonding

5
-2 Bonding

-4

-6

-8

FIGURE 12.18 Expectation of the hydrogen molecule potential (in units of 10- 12 erg)
versus internuclear distance rab (in units of Bohr radii). Shown also is a schematic of the
corresponding "bonding" and "antibonding" levels of the molecule.

state is evidenced by the monatonic increase of V(rab) with decreasing internu-

clear distance , as shown in Fig. 12.18.
In the symmetric electronic state, electrons may be said to be "shared" by the
two hydrogen atoms, thereby allowing each atom a completed (lsz) K shell. The
bond so effected is called a covalent bond . This bonding is to be differentiated
from that which couples , say, the NaCI molecule . In this case the sodium atom
gives its isolated 3s 1 electron to the vacancy in the (3sz3 p5) M shell of the chlo-
rine atom. In the resulting configuration, the positively charged sodium ion is
knitted to the negatively charged chlorine ion in what is termed an ionic bond .

Symmetric States for the Nuclear Motion of D2

Having discovered the nature of the binding of the Hz molecule, we tum next to
a discussion of the nuclear motion within this bound configuration . However, in
that we wish to address the construction of symmetric states relevant to bosons,
we will consider the isotope of hydrogen, deuterium . The deuterium atom has
at its nucleus a deuteron that has spin 1 and is therefore a boson. The Hamilto-
nian (12.48) carries over to Dz with the change that M becomes the mass of a
deuteron instead of a proton. The mass ratio then becomes Mf m ,.... 34,000 and
approximations introduced above for Hz are even more appropriate to Dz.
In the bound configuration the deuterons move within the effective potential
field V(rab) , and the Hamiltonian for the deuteron motion may be written

(12.53)
626 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

The kinetic energy of the two deuterons Tnuc may be rewritten in terms of center-
of-mass motion and motion relative to the center of mass [see (10.95)]. There
results
A 2 A2
Pr L -
+ - - 2 + VCr)
A A

Hnuc - HCM = - (12.54)

2J.L 2W

The variable r is written for the interdeuteron distance Lab - and J.L = M I2 is
the reduced mass of the two deuterons. If V (r) has its minimum at ro, then near
equilibrium one may write

-
VCr) -
= V(ro) I
+ -(r - ro) 2 (OV)
- + ... (12.55)
2 or ro

This parabolic potential gives rise to vibrational motion. At moderately low tem-
perature these vibrational modes are "frozen in." That is, they are not excited and
the deuterons assume the shape of a rigid dumbbell which is free only to rotate .l?
The Hamiltonian in this temperature domain reduces to the simple rotational form

(12.56)

where / = J.Lr02 is the moment of inertia of the dumbbell two-deuteron sys-

tem. Eigenstates of this purely rotational Hamiltonian are the spherical harmonics
liml) with corresponding eigenenergies (9.49)

Ji2 i (l + I )
El= --2-/-

The frequencies of emission due to transitions between rotational states (see Prob-
lem 9.6) lie in the infrared and are clearly distinguished from frequencies due to
transitions in the electron states which lie in the ultraviolet-visible portion of the
spectrum .
Spin-dependent eigenstates of fi nuc (12.56) are simply constructed in the prod-
uct form

Xnuc = lim I )~

Since the deuteron has spin I, it is a boson and Xnuc must be properly symmetrized
with respect to exchange of spin and space coordinates . The spin component ~ is
composed of the nine states derived from the addition of two spin-l particles .
These states were previously constructed in Chapter 9 and are listed in Table 9.5.
Of these nine states, six corresponding to s = 0 and s = 2 are symmetric with
19Characteristic rotational temperature for the D2 molecule is T = /i2/ 2IkB = 44K . Vibrational
modes are excited at 4500 K and electron states are excited at temperatures several orders of magni-
tude larger. For further reference, see G. Herzberg, Molecular Spectra and Structure, Van Nostrand
Reinhold, New York, 1950.
12.7 The Hydrogen and Deuterium Molecules 627

respect to exchange of spin coordinates, while the remaining three corresponding

to s = 1 are antisymmetric. This separation of states is listed below.

Symmetric Antisymmetric

XlsmsSISZ) = +lsmsslsZ )
:;7~~~ ] 11111)
12011)
12-111 )
s =2 11011)
11 - 111)
1= s 1

12 - 211)
10011) }s =0

In that the two deuterons form a dumbbell configuration , exchange of deuter-

ons is equivalent to inversion through the origin. It follows that this operation is
identical to the parity operation, so that the symmetry of a rotational state with
angular momentum quantum number l is (_1)1 (see Problem 12.21). Thus in
order that the states Xnuc be totally symmetric , they must be of the form

6Xnuc = Ilevenml)~S (ortho)

(12.57)
3Xnuc = Iloddml )~A (para)

For a given value of l , there are six Xnuc states corresponding to l even and three
states corresponding to l odd. As with the states relevant to helium, those con-
straining an antisymmetric spin component are denoted as para states, while those
containing a symmetric spin component are denoted as ortho states.zo
Owing to the commutation property (12.22), the exchange operator i is a con-
stant of the motion and one may expect transitions between states of different
exchange symmetry to be forbidden . Assuming a uniform population of states,
there are twice as many ortho states as there are para states. Thus in transitions
between states of different angular momentum , radiation due to (even-even) de-
cay is roughly twice as intense as that due to (odd-odd) decay.
A comparison of these properties of the nuclear wavefunctions for Hz and Dz is
listed in Table 12.5. Since the proton nuclei of Hz are fermions, the corresponding
Xnuc function must be antisymmetrized with respect to exchange of space and spin
coordinates. This reversal of symmetry requirements on Xnuc results in nearly a
complete reversal of intensity rules obtained above for the Dz molecule . (These
properties of Hz are further discussed in Problems 12.37 and 12.38.)
Finally, we note that the Xnuc wavefunction of lowest angular momentum for
the Dz molecule is the ortho ISO state. In this configuration both orbital and spin
angular momentum vanish (s = l = 0). The relative orientation of the spins of
the deuterons in this state may be described as antiparallel , although in fact this

20States of greater statistical weight carry the prefix arrha, whereas those of smaller statistical weight
carry the prefix para .
 a-
N
co

TABLE 12.5 Prop erties of the nuclear component wavefunctions for D2 and H2 n
:r
C.>
"0
D2 H2
~
N

»
"0
"'5!...
('i '
~
0'
::J

51 =I 51 = 4 8'
»
=
Xnuc Ilml) ~ Xnuc = Ilml) ~ 8'
3
Xnuc is symmetric Xnuc is antisymme tric ,n
~
o
Multiplicity Symmetry of Multiplicity Symmetry of
~
s 2s + I ~spin Classification I s 2s + I ~spin Classification I c
,....CJ
2 5 Symm etric ortho even I 3 Symm etric ortho odd (J)
Q..
I 3 Anti symmetric para odd 0 I Anti symm etric para even s;
If,
0 I Symm etric ortho even or
,ro
C.>
::J
c,
Intensity ratio for Inten sity ratio for Z
rotational transitions: I 1:2 rotational transitions: I 3: I c
o
odd-odd/even-even odd- odd/ even-even
r""CI
:r
~
State of lowest angular State of lowest angular ('i '
momentum: ISo (ortho) momentum: ISO(para) '"
Problems 629

IOO!!} state is the superposition

(The diagrammatic representation on the left is explained in Table 9.5.)

In this section we have found how symmetry requirements imposed by the
spin of constituent particles strongly influence the physical properties of diatomic
molecules . Antisymmetrization of the electron wavefunctions was found to give
rise to exchange binding of the hydrogen molecule . Symmetry requirements on
the wavefunctions for the two-bo son deuteron system were found to give rise to
intensity rules for molecular radiation. Symmetrization requirements will enter
again in the next section of this chapter, wherein the quantum mechanical basis of
superconductivity and superfluidity will be described.

PROBLEMS

12.33 It is found experimentally that hydrogen atoms with parallel electron spins repel in
scattering from each other. What is the reason for this repulsion ?
12.34 Using moment of inertia values relevant to the Hz molecule, show that the frequency
Ii/I lies in the infrared.
12.35 (a) What is the spring constant K for the vibrational coupling between the
deuterons in the Dz molecule in terms of the effective potential V(rab) and
the equilibrium radius ro?
(b) Is there a coupling between the spin of two deuterons and their vibrational mo-
tion (in one dimension)? Explain your answer.
12.36 (a) In what manner are the following two physical phenomena related? (1) The
triplet states of He lie lower than the singlet states; (2) Hz is bound in the singlet
electronic state.
(b) What is the radial probabilit y distribution for the nuclei of either Hz or Dz in
the ISO state? Where are the electrons with respect to this distribution ? In what
temperature domain is your description appropriate ?
12.37 The nuclei of the ordinary Hz molecule are protons that have spin! and are there-
fore fermions .
(a) What exchange symmetry must o/nuc have for the Hz molecule ?
(b) How many antisymmetric and symmetric spin states are there for two spin-!
particles? (See Table 11.2.)
(c) What is the ratio of intensities of spectral lines due to transitions between even
rotational states to that of lines due to transitions between odd rotational states?
How does this ratio compare to that for the heavy hydrogen molecule Dz?
12.38 As discovered in Problem 12.37, the nuclear component of the wavefunction for the
Hz molecule must be antisymmetric . In that there are three symmetric , s = 1, triplet
spin states and only one antisymmetric , s = 0, singlet spin state, the symmetric (or
630 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

ortho) states are three times more prevalent than the antisymmetric (or para) states.
At temperatures sufficiently high to populate the rotational levels, one expects to
find molecules predominantly in odd rotational states. Describe qualitatively how
this population of rotational and spin states changes with decrease in temperature.
Specifically what rotational state should prevail near 0 K?

Answer
For k B T less than the energy between rotational levels, the molecules in an S ro-
tational state cannot be excited out of that state. Owing to the Pauli principle, this
symmetric rotational state must be accompanied by the antisymmetric (para) spin
(s = 0, IS) state. Hydrogen at room temperature is a mixture of about 3:1 or-
tho to para molecules. However, near 20 K, the sample undergoes an ortho-para
conversion. Beneath this temperature, molecules that are in the IS state become
"frozen" in this state and the sample becomes comprised almost entirely of para
molecules.

12.8 • BRIEF DESCRIPTION OF QUANTUM MODELS FOR

SUPERCONDUCTIVITY AND SUPERFLUIDITY

Bose-Einstein Condensation
The spin-statistics relation, which requires that ferrnions obey the Pauli exclusion
principle and that bosons exist in totally symmetric states, has profound physical
implications. We have seen that (Section 8.4) the mechanism of conduction in
solids is intimately related to the fact that electrons, which have a spin of are t,
ferrnions and therefore obey the Pauli exclusion principle.
Bosons (i.e., particles with integral spin values) have equally significant prop-
erties. Most interesting of these perhaps is the phenomenon of Bose-Einstein con-
densation. Since bosons do not obey an exclusion principle, a gas of such parti-
cles can conceivably be in a state in which all particles have the same momentum,
same energy, and so on.
From kinetic theory we recall that the temperature T of a gas of particles is
defined as21

3 (t:.p)2 1 2
-kBT = - - == -((p - (p)) ) (12.58)
2 2m 2m

The temperature is proportional to the mean-square deviation of the momentum of

the particles in the gas. It follows that at low temperatures, there is a small spread
of momentum values away from the mean. A gas of particles which all have the
same momentum has zero temperature, even though this momentum value may

21Recall also the thermodynamic identification, T- 1 = 8Sj8E, where S is the entropy. This defi-
nition of temperature is more uniformly valid for low-temperature quantum systems than is (12.58).
Thus, whereas the latter formula implies a finite temperature for a collection of fermions in the ground
state, the thermodynamic relation gives the correct value , T = O.
12.8 Quantum Models for Superconductivity and Superfluidity 631

Px
Px Px + 6px
(a)

fip x

Px Px

(b) (c)

FIGURE 12.19 (a) The distribution function f(px) relevant to a gas of particles con-
strained to move in one dimension . The function f(px) is such that the number of particles
with momentum in the interval Px to Px + 8px is f(px) 8px (the shaded area). The total
number of particles in the gas is

The temperature of a gas is a measure of the mean-square deviation from the mean, (tlp)2,
of momentum values of the particles in the gas. Thus a one-dimensional gas of particles
with the distribution (b) is colder than the same gas of particles in the distribution (c).

be large. In a frame moving with the gas, however, zero temperature means that
all momenta are zero 22 (Fig. 12.19).
Suppose that we look at a box of bosons . The box is fixed in space. Lower
the temperature. At zero temperature, momentum values drop to a minimal value
which is consistent with the uncertainty principle (see Problem 12.40). This col-
lapse of an aggregate of bosons to a collective ground state in which they all have
22For further discussion of the kinetic definition of temperature, see R. L. Liboff, Kinetic Theory:
Classical, Quantum and Relativistic Descriptions , Prentice-Hall, Englewood Cliffs, N.J., 1990.
632 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

R/R 4 .2K

1.0 ---------~-----

0.5

3.75 3.78 T(K)

FIGURE 12.20 Resistance versus temperature for tin. The critical temperature for lead,
T = 7.18 K, is higher than that of tin.

the same minimal ground-state momentum eigenvalues is called Bose-Einstein

condensation. The interesting properties of a system in such a condensed state
may be related to the uncertainty principle. As the momentum of a particle in the
system falls to lower values, its uncertainty in position grows. Loosely speaking,
it is in many places at the same time. There are two well-established phenomena
in nature which are related to Bose-Einstein condensation: superconductivity and
superfluidity.
At very low temperatures (near 0 K), certain metals (e.g., tin and lead) be-
come superconductors .P If a current is established in a superconducting loop,
it maintains itself with zero loss. No potential difference is needed to keep the
current flowing. The resistance drops to zero below a certain critical tempera-
ture, Tc (Fig. 12.20). Also, magnetic fields become completely excluded from
a superconducting sample for temperatures below Tc (the Meissner effect; see
Problem 12.39). For tin, Tc = 3.73 K, and for mercury, Tc = 4.17 K.

Cooper Pairs
It has been established/" that below the critical temperature, interaction be-
tween electrons and the vibrational modes (phonons) of the positive ion lattice
of the metal results in a diminution of the Coulomb repulsion between electrons.
Phonons are quantized lattice vibrations. When averaged over many such phonon
emissions and absorptions, at sufficiently low temperature, the effects of these
deformations overbalance the Coulomb repulsion and yield a net attraction be-
tween electrons.P This attraction allows pairs of electrons to couple with spins
antialigned so that each pair carries zero net spin. In that their spin values are
zero, these Cooper pairs act like bosons. Beneath the critical temperature, these
pairs collapse to a collective ground state, XG . This state is described as follows.
Let l;j == (r., s.), where rj and sr, respectively, are the space and spin coordinates
23Superconductivity was disco vered by K. H. Onn es in 1911.
24J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Phys. Rev. 108, 1175 (1957).
25This attractive mechani sm was first suggested by H. Frohlich, Phys. Rev. 79, 845 (1950) .
12.8 Quantum Models for Superconductivity and Superfluidity 633

of the ith conduction electron in the sample . The ground state is comprised of the
appropriately symmetrized sum of products

XG = L (±)p <P(SI , S2)<p(S3 , S4) . .. <p(SN-I, SN)

P(I .2•....N)

relevant to an N-particle fermion system [compare with (12.33)]. The spin com-
ponents of the bound two-particle states <p(Si-l , Si) is the singlet spin-zero state,
SA (see Table 11.3). In this sense the ground state may be thought of as being com-
prised of bosons. However, in the space component of these two-particle states is
too widely spread to allow a consistent localized particle picture.P' When Fourier-
analyzed, it is found that electron pairs in the preceding ground state have equal
and opposite k vectors corresponding to zero total momentum.
Excited states in the material are separated from the ground state by an energy
gap which at temperature 0 K is centered about the Fermi energy. When Cooper
pairs are excited above the gap, they break up into two electrons that exhibit nor-
mal conduction. However, in the superconducting state, current flowing in a cur-
rent loop has been observed to persist indefinitely. Such current is attributed to a
collective motion of Cooper pairs in the superconducting ground state.

Superfluidity
A second example of Bose-Einstein condensation is superfluidity. In 1932
Kapitza-? discovered that the viscosity of liquid helium drops dramatically be-
neath the A point (2.19 K). This absence of viscous effects allows the helium to
flow freely through capillaries with diameters as small as 100 A.
At pressures less than 25 atm, helium is a liquid at 0 K. If the heat capacity
of liquid helium is measured, a singularity is observed at about 2.2 K, which
suggests a phase transition (Fig. 12.21). The viscosity in the new phase (beneath
T;,J is essentially zero, while the thermal conductivity is very high. In the new
phase , helium is a superfluid. Below TA , helium is called helium II. Helium II is a
mixture of superfluid and normal fluid.
The ground state of helium is ISO. The electron spins are antialigned with total
spin zero. The orbital angular momentum is zero. The nucleus also has zero spin.
The whole helium atom has zero angular momentum and is a boson. The super-
fluid component of helium II contains atoms condensed to the collective ground
state.

Landau Theory
The first theoretical model related to superfluidity is due to L. Landau .i" In this
analysis the liquid interacts with the walls of the capillary through which it is
flowing, via quantized vibrational modes of excitation generated in the liquid.
26For further discussion. see N. W. Ashcroft and N. D. Merman.Solid State Physics , Holt, Rinehart
and Winston, New York, 1976, Chapter 34.
27p. L. Kapitza, Nature 141. 74 (1932).
28L. Landau, J. Phys. 5. 71 (1941); see also L. D. Landau and E. M. Lifshitz, Statistical Physics ,
Addison-Wesley, Reading. Mass., 1958.
634 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

3.0

2.5
,...
.,
IlO

~ 2.0
~
';;j
s.
~ 1.5
/
'u
...
...u
0.
V
-...
.,
:z:
1.0

./
/
~

l
0.5
V
,,-
o
1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
T(K)

FIGURE 12.21 Heat capacity of liquid helium. The singularity near 2.19 K is evidence
of a phase transition . The viscosity of the liquid above the transition temperature is similar
to that of normal liquids, while below this temperature the viscosity is at least 106 times
smaller than that above the transition .

Consider that the superfluid is moving through the capillary with velocity Y.
In a frame moving with the fluid (in this frame the fluid is at rest) the capillary
wall moves with velocity - y (Fig. 12.22). Owing to friction between the wall and
the fluid, elementary excitations appear in the liquid. Let one such excitation be

+v +v

(a)
z

(b)

FIGURE 12.22 (a) In the lab frame the capillary is fixed. Fluid moves with velocity -j-v,
(b) In the frame moving with the fluid, the fluid is at rest while the capillary moves with
velocity -v.
12.8 Quantum Models for Superconductivity and Superfluidity 635

generated with momentum p and energy e(p) . Transforming back to the frame
where the tube is at rest (i.e., the lab frame) , the energy of the liquid becomes

1
E = "2Mv2 + [e(p) + p' v] (12.59)

where M is the mass of the liquid. The p . v term stems from Doppler shifting of
the phonon frequency (see Problem 12.41). If no excitation is present, the energy
!
of the fluid is M v2 • The presence of the phonon excitation causes this energy to
change by the amount [e(p) + p' v]. Since the energy ofthe flowing liquid must
decrease owing to such dissipative coupling,

e(p) + p' v < 0

This condition must be satisfied in order than an excitation appear in the liquid.
Since

(10 + P • v) 2: 10 - pv

it follows that excitations have the property

e(p)
v > -- (12.60)
p

We may conclude that an excitation of energy 10 and momentum p cannot be

created in a fluid moving past a wall with speed v unless the preceding inequality
is satisfied. If 10/ P has some minimum value greater than zero, then for small
velocities of flow beneath this minimum , dissipative excitations will not appear in
the liquid. That is, the liquid will exhibit superfluidity.
If the energy spectrum of excitations 10 (p) is plotted against p, then the minima
of 10 / p occur as those values of p where

de 10
(12.61)
dp = P
that is, at points where a line drawn from the origin of the pe plane is tangent to
the curve e(p) (see Problem 12.44).
The e(p ) curve for liquid helium has been obtained by neutron scattering ex-
periments-? It is sketched in Fig. 12.23. There are two values of p where 10/ P
is minimum, at p = 0 and p = Pl . The minimum at p = 0 is appropriate to
temperatures near zero. At such temperatures superfluidity occurs for speeds

v < vo=- del

dp p=o

29J. L. Yarnell et aI., Phys. Rev. 113, 1379, 1386 (1959) .

636 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

E:(p)

Vo =tan t/Jo
VI = tan t/JI

PI P

FIGURE 12.23 The minima of s(p) / P occur at values of P where a line drawn through
the origin is tangent to s(p ). For liquid helium this occurs at the origin and at PI ' Excita-
tions at the origin are called phonons. Those at P I are called rotons .

For slightly larger temperatures the minimum at PI comes into play. Superfluidity
occurs for this branch of excitations at fluid speeds

v< VI = de I < Vo
dp P=Pl

The speed of phonons at P = 0 is that of sound in liquid helium at 0 K.

The excitations at PI are called rotons corresponding, it is believed, to rotational
motion of small clusters of helium atoms. 30
The isotope of helium, He3 , has an unpaired neutron and is therefore a fermion.
One would not expect this isotope to exhibit Bose-Einstein condensation. Recent
experimental observation'! however, suggests the existence of a superfluid phase
in this liquid as well. As with the case of superconductivity, such phenomena
may be ascribed to a pairing process of fermions allowing for Bose-Einstein-like
behavior.

PROBLEMS

12.39 Consider a sphere of tin immersed in a uniform magnetic field at T > Te . The
finite conductivity of the tin permits the B field to penetrate. Inside the tin the field
has the value B '. If the conductivity becomes infinite as T -+ Tc, what happens to
B'?
Answer
From Faraday's law
aB 1
- = -V x 8 = --V x J
at a
30R. P. Feynman, Phys. Rev. 74 , 262 (1954 ); for further discussion of this topic, see D. L. Goodstein,
States ofMauer, Prentice-Hall, Englewood Cliffs, N.J., 1975.
31D. D. Osheroff.R. C. Richardson, and D. M. Lee, Phys. Rev. Lett. 28,885 (1972).
Problems 637

we obtain that

fora = 00
so that 13 is constant in time and is trapped inside an ordinary conductor. TIlls
is not the case for a superconductor. At T = Te, a magnetic field is excluded
from a superconductor, so that 13' = 0 for the configuration given. In this sense
a superconductor is said to have perfect diamagnetism. This effect is called the
Meissner effect.
12.40 Consider a model of superconductivity where the critical temperature Te describes
the width of the Fermi sea (see Sections 2.3 and 8.4). Estimate the spatial spread
(Ar) of the wavefunctions for such electrons in tin.

Answer
I
_(~p)2::::: kBTe
2m
(~p)2 ::::: mVF~p

where

Together with the uncertainty principle , this gives

h hVF 4
~x > - ::::: - - ::::: 10- cm
- ~p kBTe

which is a macroscopic length.

12.41 An excitation has energy £ = Iuo and momentum p = hk in a frame S. Show that
in a frame S' moving with velocity v with respect to S,

£' =£ - p' v

Answer
The energy of a photon of frequency co in the S frame is

£ = hw

In a frame moving with v with respect to S, the frequency is Doppler-shifted to the

frequency

w' =w-k ·v

(If k is parallel to v, w' decreases . If k is antiparallel to v, w' increases.) Multiplying

through by h gives the desired result.
12.42 (a) Use the uncertainty relation to estimate the lowest temperature that a collection
of bosons confined to a box of edge length L can have.
(b) What temperature does this correspond to for helium in a box with edge length
I cm?
638 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

12.43 A classical gas is said to be degenerate when thermodynamic properties of the sys-
tem (equation of state, specific heat, conductivity, etc.) are governed by quantum
statistics, as opposed to classical, Boltzmann statistics . A criterion that determines
if a gas is degenerate may be obtained by comparing the mean interparticle distance
n- I / 3 (n = particle number density) with the average de Broglie wavelength of
particles .
(a) What is this criterion in terms of n, T, and m, where T is the temperature of the
gas and m is the mass of a particle ?
(b) Use this formula to estimate the temperature at which a neutron star of mass
density 1014 g/cm3 becomes degenerate .
12.44 Show that s(p ) / p is minimum at those values of p for which
ds s
dp = P
12.45 Show that superfluidity does not occur if excitations have the free-particle spectrum

[Note: If the liquid is a system of uncoupled bosons, one expects that excitations
follow this spectrum . N. N. Bogoliubov 32 was the first to show that a gas of bosons
with weak interactions has a spectrum of excitations et p ), which has a finite slope
at p = 0 (such as sketched in Fig. 12.23).]
12.46 A certain Bose liquid has the excitation spectrum

s( p )
2
= P [b + (p _ b)2]
So 3
- p
p=-
Po
where so' Po, and b are constants.
(a) What are the maximum superfluid speeds for the photon and roton branches of
excitations , respectively?
(b) What is the energy gap for the roton branch of excitations ?
12.47 Give a qualitative explanation of the fact that superconductors are poor normal con-
ductors .

Answer
Superconductivity is due to electron-phonon interactions. Metals with strong
electron-phonon interaction will show large resistance at room temperature and
therefore be poor conductors. On the other hand, strong electron-phonon interac-
tion will raise the critical temperature beneath which superconductivity becomes
evident. Such, for example, is the case for lead, which is a poor conductor but
has one of the highest critical temperatures . On the other hand, superconductiv-
ity in gold and silver, which are very good conductors at room temperature and

32 N. N. Bogoliubov, J. Phys. USSR 11, 23 (1947).

Problems 639

must therefore be typical of a weak electron-phonon interaction, proves difficult to

exhibit.
12.48 When He II is constrained to flow in a circular channel, the circulation maintains
itself with no dissipation. Let particles in the ground state have the wavefunction
rp = Aexp[iifi(r)].
(a) The velocity field u of particles in this state is related to mass current 33 through
the relation Jm = mA 2u . Show that u = (li lm)Vifi .
(b) Show that the values of the circulation are restricted to the discrete quantum
values

f u ·dl=na (n=O ,±I,±2, .. . )

h
a '=-
m

The constant a :::: 10- 3 cm 2 Is for He.

Answers
(a) J = 1iA2Vifi.
(b) In that u is the gradient of some function, V x u = O.
(c) Around the path of flow we have

f u • dl = ~ f Vifi • dl

To ensure that the wavefunction is single-valued, change in ifi about the closed
loop is restricted to integral multiples of Zn .

12.49 The halogens, which comprise group VII of the periodic table, are characterized
by the common property of missing one electron in the outermost p subshell. It
is found that the ground states of these atoms are well described by the equivalent
configuration of a single hole bound to an atom in an orbital p state. Using this
model, obtain the possible ground states of a halogen atom. Check your answer
with the ground states given in Table 12.3. The notion of a hole was introduced in
Chapter 8 and is discussed further in the following section.
12.50 In the Heisenberg model for ferromagnetism , the Hamiltonian for an array of mag-
netic moments is given by
N
H=- ALLl-ti .l-tj- Ll-ti· B
j i-I

where A is a positive constant. The first sum is over "nearest neighbors," and the
remaining sum extends over the N moments in the sample. The form of the first
sum in this Hamiltonian presupposes that aligned magnetic moments are lower in
energy than antialigned moments.
(a) Is this description consistent with classical physics?
(b) Offer a quantum mechanical explanation for the Heisenberg model, citing the
Pauli principle appropriate to electrons.
33Note the relation Jm = mJ. where the particle current J is defined by (7.107).
640 Chapter 12 Appl ication to Atomic , Molecular, Solid-State, and Nuclear Physics

12.51 (a) Referring to Fig. 12.15, give the "first" ionization energy of helium.
(b) Recalling that the remaining electron after ionization is left in the ground state
of the Z = 2 atom, calculate the "second" ionization energy of helium. Com-
pare your answer with the observed value, 54.4 eV.
12.52 The relativistic wave equation for bosons of rest mass m may be obtained by trans-
forming the relation (see Problem 2.26)

E2 = p2 e2 + m 2e4
through the identifications

, a
E -+ E = ili-
at
p-+p=-iliV

(a) Obtain the wave equation relevant to bosons of rest mass m. This equation is
called the Klein-Gordon equation. (This equation is discussed in detail in Sec-
tion 15.2.)
(b) What form does this equation assume for photons?
(c) Obtain a time-dependent isotropic solution to the Klein-Gordon equation for
bosons of finite rest mass . What characteristic decay length does this solution
imply?
(d) What is the length for n mesons (mrr ::: 260me)?

Answers
a
1 2 1fr
(a) V21fr - - -
m 2e2
= -1fr
e 2 at
2 li2

~2 a 1fr2
2
(b) V21fr _ = 0
e at
Equation (b) is appropriate to the propagation of electromagnetic fields in vacuum,
for either the scalar or vector potential.
(c) Referring to (C.16), we find (for r > 0)

Ae- r/ a
1fr(r) = --
r
Ii
a=-
me

(d) arr = Xc/ 260 = 1.49 x \0-13 cm = 1.49 fermi

12.53 A deuteron is composed of a bound neutron and proton. Its angular momentum
quantum number is j = 1. The spin-orbit component of the Hamiltonian is given
by

where u. is the reduced mass of the two-particle system and e is the speed of light.
 12.9 Impurity Semiconductors and the p-n Junction 641

(a) Neutron and proton both have spin !.

What are the possible values for the total
spin quantum number s of the deuteron?
(b) Rewrite Ii' as a function of 1, t, and S.
(c) What are the possible eigenvalues of H'? Leave your answer in terms of f.J.,C 2 .
12.54 The outer shell of neon in the ground state is in the 2 p 6 configuration. If one of these
electrons is excited, one of four lines are observed in subsequent decay. Employing
a model similar to that discussed in Problem 12.49, argue how these four lines
might emerge from a coupling between the excited electron and the 2 p 5 unexcited
electrons .
12.55 Lithium is a three-electron atom. Working in the central field approximation and
neglecting L-S coupling:
(a) Write down an appropriately symmetrized ground-state wavefunction, o/G(l,
2,3) , for lithium with spin taken into account. Use the notation o/nlm(i)a(i)
and o/nlm(i)f3(i) for single-electron wavefunctions . The variable i is electron
number and runs from 1 to 3.
(b) What is the ground-state energy EG of lithium corresponding to your answer
in part (a)?
(c) Your answer to part (b) presumes that EG is the eigenvalue of Hcorresponding
to the eigenfunction vo- This is,

What is the explicit form of H(I , 2, 3), and what property of EG allows it to
be an eigenvalue of the extended wavefunction 0/G ?
(d) What is the answer to part (b) if electrons have spin zero?

Answers (partial)
0/100(l)a (l) o/lOO(l) f3 (l) o/200(l)a(l)
1
(a) o/G(l, 2, 3) = Mi o/lOO(2)a(2) 0/\ 00(2) f3 (2) o/200(2)a(2)
v3!
o/lOo(3)a(3) o/lOo(3)f3(3) o/200(3)a(3)

(b) EG = -Z2JR (~
12
+ ~)
22
= - (~) JR
4
(c) EG is degenerate, so that in the present instance, all six wavefunctions in
o/G(l , 2, 3) correspond to the same energy, -9Z 2JR/4.

12.9 • IMPURITY SEMICONDUCTORS AND THE p-n JUNCTION

The concepts of valence and conduction bands were described in Section 8.4
in reference to electrical conductivity in solids. Having developed the theory of
atomic structure in the early part of this chapter, we are now prepared to pursue
the various aspects of impurity semiconductors. Semiconductors which conduct
by virtue of impurities are said to be extrinsic. Intrinsic semiconductors, on the
other hand, are not dependent on impurities for conduction .
642 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

Donor (n)-Type Impurities

Silicon, which is in the IVth group of the periodic chart, has four valence electrons
(3s23 p 2) .
Antimony lies in the Vth group of the periodic chart and has five valence elec-
trons (5s25 p 3 ) . Thus, if an impurity antimony atom replaces a silicon atom, four
of the five valence electrons contribute to the bonding of the impurity atom to
the silicon crystal. The fifth electron remains loosely bound to the parent anti-
mony atom. A schematic two-dimensional illustration of this bonding is shown in
Fig. 12.24. Under minor thermal agitation, the impurity atom is ionized and the
loosely bound electron is excited to the conduction band. See Fig. 12.25. Once
in the conduction band, the electron becomes a charge carrier and contributes to
current. Since it is considerably more difficult to activate the valence electron of
silicon to the conduction band, there will, in general, be a larger fraction of elec-
trons in the conduction band from the impurity antimony component than from
silicon atoms. This explains the experimentally observed large change in conduc-
tion for such crystals due to the presence of impurities .
Since conduction electrons are supplied by the antimony atoms, this form of
impurity is called a donor-type impurity . Extrinsic semiconductors which contain
a donor-type impurity are called n-type semiconductors. (Think of n for nega-
tive.)

0•
•
• Conduction •
electron

0• cf (0
• •
• • • •
•

(0 (0 0
• • • •
• • •
•
0) 0)
• •
• •
0)
FIGURE 12.24 An antimony atom in a silicon crystal contributes a conduction electron.
12.9 Impurity Semiconductors and the p-n Junction 643

Conduction Conduction
o electron band

Eg '-- ~-_---I._--_--I

Ed ------'------r-----
i
E

o
Filled valence
band

FIGURE 12.25 Energy level Ed of an impurity antimony atom in a silicon crystal lies
close to the bottom of the conduction band. The energy gap between valence and conduc-
tion bands is E 9 .

Acceptor (p)-Type Impurities

Consider now that a silicon atom is replaced with an atom from group III of the
periodic chart, such as indium. Since In has a valence of three (5s25pI), one of
the four bonds to nearest neighbors will be missing an electron. See Fig. 12.26.
This unsaturated indium bond effects an energy level slightly above the top of the
valence band as shown in Fig. 12.27. Under thermal agitation an electron from
the valence band of the host silicon crystal is excited to this level, leaving an
unfilled level in the valence band. This unfilled level, which affects the presence
of positive charge, is called a hole. If an electric field is applied, the hole will drift
in the direction of the field and contribute to the current.
Since an electron is accepted by the impurity atom in the creation of a conduct-
ing hole, the impurity atom is called an acceptor and the semiconductor is said to
be p type. (Think of p for positive.)
Note that of the three atomic groups considered above, type IV atoms act as
hosts to impurity atoms from type III (p-donor) and type V (n-donor) groups.
Molecular semiconductors, on the other hand, are commonly comprised of
 0•
•
• •

0• 0•
• •
• • •
•

0• 8 0
• • •
• • •
•
0• 0
•
• ~conduction hole due

0
to electron removed
to complete bonding
of In impurity atom

FIGURE 12.26 An indium impurity atom in a silicon crystal results in the presence of a
hole.

Conduction
band

E L..- --J

Ea-------------r--------

Valence
band

FIGURE 12.27 An electron from the valence band ionizes an impurity acceptor atom
and leaves behind a hole.

644
12.9 Impur ity Semiconductors and the p-n Junction 645

type III and V atoms (e.g., GaAs, InP) and may be doped with impurity atoms
from a number of other groups .34
For a semiconductor which contains both acceptor and donor impurities, one
writes

a = entu, + ep/Lp
for the conductivity, where /L denotes mobility and n, p denote electron and hole
densities , respectively. (See Problem 8.27.) In germanium at room temperature
(in MKS units),

/Ln = O.36(m/s)/(V /rn)

/Lp = O.17(m/s)/(V/m)

General Results. Fermi-Dirac Distribution and Densityof States

Consider a box of noninteracting electrons . The box has volume L 3 and is main-
tained at temperature T. Electrons may be found in any of a discrete infinity of
states. Owing to the exclusion principle , not more than two electrons can be in
anyone energy state at any given time.
The probability that a given energy level at the value E is occupied is given by
the Fermi-Dirac distribution relevant to ferrnions: 35 , 36
1
f(E) = e(E-EF) /ksT +1 (12.62)

One may also interpret f(E) as giving the fractional occupation of an electron in
the energy level E . The Fermi energy E F was discussed previously in Chapter 2.
At T = 0 K we see that f(E) = 1 for E < EF and f(E) = 0 for E > EF .
All levels up to EF are occupied at zero temperature. See Figs. 2.5 and 12.28.
To obtain the density (number/cm-') of electrons in the energy interval (E, E +
d E), one must multiply f(E) by the density of states (number of states/energy-
crrr'), which we have previously labeled 9 (E) (see Section 8.8). It has the value

9(E) = _1_ (2m*)3 /2 EI /2 (12.63)

2JT 2 li2

Here we have written m" for the effective mass introduced previously in Sec-
tion 8.2. Thus the density of electrons with energy in the interval (E , E + dE) is
given by f(E)9 (E) dE .
34This topic is discussed further by J, C. Phillips, Bonds and Bands in Semiconductors, Academic
Press, New York, 1973.
35Recall Section 12.3 and see Appendix A.
36The parameter E F in (12.62) is the Fermi energy at 0 K. For T > 0 K, E F in (12.62) is more cor-
rectly termed the chemic al potential. If two systems of a single species have equal chemical potentials,
there is no particle flow between the systems. If chemical potential s are different , particle flow occur s
in a direction to equalize the chemical potentials.
646 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

I(E)
T=O

T> 0

o E

FIGURE 12.28 At 0 K all levels are occupied up to the Fermi level. At finite tempera-
tures, levels above E F are occupied.

General Results
a. Electron Density First we tum our attention to the density of electrons , n, in
the conduction band. This is obtained by integrating the product 1 (E) 9 (E) dE
over conduction band energies . There results

n= 1cond oband
I(E)9 (E) dE (12.64)

If we consider electrons in the conduction band to be free, then 9 (E) is given by

(12.63). Since energies in (12.64) are measured from the bottom of the conduction
band, we write

9(E)=-1 (2m*)3
_e
/2 (E-E)I /2
(12.65)
2rr2 li2 c

where E; denotes the energy at the bottom of the conduction band. Equa-
tion (12.65) reflects the property that there are no states available in the energy
gap. With (12.62), we see that I(E) decays rapidly with increasing E . This
property permits the upper limit in (12.64) to be set equal to infinity. Combining
results and setting
1
f3=.-
kBT
(12.64) becomes
OO
n = _1_ (2m;)3/2
2rr 2 li2 1s;
(E - E c)I /2dE
e{3(E-EF) + 1
(12.66)
12.9 Impur ity Semiconductors and the p-n Junction 647

Here we have assumed that m; is constant over the range of integration. Changing
variables to

rJ = E - E;

(12.66) becomes

1 (2m;)3/2 r rJI /2drJ

(12.67)
n = 2n2 ~ 10 ef3'1ef3(Ec-EF) +1
To simplify this integration, we assume that E c - EF » kBT or, equivalently,
{3 (E c - E F) » 1. A material which satisfies this assumption is said to be nonde-
generate. 37 If E F lies in the conduction band, the material is said to be degenerate.
Given the preceding inequality, one may neglect the unity term in the denomi-
nator of (12.66) to obtain 38

(12.68)

which gives

(12.69)

where

1 (2m;k BT)3/2
Nc(T) = 4: nli2 (12.70)

represents an effective density of states in the vicinity of E ::::: E c , i.e., near the
bottom of the conduction band.

b. Hole Density We tum next to a parallel calculation for density of holes in

the valence band. A hole will be present at a given level providing an electron is
missing from that level. So the Fermi distribution for holes becomes 1 - f(E).
Thus the density of holes in the valence band is written

p= 1 valence band
9p(E)[1 - f(E)] dE (12.71)

When electrons in a partially filled band gain energy, they move upward in the
band, leaving behind vacated states. Thus holes which gain energy move down-
ward in their band. To incorporate this property and further stipulate that the zero
37That is, classical stati stics applies. See Problems 2.48 and 12.43. At room temperature, kB T ::::
0 .025 eY. A measure of Ec-EF is roughly given by E9 . Here are some typical value s: E9 = 1.14 eV
for Si, 1.25 eV for InP, and 1.4 eV for GaAs, for which cases kB T « E 9 . Note that, in this limit ,
mean occupation numbers per state grow small, which is a basic property of the classical domain.
10
38Here we recall 00 '11 /2e-fJ'I dn = ./ii/2f33 /2 .
648 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

of energy lies at the top of the valence band, the energy of a hole is written E; - E,
where E varies from E; to -00 . With this expression for energy substituted in the
density of states (12.63), the preceding expression becomes

1 (2m* Ev
1]
P= 2rr2 11/
)3/2
1 -00 (E; - E)I /2
[
1- +1
eP(E-EF) dE (12.72)

Following steps leading to (12.65), we introduce the variable 7} = tE; - E) and

again assume nondegeneracy, i.e., f3(EF - E v ) » 1. There results

(12.73)

where

N (T) = _1 (2m*k
h B
T)3/2
(12.74)
v 4 n li 2

is the effective density of holes near the top of the valence band. Multiplying
(12.69) and (12.73), we find

np = -1 (k~T)3 (m*m*)3 /2 e-Eg / k B T (12.75)

2 n li2 e h

This gives the important result that the product of nand p for a given semicon-
ductor is constant at a given temperature. The key assumption in obtaining the
preceding result is that the semiconductor is nondegenerate. Thus, with the rule
(12.75), we see that introducing an impurity which would, say increase n would
diminish p, since the product np is constant. The generality of the result (12.75)
follows from the observation that it is an application of the law of mass action to
electron-hole dynamics .V

Expressions for the Fermi Energy

a. Intrinsic Case For intrinsic semiconductors, for every electron in the conduc-
tion band, there is a hole present in the valence band. Thus we may set

n; = Pi (12.76)

where the subscript i denotes intrinsic . Equating (12.69) to (12.73) gives

Solving for E F gives the intrinsic Fermi energy :

39That is, for the chemical reaction A :;=: B+C, one may write nBnC I n A = f(T) , where nA denotes
number density of A particles, etc., and f(T) is an arbitrary function of temperature . Thus, with A
denoting neutral atoms, B electrons , and C ionized atoms, one obtains the form of (12.75).
12.9 Impurity Semiconductors and the p-n Junction 649

Intrinsic

1 v
EF = -(E 3
+ E c ) + -kBT In 2 (m*) (12.77)
24m;

If the zero of energy is taken to be at E v , then E; becomes the value of the energy
gap E 9 and (12.77) may be rewritten in the more concise form

1
EF = -E 9+
3
-kBTln 2 (m*) (12.78)
2 4m;

Note in particular that energy values in semiconductor theory are not absolute but
are relative to a given reference level. Thus in (12.78), the E = 0 value is set at
t
the top of the valence band and EF (for m'h = m;) has the value of E 9 , whereas
in (12.77) EF is equal to the average of E v and E c • In all reference schemes EF
(for equal m*) lies halfway between E; and E c . However, in practical usage, this
is not the case. For example, for Ge, mh/ m; = 0.67 and for Si, mh/ m; = 0.54 .
In both cases EF falls slowly with increasing temperatures.
Returning to the density of charge carriers, for intrinsic semiconductors we
may write

(12.79)

With (12.69) and (12.73) there results

Intrinsic

(12.80a)

The effective density of states in the conduction band Nc<T) is given by (12.70),
and Nv(T), corresponding to the valence band, is given by (12.74).
Returning to (12.75) we note that np at a given temperature is constant. Eval-
uating this constant in the intrinsic domain gives the relation

(12.80b)

This is a very useful relation as it is valid for both intrinsic and extrinsic semicon-
ductors (in the nondegenerate limit and in the absence of recombination).

b. n-Type Semiconductors. Low-, Intermediate-, and High-Temperature Parame-

ters Consider a crystal doped with donor-type impurities. Let Nd represent the
density of impurity donor atoms. Then if density of unionized donor atoms is Ndo'
density of ionized donor atoms is Nd - Ndo' Charge neutrality may then be written

(12.81)
650 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

At relatively low temperature there are very few electron-hole pairs created , and
the preceding equation becomes

(12.82)

Thus, in n-type semiconductors, electrons are majority carriers and holes are mi-
nority carriers.
To calculate the Fermi energy for this case, first note that the density of neutral
donors Ndo is equal to the number of donor states per volume that are occupied .
Thus we may write

(12.83)

Substitution of (12.82) gives

n = Nd [1 + etJ:ErEd) ]
(12.84)

Passing to the limit kBT « EF - Ed gives

n = Nde-(EF-Ed) /ksT (12.85)

At low temperatures, n < Nd so that EF > Ed, and we may conclude that for
the present case EF lies between Ed and E c . See Fig. 12.29. Equating this latter
expression to our previously derived expression for n (12.69) gives

(12.86)

from which we find

LowT

(12.87)

At T = 0, E F is midway between E; and Ed. As temperature rises, E F falls

since N c > Nd .
At intermediate temperature nearly all donors are ionized . Assuming suffi-
ciently large impurity density so that Nd »ni, (12.82) gives

(12.88)

Again employing (12.69), together with the preceding equation, we find

Nd = Nce-(Ec-EF) /ksT (12.89)

Solving for the Fermi energy gives

Int. T

(12.90)
12.9 Impurity Semiconductors and the p-n Junction 651

- - - - - - - - Ed

feE) o
FIGURE 12.29 At low temperatures, E F lies between Ed and E c for an n-type semi-
conductor. A similar diagram applies to p-type semiconductors for which, under similar
conditions, E F lies between Ea and E v . At such temperatures, electrons are said to be
"frozen out" of the conduction band (donor levels remain occupied). Likewise, holes are
frozen out of the valence band (acceptor levels remain occupied with holes).

At sufficiently high temperature, electron-hole generation across the energy

gap grows large compared with impurity contributions. In this limit the semicon-
ductor behaves intrinsically and (12.77) et seq. apply.

c. p-Type Semiconductors. Low-, Intennediate-, and High-Temperature Parame-

ters For a semiconductor doped with acceptor-type impurities, charge neutrality
is written
(12.91)
Again note that at low temperature there are very few electron-hole pairs, and the
preceding relation becomes
p = Na - Nao (12.92)

The right-hand side of this equation represents the density of ionized acceptor
atoms. Each such ionized atom represents a filled acceptor state, and we may
write
(12.93)
652 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

Again we assume f3(E a - EF) » 1. Substituting the resulting expression into

(12.92) and recalling (12.73) gives
LowT

(12.94)

1
EF = -(E v - Ea ) + -kBTln
1 (Nv)
- (12.95)
2 2 ».
Again , the Fermi level at T = 0 is midway between the top of the valence band
and the acceptor levels . Since N v > Na , the Fermi level rises with increasing tem-
perature. At intermediate temperature, electrons in the valence band are thermally
excited to the E a level, ionizing all acceptor atoms . For sufficient doping density ,
charge neutrality becomes

p=Na

or, equivalently,

Taking the logarithm of both sides gives

IntT

(12.96)

At sufficiently high temperature, it is again the case that electron-hole gener-

ation across the gap grows large compared with impurity contributions, and the
semiconductor behaves intrinsically and (12.77) et seq. apply.
These effects are depicted in Fig. 12.30 for the case mit ;:::: A general m;.
compilation of results is presented in Table 12.6.

Compensation
An intrinsic semiconductor section is often an important component of a semi-
conductor device. 4o However, it is frequently the case that semiconducting ma-
terials are not free of impurities, which render the material either n type or p
type . Thus, for example, in high purity, both silicon and germanium tend to be
p type. In this case, the method of compensation refers to the process of adding
sufficient concentration of n-type impurities to counterbalance the p-type impu-
rities and thereby produce an effectively intrinsic material. For the case at hand,
in principle, alkali metals such as lithium, sodium, and potassium tend to form in-
terstitial donors in silicon and germanium. However, in practice only lithium can

4OForexample, this is the case for the p-i -n diode, important to infrared detection as well as particle
detection . For further discussion, see S. M. Sze, Physics of Semiconductor Devices, 2d ed., Wiley,
New York, 1981.
12.9 Impur ity Semiconductors and the p-n Junction 653

FIGURE 12.30 Change in E F with temperature for n- and p-type semiconductors (for
miz ;:: m;).Both curves are asymptotic to E F (intrinsic) .

be introduced in these materials in sufficient concentration to produce practical

compensation.v'

The p-n Iunctiorr'?

This section deals with a description of the p-n junction that plays a fundamental
role in many solid-state devices and involves a number of properties of impurity
semiconductors developed above. The p-n junction is composed of a p-type semi-
conductor (uniformly doped with acceptors) joined across a plane interface with
an n-type semiconductor (uniformly doped with donors) . If no voltage is placed
across the junction it is referred to as unbiased. A junction with voltage placed
across it is called a biased junction (Fig. 12.31). In what follows , stemming from
a basic property of the Fermi energy, we first describe the phenomenon of band
bending. The carrier densities, nand p , are then obtained in terms of a potential
function for the unbiased case, which leads to an expression for the "built-in po-
tential," Vo , across the junction. The region of interaction of electrons and holes in
the domain of the junction is called the depletion zone . Working in the depletion-
zone approximation, expressions for the potential function V (x) and nand p den-
sities in the depletion zones are obtained as well as a relation between Vo and
the width of the depletion zone . The discussion continues with an estimate of the
I -V characteristic for the biased p-n junction in the low-bias approximation. The
resulting I-V relation is found to give good agreement with measurement in the
small-bias potential domain.
41G. F. Knoll, Radiation Detection and Measurement , 2d ed., Wiley, New York, 1989.
42For further discussion, see S. Wang, Fundamentals of Semiconductor Theory and Device Physics,
Prentice-Hall, Englewood Cliffs, N.J., 1989.
654 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

TABLE 12.6 Semiconductor relations

General Results (Nondegenerate Crystal)

Conduction Band Valence Band

geE) = - 1 (2m*
2n 2
_e )3/2 (E - E c)I /2
li2 2n
I (2m* )3/2
gp(E) = - 2
Ii
-f
(E v - E)I /2

«r, = Nce-(Ec-EF) /ksT peT) = Nve-(EF-Ev)/ksT

N =~ hk
(2m BT)3 /2
v 4 nli2

np = -I (k...!!..-
T)3 (m*m*)3 /2 e-Eg /ksT
2 nli2 e h

Intrinsic Semiconductor

EF = ~(Ev
2
+ Ed + ~kBTln
4
(m~)
me

EF = ~E9
2
+ 4~kBTln (m~)
me
Extrinsic, n type
1]ow

Extrinsic, p type
Tlow

I v + Ea)
EF = -(E
2
+ -kBTln
1
2
(N-N v )
a

3Presumes large doping density.

Preliminary Relations
Here is a brief review of some physical relations relevant to the present topic. The
diffusion equation relates to a fluid of particles in which a gradient of number
density, ncr), is present. It is given by

nu = -DVn (l2.97a)

which indicates that a flux of particles comes into play to restore the fluid to
uniformity. The fluid velocity is u(r) and D is the diffusion coefficient. Suppose
the particles carry a charge, e. Then current density is given by

J =enu (12.97b)
12.9 Impurity Semiconductors and the p-n Junction 655

~ p type n type 1
(a) (b)

FIGURE 12.31 (a) Unbiased p-n junction. (b) Forward-biased p-n junction in which
electrons and holes are driven to the contact region where they recombine and current
flows. (c) Reverse-biased p-n junction in which electrons and holes are driven away from
the contact region and current drops dramatically.

and (12.97a) becomes, for electron carriers (n) and hole carriers (p), respectively,

JDn = -DneVn = DnlelVn

(12.97c)
JDh = -DheVp = -DhlelVp

If an electric field, 8 , is present in the device, then current is due to the applied
field as well as diffusion current (l2.97c). There results

I n = enJ-Ln8 + eD nVn
(12.97d)
Jh = epJ-Lh8 - eDh VP

where mobility, J-L , is related to the diffusion coefficient through the Einstein re-
lations
J-LnkBT
Dn = - - - , (12.97e)
e
Following (l2.97c), e is written for lei, so that electronic charge is -e, and holes
have charge e. (With respect to this notation, mobilities in the preceding relations
are positive.) Lastly we recall Poisson's equation (SI)

V 2 V = -~ = -~(p + Nd - n - Na ) (l2.97f)
£ £

where p represents charge density ; Nd, Na represent ionized donor-atom density

and ionized acceptor-atom density, respectively; e is the dielectric constant of the
656 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

junction; and

8=-VV (12.97g)
We recall that (12.97f) stems from the preceding relation and the first of Maxwell 's
equations (V • 8 = p/ e).

Unbiased p-n Junction

When the n- and p-type materials are brought into contact, electrons diffuse to
the p side and holes diffuse to the n side. Recall that holes on the p side stem
from acceptors , and conduction electrons on the n side stem from donors . When
an electron from the conduction band of the n side diffuses across the junction,
it recombines with a hole in the valence band of the p side. This recombination
leaves a negatively ionized acceptor on the p side and a positively ionized donor
on the n side (Fig. 12.32). This charge buildup gives rise to an electric field that
retards the diffusion . When current ceases, the domain of interaction about the
junction is void of carriers and for this reason is called the depletion zone. In the
state of no current flow, with (12.97d) , and working in one dimension , we write
dn
I n = 0 = enILn8 + eD n - (12.98a)
dx
dp
Jh = 0 = epILh8 - eDh- (12.98b)
dx
Writing
dV 1 dp dlnp 1 dn dlnn
8=--,
dx p dx ----;;;- , n dx dx
(12.98c)

p side n side

-------E;

-------E:
FIGURE 12.32 An electron from the n side of the depletion zone annihilates a hole from
the p side, leaving an uncompensated , negatively charged acceptor on the p side and an
uncompensated . positively charged donor on the n side.
12.9 Impurity Semiconductors and the p-n Junction 657

E; --------.. .

Ea - - - - - - - -
'\ En
c

E:---------., EF

"\
Ed

En
v

FIGURE 12.33 Band bending in the p-n junction maintaining constant E F . At xo, E F is
midway between E v and Ee , which is identified as an intrinsic property. For equal doping
densities, xo lies at the midpoint of the depletion zone.

together with the Einstein relation (12.97e), equations (12.98) give

-d
dx
(ev
- - +lnp ) =0
kBT
(12.99a)

d (ev )
dx kBT -Inn = 0 (12.99b)

Prior to integrating these equations, we note the following . At low temperature,

the Fermi energy in the p-type material lies midway between the acceptor level
and the top of the valence band. In the n-type material it lies midway between the
donor level and the bottom of the conduction band (12.30). When the materials are
brought into contact , the relative positions of the conduction and valence bands
adjust to maintain a constant Fermi energyi'' (Fig. 12.33). It follows from the
continuity of the various levels in this diagram that there is value of x, which we
label xo , at which E F lies at the midpoint of the energy gap. This condition was
previously identified as an intrinsic property (12.78). It follows that at this point
n = p = n i , As we will find, it is consistent to set V (xo) = O. Integrating (12.99)
from x to xo, we obtain
eV eV n
- - - In n
kBT
=0 - In n-
t «
--=In-
kBT ni
(12.100a)
eV eV p
- - +lnp = O+lnni, - - =-In-
kBT kBT n;

or, equivalently,

eV(x)
n(x) = ni exp - - ,
kBT ( ev(x))
p(x) = n, exp - kBT (12.100b)

43At T > 0 K, the Fermi energy is more appropriatel y called the chemical potential . See footnote 36,
page 64S.
658 Chapter 12 Application to Atomic , Molecular, Solid-State, and Nuclear Physics

Note that, as previously stated, at x = xo, with V = 0, n = p = ni , At

x = +00, n(x) -+ Nd and at x -+ -00, p(x) -+ Na, which, with (l2.100a) give

eV(oo)/kBT = In(Nd/ni)
(l2.l01a)
eV(-oo)/kBT = -In(Na/ni)

The built-in potential, Vo, is given by

kBT NaNd
Vo = V(oo) - V(-oo) =- I n -2- (l2.101b)
e ni

(In the present notation , Nd , Na represent ionized donor and acceptor densities,
respectively.)

Depletion-Zone Approximation
The relations (l2.100b) are incomplete without knowledge of Vex) . A model that
affords evaluation of the potential function is called the depletion-zone approx-
imation. Profiles of various components of the p-n junction in this model are
shown in Fig. 12.34. In this picture regions I and II denote the p and n domains of
the depletion zone, respectively. As charge densities in the present approximation
are constant in each p and n domain of the depletion zone, conservation of charge
requires that

(l2.102)

As p is constant in regions I and II (see Fig. 12.34), the solution of Poisson's

equation (l2.97f) in each of these respective domains has the general form

px 2
V(x)=--+Bx+K
2s
where Band K are arbitrary constants . Solutions of Poisson's equation over do-
mains I and II of the depletion zone which include the built-in potential (l2.101b)
with respect to points x p and X n are then given by

VI(x ) = -eN
2s
a
(x -
2 kBT
x ) - -
p e
Na
In -
n;
(12.103a)

Vn(x) = -eNd
(x -
2s
xn )
2 kBT
+- Nd
In-
e ni
(12.103b)

Potentials are defined with respect to positive charge flow so that, as evident from
Fig. 12.35, the preceding potentials retard positive charge flow from p to n re-
gions. The potential "seen" by electrons is obtained by reflecting the curve of
Fig. 12.35 through the x axis, from which it is evident that this potential equally
retards negative charge flow from n to p regions. The constants in (12.103) return
12.9 Impurity Semiconductors and the p-n Junction 659

p side n side

IJ
Na

IN,
xp Xo xn

(a)

p
n
I II II
I ',', ',',
xp xp Xn xn
(b)

xp +
I +eNd

Xo xn

(c)

FIGURE 12.34 Densities in the depletion-zone approximation: (a) Doping densities.

(b) Carrier densities. (c) Charge densities. The bulk material domains I and II are shown as
well. Charge densities in these latter domains are zero as charges due to ionized impurity
atoms are compensated by carrier charges.

the correct built-in potential

Vn(x n ) - VI(Xp ) = Vo

where Vo is given by (12.101b).

Electron and Hole Densities and Depletion-Zone Width

Substituting the potentials (12.103) into the nand p expressions (12.100b) gives

(X- x )2] x in region I (x > x p ) (12.104a)

p(x) = Naexp - ; ,
[ 2A p
660 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

V(x)

(a)

e(x)

(b)

FIGURE 12.35 (a) Built-in potential in the depletion -zone approximation illustrating a
barrier to positive charge flow from p to n domains or negative charge flow from n to p
domains . (b) Built-in electric field corresponding to the potential in (a). This electric field
drives holes to the left and electrons to the right, away from the contact region.

x in region II (x < x n ) (12.104b)

where the A lengths are given by

2 ekBT
Ap == -2-- (12.104c)
e N«
12.9 Impurity Semiconductors and the p-n Junction 661

2 ekBT
An == -2-- (12.104d)
e Nd
The relations (12.104a,b) indicate an exponential decay of charge carrier densities
into the depletion zone. This decay increases with increase of the A coefficients.
This smooth decay is an improved estimate of carrier densities in the depletion
zone compared with the sharp cutoff of starting values of these parameters in
Fig. 12.34.
We wish to apply these latter equations to obtain an expression for the
depletion-zone width

W ==X n <s»

in terms of the built-in potential, Vo . This relation may be obtained from the con-
straint that VI and VII must match xo, which gives
eNa 2 kBT n; eNd 2 kBT Nd
-(xo - Xp) - - I n - = --(xo - Xn) +- In - (12.105a)
2e e n; 2e e n,
Referring to the charge-conservation equation (12.102) and the built-in potential
relation (12.1Olb) we see that the latter equation implies the following two equa-
tions,

(12.l05b)

from which we find

Vo = .: ( NaNd ) W 2 (12.l05c)
2e N a + Nd

This is the desired relation between the width of the depletion zone and the built-
in potential. With (12.101b), the latter relation gives W in terms of known doping
densities and intrinsic electron density.
The preceding results, (12.104) and (12.105c), represent equilibrium properties
of an unbiased p-n junction at the temperature T .
One may associate a capacitance with a p-n junction. Thus, if A represents the
cross-sectional area of the junction, the capacitance of an unbiased p-n junction
is given by
eA
c=-
W
(12.l05d)

where the width, W, of the depletion zone is given by the preceding relation,
(12.105c).

Biased p-n Junction

A biased p-n junction is one with an external voltage, Vb, applied across the
junction. Recall that the built-in voltage, Vo, opposes further current flow in the
662 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

unbiased junction. The resulting voltage across the junction is

(12.106)

That is, the total potential drop occurs across the depletion zone. Again, potentials
are defined with respect to positive charge flow. (Recall Fig. 12.35.) For forward
bias, Vb > 0, and Vb diminishes Vo and current flow is enhanced. For reversed
bias, Vb < 0, and the built-in potential is enhanced, thereby inhibiting current
flow. In forward bias, electrons and holes are driven together and they recombine.
The external potential resupplies electrons to the n domain and current flows. In
reverse bias, electrons and holes are driven away from the junction domain in
opposite directions and current falls dramatically. A typical I-V characteristic for
a p-n junction is shown in Fig. 12.36. As evident from this figure, a p-n junction
serves as a rectifier.

Low-Bias Approximation
We wish to obtain an estimate for the I-V characteristic of a p-n junction in the
limit of small applied voltage. The estimate is based in large part in evaluating
n(x) and p(x) in respective minority domains outside the depletion zone. The
reason for this procedure is that, to this same order, carrier densities in their re-
spective majority domains are approximately constant and equal to related donor
densities (Fig. 12.34b). In the first step of this estimate, one assume that (12.98)
maintains, leading again to (12.99). Integrating this expression between X n and
x p gives

e[V(xn) - Vex )]}

p(x n) = p(xp) exp { - kBT p (12.107a)

e[V(xn) - Vex )]}

n(xp) = n(xn) exp { - kBT p (12.107b)

With (12.106) and recalling the expression (12.101b) for Vo, the preceding equa-
tions are rewritten
nj2 eVb
p(x n) = p(xp)--exp- (12. 108a)
NaNd kBT
nj2 eVb
n(x p) = n(xn ) - - exp-- (12.108b)
NaNd kBT
With p(xp) ::::: Na, n(x n) ::::: Nd, these latter equations become

nj2 eVb
p(x n ) ::::: - exp - - (12.108c)
Nd kBT
nj2 eVb
n(x p) ::::: - exp - - (12.108d)
Na kBT
12.9 ImpuritySemiconductors and the p-n Junction 663

(a)

--.l.-------.l.-----II--------L--------L---~ x

FIGURE 12.36 (a) I-V characteristic of a p-n junction. Typical currents are of the or-
der of rnA at I V. (b) Minority charge-carrier densities in the depletion-zone approxima-
tion from which it is seen that electrons and holes in their respective minority domains are
driven to the depletion zone, thereby increas ing the occurrence of recombination and cur-
rent flow. For reverse bias, (V --+ - V), electrons and holes are repelled from the depletion
zone, sharply diminishing the probability of recombination, with corresponding decrease
of current flow.

Note that, as indicated above, p(x n ) and n(x p ) are minority carrier densities. We
introduced the two displacements:

These displacements lie far beyond the depletion zone (Fig. 12.34b). It is assumed
that gradients of minority carrier densities as well as drift terms in minority carrier
currents are negligibly small at respective displacements in and i p . As all doping
atoms are ionized , in region I, p = N« , and in region iI, n = Nd. With reference
to (12.80b), we then write

(12.108e)
664 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

The latter three relations are boundary conditions for n (x) and p(x) in their re-
spective minority domains.
Before proceeding further, we recall the continuity equation (7.97). In the
present situation this equation is written for both nand p. However, for transport
in a semiconductor, one must take into account the possibility of recombination of
holes and electrons . Thus, for example , the continuity equation for holes is given
by (in one dimension)
op 1 oJh 1 _
- +- - = --[p - p(xn ) ] (12.109a)
ot e ox Th

This equation indicates that p in its minority bulk domain, IT (Fig. 12.34b), relaxes
in the characteristic recombination time, Th, to its equilibrium value, p(xn ) . [As
we are employing (l2.80b), in the following one assumes small recombination
rates, l/T .] In the steady state we set op/ot = O. Employing the drift current in
(12.97c)
dp
Jh = -eDh- (12.109b)
dx
together with the steady-state version of (l2.109a) and (l2.l08e), we obtain
d 2P 2 n,
2
2
- -Kh P = --Kh (12.109c)
dx 2 Nd

2 1
Kh - - - (l2.109d)
DhTh
The solution to (12.109c), subject to the boundary conditions (12.108c,e), is given
by

p(x) = -'-
Nd
2
n- [ 1 + S(x) (ev)]
exp- --:- 1
kBT
(12.11Oa)

(x in region IT)
sinh Kh(X n - x)
S(x)= 'h - ) (12.11Ob)
Sill Kh(X n - Xn

=
where we have set Vb V . At x = xn, the preceding solution returns (12.108e),
whereas at x = x n , it returns (12.108c). In this manner we find that for forward
bias, eV / kBT > 0, p(x) grows from its value at xn to its value at x n .
With (12.109b) we obtain the minority current density on the n side of the
junction (region IT).

eni2Dh /
Jh =- - S (x) ( exp--1
eV )
(12.11Oc)
Nd kBT
where a prime denotes differentiation.
The equation for n(x) in region j is the same as (l2.109c) with the parame-
ter,K;, defined parallel to (12.109d). The solution for n(x) on the p side of the
12.9 Impurity Semiconductors and the p-n Junction 665

junction (region I) is given by

n- [ 1
n(x) = _1_
Na
2
+ Sex) (ev)]
exp--l
kBT
(12.11Od)

(x in region I)
- sinh Kn(X
S(x)=--------"--
p) -x (12.1I0e)
sinh Kn(Xp - x p)

which returns the correct boundary conditions, (12.108d) and the second of
(12.108e) . We note that for forward bias, e V / k B T > 0, n (x) grows from its
value at Xh to its value at Xh . In this manner we find that a forward bias forces
both nand p minority carriers to the region of the depletion zone (Fig. 12.36a), in
opposition to the built-in field (Fig. 12.35). For reverse bias (V ~ - V), electrons
and holes are driven away from the depletion zone, in accord with the built-in
potential, diminishing the occurrence of recombination. That is, in reverse bias,
p (x) diminishes from its value at xn to its value at X n , and n (x) diminishes from
its value at Xh to its value at Xh .
The relations (12.11Od,e)give the minority current density on the p side of the
junction.

entDn
in = - - S-, (x) ( e xeV
p--l ) (12.11Of)
Na kBT

If there is no generation or recombination of carriers in the depletion zone, K ex

'["-I ~ 0, and in and ih are constant in these regions. To demonstrate this prop-
erty consider the K dependence of, say, in.

cosh xfx - x )
in ex - - - - - - - 'p- -
[sinhK(x p - Xp)]/K

In the limit, K ~ 0, this form goes to the constant, 1/(x p - xp ) . It follows that
(12.1I1a)

These relations permit preceding expressions for minority currents to be em-

ployed in calculation of total current, and we write

(12.1 lIb)

so that with (12.110d) and (12.110f) there results

(12.11Ic)

(12.1 lId)
666 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

With K I =J A, where A is the cross-sectional area of the junction and I represents

line current , we obtain

I = Is (exp eV -
kBT
1) (12.112a)

Is = AJs (12.112b)

This finding represents the first-order solution in the low-bias approximation

above the zero-current starting value (12.98). With V replaced by Vb, the current
(12.112) is illustrated by the small-bias portion of the I -V characteristic shown
in Fig. 12.36 (where V is again written Vb)' Replacing V with -V in (12.112a)
indicates that Is is the saturation current in reverse bias.
We close this discussion with the generalization of our previous experience for
the capacitance for an unbiased junction, (12.10Sb), to the case of the capacitance
of a biased junction. With (12.106), we recall the potential across the depletion
zone of a p-n junction changes in the presence of an applied potential: Vo -+
Vo - V . The related generalized width of the depletion zone is given by

(12.1 13a)

When inserted in (12.1OSb), the preceding relation gives the desired generalized
expression for the capacitance of a biased p-n junction,

C- A ee N N
d a
]1 /2
[ 2(Vo - V) (Nd + N a ) (12.113b)

An important application of capacitance of a p-n junction is as follows. Con-

sider a p+ -n junction (which denotes a p-n junction with a highly doped p side,
i.e., n a » Nd) . In this limit the capacitance of the device becomes

ee ]1 /2 (12.113c)
C=A [ Nd
2(Vo - V)

For large reverse bias, V -+ - V, IV I » Vo, and Vo may be neglected compared

to V. For this case, the preceding relation affords a measurement of Nd in terms
of the cross-sectional area of the device.

PROBLEMS

12.56 Consider an n-type semiconductor.

(a) Give two relations in terms of density of ionized donors, Nd - Ndo' and con-
duction electrons n which are obeyed in the 1iow domain.
(b) Give two relations involving these variables which are obeyed in the 1Jnt do-
main.
Problems 667

12.57 A p-type semiconductor has a hole density p = 1.7 x 1013 cm- 3 at T = 388 K.
If E 9 = 1.8 eV and m;m = 0.25me2 at this temperature, what is the electron
h
carrier concentration n ?
12.58 The effective density of states, Nc(T), of a semiconductor at T = 370 K is 0.42 x
1020 cm- 3 . What is the value of m; / me for this material at this temperature?
12.59 An intrinsic semiconductor has an energy gap of 1.03 eV with effective masses
m; = 1.8mh.
(a) What are the values of EF at 300 K and 500 K?
(b) What are the charge carrier densities nand p, respectively, at these tempera-
tures?
12.60 In the text we discussed Sb and In as impurity atoms in an Si host crystal. With
reference to the periodic chart and an appropriate handbook for required additional
data, choose another host element from the same group that Si lies in and two other
impurity elements which would serve as donors and acceptors, respectively, for
your host element. The device is to operate in the temperature interval 300 K ~
T ~ 500 K.
12.61 Where does the Fermi energy of a conductor lie at 0 K with respect to valence
and conduction bands? Your answer should include reference to the Fermi-Dirac
distribution.
12.62 A one-dimensional model of a certain intrinsic semiconductor with lattice constant
a has the following E versus k relations for the conduction and valence bands,
respectively.

2 ka
EC<k) = 2Eo - bEo cos 2
ka
Ev(k) = bEo cos2 2
The dimensionless parameter b is less than I. At 0 K the conduction band is empty.
(a) Sketch these curves on the same E versus k graph in the reduced zone.
(b) Obtain expressions for the effective electron and hole masses, m;and mh,re-
spectively. (See appropriate expressions in Problem 8.57.)
(c) If Eo = 2.5 eV, b = 1,and a = 1.5 A, obtain the value (in eV) of the Fermi
energy for this semiconductor.
(d) At what frequency (Hz) would incident photons cause this semiconductor to
conduct at 0 K?
12.63 A semiconductor is doped with donor impurity atoms of density Nd and acceptor
atoms of density Na . What is the equation of charge neutrality for this extrinsic
semiconductor?
12.64 Electronegativity is a measure of the force required to remove the ionizing electron
of negatively charged ions. Consider, for example, that an H atom in the ground
state is negatively ionized, resulting in a Is2 configuration.
(a) Repeating the calculation in the text relevant to helium, calculate the new
ground-state energy of the H- ion.
668 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

(b) Is the H- a bound system?

(c) Repeat part (a) for a one-electron atom with nuclear charge number Z . How
does the energy you obtain depend on Z ?

Answer (partial)
(a) Referring to Problem 12.28 and Table 10.3, we find

2e 2 5e2
E=2lR+l::..E=--+- <0
2a Sa
Thus, to within the stated approximation, we find the ion to be a bound system.

12.65 Employing rules of electrostatics, show that the Coulomb interaction energy A
(12.45) may be written as a single integral over electric field.

Answer
This integral is in the form

A = ff drdr,p(r)p(r')
Ir-r'l

where p = ellpl2 is charge density. Introducing the field (cgs)

V·8 = 4JTp
permits the preceding integral to be rewritten

2
A = 2f 18(r)1 dr > 0
8JT

12.66 Write down the condition of charge neutrality in a p-n junction in

(a) The p domain beyond the depletion zone , i.e., x < x p .
(b) The n-domain beyond the depletion zone, i.e., x > Xn . (See Fig . 12.34.) In the
depletion-zone approximation,
(c) What is the reason for evaluating only minority carrier densities in calculation
of current across a p-n junction?
(d) What entities are depleted in the depletion zone of a p-n junction?

Answers (partial)
(a) p(x < xp) = Na + n(x < xp)
(b) n(x > x n) = Nd + p(x > xn)
12.67 (a) Show that the potentials (12.103), relevant to the depletion-zone approxima-
tion, satisfy Poisson's equation (12.97f) (in one dimension) with correct charge
densities.
(b) Again employing these potentials, obtain expressions for the electric field, 8,
across the p-n junction.

12.68 Assuming doping densities, N a :::::: Nd :::::: 10 17 cm- 3 and e = 1.02 at T = 300 K,
(a) Find the values for the characteristic lengths, Ap and An (12.104) .
 12 .10 Elements of Nuclear Physics. The Deuteron and Isospin 669

(b) Make a rough sketch of n(x) in region II and p(x) in region I correspond-
ing to a p-n junction with the preceding properties in the unbiased state. (See
Fig. 12.34.)

12.69 (a) Obtain the relations (12.105b) from (12.105a).

(b) From these relations, obtain (12.105c) for Vo in terms of W. A p-n junction
operates at T = 300 K with densities n, = 10 18 cm- 3 , Na = 1020 cm- 3 ,
Nd = 1019 cm- 3 . In addition , e = 1.2.
(c) What is the width of the depletion zone, W? Answer in JLm.
(d) A reverse bias is applied across the junction so that Vo - V = 5 V. The cross-
sectional area of the junction is A = 15 JLm2 . What is the capacitance of the
device (in F)?
(e) If Nd » N« , what is the value of Nd (in cm- 3)?

12.10 • ELEMENTS OF NUCLEAR PHYSICS. THE DEUTERON AND ISOSPIN

The Pauli principle was applied earlier in this chapter in constructing properly
symmetrized atomic and molecular states. In the preceding section we saw how
this principle, through use of the Fermi-Dirac distribution, comes into play in
describing electron and hole properties in semiconductors. In this concluding sec-
tion of the present chapter we will find how the Pauli principle, through the notion
of isotopic spin, comes into play in nuclear physics. This example addresses con-
struction of the ground state of the deuteron.
The first section of the present discussion addresses the angular and spin com-
ponents of the ground state of the deuteron. Recall that an estimate of the radial
component of the ground state was obtained in Problem 10.31. In the present
discourse, the angular component is obtained by employing conservation of par-
ity and magnetic moment data. In this construction, noncentral, spin-dependent
forces are encountered.
In the second component of the discussion , the notion of isotopic spin is de-
veloped, which, as noted above, permits quantum statistics to be applied, and the
appropriately symmetrized ground state of the deuteron is obtained.

Magnetic Moment of the Deuteron

The magnetic moment of the deuteron is related to its angular momentum. In
general , the angular momentum of a nucleus may be written as a sum of orbital
and spin angular momenta of nuclear constituents, and we may write 44

J=L+S (12.114)

For the deuteron, L is attributed to rotational motion of the neutron and proton
about each other, while S is the resultant of their spins.

44Nuclear spin is often denoted by the symbol I.

670 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

As noted in Section 12.7, the spin of the deuteron is unity (i.e., j = 1). The
!-
spin of the neutron and proton are both so that allowed coupled spin values are
s = 0 , 1. The values of orbital angular momenta l which combine with these spin
values to give j = 1 are listed below.

s=1 [=0: j=l 3S1

[=1 : J=0,1,2 3PI

[=2: j=l, 2, 3 3DI (12.115)

s=O [=0: j=O
[=1 : j=l IPI

Thus, the experimentally observed nuclear spin of unity for the deuteron in its
ground state implies that the angular component of the wavefunction is some su-
perposition of 3SI , 3PI, 3DI, and I PI states. We will now find how parity conserva-
tion properties of the Hamiltonian of the deuteron further specify the composition
of the ground-state wavefunction.
The Hamiltonian of the deuteron may be written H (r p , r n ; SpSn), where the
subscripts nand p denote neutron and proton, respectively. The parity operator
JP (see Section 6.4) which reflects r p and r n through the origin leaves H un-
changed.P

JPH(r p , r n ) = H(-r p , -rn ) = H(r p , r n )

(12.116)
[JP, H] = 0

It follows that the parity is a good quantum number and may be listed along with
other quantum numbers in specifying the state of a deuteron. That is, states are of
definite parity.
The combination of states 3S1 and 3DI corresponds to even parity, while the
combination 3PI and IPI corresponds to odd parity (recall Problem 9.46) . The
specific combinations which comprise the angular component of the ground-state
wavefunction is obtained by finding the combination which gives the closest value
to that of the observed magnetic moment of the deuteron,

J1-d = 0.857 J1-N

The nuclear magneton J1-N has the value

lie
J1-N = - - = 0.505 x 10- 23 erg gauss- I
2Mp c

The proton rest mass is M p-

45This statement is valid for the strong nuclear forces.

12.10 Elements of Nuclear Physics. The Deuteron and Isospin 671

There are three distinct contributions to the deuteron's magnetic moment:

ltd = Itp + Itn + ItL (12.117)

As with the electron, the proton and neutron magnetic moments are proportional
to their respective spins. Experimentally, one obtains the values (noted previously
in Section 11.9)

Itp = 9p (JL;) S(p) 9p = 2(2.79)

Itn = 9n (JL;) Sen) 9n = 2(-1.91)

The contribution to ltd

due to orbital motion of the neutron and proton about each
other is (see Problem 12.72)

ItL = ~ (JL;) L
It follows that (12.117) may be rewritten

ltd = (JL;) [9 pS(P) + 9nS(n) +0.50L] (12.118)

As stated in the previous chapter, it is conventional to measure the expectation

of the z component of ltd
in the state m j = j and to call this the magnetic
moment of the deuteron JLd.
46
We will calculate the contributions to this value
from each of the four states listed in (12.115) and see which contribution of parity-
conserving pairs of states ePI and IPI or 3S1 and 3DI) gives a value closest to the
experimentally observed value.
Introducing the spin operators

S= S(p) + Sen)
S- = S(p) - Sen)
permits (12.118) to be rewritten

ltd = (JL;) [( 9p ; 9n ) S + ( 9p ~ 9n ) S- + 0.50L]

We are interested in the expectation of the z component of ltd

in the three-triplet
and one-singlet coupled spin states listed in Table 11.3. Calculation readily shows
that (Sz-) is zero in any of these four states (see Problem 12.70). Thus, for pur-
poses of the said calculation, we are permitted to write

ltd = (JL;) [0.88S + 0.50L] (12.119)

= (JL;) [0.38S + 0.50J]

46See also H. Von Buttlar, Nuclear Physics , Academic Press, New York, 1968.
672 Chapter 12 Appl ication to Atomic, Molecular, 501 id-State, and Nuclear Physics

Construction of the Ground State

Following the description given in Problem (12.16) relevant to the Zeeman effect,
we again consider that rotation of J about the z axis (direction, say, of a measuring
B field) is slow compared with that of JL about J. With (12.119) this gives the form

• J) Jz )
(fJ., d2 ) -_ (fJ.,d j2

=
fJ.,N
{0.50+0.38 [j(j + I) +s(s +
2j(j
I)
+ I)
-l(l + I)]} m i
J

Using this expression, one obtains the following contributions to (fJ.,d) for the four
states in question (setting m j = I as indicated above):

(fJ.,deSI)) = fJ.,N [0 .50 + 0.38 (~: ~) ] = 0.88fJ.,N

(fJ.,dePI)) = fJ.,N [0.50 + 0.38 (~: ~)] = O.50fJ.,N

(12.120)
(fJ.,d( 3Pj)) = fJ.,N [ 0.50 + 0.38 (2+2-2)]
2x2 = 0.69fJ.,N

(fJ.,d( 3Dj)) = fJ.,N [ 0.50 + 0.38 (2+2-2

2 x 2 X3)] = 0.31fJ.,N

Since (fJ.,dePI)) and (fJ.,d(lpj)) are both less than (fJ.,d) , there is no combination
of the form

(fJ.,d) = (a3p)2(fJ.,depj)) + (al p)2(fJ.,dePI))

(a3p)2 + (al p)2 =I
which gives the observed value of (fJ.,d). On the other hand, fitting (fJ.,d) to the
combination of even parity states ,

(fJ.,d) = (as)2(fJ.,deSI)) + (aD)2(fJ.,deDI)) = 0.86fJ.,N

(as)2 + (aD)2 = I

does allow the solution

(as)2 = 0.96, (12.121)

One may conclude that the ground state is a mixture of ..10.96 3S1 and .jO.04 3DI
states. That is,

(12.122)
12.10 Elements of Nuclear Physics. The Deuteron and Isospin 673

The coordinate representation of this state is given by

(12.122a)

where the yt (0, ¢) spherical harmonics are listed in Table 9.!.

As ao « as, for purposes of calculating the magnetic moment of the deuteron ,
it suffices to take ItjlG) to be entirely an S state. However, in evaluating the
electric quadrupole moment , the D contribution to the ground state is necessary
to obtain a finite result. Thus it is found that the superposition state (12.122)
gives agreement with the experimentally observed quadrupole moment of the
deuteron , Q = 0.0027 X 10-24 cm2.47 This moment is a measure of the deviation
from spherical symmetry of the charge density of the deuteron . Furthermore, the
sign (+) of the quadrupole moment indicates that the charge distribution of the
deuteron is prolate (resembling an egg) rather than oblate (resembling the earth).

Noncentral Forces
We note the following important fact about the neutron-proton interaction. Since
the ground state of the deuteron is a superposition of states of different 1, it is not
an eigenstate of L2. If the interaction between these two particles were a central
force,48 then eigenstates of H would be common with eigenstates of L2, and i;
and would appear as in the form of (10.108). We conclude that the intemucleon
interaction is not a central-force field but rather is spin-dependent. That is, the po-
tential of interaction between the particles is dependent on the relative orientation
of the spins Un and Up of the neutron and proton, respectively, and the radius vec-
tor r separating them. An example of such a noncentral spin-dependent potential
of interaction is given by the form 49

H~ S -_ f S (r ) [(un' r)(u
r2
p • r)
-
I A A]
-Un ' Up
3
(12.123)

u
where denotes the relevant Pauli spin matrices and Is'r) is a scalar function
of r . Note that Hs preserves parity.
The fact that the deuteron has only one bound state (Problem s 8.1 and 10.31)
with parallel neutron-proton spin orientations (triplet state) is also evidence for the
spin-dependent nature of the nuclear force, for if this were not the case, a bound
state would also exist for the antiparallel case (singlet state). See Fig. 12.37.

Isotopic Spin
In nuclear physics, the neutron and proton are viewed as two distinct states of
a single particle called a nucleon. This point of view (due to Heisenberg, 1932)
47This value takes electronic charge equal to I.
48Central potential was previously discussed in Section 10.5.
49H. Feshbach and J. Schwinger, Phys. Rev. 84, 194 (1951) . An assortment of forms for two-body
nuclear interactions is presented in L. Eisenbud and E. P. Wigner, Nuclear Structure, Princeton Uni-
versity Press, Princeton, N.J., 1958.
674 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

®® (a) (b)

FIGURE 12.37 Relative possible orientations of the neutron and proton spins in the
deuteron. The fact that the triplet spin state depicted by (a) is the ground state, whereas the
singlet state depicted by (b) is not observed in nature, is evidence of the spin dependence
of nuclear forces.

stems from the assumption that the nuclear force is far larger than the Coulomb
force within the nucleus, or, equivalently, that the intemucleon force is charge-
independent.
These features are incorporated in a property called isotopic spin , or, equiva-
lently, isospin. The operator corresponding to this property is written / and has
three orthogonal components which obey the relation

(12.124)

These operators exist in isotropic spin space . Eigenvalues of the third component
/3, relevant to a particle of charge Q (measured in units of electronic charge), are
given by

1
h=Q--2 (12.125)

It follows that h +!
= for a proton and h = for a neutron.-!
Let us ascertain the isospin states for a two-nucleon system . For this purpose
we introduce the single-particle isospin states (see Fig. 12.38)

(12.126)

_
--« ~--- /3 = + j, Ip>

[l= j,Q=lj .

' - - - - - - - - - 13 =-j, In>

1
FIGURE 12.38 Nucleon with isospin and charge number I has two /3 components
corresponding to the eigenstates Ip} and In}.
12.10 Elements of Nuclear Physics. The Deuteron and Isospin 675

TABLE 12.7 Two-nucleon coupled isospin states

State / /3
Ipp)s = Iph jph

Ipn)s = ~[jph Inh + jphlnhl 0

jnn)s= InhInh -I

Ipn)A = ~[jphlnh -Iphln)il 0 0

so that Ip} and In} correspond, respectively, to !J = and /3 = 1 -1.

These states
are written in analogy with the ordinary spin states a, fJ defined in Section 11.6.
Thus , raising and lowering operators may be introduced, defined as

(12.127)

Applying these operators on the states Ip} and In} gives

(12.128)
41n} = Ip}, LIp} = In}
With these results at hand, paralleling the construction of the coupled states
shown in Table 11.3 gives the coupled isospin states listed in Table 12.7.
In the framework of isospin, as noted above, neutron and proton are viewed as
separate states of a nucleon . As the nucleon we are considering has spin the 1,
coupled quantum states of a pair of such nucleons must be antisymmetric with
respect to simultaneous interchange of particle coordinates, spins, and isospins .l"
Accordingly, the ground state of the deuteron is written

(12.129)

where the overbar on the left side indicates symmetrization. The component Il/rG}
represents the state (12.122) comprised of 3S and 3D states, with definite j = 1
and s = 1. The middle factor on the right side of (12.129) represents the triplet,
symmetric, two-particle spin functions corresponding to spin 1 (see Table 11.3).
Since these contributions are both symmetric states, the combined state is sym-
metric as well. It follows that 1/ !J} must be the antisymmetric state Ipn} A·
Suppose this were not the case and that, instead, the proper isospin state were
the symmetric / = 1 states. The Hamiltonian for our system is assumed to be
rotationally invariant in isotopic spin space. Consequently, all three / = 1 states
correspond to the same energy. This triplet includes the states Ipp}s or Inn}s that
5013 is sometimes called the "charge variable" and is labeled It; . For further discussion, see L. D.
Landau , E. M. Lifshitz, and L. P. Pitaevskii , Quantum Mechanics, 3d ed., Pergamon Press, New York,
1974.
676 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

correspond, respectively, to the diproton and dineutron which are not found in
nature .
Thus , the appropriately symmetrized ground state of the deuteron (apart from
the radial factor) is given by
- 3
I1fJG} = I1fJG} I x}lpn}A (12.130)

which is a common eigenstate of j2, 52, 5z , [2 , [3.

Summary
The preceding result (12.129) for the ground state of the deuteron was based on
the following concepts and properties: (1) Parity is conserved for the strong nu-
clear forces ; (2) the value I for the spin of the deuteron; (3) values of the magnetic
dipole and electric quadrupole moments of the deuteron. These properties gave
rise to the even-parity, unsymmetrized component I1fJG} , (12.121).
Notions of the nucleon , charge-independent nuclear forces, and isotopic spin
were then introduced. With these concepts and the Pauli principle at hand, the
isospin factor Ipn}A was obtained. When coupled with I1fJG}, this factor produced
the appropriately symmetrized ground state (12.129).
It should be noted that until the notion of isotopic spin is introduced, there
is no reason to incorporate the Pauli principle in the formalism, as neutron and
proton are distinct particles. However, once isospin is introduced, neutron and
proton become different states of the same particle , the nucleon. As this particle
is a fermion , coupled states of such particles must be antisymmetric.

PROBLEMS

12.70 Show that (Sz-) [defined following (12.118)] is zero for any of the four coupled
spin states listed in Table 11.3.
12.71 Is the Pauli principle (neglecting isotopic spin) relevant to a proton-neutron system?
Explain your answer.
12.72 Let the neutron and proton in a deuteron rotate about a common origin with com-
bined orbital angular momentum L. Assuming circular motion, show classically
that the magnetic moment due to this orbital motion is

ILL = (_e_)c
4Mp
L = ~ (JJ-N) L
2 h

where JJ-N is the nuclear magnetic moment. Note that the orbital neutron motion
contributes zero current.
12.73 Consider an ideal experiment in which a gas of deuterons is immersed in a steady
:B field of 103 gauss. Radiation of frequency v is transmitted through the sample.
At what frequency might you expect to see a diminishment in the transmission ?
Problems 677

12.74 (a) What is the value of up • un~ for each of the four spin states ~ listed in Ta-
ble 11.3?
(b) Let Icir) and JS(r) represent central potentials. Is the interaction represented
by the Hamiltonian

H = Je(r) + Js(r)up • Un
a central potential? Explain your answer.
(c) What is [H, [2] for this Hamiltonian?
12.75 At which orientation of spins relative to interparticle radius does the interaction Hs
given by (12.123) lose r dependence?

12.76 A two-nucleon system is in a state with zero orbital angular momentum .

(a) List the possible spin states for this system . (Each nucleon has spin t.)
(b) List the possible isotopic spin states for this system .
(c) Employing the preceding results, write down the appropriately symmetrized
spin-isospin states for this system .

12.77 Derive an expression for the current vector, J, for a particle of charge q and mass
m moving in a steady magnetic field with vector potential A. This derivation should
follow (7.103) et seq., and the Hamiltonian (10.85) appropriate to a charge q . Hint:
Your answer should agree with the form (7.107) with momentum replaced by p-
(qA /c) .

12.78 Let the wavefunction for the particle described in Problem 12.77 be written 1/I(r) =
11/1 1exp i eP and assume further that the significant spatial variation of 1/1 is contained
solely in the phase <1>. (Here we are outlining the Ginzburg-Landaus! formalism
in which 1/1, called the order parameter, vanishes above Tc and whose magnitude
below Tc measures superconducting order at r .) With these assumptions, show that
your expression for J, obtained in Problem 12.77, reduces to

12.79 As appreciable current in a superconductor can flow only near the surface, with
the expression for J obtained in Problem 12.78 and the assumption that 11/11 2 is
essentially constant in the superconductor, we find

o= f J . dl = f [~: A+ ~ 'YeP] . dl
where the path of integration lies deep within the superconductor. Employing
Stokes 's theorem to reduce the A integral and the single-valuedness of eP to reduce
the 'YeP integral in the preceding relation, show that the magnetic flux enclosed by
the ring, <1> = ff B • dS, is quantized according to the rule 1<1> 1 = n <1>0, where n is

51V. 1. Ginzburg and L. D. Landau , Zh. Exps. Tear. n « 20, 1064 (1950). See also , N. W. Ashcroft and
N. D. Mennin, Solid State Physics, Holt, Rinehart and Winston, New York, 1976, Chapter 34.
678 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

an integer and <1>0 = h ejq . For Cooper pairs. one sets q = 2e. There results
<1>0 = he j 2e = 2.068 x 10- 7 gauss-em/

which is known as the .fluxoid or the .flux quantum .

12.80 Consider a hydrogen atom confined to a spherical box of radius a » ao, where ao is
the Bohr radius . Separating the motion into center-of-mass and relative coordinates,
obtain explicit expressions for wavefunctions and eigenenergies of the hydrogen
atom. In what domain of the box is your solution valid?
Answer (partial)
With the boundary condition, q:>CM(R = a) = 0, this solution is valid in the domain
R < (a - ao) .
12.81 (a) Show that all classical electrostatic equilibrium states are unstable (Earnshaw 's
theorem) .
(b) Does this theorem apply in quantum mechanics?
[Hint: For part (a), expand the potential in a Taylor series about the given equilib-
rium point. Then apply Laplace's equation and consider a propitious displacement
away from equilibrium.]
12.82 (a) What is the numerical value of the Fermi-Dirac function, f eE ), at E = EF at
T = 100 K?
(b) Given that the density of states of a given system is geE), write down an ex-
pression for the number of states, /:!"N, this system has in the energy interval,
E = EO to 10EO, where EO is the ground state of the system .
12.83 A schematic for the low-lying levels of barium ([Xe] 6s 2 ) is shown in Fig. 12.39.
(a) What transition out of the ground state of this atom is of minimum energy?
Explain your answer.
(b) If the atom is in one of the 3PI states. what related states may it fall to? If it is
in the ID state, which states may it fall to? What transitions are possible out of
the 3D states? Explain your answers .
(c) The decay time for the transition 3p ~ ID is significantly shorter than that of
the ID ~ 3 D transition. Incorporating this property and the preceding answers.
suggest a three-level lasing scheme for ideally isolated barium atoms .
12.84 The commutator relation for the operator52 corresponding to the number of photons,
N , in a cavity and that corresponding to the phase, 1/>. of these photons. is given by
[N.I/>] = - i .
(a) What is the related uncertainty relation 53 for these variables ?
(b) If /:!,.I/> = 10-4 rad, what is /:!"N?
Note: With the answer to part (a) we note that /:!"N may be made vanishingly small
at the expense of very large /:!,.I/> . The resulting field54 is called "squeezed light." The
field is "squeezed" in the sense that intensity fluctuations are vanishingly small.

52The number operator for photons is discussed in Problem 13.38.

53The uncertainty relation derived here is relevant to a system of many quanta . For further discussion,
see W. Heitler, Quantum Theory of Radiation, Oxford University Press, New York, 1944, page 68;
W. H. Louisell , Phys. Leu. 7,60 (1963) .
54For further discussion , see M. C. Teich and B. E. A. Saleh, Physics Today, June 1993, page 26.
Problems 679
___ Ip

_ _ _ 3p

---ID

===3D

- - - ISO

FIGURE 12.39 Schematic of the energy levels for the low-lying states of barium. (See
Problem 12.83.)

12.85 An important test for the presence of atomic hydrogen in the universe is detection
of the 21-cm radiative wavelength related to the electron-proton spin coupling in
hydrogen . This radiation is emitted in the decay of the excited antiparallel electron-
proton magnetic -moments state to the aligned electron-proton magnetic-moments
ground state. The Hamiltonian for this interaction is given by

H = -AILe' ILp
(Values of magnetic moments in the preceding expression are given in Section 11.9.)
(a) Given that the said transition emits 21-cm radiation ,obtain a value of the con-
stant A . State units of this constant. (The lifetime of this excited state is r ::::
1015 s :::: 108 yr. The frequent detection55 of this transition on earth indicates
an abundant presence of atomic hydrogen in the universe .)
(b) What is the ratio, K , of the energy of this 21-cm photon to that of a Lyman ex pho-
ton? (The resulting small value of K indicates that the energy of electron-nuclear
transitions is far less than that of atomic transitions. Corresponding levels are
labeled "hyperfine ,")

55First detected by E. M. Purcell and H. I. Ewen , Nature 168, 356 (1951).

680 Chapter 12 Application to Atomic, Molecular, Solid-State, and Nuclear Physics

12.86 A rare gas of lithium +2 ions (M = 6.9 amu ; Z = 3; electronic configuration of

[ls 2 2sj ]) is to be used to filter out radiation at a frequency corresponding to the
transition from the ground to the first excited state of an ion.
(a) What is the ground-state energy of this ion? (Answer in eV.)
(b) What is the frequency corresponding to the above-stated transition? (Answer in
Hz.) To what frequency domain does this radiation correspond?

12.87 Consider two cubical boxes with smooth flat walls and respective edge-lengths a, ii.
Show that respective emission frequencies, ws, Ws of an electron confined to each
of these boxes satisfy the relation

WsV 2j3 = WsV2j 3

where

s, = fiws = s2 EG
s2 =ni + n~ + n~
In these expressions, EG is the ground state of a one-dimensional box of edge-
length a, {n;) are positive integers , and V, V are respective volumes of box a and ii .

12.88 Consider three electrons with respective orbital and spin angular momenta, L j ,
Sj ; L2, S2; L3 , S3·
(a) Working in LS coupling, show that the following are sets of commuting opera-
A2 A A2 A2 A A 2 A A 2
tors:{J ,Jz,L,S ,(L;+Lj) ,(Sk+Sq»).
(b) Working in jj coupling show that the following are sets of commuting opera-
A2 A A2 A2 A2 A A 2 • • •
tors: {J , Jz , J j , J 2 , J 3 , (J; + Jj) ). (In the preceding, I 1= J, k 1= q and all
four integers run from I to 3.)
(c) Write down a ket vector for one set of commuting operators in each case. (See
also Problem 9.43 .)
12.89 Working in cylindrical coordinates (Appendix D):
(a) What is the Hamiltonian of an harmonic oscillator of mass M which oscillates
in the x, y plane with natural frequency w, and whose center of mass is free to
move in a domain of height L in the z direction?
(b) Show that eigenfunctions may be written

l{JAl .A2. A3 (p, z. (/J) = pm exp[ -~ j2] G (~ ) exp(±imt/l)ZK (z)

~ = Mwp2 jfi
where G(~) are Associated Laguerre polynomials (Table 10.3).
(c) What parameters do you find for the quantum numbers ()'1, ),,2, ),,3)?
(d) Obtain an explicit expression for the function Z K (z).
(e) Let Ez denote an eigenenergy of the two-dimensional oscillator confined to a
plane of constant z. What is the form of the corresponding eigenenergy, E , of
the oscillator described in the given problem, in terms of E z?
CHAPTER

13 Perturbation Theory

13.1 Time-Independent. Nondegenerate Perturbation Theory

13.2 Time-Independent, Degenerate Perturbation Theory
13.3 The Stark Effect
13.4 The Nearly Free Electron Model
13.5 Time-Dependent Perturbation Theory
13.6 Harmonic Perturbation
13.7 Application ofHarmonic Perturbation Theory
13.8 Selective Perturbations in Time
13.9 Atom -Radiation Interaction
13.10 Hartree-Fock Model

In this chapter perturbation techniques are described which serve to generate

approximate solutions to the Schrodinger equation. Such solutions appear in the
form ofan expansion away from known. unperturbed values. A special procedure
is developed for systems with degenerate eigenenergies. Application is made to
problems in atom ic physics and the problem ofan electron in a periodic potential,
encountered previously in Chapter 8. Harmonic perturbation theory is applied in
a rederivation of the Planck radiation formula and the theory of the laser. The
chapter continues with a description of the interaction of a radiation field with
an atom. Here we encounter the notion of oscillator strengths and the important
Thomas-Reiche-Kuhn sum rule. The concluding section addresses the Hartree-
Fock model for obtaining approximate atomic wavefunctions and eigenenergies.

13.1 • TIME-INDEPENDENT, NONDEGENERATE PERTURBATION THEORY

Approximate methods of solution to an assortment of problems in quantum

mechanics were described previously in Chapter 7 (WKB analysis), Chapter 8
(LCAO approximation), Chapter 10 (Thomas-Fermi model), and Chapter II
(interaction picture).
Our present concern lies in refinement of the approximation method described
in Chapter 12, used in calculating both the ground-state wavefunction of the he-
lium atom (Problem 12.28) and the fine-structure spectrum of hydrogen (12.13).

681
682 Chapter 13 Perturbation Theory

TABLE 13.1 Examples of perturbation Hamiltonians

Name Description Hamiltonian
L-S coupling Coupling between orbital and H = Ho + j(r)L. S
spin angular momentum in a if' = j(r)L . S
one-electron atom Ho = p2 / 2m - e 2 Z I r
Stark effect One-electron atom in a constant, H = Ho + e80z
uniform electric field 8 = ez 80 H' = e80z
Ho = (p 2 / 2m ) - e2ZIr
Zeeman effect One electron atom in a constant, H = Ho + (eI2mc)j • :B
uniform magnetic field :B if' = (eI2mc)j • :B
Ho = (p 2 / 2m ) - e2ZIr
Anharmonic Spring with nonlinear restoring H = Ho + K'x 4
oscillator force H' = K'x 4
Ho = (px 2/2m) + 1Kx 2
Nearly free Electron in a periodic lattice H = Ho + V(x)
electron
model
V(x) =L Vn exp[i(2nnxla)]
n
Ho = px 2/2m

In both these problems the Hamiltonian encountered was of the form

H = Ho + fI'
This breakup of a Hamiltonian into a part Ho, whose eigenfunctions are known,
and an addition term fI', which is in some sense small compared to Ho, is typical
of many practical problems encountered in quantum mechanics . The theory that
seeks approximate eigenstates of the total Hamiltonian H is called perturbation
theory. In the expression above, the Hamiltonian, Ho, is called the unperturbed
Hamiltonian, while H' is called the perturbation Hamiltonian. Some typical per-
turbation problems are listed in Table 13.1.
The perturbation analysis we will develop in this chapter divides into three cat-
egories : (1) time-independent, nondegenerate; (2) time-independent, degenerate;
(3) time-dependent. In the last category one investigates the time development of
a system in a given state due to a perturbation on the system which is turned on at
a given instant of time.

Smallness of the Perturbation

Perturbation theory begins with the assumption that the perturbation Hamilto-
nian, H', is in some sense small compared to the unperturbed Hamiltonian, Ho.
The criterion that establishes the smallness of H' compared to Ho will emerge in
the course of the analysis. Another underlying assumption in perturbation theory
13.1 Time-Independent, Nondegenerate Perturbation Theory 683

is that the eigenstates and eigenenergies of the total Hamiltonian H, do not dif-
fer appreciably from those of the unperturbed Hamiltonian, Ho. That is, suppose
that {fPn} and {En} are, respectively, the eigenstates and eigenenergies of the total
Hamiltonian H,
,.. A A /

HfPn = (Ho +H )fPn = EnfPn

while {fPn (O)} and {En (O) } are, respectively, the eigenstates and eigenenergies of
the unperturbed Hamiltonian

HOfPn (0) = En (0) fPn (0)

Then it is always possible to write

_ (0)
fPn - fPn + ufPn A

En = En (0) + !1E n

where, owing to the smallness of H', !1fPn is a small correction to fPn(O) and !1E n
is a small correction to En (0) .
To keep the smallness of H' in mind, we rewrite it as AH', where A is an
infinitesimal parameter and is introduced for "bookkeeping" purposes only. The
equation to which we seek a solution is of the form

(13.1)

The Perturbation Expansion

The eigenstates and eigenenergies of Ho are assumed known. Since fPn -+ fPn (0)
as A -+ 0, it is consistent to seek solution to (13.1) in the form of a series with
fPn (0) entering as the leading term. In a similar manner, En is expanded, with En (0)
entering as the leading term.

fPn = fPn (0) + AfPn (I) + A2fPn(2) + .

(13.2)
En = En (0) + AEn(I) + A2 En (2) + .
Substituting these expansions into (13.1) and arranging terms according to powers
inA gives

[HOfPn(O) - En(O)fPn(O)] +A[HWn(l) + H'fPn(O) - En(O)fPn(l) - En(l)fPn(O)]

+ A2 [HOfPn (2) + H'fPn (I) - En(O)fPn (2) - En(l)fPn (I) - En(2)fPn (0)]

+ ... =0 (13.3)

This equation is of the form

F(O) + AF(I) + A2 F(2) + A3 F(3) + ... = 0

&84 Chapter 13 Perturbation Theory

If this equation is to be true for arbitrarily small values of A., then

F(O) = F(I) = F(2) = ... = 0

In this manner (13.3) gives the coupled set of equations

(a) HOCfJn (0) = E6°)CfJn (0)

(b) (Ho - En(O»CfJn (I) = (En (I) - iI')CfJn (0)
(c) (Ho - E n(0»CfJn(2) = (En(l) - iI')CfJn(l) + E n(2)CfJn(0) (13.4)
(d) (Ho - E n(0»CfJn(3) = (En(l) - iI')CfJn(2) + E n(2)CfJn(l) + E n(3)CfJn(O)

In the lowest approximation, (13.4a) returns the information that {CfJn (O)} and
{En(O)} are, respectively, the eigenstates and eigenenergies of Ho. The second (as
well as all of the higher-order equations) has the following interesting property.
The left-hand side of this equation remains the same under the replacement

CfJn (I) -+ CfJn (I) + aCfJn (0)

where a is an arbitrary constant. Suppose that one solves (13.4b) for CfJn(I) and
En (I) . Then CfJn (1) + aCfJn (0) ; En (I) is also a solution . An extra constraint is needed
to remove this arbitrary quality of solution. This constraint may be taken as fol-
10ws. I We assume that all corrections to CfJn (0) in (13.2) are normal to CfJn (0).

(for s > 0 and all n)

In Hilbert space this relation indicates that I:i.CfJn is normal to CfJn (0) . This condition
will aid us in the construction of CfJn (s).
Returning to (13.4b) we note that Ho operates on CfJn (I) in this equation, which
suggests that the solution may be obtainable through expansion of CfJn (1) in a su-
perposition of the eigenstates of Ho.

(13.6)

If this expansion is substituted into (13.4b), there results

(Ho - En (0» I:>ndCfJn(O») = (En(I) - iI')ICfJn(O»)

i

1Another popular constraint is the constru ct 'Pn (s ) so that it is normalized . Both choices of constraint
yield the same corrections to the energy, (En (s»), while the wavefunctions that emerge differ by at
most a phase factor (see Section 4.1).
13.1 Time-Independent, Nondegenerate Perturbation Theory 685

Multiplying from the left with ((fi /0) I gives

(E /0) - En (O»)cnj + Hjn = En (I) 8jn (13.7)

where «; are the matrix elements of iii in the {(fin(O)} representation

First-Order Corrections

With j f= n, (13.7) gives the coefficients, {cnj}, which when substituted into
(13.6) gives the first-order correction to (fin.

H!In
Cni = En(O) _ Ei(O)
(13.8)
(I) _ "" H [n
. (0) (0)
(fin - L E (0) _ E '(O) (fi, + Cnn(fin
ifn n I

Here one assumes that all corrections {(fin (5)} lie in a Hilbert space that is spanned
by the unperturbed wavefunctions, {(fin(O)}.
The coefficient Cnn is obtained from (13.5), which yields

Cnn =0

With j = n, (13.7) gives the first-order corrections to the energy En .

En (1) = H'nn (13.9)

These are the diagonal elements of fi' . Substituting (13.8) and (13.9) into (13.2)
and setting A = 1 gives

_ (0) H[n . (0)

(fin - (fin + "L " E (0) _ Eo (0) (fir
ifn n I
(13.10)

I En = En (0) + H'nn I
The first of these equations tells us that in order for the expansion (13.2) to make
sense, the coefficients of expansion should be less than 1.

The matrix elements of iii should be small compared to the difference between
the corresponding unperturbed energy levels. In similar manner, the second equa-
686 Chapter 13 Perturbation Theory

tion in (13.10) reveals that

The diagonal elements of the perturbation Hamiltonian should be small compared

to the corresponding unperturbed energy level.

Second-Order Corrections

To find the second-order correction to IfJn and En, we must solve (13.4c) . Again
we note that Ho operates on IfJn(2) , and it is again advantageous to expand IfJn (2) in
the eigenstates of Ho.

(13.11)

Substitution into (13.4c) gives

L Ei(O)dndlfJ/O)} + H 'llfJn (I)}

i

= En(O) LdndlfJi(O)} + En(I)llfJn(l)} + E n(2)llfJn(0)}

i

Multiplying from the left with (1fJ/ 0) 1 gives

(E/O) - En(O»)dnj + (IfJ/O)IH'llfJn(l)} = E n(2)o jn + En(l) (1fJ/0) IlfJn(l)}

(13.12)

With j = n, this equation gives

Owing to the Henniticity of H', this equation may be rewritten

IH~i
2
E (2) _ '" 1
(13.13)
n - L..- E (0) _ £-(0)
i i=n n I

Note that in obtaining this result we have used the result that Cnn = O. Substituting
this expression for En (2) into (13.2) together with the expression for En (I) given
by (13.9) gives the following second-order expression for En:
Problems 687

E - E
n- n
(0)
+ H'nn + "" IH~iI2
£....J E (0) _ Eo (0) (13.14)
i,",n n I

To calculate the second-order corrections to the wavefunction ({In, we must

obtain the coefficients dni in (13.11). These are directly obtained from (13.12).
With n =1= j this equation gives

E ·(O»)d . = (m .(O)liI'''''
H'
( E n (0) _
J n] "r J
kn Imk(O)}
£....J E (0) _ E (0) "r
k,",n n k

I (
- Hnn ({Jj
(0) I ""
£....J E (0)
HIm
_ E (0)
I
({Jk
(O))
k,",n n k

In the second sum, only the k = j term survives the (({J /0) l({Jk (O)} inner product.
All terms in the first sum remain. There results

dnj = E
1
(0) _ E . (0)
("" HjkH~n)
£....J E (0) _ Ek(O) - (E (0)
o:»;E .(0»)2
-
n J k,",n n n J

Again, with (13.5), one finds that

dnn = 0

In this manner one obtains the following expression for ({In, good to terms of
second order in iI'.

(13.15)

PROBLEMS

13.1 Calculate the first-order correction to £3(0) for a particle in a one-dimensional box
with walls at x = 0 and x = a due to the following perturbations:
(a) H' = 10- 3 Esx ]«
(b) H ' = 10-3 £, (x /a)2
(c) H' = 10-3 £, sintx /«)
688 Chapter 13 Perturbation Theory

13.2 What is the eigenfunction lpn for the same configuration as in Problem 13.1, to terms
of second order, for the constant perturbation

H' = 1O-3E1?
13.3 Calculate the eigenenergies of the anharmonic oscillator whose Hamiltonian is listed
in Table 13.1, to first order in iI'.

Answer
In terms of raising and lowering operators, (at, a) , the perturbation Hamiltonian
appears as

The corrections to En (0) which we seek are given by

En (I) = H~n = (nliI'ln)

The only terms in the expansion of (a + at )4 which give nonzero contributions are
those which maintain the eigenvector In). All other terms vanish because of orthog-
onality with (nl . Of the 16 terms in the expansion of (a + at )4, only six survive this
orthogonality condition. The energy En (1) may be determined by a graphical anal-
ysis, according to which a diagram is associated with each integral that contributes
to (nliI'ln ). The eigenvector In) is represented by a dot drawn at the right of the
diagram. Another dot drawn at the left of the In) dot, but on the same horizontal ,
represents the eigenbra (nl. The creation operator, at, is represented by a diagonal
arrow from the right and inclined upward at Jr / 4, while the annihilation operator, a,
is an arrow from the right at -Jr / 4. Thus the diagram related to (nlaatln) is

The diagram that represents the fourth-order term, (nlat aat aln) , is

while the second-order term, (nla2In), is represented by

Any sequence of arrows that do not join the two horizontal dots represents a zero
contribution. Continuing in this manner, we find that, in all, there are 16 fourth-order
diagrams . Of these, only six gave nonzero contributions. These six diagrams are
Problems 689

Summing these contributions gives the desired result:

3K'
En (I) = - 4 [2n(n + I) + I)
4{3

13.4 Consider a particle of mass m in a potential well shown in Fig. 13.la. Suppose that
the bump at the bottom of the well can be considered a small perturbation .
(a) Calculate the corrected second eigenenergy and eigenfunction to first order in
the perturbation . [Recall (6.100).]
(b) What dimensionless ratio must be small compared to I in order for your approx-
imation to be valid?
(c) Is the parity of the unperturbed wavefunction maintained by this perturbation ?
Explain your answer.

V(x) V(x)

-L/2 -a/2 a/2 L/2 x -L/2 a/2 L/2 x

(a )

FIGURE 13.1 Potential configurations for Problems 13.4, 13.5, and 13.6.
690 Chapter 13 Perturbation Theory

13.5 (a) Consider the perturbation bump shown in Fig. 13.1b. What are the first-order
corrected eigenenergies of a particle of mass m confined to this well?
(b) Calculate the eigenenergies of this configuration, in the domain E » Vo , us-
ing the WKB formula (7.191) and compare your results with those obtained in
part (a).
13.6 Again consider the potential shown in Fig. 13.1a. What is the unperturbed ground
state for:
(a) Two identical bosons of mass m confined to the box.
(b) Two identical fermions of mass m confined to the box.
(c) What are the unperturbed ground-state energies Es and EA for these two cases?
(d) Use first-order perturbation theory to obtain the new ground-state energies for
these two cases.

Answers (partial)

(a) rps(Xj, X2) = ~ cos C2 )cos C2)

(b) rpA(Xj ,X2) = - : [cos C~j) sin CJr;2) - cos C2) sin CJrt)]
13.7 A particle of mass m is in an asymmetrical one-dimensional box, depicted in
Fig. 13.2.
(a) Use first-order perturbation theory to calculate the eigenenergies of the particle.
(b) What are the first-order corrected wavefunctions?
(c) If the particle is an electron, how do the frequencies emitted by the perturbed
systems compare with those of the unperturbed system?
(d) What smallness assumption is appropriate to Vo?

Answers
(a) En = En(O) + En(l) = n2Ej + Vo
2

Vex)

o L/2 L x

FIGURE 13.2 Potential configuration for Problem 13.7.

 Problems 691

n22
(b) i{ln = G.
V-;; [Sin (mrx) + Vo L
a
;2
E] Ifn n -I
".

Al
n
== !.. [Sin(n - I)Jr/2 _ sin(n + I)Jr/ 2 ]
it n-l n+l
(c) They are the same.
(d) VO« E]

13.8 Show that the matrix elements of the perturbation Hamiltonian iI' obey the equality

L IH~m 1
2
= (H t2 ) n n
m

Answer

LIH~mI2 = L(nliI/lm)(mliI/ln)
m m
A /AA / A /2 a
= (niH / H In) = (niH In) = (H )nn

Here we have recalled the relation

i = Llm)(ml
m

13.9 A hydrogen atom in its ground state is in a constant, uniform electric field that points
in the z direction. The electric field polarizes the atom. Show that there is no change
in the ground-state energy of the atom to terms of first order in the electric field. The
interaction energy is

H' = e8z = e8rcosO

13.10 The conditions are those of Problem 13.9. In calculating second-order corrections to
the ground state, one must know the values of the matrix elements (IOOIH'lnlm) ,
where Inlm) denotes a hydrogen eigenstate. Show that these matrix elements vanish
if I =1= I.
13.11 Consider again, as in Problem 13.9, that a hydrogen atom is in a constant, uniform
electric field 8 that points in the z direction. If a is the polarizability of hydrogen,
then the change in the ground-state energy is ia
8 2 • so we may write

Hfn == (IOOIH/lnIO)

(a) Use the result of Problem 13.8 to show that

I 82 (H' 2 ) 11
-a < -~-..:..;;...~
2 E2(0) - E](O)

(b) What maximum value for the polarizability of hydrogen does this inequality im-
ply? How does this value compare with the correct value of a?
692 Chapter 13 Perturbation Theory

Answers
(a)

so

Using the resultsof Problem 13.8, we obtain

00 00
L IH{n I2 = IH{l + L IH{nI2 = (H,2)1l
n=l n >l

so that
L IH{nI2 = (H,2 h 1 -IH{,12 < (H'2 h ,
n>'
hence

(b)

whichgives
16 3
Q"max = 3ao
The morecorrectvalueis Q" = 9ao3/2 .

13.2 • TIME-INDEPENDENT, DEGENERATE PERTURBATION THEORY

Again we consider a system whose Hamiltonian has the form

" " A /

H = Ho+H
where H' is a small perturbation about the unperturbed Hamiltonian, Ho. In the
present case , however, Ho has degenerate eigenstates. We have found previously
(see Section 8.5) that degeneracy in quantum mechanics stems from symmetries
inherent to the system at hand . Any distortion of such symmetry should therefore
tend to remove the related degeneracy.
Suppose, for example, that the ground state of Ho is q-fold-degenerate. If the
symmetry producing this degeneracy is destroyed by the perturbation H', the
13.2 Time-Independent, Degenerate Perturbation Theory 693
£(0) £(0)
q+l q+1

£lO) \

q-fold degenerate
state, before
perturbation After perturbation

FIGURE 13.3 Perturbation causes a removal of degeneracy.

ground state En (0) separates into q distinct levels (Fig. 13.3). The primary aim
of degenerate perturbation theory is to calculate these new energies . Suppose that
we proceed as in the nondegenerate case described in the previous section and
expand the first-order wavefunctions of H in the eigenstates of Ho (13.6).
m (I) -
-rn - "cn1-r1
~
. m. (O)

The formula that emerges for the coefficients {en;} is given by (13.8):

cn; = En(O) _ E ;(O)

If EI (0) is q-fold-degenerate, then

El (0) = E2(O) = ... = E q (0)

and Cn ; is infinite for n, i ::: q . This situation is remedied by constructing a new set
of basis functions from the set {l{Jn (O)} which diagonalize the submatrix, H(n (for
n, i ::: q) . With the off-diagonal elements of H(n vanishing, the corresponding
singular C;n coefficients also vanish and we may proceed as in the nondegenerate
case.

Diagonalization of the Submatrix

Thus the primary aim in degenerate perturbation theory is to diagonalize this sub-
matrix of H(n. As it turns out, the diagonal elements so constructed are the incre-
mental energies, which when added to E I (0) separate the q energies contained in
the ground state.
Let the q functions that diagonalize H(n (i, n ::: q) be labeled CPn:
q
m -~
'r n -
" an1-r1
.m. (Ol (13.16)
;= 1
694 Chapter 13 Perturbation Theory

These linear combinations of the degenerate eigenstates {lfli(O)} diagonalize H(n'

so

(n ,ps.q) (13.17)

These functions, when joined with the complementary set of non-degenerate

states, {Ifli(0) .i > q}, give the basis/

B= {qJI ' qJ2 " " ,qJq ' Iflq+1 (0), Iflq+2(0), . . . } (13.18)

The matrix of if' calculated in this basis appears as

,,,
o ,, H{ ,q+1
,
,,
,,
,,
o ,,,
,
_______________________________~i_~_J (13.19)

First-Order Energies

We will now show that the diagonal elements of the q x q submatrix of H' are
the first-order energy corrections En' to En (0) , n S. q . That is,

(n S. q) (13.20)

If these diagonal elements are mutually distinct, then the q-fold degeneracy of Ho
is removed by the perturbation H'. To establish the equality (13.20), we proceed
as follow s.
The Schrodinger equation for the total Hamiltonian appears as
I
+H
A A A

Hlfln = (Ho )Ifln = Enlfln

The ground state of Ho is a q-fold degenerate. Substituting

Ifln = qJn
En = En(O) + E n'
I (n < q)
-
(13.21)

into the Schrodinger equation gives

H -Ifln = En1-Ifln
A /
(n S. q) (13.22)

Here we have recalled that qJn' being a linear combination of degenerate states,
is itself a degenerate state (corresponding to the eigenvalue E 1(0»). With the el-
2The sequence (13.16) spans the same subspace of Hilbert space spanned by the degenerate states
{'Pn (0) I, n ~ q . Thus the basis (13.18) spans the same Hilbert space spanned by the basis {'Pi(O)I,
i ?: I.
13.2 Time-Independent, Degenerate Perturbation Theory 695

ements of {qJn} (as well as those of {Ifln (O)}) taken to comprise an orthogonal se-
quence.l one is able to identify (13.17) as being the matrix counterpart of the
operator equation (13.22). This is so, provided that we set

(n S q)

which again is the relation (13.20). This equality establishes the fact that the di-
agonal eleme~ts of the submatrix »:
are the first-order corrections to the total
Hamiltonian H (for n S q).
Let us now construct the new basis functions {qJn} which diagonalize the said
submatrix of tt' . These are given in terms of the ani coefficients in (13.16), which
make {qJn} obey the eigenvalue equation (13.22). Substituting the former into the
latter gives

Multiplying from the left with (Ifl/O) I gives

'L,aniH~i = En''L,ani8pi = En'a np (for fixed n , p S q)

i i

This equation may be rewritten as

q
'L,(H~i - En'8 pi)ani = 0 (n,psq) (13.23)
i=1

The coefficients {an i} for a fixed value of n comprise the column vector represen-
tation of qJn in the subbasis (lflt(O) , l S q} . Similarly,

H~i = (Ifl/O)IH'llfli(O»)

are the matrix elements of H' in this same basis.

The Secular Equation
For each value of nand p, (13.23) is one equation for En' and the q components
{ani}. There are q such equations corresponding to the q values of p . For n = 1,
for example, these equations appear as

H~q) (all)
H ; I - EI' H;2 H;3
H~I H~2 - EI' H~3 H 2q an
(13.24)
(
··· ... =0

H~I al q

3One may alwaysconstructan orthogonalset of functionsfrom a given sequenceof degeneratefunc-

tions throughthe so-calledSchmidtorthogonalization procedure. See Problem 13.53.
696 Chapter 13 Perturbation Theory

This is the matrix equivalent of (13.22) in the basis {lpn (0), n :s q} . Setting n = 2
in (13.23) generates (13.24), with the modifications that EI ' is replaced with E2'
and the column vector {aId is replaced by {a2d . As n runs from 1 to q , one
obtains q such equations. The condition that there be a nontrivial solution {ani}
for anyone of these q matrix equations is that the determinant of the coefficient
matrix vanish, which gives the secular equation

det IH;i - En'opil = 0 (13.25)

This equation may be rewritten in a purely operational form,

detlH' - En'II =0

The identity operator is I (or, equivalently, the q x q unit matrix). The q roots
of the algebraic equation (13.25) are the eigenvalues of (13.22). They are the
diagonal elements of the submatrix of H' depicted in (13.19). Substituting any
value of E ' so obtained, say EI ' , back into (13.24) permits one to solve for the
coefficients {ali} . In similar manner, E2' permits calculation of {a2i}, and so on.
These coefficients, in turn, give the new basis functions FPn} in (13.18).
Using this new basis (13.18), the ambiguities due to the degeneracy of Ho
are removed" and one may proceed with the analysis developed in the previous
section for the nondegenerate case. For example, (13.2) appears as

lpn = lpn + A({in (1) + A2 ({in (2) + ... n :s q

lpn = lpn (0) + Alpn (1) + A2lpn (2) + ... n > q
En = En (0) + AEn' (1 ) + A E/(2) + ...
2 n :s q (E 1(0) = ... = E q (0»

En = En (0) + AEn (1) + A2 En (2) + ... n > q

En ' = (((inlH'l({in) n :s q
E n(I) = (lpn(O)IH'llpn(O») n > q

An outline of this analysis is shown in Fig. 13.4. An important feature of de-

generate perturbation theory is that solution of the matrix equation (13.24) gives
(1) first-order corrections to the energy; and (2) corrected wavefunctions which,
together with the nondegenerate states, serve as a proper basis for higher-order
calculations .

Two-Dimensional Harmonic Oscillator

In this section we are primarily concerned with the degeneracy-removing property
of the perturbation H'. As a simple example of these procedures , consider the case
4It may be that first-order calculation does not remove the degeneracies of HO' For example, this
occurs if the off-diagonal elements of fIl are zero. For such cases it becomes necessary to include
higher-order terms to remove the degeneracy. For a discussion and problems relating to such second-
orderdegenerate perturbation theory, see L. 1. Schiff, Quantum Mechanics , 3d ed., McGraw-Hili,
New York, 1968.
 13.2 Time-Independent, Degenerate Perturbation Theory 697

Construct (13 .16)

q
"
" " +H'
H=H o . ,~. (0)
L...J a m""
;;; = "'"
?"n
ifO?"n
,,, (0) =En (0),,,"'n (0) i= I
which diagonalizes submatrix
E( (0) is q-fold degenerate
ofH' ;
<~" IH'I¥'k > =s's.,

Nondegenerate
perturbation theory
with basistHI3 .18)
(13 .23)
o (13 .25)

Gives new
basis~
(\3 .18)

FIGURE 13.4 Elements of degenerate perturbation theory.

of the two-dimensional harmonic oscillator whose Hamiltonian is

or, equivalently,

where

1 " "t 1 " "t 2 mwo

x = --(a+a) Y = -(b+b) f3 =-
v'2f3 v'2f3 Ii

The eigenstates of Ho are the product forms (8.111)

&98 Chapter 13 Perturbation Theory

which we will label lnp). The corresponding eigenenergy

E np = fUvo(n + p + I)
is (n + p + I)-fold degenerate. It follows that the energy
E 10 = EOl = 2fUvo

is twofold degenerate, with corresponding eigenstates 110) and 101) .

Let us apply the analysis developed above to determine how this energy sepa-
rates due to the perturbing potential

HI =Klxy

Furthermore, we wish to find the two new wavefunctions that diagonalize fjl.
These are given by the linear combinations (13.16)

WI = a!p1O + h!pol
W2 = al!p1O + hI!POI

The submatrix of fjl in the basis {!PIO, !pod appears as

fjl = KI ((IOIXYIIO) (1OIXYIOI)) = lE(0 01)

(OllxyllO) (OllxyIOI) I
K
I

lE=-
2f32
Consider, for example, the calculation of the (1, 2) elements of HI .

(IOlxyIOI) = 2~2 (1Olah + ath + aht + athtlOI)

I t~ I
= -(lOla biOI) = -
2f32 2f32

With these values of the matrix elements of fjl, we are prepared to solve (13.25)
for the incremental energies E I • This equation appears as

which has the solutions

I
E = ±lE

Thus we find that the perturbation separates the first excited state by the amount
2lE.
Problems 699

E+ = E IO + IE
EIO
-C E_=EIO-IE
(13 .26)

The corresponding new wavefunctions are obtained by substituting these value s

into the matrix equation (13.23) , which for the present case takes the form

- E'
( IE
IE)
-E'
(a)b = 0

The two values of E' given above serve to determine two column vectors, (a, b)
and ta' , b') . The value E' = IE gives a = b, while E' = -IE gives a/ = -b'.
Thus we obtain for the new wavefunctions, qJ, and qJ2, the values

(13.27)

These new wavefunctions serve to diagonalize the perturbation Hamiltonian H '.

PROBLEMS

13.12 How does the threefold-degenerate energy

E = 31Uvo

of the two-dimensional harmonic oscillator separate due to the perturbation

H' = K'xy?

13.13 Consider a particle confined in a two-dimensional square well with faces at x =

0, a ; y = 0, a (see Section 8.5). The doubly degenerate eigenstates appear as

IfJnp(x, y) = ~ sin C: X
) sin (P;Y)
Enp = E, (n 2 + p2)
What do these energies become under the perturbation

H' = 10-3 E, sin (:X)?

13.14 The eigenstates of a rotating dumbbell , with moment of inertia I,

are (21 + Ij-fold degenerate . In the event that the dumbbell is equally and oppositely
charged at its ends, it becomes a dipole . The interaction energy between such a
700 Chapter 13 Perturbation Theory

dipole and a constant, uniform electric field 8 is

iJl = -d. 8
The dipole moment of the dumbbell is d . Show that to terms of first order, this
perturbing potential does not separate the degenerate E[ eigenstates,

13.15 Consider again the dipole moment described in Problem 13.14. If both ends are
equally charged, the rotating dipole constitutes a magnetic dipole . If the dipole has
angular momentum L, the corresponding magnetic dipole moment is
e
p,=-L
2mc

where e is the net charge of the dipole . The interaction energy between this magnetic
dipole and a constant, uniform magnetic field 13 is

'I e '
H =-ji.13=--L.13
2mc

(a) If 13 points in the z direction, show that iJ' separates the (21 + I)-fold degen-
erate E[ energies of the rotating dipole .
(b) Apply these results to one-electron atoms to find the splitting of the P states.
(Neglect spin-orbit coupling.) (Note : This phenomenon is an example of the
Zeeman effect discussed previously in Problems 12.15 et seq.)

13.3 • THE STARK EFFECT

In Problem 13.9 we found that to within a first-order calculation, an electric field

does not remove mi degeneracy of states of definite orbital number 1. However,
as we shall now see, a similar calculation reveals that such a field will induce a
partial separation of the n 2 degeneracy of eigenenergies related to one-electron
atoms. This effect was first noticed in 1913 by Stark. He observed the splitting of
the Balmer lines in a field of 100,000 V/cm. (The more readily observed Zeeman
effect was first observed in 1897.)
The Hamiltonian of a one-electron atom in a constant, uniform electric field 8
which points in the z direction, neglecting spin, is

, iJr2 f} Ze 2
H= -+
2m
- - - -r
2mr 2
-e8z
A 'I
=Ho+H
iiI = -e8z = -e8r cos e

The eigenstates of the unperturbed Hamiltonian are n2-fold degenerate. Let us

consider how the perturbing electric field removes this degeneracy. Specifically,
let us consider the fourfold degenerate n = 2 states. The related degenerate wave-
13.3 The Stark Effect 701

functions are, in the Inlm) notation,

1200),1211),1210),121-1)

To calculate the incremental changes in the energy E2, we must solve the deter-
min antal equation

(2001H' 1200) - £' (2001H'1211) (2001H' 121O) (200IH'121-1 )

(2111H'1200) (2111H'1211) - £ ' (211 IH' 1210) (2111H'121 -1)
0=
(21OIH' 1200) (21OIH'1211 ) (2101H' 121O) - £' (21OIH'121 -1)
(21-11H'1200) (21-1IH' 1211 ) (21-11H' 121O) (21-11H'121-l) - £'
(13.28)

Only two elements survive integration. All elements with different mi numbers
vanish by orthogonality of the Inlmt) states. Equivalently, one says, "if' does not
connect states of different m; " Integration gives.'

(2101if'1200) = (2001if'121O)

= _ eao8
32rr Jo
rex:> p\2 _ p)e-P dp 11
-1
d cos e cos 2 er
Jo
2Jr
d¢

e8h?
= - - = -lel38ao = -IE
mZe
With these values inserted into the determinant above, (13.25) becomes

-E' o -IE o
o -E' o o =0
-IE o -E' o
o o o -E'

which has the four roots

E' = 0,0, +IE, -IE
(13.29)
IE = 31el8ao

Thus we find that to terms of lowest order in the electric field 8, the degenerate
n = 2 state separates into three states :

To calculate the new n = 2 wavefunctions

({J = a1200) + b1211) + c121O) + dI2l-1)

5The nondimensional radius P is defined in Table 10.3. See also Tables 9.1 and 10.5.
702 Chapter 13 Perturbation Theory

we substitute the values (13.29) into the matrix equation

o -II':
-£' 0
(13.30)
o -£'
o 0

There results
1
£z + = £z(O) + II': ---+ 1fJ+ = .j2(1200) - 1210))

1
£z- = £Z(O) - II': ---+ 1fJ- = .j2(1200) + 1210)) (13.31)

£zo = £Z(O)<1fJ = 1211)

IfJ = 121-1)

The perturbation mixes the m = 0 states, while the m = 1, -1 states are left
degenerate. The values ±II': represent the average values of the interaction ii' in
the respective states , 1fJ±.

Symmetry Breaking and Removal of Degeneracy

Degeneracy, such as in the example of the two-dimensional harmonic oscillator
considered above, is due to symmetry properties of iio. In this example the poten-
tial x Z+yZ is noted to be invariant under rotation about the origin in the plane . The
perturbation xy destroys this symmetry, thereby removing the degeneracy. (Recall
related discussions in Section 8.5.) It has been noted that an eigenenergy which is
n-fold degenerate corresponds to n independent eigenfunctions. When the sym-
metry related to this degeneracy is broken , an n-fold degenerate eigenenergy may
split into as many as n new eigenenergies. Another example addresses the case
when a hydrogen atom is placed in an external field. This perturbation breaks ro-
tational symmetry of the Hamiltonian about the nuclear center of the atom . In the
problem of the Stark effect discussed immediately above, a uniform electric field
breaks rotational symmetry and eigenenergies proliferate.? In addition, as noted
in (13.31), there is a decrease in the number of "good quantum numbers" from
(n , l , m) to (n). Consider the case in which the strength of the symmetry-breaking
perturbation is represented by a parameter in the Hamiltonian, which we label g .
If eigenenergies are plotted versus g, then at some value of g, eigenenergies bifur-
cate . Such behavior of eigenvalues is sometimes referred to as "level repelling."
The properties of repelling of levels and loss of good quantum numbers are often
associated with nonintegrable systems.

6The order of degeneracy of a given system may exceed that implied by symmetry principles (i.e.,
group theory). Such, for example, is the case for the two-dimensional harmonic oscillator as well
as for the hydrogen atom. For further discussion, see F. A. Cotten, Chemical Applications of Group
Theory, 3d ed., Wiley, New York, 1993.
 13.4 The Nearly Free Electron Model 703

PROBLEMS

13.16 Show that the n = 2 matrix of the interaction Hamiltonian iI ', of the Stark effect,
is diagonal in the basis (13.31) .

13.17 What is the dipole moment of the hydrogen atom in the CfJ± states (13.31)?

Answer
The interaction energy between an electric dipole d and the electric field 8 is

H' = -d · 8

In the CfJ± states , the average value of H ' is E = ±3Ielao8. We may infer from this
result that the magnitude of the dipole moment in the CfJ± states is 3lelao. The di-
rections of these moments are parallel or antiparallel to the z axis (i.e., the direction
of 8).

13.18 What is the charge density q(r, B) of the hydrogen atom associated with the state
CfJ- (13.31) ?

Answer

q(r, B) = e[CfJ_1 2 = _e_

16JTao3
[I - s. 2(~)J2
ao
sin
2
e- r/ ao

13.19 Of the two states CfJ± in (13.31), CfJ- is said to be more stable than CfJ+ . Why ? Discuss
your answers in light of the interaction energy, -d . 8 .

13.4 • THE NEARLY FREE ELECTRON MODEL

In this section we return to the problem of an electron in a periodic potential V (x),

discussed in Sections 8.2 and 12.9. Wavefunctions are in the Bloch form

rp(x) = u(x)e i kx

where the periodic function u(x) has the same period a as Vex) . Eigenenergies
are functions of the crystal momentum wavenumber k. For a lattice of length L,
the nearly continuous wavenumber k has the discrete values [see (8.42)]

We recall (see Problem 8.13) that in the high-energy domain (E » V), the
n
energy spectrum reduces to the free-particle values 2k2/ 2m , or, equivalently,

n2k/
Ek ' = - - = - -
j 2h 2
J 2m 2mL2
which is doubly degenerate: Ekj = E-kj '
704 Chapter 13 Perturbation Theory

We now wish to obtain an expression for the energy gap oEn at the nth band
edge. A band edge, we recall, is a break in the energy spectrum which occurs at
the k values k ja = nn .
In the present analysis the periodic potential is considered a small perturbation
to the free-particle Hamiltonian

A PA2
Ho=-
2m
The electron is "nearly free," which is the same as saying that E » V .
Unperturbed eigenstates are normalized to the sample interval L.

cpk . (0) = _1_ exp(ik rx )

J .Ji J

With k j = j (27l' / L) these functions comprise an orthonormal sequence, as may

be seen as follows:

= -1 jL .
12 sin[(kj - kq)L/2]
(kqlkj) exp[l(kj - kq)] dx = ----"---'--,-- (13.32)
L -L12 (kj - kq)L/2

= sin(j - q)7l' = o.
(j _ q)7l' Jq

The Perturbation Potential

The perturbing potential is periodic and may be expanded in the Fourier series
(see Problem 13.58).

H' = Vex) = f
n=-oo
Vn exP [ i 27l' n (~)] (13.33)

The zero energy line in the present analysis is set at the average of Vex) . This
ensures that the de component Vo of V vanishes (see Fig. 13.5).
LI 2
Vo= j
-L12
V(x)dx=O

Application of the first-order perturbation theory necessitates calculation of the

matrix elements of H'.

(13.34)

= L VnOkq .kj+(2rrnla)
n
13.4 The Nearly Free Electron Model 705

V(x)

FIGURE 13.5 The zero in potential is chosen so as to eliminate Vo.

L/ 2
Vo =
l-L/2
V(x)dx =0

Substituting these matrix elements into the first equation of (13.10) gives the
first-order corrected wavefunctions

_ (0) _1_ ~ Vn exp{ix[kj + (2rrn/a)]} (13.35)

rpkj - rpkj + v ITL n=-oo
Z:: E
kj
(0) _ E
k(O)
j+(27rn /a)

Calculation of the first-order corrected eigenenergies as given by the second

equation in (13.10) gives

Ek
J
= Ek(O)
J
+ (kJ·lVlkJ· ) = Ek(O)
J
+ Vo (13.36)

nh·J_ 2
= Ek ' (0) = __
J 2m
To first order, the energy remains unperturbed. This free-particle spectrum was
found in Chapter 8 to maintain in the domain E » V . However, in the present
analysis there is explicit evidence which indicates that this result is invalid at the
band edges of the energy spectrum . Namely, the summation in (13.35) for the
first-order corrected wavefunctions becomes singular if the denominator of any
term vanishes.

0) fj,2 4rrn
Ek '( - E (0) = ---(k
2
j a +nrr) = 0 (13.37)
J kj +(27rn/ a) 2m a
This singularity arises from the zeros of (13.37), corresponding to the degeneracy
at the band edges, k ja + nn = O. To obtain correct energies at these values of
k] , one must use degenerate perturbation theory. As described in Section 13.2,
the first step in this procedure is to construct a new basis {W} that diagonalizes the
relevant 2 x 2 submatrix of H' . This new subbasis {W} is constructed from linear
706 Chapter 13 Perturbation Theory

combinations of the degenerate portion of the unperturbed basis and we may write

(13.38)

Diagonalization of H' in the basis (13.38) yields the matrix equation

(13.39)

The matrix elements of H' are given by (13.34), from which we find

HkJ kJ = H~k J'

-kJ = Vo = 0
Only the off-diagonal elements survive. Choosing the origin so that Vex) is an
even function gives

HkJ ' -kJ· = Vn = V- n = H~k · k·

l
J. J

The determinant equation (13.25) then becomes

which gives the roots

Resubstituting into the matrix equation (13.39) gives two nondegenerate eigen-
states, which written together with the first-order corrected energies appear as

e..: =
J
Ek(O)
J
+ Vn qJ+ = 2a sinkjx

Ekj - = Ek j (0) - Vn qJ_ = 2acoskjx

These eigenstates , we recall, are the standing waves at the band edges previously
obtained in Section 8.3 relevant to the Kronig-Penney potential. The present cal-
culation affords an estimate (in the domain E » V) of the width of the energy
gap at the band edges.

Thus we find that the nth gap in the energy spectrum E = E(k), at the values
k = ±nJr[a, has width which is twice the nth Fourier coefficient of the potential
V (x) (Fig. 13.6).
Problems 707
E

FIGURE 13.6 Energy gaps at the band edges in the nearly free electron model are given
by twice the corresponding Fourier coefficient of the periodic potential.

In the nearly free electron model described above, conduction electrons are
weakly bound to atoms of the lattice. This is opposite to the situation in the tight-
binding approximation described in Problem 8.16 and in the LCAO analysis pre-
sented in Section 8.7.

PROBLEMS

13.20 Resketch the E(k) curve shown in Fig. 13.6 for the case that the zeroth Fourier
coefficient Vo > O.
13.21 A periodic potential has Fourier coefficients Vn = VIIn 2 . The width of the tenth
energy gap is 0.031 eV. What is the value of VI?
13.22 What are the energy gaps at the band edges for a particle in the periodic delta-
function potential
00

Vex) = VOa L o(x - qa)

q=-oo

defined over the entire x interval? [Hint: Consider the following delta-function rep-
resentation
00
o(y) = L exp(i2nny)
n=-oo
The right-hand side is periodic with period 1 and represents a delta function at
y = 0, ±1 , ±2, ... . The left-hand side is a delta function only at the origin. The
domain of validity of the equation may be extended to the whole y axis if one writes

; , o(y - m) = ; , eXP(i2nny).]

13.23 (a) Show that the first-order corrected wavefunction C{Jk as given by (13.35) may be
cast in the form of a Bloch wavefunction C{Jk = u(x)e i kx •
(b) Show that the expression you obtain for u(x) has period a .
708 Chapter 13 Perturbation Theory

13.24 Estimate the energy gaps at the band edges for the Kronig-Penney potential of
potential height Vo, well width a, and barrier width b.

Answer

liEn ~
2Vo .
--sm --(mra)
nit a +b

13.25 Estimate the band-gap widths for the potential

Vex) = 2Vocos
27fX)
( ---;;-
Answer
Rewriting the cos term in terms of exponentials,

Vex) = 2Vocos 27fX) =

( ---;;- [(i27fX)
Vo exp-a- + exp (-i27fX)]
-a-

reveals that this potential has only two nonvanishing Fourier coefficients. First-order
perturbation theory implies the existence of the gaps at k = ±7f/ a, of width 2 Vo.
Higher-order perturbation theory would uncover energy gaps at subsequent band
edges as well. This may be concluded directly by writing the Schrodinger equation
for the given potential'?

CPxx + 2m
1i2 [E - 2V cos (27fX)]
---;;- cP = 0

This is a well-known equation in mathematical physics and is called the Mathieu

equation . As with most such equations, it stems from writing the "wave function,"
(''12 + k 2 )cp = 0, in a particular orthogonal coordinate frame. For the Mathieu
equation these are elliptic cylinder coordinates . Solutions of the equation are called
Mathieufunctions and have been studied in detail. These analyses reveal a sequence
of intervals on the E axis in which solutions to the equation are unstable .S
13.26 At 0 K a certain semiconductor has its valence band filled and conduction band
empty. The potential "seen" by electrons may be approximated by the function

Vex)
27fX)
= 2Vocos ( ---;;-
Vo = 0.1 eV, a = 0.5 A

(a) Assuming that the valence band is the band of lowest energy, estimate the width
of the gap li E between the valence and conduction bands.

7This case has been studied in detail by P. M. Morse, Phys. Rev. 35, 1310 (1930) . Here it is also shown
that the energy bands approach the line spectrum of an infinitely deep well as the amplitude Vo of the
periodic potential grows infinitely large.
8That is, serie s solutions in these domains do not converge . For further discussion of the properties of
Mathieu functions, see P. M. Morse and H. Feshbach, Methods of Theoretical Physics , McGraw-Hill,
New York, 1953, Chapter 5; also, E. T. Whittaker and G. N. Watson, A Course of Modem Analysis,
Cambridge University Press, New York, 1952, Chapter 19.
 13.5 lime-Dependent PerturbationTheory 709

(b) Setting 8 E = k B T, estimate the temperature at which the sample will begin to
conduct appreciably. (kB = 8.6 x 10- 5 eV/K.)

Answers
(a) 8E::: 2Vo = 0.20 eV
(b) T::: 2300 K

13.5 • TIME-DEPENDENT PERTURBATION THEORY

Time-dependent perturbation theory addresses the following problem . Initially,

the unperturbed system is in an eigenstate of No. Then the perturbation, if' (z), is
"turned on." What is the probability, after a time t, that transition to another state
(of No) occurs ?
The total Hamiltonian for these problems is of the form
A

H (r, t) = Ho(r) + AH (r, t)

'" "I
(13.40)

where Ais again a parameter of smallness .

Let the time-dependent eigenstates of No be written

1/In(r, t) = rpn(r)e- i Wnl

(13.41)
A _ (0) _
HOrpn - En rpn = liwnrpn

Suppose that at time t > 0, the system is in the state

1/I(r, t) = L Cn(t)1/In (r, t) (13.42)

n

Then , by the superposition principle , ICn 12 is the probability that measurement

finds the system in the state 1/1n at the time t . Thus the primary aim of the present
discussion is to calculate these coefficients. They are determined in the following
manner. The wavefunction 1/I(r, t) is a solution of

01/1
in- = (Ho + AH')1/I
A A

(13.43)
ot
Substituting the expansion (13.42) into this equation and then operating from the
left with f dr 1/Ik * (r, t) gives

dq = A "L....J{kIH ' In)cn

in- (13.44)
dt n

This is an infinite sequence of coupled equations for the coefficients {q(t)}. In

the limit that A ---+ 0, the q coefficients are all constant. It is therefore consistent
to seek solution in the form
710 Chapter 13 Perturbation Theory

(13.45)

Substituting this series into (13.44) and equating tenus of equal powers in A. gives
(with a dot denoting time differentiation and Hf<n written for the matrix elements
of if')

i nck (0) = 0
i nck (I) = L Hf<n cn (0)
n

i nck(2) = LHf<ncn(l ) (13.46)

n

. i nck(s+ l ) = L H~ncn (S)

n

The lowest-order equations for Ck(0) indicate that these coefficients are all con-
stant in time. They are the initial values of the coefficients {Ck (t)} .
We now specialize to the problem in which it is known that initially the system
is in a definite eigenstate of Ho , say 1/F/(r, t) . With (13.42) this implies that as
t ~ -00,

1/F(r, t) ,...., 1/F/(r, t) = L on/1/Fn(r, t)

n (13.47)

Note that we have taken "initially" to denote the time t = -00. Substituting this
value into the second equation in (13.46) gives (dropping the superscripts 0 and 1)

inck(t) = L H~ncn(-oo) = H~/ (13.48)

n

For n f= 1, Cn (-00) = 0, so the first-order solution for Ck(t) is given by

Ck(t) = ~
1
In
if
-00
I I
Hk/(r, t) dt
I
(k f= 1) (13.49)

If the time dependence of if' (r , t) is factorable , then

H '(r , t) = lHr'(r)J(t)

and the matrix elements of H becomes [with (13.41)]

H~/(t) == (1/FkIH'(r, t)I1/F/) = (tpkllHI'(r)ltp/)ei wk/f J(t)

= lHI k/eiWklf J (t) (13.50)

liwk/ == Il,(Wk - wI) = Ek - E/
Problems 711

(Note that we have deleted the zero superscripts of Ek and El .) This gives the
more explicit form of edt),

q(t) = -!£
lHl
'
In
L-00
eiwklt'f(t')dt' (13.51)

These coefficients determine the effect of the perturbation on the initial state , VI[.
The probability that the system has undergone a transition from this state to some
other eigenstate of He; Y,k, at the time t, is

(13.52)

Application of these results follows. The transition probability Pl-'>k is hereafter

written Plk .
Matrix elements in time-dependent perturbation theory are conventionally
written with the initial state on the right and the final state on the left, as illus-
trated in the following symbolic form :

(final state I interaction I initial state)

This convention is employed in the remainder of the text.

PROBLEMS

13.27 A system with discrete eigenstates {CPn} and eigenenergies {En} is exposed to the
perturbation
A A e-t2/r2
H' = lHI' ( r ) - - -
r.fir
The perturbation is turned on at t = -00, when the unperturbed system is in its
ground state, 1/fo. What is the probability that at t = +00 the system suffers a
transition to the state 1/fk' k > O?
Answer
To obtain the answer to this problem, we must calculate the time integral in (13.52).
Writing w for WkO, we have

I = -I
.fir
1 00

-00
e,wt
.
«:' 2 Ir 2 d(t Ir )

= _1_1
.fir
00

-00
eiw~-~2 d~ (~ == t i t ; W == rw)

1
= .fire--2 4
1
W 1
00

-00 exp [- (~ - iW)2] d~

2
712 Chapter 13 Perturbation Theory

Substituting this into (13.52) gives the desired result

~O
2
r2 2
POk = I 1 e-(Eo-Ed j2fi

13.28 Consider that the unperturbed system in Problem 13.27 is a particle of mass m
confined to a one-dimensional box of width a. The spatial factor in H ' (x, t) is

10- 4 2
A

Ifu' (x) = Px
2m
What state does the perturbation leave the system in at t = -l-oo?

13.6 • HARMONIC PERTURBATION

Stimulated Emission
As a first application of the analysis above, we consider a perturbation that is
switched on at t = 0 and is subsequently monochromatically harmonic in time.
The perturbation acts on a system whose Hamiltonian is flo. If it is definitely
known that the unperturbed system was in one of its own stationary states before
the perturbation was applied (t < 0), in what state will measurement find the
system after the perturbation has been turned on for t seconds? This problem is
appropriate, for example, to an atom that interacts with a (weak) electromagnetic
field. The explicit form of the perturbation fI' is (with w > 0)

HA, (r, t) = 1 0
A

2H ' (r ) coswt
t <
t2:0
0
(13.53)

(see Fig. 13.7). Substituting this form into (13.51), with f(t) = 2 cos on, gives

H' "
=1HJ'(r)f(t)
f(t)=2coswt

system in the
state ~l at
time t < 0
measurement made
at time t > 0
FIGURE 13.7 Harmonic perturbation.
 13.6 Harmon ic Perturbation 713

lHI 'kl
q(t) = -.-
In
it . , .,+ .,
0
e'Wklt (e- tWf e'Wf ) dt'

_
---
n
I
lHI~1 [ei(Wkl-W)t -
Wkl - W
ei(Wkt+W)t -
+-----
Wkl + W
I]
Employing the relation (see Problem 1.21)

e iB - I = 2ie iB/ 2sin(()/2)

permits the equation above to be rewritten

i2lHI ~l [e i(Wkl-W)t/2 sin(wkl - w)t / 2 e i(Wkl+W)t /2 sin(wkl + w)t /2]

cdt) = - - - + ---------
n Wkl - W Wkl + W
(13.54)

The dominant contributions to Ck come from the values W ~ ±Wkl. At these

values, the moduli of the two bracketed terms in q, respectively, assume their
maximum value, t /2. These resonant frequencies correspond to the energies

W ~ +Wkl ----+ Ek > E,

W ~ -Wkl ----+ Ei > Ek

In the first case, the "final" energy Ek is larger than the "initial" energy Ei. The
system absorbs energy and jumps to a higher energy level.

The energy absorbed, lu», is that of a photon in the incident radiation field
(Fig. 13.8).
For the second case the perturbation induces a decay in energy

A photon of energy Iuo is radiated away from the system. This decay process is
stimulated by a photon of the same frequency in the perturbation field.

- - - . - - - - Ek

_ _ _...1- E _ _ _...1- E
l k
Resonant absorption Stimulated emission

FIGURE 13.8 Dominant transition processes due to harmonic perturbation.

714 Chapter 13 Perturbation Theory

sin2 ~ (wk/ - w)t

(Wkt - w)2

-61T -41T -21T 41T

t t t t

FIGURE13.9 Frequency dependence of probability of transition from the Ith to the kth
state at the time t due to harmonic perturbation of frequency co.

Let us consider the case that the incident radiation field excites only higher
energies, so that Wkl > O. Under such conditions, the first term in (13.54) dom-
inates and the probability that the perturbing field causes a transition to the kth
state becomes

Plk = Ickl
2
= ~2(
411HI~112 2 sin2 [I-2(Wkl- W )t ] (13.55)
It Wkl - w)

The frequency dependence of this function is plotted in Fig. 13.9.

Energy-Time Uncertainty

From this sketch it is evident that states falling in the interval

(13.56)

have the greatest probability of being excited, after the perturbing field has acted
for t seconds. If this perturbation is applied many times to an ensemble of in-
dependent, identical systems, all initially in the state 1{!1, then after a time t the
energies excited among the members of the ensemble will lie primarily in the in-
terval /),.E. Thus (13.56) gives the uncertainty in the values of energy observed.
The final energies Ek are spread throughout the interval

/),.E::: -
n
t
13.6 Harmon ic Perturbation 715

(13.57)
Efinal ~ E inilial ±!:J.E

Note that Einilial refers to the energy of the incident photon plus that of the system
in its initial state. Our perturbation analysis returns the principle of conservation
of energy, properly modified by the uncertainty relation , !:J.E ~ hit.

Long-lime Evolution
The transition probability formula for absorption (13.55) and its counterpart for
stimulated emission may be written together as

41lHI ~112 2 [ 1 ] (13.58)

Plk = 2 2 sin -(Wkl =f w)t
h (Wkl =f w) 2

where the =f signs refer to absorption and stimulated emission, respectively (see
Fig. 13.8). This expression takes a convenient form in the long-time or, equiva-
lently, high-frequency limit. This form follows from the delta-function represen-
tation

~ 2 . sin 2 (wt 12)

o(w) = - hm 2 (13.59)
T( 1-+00 tea

so that in the same limit,

(13.60)

The corresponding transition probability rate appears as

(13.61)

In this or the formula above, the delta function expresses the fact that in the
long-time limit, the Fourier transform of a monochromatic perturbation becomes
sharply peaked about the frequency of perturbation. The system "sees" only a
single frequency. Since the uncertainty in energy hi t vanishes in this limit, the
argument of the delta function is also an expression of conservation of energy.

Short-lime Approximation
If a harmonic perturbation is applied to a system for a short-time interval such
that (Wkl - w)t « 1, (13.55) may be expanded to yield

(13.62)
716 Chapter 13 Perturbation Theory

The related transition probability rate is

(13.63)

At early times, the rate at which transitions to the kth state occur grows linearly
with time.

The Golden Rule

In many problems of practical interest, the final excited states lie in a band of
energies (Fig. 13.IO). Such is the case, for example, for ionization or free-particle
scattering states. Such states comprise a continuum. If the density of final states
is geEd, then

is the number of energy states in the interval Ek to Ek + d Ei , The probability that

a transition occurs to a state in a band of width 2~ centered at Ek is

Inserting (13.55) gives

_ _ _ _...l- £1

FIGURE 13.10 Resonant absorption to a band of energies . If the density of states about
Ek is 9(Ek), then the probabilit y rate of transition to the band is

2n / 2
Wlk = /i9(Ek)llHI kll
Problems 717

For fixed E/, t and w

dE" = 2Mf3
t
and

where 28 is the corresponding spread in f3 values. Owing to the rapid decay of the
sin 2 f3 /f32 function (see Fig. 13.9), only a slight error is introduced in the expres-
sion above if we replace the interval (-8, +8) by (-00, +00). Furthermore, if
we assume that 9 and JH[ k/ are slowly varying functions of Ei ; they may be taken
outside the integral. There results

The related transition probability rate is

27T , 2
W/k = Ii'"g(EdllHI k/1 (13.64)

This formula was found to have such widespread application that Fermi" dubbed it
"Golden Rule No.2." It will be applied in Section 13.9 in study of atom-radiation
interaction and in Chapter 14 in study of the Born approximation in the theory of
scattering.

PROBLEMS

13.29 Show that if the perturbation field

iI ' = 2H (r) cos (J)(

acts on a system initially in the lth state of a very long time, and (J) :::::: (J)k/, then the
only state that will be excited is the kth state. Interpret this result in terms of the
Fourier decomposition (in time) of iI '.
13.30 The expression in the text obtained for P/k , the probability of transition to a band of
states, is seen to be independent of the frequency (J) of the perturbing field. In what
approximation is this result valid?
13.31 What does the transition probability P/k (13.55) become if the perturbing field is
precisely "on resonance" ; that is, if (J) = (J)k/?

9E. Fermi, Nuclear Physics, University of Chicago Press, Chicago , 1950.

718 Chapter 13 Perturbation Theory

13.32 A polarized beam of current, J cm- 2/s, contains electrons with spins aligned with
a steady magnetic field of magnitude jJ which points in the z direction. The beam
propagates in the x direction . The wavefunction for an electron in the beam is

1fi = _l_ei(kx-wt)
(27l')1/2
(0) 1

A monochromatic electromagnetic field of frequency

extends over a length of beam path, L cm long. A Stern-Gerlach analyzer is in

the path of the beam at a point beyond the domain of the electromagnetic field. Its
orientation is such that spins aligned with B are not deflected from the beam. If
the beam moves with the speed v, what is the current of electrons scattered out of
the beam by the S-G analyzer? Assume that the interaction between electrons in the
beam and the electromagnetic wave is

ii' = -II-' jJ' coscot

where jJ' is the amplitude of magnetic field of the wave. The component of B' in
the z direction is small compared to B .
13.33 A one-dimensional harmonic oscillator of charge to mass ratio e/ m and spring con-
stant K is in its ground state. An oscillating uniform electric field

8(t) = 28 cos wot,

is applied for t seconds, parallel to the motion of the oscillator. What is the proba-
bility that the oscillator is excited to the nth state given that (wnO - WO)t « I?
Answer

13.7 • APPLICATION OF HARMONIC PERTURBATION THEORY

In this section we apply the transition probability fonnula (13.61) found above to
Einstein's derivation of the Planck radiation formula (2.3) and to a brief qualitative
description of the laser.
Einstein considered the equilibrium state between the walls of an enclosed
cavity and the radiation field interior to the cavity. Atoms in the walls constantly
exchange energy with the radiation field. The excited states of these atoms are
very closely spaced and essentially comprise a continuum. Consider that the two
energies EI and Ek are representative of two states in the continuum (Fig. 13.11).
Photons with energy hv = Ek - EI can be absorbed by atoms in the EI state
raising them to the Ek state. Atoms may decay from the Ek to the EI state by
stimulated or spontaneous emission. Stimulated emission was discussed in the
preceding section. Spontaneous emission is related to the natural lifetime of the
excited state and is more dependent on internal properties of the radiating system.
13.7 Application of Harmonic PerturbationTheory 719

FIGURE 13.11 Energy states of closely packed atoms in the walls of a radiation cavity.

Einstein A and B Coefficients

The rate at which atoms in the Ek state decay by stimulated emission is propor-
tional to the number of such atoms (Nk) and number of photons of frequency v
in the radiation field. This number of photons is proportional to photon energy
density, u(v) . The rate at which the atoms in the Ek state decay by spontaneous
emission, on the other hand, is proportional only to the number of atoms Nk in
the Ek state. Thus the total transmission rate of decay of atoms with energy Ek to
energies E/ is

(13.65)

The transition probability rate per atom is written Wk/. The proportionality
constants A and B are called Einstein A and B coefficients.
Atoms in the E/ state can be excited to the Ek state only by absorption of a
photon of frequency (Ek - E/) / h. Thus the rate of elevation of atoms in the E/
state to the Ek state is

(13.66)

Planck Radiation Formula

In equilibrium, these rates must equal:
720 Chapter 13 Perturbation Theory

Equivalently, we may write

Ni Akl + BkIU(V) (13.67)

Nk BlkU(V)
The ratio N; / Nk may be obtained from elementary statistical mechanics. If the
energy of an aggregate of atoms at the temperature T is partitioned such that NI
of the atoms have energy E I, N2 have energy E2, and so on, and the total energy of
the aggregate is constant, then the most probable distribution of energies is given
by the Boltzmann distribution . 1O In this distribution the number of atoms N, with
energy E, is proportional to the Boltzmann factor, exp( - Ed ke T), where ke is
Boltzmann's constant. It follows that the ratio Ni/ Nk has the value

Nl = e(Ek-E/) /kBT = ehv /kBT

Nk
Substituting this into (13.67) gives

(13.68)

The basic structure of the Planck radiation formula (2.3) follows if Blk = Bkl .
This equality may be obtained from the harmonic perturbation theory developed
above. To these ends we consider the interaction between a wave mode in the
radiation field,

8 = 802coswt

and an atom in the wall of the cavity. If d is the dipole moment of the atom, this
interaction is

H' = -d· 8 = -80' d2coswr (13.69)

which is identical to the perturbation (13.53). Substituting into (13.61) gives the
transition probability rate for radiative absorption,

2rr 2
Wlk = li2 l(kI80· dll}1 8(Wkl - w)

If the electric field is isotropic (80 is randomly oriented), then

2 1 2 2
l(d·8o}l = 3(80}ldlkl

Furthermore, the energy density associated with this mode is

1 2
U(w) = -(80 )
2rr
IOThis distribution was employed previously in Problem 2.36. For further discussion, see F. Reif,
Fundamentals a/Statistical and Thermal Physics, McGraw-Hili, New York, 1965.
13.7 Application of Harmonic Perturbation Theory 721

It follows that

4n 2
vn« = 3li2 U(W)ldlkI28(Wkl - w)

With u(w) = U(W)8(Wkl - w) , (13.66) may be rewritten

Equating this value to the preceding expression for Wlk gives the desired result,

(13.70)

The square moduli of the matrix elements of d obey the relation

l(lldlk}1 2 = (lldlk) . (kldll)

= (kldll) • (lldlk) = l(kldll}1 2

It follows that

and (13.68) reduces to the desired form,

AlB
u (v) - --=---=-=-- (13.71)
- ehv/kBT - 1

To obtain the ratio A I B we will use the correspondence principle . This rule stip-
ulates that (13.71) should reduce to the classical Rayleigh-Jeans law (discussed
in Section 2.2) in the limit h -+ O.

8nv 2
URJ = -3-kBT
c
Expanding the exponential in (13.71) in this limit gives
2
AIB = 8nv kBT
hvlkBT c3
from which we obtain the desired result (see Problem 13.52)

Substituting this value into (13.71) gives Planck's formula , (2.3).

722 Chapter 13 Perturbation Theory

The laser
The concepts of stimulated and spontaneous emission play an important role in
the theory of the laser. A laser is a device for producing an intense beam of co-
herent monochromatic light. A coherent beam may be defined as follows. Two or
more collinear, unidirectional, monochromatic beams of electromagnetic radia-
tion which propagate in the same region of space , and are in phase, form a coher-
ent beam. In 1954, C. H. Townes II conceived of a process for the generation and
amplification of such coherent radiation in the microwave domain. The device was
called a maser. The term is an acronym for the words microwave amplification by
the stimulated emission of radiation. Shortly after, these concepts were extended
to the optical region . 12 In this domain the corresponding device is called a laser.

Coherent Photons
The central principle in the realization of the laser is as follows. In constructing
formula (13.65) for the transition rate from the Ek state to the EI state, decay due
to stimulated emission was taken to be proportional to the number of resonant
photons present in the radiation field. Consider that a number of atoms in a gas
are in the excited state Ei, Then when a photon of frequency

falls on one of these atoms it stimulates the emission of another photon of the
same frequency. These two photons , the emitted and incident, travel in phase in
the same direction and combine coherently. A second atom, in the vicinity of the
first and also excited to the Ek state, suddenly "sees" a duplication of resonant
photons and is stimulated to emit another coherent photon, thereby adding to the
intensity of the coherent radiation and amplifying it.
We found above that the matrix elements for radiative excitation, Blk, and stim-
ulated emission , Bkl ' are equal [(13.70) et seq.]. To ensure that resonant photons
stimulate decay to the EI state more than they are absorbed in exciting the atom
to the higher Ek state, the number of atoms , Ni ; in the Ek state must outweigh the
number of those in the lower EI state, NI . At any finite temperature, T, the ratio
of these numbers of atoms, Nk! NI , is given by the Boltzmann formula [preceding
(13.68)] .

II J. P. Gordon , H. J. Zeiger, and C. H. Townes, Phys. Rev. 99, 1264 (1955).

12A. L. Schawlow and C. H. Townes, Phys. Rev. 112, 1940 ( 1958). For further discussion and ref-
erence, see B. L. Lengyel, Introduction to Laser Physics, Wiley, New York, 1966; T. B. Melia, An
Introduction to Masers and Lasers, Chapman & Hall, London, 1967.
13.7 Application of Harmonic Perturbation Theory 723

Population Inversion and Optical Pumping

The number of atoms in the higher Ek state decreases exponentially with energy
difference, hv . To effect a population inversion, so that Nk > Ni , an outside
source must be brought into play.13 In optical pumping, Nk is increased by irra-
diation with light of frequency v ~ (Ek - Eo)/ h, where Eo is the ground state
of the atom. Another technique for effecting a population inversion is by bom-
bardment of electrons with energy E > Ek - Eo, thereby exciting atoms in the
ground state to higher energy states. In a third technique, inelastic collisions be-
tween atoms in the ground state with those of a foreign gas which are in an excited
state E > Ek - Eo serve to populate the higher-energy states.
Consider three atomic states: Eo, E I, and E2. The ground state is Eo, while E2
is a short-lived state with a lifetime of the order of 10 to 100 ns. The more stable
("metastable") E I state has a lifetime of the order of J1S to ms. The upper E2 state
replenishes the metastable state through spontaneous decay [the A coefficient in
(13 .65)]. These randomly emitted photons comprise an incoherent radiation field.
In Fig. 13.12 a scheme is depicted where the E2 state is populated through pump-
ing with an external source. When an atom in the metastable E I state decays to the
Eo state, neighboring atoms in the populated E I state decay, through stimulated
emission to the ground state, and a coherent beam is generated.
Such a device is realized in an optical cavity of cylindrical shape with end
mirrors positioned so that the cavity is tuned to the frequency mode (EI - Eo)/ h.
Photons propagate parallel to the axis of the tube. The radiation field is coherently
amplified by a stimulated emission (E I ~ Eo) and may be tapped through a small
aperture on the axis, in one of the mirrors. The spontaneously emitted photons
(E2 ~ EI) propagate in random directions and are dissipated in collisions with
the walls.

Spontaneous

E2 - - - r - - - - " " " T - - + - - (Short-lived)

Incoherent
Pumping E t --'--r---(Metastable)
~

Eo '-- +-..1..-_ _

Stimulated

FIGURE 13.12 Schematic for the three-level laser.

13Including the effects of degeneracy, the number of atoms in the nth state is given by Nn = 9nNn .
The condition for growth of radiation then becomes Nk /9k > Nj /91' which returns the inequality
stated in the text. For further discussion, see A. Yariv, Quantum Electroni cs, Wiley, New York, 1968.
724 Chapter 13 Perturbation Theory

PROBLEMS

13.34 Obtain a relation between the spontaneous emission coefficient Akl and the dipole
matrix element Idkil , analogous to (13.70).

13.35 The electric dipole moment of the ammonia molecule, NH3, has magnitude d =
1.47 x 10- 18 esu-cm . A beam of these molecules with dipoles polarized in the +z
direction enters a domain of electric field of strength 1.62 x 104 Vfcm, which points
in the - z direction . The resulting interaction between the molecules and the field
gives rise to coherent radiation . What is the frequency of this radiation (Hz)? (Note:
Units of voltage in cgs are statV; I statv = 300 V.)

Answer

v = 24GHz

13.36 In deriving Planck's formula (see Problem 2.36 et seq.) for the density of photons
in frequency interval v, v +dv, in a radiation field in equilibrium at the temperature
T, in cubic volume V = a 3 ,

8Jfv2 dv I
n(v) dv = 3 h fk T
c eVB-I

one assumes the relation

n(v) dv = fBE(v)g(v) dv
The term g(v)dv is the density of available states in the said frequency interval,
while fBE is the Bose-Einstein factor, giving the average number of photons per
state,

Using the momentum-position uncertainty relation (appropriate to a three-dimen-

sional box), derive the following expression for the dens ity ofstates g , relevant to a
radiation field:

8Jfv2 dv
g(v)dv= 3
c

Answer
With p = hv ]c, we look at Cartesian p space and note the volume of a spheric al
shell of radius p and thickness dp .

Volume of shell = 4Jf p2 dP

All momenta in this shell correspond to nearly the same energy-and therefore
the same frequency. Owing to the uncertainty principle for a particle confined to a
three-dimensional box (see Problem 5.42), the smallest volume (a cell in p space)
that may be specified with certainty to contain a state is
Problems 725

It follows that the number of states in the dp shell is

Volume of shell
Volume of cell

Finally, we note that one counts states only in the first quadrant of p space insofar
as photons have only positive frequencies . Furthermore, to each such p state there
are two photon states, corresponding to two possible polarizations. This gives

I 4JrVp 2dp
V g(v) dv =- x 2 X ---'3~---C-
8 h /8
The desired answer follows using the given relation between p and v for a photon.
13.37 (a) What are the Hamiltonian Ho, eigenstates and eigenvalues of a collection of
harmonic oscillators with natural frequencies , WI, W2 , . .. , in the number op-
erator representation?
(b) The Hamiltonian constructed in part (a) represent s a radiation field in second
quantization. In this representation the radiation field is viewed as a collection
of harmonic oscillators . A perturbation H' is applied to the radiation field such
that
A I A in A

[H ,Ntl=-NI
r
NI=Gl tGI

The subscript I denotes the frequency WI. Show that [Ho , Ntl = O. Then show
that the perturbation H' represents a sink that diminishes the number of photons
of frequency WI exponentially with an e-folding time r .

Answers
(a)

Eigenstates take the form

The number of photons in the co; mode is ru. These wavefunctions have the
properties

Gi I· .. . ni, ) = -Jliil ... , n l - I , . . .)

Gi t I.. . . n. , ) = J1li+l1. .. . n; + I, . ..)

It follows that

Holnl ,n2" " ) = ~nwi (n i +

I
D
I n l , n2, " ' )
726 Chapter 13 Perturbation Theory

(b) The average number of photons of frequency co 1 is

(N l) = (n l, n2,· · .IN l ln l , n2, ·· .) = nl

The equation of motion for (N 1) is [recall (6.68)]

d (N l ) i A A i A I A 1
----;j( = h ([H , Nil ) = h ([H , Nil ) = -~ (Nl )

-dnl = --nl
1 /
nl =nl(O)e- t r
dt r

13.38 In Problem 13.37 we found that the Hamiltonian of an electromagnetic radi ation
field may be writtenl''

HR = Lhwjaj taj
j

Rec alling that the frequency of a photon of momentum lik is co = ck [see (2.28)] , it
follows that the operator corresponding to the total momentum of the radiation field
is

p= Llikja/aj
j

Here we are assuming that the field is contained in a large cubical box with perfectly
reflecting walls. Boundary conditions then imply a discrete sequence of wavenum-
ber vectors [k.} , (See Problem 2.37.) Consider that a charged particle bound to a
point within the radiation field vibrates along the z axis . Its Hamiltonian is

Hp = hwo (ata + D
(a) If the oscillator and the field are uncoupled from each other, what are the Hamil-
tonian and eigenstates of the composite system?
(b) Consider now that a small coupling exists between the particle and the field
whose interaction energy is proportional to the scalar product of the displace-
ment of the particle and the total momentum of the field, with coupling constant
exhwo2 2, ex
j mc where is the fine-structure constant. Assuming (13 .61) to be ap-
propriate, calculate the probability rate that as a result of this interaction, the
oscillator emits a photon of energy hwo.
Answers
(a) HPF = Hp + HF = hwo (ata + D+ 'l;hwja/aj
J

11ft) = l1ftp)I1ftF) = In; nl , n2, " ' )

HPFI~} ~ [hwo (n + D+ ~liwjnj] I~)

14Here we are omitting the infinite zero-point energy I: 1fiwi -
 13.8 Selective Perturbations in Time 727

The notation is such that n j denotes the number of photons with momentum
nkj .
(b) The interaction Hamiltonian is

H" = -cxliwo2
- - - (1a +
•
a·t) L .
Jik·za · t a· '
me2 .j2 f3 . J J J
J

where f32 == mwoln [recall (7.26)] and kjz denotes the z component of kj .
To apply (13.61) we must first calculate the matrix element of if' between the
initial state

l1fri) = In;n\ • . . . • nk. · · ·),

and the final state

(1fr/ 1= (n - 1; n\, .. . • nk + 1, .. .1,

EI = liwo (n - ~) + liwO(nk + 1) + L liwjn j
2 j#k

Here nk represents the number of photons with wavenumber wole. This choice
of (1fr I I guarantees conservation of energy in the transition, as prescribed by
the delta-function factor 8(Ei - E I) in (13.61) . Since l1fri) (as well as (1fr I I)
is an eigenvector of La a
j t l: it may be brought through the field component
of fJ'. When completed with the bra vector (1fr I I, the inner product of the field
component factors of the wavefunctions vanishes and we conclude that for the
given perturbation,

(1fr/ IH' I1fri ) = H!I =0

The hypothetical field-particle coupling does not induce a transition of the state
of the harmonic oscillator.

13.8 • SELECTIVE PERTURBATIONS IN TIME

The Adiabatic Theorem

Let us now consider the case that H' is adiabatically turned on, which means
that H' changes slowly in time (Fig. 13.13). Consequently, at any instant of time,
the Hamiltonian may be treated as constant and an approximate solution may be
obtained by regarding the Schrodinger equation as time-independent. This will be
shown below.
The slowly changing quality of fI' may be incorporated into the analysis
through a parts integration of (13.51).
728 Chapter 13 Perturbation Theory

[(t)

H'(r, t} = HI' (r)[(1)

FIGURE 13.13 Adiabatic perturbation . The rise time, w- 1, of the perturbation obeys
the inequality

to e; Wkl

where Wkl are the natural frequencies of the unperturbed system.

I
Ck(t) = :-
In
it
-00
. I I
Hkl(t')elWk/t dt' = - - -
liwkl
it
-00
a . I
Hkl(t')-,elWk/t dt'
at
(13.72)

= - -I - { Hkl(t)elWk/t
liwkl
. - it .
-00
a
elWklt - , Hkl(t ') dt'
I

at
}

where we have set Hkl(t) == IHI kd(t) . If k(t) is slowly varying, the second term
is small compared to the first and Ck (r) is well approximated by

I . (klkll}eiwklt
Ck (t ) -~ - ~ . . H'kl (t ) e lWk/t -- - (0) (0) (13.73)
tuoi] Ek - EI

Let us recall that the analysis leading to (13.49) presumes that the system is in
the stationary state Vrl at t < O. To terms of first order in the perturbation H', the
series (13.42) then appears as

Vr(r, t) = Vrl +L CkVrk

ktl

As the perturbation is slowly varying, we write Vr(r, t) :::: exp( -iwlt)fIJ(r).

Substituting this expression into the preceding and employing (13.73) gives

'""' HklfIJk
fIJ(r) = 'PI + Z:: E (0) _ E (0) (13.74)
ktl I k

This is precisely the first-order result which stationary (time-independent) pertur-

bation theory gives, (13.10). But the solution (13.10) represents an eigenfunction
of the new Hamiltonian, which in the present case is H = Ho + k (t). That is, the
wavefunction (13.74) represents the lth eigenstate of Ho + k (t) , to first order in
H'. [Note that we are writing k(t) for the operator H' evaluated at the specific
13.8 Selective Perturbations in lime 729

time t.] Since the system was originally in the lth state of Ho, we may conclude
the following. A system originally in the lth state of an unperturbed Hamiltonian
will at the end of an adiabatic perturbation be found in the lth state of the new
Hamiltonian . One says that the system "remains in the lth state."
Having established that C{J as given by (13.74) is an eigenstate of the new
Hamiltonian, it follows that expectation of energy in this state is the same as
the eigenenergy of the state. The expectation of energy to first order in the per-
turbation Hamiltonian is easily calculated with the explicit form of C{J as given by
(13.74). There results

(13.75)

This is the lth eigenenergy of the Hamiltonian Ho + if' (t) . So under an adi-
abatic perturbation, a system originally in the lth eigenstate of an unperturbed
Hamiltonian will at time t be found in the lth eigenstate of the new Hamiltonian
with the lth eigenenergy of the new Hamiltonian. With C{J/ (t) written for the time-
independent wavefunction evaluated at the time t , we may write that under an
adiabatic perturbation

C{J/ (0) -+ C{J/ (t)

(13.76a)
E/(O) -+ E/(t)

where

[Ho + if'(t)]C{J/(t) = E/(t)C{J/(t) (13.76b)

This equation is not a time-dependent equation . One first evaluates H' (r) and
then finds the solution C{J/(t) . These results constitute the adiabatic theorem, first
proved by Born and Fock in 1928. 15
As an example of the use of this theorem, consider that a particle of mass m
is in a one-dimensional box of edge length a. The width of the box is slowly in-
creased to eta over a long interval lasting t seconds, where a > 1. Let us calculate
the new wavefunction and new energy of the system given that the particle ini-
tially was in the ground state. The adiabatic theorem tells us that the new state is
the ground state of the new Hamiltonian.

C{JI ~ C{JI (r) = V~

/2 sin (JretaX)
Here, as above, we are writing C{J(t) for the time-independent wavefunction eval-
uated at the time t. The new energy is the ground state of the new Hamiltonian .

15M. Born and V. Fock , Z. Physik, 51,165 (1928) .

730 Chapter 13 Perturbation Theory

Note that in slowly expanding (as in the classical case), the particle loses energy
to the receding walls in slowing down. The amount of work absorbed by the walls
in the present example is

2
W = E1 - E(t) = -h- ( 1 - I-)
8ma 2 a2

Domain of Validity
Our conclusions regarding an adiabatic perturbation rest on neglecting the integral
term in (13.72). Let us obtain a quantitative criterion which allows this term to be
discarded. Let the perturbation be gradually turned on at t = O. If iI' changes
slowly, then af!' jat may be taken outside the integral term in (13.72) and we
obtain

It follows that the nonintegral term in (13.72) dominates provided that

(13.77)

Thus a perturbation that undergoes a small fractional change in a typical period

of the unperturbed system may be termed adiabatic.

Transition Probability
We have found that in the adiabatic limit, neglecting the integrated term in (13.72)
leads to the system remaining in the initial state of the unperturbed Hamiltonian.
What is the probability in this same limit that there is a transition out of the initial
state? Suppose, for example, that an adiabatic perturbation is turned on at t = 0
and that the unperturbed system is originally in the lth state of the unperturbed
Hamiltonian. The probability for a transition from the I to the k state is given by
ICkI 2 . With (13.72) in the adiabatic limit, we obtain
13.8 Selective Perturbations in lime 731

Over a long time interval, the last term averages to zero, whereas the sin 2 term
in the middle expression averages to !.
As we have found above, the first term
represents the probability that there is no transition out of the lth state. It follows
that the probability that there is a transition out of the initiallth state to some kth
state (k :j:. I) in the adiabatic limit is

(13.78)

Once more we find that there is vanishingly small probability of transition out of
the lth state for sufficiently slowly changing perturbation.
An instructive laboratory example of the adiabatic theorem is given by elastic
collision of molecules in a gas. During the collision, electron states are acted
upon by intermolecular forces. This collisional interaction varies roughly as the
intermolecular velocity, which, typically, is far smaller than mean atomic electron
speeds . Thus, collisional forces which act on atomic states may be viewed as being
adiabatically switched on. Consistent with this observation, one finds that whereas
final rotational and vibrational molecular states are altered in the collision, atomic
states suffer minimal change .l?

Sudden Perturbation
The next type of perturbation problem we wish to examine involves a sudden
change in a parameter in Ho. For example , suppose that the spring constant of a
simple harmonic oscillator is suddenly doubled . If the oscillator is in its ground
state before the perturbation, in what state is it after perturbation? For such prob-
lems it is presumed that the eigenstates of both Hamiltonians (i.e., before and after
perturbation) are known. This , together with the assumption of an instantaneous
change of Ho, are basic to the sudden approximation.
Let us call the Hamiltonian before the change in parameter, H, and the Hamil-
tonian after the change in parameter, H', so that

HC{Jn = EnC{Jn t < 0

H C{Jn =
~ I I E I I
t ~0
n C{Jn

(Note that Ii' is now a total Hamiltonian.) Initially, the system is in an eigenstate
of Ii.
C{Jt

1fr' (r, 0) = C{Jt (r)

At later times the system is in a superposition state of H'.

1fr' (r, t) = L cnC{Jn' e-iwn't
n

16For further discussion, see R. D. Levine and R. B. Bernstein. Molecular Reaction Dynamics and
Chemical Reactivity, Oxford, New York, 1987.
732 Chapter 13 Perturbation Theory

Equating this function to its initial value at t = 0 gives

IfJI = LCnlfJn '

n

so that

Cs = (1fJs'11fJ1)
The probability that the sudden change from iI to iI ' causes a transition from the
lth state of iI to the sth state of iI ' is

(13.79)

A criterion for the validity of this approximation scheme may be drawn from
the preceding relation. As described above, let the sudden change in Hamiltonian
occur at t = O. For the relation (13.79) to have meaning it is necessary that the
initial wavefunction IfJI(r) : (1) maintain its form at t = 0+ and (2) lie in the
Hilbert space spanned by the eigenstates of the new Hamiltonian . Thus, for a
particle confined to a rigid-walled box, we may conclude that the approximation
is inappropriate to sudden compression of the box but is appropriate to sudden
expansion of the box.
To further demonstrate this approximation, let us apply the formalism to the
latter case. Thus we consider a particle of mass m confined to a one-dimensional
box which suffers a sudden expansion. With no loss in generality we take the box
to be of unit width (a = 1). The particle is in the energy eigenstate IfJI at t < O.
At t = 0 the box is suddenly expanded to the edge length a > 1 (recall Problem
5.11). With the preceding formalism we may establish the somewhat surprising
result that although no work is done on the particle in the expansion, energy is not
necessarily conserved.
The new Hamiltonian iI ' (after expansion) is given by (4.23) with a replaced
by a . New basis wavefunctions are the eigenfunctions

, (2 (S7CX)
IfJs =V~sin ~

The probability for transition is given by

jP;; = (lfJillfJ;) = Ja 1 1
sin Irr,c sin C: X
) dx (13.80)

= _1_ {Sin7C[I- (sla)] _ sin7C[1 + (sla)]}

.;a 7C[I- (sla)] 7C[1 + (sla)]
Note that the integral in the preceding expression goes over (0, 1) and not (0, a)
because IfJI(X) is zero over (1, a) .
The transition which conserves energy occurs for the s value s = al , With
A == li 27C 2/ 2m , we obtain
Problems 733

Es'I _
s=s
= A 2ClS
2
I
s=s
_ = Al 2 = Ei (13.81)

At this value of s, (13.80) gives

1
Prs=-<1 (13.81a)
a
Thus , as stated , it is not certain that the transition conserves energy. However, it is
simply shown that expectation of energy is conserved. To show this, we note that
at t = 0+ the wavefunction for the system at hand is given by

CPt = ~t
CPt = 0

It follows that

(13.82)

Note that the third term is defined over (0, 1) and is the value of (E) at t < O. The
last term is defined over (0, Cl) and is appropriate to t = 0+. We conclude that
(E) is conserved in the expansion. Further developing (13.82) gives

s, = L(cptlH'I~k)(~klcpt)
k
(13.83)
s, = LPtkEk
k

or, equivalently,

2 1 ~ 2
I = 2 L.Ptk k (13.84)
Cl k=!

with Ptk given by (13.80). This result is valid for arbitrary a > 1, i.e., for irrational
as well as rational values. In particular, note that if k/ a =1= I for all k, then no direct
transition conserves energy. However, with (13.83), expectation of energy is still
conserved.
Expressions obtained above for the transition probability in various limiting
cases are listed in Table 13.2.

PROBLEMS

13.39 A neutron in a rigid spherical well of radius a = 0.1 Ais in the ground state. The
radius of the well is slowly decreased to 0.9a.
734 Chapter 13 Perturbation Theory

TABLE 13.2 Transition probabilities for time-dependent perturbations

1. fI' is separable. fI l(r, t) = JH[ /(r)f(t)

Pik = 'JH[;i Ii:

e e
iaw'
f(tl)dtf

Harmonic perturbation, f = 2coswt, DC perturbation (w = 0, f = 1)

turned on at t = 0 : turned on at t = 0 :

Long -time or high-frequency limit : Short-time or low-frequency limit:

27rtlJH[~lI2
Plk = fj} 8(Wkl - w)

21fIJH[ ~l12
Wkl = fj} 8(Wkl - w)

Probability for transition to a band centered at Ek with 91JH[ I I slowly varying in energy:

- Zn t I 2
Plk = ---n9(Ek)lJH[kll

21f I 2
lihl = T9(Ek)lJH[kll
2. fI l = fI' (r, t) is slowly changing
Probability for trans ition out of the initial I state :

Adiabat ic theorem:

(k =1= l)

3. fI changes suddenly to fI'

Eigenstates of both fI and fIl are known .

(a) What is the energy and wavefunction of the neutron after the decrease in the
well radius? (For normalization, see Table 10.2.)
(b) How much work (in eV) is performed during the compression of the well?
13.40 A collection of No = 1013 independent electrons have spins polarized parallel to a
uniform magnetic field that points in the z direction, of magnitude 20' A perturba-
tion field is applied in the x direction of magnitude

(t 2: 0)
Problems 735

(a) Obtain a criterion involving r which ensures that the perturbation is adiabatic .
(b) Given that nr = 102 and .:80 = 104 gauss, estimate the number of electrons
t>N that are thrown out of the ground spin state at t = lOr.

Answers
(a) r :» n- I = mcle.:8o
(b) A rough estimate is obtained from (13.78).
2 2
1O- 3 JLb .:80 1 Ie-t iT 1
t>N ~ 2No lin ~
1

13.41 A one-dimensional harmonic oscillator has its spring constant suddenly reduced by
!.
a factor of The oscillator is initially in its ground state. Show that the probability
that the oscillator remains in the ground state is P = 0.986.
13.42 A particle of mass m in a one-dimensional box of width a is in the third excited state
(i.e., n = 3). The width of the box is suddenly doubled. What is the probability that
the particle drops to the ground state?
13.43 A one-dimensional harmonic oscillator in the ground state is acted upon by a uni-
form electric field

switched on at t = -00. The field is parallel to the axis of the oscillator. What is the
probability that the oscillator suffers a transition to its first excited state at t = +00
in the limits : (a) W()r » I, (b) wOr ~ l. For case (b), are any other transitions
possible? [Hint: See Problem 13.27.]
13.44 Radioactive tritium, H3, decays to light helium, (He3+), with the emission of an
electron . (This electron quickly leaves the atoms and may be ignored in the follow-
ing calculation .) The effect of the f3 decay is to change the nuclear charge at t = 0
without effecting any change in the orbital electron. If the atom is initially in the
ground state, what is the probability that the He+ ion is left in the ground state after
the decay?
13.45 A hydrogen atom in the ground state is placed in a uniform electric field in the z
direction,

8 = 80e- t1T
which is turned on at t = O. What is the probability that the atom is excited to the
2P state at t» r?
13.46 The perturbation

is applied to a particle of mass m in a one-dimensional box of width a at t = -00.

At this time the particle is in the ground state. If lilr « E I, in what state is it most
probable that the particle will be at t = +oo?
736 Chapter 13 Perturbation Theory

13.47 An electron in a one-dimensional potential well

V= ~Kx2
2
is immersed in a constant, uniform electric field of magnitude 8 which points in the
x direction. The corresponding perturbation to the Hamiltonian is

H' = e8x
(a) Find the exact eigenenergies of the total Hamiltonian,

jJ2 1 2
+ -Kx + e8x
A

H = -
2m 2
(see Problem 7.16). Discu ss your findings with respect to the corresponding
classical motion .
(b) Show that the first-order corrections to the energy vanish . Then calculate the
second-order corrections. Show that these agree with your answer to part (a),
so that the second-order corrections give the complete solution for this problem.
Answers
(a) Setting
e8
X=-
K
together with the transformation of variables

x' =x + X
permits Hto be rewritten
p2 1
= 2m + 2K (x 2 + 2xx)
A

p2 1 /2 1 2
= 2m + 2Kx - 2Kx
= Ho-IE
Since

is constant, the eigenenergies of H are simply

All levels are equally depressed by the constant energy IE. The new wavefunc-
tions are

f{Jn = f{Jn(x /) = f{Jn(x + X)

The center of symmetry of these wavefunctions is at x = - X.
Problems 737

In the corresponding classical problem, the potential of the electron in the

presence of the uniform electric field is

This parabola is congruent to the original potential Kx2 /2 . The new equilibrium
at x = - X occurs where the electric force is balanced by the spring force. This
new potential minimum is lower than the original minimum by the amount IE.
(Work must be done to move the electron, quasi-statically, from x = - X to
x = 0.) The classical analog of the quantum mechanical problem considered
above involves an electric field that is very slowly (adiabatically) turned on. If
the energy of the oscillator initially is E(O), what is it after the electric field is
established? In classical mechanics, for such adiabatically changing harmonic
motion, the ratio A /w is constant (it is an adiabat ic invariant) . The amplitude
of oscillation is A . Since the new potential well is congruent to the old well, w
is the same for both wells. This means that the amplitude of oscillation must
also be preserved (during the adiabatic change). Owing again to congruency of
the parabolas, this is ensured (only) if the distance between E(O) and the bottom
of the new well is the same as the distance between E(O) and the bottom of the
original well. That is, if

E(O) = E + IE
It follows that the new energy of the oscillator

E = E(O)-IE
is depressed from the initial value, E(O), by the amount IE. This is identical to
the quantum mechanical result (Fig. 13.14).

IE ;:! KX2
FIGURE 13.14 The classical adiabatic change from a parabolic potential, V , to a con-
gruent (wo = wb) parabolic potential, V' . with a new center of symmetry, preserves am-
plitude A . We see that the amplitude of oscillation is preserved, A = A', provided that
E = E(O) -IE. This result for the new energy of oscillation, E, is identical to the quantum
mechanical result. (See Problem 13.47.)
738 Chapter 13 Perturbation Theory

(b) The perturbation Hamiltonian may be rewritten

H, = e8
M (a' t ' = 8"
+ a) (a t+'a)
v2f3

f32 =mli.!(),
Ii
8
,
=--
e8
v'2f3
It follows immediately that

(nIH' ln) =0
so there is no first-order correction to En (0) . To calculate the second-order cor-
rections, we must evaluate the off-diagonal matrix elements of H'.

(niHil) = 8 ' (nla t + all)

= 8'(JJ+lon,I+1 + Jlon ,l-l)
Substituting this expression into (13.14) gives the desired result .

13.48 A system with discrete energy states and Hamiltonian Ho has the density operator
Po, which is diagonal. Furthermore, [PO , Hol = 0, so Po is constant in time. Show
that after a perturbation Ii' is applied, the diagonal elements of Pchange according
to the Pauli equation

The transition rates Wnk are given by (13.63).

The density operator was discussed in Section 1l.1l.

Answer
For a short-time interval after the perturbation H' is applied, we may expand P to
obtain

p) 2p)
,
p(M) ,
- p(O) = (a- ilt + (a- 2 -(ilt)2
- + ...
at 0 at 0 2
Using the equation of motion (11.122) for p permits the last equation to be written ,
with p(O) = Po,

ap
at = tR~o
[P(ilt)-PO]
ilt
{I,
= tR~o iii [H , plo-
ilt"
2li 2 [H, [H, p]]o
}

The diagonal form of Po leads to the following properties:

(nl[Ho + H', polin) = 0

(nl[Ho, [H' , Po]] In) = 0
 13.9 Atom-Radiation Interaction 739

Forming the diagonal elements of ap;at then gives

apnn M
-a- =- .00]] In)
A A

lim - 2 (n l[ H ' , [H',

t ~t--->O
2h

The diagonal element on the right-hand side redu ces to

A, A, "\' , 2 , 2
(n l[ H ,[H ,.Bollin) = 2 L..-{IHnkl Pnn(O) -IHnkl Pkk(O)}
k

which when substituted into the preceding equation gives the desired result .

13.9 • ATOM-RADIATION INTERACTION

Our first description of the interaction between an atom and a radiation field was
given in terms of Einstein's derivation of the Planck radiation law (Section 13.7).
We now wish to present a Hamiltonian formulation of this problem. The Hamilto-
nian of an electron in an electromagnetic field with vector potential A (previously
introduced in Section 10.4) and electric potential V (r) is given by

H = 2m
I [e
p- ~A(r,t) ]2 + VCr) (13.85)

In the full quantum electrodynamic analysis of this problem, electrodynamic

fields are quantized. The present analysis is termed a "semiclassical" description
in that the vector potential A in (13.85) is taken as a classical field. 17
Expanding the kinetic energy term in (13.85) gives

H = H(O) +k +H"
A2
e
H(O) == L + V ; A ,

H =--[p.A+A.p]
A A

(13.86)
2m 2mc
2
H"=_e_ A 2
2mc 2

We neglect HI! and consider H' a perturbation term to the unperturbed atomic
Hamiltonian H(O) .
The matrix element of k between initial In} and final In'} states of H(O) is
given by

(n'IH'ln) = __e_
2mc
f 1fr;,(p. A + A· P)1frn dr (13.87)

Note that the index n denotes a quantum state, not the principal quantum number.

17For a more complete quantum analysis , see E. G. Hams, A Pedestrian Approach to Quantum Field
Theory, Wiley, New York, 1972; A. S. Davydov, Quantum Mechanics , 2d ed., Pergamon, Elmsford,
N.Y., 1973.
740 Chapter 13 Perturbation Theory

The first term in this integral may be written

P • A1/fn = -;-liI v . (A1/fn) = -;-IiI [1/fn (V • A) + (A • V)1/fn] (13.88)

The gradient and curl qualities of the potentials which enter electrodynamics per-
mit certain gauge conditions to be imposed . In the so-called Coulomb gauge one
sets

V·A=O

so that (13.88) reduces to

Ii
A

P • A1/fn = -;-A
I
• V1/fn

Thus we may write (13.87) in the equivalent forms

(n'IH'ln) = _-!...-
mc
f 1/f;,A . p1/fn dr (13.89a)

(n "IH In) = - -
e
mc
f * A1/fn dr
1/fn'P'
A (13.89b)

A revealing insight into this analysis is gained if one associates the vector poten-
tial A with the wavefunction of a photon of energy luo. This permits the preceding
two matrix elements to be symbolically written

(emission) (l3.90a)

(absorption) (13.90b)

Thus, (l3.90a) corresponds to the case where a photon is emitted by the system,
whereas (13.90b) corresponds to the case of the absorption of a photon. Since
these two integrals are equal, we conclude that probabilities of emission and ab-
sorption, at the same frequency, are equal. This is an example of the principle of
microscopic reversibility.
To obtain a more explicit representation of the matrix elements (13.89), we
consider the atom in interaction with a plane electromagnetic wave whose vector
potential is of the form

A(r, t) = aAo cos(k . r - wt) (13.91)

where a is a unit polarization vector (lal 2 = 1) which is normal to the propagation

vector k. We wish to construct the amplitude Ao so that the corresponding wave
carries one photon per unit volume. The time-averaged energy density carried in
13.9 Atom-Radiation Interaction 741

a plane electromagnetic wave is

(13.92)

With

B=VxA (13.93)

and given the form (13.91) , we find that

(13.94)

For one photon per unit volume we set

(13.95)

which, with co = ck, gives

(13.96)

Henceforth we set the volume V = 1.

In what follows , we will employ Fermi's golden rule (13.64). In formulating
this relation it was found that for the time-dependent perturbation, cos on, the
exp( - i wt) term was responsible for resonant absorption and the exp(+i wt) was
responsible for decay. For consistent application of Fermi's rule, these observa-
tions must be incorporated into the present analysis. Thus, we first write (13.91)
in the form

(13 .97)

which permits the identification

Ao ±i(k·r-wt) (13.98)
A±= -ae
2
where A+ corresponds to photon absorption and A_ to photon emission. So we
may write

2rrli) 1/2 ±i(k·r-wt)

A±=c ( - - ae (13.99)
w

With

(13.100)
742 Chapter 13 Perturbation Theory

and (13.89) we obtain

(13.101)

The Dipole Approximation

For typical atomic transitions A » ao, where the Bohr radius ao is a length char-
acteristic of atomic dimensions. Thus, over the domain of integration in the pre-
ceding matrix element, k • r « 1, and we may write

ei k • r = 1 + ik • r + ...
In the dipole approximation, one sets eik • r = 1. There results

(n'IIHI±ln) = - m
e (27f1i)1/2
---;;;- (n'la. pin) (13.102)

With the aid of the commutator relation (5.62) this matrix element may be trans-
formed to one only involving r (see Problem 13.60):

A im (0) A

P = T[H ,r] (13.103)

It follows that

(13.104)

Recalling that In'} and In} are eigenfunctions of the unperturbed Hamiltonian
fJ(O), we obtain

(n'lpln) = - im (En (0) - E n/O») (n'lrln)

Ii (13.105)
(n'lpln) = - i mw (n' lr ln }

Note that we are writing w for W nn' > O. With these relations at hand, (13.102)
may be rewritten more explicitly as

(n'IIHl+ln, k) = (n', kIIHI_ln)

= i e(27fliw ) 1/2(n' la . rln} (13.106)

Here we have specified that in the absorption process , the initial state contains a
photon of wavevector k, whereas in the emission process, the final state contains a
photon of wavevector k. Again as in (13.90), we find that the related probabilities
of these two processes are equal. Spontaneous decay may be associated with the
matrix element (n', kIIHI_ln).
13.9 Atom-Radiation Interaction 743

Spontaneous Decay
As described previously, the Einstein A coefficient represents the probability rate
for spontaneous decay. [See, for example, (13.65).] Ordinarily, one would suspect
that such spontaneous decay occurs in the absence of any perturbation. However,
in our earlier description of hydrogen (Section 10.6) we found atomic states to
be stationary. So in reality, spontaneous decay must have a triggering mechanism.
What is this mechanism?
The answer to this question stems from the observation that (as described in
Problem 13.37) an electromagnetic field may be represented as a collection of har-
monic oscillators. We have found previously that a harmonic oscillator always has
a residual energy which is called its "zero-point energy." In like manner, no region
of space is ever free of electromagnetic energy. It is such vacuum fluctuations of
electrodynamic fields which are responsible for spontaneous decay.

Golden Rule Revisited

In describing spontaneous decay we concentrate on the emission matrix element
(n', kIIHLln), and consider that the emitted photon lies in the differential of
solid angle dQ . The corresponding transition probability rate is given by Fermi's
golden rule (13.64), which, in the present case, assumes the form

,,27f 2-
dwnn, = L...J TilHIn1nl g(E)dQ (13.107)
a,

Here a, denotes possible photon polarization and geE) dQ is the density of states
of photons emitted in the solid angle dQ . This value of geE) may be obtained
from the value of 9( v) given in Problem 2.37 by setting

V9(V)dV=V9(W)dW=2[V f geE)dQ]dE (13.108)

The factor 2 in this equality accounts for photon polarizations. Setting the volume
V = 1, we obtain

(13.109)

Combining these results gives the following probability rate for spontaneous de-
cay:

(13.110)

The summation over polarizations in (13.110) is performed as follows. First, we

note that

(n', kla- rln) = a- (n', klr]») (13.111a)

744 Chapter 13 Perturbation Theory

We recall that in the dipole approximation the presence of k in the preceding

matrix element is cosmetic. It merely reminds us that the element is relevant to
spontaneous decay and includes a photon of wavevector k in the final state. It
follows that the only vector in the matrix element on the right side of (13.111a) is
r . Consequently, this matrix element is likewise in the direction of r and we may
conclude that the entire matrix element on the left side of (13.111a) is proportional
to 3 • r . With this observation, the sum in (13.110) is evaluated as depicted in
Fig. 13.15. The propagation vector of the emitted photon, k, makes an angle e
with the radius vector r, which is held fixed and taken as the polar axis. The unit
polarization vectors 31,32 are normal to each other as well as to the propagation
vector k. To facilitate calculation, the vectors k, 31, and r are taken to lie in the
same plane (see Fig. 13.15). It follows that the vector 32 points out of the plane
of the paper, so that 31 • r = sine, 32 • r = 0 (r is a unit vector). Thus we
obtain

L I(n ' , k13; • rln)12 = 131 • in , klrln)1 2 + 132 • (n', klrln)1 2

a,

Inserting this result into (13.110) gives the following probability rate for sponta-
neous decay:
2 3
e W I 2 . 2
dwnn, = ~I(n ,klrln)1 sm edQ (13.111b)
2rr 1/,<;
So we reach the conclusion that the differential transition probability rate dwnn,
depends only on the angle e between rand k. Consequently, the subsequent inte-
gration over dQ is independent of the azimuthal angle, and with r still held fixed,

al·r=sin8
k a2'r= 0

a2 points out of
plane of (r, k)

FIGURE 13.15 Configuration for summation over 3; in (13.110). The propagation vec-
tor k defines the direction of the differential of solid angle dQ.
13.9 Atom-Radiation Interaction 745

we obtain

(13.112)

Note that we are writing

l(n'lrln)1 2 = l(n'lxln)1 2 + l(n'IYln)1 2 + l(n'lzln)1 2

The coefficient Ann' represents the probable rate of transition for spontaneous de-
cay and may be identified with the Einstein A coefficient in (13.65) . To obtain the
corresponding expression for radiated power, one multiplies Ann' by lua. There
results

(13.113)

which agrees with our previous result (10.139).

The Einstein B coefficient may be found from our previously derived relation
(see Problem 13.52):

which, with (13.112), returns our earlier finding (13.70).

It should be kept in mind that the preceding results are relevant to weak fields
and the assumption that A » aQ .
To apply the above formulas to an atom with Z electrons, one makes the re-
placement

z
r nn , = (n'lrln) -+ (n'ILriln) (13.114)
i=!

in the preceding expressions for A and P. The sum in (13.114) runs over all
the electrons in the atom. For such cases er nn , represents the total dipole matrix
element of the atom.
The total probable rate of spontaneous decay of an atom in the nth energy state
is given by

(13.115)

The summation runs over all states of lower energy than En. The corresponding
mean lifetime of the nth excited state is then given by

1
rn = - (13.116)
An
746 Chapter 13 Perturbation Theory

OscillatorStrengths
An important parameter in radiation analysis is the oscill ator strength, defined as
the dimensionless form

(13.117)

It follows that
2e 2(j}
Ann' = --3- inn' (13.118)
me
2e 2 1Uu3
Pnn, = - me
- 3 - i nn' (13.119)

Oscillator strengths obey the so-called Thomas-Reiche-Kuhn sum rule ,

Linn' = 1 (13.120)
n'

from which, together with (13.117), it may be concluded that inn' < 1.
The sum rule (13.120) is simply derived with the aid of the basic commutator
relation (5.61), according to which , in one dimension, one obtains

(nl[x , Pxlln) = iii (13.121)

Equivalently, we may write (see Problem 11.1)

L[(nlxln')(n'lpxln) - (nlpxln')(n'lxln)] = iii (13.122)

n'

With (13.105) we obtain

L[imwn'nl(n'lxln)12 - imwnn'l(n 'lxln)1 2l = iii

n'

'\' 2mwn'n , 2
L..J -li-I(n Ixln)1 = 1
n'

The same result follows with x replaced by y or z. There results

'\' 2mw , 2 '\'

L..J ~I(n Irln)1 = L..J inn' = 1 (13.123)
n' n'

For an atom with Z electrons, with (13.114) one finds

Linn' = Z (13.124)
n'
Problems 747

Values of the oscillator strengths of hydrogen corresponding to the first four n P -

1S transitions are listed below. IS

n 2 345
fnP- IS 0.416 0.079 0.029 0.014

We conclude this section with an estimate of the lifetime of the excited states
of one-electron atorns.l? With Irnn,1 ::::: ao, we rewrite Ann' (13.112) as

(13.125)

Introducing the effective nuclear charge Z,

Ze2
Iuo ==- (13.126)
ao
permits the preceding formula to be written

4 3 -2
Ann'::::: 3"a wZ (13.127)

where a = e 2I tic is the fine-structure constant. With Z ::::: 1, and Iuo ::::: IR =
a 2m e2 / 2, we find

a 5 me 2 a 5e
A , ~ - -- - -­ (13.128)
nn - !i 2 - lXc

In this expression, X.C = !il me is the Compton wavelength. Inverting Ann' gives
the lifetime/"

T ::::: a- 5
(2~c) ::: 10- 10 s = 0.1 ns (13.129)

PROBLEMS

13.49 The radioactive isotope C II decays through positron emission to B 11- . With the
same assumptions holding as described in Problem 13.44, estimate the probability

18R. Loudon, The Quantum Theory of Light, Clarendon Press, Oxford , 1973.
19More detailed discussions of these topics may be found in H. A. Bethe and E. E. Salpeter, Quantum
Mechanics ofOne- and Two-Electron Atoms, Plenum , New York, 1977; F. Constantinescu and E. Mag-
yari , Problems in Quantum Mechanics, Pergamon Press, Elmsford , N.Y., 1971; S. Flugge , Practical
Quantum Mechanics, Springer, New York, 1974.
20Nanosecond (10- 9 ), picosecond (10- 12 ), and femtosecond (10- 15 ) measurements are presently
commonplace in many laboratories. A review of these topic s is given by P. W. Smith and A. M.
Weiner, IEEE Circuits and Devices, 4, 3 (May 1988). Note the powers of lOin physics: speed of light,
3 x 10 10 crn/s; decay time of an atom , 10- 10 s; age of the universe , 10 10 yr.
748 Chapter 13 Perturbation Theory

that B 11- is born in the ground state. (Hint: For your estimate, consult Problem
12.28 and pay attention to Z dependence of wavefunctions.)

Answer
As discussed in Problem 12.28, two-electron , ground-state wavefunctions (with
Hso = 0) are given by

1
If!z(q, rz) = - 3 exp
rra
[-(q + z)]
a
r

a::-
ao
Z
with normalization

The transition probability to the ground state of B 11- is given by

where Z :: Zc + ZB . The preceding integral may be rewritten

There results

.JP = (4Z; zZc)

3

With ZB = 5, Zc = 6, Z = 11, we obtain P = 0.951.

13.50 (a) The density matrix (in a representation where 1;2 and Sz are diagonal) describ-
ing a beam of spinning electrons has the value

at t < O. What are the values of (Sz), (Sx ), and (Sy ) for an electron in the
beam?
(b) The beam interacts with a field which is turned on at t = O. The corresponding
interaction Hamiltonian has the matrix

,
H= 1iwo 1
(0 ~)
Problems 749

where Wo is a characteristic frequency. Estimate the value of the matrix p at the

time 6.t, where 0 < !1t « Wo-1 .
(c) What is the value of (Sy) at this value of time !1t? [Hint: Recall (11.122).]

Answers
(a) (Sz) = (Sx ) = n 12, (Sy) = O.
(b) Set

,
p(!1t) = p(O), (aat
p
+ - )
0
!1t + ...
,
= p(O) + inI [H" , p]o!1t + ...
Employing the given value of p(O) gives

p(!1t) = (0~5 °en - iWO !1t (~ -1)

o
= (1 a)0
a*

where

a == 0.5 + iwO!1t
(c) Att =!1t

(Sy) "
= r-ss, = Tr [,in
P"2 (0 1 ~1) ]
in
=2 Tr
[( I
a* ~) G ~I)]
in - a *) = -nwo!1t
= -(a
2
Thus , the perturbation causes a change in (Sy).

13.51 A one-dimensional harmonic oscillator of charge-to-mass ratio elm, and spring

constant K oscillates parallel to the x axis and is in its second excited state at t < 0,
with energy

An oscillating, uniform electric field

8(t) = 28ocoswot
is turned on at t = 0, parallel to the motion of the oscillator.
(a) What is the new Hamiltonian of the oscillator at t > O?
(b) What are the matrix elements !HI ' 2n for this system? You may leave your answer
in terms of f3 = Jmwoln.
(c) What are the probabilities P2n that the oscillator undergoes a transition to the
nth state at the end of t < wi) 1 seconds? [Hint: Use harmonic perturbation
theory and look at the short-time limit.]
750 Chapter 13 Perturbation Theory

(d) Using your answer to part (c), offer a technique for 'pumping' a harmonic os-
cillator to higher states.
13.52 In our discussion of Planck's radiation law in Section 13.7, an expression for AlB
was obtained as a function of frequency, v. What is the corresponding expression
for AlBin terms of angular frequency, w? [Hint : See Problem 2.43.]
13.53 Suppose «PI and «P2 are two normalized eigenfunctions of an operator A with the
same eigenvalue. Construct two new normalized , orthogonal functions, 0/1 and 0/2,
which are linear combinations of «PI and «P2. Offer a geometrical description of
construction in the appropriate Hilbert space.

Answer (partial)
Let

0/1 = «PI,
and

f «pi «P2 dr = K 1= 0

Orthogonality of 0/1 and 0/2 gives

ex+f3K=O

whereas normalization of 0/2 gives

There results
K
f3K = -ex = -r====:::;;:2
)1 -IKI
which determines 0/2 , This construction gives the essentials of the Schmidt orthog-
onalization procedure.

13.54 (a) Show that the time-dependent Schrodinger equation may be written [compare
with (11.144)]

o/(t)
IJot
= 0/(0) + iii H(t')o/(t ') dt'

(b) Show that for small H the preceding equation gives the series (Neumann-
Liouville expansion) with t > i', etc.

1 fot .1 fot dt' fottdt" H(t')H(t") + . . . ] 0/(0)

o/(t) = 1+-;- dt' H(t') + - '-2
[ Iii 0 (Iii) 0 0

(c) Show that the nth-order term in the preceding series may be written

1 1 fot
0/(n) (t) = --.- ... fot dtl " ~ ~ ~
· d tn T [H (t l )· ·· H(tn)]o/(O)
n! (11i)n 0 0
Problems 751

where the time-ordering operator

for la > tb
for tb > ta

13.55 A particle of mass m is confined to a one-dimensional partitioned box with walls

at (-L /2 , 0, L/2) . The Hamiltonian for this configuration is labeled Ho. See
Fig. l3.l6a. One of two perturbations, H~ b: is applied as shown in Fig. 13.l6b, c.
The total Hamiltonians for the particle are .
A A A I
Ha,b = HO + Ha,b

A
Vin HO

Wall
.> Wall

-L/2 L /2 x
(a)

V Vin H;

l.~.,/·.vo
I
-0 /2
I
a{2 x
-L/2 L/2
(b)

V
Vin H;

-L{2 L{2 x

(e)

FIGURE 13.16 Potential configurations for Problem 13.55.

752 Chapter 13 Perturbation Theory

Let the perturbation potentials have magnitude Vo = 10-4 EG, where EG is the
ground state of Ho, and let the normalized ground states of Ho be written [r) and
If} corresponding to the particle trapped in the right and left boxes , respectively .
(a) What are the coordinate representations of Ir} and If)?
(b) What is the ground-state eigenenergy E G of Ho? What is the order of the de-
generacy of EG?
(c) What is (fIHolr) ?
(d) What are the values (zero or nonzero) of the commutators [Ho ,lP], [Ha, Jp],
[Hb , JP)? Here we have written JP for the parity operator (6.90).
(e) Construct two simultaneous eigenstates of Ho and JP in terms of If} and [r) .
Label these states IS} (symmetric) and IA) (antisymmetric). What are HoIS},
HoIA), IP'IS}, JPIA}?
(I) Using degenerate perturbation theory, obtain the splitting of EG due to H~ and
H~, respectively. Do both H~ and H~ remove the degeneracy of EG? Explain
your answer. Call the new energies E+, E_ . What effect does the nonsplitting
perturbation have on the ground-state energy?
(g) The trapped particle is an electron and 2a = L = 10 A. What are E+ and E_
(in eV)?
(h) Can you suggest an alternative way to evaluate E± which does not employ
degenerate perturbation theory?
Answers (partial)
(a) (fix) = 0 on (0, L/2) and (rlx) = 0 on (-L/2, 0).
(e) IS} = !fUl} + Ir}), IA) = !f[ll} -Ir})

JPIS} = IS}, JPIA} = -IA}

HoIS} = EGIS}, HoIA} = EGIA}
(I) We must examine the secular equation

(fIH~,blf) - E' (f I H~,b lr ) 1_

1
(rIH~,blf) (rIH~,blr) - E' - 0
with

8a == (lIH~lf) = (rIH~lr)
8b == (lIH~lf) = -(rIH~lr)
t:.8 a == (rIH~lf) = (fIH~lr) = 0
t:.8b == (r I H~ l f ) = (rIH~lf) = 0

We find that H~ splits EG but H~ does not. The explanation of this result is
that H~ maintains the symmetry of Ho . That is, for both Ho and H~ , there is no
difference between right and left boxes. New energies (for Hb) are given by

E+ = EG +8b
E_ = EG - 8b
Problems 753

13.56 An approximate fonn of the Hamiltonian of a positronium atom in the 1S state

immersed in a weak magnetic field 13 is given by

H = A81 • 82 + Q • (81 - 82) == Ho + "AH'

e13
Q=-
J-tc

where H' is the magnetic field tenn. The subscripts 1 and 2 denote electron and
positron, respectively, A is a constant, and J-t is reduced mass.
(a) Show that the coupled spin states ~s and ~A relevant to two spin-! particles are
eigenstates of Ho . What is the ground-state energy of Ho and which eigenstate
does this correspond to? [See Table 11.3 and recall (11.111).]
(b) Employing ~S and ~A as basis states, use first-order perturbation theory to ob-
tain eigenvalues and eigenstates of H. Assume that 13 is aligned with the Z
axis.
13.57 In Section 11.12 we encountered an approximation scheme centered about the in-
teraction picture . Again consider the Hamiltonian given in Problem 11.102, which
we now write in more symbolic form,

Let the system be in the initial state

= In)
11/1(0))
Holn) = En In)
where , we recall Ho is appropriate to an harmonic oscillator with natural frequency
wO°
(a) Calculate the wavefunction 11/11 (t)) to first order in "A.
(b) For the problem at hand, show that
2
l(mI1/l(t))1 = I (m11/l1 (t))1 2
(c) Again, to O("A), obtain an integral expression for the time-dependent transition
probabilities Pn -'> m .
(d) In the event that V is time-independent, show that your answer to part (c) re-
duces to (13.55) corresponding to the DC perturbation, w = o.
Answers
(a)

l'iwnn = Em - En = l'iwo(m - n)

(b) First, note that

11/I(t)) = le-iHot /n1/l1 «»

The desired equality follows since Holm) = Em 1m).
754 Chapter 13 Perturbation Theory

(c) With the result (b), we write

Pn ...... m = l(ml1fr/(t» )1 2
where , for m =1= n ,

(m l1fr/ (1») = ado

). r • elWmnl
f A

(m lV (t')ln) dt'

(d) In the event that V is time-independent, integration of the preceding finding

gives

mn
(ml1fr/(t») = V
fiw mn
(1- i Wmn ' )
Vmn =).(mIVln)
which yields

2 Vmn 12 · 2 w mn t
Pn......m = - - sm - -
1 fiw mn 2
This result is seen to agree with (13.55) for the DC perturbation, W = O.
13.58 Establish the following relations for the coefficients of the Fourier expansion
(13.33) of the real potential function V(x) .
(a) V; = V-no
(b) If V(x) is even, then

(c) If V(x) is odd, then

(d) For V( x) (even, odd) and with period 2a , one writes

00 an cos - - nJrx)
V(x) (even) = ' " a
odd L..
n=! ( . nit x
bnsm--
a
Show that (with ao = 0)

bn = 2iVn = -2iV-n
Answers (partial)
(a) As replacing n with -n in the series (13.33) does not change the sum, we may
set

L Vnei2nn x/a = L V_ne-i2nnx /a

Vn Vn
Problems 755

For V(x) real, V(x) = V (x)" . With the preceding we then obtain
L V;e-i2rrnx /a = L V_ne-i2rrnx /a
Vn Vn

whence V; = V-no
(b) If V (x) is even, then
L Vnei2rrnx/a = L Vne-i2rrnx/a = L V_nei2rrnx /a
Vn Vn Vn

whence Vn = V-n o
(c) If V(x) is odd, then

L Vnei2rrnx /a = - L Vne- i2rrnx/a = - L V_nei2rrnx/a

Vn Vn Vn

'whence Vn = - V-n .
13.59 (a) Evaluate the average dipole moment of hydrogen for the 3P --+ IS transition
from the corresponding value of oscillator strength given in the text. Work in
cgs units and state the dimensions of your answer.
(b) Compare your answer with the classical estimate , d ~ eao.
13.60 Establish the commutator relation (13.103).

Answer
Since r commutes with V(r) , the relation reduces to

With reference to Problem 5.45, the preceding is rewritten

p= ;/i IPlp, r] + [p, r]p}

where j is the identity operator,

which establishes the said relation.

13.61 A particle of mass m is in a three-dimensional, rigid-walled cubical box of edge
length a. Edges of the box are aligned with the Cartesian axes, with one comer of
the box at the origin.
(a) Write down the normalized ground state for this configuration.
(b) The face of the cube at x = a is suddenly displaced to x = 2a. Obtain an
expression for the probability that the particle remains in the ground state.
(c) Given that the particle is an electron and that a = 2 A, what is the numerical
value of this probability ?
756 Chapter 13 Perturbation Theory

13.62 A particle of mass m is confined to the interior of a rigid-walled spherical cavity of

radius a. The particle is in the ground state . At t = 0 the radius of the sphere begins
to expand according to

where r » mao2 //i. What are the wavefunction and energy of the particle at t =
r /2?
13.63 At a given instant of time an harmonic oscillator undergoes a sudden change in
spring con stant from K to K' . Show that for energy to be conserved in the accom-
panying transition, J K/K' must be the ratio of two odd numbers.

13.64 The same cond itions as in Problem 13.26 apply . However. now the periodic poten-
tial is as shown in Fig. 13.17.

x
• • •

FIGURE 13.17 See Problem 13.64.

If Vo = 1.5 eV and a = 2.3 A, at what energy (eV) will photons incident on the
crystal cause it to conduct? [H int: Before you start , decide on the period of V(x) .]

Answer
You should obtain

4Vo
£9 =2V\ = - 2 =0.61 eV
it

13.65 In our analysis of the nearly free electron model we set

(a) H~ k = Vo =0
i- }

(b) H~i - -k }' = Vn

(c) H'.-k} , k} = V- n
(d) Vn = V- n
Establish the validity of these relations and state the relat ion between nand k j '
 13 .10 Hartree-Fock Model 757

13.66 The oscillator strength between the 4P and the 3S states of sodium is 134
0.484 .2 1 Working in cgs units:
(a) What are the dimensions of /34?
(b) What is the dipole moment, d34, of the atom, corresponding to this transition?
State the units of d34 (in cgs).
(c) Compare your values for d34 with the value of the dipole moment , lei ao, where
ao is the Bohr radius. [Note: Recall (lO.138b).]
13.67 Derive the expression for time-averaged electromagnetic energy density, (U), as
given by (13.94), carried in the plane-wave vector potential (13.91).
13.68 Show that the basis (13.11) comprises an orthogonal sequence .

13.69 (a) An electron with energy E » Vo propagates through a one-dimensional crystal

with potential

V(x)
21l' X) + 21 cos (41l'X)]
= 2VO [cos ( --;- --;-

What are the values of the nonzero band gaps, 8En , for this material? How
many energy bands exist for this material?
(b) The second energy band is identified with the valence band of the material and
the next-higher band is identified with the conduction band, which at T = 0 K
is empty. Given that Vo = 1.5 e V and that a = 2 A, estimate the temperature at
which this material will start to conduct .
(c) What are the new wavefunctions related to the first energy gap of this system?

Answers (partial)
(b) Tcond = 0.17 x 105 K

13 .10 • HARTREE-FOCK MODEL22,23

In Section 10.8 we encountered the Thomas-Fermi statistical model for determin-

ing an effective atomic potential. The Hartree-Fock self-consistent model more
realistically takes into account the effects of atomic electrons. In addition to an
effective atomic potential, the model also determines eigenenergies and wave-
functions. Three primary elements of this model are: (1) Each electron moves
in a central field equal to the nuclear potential and that due to the charge den-
sities of the remaining atomic electrons. (2) The Schrodinger equation is solved
for each electron in its own central field, and resulting wavefunctions are made
self-consistent with the fields from which they are calculated. (3) The atomic

21R. D. Cowan , The Theory ofAtomic Spectra and Structure , University of Cali fomi a Press, Berkeley,
Calif., 1981, Section 14-14.
22This analysi s is due to D. R. Hartree, Proc. Cambr idge Soc. 24, III (1929).
23Antisymmetrization of wavefunctions in this scheme was formulated by V. Fock, Z.f Phys. 61, 126
(1930) .
758 Chapter 13 Perturbation Theory

wavefunction is a product of single-electron orthononnalized wavefunctions,

1/!(rl , ... , rz) = ({J1(r» ... ((Jz(rz) (13.130)

With statements (I) and (2) we note that each one-electron wavefunction satisfies
the equation

fh({Jk(rk) = Ek({Jk(rk) (l3.13la)

[n
'
Hk== -~+W(rd
A ]
2 ih 2
(l3.131b)

W(rk) == Vc(rk) + Vk(rk) (l3.131c)

Ze 2
Vc(rk) ==-- (13.131d)
rk

(13.131e)

(13.131f)

The term VC represents electron-electron Coulomb energy. Summation over k

(or j) runs from 1 to Z relevant to an atom with Z electrons. Note that W(rk) is
a central potential and that spin-orbit effects are neglected in the preceding equa-
tions. As there are Z electrons in the atom, (13.131) constitutes Z simultaneous
nonlinear integrodifferential equations for the Z functions . An iterative scheme
is used to solve these equations. This approximation involves the following four
steps:

1. An approximate central potential representing W(rk) is assumed, labeled

W(I)(rk) . (Such a choice might be the Thomas-Fermi potential.)

(13.132a)

2. Electron wavefunctions are computed employing this approximate poten-

tial.

(13. 132b)

3. With first-order wavefunctions, {({Jk (I) (rk)} , at hand, charge densities,

e21({Jk(l)(rk)1 2 , are calculated .
4. Employing the preceding results, the second-order atomic potential,
W(2) (rk), is calculated .

The iterative scheme (steps 1 through 4) is repeated until

W(n+I)(rk) ::::: W(n)(rk) (13.132c)
Problems 759

Calculate
<p(n)(rk)' Ek(n)

Calculate Calculate
w(n+ I)(rk) e 2 1 <p(n )(rk) I 2

y
END

W i") ~ Wi"'" , - -......._ - Goto[j] ,

FIGURE 13.18 Flowchart illustrating the nth cycle of the iterative scheme for solution
of the Hartree-Fock equation.

that is, until w(n) (rk) does not change appreciably with the next iterative cycle. At
the conclusion of this iteration one has a self-consistent effective atomic potential
together with eigenfunctions and eigenenergies relevant to an atom with atomic
number Z, to within a given order of accuracy. A flowchart illustrating this scheme
is shown in Fig. 13.18. In step 5 of this chart we have written II W II for the norm
of W (see Section 4.4). The value of the parameter e determines the accuracy of
the solution.

PROBLEMS

13.70 (a) Neglecting spin-orbit effects, write down the Hamiltonian, Hz , of an atom with
atomic number Z .
(b) In what manner does Hz differ from Lk Hk of the Hartree-Fock analysis?

Answers
(a)

The inequality j < k restricts the double sum to distinct (j, k) pairs.
760 Chapter 13 Perturbation Theory

(b) The difference between Hz and Lk Hk is due to the substitution of the effec-
tive potential Lk Vk( rk) for the exact double-sum electron-electron Coulomb
energy in the preceding expression .

13.71 With reference to the preceding problem, show that the expectation (Hz) , in the
state given by the wavefunction (13.130), may be written as a sum over the single-
particle expectations (Hk ).

Answer
With orthononnality of !Pk(rk) wavefunctions we obtain

(H ) = f ···f 1/1* H1/I drj . . . dt:z

=~f ~:2) !Pk(rk)drk

2
f !Pk(rk) (- h:V: -

+L ff!pj(rj)!Pk(rk)~!pj(rj)!Pk(rk)drk
jfk 0k

With reference to (l3 .13lb) et seq., and the preceding result, (H ) may be written

13.72 (a) Show that (H ) may be made stationary through variation of individual one-
electron atomic wavefunctions.
(b) With reference to Problem 13.70, what can you conclude regarding the eigen-
states of Hz?
Answer (partial)
(b) In problem 4.30 we found that wavefunctions that make (H) stationary are
eigenstates of H and, conversely, eigenstates of H render (H) stationary. With
part (a) of this problem, we may conclude that 1/1 as given by (13.130) is an
eigenstate of Hz , providing !Pk(rk) are eigenstates of Hk .

13.73 Obtain an expression for the eigenenergies of Hz in terms of the eigenenergies of

Hk and electron-electron Coulomb energy.

Answer
Employing expressions obtained in Problem 13.71, we find

13.74 A particle of mass m is confined to the interior of a spherical cavity of radius G .

(a) What is the ground-state wavefunction, 1/IG (r , e, cf», and energy, EG ' of this
system?
(b) A potential sphere of radius G / I0 and height Va is placed concentric and interior
to the original sphere. Employing nondegenerate perturbation theory, evaluate
Problems 761

the new first-order ground-state wavefunction, vc' , and new ground-state en-
ergy, Ed.

13.75 An harmonic-oscillator Hamiltonian includes the perturbation field, (C / a)x 3 where

C and a are constants. In the representation of (7.IS), if dimensions of a are length,
what are the dimensions of C? What is an appropriate statement of smallness for the
given perturbation? Calculate the first-order correction for the nth eigenstate energy
for the given perturbed harmonic oscillator.

13.76 Throughout this text, it was stated that the Hamiltonian of a confined particle is that
of a free neutral particle interior to the domain of confinement and infinite beyond
the domain of confinement (cf., 4.2) . Furthermore, it is assumed that the sudden
expansion formalism is valid. Are these assumptions correct?

Answer
First note that we are discussing a particle with no internal degrees of freedom.
To the degree that a perfectly rigid wall exists, the Hamiltonian form (4.2) is valid .
Consider that the confining 'box' undergoes a sudden expansion. In this event , there
is a finite probability for a transition of the system to a lower or higher energy. In a
sudden expansion, no energy is lost to the particle in wall collis ions . As the particle
is neutral, there is no radiative mechanism to support either of these transitions, and
the particle suffers no change in energy, in contradiction to standard perturbation
theory. In a sudden compression, the particle can gain energy from the walls , but
once in an excited state , it cannot radiatively decay as it is neutral.
If the particle is charged, and the confining walls are metallic or dielectric, then
the simple Hamiltonian (4.2) is incorrect. The valid Hamiltonian is given by the
electromagnetic form , (10.8S). If it is the case that the confining walls are flat, then
for either type confinement, the charged particle interacts with its image.

13.77 Consulting the periodic chart, we note that the elements palladium and platinum,
in the solid state , have the respective electronic structures: [Kr)4d lO; [Xe)Sd lO. We
recall that the p shell can accommodate 10 electrons, so that the respective d shells
in these atoms are closed. How is it that these solids are metals (i.e., have finite
conductivity)?

Answer
There is overlap between the d shell and the s shell , so that electrons in the closed
d shell can be promoted to the two-state s shell and contribute to conductivity. As
noted previously, in this instance the material is said to be degenerate (Section 12.9).
To denote this configuration in the periodic chart, the complete electronic structure
is written : [Kr)4d IOSso ; [Xe)Sd I06so .
CHAPTER

14 Scattering in Three Dimensions

14.1 Partial Waves

14.2 S- Wave Scattering
14.3 Center-of-Mass Frame
14.4 The Born Approximation
14.5 Atomic-Radiative Absorption Cross Section
14.6 Elements of Formal Scattering Theory. The Lippmann-Schwinger
Equation

In this chapter an elementary description is offered of the quantum mechanical

theory of scattering in three dimensions. Application of low-energy scattering is
made to the Ramsauer effect, formerly encountered in Chapter 7, and scatter-
ing from a rigid sphere. The chapter continues with a discussion of the Born
approximation. This important analysis permits certain scattering problems to
be formulated in terms of harmonic perturbation theory developed previously in
Chapter 13. The cross section of an atom interacting with a radiation field is ob-
tained. For off-resonant incident phonons one encounters the line-shape facto r.
The chapter concludes with a description of the formal theory of scattering and
derivation of the Lippmann-Schwinger equation in which the formalism of the
interaction picture (Chapter 11) comes into play.

14.1 • PARTIAL WAVES

The Rutherford Atom

One of the most fundamental tools of physics used for probing atomic and sub-
atomic domain s involves scattering of known particles from a sample of the el-
ement in question. Thus, for example, the description of an atom as being com-
prised of a positively charged central core of radius ~ 10- 13 em, with external
satellite electrons, is due to scattering experiments performed by E. Rutherford in
1911. In these experiments ex particles in an incident beam were deflected in pass-
ing through a thin metal foil. The prevalent model for an atom at the time was J. J.
Thomson's "plum pudding" model , in which negative electrons floated in a ball
of positive charge . The relatively large angle suffered by a small fraction of the ex

762
14 .1 Partial Waves 763

particles in the incident beam in Rutherford's experiments was found to be incon-

sistent with Thomson's model of the atom. For it is easily shown that ex particles,
after passing through hundreds of such spheres of distributed charge, are deflected
at most only by a few degrees. On the other hand, the actual scattering data are
consistent with an atomic model in which the positive charge is concentrated in
a central core of small diameter. Large angle of scatter is then experienced by ex
particles which pass sufficiently close to the positive nucleus.

Scattering Cross Section

The typical configuration of a scattering experiment is shown in Fig. 14.1. A
uniform monoenergetic beam of particles of known energy and current density
J inc (7.107) is incident on a target containing scattering centers . Such scattering
centers might, for example, be the positive nuclei of atoms in a metal lattice. If
the particles in the incident beam are, say, ex particles, then when one such particle
comes sufficiently close to one of the nuclei in the sample, it will be scattered. If
the target sample is sufficiently thin, the probability of more than one such event
for any particle in the incident beam is small and one may expect to obtain a
valid description of the scattering data in terms of a single two-particle scattering
event.
Let the scattered current density be Jsc. Then the number of particles per unit
time scattered through some surface element dS is J sc . dS. Let dS be at the radius
r from the target. Then if dO is the vector solid angle subtended by dS about the
target origin, dS = r 2 dO (see Figs. 9.9 and 14.1). The vector solid angle dO is
in the direction of er ; that is, dO = er dQ . It follows that

Inciden t beam

FIGURE 14.1 Scattering configuration.

764 Chapter 14

Number of particles
passing through dS
per second
I
Scattering in Three Dimensions

= dN = J se • dS = r 2 Jse • dO

Since the number of such scattered particles will grow with the incident current
one may assume this number to be proportional to J ine and can equate
J ine ,

(14 .1)

The proportionality factor do is called the differential scattering cross section and
has dimensions of crrr'. It may be interpreted as an obstructional area which the
scatterer presents to the incident beam. Particles taken out of the incident beam
by this obstructional area are scattered in dO. The total scattering cross section
a represents the obstructional area of scattering in all directions .

(14.2)

Scattering cross section has a classical counterpart . Classically, the total cross
section seen by a uniform beam of point particles incident on a fixed rigid sphere
of radius a is a = na 2 . If the incident beam has current JJine, the number per
second scattered out of the beam in all direction s is na 2JJine.

The Scattering Amplitude

Returning to quantum mechanics, let the particles in the incident beam be inde-
pendent of each other so that prior to interaction with the target a particle in the
incident beam may be considered a free particle. If the z axis is taken to coincide
with the axis of incidence, then a particle in the incident beam with momentum
lik and energy li 2 k 2 / 2m is in the plane-wave state,
. _ ikz
Ipme - e (14.3)

When this wave interacts with a scattering center, an outgoing scattered wave Ipse
is initiated. If the scattering is isotropic so that scattering into all directions (a1l4n
steradians of solid angle) is equally probable, we can expect the scattered wave
Ipse to be a spherically symmetric outgoing wave. The specific form of an isotropic
outgoing wave was described previously [(10.65) and Problem 10.6].

ei kr
Ipse,iso = -
r
More often, however, the scattered wave is anisotropic. Anisotropy of the scatter-
ing component wavefunction Ipse may be described by a modulation factor f«(}),
and in general we write

f«(})e ikr
Ipse = -'----- (14.4)
r
14 .1 Partial Waves 765

The modulation fee) is called the scattering amplitude and will be shown to
determine the differential scattering cross section da .
The number of particles scattered into dO , which is in the direction of e-, is
obtained from the radial component of J se [recall (7.107)] :

*
]se, r = 2mni ( Ipse aar Ipse - Ipse aar Ipse*) (14.5)

= Ii\ If(e)1
mr
2

Since the vector element of solid angle dO is in direction e-, it follows that

r 2 J se • dO = r 2 ]se, r dQ = ]ine do
In that the current vector of the incident beam only has a z component with mag-
nitude Ilk/m [see Fig. 14.2], the preceding equation becomes

2 Ilk
r ]se
, rdQ = -da
m
Substituting (14.6) into this equation gives the desired relation,

2
I da = If(e)1 dQ I (14.6)

Thus the problem of determining do is equivalent to constructing the scattering

amplitude f (e) .

FIGURE 14.2 The scattering cross section is independent of the azimuthal angle ¢ for
central potentials of interaction V (r) .
766 Chapter 14 Scattering in Three Dimensions

Owing to the rotational symmetry of the scattering configuration about the axis
of the incident beam and the assumed radial quality of the interaction potential be-
tween incident particle and scatterer, the scattering cross section depends only on
the scattering angle 0 (and incident energy) and not on the azimuthal angle ¢ (see
Fig . 14.2). It follows that, in integration (14.6) over all directions, the integration
over d¢ may be done separately to obtain 2][ . There results

a = f do = 2][ l,r If(0)1 2 sinO dO (14.7)

The total cross section is a simple integral over the square modulus of the scat-
tering amplitude. Referring again to Fig . 14.2, we see that the same symmetry
implies that f(O) is an even function of 0 or, equivalently, f(O) = f(cos 0) .

Partial-Wave Phase Shift

The form of the wavefunction for the steady-state scattering configuration de-
scribed above, at positions far removed from the scattering target, will contain a
plane-wave incident component and an "outgoing" scattered component:

. f(O)e ikr
<per, 0) = e'kz + (r --+ 00) (14.8)
r
(Fig. 14.3). The scattering amplitude is determined by matching (14.8) to the
asymptotic form of the solution of the Schrodinger equation relevant to the con-
figuration at hand. Such configuration includes a particle of mass m with known
energy ti2 k 2 / 2m, interacting with a fixed scattering center through the central po-
tential VCr) . The radial Schrodinger equation is given by (10.109) .
2
1 d
- -r -
I (l + 1) +k 2 - 2mV ]
-- Rkl(r) = 0 (14.9)
2
[ r dr r2 ti2

scattering
center

scattered wave

FIGURE 14.3 Incident plane wave and scattered outgoing spherical wave.
14.1 Partial Waves 767

In the far field where VCr) is rapidly approaching zero, one may expect the so-
lution to this equation to be given approximately by the asymptotic form of the
free-particle solution j/(kr) [see (10.55) and Table 10.1).

Rk/ ~ - 1 SIll
. ( kr - -ITC) (14.10)
kr 2
Provided that V (r) decreases faster than r -I , this free-particle asymptotic form
remains intact I save for a change in argument through a phase shift 8/.

Rk/
asm . ( kr - -ITC
= -1 SIll + 8/) (14.11)
kr 2
A superposition state comprised of these wavefunctions at fixed k has the form
00

gYk(r,O) = LC/Rk/asmPI(COSO) (14.12)

1=0

The lth term in the sum is called the lth partial wave and 81 is the phase shift that
the partial wave incurs in scattering.
We must now match the asymptotic form of the general solution (14.12) to the
form (14.8). With the expansion of exp(ikz) given in Problem 10.12, we obtain
the asymptotic expression

'k ~ / sin(kr-ITC/2)
e' z ~ L...J(21 + l)i p/(cosO)
/=0 kr
The coefficients C/ and the scattering amplitude f(O) are found from the matching
equation
" ( sin(kr - ITC / 2 + 81)
L...J C/ PI cos 0) - - - - - - -
/ kr
ikr
= L(21 + l)i/ p/(cosO) sin(kr -ITC/2) + f(O)e
/ ~ r
Expanding f(O) in a series of Legendre polynomials , one obtains, after some
trigonometric gymnastics,

CI = i/(21 + 1)exp(i8/)
(14.13)
f(O) = -1 ~ C/ . ~
L...J ""T ( 0
SIll UI Pi cos )
k /=0 I
The problem of calculating do or, equivalently, f(O) is reduced to one of con-
structing the phase shifts of 8/.
1For example, the analysis is not valid for the Coul omb potenti al V (r) = r -1 . Proof of the valid ity of
the stated criterion may be found in L. Landau and E. Lifsh itz, Quantum Mechanics, Addison-We sley,
Reading, Mass., 1958.
768 Chapter 14 Scattering in Three Dimensions

Two immediate results are evident: First, substituting the series (14.13) into
(14.7) and taking advantage of the orthogonality of the Pi(cos (J) polynomials , we
obtain
4rr 00
(J = 2" L(21 + 1) sin2 s, (14.14)
k t=o
The second result follows from setting (J = 0 in (14.13), which yields

1 i
f(O) = - L(21 + 1) cos Ot sin Ot +- L(21 + 1) sin2 Ot
k t k t

Compari son with (14.14) reveals that

4rr
(J = k Im[f(O)] (14.15)

This result is known as the optical theorem. It is a widely used relation connecting
the forward scattering amplitude, f (0), to the scattering in all directions, (J . 2

Relative Magnitude of Phase Shifts

The problem of determining the partial wave phase shifts Ot is often difficult.
However, under certain conditions one may make simplifying assumptions which
greatly facilitate calculation. In classical scattering one introduces the impact pa-
rameter. If Land p are the incident particle 's angular momentum and linear mo-
mentum, respectively, then the impact parameter s is given by (see Fig. 14.4)

L = ps

Quite clearly, if the potential of interaction is appreciable only over the range
ro, then the interaction between incident particle and scatterer will be negligible
for s > ro oThis criterion provides a useful rule of thumb applicable in quantum
mechanics . With L = liJI(l + 1) ::::: Itl and p = hk, interaction will be negligible

p
s
---------- .....
scattering
center

FIGURE 14.4 Classical trajectory and impact parameter s.

2For inela stic scattering, (14 .15) is still valid with a replaced by the total cro ss section, aT = as+aA,
where as is the elastic cros s section and a A is the absorption cro ss section.
Problems 769

if

I > rok (14.16)

The incident energy is fi,2k2/2m .

Each partial wave in the superposition (14.12) represents a state of definite an-
gular momentum. From (14.16) we can expect that partial waves with I values in
excess of rok will suffer little or no shift in phase. In the corresponding expansion
of the scattering amplitude 1 (0) as given by (14.13) it follows that only those 8[
values will contribute for which I < rok. For low-energy scattering with kr« « 1,
only the I = 0 phase shift will differ appreciably from zero. When such is the case
(14.13) reduces to

1 ./)
1(0) = ke' 0 sin 80 (14.17)

which is independent of O. The scattering is isotropic and is called S-wave scatter-

ing. Only the S partial wave (I = 0) contributes to the scattering. In the opposite
extreme of large incident energies, kro » 1, we can expect all partial waves to
suffer phase shifts and the cross section to be anisotropic .

PROBLEMS

14.1 From (14.1) we find that the number of particles scattered into the solid angle dQ per
second is

dN =]du

or, equivalently,

dN = (dU) dQ
] dQ

The Coulomb cross section for the scattering of a charged particle of energy E and
charge q from a fixed charge Q is

do (qQ)2 I
dQ = 4E sin 4 (e/ 2)

(a) What is the expression for the fraction of particles scattered into the differential
cone (e, e + de) from a target comprised of A scattering centers per unit area?
(b) Employ the expression you have obtained to find the fraction of ex particles with
incident energy 5 MeV which are scattered into a differential cone (e, e +de) at
e
= n / 2, in passing through a gold sheet I JIm thick.
Answers
(a) If we assume that each particle in the incident beam sees only one scatterer and
that there is a scattering event for each scatterer, then
770 Chapter 14 Scattering in Three Dimensions

oN = A (:~) dQ
For the scattering into the cone (e. e + de)

oN=A fo2rr d4J(:~)Sinede

=2rrA (:~) sinede

(b) For a sheet of mass density P. thickness I , comprised of atoms with atomic
mass A,

A = pNol
A
where No is Avogadro's number (NO atoms have mass A grams) . For a gold foil
I em thick with p = 19.3 g/cm 3 and A = 197, we obtain A = 5.9 x 1022 I
atoms/cm- . For ex particles of energy 5 MeV scattered by the nuclei of gold
atoms, q Q/ E = e 2 • 2 x 79/ E = 4.6 x 10- 12 em. Thus we obtain

oN(rr /2) = ~ (qQ)2 pNol de ~ 2 x 10-4 de

2 E A

14.2 • S-WAVE SCATTERING

Let us consider the configuration of a low-energy beam of point particles of mass

m scattering from a finite spherical attractive well of depth Vo and radius a.

- Vo for r < a
VCr) = (14.18)
[o for r > a

If we assume that energies are sufficiently small that ka « I , we need only look at
the S-wave scattering. The corresponding Schrodinger equation is obtained from
(14 .9). Setting I = 0 and u == r R there results, for r < a,

£1------
a
o ~-_;_---- r

The solution to this equation which corresponds to R(r) remaining finite at r =0

is

(r < a)
14.2 S-Wave Scattering 771

For r > a, V = 0 and we obtain the general solution

u2 = B sin(kr + 80) (r > a)

n,2k2
-=E
2m
Boundary conditions require continuity of dIn u f dr at r = a , which gives

k, cot k, a = k cot(ka + 80) (14.19)

In that k, is finite, in the limit that k goes to zero,

k,cotk,a
cot(ka + 80) = k

grows large so that sin(ka + 80) grows small and we may set
sin(ka + 80) ::::: ka + 80
Since ka « I, this equation implies that 80 « I as well. Under these conditions
(14.19) reduces to

or equivalently

In that 80 is small, we may also set

We may now construct the scattering amplitude (14.17) and cross section (14.7).

a= 4tt a 2 (tank,a
---I
)2 (14.20)
k,a

Two significant observations relevant to this study of attractive well scattering are
discussed next.

S-Wave Resonances and Ramsauer Effect

First we note that when k,a is an odd multiple of (nI2), tank,a is infinite and
the cross section as given by (14.20) becomes singular. In that 80 is also infinite
at these values of kia , assumptions leading to (14.19) are violated and we must
772 Chapter 14 Scattering in Three Dimensions

seek an alternative procedure to construct the cross section. Consider the relation
(14.19), which assumes only that ka « 1. Let k) a = n (rr /2) , where n is an
odd number. At these values , (14.19) gives sin(80 + ka) = 1, which with the
condition ka « 1 yields sin 80 ::::: 1. Thus the maximum cross section at these
S-wave resonances is

4x
O"max = JZ2' (14.21)

A more careful analysis pursued to higher angular momentum states , appropriate

to larger incident energies, reveals corresponding resonances at I = 1, termed
P-wave resonances, and so forth .
Whereas (14.20) suggests resonant scattering at odd multiples of n /2, it also
indicates that the attractive scattering will become transparent to the incident
beam at values of k) a which satisfy the transcendental relation

As noted in Section 7.8, such resonant transparency of an attractive well is ex-

perimentally corroborated in the scattering of low-energy electrons ("'-'0.7 eV) by
rare gas atoms and is termed the Ramsauer effect.

The Repulsive Sphere

The second observation related to our study of low-energy scattering by a scat-
tering well is that merely changing the sign in the defining equations (14.18)
produces the potential for a repulsive sphere of radius a. Solution for the cor-
responding scattering problem is effected by simply replacing k) by i K , in the
relations following (14.18) . For the interior wavefunction we obtain

u) = A sinh xr (r < a)

!i2K2
--=Vo-E > 0 E~-i---
. - L -............_ _
2m ~ r

The exterior wavefunction U2 maintains it sinusoidal dependence for r > a, as

given in the equation preceding (14.19) . Imposing boundary conditions at r = a
and, again assuming low-energy incident particles, we obtain the total scattering
cross section,

0"
2
=4xa (tanhKa
----1 )2 (14.22)
Ka
Problems 773

In the limit that Vo -+ 00, the sphere becomes impenetrable and the total cross
section reduces to

(14.23)

Since this formula does not contain h, our suspicion is that it is also appropriate to
the classical domain . However, the obstructional area imposed by a rigid sphere
of radius a to an incident beam of classical particles has the value na 2 , so the
quantum cross section is larger than the classical one by a factor of 4. Although
the cross section (14.23) does not contain Planck's constant, nevertheless one
might still object to considering it a classical result in that it is relevant to the
strictly nonclassical domain of large de Broglie wavelength . If a classical result
is to be obtained, it should emerge in the limit of large incident energy, ka » 1.
Such analysis, which includes the phase shifts of all waves.I again yields a cross
section independent of ti; namely

ka» 1

which is still larger than the classical result. Thus the classical cross section does
not emerge in the limit of large incident energy. This discrepancy may be ascribed
to the sharp edge of the spherical potential barrier for the configuration at hand .
Across the sharp potential step, dV[dx is infinite and it is impossible for the
classical criterion (7.166) to be satisfied .

PROBLEMS

14.2 The scattering amplitude for a certain interaction is given by

fee) = ~(eika sinka + 3iei2ka cos e)

k
where a is a characteristic length of the interaction potential and k is the wavenumber
of incident particles.
(a) What is the S-wave differential cross section for this interaction?
(b) Suppose that the above scattering amplitude is appropriate to neutrons inci-
dent on a species of nuclear target. Let a beam of 1.3-eV neutrons with current
1014 cm- 2 s-I be incident on this target. What number of neutrons per sec-
ond are scattered out of the beam into 47T x 10- 3 steradian about the forward
direction?
14.3 Analysis of the scattering of particles of mass m and energy E from a fixed scattering
center with characteristic length a finds the phase shifts

<
o
• _I [
-sm (iak)l ]
I - "j(21 + l)l!

3The calculation may be found in L. 1. Schiff, Quantum Mechanics, 3d ed., McGraw-Hili, New York,
1968.
774 Chapter 14 Scattering in Three Dimensions

(a) Derive a closed expression for the total cross section as a function of incident
energy E.
(b) At what values of E does S-wave scattering give a good estimate of a?
Answer (partial)

(a)

14.3 • CENTER-Of-MASS fRAME

In all of the preceding analysis, it has been assumed that the target particle remains
fixed during the scattering process. This is the case if the mass of the target par-
ticle far exceeds that of the incident particle. More generally, however, the recoil
motion of the target particle must be taken into account in any scattering analysis .
Thus the general formulation of a scattering event involves two particles, of mass
ml and m i -
As described in Section 10.5, the motion of such two-particle systems may
be described in terms of the motion of the center of mass and motion relative to
the center of mass. The Hamiltonian of the relative motion (10.99) describes a
single effective particle with reduced mass u. = m I ma!(m I + m2) at the radius
r = fl - f2. This is the motion observed in a frame moving with the center of
mass. So, in fact, in this center-of-mass frame, the scattering event may be de-
scribed by a single particle of mass fL interacting with a potential V (r) centered
at a fixed origin. It follows that the preceding formulation of the cross section
0'(8) describing scattering from a fixed scattering center is appropriate to scat-
tering in the center-of-mass frame. The only change is that the mass m of the
incident particle is set equal to the reduced mass u , In addition, we must note that
the angle of deflection 8 is measured in the center-of-mass frame. For example,
in the expression (14.13) for the scattering amplitude, 8 is the angle of scatter in
the center-of-mass frame, which will henceforth be called 8e. To obtain a relation
between the scattering cross section a[(8d in the frame of the experiment, or
what is commonly called the lab frame and the cross section ac(8c) as measured
in the center-of-mass frame, we note the following . The number of particles scat-
tered into an element of solid angle in the lab frame JJinc(daL/dQd dQL is equal
to the number scattered into the corresponding solid angle in the center-of-mass
frame, JJinc(dae/dQc) dQe . This gives the equality
dae dcos8e
=------ (14.24)
dcos8e dcos8L

The relation between cos 8e and cos 8L is obtained by examining the scattering in
both frames. In transforming from one frame to the other, it is convenient to speak
in terms of velocities. Such velocities are related to linear momentum through the
prescription v = hk/m . When one describes an "orbit" in this description , one
has in mind a picture inferred by the direction of momentum k vectors. Thus,
14.3 Center-oF-Mass Frame 775

(a) (b) (c)

FIGURE 14.5 (a) In the center-of-mass frame, the total momentum is zero. (b) The
relative velocity vector v rotates through the angle Be . (c) In the lab frame, m2 is assumed
to be at rest before collision.

before collision, m2 is at rest and the incident particle has velocity VI = tiki mi.
After collision, m I is scattered through the angle fh.
The center-of-mass frame is characterized by the property that total momentum
in that frame is zero before and after collision (Fig. 14.5). Letting barred variables
denote values in the center-of-mass frame, and v the relative velocity,

one obtains , for before the collision

We may immediately conclude that

In a similar manner, after collision we write

-PI, = /-LV
,
p/ = -/-LV'

or, equivalently,
- , /-L ,
VI = - v
ml
- , /-L,
V2 = --v
m2

The corresponding relations in the lab frame are obtained by adding VCM to the
right-hand sides of these equations . Multiplying the resulting equations by m I and
ma. respectively, gives
, , /-L ~
PI = /-LV + -:J~
m2
776 Chapter 14 Scattering in Three Dimensions

FIGURE 14.6 Orientation of momentum and relative velocity for m I / m2 < I .

(14.25)

/P = PI + P2
Since mz is at rest before scattering, PI = mlV = /P. It follows that (14.25) may
be rewritten
, , ml
PI = fLV + -fLV
m2 (14.26)
P2' = -fLv' + fLV

FIGURE 14.7 Triangle These vector equations imply the vector diagrams shown in Fig. 14.6. The desired
used to obtain relation be- relation between fh and ec is obtained by constructing taneL from the partial
tween fh and (jc. diagram shown in Fig. 14.7:

fLv'sinec
taneL = -----=------=~-- (14.27)
(m 1/m2)fLv + fLV' cos ec
Now Hrel is a conserved quantity throughout the scattering . Prior to, and after
collision, Hrel is purely kinetic and has the respective values fLv2/2, fLV,2/2. It
follows that the magnitude of the relative velocity is maintained in scattering

v = v'
Substituting this equality into (14.27) gives the desired relation ,
sinec
tan eL = ---=--
E + cosec
(14.28)

This relation permits completion of (14.24):

daL dac (1 + E2 + 2E cosec)3 /2

(14.29)
d COSeL d cosec 1 + E cosec
If the mass of the scatterer is very much larger than that of the incident particle ,
we may set E = 0 and the cross sections in both frames are equal. From (14.28)
in this same extreme we obtain eL = ec .
 14.4 The Born Approximation 777

In general, as (14.29) implies, scattering that is isotropic in the center-of-mass

frame is not isotropic in the lab frame. For example, the isotropic cross section
obtained for S-wave scattering (14.17),

when substituted in (14.29) yields [with (14.28)] an anisotropic cross section in

the lab frame:

(da)
dr? L
sin2 80 (I + E2 + 2Ecos Bc)3 /2
= ~ I +EcosBc
(14.30)

Applications of results developed in this section appear in problems to follow.

Whereas our primary example in the preceding analysis is relevant to low-energy
scattering, where the potential of interaction plays a dominant role, the analysis to
be developed in Section 14.4 addresses the case where the potential of interaction
acts as a small perturbation on the incident plane-wave state. This analysis, known
as the Born approximation, has many applications.

PROBLEMS

14.4 Assume that the differential cross section for a given interaction potential do /dn is
isotropic in the center-of-mass frame . For mass ratio E « 1, what is the ratio of the
differential cross section in the forward direction to that in the () = it / 2 direction in
the lab frame?

Answer
da(O)
---= 1+ 2E
do (rr / 2)

14.5 At what value of ()c will the cross section vanish in the lab frame for S-wave scatter-
ing of two particles with mass ratio E?

14.4 • THE BORN APPROXIMATION

Harmonic perturbation theory, developed in Chapter 13, includes as a special case

the example of a constant potential that has been turned on for t seconds. The per-
turbation Hamiltonian'' is then given by (13.53) with (J) = O. As was shown in
Section 13.6, the theory of harmonic perturbation leads naturally to Fermi's for-
mula (13.64) for cases where final states comprise a continuum. Such, of course,
is the situation for scattering problems .

4For the case w = 0, the factor 2 is deleted in (13.53).

778 Chapter 14 Scattering in Three Dimensions

Domain of
interaction

FIGURE 14.8 In the Born approximation, incident and scattered particles are in plane-
wave states.

For these problems the perturbation Hamiltonian is the interaction potential,

which is viewed as being "turned on" during the time that the incident particle is
in the range of the potential. The incident particle enters the range of interaction
with momentum lik and leaves the range of interaction with momentum lik'. Such
states of definite k before and after interaction correspond to plane-wave states
(Fig. 14.8). Let us suppose that the scattering experiment is performed in a large
cubical box of volume L 3 . Normalized plane-wave states corresponding to k and
k' are then given by

(14.31)
, eik" r
Ik} = 0 /2

To apply Fermi 's formula (13.64) for the rate of transition from the k to the k'
state, caused by the perturbing potential VCr) ,

(14.32)

we must know the density of final states g(Ek ') ' Having prescribed that the scat-
tering experiment is performed in a large box of volume L 3 , we may employ the
expression for geE) as given by (8.141). Written in terms affinal momentum hk',
this expression becomes for nonspinning particles''

mL 3k'
g(Ek') = 2n 2li 2 (14.33)

5The 9 factor in Problem 2.42 represents density of states per unit volume.
14.4 The Born Approximation 779

Now we wish to use the rate formula (14.32) to obtain an expression for the differ-
ential scattering cross section da. This parameter was defined by (14.1) according
to which the number of particles scattered into dO per second is ]ine da . To relate
the transition rate Wkk ' to da , we note that the incident plane wave [k) given in
(14.31) corresponds to an incident current

lik
Jine = mL3 (14.34)

In that g(Ek) as given by (14.33) is isotropic in k', it represents the density of

final k' states in all 4rr solid angle. To select those scattered states that lie in
the direction dO about k' , we multiply 9 by the ratio do./4rr . With g(Ek') so
augmented, Wkk' then represents the rate at which particles of the incident flux
(14.34) are scattered into dO in the direction of k' . This rate is by definition the
product ]ine da. Thus we obtain the desired relation

dO.
]ine da = -Wkk'
4rr
Inserting previous expressions, we obtain

2
da mL 3 k'
- = - -1(k'lVlk}1 2 (14.35)
dO. ( 2rrli2 ) k

Since particles suffer no loss in energy in the scattering process, we may equate

k =k'

The Scattering Amplitude

Recalling (14.6), which relates do to the scattering amplitude feB), and inserting
the explicit forms (14.31) for incident and scattered states into (14.35), allows the
identification (with a conventional minus sign)

feB) = .:»:
2rrli
f V(r)eir'(k-k') dr
2
(14.36)

This formula for the scattering amplitude may be further simplified through the
substitution

K=k-k'

As is evident in Fig. 14.9a, owing to the equal magnitudes of k and k', we may
set

K = 2ksin (~)
780 Chapter 14 Scattering in Three Dimensions

0/2 k
~K/2
(a)

(b)

FIGURE 14.9 (a) Transformation K =k- k' . (b) Spherical coordinate frame with K
aligned with the polar axis.

where () is the angle of scatter. With the differenti al volume of integration dr in

(14.36) written in spherical coordinates and the polar axis taken to be coincident
with K (Fig. 14.9b), we obtain

f«()) = -~2 t" icp r d8 sin8 [00 dr r2V(r)eiKrcos9

2n ti 10 10 10
= -m
-
ti2
100
0
drr 2V(r) 11
-1
drJeiKrT/

Integration over rJ == cos 8 gives

f«()) = - -2m
2-
ti K
10
00
dr rVer) sin Kr (14.37)

This expression for the scattering amplitude is called the standard form of the
Born approximation .
In applying this formula for the scattering amplitude, one should keep in mind
that it is derived on the basis of perturbation theory according to which the scatter-
ing potential should be small compared to the free-particle (unperturbed) Hamil-
tonian. This will be the case for sufficiently large incident energies or sufficiently
weak strengths of potential.
14.4 The Born Approximation 781

The Shielded Coulomb Potential

Let us apply (14.37) to calculate the cross section for the shielded Coulomb po-
tential,"

Ze2 exp(-rja)
(r = - ---=-------'-
V)
r

The exponential factor for r > a acts to shield the bar Coulomb potential Ze 2 j r
between two particles with respective charges Ze and e. Thus beyond the range
a, the potential is exponentially small. Within the range, r < a, the potential is
essentially Coulombic. Substituting the shielded Coulomb potential into (14.37)
gives

2mze21°O r ja
f(8) = - 2 - - e- sinKrdr
Ii K 0
2mZe 2
(~)
I
K = 2ksin
~ K2 + (1ja)2 '

The corresponding scattering cross section is obtained from (14.6) .7

da (2mZe 2jli2)2
(14 .38)
dQ [K2 + (1ja)2]2

In the limit of large incident energies K 2 » a- 2 , the predominant contribution to

do is due to the bare Coulomb potential. The resulting cross section, employed
previously in Problem 14.1, appears as

(14.39)

li2 k2
E=-
2m

This is the precise expression for the Rutherford cross section for the scattering
of a charged particle with charge e and mass m from a fixed charge Ze, which is
seen to be dominated by forward scattering (8 ~ 0). Furthermore, the classical
evaluation of the Rutherford cross section also gives (14.40), with E = p2 j2m .

6This potential enters in three independent areas of physics, where it carries the following names:
in plasma physics, the Debye potential; in high-energy physics, the Yukawa potential ; in solid-state
physics, the Thomas-Fermi potential.
7With m replaced by the reduced mass J1. , (14.38) represents the cross section in the center-of-mass
frame.
782 Chapter 14 Scattering in Three Dimensions

PROBLEMS

14.6 Using the Born approximation, evaluate the differential scattering cross section for
scattering of particles of mass m and incident energy E by the repulsive spherical
well with potential

Va, O <r <a

V(r) ={
0, r > a

Exhibit explicit E and 8 dependence .

Answer
do 2mVO).
dQ = ( /i2K3 (smKa - KacosKa)
2

n« = 2·J2mE sin(8 j2)

14.7 Using the Born approximation, obtain an integral expression for the total cross sec-
tion for scattering of particles of mass m from the attractive Gaussian potential

14.8 An important parameter in scattering theory is the scattering length a. This length is
defined as the negative of the limiting value of the scattering amplitude as the energy
of the incident particle goes to zero.

a =- lim f(8)
k---+O

(a) For low-energy scattering and relatively small phase shift, show that

a =- lim 00
k
k---+O

(b) For the same conditions as in part (a), show that

a = 4Jra 2
(c) What is the scattering length for point particles scattering from a rigid sphere of
arbitrary radius a?

14.5 • ATOMIC-RADIATIVE ABSORPTION CROSS SECTION

Returning to our analysis of Section 13.9, again we consider a flux of photons

incident on an atom, carrying one photon per unit volume. Since the photon moves
with speed c, this gives an incident photon current

I photon (PhotonS)
line = C X 3 = C 2 (14.40)
cm cm s
14.5 Atomic-Radiative Absorption Cross Section 783

There is a probability rate for each photon in the incident current to be absorbed
by the atom . The principle of microscopic reversibility allows us to equate this
probability rate for absorption to the corresponding probability rate of atomic
decay .
With Fermi's golden rule (13.64) we may then write the probability rate for
atomic absorption as

2Jr I 2g(w)
W nn' = hl(n , k IIIL ln}1 -li- (14.41)

where we have made the replacement

g(E) = g(w)
n
[See (13.11lb) and recall that g(E) = 8Jrg(E) and that V = 1.] With (13.106),
our preceding equation (14.41) becomes

2Jr 2 g(w) I 2
W nn' = he (2JrIiw)T1(n [a- rln}1 (14.42)

If the polarization unit vector a of the incident field is randomly oriented, then
as was previously demonstrated in our discussion of the Einstein B coefficient
(Section 13.7), we may set

I 2 I 2 1dnn ,2
I(n la - rln}1 = 3rnn' = 3 72 (14.43)

Here we have reintroduced the dipole moment d (13.69).

With (14.43) placed into (14.42) we find

(14.44)

This is the probability rate for an atom to absorb a photon from an incident cur-
rent carrying one photon per unit volume . Since the incident current in the present
configuration is so normalized, W nn' represents the rate at which photons are ab-
sorbed by the atom from the incident beam. With our previous definition of cross
section we may then write

(14.45)

Note the dimensions:

PhotonS) 2 (1)
line ( cm 2 s X a(cm ) = W nn, -;

With (14.40) and (14.45) we obtain the total cross section,

(14.46)
784 Chapter 14 Scattering in Three Dimensions

for resonant absorption at the incident frequency

A more accurate description of this process includes the possibility of absorp-

tion of incident photons which are off-resonance. For such cases, with W == W nn' ,
(14.46) is generalized to the form

(14.47)

where we have set

9(W) -+ 9(w, w)

The function 9(w, w) is the so-called line-shape factor.

A realistic expression for 9(w, w) which is appropriate to many line-broadening
processes is the Lorentzian line-shape factor,

(14.48)

See Fig. 14.10.

The spreading of an absorption line is attributed to relaxation processes-such
as, for example, the relaxation of excited atomic states incurred in atomic colli-
sions. If r represents the decay time for such processes, then one sets

In the idealized limit that these states last indefinitely, I::!.wL -+ 0, and 9[(w, w)
becomes sharply peaked about w = W. In this limit (14.47) becomes

FIGURE 14.10 Lorentzian line shape.

 14.6 Elementsof Formal Scattering Theory 785

(14.49)

Here we have employed the delta-function representation (e.9).

14.6 • ELEMENTS OF FORMAL SCATTERING THEORY.

THE LIPPMANN-SCHWINGER EQUATION

In this concluding section we present a brief introduction to the formal theory of

scattering, central to which is the Lippmann-Schwinger equation.8 An elementary
derivation of this equation is presented based on the interaction picture described
previously in Section 11.12. The Lippmann-Schwinger equation so derived ap-
pears in a form independent of representation. Writing this equation in coordinate
representation is found to give an integral equation for scattered states, which in
tum gives a general expression for the scattering amplitude. In the Born approxi-
mation this relation returns our previous expression for f«() given by (14.36).
We consider the Hamiltonian

(14.50)

where Ho is the free-particle Hamiltonian and the infinitesimal parameter 8 has

dimensions of energy. The interaction Yis assumed to be independent of time.
For small 8 the exponential factor has the effect of "turning on" the interaction Y
in the interval about t = O. We will also find that the presence of this factor insures
convergence of integration in the derivative to follow (in both limits t -+ ±oo).
Recall that the Schrodinger equation in the interaction picture has the integral
form (11.144) ,9

(14.51)

where, we recall,

11fJ/ (t)) = eitHo/nl1fJ(t)} (14.52a)

Y/(t) = eitHo/nYe- itHo/n (14.52b)

As we wish to apply (14.51) to scattering theory, we stipulate that Y/(t) -+ 0 in

the limits t -+ ±oo. At these asymptotic values, with the interaction vanishingly
small, 11fJ/ (t)} loses its time dependence, and we define

8For further discussion, see E. Merzbacher , Quantum Mechanics, 2d ed., Wiley, New York, 1970,
Chapter 19.
9Ket notation is employed to obtain a relation independent of representations.
786 Chapter 14 Scattering in Three Dimensions

I.., > ft.. • •• --- -.0'.~

... . . .
, x(+) ~
1-
(a)

-I

(b)

FIGURE 14.11 (a) "In" solution . (b) "Out" solution .

(14.53)
which may be identified as free-particle states.
Rewriting (14.51) over the interval to = ±oo, t = 0, and identifying the scat-
tering states

(14.54)
gives the equation

(14.55)

Note that with this choice of interval, 11jJ.(±)}, given by (14.54), represents scat-
tered states in the domain of interaction . Furthermore , as 11/1 (+)} is relevant to the
time interval (-00 :s t :s 0), we may identify it with incoming incident waves,
commonly called the "in" solution . As 11/I(-)} relates to the interval (00 ,0), it is
the time-reversed state of 11/I(+)} and is commonly called the "out" solution. See
Fig. 14.11. In the limit 8 ~ 0 we take 11/I(±)} to be an eigenstate of the total
Hamiltonian and write
(14.56)

Reduction of Interaction Integral

Consider the integrand in (14.55). With (14.52a,b) we write

VI (t) 11/11 (t)} = ei HOI/n V e-ei/i /ne-i HOI / nedlOI/ nI1/l (t)} (14.57)
= ei HOI /nVe- el/l/ ne-iHI /nI1/l(0)}
14.6 Elements of Formal Scattering Theory 787

and note

Substituting this identification into (14.57) followed by replacement of (14.57)

into (14.55) gives

(14.58)

With (14.56) the preceding becomes

(14.59)

For "in" solutions we encounter the integral

G~(+) = -
i li
11 0

-00
dtexp [-t (iHO-iE+ t:)]
li
1
E - Ho + i t:

Simil arly,

GH = ~
E - Ho - it:

Note that without the presence of e the integrals G (±) do not converge. Sub-
stituting these expressions for G (±) into (14.59) gives the Lippmann-Schwinger
equation,

(14.60)

which, as previously noted, is independ ent of specific representation.

Scattering Amplitude Revisited

In Problem 14.11 you are asked to show that the coordinate representation of the
"in" solution to (14.60) assumes the form

1/JK(+)(r ) = ((Jk(r) - ~
2n li 2
f exp(ik lr - r' 1) V (r ') 1/JK+ (r') dr'
[r - r ' l
(14.6 1)
788 Chapter 14 Scattering in Three Dimensions

Here we have made the identifications

(rllp) = Ipk(r) (14.62a)

(rl1/J<+») = 1/Jk<+)(r) (14.62b)
(rIV1/J<+») = V(r)1/J<+)(r) (14.62c)

At large distances from the interaction domain we may write

1 1
--::::::-+ ...
Ir- r 'l r

and

, r ' ) 2 2r · r '] 1/2

klr - r I = kr 1+ ( --;: - -----;:2
[

:::::: kr r·r']
[ ---;:2
1- + ... = kr - k • r "
where we have set
, kr
k=:-
r
Substituting these expansions in (14.61) gives

m eikr
1/Jk<+)(r) = Ipk(r) - - - --(lpkllVl1/J<+») (14.63)
2nn2 r k

Comparison with (14.7) gives the following expression for the scattering ampli-
tude :

(14.64)

In the Born approximation

and (14.64) returns our previous finding (14.36). However, one should bear in
mind that (14.63) is, more generally, an integral equation for 1/Jk<+), solution
to which gives a more accurate expression for the scattering amplitude through
(14.64).

PROBLEMS

14.9 A beam of photons at a given frequency propagates into a medium of atoms with
density n(cm- 3 ) . The cross section for absorption of photons at this frequency by
Problems 789

the atoms is a. If J is incident photon flux, then argue that the decrease in J in the
distance dx due to absorption is

dJ = -KJ dx
where

K = no
14.10 A monochromatic beam of photons at frequency v = 10 14 Hz and intensity
l.4keV/ cm2 s is incident on a gas of atoms of density n = 1018 cm- 3. At the
given frequency the radiation is very near resonance of the atoms . The related tran-
sition dipole moment of atoms in the gas has magnitude 0.4aoe. At what distance
(em) into the gas will the intensity of the beam be e- I times its starting value?
(H int : Use results of the preceding problem .)

14.11 Working in the coordinate representation, employ the Lippmann-Schwinger equa-

tion (14.60) to derive its coordinate representation (14.61).

Answer
Let us label

G± == lim G(±)
8---+0

The Lippmann-Schwinger equation (14.60) may then be written

To obtain the coordinate representation of this equation, we operate on it from the

left with (r ] to obtain [see (14.62)]

1frk(±)(r) = !Pk(r) + (r IG ± VI1/r(±») == !Pk(r) + I±

which serves to define the interaction term I±. Developing this term, we obtain lO

We recall I I

(r lk ) = __l_e ik •r
(271' )3/4

and

WHere we employ the spectral resolution of unity (see Problem 11.1).

II Note that this form gives the proper normaliz ation

f dk (rlk}(klr/} = ~(r - r')

See (C.II).
790 Chapter 14 Scattering in Three Dimensions

whence

Thus we obtain

I± = --3
1 f f dr' dk
ik o(r - r' )
e -2 (r' IVI1Jr(±»)
(2Jr) E - (l1h 12m ) ± i f

Consider the k integration,

where

~r == r- r '
11,2/,2
R± = E- -2m
-±if

With

f dk = 2Jr loo [II

2
k dk dJ.1

koM
J.1 == cos e = ----
Ik oMI
integration over J.1 gives

1 - 1 - -
1 -I
dJ.1exp( ikMJ.1) = -;=--[exp(ik
ik Sr
M) - exp(-ik ~r)l

As kl R± is an odd function oi]: we find

<I>± = 2Jr roo dH2 ~ [exp(ik M) - exp(-ik M)]

Jo i k Sr R±

= 2Jr
i ~r
1 00

- 00
dkki M
R±
f

Next we set
2 2
1i k
E=-
2m
which, by conservation of energy, is the same as the free-particle energy of the
incident wave, <tJk(r) . We obtain

2m R± = (k2 _ k2 ± i 2mf)
1i2 1i2

= (k - k ± is )(k + k ± is )
Problems 791

Imk

-k +if k+i"f
• •
Re k

FIGURE 14.12 The contour for the integration in <t>± . The pole at k + is' contribute s to
1/1(+) and the one at -k + is' contributes to 1/1(-) .

where s == ms/ti 2k. Thus

t 2m ( 1 /2 1/ 2 )
R± =- ti2 k - k =f is + k + k ± is
We are now prepared to integrate over k by contour integration . As the integrand
contains the factor exp ik Sr , it must be closed in the domain Irn k > 0, that is, the
upper half k plane. For <t> + there is a pole at

k =k+ie
whereas for <t> _ there is a pole at

k = -k + is
See Fig. 14.12.
Passing to the limit e --+ 0, we obtain

<t>± 27l' -27l'

= --- 2m i (1)
- e
± I'k I..>r
A

iM2 ti 2
47l'2 m e±ik t>r
= ~~
whence

which when inserted in our starting relation

returns (14.61).
792 Chapter 14 Scattering in Three Dimensions

14.12 Is (14.61) a valid relation for inelastic collisions-for example, an ionizing collision
in which an electron is emitted from an atom due to, say, electron scattering?

Answer
As the energy of the incident electron is not conserved (it loses energy in releasing
the bound electron), one cannot equate energy of the scattered electron to its incident
free-particle value, and the derivation in the preceding problem is invalid.
CHAPTER

15 Relativistic Quantum Mechanics

15.1 Preliminary Remarks

15.2 Klein-Gordon Equation
15.3 Dirac Equation
15.4 Electron Magnetic Moment
15.5 Covariant Description

In this chapter we present a description of relativistic quantum mechanics. The

chapter begins with a review of basic relativistic notions and continues with
derivation of the Klein-Gordon equation relevant to relativistic bosons. Incor-
porating elements of this equation, the Dirac equation appropriate to fennions
is obtained. Plane-wave solutions to this equation for a free particle are con-
structed, components of which are shown to correspond to "spin up" and "spin
down" wavefunctions . The energy spectrum of these solutions is noted to have
a forbidden gap and an infinite sea of negative values. These properties imply
the existence ofpositrons. The Dirac equation is then shown to imply the correct
magnetic dipole moment of the electron. The chapter continues with a derivation
of the four-dimensional spin operator in the Dirac formalism and closes with a
briefintroduction to the covariant formulation of relativistic quantum mechanics.

15.1 • PRELIMINARY REMARKS

Postulates, World Lines, and the Light Cone

The two postulates of special relativity are as follows :

1. The laws of physics are invariant under inertial transformations.

2. The speed of light is independent of the motion of the source .

The first postulate states that the result of an experiment in a given inertial
frame of reference is independent of the constant translational motion of the sys-
tem as a whole. An inertial frame is one in which a mass at rest experiences no
force . Thus there is no absolute frame in the universe with respect to which motion

793
794 Chapter 15 Relativistic Quantum Mechanics

ct

x
FIGURE 15.1 The world line of a particle moving in one direction from rest.

of an arbitrary inertial frame is uniquely defined. Only relative motion between

frames is meaningful.
Concerning the second postulate, consider a light source fixed in a frame S.
The frame moves relative to the observer in a frame Sf with speed v. The observer
measures the speed of light c, independent of the speed v. This situation is evi-
dently equivalent to one in which Sf moves relative to S with speed v. Thus the
speed of light is independent of the motion of the receiver, as well as that of the
source. This conclusion is alien to our intuitive picture of either wave or particle
motion.
Einstein defined an event as a point in space-time coordinates . The locus of
events of a particle is called the world line of the particle. For one-dimensional
motion, the world line of a particle is a curve in (x, ct) space (see Fig. 15.1).
Of particular interest in the study of relativity is the concept of the light cone.
This is the world line of the leading edge of a light wave stemming from a source
switched on at a given instant at a given location. The notion of past and future
may be defined with respect to the light cone (see Fig. 15.2).

ct

FIGURE 15.2 Light pulse is initiated at O. Points in the domain marked "Elsewhere"
cannot be reached from 0 at speeds less than or equal to c.
15.1 Preliminary Remarks 795

Four-Vectors
An event at a given location x and time t may be described by the four-
dimensional vector

x= (x, ict) (15 .1)

which is called e four-vector. The momentum four-vector is given by

(15.2)

Here we have written

(15.3)

for the total energy of a particle (in the absence of potential) of rest mass m with
kinetic energy T . The parameter y is written for

2 1 v
Y == 1- f32' f3==- (15.4)
c

Note that y increases monotonically from 1 at f3 = 0 to 00 at f3 = 1. The rela-

tivistic momentum three-vector is given by

dx
p = ymv, v=- (15.5)
dt
where t is time measured in the lab frame.
Let us write Pf.L for the components of the four-vector p. Thus
iE
P4=-, (15.6)
c
In this notation we may write

(15 .7)

where uf.L is the velocity four-vector,

_ Pf.L _ dx., _ •
uf.L - -;; - dr: = xf.L (15.8)

and r denotes proper time . This is the time measured on a clock attached to the
moving particle (discussed further below) .
Three other important four-vectors are

J= (J, icp) (l5.9a)

A= (A, i<l» (15.9b)
796 Chapter 15 Relativistic Quantum Mechanics

(l5.9c)

In these expressions , J is current density, p is charge density, A is vector po-

tential, <I> is scalar potential, k is a wave vector, and w is frequency.

lorentz Transformation
Consider that a frame S' moves at constant speed v in the z direction with respect
to a frame S, as shown in Fig. 15.3. It is readily shown from the two postulates
above that, if At! is a four-vector in S, an observer in S' observes the compo-
nents

(15.10)

where
o

i~)
I 0
L = 0 I o (15.11)
o 0 Y
(o 0 -iyfJ
[In (15.10), we have employed the "Einstein convention" in which repeated
indices are summed (from I to 4).] Note that L is orthogonal; that is,

(15.12)

where, we recall, I is the transpose of L. Note further that

detL = I (15.13)

(see Problem 11.96). Thus L effects a rotation in complex four-dimensional

space. Consequently, the "length" of a four-vector is preserved under a Lorentz
transformation. That is,

A~A~ = At!At! (l5.14a)

x x'

S S'

v
z z'

y y'

FIGURE 15.3 The frame S' moves at velocity v with respect to the frame S.
15.1 Preliminary Remarks 797

Let us show this formally:

A~A~ = L/LvAvL/L).A). = L/LvL/L).AvA).

= L~~L/L).AvA). = ov).AvA). = A).A).

For example, consider the length of the momentum four-vector,

(l5.14b)

This returns the useful relation

(15.15)

For X, we find

(15.16)

Such entities, which remain invariant under a Lorentz transformation, are called
Lorentz invariants. Note in particular that the Lorentz invariant x/Lx/L as given by
(15.16) is a reiteration of the second postulate stated above that the speed oflight
is c in all frames .
The Lorentz matrices (15.11) have the following group property.

(15.17)

where

(15.18)

Note that (15.18) precludes the speed of any object from exceeding c. Consider
that one of the frames moves with relative speed c, so that, for example, fh = 1.
Then (15.18) returns the value 1312 = 1. Furthermore, L(O) = I, the identity
operator, and L -I = L. The Lorentz transformations comprise a group called the
Lorentz group . 1

Length Contraction, Time Dilation,and Proper Time

With relative interframe motion again confined to the z direction, Lorentz trans-
formation of the event four-vector gives

z' )
( ict ' = Y
(I-if3 if3 ) ( z )
I ict
(15.19)

1Equations written in tensor notation, such as (15.8), (l5.14b), and (15.16) , are valid in all coordinate
frames. Equations written in this form are said to be covariant. See Section 15.5.
798 Chapter 15 Relativistic Quantum Mechanics

We find

z' = y(z - f3ct) (15.20a)

ict' = iy( -f3z + ct) (15.20b)

Let a rod of given length !:!.z' lie fixed in a frame that we label S'. The frame
S' moves with speed f3c relative to the frame S. At a given instant, the length
of the rod is measured in the S frame . This means that the locations of the ends
of the rods are measured simultaneously in S. Calling the length so measured
!:!.z = Zb - Za, with tb = t a, (15.20a) gives

1, ,
!:!.z = -!:!.z <!:!.z (15.21)
y -

The rod moving past the frame S is measured to be shortened by an observer in

S. This is the phenomenon of length contraction.
Consider next a clock that is at a fixed location in the moving S' frame (z~ =
z~). To find the manner in which intervals t' on this clock are observed in S, we
write the inverse of (15.20b):

ict = iy(f3z' + ct ')

Choosing z' = 0 gives

t = yt' 2: t' (15.22)

Thus an observer in S concludes that intervals on his clock , t, are longer than those
on the S' clock or that the S' clock is "running slow." This is the phenomenon of
time dilation .
An important parameter in relativity is that of proper time. As noted above,
the proper time of a particle is the time measured on a clock that moves with the
particle . Thus, if t denotes time in the lab frame and r proper time, we write

dt = y d t (15.23)

Comparison with (15.22) reveals that we have identified proper time with the sin-
gle clock fixed in S'. The square differential element, (d-r)2 is a Lorentz invariant.
(See Problem 15.18.)

15.2 • KLEIN-GORDON EQUATION

A relativistic wave equation may be obtained from (15.15) (in which we recall
that p is a three-vector) together with the substitutions

E ~
a
ili-, p ~ iliV (15.24)
at
15.2 Klein-Gordon Equation 799

There results

(15.25)

This equation is called the Klein-Gordon equation and is relevant to the quantum
dynamics of spinless bosons such as, for example, JT mesons. As developed for
the nonrelativistic case (see Section 7.5), expressions for probability and current
densities relevant to (15.25) follow from the continuity equation
op (15.26)
-+V·J=O
ot
To obtain the image of this equation in the present development, we multiply
(15.25) from the left by lfr* and the complex conjugate of this equation from the
left by lfr . Subtracting the resulting two equations gives''

per, t) = -iii-2 (*Olfr olfr*)

lfr - -lfr- (15.27a)
2mc ot ot
fer, t) = J!-(lfr*Vlfr -lfrVlfr*) (15.27b)
21m

Electromagnetic Potentials
The Klein-Gordon equation for a particle of charge q in an electromagnetic field
with potentials A(r, t), <I»(r, t) is obtained as follows. With (15.2) and (15.9) it
is noted that (A, i<1» and (p, i E[c) are both four-vectors . Thus (15.15) may be
generalized to read

(15.28)

Substituting the operator equivalents (15.24) for E and p into the preceding equa-
tion gives the generalized Klein-Gordon equation for a charged particle in an
electromagnetic field. [In this equation and in the following , unless otherwise
specified, H as well as E and p, as given by (15.24), are operators.]
We wish to apply (15.28) to the case of charged particle in a static field <1>(r).
There results

(15.29)

Substituting

lfr(r, t) = u(r)e- i Et / 1i (15.30)

into the preceding equation gives the time-independent equation

2This form of p may be negative so that it is not a valid representation of position probability density.
800 Chapter 15 Relativistic Quantum Mechanics

(15.31)

where E is now a c number (i.e., a constant) .

We wish to ascertain the manner in which this Klein-Gordon equation reduces
to the time-independent nonrelativistic Schrodinger equation . Toward these ends
we set E = mc 2 + E' and assume that E' « mc 2 and Ilq<l>ll « me". In-
serting these inequalities in (15.31) returns the time-independent nonrelativistic
Schrodinger equation

(15.32)

See Problem 15.3.

15.3 • DIRAC EQUATION 3

Suppose one attempts to construct a relativistic wave equation from the time-
dependent Schrodinger equation

a (15.33)
iii at 1fr(r, t) = H1fr(r , r)

Again referring to (15.15) and substituting p = -iliV into (15.33) we find that
the resulting expression is not symmetric in space and time derivatives and there-
fore is not relativistically invariant. Dirac circumvented this inconsistency by con-
structing a Hamiltonian linear in space derivatives. The simplest such form that
maintains Hermiticity is given by

H = ca . p + f3mc2 (15.34)

where the Hermitian, dimensionless parameters a and f3 are to be determined and

p denotes the three-dimensional momentum vector. [The parameter f3 in (15.34)
should not be confused with f3 in (15.4).] Substituting the latter expression into
(15.33) gives

(E - ca· p - f3mc2 )1fr (r , t) = 0 (l5.35a)

which, with replacements (15.24), gives

(iii :t + ihca - V - f3 mc2) 1fr(r, r) = 0 (l5.35b)

This latter equation is the Dirac equation for a free particle . As the particle is
free, the parameters a and f3 must be independent of r, t , p, and E (and commute

3p. A. M. Dirac, Proc. Roy. Soc. (London) All?, 610 (1928).

15.3 Dirac Equation 801

with all these variables). Otherwise the Hamiltonian would contain space-time
dependent energies that would give rise to forces .
The structure of the parameters a and 13 may be obtained from the require-
ment that any solution to (15.35b) is likewise a solution of the free-particle
Klein-Gordon equation (15.25), corresponding to the fundamental energy equa-
tion (15.16). (The converse is not true.) Multiplying (15.35a) from the left by
(E + a . p + f3mc2 ) there results

{E 2 - c2[a/px 2 +alpl + a z2p z2 + (axa y +ayax)pxpy (15.36)

+ (aya Z + a Zay)pyPZ + (a za x + axaz)pzpx]
- m 2c4f32 - mc3[(a xf3 + f3ax)px
+ (ayf3 + f3ay)py + (a zf3 + f3az)pz]}1fr(r, t) = 0

This equation agrees with (15.25), providing

2 2
ax = a y = a z =
2
13 2 = I
[ax, a y ]+ = [ay, a z ]+ = [az, a x ]+ = 0 (15.37)
[ax, 13]+ = [ay, 13]+ = [az, 13]+ = 0

where I denotes the 4 x 4 identity matrix and

[A, B]+ == AB + BA
denotes the anticommutator. It follows that the parameter 13 and the components
of a are operators. As H given by (15.34) is Hermitian, each of four parame-
ters (a, 13) must likewise be Hermitian. The smallest dimension for the operators
(a, 13) to satisfy the anticommutator relations (15.37) and maintain Hermiticity is
four," As these operators appear in the Dirac equation, it follows that the Dirac
wavefunction, 1fr, must likewise be four-dimensional.
As noted in the first equation of (15.37) the squares of all four operators (a, 13)
are unity. Each operator has eigenvalue ± 1. Choosing 13 as a diagonal matrix , we
obtain

13 = (Io 0)-I
(15.38)

where I is the 2 x 2 identity matrix . The anticommutator properties (15.37) are

obeyed by the Pauli spin matrices (Problem 11.50). The simplest forms that obey
the relation (15.37) in agreement with (15.38) and Hermiticity requirements are
given by

4For further discussion, see L. I. Schiff, Quantum Mechanics, 3d ed., McGraw-Hill, New York, 1968,
Section 52.
802 Chapter 15 Relativistic Quantum Mechanics

ax = (0 ° '
ax
ax) ay = (0 ° '
ay
a
y)
az = (0
az ~) (15.39)

The forms (15.38, 15.39) may be more concisely written -

a = (~ ~) (15.40)

As the a matrices are composed of Pauli spin operators, they obey commutation
relations parallel to (11.99), namely,

(15.41a)

With (15.37) there results

(15.41b)

As noted above, 1jJ(r, t) is a four-component column matrix

1jJI(r , t)]
1jJ(r, t) = :
[
1jJ4(r, t)

The Dirac equation (15.35b) is then equivalent to four coupled linear homoge-
neous first-order partial differential equations for the components of 1jJ.

Density and Current Expressions

A continuity equation may be obtained from the Dirac equation (15.35b) by mul-
tiplying it on the left by 1jJ t and multiplying the Hermitian adjoint equation

on the right by 1jJ and subtracting the two equations. There results

(15.42)

It follows that probability and current densities for relativistic fermions are given
by

(15.43)

5Note that with (15.39) the relativistic free-particle Hamiltonian (15.34) is spin-dependent through
the inner product term , a· p . (Recall related discussion in Section 11.7.)
15.3 Dirac Equation 803

This form of p is nonnegative and may be consistently identified as position prob-

ability density. The form of J given by (15.43) may be shown to reduce to (7.107)
in the nonrelativistic limit. (See Problem 15.6.)

Plane-Wave Solutions
Plane-wave solutions to the free-particle Dirac equation (15.35b) are given by

1/I(r, t) = aexpi(k. r - cot) (15.44a)

E = tuo (15.44b)

P = lik (15.44c)

where a is a four-component column vector. The components al> j = 1, .. . ,4,

of a as well as the parameters E and pare c numbers . As in the nonrelativistic
counterpart of (15.44a) , k and E are specified in this plane wave. Substitu-
tion of (15.44) into (15.35b) gives the following matrix equation for the column
vector a.

(lick. a- E + {3mc z)a = 0 (15.45a)

which in tum yields the following set of four algebraic equations for the a coef-
ficients.

(E - mcz)al - Cpza 3 - ct p; - ipy)a4 = 0

(E - mcz)az - c(Px + ipy)a3 + Cpza 4 = 0

(15.45b)
(E + mc z)a3 - cp zaJ - c(Px - ipy)az = 0
z)a4
(E + mc - ctp, + ipy)aJ + cpzaz = 0

These equations have a nontrivial solution only if the determinant of the coeffi-
cient matrix vanishes . There results

(15.46)

which is noted to be in agreement with the fundamental relativistic energy relation

(15.15).
Explicit forms of the aj coefficients correspond to choices of the sign of the en-
ergy. For each such choice there are two linearly independent solutions . In the fol-
lowing , the four sets of coefficients are written as column vectors and are labeled
it),
1/1 corresponding to (+) or (-) energy and "spin up" t and "spin down" {,. .
(Motivation for this spin notation is given below.)
804 Chapter 15 Relativistic Quantum Mechanics

(15.47a)

cpz/ 11_ ]
.,,(-) _ C(Px + ipy)/l1_
'i't - 1 ' (15.47b)
[
o
where

(15.47c)

With N p representing the normalization coefficient of these eigenstates, we write

for, say, .,,(+)
'i't

(1/J(+)I1/J(+))
t t
= N p 2 [1,0, cPz, c(Px - i PY ) ]
11+ 11+
[ ~
cpz/11+
] = 1

ct p; + ipy)/l1+
(15.48a)

It follows that

(15.48b)

(See Problem 15.8.)

Relation to Electron Spin

To demonstrate the consistency of spin-up, spin-down notation of the plane-wave
states (15.47) we note the following . Nonrelativistic energy of a free particle is
positive . It follows that the 1/J(+) functions relate to the nonrelativistic limit. This
limit is described by the condition vI c ~ O. The two lower entries of 1/J behave i1)
as cpll1+ ,...., vic ~ o. It follows that in this nonrelativistic limit one may write

1/Jt)(r, t),...., G) expi(k. r- wt)

(15.49)
1/Jl+\r, t) ,...., (~) exp i (k . r - wt)
15.3 Dirac Equation 805

FIGURE15.4 An energy gap of width 2me2 = I MeV exists in the Dirac theory of the
electron .

These plane waves were encountered previously in Section 11.7 relevant to the
1
propagating of free spin electrons with "spin up" and "spin down," respectively.
We may conclude that the notation of (15.47) is consistent.

Electrons and Positrons

Eigenenergies of the plane-wave solutions (15.47) are given by

E+ = +(e 2 p2 + m 2e4 ) 1/2 2: me 2

(15.50)
E_ = _(e 2 p2 + m 2e4 ) 1/2 :s -me 2

Thus, the energy spectrum of a free particle in the Dirac theory is continuous and
ranges from -00 to +00 except for a forbidden gap of width

2me 2 = 1.02 MeV

about the value E = O. The negative-energy states are assumed to be com-

pletely filled. Because of the exclusion principle there cannot be any transitions
of negative-energy electrons . (See Fig. 15.4.) A manner of interacting with this
"Fermi sea" of negative-energy electrons is through photon interaction . Consider
a photon of energy E w > 2me 2 that excites an electron in the Fermi sea to positive
energy. In so doing, a "hole" is left in the Fermi sea. This hole permits a second
negative-energy electron to make a transition to it, leaving a hole elsewhere in the
Fermi sea. As in semiconductor physics (Section 12.9), this hole acts as a posi-
tive charge and is called a positron." Thus, absorption of a photon in the Fermi
sea of energy in excess of 2me 2 leads to the creation of an electron-positron pair.

6The positron was first observed by Carl Anderson in a cloud-chamber experiment [Phys. Rev. 41,405
(1932»). (The closely alliedposilronium atom is discussed in Problems 12.32 and 13.56.)
806 Chapter 15 Relativistic Quantum Mechanics
, , , , , , , ,
I I / / / I / /
, /
I
/
/
I
/
/
,/
/
/
/
/
/
/

/ / / / / / / /

e .:
/ I I / / / /
/ / / / / I /

/ / / / / / /

I
/
e /
+/
,
/
/
I
0 /
/

/ / / / I / /
/ / I / / / /
/ / / / / / /
/ / I / I I I

(a) (b)

FIGURE15.5 (a) Electron-positron creation through photon absorption. (To ensure con-
servation of energy and momentum, this event must occur in the presence of a third body,
which carries away momentum.) (b) Electron-positron annihilation and photon emission.
This is a composite transition in which the electron first emits a photon and then annihilates
with a positron with emission of a second photon . (See Fig. 15.6.)

Now suppose a vacancy exists in the Fermi sea. A positron-energy electron can
fall to this hole and fill the Fermi sea. The decay energy of this electron-positron
annihilation is carried away in two photons . (See Fig. 15.5.)
A revealing diagrammatic description of these events may be obtained from the
plane-wave form (15.44) . A charge conjugation operation on the wavefunction of
an electron changes it to that of a positron. With the preceding discussion we note
that such transformation is effected by the change E --+ - E, which gives

1/!pos(r, t) = a(+) expi(k. r + wt)

The difference between the phase of this function and that of the original electron
is the change t --+ -t. This observation was employed by R. P. Feynman to sug-
gest that in a space-time diagram depicting interacting electrons and positrons, it
is consistent to view the positron trajectory moving backward in time. An exam-
ple of this type of diagram is given in Fig. 15.6 for electron-positron annihilation
which, as noted above, is a composite event."

15.4 • ELECTRON MAGNETIC MOMENT

Two-Component Column Wavefunctions

The Dirac equation for a particle of charge q in an electromagnetic field described

by the vector and scalar potentials, A(r) and <I>(r) (where B = V x A and E =
- V~) stems from generalization of the Hamiltonian (15.34).
7For further discussions, see E. G. Harris, A Pedestrian Approach to Quantum Field Theory, Wiley-
Interscience, New York, 1972.
15.4 Electron Magnetic Moment 807

x
FIGURE 15.6 Feynman diagram for electron and positron annihilation. Time, t, in this
graph is nondimensional. The two vertices in this diagram indicate that the interaction is a
composite event. Note that the positron is depicted as propagating backward in time.

H = ea • (p - ~A) + [3me 2 + q<t>(r) (15.51)

The related time-independent Dirac equation is given by

[ea. (p - ~A) + [3me 2 + q<t>] 1ft = E1ft (15.52)

where

(15.53)

(and, say, a . p is a sum of 4 x 4 matrices) . The off-diagonal, symmetric property

of a (15.40) leads to separation of (15.52) into the coupled equations

[e(p-~A),uJV+(me2+q<t»W=EW (15.54a)

[e (p - ~A). uJ W - (me2 -q<t»V = EV (15.54b)

where

(15.55)

(and, say, p' o is a sum of 2 x 2 matrices). Solving for V in (15.54b) in terms of

W gives
808 Chapter 15 Relativistic Quantum Mechanics

v= e(p - (qle)A) • U] W (15.56)

[ E - qct> + me2

Inserting this expression into (15.54a) gives

2
e [ (p - ~ A) . u( E - q ct> + me2 ) -I (p - ~ A) . u] W
+ (me2 + qct» W = EW (15.57)

Nonrelativistic Limit
As noted from (15.56), in the nonrelativistic limit (vic -+ 0), IIVII is negligible
compared with IIWII. In this same limit, as noted above (15.32), E ' = E - me2
and Ilqct>11 are far smaller than the rest energy me" . This observation permits the
expansion

(E - qct> + me)-
2 I = (E ' - qct> + 2me)-
2 1= -1-2 ( I - E' - qct> + ... )
2me 'lmc ?
(15.58)

Keeping the leading term in the preceding expression, (15.57) reduces to

I ([p - (q~~A]. u)2 + qct> I W = E'W (15.59)

To further simplify this equation we recall the vector operation relations described
in Problem 11.101, with which (15.59) reduces to

_1
[ 2m
(p _ ~A)2
e
_ !LS . B + qct>] W = E'W
me
(15.60)

where we have set

Ii
S= -u (15.61)
2

Equation (15.60) is the nonrelativistic time-independent Schrodinger equation of

a particle of charge q and spin S in a magnetic field B and scalar field ct>. It implies
a spin dipole moment of value

qS
p.=- (15.62)
me

Setting q = lei, the preceding expression for p. agrees with measured values [see
(11.86)] .
15.4 Electron Magnetic Moment 809

Spin Operator in Dirac Formalism

The 4 x 4 matrix representation of the Pauli spin operator in Dirac analy sis is
given by the form

(T'=(~ ~) (15.63)

(T
where denotes the Pauli spin matrices. To demonstrate this property we work
in the Heisenberg picture and the Dirac Hamiltonian of a particle in a central
field VCr) .

H = ca · p + {3m c2 + VCr) (15.64)

We note first that orbital angular momentum L = r x p is not a constant of motion

for the Hamiltonian (15.64) . Thus , as L commutes with VCr) , we write

(15.65)

With the relation (Problem 15.9)

[L, a . p] = ina x p (15.66)

(15.65) reduces to

dL
- = ca x p =1= 0 (15.67)
dt
We may conclude that orbital angular momentum is not a constant of the motion
for the Dirac Hamiltonian (15.64) . To discover the angular momentum constant
we write

(15.68)

which obey s standard angular momentum commutation rules . If dJ jdt = 0, then

J may be identified with total angular momentum and(nj2)Q = S, with intrinsic
spin angular momentum. With (15.67) we note that one must find

dQ 2c
-=--axp (15.69)
dt n
Consider the matrix a x a whose time derivative is given by

d i
-(a x a) = --[(a x a) , H]
dt n
ic 4i c
= -"h[(a x a), a· p] = - n a x p (15 .70)
810 Chapter 15 Relativistic Quantum Mechanics

(see Problem 15.10). It follows that

i
Q = --a x a (15.71a)
2
and
-S= -Q
n (15.71b)
2
Note that
dJ dL n dQ
-dt = -dt + -2 -dt = ca x p - ca x p =0
We may conclude that the consistent expression for the spin operator in the Dirac
formalism is given by (15.71). Explicit components of the 4 x 4 matrix S are
obtained from the rules (15.37) and (15.41). Thus, for example,

- in in n
Sz = --(axay - ayax) = - - axa y = -a z
4 2 2

Similar results pertain to Sx, Sy. These findings are summarized by the form

(15.72)

which is the appropriate four-dimensional spin operator in the Dirac formalism.

Note in particular that the Dirac equation for a free particle (15.35b) does not
include the diagonal spin operator S, but the off-diagonal symmetric a operator
(15.40). As noted above, the off-diagonal, symmetric nature of the operator a
generates separation of the Dirac equation into coupled equations for a pair of
two-component column-vector wavefunctions (15.54) in which the standard 2 x 2
Pauli spin matrices emerge.

15.5 • COVARIANT DESCRIPTION

In this concluding section we present a brief introduction to a covariant descrip-

tion of relativistic quantum mechanics.

The Dirac Equation

To obtain a covariant Dirac equation we introduce the y matrices

Yk = -i{3ak, k = 1,2,3
(15.73)
Y4 = {3, J1- == mcfli
15.5 Covariant Description 811

[The parameter y denoting these matrices should not be confused with y defined
in (15.4) .] The Dirac equation (15.35b) may then be written in the covariant form
(with the Einstein summation convention assumed)

( yv ~
oX v + Jl) \{I = 0 (15.74)

[To demonstrate the equivalence of this equation and the Dirac equation (15.35b),
multiply (15.74) from the left with f3 and recall f32 = I.] To examine the Lorentz
invariance of (15.74), we introduce the transformation

\{I' = A \{I (15.75)

where A is a unitary 4 x 4 matrix (Problem 15.19) that transforms the four-

column-vector wavefunction \{I in the S frame to the wavefunction \{I' in the S'
frame. The y matrices remain unchanged in this transformation. We note further
that

(15.76)

where L VK is the Lorentz transformation (15.10). Our demonstration of the

Lorentz invariance of the Dirac equation starts by writing (15.74) in the S' frame:

(15.77)

Multiplying this equation from the left with A-I gives

(15.78)

Setting

(15.79)

in (15.78) returns the Dirac equation (15.74) in the S frame, thereby establishing
the Lorentz invariance of the equation . Note that (15.79) indicates the manner in
which the y matrices transform under a Lorentz transformation.
The y matrices (15.73) play an important role in relativistic quantum mechan-
ics. We note the following anticommutation relations

y} = I (15.80)

where I denotes the 4 x 4 identity matrix .

812 Chapter 15 Relativistic Quantum Mechanics

Covariant Conservation Equations

The covariant structure of the Klein-Gordon equation (15.25) is given by

(15.8Ia)

where Pv represents the quantum four-momentum vector

. 0
Pv = zli- (15.8Ib)
oX v
The corresponding form of (15.8Ia) for a particle of charge q in an electromag-
netic field is given by

(15.82)

The covariant form of the continuity equation that follows from the Klein-Gordon
equation (15.82) is given by

sr;
-=0 (15.83a)
oX v
Jv = - Ii . ( \II * -0\11 - \11--
0\11*) - -Av\ll
e * \II (15.83b)
2mz oX v oXv me

This four-current vector is in accord with the basic form (15.9a). In the limit that
A v = 0, (15.83b) returns the two relations (15.27).
Let us derive the covariant form of the continuity equation which follows from
the Dirac equation (15.74) . First consider the Hermitian adjoint of (15.74) given
by

(15.84a)

Here we have recalled that X4 is imaginary . Multiplying the latter equation from
the right by Y4 and employing (15.80) gives

(15.84b)

where

is labeled, the "Dirac conjugate" of \II. Multiplying the right equation in (15.84b)
by \II and adding the resulting equation to that obtained by multiplying the Dirac
equation (15.74) by \II gives an equation which when multiplied by the con-
Problems 813

stant i c gives the normal form of the covariant continuity equation appropriate
to fennions .

(15.85)

This equation implies the four-current

(15.86a)

with related position probability density

(15.86b)

and current density

(15.86c)

or, equivalently,

J= cqJ'aqJ' (15.86d)

Equations (15.84b through d) indicate that the covariant continuity equation

(15.85) is equivalent to the noncovariant continuity equation (15.42) .
Additional discussion on covariant relativistic quantum mechanics may be
found in the cited references/ and in Problems 15.19 et seq.

PROBLEMS

15.1 (a) Evaluate the squared lengths of each of the three four-vectors (15.9).
(b) What form do each of these squared lengths assume under a Lorentz transfor-
mation to a primed frame?
(c) What is the name given to the squared length of a four-vector?

15.2 Show that the Klein-Gordon equation (15.25) has a plane-wave solution,

1/!(r, t) = exp i (k • r - wt)

where

15.3 Obtain the first-order relativistic correction to the time-independent Schrodinger

equation (15.32) from the Klein-Gordon equation (15.31) .

8J. D. Bjorken and S. D. DreIl, Relativistic Quantum Mechanics , McGraw-Hili, New York, 1964,
Section 2.4; V. B. Berestetskii, E. M. Lifshitz, and L. P. Pitaevskii, Relativistic Quantum Theory,
Pergamon, New York, 1979, Chapter III; S. Flugge, Practical Quantum Mechanics , Springer, New
York, 1974, Problems 192, 193, 199; A. S. Davydov, Quantum Mechanics , 2d ed., Pergamon, New
York, 1965, Section VIII, 61; F. Constantinescu and E. Magyari, Problems in Quantum Mechanics,
Pergamon, New York, 1971, Chapter XII.
814 Chapter 15 Relativistic Quantum Mechanics

15.4 What is the dimension of the relativistic wavefunction.wrr, t) , for a spin ~ particle?

15.5 What are the probability and current densities (P. J) for the plane-wave state, 1/11-)?
15.6 Show that the relativistic expression of current density J, (15.43) , reduces to (7.107)
in the nonrelativistic limit , vic -* O.

15.7 Show that the four free-particle eigenstates (15.47) of the Dirac equation are or-
thogonal.

15.8 Evaluate the normalization constant, N~-) , corresponding to the normalization,

(1/1+-)11/1+-») = 1.
15.9 Demonstrate the validity of the commutator relation (15.66) .

15.10 Demonstrate the validity of the third equality in (15.70) .

15.11 Show that the two equations (15.54) for the Wand V are equivalent to the single
Dirac equation, (15.52).

15.12 Evaluate components of the 4 x 4 matrix. Sx. explicitly from (15.71) , where S is the
spin operator in the Dirac formalism. Show that your result agrees with the general
form (15.72).

15.13 (a) Write down the wavefunction for a spin ~ free particle propagating in the z
direction with "spin up" and positive energy.
(b) Employing results of Problem 15.12 and part (a) of this problem, evaluate the
expectation, (1/1 t) 1S 1/1t \ for the particle in this state.
x

15.14 In Section 15.1 we encountered "lab time" and "proper time ." Which of these is
the time appearing in the Klein-Gordon equation (15.25) and the Dirac equation
(15.35b). respectively?

15.15 Consider the Feynman diagram of Fig. 15.6. Suppose the time axis is labeled ct.
where c is the speed of light. Which components of the diagram should be changed
and in what manner should these changes be made? Explain your answer.

15.16 Introduce the two- and four-dimensional unit matrices. lz and [4 , respectively.
(a) Employing [4, rewrite the Dirac equations (15.45a) and (15.52) and the Hamil-
tonian (15.51) in a more self-contained manner.
(b) Employing h , rewrite (l5 .54a, b) in a more self-contained manner.
15.17 (a) Rewrite the set of equations (l5 .45b) as a 4 x 4 matrix times the a column
vector.
(b) Evaluate the determinant of this matrix and verify (15.46).

15.18 Show that (dr)2 is Lorentz invariant.

Answer
We may set
Problems 815

which is evidently a Lorentz invariant. Let us evaluate the right side of this equation
in a frame S' in which the particle is at rest. In this frame components dx~ are
(0,0 ,0, icdt'Y, in which case (dr)2 = (dt')2 , and again we find that r is the time
measured on a clock moving with the particle.
15.19 Argue that the A operator introduced in (15.75) is unitary and describe the geomet-
rical significance of this property.

Answer
The Lorentz transformation rotates the four-vector, x , in 4-space. It follows that in
the transformation A 1fr = 1fr' (x/), x' is rotated from x . Recalling the discussion in
Problem 9.17, we may conclude that for an increment of rotation in 4-space , ocjJ ,
one may write A = exp i ocjJ • G, where G is the generator of the rotation. It follows
that A is unitary.
15.20 Verify explicitly that axay = -ia z .
15.21 Establish the following properties of the Y matrices:
(a) Yv2 =1
(b) YvYv = 41
(c) YAYvY),. = -2yv
(d) Y),.YvYKY),. = 40 KV
(e) YKYVYAY/lYK = -2Y/lY),.Yv
(0 Tr(YKYv) = 40 KV

15.22 It is possible to construct 16 linearly independent 4 x 4 products of the Ya matrices

that span the space of 4 x 4 matrices . These are given by the following forms:

r~~ == iYvY/3
(2) _
r(l) == I, rv = Yv,
(5)
r(4) == Yl Y2Y3Y4 == Y5, r; == Y5Yv
These matrices play an important role in construction of Lorentz covariants. In the
following problems, r(n) denotes any of the preceding r matrices.
(a) Argue that the space of 4 x 4 matrices is spanned by 16 independent 4 x 4
matrices.
(b) What is the matrix representation of Y5 ?
(c) Show that (r(n»2 = ±I
(d) Show that except for r(l), for each r(n) there exists a r(m) such that

[r(n), r(m)l+ = °
(e) With part (d) show that (again except for r(l) , Tr r(n) = 0.
(0 Show that the product of any two elements of the set of r matrices is propor-
tional to a third element of the set:
r(n)r(m) = ar(q)
where a = ±1 or ±i, and a = +1 if and only if n = m.
(g) If r(n) f= r(l), there exists r(K) such that r(K)r(n)r(K) = _r(n) .
816 Chapter 15 Relativistic Quantum Mechanics

(h) Show that [Yoo, Ysl+ = O.

(i) Show that if r(n) =f. r(l), Tr r(n) = o.
U) Suppose a set of numbers an exists such that

L:>nr(n) =0
n

Show that there is at least one value of an that vanishes, thereby establishing
the linear independence of the r(n) matrices.

Answers (partial)
(a) The indices of a 4 x 4 matrix each run from I to 4. It follows that 16 independent
4 x 4 matrices span the space of such matrices. As part (g) establishes that the
16r(n) matrices are independent, it follows that they are a basis for this space.

(b) Ys = (~I ~)
(c) This property follows from (15.80).
(g) Multiply both sides by r(K) .
(i) Trr(n) = Trr(n)(r(m»2 = -Trr(m)r(n)r(m)

= +Trr(m)r(n)r(m) = 0
The second equality follows from part (d). The third equality follows from
TrAB =TrBA .
(j) Multiply the given equation by r(n) =f. r(I) , take the trace, and use part (i).
This gives an = O. Now multiply the given equation by r(l) to find al = O.
CHAPTER

16 Quantum Computing

16.1 Binary Number System

16.2 Logic Gates
16.3 Turing Machine and Complexity Classes
16.4 Qubits and Quantum Logic Gates

16.1 • BINARY NUMBER SYSTEM

Constructing Binary Numbers

In the decimal number system , the value of a number is represented by the respec-
tive locations and values of digits. Thus, for example , the number 604.23 means

2 3
604.23 = 600 + 0 + 4 + 10 + 100 (l6.1a)

or, equivalently,

604.23 = 6 x 102 +0 X 101 + 4 x 10° + 2 x 10- 1 + 3 x 10-2 (l6.1b)

In this scheme, lOis the base and positions of numbers in the sequence represent
powers of 10. Coefficients run from 0 to 9.
In the binary number system the base is 2 and coefficients are 0 or 1. Here is a
list of low-valued numerical equivalents in the binary system :

Binary No. 00 01 10 II 100 101 110 III 1000 1001 1010 lOll
Decimal No. 0 I 2 3 4 5 6 7 8 9 10 II

For example, the binary representation of lOis

10 = 1 x 23 + 0 x 22 + 1 x 2 1 + 0 x 2° == 1010 (l6.2a)

The number 48.5 is given by

48 .5 = 1 x 25 + 1 X 24 + 0 X 23 + 0 x 22 + 0 x 2 1 + 0 x 2° + 1 X r l

(l6.2b)

817
818 Chapter 16 Quantum Computing

or equivalently,

48.5 = 110000.1 (l6.2c)

Note that as in the decimal number system , digits to the left of a decimal point are
greater or equal to 1 with the first space to the left of the decimal point represent-
ing the power, 2° and the first place to the right of a decimal point representing
the power 2- 1 .
In electronic logic circuits, the numbers 0 and 1 (bits) relate to two separate
voltage levels. Typically, the number 0 corresponds to a voltage near zero and the
number 1 to a voltage near 5 volts.
In general , the binary representation of a number less than one is obtained as
follows . Consider the number 0.3.
Step J: Carry out a multiplication by 2.

0.3 x 2 = 0.6 (l6.3a)

The zero in this expression gives the first number (to the right of corresponding
decimal point) in the binary representation of 0.3, so that

0.6 ~ 0 x r 1~ 0.0

Step 2: Carry out a multiplication by 2 on the resulting product in the preceding

step.

0.6 x 2 = 1.2 (l6.3b)

The coefficient 1 is the second entry in the binary representation of 0.3, so that

1.2 ~ 1 x r 2
~ 0.01

Step 3: Change 1.2 to 0.2, and repeat the process to obtain

0.2 x 2 = 0.4 (l6.3c)

which corresponds to

0.4 ~ 0 x r 3
~ 0.010

Continuing this process gives the binary representation

0.3 = 0 x 2- 1 + 1 x 2- 2 + 0 x 2- 3 + 0 x 2- 4 + 1 X 2-5 + 1 x 2- 6 (l6.3d)

0.3 = 0.010011 (l6.3e)

This representation corresponds to the decimal value, 0.296875.

The preceding process is described by the mapping

X n+l = 2x nmod(l) (l6.3f)

16.1 Binary Number System 819

where A = B mod (m), iff A = B + km , where k is an integer. Note that x

mode 1) = x if x S 1. Thus for example, in the problem above, X2 = 2 x
0.3 mod(l) = 0.6, which is (l6.3a) .
A number in the decimal number system greater than one may be converted
to one in the binary representation by the following sequence of divisions by 2.
Consider the number 28:

28
- = 14 with no remainder: 0 (l6.4a)
2 '

The number 0 is the entry to the far right (i.e., the smallest entry) of the binary
representation of 28. In the next step the number 14 is divided by 2 to obtain

214 . der : 0
. remain
= 7 , with (I6.4b)

So 0 is the second zero in the binary representation of 28 . Continuing, we find

~2 = 3
'
with remainder: 1 (l6.4c)

3
2 = 1, with remainder: 1 (l6.4d)

~2 = 0
'
with remainder: 1 (l6.4e)

Note that in (I6.4c), for example, 7 = 2 x 3 + 1. The binary equivalent so obtained

is given by

28 == 11100 (I6.4f)

where the first zero on the right corresponds to 0 x 2° and the 1 on the far left to
the value 24 .

Binary Arithmetic
Addition
Binary arithmetic derives from the fact that entries in a binary number represent
the exponential 2n , and that a transition to the left of an entry increases the expo-
nent by 1 and a transition to the right diminishes the exponent by 1. Thus, consider
the addition (in the decimal system, 24 + 24)

11000
11000
110000 (24 + 24 = 48)
820 Chapter 16 Quantum Computing

This result gives the rule: 1 + 1 gives 0 with a carry to the left of 1. Here are
some more examples:

11 11 11
01 10 11
100 101 110

More generally in any binary sum, one adds column by column , carrying where
needed to higher value columns. The rule for the summation, 1 + 1 + 1 = 11,
occurs when a carry-over adds to a column of two 1's, and stems from the relation

(I6.5a)

This rule occurs in the third summation in the preceding example.

Complements and 2's Complement of Numbers

The complement a of a single digit a in a base b is given by
a = (b - 1) - a (I6.5b)

Thus, in the base 2, 6 = 1 and 1 = O. In the base 3, 6 = 2, 1 = 1 and 2 = O.

The complement (or inverse) of a number, x , more than one digit long , is de-
fined as

x+x=III ·· ·

Here are some examples of complements of binary numbers:

0010 = 1101
1010 = 0101
1000 = 0111
1111 = 0000 (I6.5c)

Thus if a is a digit n bits long then

a+ a = (1111 . . . ) (n bits long)

The 2's complement (or negative) of a number a' is defined as

a +a' = 000 · ·· (16.5d)

Thus,

(I 111)' = (0001)

That is,
16.1 Binary Number System 821

1 1 1 1
000 1
000 0
with a carry of + 1. The 2's complement of (l 010) is (0110). Thus if a is a digit n
bits long, then

a + at = (0000 · . . )(n bits long) .

Subtraction
Binary subtraction is accomplished by constructing the 2's-complement of the
number subtracted . Consider the subtraction 1101 - 1010. To obtain the 2's-
complement of 1010, first complement this number to obtain 0101 and then add
1 to obtain
o 1 0 1
000 1
o 1 0
Then add this number to 1101 to obtain
1 0 1
o 1 0
o 0 1
The carry-over 1 bit is deleted in this process but signifies that the sign of
the answer is positive. A carry-over 0 bit signifies that the sign of the answer is
negative. Note that in the decimal system, the original subtraction is 13 - 10 = 3.
This answer agrees with the preceding result (i.e., 0011). Note also that 1010 -
0110 = 0100 with a carry-over 1 bit. (In the decimal system, 10 - 6 = 4.) This
technique of subtraction of binary numbers is employed in computer software.

Multiplication
In multiplication one obtains partial products with the following binary multipli-
cation table:

o x 0 = 0, 0 x 1 = 0, 1x 0 = 0, 1x 1= 1 (l6 .5~)

Here's an example :

1110.1
x 1.01
11101
00000
11101
10010.001
822 Chapter 16 Quantum Computing

Note that as with decimal multiplication, the decimal point is inserted in the
answer. In the decimal system this multiplication is equivalent to 14.5 x 1.25 =
18.125.

Division
Division proceeds as with the decimal system. Consider, for example, the ratio
19.25/3 .5, which in binary numbers is given by 10011.01/11.1.
10 l.l
1 l.l ) 1 0 0 1 1.0 1
111
1010
.z.u
III
UJ.
000
The first subtraction of III from 1001, with the 2's-complement, is equiv-
alent to the addition of 001 and 001 to give 10. The 0 multiplication is sub-
sumed in the calculation and brings down the two digits, 10. We conclude that
10011.01/1l.l = 1Ol.l = 5.5.

Information Measure
Consider a message SI elements long. Each element is chosen from a box of b
symbols. Another message is S2 elements long, etc. There are
b
n = ITs; (l6.6a)
;=1

distinguishable messages in all. If two messages carry information a I and a2 units,

respectively, when combined, they have the content of al +a2 units. In this manner
one speaks of information as being an additive function. With (16.6a) this property
is satisfied by the form
b
1= log2 n = L log2 s 1 (16.6b)
;= 1

The fundamental unit of information is called a bit (contraction of binary and

digit) and, as noted above, in binary code is equal to either 0 or 1. Eight bits
comprise a byte.
The result (l6.6b) follows provided the S cases are of equal probability. For
example, this is not the case for letters of the alphabet, chosen at random from a
word. Let the b symbols {s;} have a priori probabilities, PI , P2, P3, ... , pi: Let
each element in this message be written as a fraction whose denominator is the
least common denominator of the set, that is, PI = ci j m, P2 = C2 / m .. . . ,Pb =
Cb/ m. The average information content of this sequence is given by the following.
If the symbols are all distinct, the information content of each term is, I = log2 m,
 16.2 Logic Gates 823

by definition. The indistinguishability from other elements of its class in the se-
quence is equal to the information content of a symbol required to distinguish
members of that class, that is, I = log2 ci , Symbols of the class k occur with
probability Ck /m, and each carries I = (log, m - log, Ck) bits, which gives the
form
b c b
(l) = L ~(log2 m -log2 Ck) = - L Pk log2 Pk
~Im ~I
(l6.6c)

This is the average information of a message composed of b symbols with distinct

a priori probabilities {Pk} .

PROBLEMS

16.1 Convert the following binary numbers to their decimal equivalents: (a) 0001110, (b)
111000, (c) 10.00011. [In (a) and (b), recall that the entry to the far right of these
expressions corre sponds to 2° .]
16.2 Convert the following decimal numbers to their binary equivalents: (a) 0.7, (b) 203,
(c) 456.

16.3 Rewrite the addition 7 + 5 = 12 in binary numbers.

16.4 (a) Rewrite and calculate the subtraction (18 - 5) in binary numbers . (b) Show that
the binary number you obtain has the value 13 in decimal numbers .
16.5 (a) What is the value of the binary product 1110.1 x 1.001 in binary numbers ? (b)
What is value of this product in decimal numbers ?
16.6 What is the representation and value of the fraction 181/21 in binary numbers?
Calculate the result to four decimals .
16.7 What are the complements of (11011) . (11101) . (10001)?
16.8 What is the 2's complement of the binary number (0101O1)?
16.9 Describe the difference between a bit and a byte.
16.10 Lets denote the spin of a particle and let {n2} denote the elements of base 2 numbers
and {n3} denote the elements of base 3 numbers . What respective values of s are
related to: (a) {n2} -!.(b) {n3} - I?
16.11 What is the nature of the following binary numbers ? (a) (xyzlO) , (b) (xyzl)?
Answer
(a) As 2G + 2 = even, (a) represents an even number.
(b) As 2G + 1 = odd, (b) represents an odd number.

16.2 • LOGIC GATES

The NOT Gate

In computer logic gates the output binary numbers 1 and 0 correspond , respec-
tively, to TRUE and FALSE. Suppose , for example that the Boolean variable A
824 Chapter 16 Quantum Computing

A
A-[>o~
(a)
FIGURE 16.1
rn o
1

Equivalent NOT gate and truth table.

(b)
1
0

represents the proposition, 'this circuit is open .' Application of the NOT connec-
tive to this variable produces the propo sition 'this circuit is closed.' Thus, if A
has the value TRUE, (say, 1) then NOT A has the value FALSE (i.e., 0) and vice
versa. The Boolean symbol 1 is referred to as the complement of the symbol O.
Thu s, a NOT operation is a complementation operation. A small circle ('invert
bubble ') at the start of the output line of a logic gate signifies the complementa-
tion of the input proposition. The symbol for a NOT logic gate and its truth table
are shown in Fig. 16.1. A bar over an element indicates the complementation of
that element. The circuit of a NOT connective is an inverter. This element con-
verts a high voltage to a low one, and vice-versa, and inhibit s interaction between
input and output signals.

The AND Gate

If A and B are two logical propositions, then the proposition, Z = A AND B
is TRUE (i.e., output, 1) if and only if both propositions A and B are TRUE
(input 1,1). We note that the AND connective can be applied to more than two
variable s. For example, the function, Z = A AND BAND C AND D is TRUE if
and only if the values, A, B, C, D are TRUE. For the A AND B gate the output is
written A . B or simply AB. The symbol for an AND gate, together with its truth
table are shown in Fig. 16.2. Note that in the truth table , Z is obtained by normal
multiplication .

A B Z=AB

-I~=AB
0 0 0
: __ 0 1 0
1 0 0
1 1 1

(a) (b)

FIGURE 16.2 An AND gate and its truth table.

16.2 Logic Gates 825

A B Z=A+B

:----D>--- Z=A+B 0
0
1
1
0
1
0
1
0
1
1
1

(a) (b)

FIGURE 16.3 An OR gate and its truth table.

The OR Gate
If A and B are two logical propositions, then the proposition Z = A OR B is
TRUE if and only if propositions A or B are TRUE, and Z is FALSE if and
only if A and B are FALSE. The OR connective can be applied to more than two
variables. For example, the function, Z = A OR B OR C OR D is FALSE if
and only if the propositions A or B or C or D are FALSE. The symbol for an
OR gate and its truth table are shown in Fig . 16.3. The output of an OR gate is
written A + B. In these definitions, A AND B represents the intersection, A n B,
of A and B whereas A OR B represents the union, A U B, of these sets . Note that
the symbol for an AND gate includes a smooth output segment whereas that of
an OR gate includes a pointed output segment. Note that in the truth table, apart
from (A, B) = (1, 1), values of A + B are obtained by normal addition.

The NOR Gate

The phrase NOR is a contraction of NOT and OR and the output of a NOR logic
gate is 1 if neither the propositions A nor B are TRUE. A NOR operative is the
complement of an OR operative. The output of a NOR logic gate is written A + B.
A NOR logic gate and its truth table are shown in Fig. 16.4. Note that the symbol
for a NOR gate is an OR gate with an invert bubble attached to the start of the
output that indicates complementation.

A B A+B
0 0 1
0 1 0
1 0 0
1 1 0

(a) (b)

FIGURE 16.4 A NOR gate and its truth table.

826 Chapter 16 Quantum Computing

A B A·B
0 0 1
0 1 1
1 0 1
1 1 0

(a) (b)

FIGURE 16.5 A NAND gate and its truth table.

The NAND Gate

The phrase NAND is a contradiction of NOT and AND and the output of a NAND
gate is 1 if the propositions A or B are FALSE. The NAND operative is the com-
plement of an AND operative . The output of a NAND logic gate is written A . B
or, simply, AB. A NAND logic gate and its truth table are shown in Fig. 16.5.

The XOR Gate

The exclusive OR gate is labeled the XOR gate. As opposed to the OR gate, the
output of this gate is the same as that for an OR gate except for the case that either
A or B, but not both, have the value 1. That is, when A or B is TRUE and at
the same time, both are not TRUE. An XOR gate and its truth table are shown in
Fig. 16.6. The symbol for an XOR operative is EB, which designate s addition mod
(2). (Recall A = B mod (m) , iff A = B + km , where k is an integer.) From the
truth table shown below we can see that for inputs A and B, the output Z obeys
the relation

Z=AEBB=A+AB

Note also that

AB+AB = AEBB

A B A@B
0 0 0
0 1 1
1 0 1
1 1 0

(a) (b)

FIGURE 16.6 An XOR gate and its truth table.

 16.2 Logic Gates 827

From the truth table shown below we see that an XOR logic gate comprises a
summational computer element. Note that the 1 + 1 = 0 entry represents a carry-
over bit.
The preceding properties of logic gates together with Boolean postulates are
listed in Tables 1,2,3. In these we recall that numerical values 1 and 0 are repre-
sentation of TRUE and FALSE, respectively. Thus, for example, the proposition
Z = A OR B, indicates that Z is TRUE if either A or B or both are TRUE. The
numerical equivalent of this statement is 1 + 1 = 1. In Boolean relations, A + B
means A or B, whereas AB means A and B.

TABLE 16.1 Boolean postulates in 0 and 1 TABLE 16.2 Boolean postulates in 0, 1 and A

OR AND NOT OR AND NOT

0+0=0 0 ·0= 0 0=1 A+O=A A ·O=O A=A
0+1=1 0 ·1 = 0 1=0 A+l=1 A ·l =A
1+0=1 1 ·0= 0 A+A=A A ·A=A
1+1=1 1·1 = 1 A+A=1 A ·A=O

Let us demonstrate the relations: (a) A + A = 1 and (b) A . 1 = A. (a) If

A = 0, A = 1, and if A = 1, A = O. (b) IfA = 0 orl , A ·l = A.
The following is a list of Boolean postulates for three sets.

TABLE 16.3 Boolean postulates in A, Band C

Complementation laws: A + A = 1, A . A = 0

Involution: (A) = A
Commutative laws: A + B = B + A, A . B = B . A
Associative laws:

A + (B + C) = (A + B) + C
A . (B . C) = (A . B) . C
Distributive law:

A · (B+C) =A ·B+A·C
A + B . C = (A + B) . (A + C)

De Morgan's laws:

(A +B) = A· B
A·B=A+B

Note that in general, the complement of a complementation operative NQ is

the operative Q .
828 Chapter 16 Quantum Computing

NAND = AND
NOR=OR
The validity of relations among Boolean variables is shown, in general, with truth
tables. For example. consider the relation,
A+A ·A=A
The corresponding truth table is given by
A A ·A A+A·A
1 1 1
o 0 o
which establishes the given relation.
A truth table for the preceding logic gates is shown below. Each row includes
the output corresponding to the input A, B, except for the last column, which
includes only the A input.
A B AND OR XOR NOTA
o 0 o 0 o 1
o 1 o 1 1 1
1 0 o 1 1 o
1 1 1 1 o o
As noted above, A . B (AND) represent the intersection A n B of the sets A
and B, whereas A + B (OR) represents the union A U B of the sets A and B.
These concepts are well described in terms of Venn diagrams. Each point of a
Venn diagram represents a point of the given set. See Figure 16.7.

PROBLEMS

16.12 Employing the rules of Tables 1,2,3 establish the following relation for the Boolean
variables, A, B, C.

(A + B) . (A + C) = A +B . C
Answer
The validity of this relation is demonstrated in the following truth table.

A B C BC A+BC A+B A+C (A + B) (A + C)

0 0 0 0 0 0 0 0
0 0 1 0 0 0 1 0
0 1 0 0 0 1 0 0
0 1 1 1 1 1 1 1
1 0 0 0 1 1 1 1
1 0 1 0 1 1 1 1
1 1 0 0 1 1 1 1
1 1 1 1 1 1 1 1
Problems 829

(a) A · B (b) A + B (c) A . B (d) A + B

~ ~
(e) A +B (f) A . B (g) A . B (h) A· B

FIGURE 16.7 Venn diagrams corresponding to Boolean relations of the sets A and B
in which the shaded area of a diagram represents the outcome of the given process. (DIG-
ITAL ELECTRONICS WITH ENGINEERING APPLICATIONS by SifferlenIVartanian,
© Reprinted by permission of Pearson Education, Inc., Upper Saddle River, NJ).

16.13 (a) What is the Boolean relation corresponding to the equation

Z=AORBORC?

(b) What is the corresponding truth table?

Answer
(a)Z =A+B+ C
(c) The related truth table is given by:

ABC Z
1 1 1 1
110 1
1 0 1 1
1 0 0 1
011 1
010 1
010 1
000 0

Note that this table includes 23 = 8 rows.

830 Chapter 16 Quantum Computing

16.14 How many rows does the truth table have for a logic gate with N input components?
Answer
This is the number of distinct ways of distributing two letters in N spaces, namely,
2N .
16.15 In general, the circuit diagram corresponding to a Boolean logic expression may be
built up from NOT , OR and AND gates. In this context, what is the circuit diagram
corresponding to the Boolean relation, A . B + K + (G . E) = Y?
Answer

A
B

K A·B+K+G ·E=Y

G
E

16.3 • TURING MACHINE AND COMPLEXITY CLASSES

Turing Machine
A Turing machine! contains four principal components:

(a) A program,
(b) A finite state control (like a stripped-down microprocessor),
(c) A tape (that acts as a computer memory),
(d) A read-write tape-head which connects to the tape that is currently readable
(Fig. 16.8) .

The Turing machine tape is a one-dimensional item that extends to infinity

in one direction and includes symbols drawn from an alphabet r . The tape in

Program I Finite State Control

FIGURE 16.8 Basic elements of a Turing machine.

IA. M. Turing, Proc. London Math Soc. 42,230 (1936) .
16.3 Turing Mach ine and Complexity Classes 831

the figure contains four symbols: 0, 1, e-, and b, that denotes a blank space. Tape
squares each contain a symbol drawn from some alphabet r . The symbol c- marks
the left edge of the tape. The read-write tape-head locates a single square on the
tape as the one being assessed by the machine. The machine starts its operation
with the finite state control in a given state and the read-write head at the left-
most tape square. The computer then proceeds in a step-by-step manner according
to a given program, which is a finite-ordered list of program lines of the form
(q, x , s', x', s) . The first parameter, q, is a state from the internal states of the
system. The second parameter is a term from the alphabet, r. In each cycle of
the machine the Turing machine searches for a program line with (q, x) values
such that the current internal state of the computer is q and the symbol being
read on the tape is x . If such a program line is not found, then the internal state
of the machine is changed to % and the machine ceases operation. If the line is
found, then the program line is executed. In this process, the internal state of the
machine changes to s', the x on the tape is overwritten by the symbol x' , and the
tape head moves left, right or stands still depending on whether s is -1 , + 1, or
0, respectively. The importance of a Turing machine to computer analysis lies in
the Church-Turing theorem: Any algorithmic process can be simulated efficiently
using a Turing machine.2

Complexity Classes; Polynomial and Exponential Times

Computational complexity is the area of research that addresses the subject of
finding a measure of the difficulty of computational problems. In this study one
encounters complexity classes. These may be viewed as a set of computational
problems with a common feature. More specifically a complexity class is a col-
lection of languages. A problem is in TIME [f(n)] if a Turing machine exists
that decides if a candidate-string, x, is in the language (i.e., set of strings) in time
O(f(n», where n is the string-length of x. A problem is soluble in polynomial
time if it is in TIME (n q ) , for some finite, q . The collection of all languages
that are in TIME (n q ) , for some finite q, comprise the P complexity class. The
(decision) problem to discern whether a number is prime is a problem in the P
complexity class. Roughly speaking, the complexity class NP includes problems
whose solutions may be readily checked on a classical computer, but whose anal-
ysis is not soluble in polynomial time on a classical computer. Thus, the problem
of factoring an integer into its prime factors (to be discussed below) is a problem
in the NP complexity class. Once a factor is discovered, it may simply be checked
on a classical computer. However, there is no fast (polynomial time) means of
solving the factoring problem on a classical computer whence it is an NP prob-
lem. The class P is a subset of the class NP. The complexity class EXP (for
exponential time) contains all decision problems that may be decided by a Turing
Q
machine running in expon ential time that is, in time O(2n ) , for a constant q .

2For further discussion see : M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum
Information , Cambridge University Pres s, New York (2000).
832 Chapter 16 Quantum Computing

(the computer facility is 'small'), but not necessarily in time. The complexity
class BPP includes problems that can be solved using randomized algorithms in
polynomial time with a bounded error probability. The complexity class BPP ad-
dresses classical computers. It contains languages L with the property that a Tur-
ing machine M exists such that if x E L then M accepts x with probability at least
3/4, and if x If. L, then M rejects x with probability at least 3/4. The complexity
class BQP addresses quantum computers . It denotes the class of languages that
can be computed on a quantum computer in polynomial time, yielding the correct
answer 2/3 of the time. It is known that all problems in P can be solved efficiently
on a quantum computer and that in general, BQP lies between P and PSPACE.
Furthermore, BQP~ PSPACE and BPP ~ BQ which, together with preceding
relations, gives the following sequence of inclusions

BPP ~ BQP ~ PSPACE, P ~ NP ~ BSPACE ~ EXP

Thus, for example, a problem in the class BPP can be solved in the class BQ
but not vice versa.

16.4 • QUBITS AND QUANTUM LOGIC GATES

It has been envisioned that quantum computers will perform at the atomic scale.
The primary difficulty faced by conventional computers at this scale length are
limitations due to dissipation. From thermodynamics, we know that a cyclic sys-
tem that dissipates heat is irreversible. In a conventional logic gate, as described
above, the process is irreversible . Consider for example, an AND, OR, or XOR
gate on two bits. The four possible inputs {OO, 01,10, II} map onto one of two
possible outputs, 0 or 1, and the respective process is not invertible.
In a quantum computer, programs are performed by unitary evolution (Sees.
3.4 and 11.3) of the input state of the system . Since all unitary operators are
invertible, V-I = u', a quantum computation may always be reversed. In the
quantum version of the preceding example, four outputs are required.
As noted in Sec. 16.1, in binary code, either of the two states represented by a
zero or one is called a bit. In quantum computing the analogue of these two states
are two orthogonal quantum states that comprise a quantum bit, or more simply,
a qubit. In this formalism one is interested in wavefunctions that readily translate
into a binary representation. Such, for example, is the case for uncoupled spin
states and spin functions , la) and 1,8) (Sec. 11.6). In binary notation, these are
labeled, respectively, 10) and 11) and comprise two orthogonal basis states. Both a
single classical bit and a quantum qubit can represent, at most, two different num-
bers. However, qubits are distinguished from classical bits in that infinitely many
superposition states may be built up from these two basis states. For example,
consider the single-qubit superposition state

I1/!) = alO) + b11) (l6.7a)

lal 2 + Ibl 2 = 1 (l6.7b)
16.4 Qubits and Quantum Logic Gates 833

Thus there is a state corresponding to each (a , b) value on the unit circle (16.7b).
In general, a quantum computer includes architecture that creates superposition
(binary) states. For our second example, we consider the two-qubit superposition
state,

11Jr} = cooIO}IO} +colIO}II} + clOII}IO} + cllll}11} (16.7c)

With normalization,
1
L ICi,j 12 = I (16.7d)
i ,j=O

the state (16.7c) represents infinitely many quantum superposition states. An m-

bit wavefunction has 2 m different components and exists in a 2m-dimensional
Hilbert space. Thus, numbers that enter in quantum computing typically require
many qubits for representation.
Let us consider logic gates for a single qubit, namely, IO} or II} . With these
states as two-component column vectors, (?) and (b), respectively, the most gen-
eral unitary transformation on these vectors is given by the 2 x 2 matrix,
_ i¢ ( e- i (a + . )/ 2 cos(e / 2) ei (- a + . )/ 2 sinCe/ 2) )
UI) - _ei(a+T )/2 sinCe / 2)
(16.8a)
e e i (a + . )/ 2 cos(e / 2)

For demonstration purposes we set ¢ = a = T = O. Consider first the 'flip'

unitary operation

UrrIO} = II} (16.8b)

Our second example is the operation U -rr / 2 that maps the IO} state to a corre-
sponding superposition state.
I
U- rr/21 0} = ~(IO) + II}) (16.8c)

This mapping may be generalized to a string of m single-qubit states all in the

IO} state. Applying the preceding mapping to each qubit gives a superposition of
every possible qubit string of length m. A schematic of a quantum gate for the
preceding processes is shown in Fig. 16.9.
Let us consider the quantum gate corresponding to the NOT operation on one
qubit. Labeling the operation X, then

Xlk} = Ik E9 I} (16.9a)

11Jr}~ UI) ~UI)I1Jr}

FIGURE 16.9 Schematic for the quantum-gate processes (16.8b,c) in which the unitary
operator Uo maps 11/1) to Uoll/i).
834 Chapter 16 Quantum Computing

\j) - - - - - X - - - - - l j @ l )

Ik)---------Ik)
FIGURE 16.10 Quantum logic NOT gate in which Ij) goes to Ij 611 ). The flow of the
process is from left to right. Note that for j = 0 or 1, Ij 611 ) i= [j). (1 + 1 = 2 = 0 mod
2, 0 + 1 = 1 = 1 mod 2).

\j) - - - - - r - - - - - - lf >

Ik) - - - - - { +
of. } - - - - - Ik@j)

FIGURE 16.11 Quantum logic XOR gate. In this gate, qubits in the final register include
= =
information of the starting state. Thus, for example, if bit values, k 61 j 0 and j 1, then
k= 1.

where, we recall, EB represents addition mod (2). In a two-qubit example , the NOT
operation on Ii) in the state IJ) Ik) is given by

Xli)lk) = Ii EB 1)lk) (16.9b)

This operation is represented by the quantum logic gate shown in Fig. 16.10.
The XOR quantum gate is shown in Fig. 16.11, relevant to the coupling of two
qubits.

Single-Qubit Logic-Gate Matrices

A single qubit state is written I1/!) = alO) + bll}, with normalization lal 2 +
Ibl 2 = 1. Operations on this qubit are 2 x 2 unitary matrices that preserve this
norm. In addition to the Pauli spin matrices (11.82) one has the following three
matrices, important to quantum logic gates .

H=.j2
1(1 1 (16.9c)

The Hadamard gate matrix is labeled H, the phase gate is labeled S and the n 18
gate is labeled T . Note that H = (ax + a z ).j2, and S = T 2 . The rr/8 notation
has its origin in the equivalent symmetric form
i
_ i (rr/ 8) ( e - (rr/ 8) 0 )
T- e 0 ei(rr /8 ) (l6.9d)
16.4 Qubits and Quantum Logic Gates 835

Computing Functions
Let us consider a quantum computer that evaluates functions in a unitary re-
versible manner. Two quantum registers are required, one of length m and the
other of length n. For example, consider the function, f, defined on the binary
space (0, 1, . . . ,2 m - 1). A classical computer evaluates f at each labeled input,
producing the output,

f(O), f(1) , . . . , f(2 n - 1) (16.l0a)

where m and n are positive integers. To compute functions that are not one-to-
one, and maintain reversibility of computation, quantum computers must keep a
record of the input, which necessitates the use of two registers. The first of these
stores the input data, the second records the output. The quantum state of the first
register is written [x) and represents, for example, an m-product state of single
particle spin states. Each possible output is written IY} where, Y = f(x) . As noted
above, [x) is a member of a 2m-dimensional Hilbert space and IY} is a member of
a 2 n-dimensional Hilbert space. These states are individually orthogonal.

(xix') = ox .x" (yly') = Oy,yl

The function evaluation is determined by evolution of the two registers. If U is

the unitary mapping of the transformation, then

U: Ix}IO} ~ Ix}lf(x)} (16. lOb)

One may prepare x in the first register and read f(x) in the second register. More
generally, one may prepare a superposition of all input functions as a single state.
Running the computer only once, one obtains

1
U: 2m / 2
[2
~
m-I]

[x)
1
10} ~ 2m / 2 [IO}lf(O)} + 1l}lf(l)} ... 12
m
- I}lf(2
m
- I)}]

(16.l0c)

Note that the states, 10} , II}, .. . ,12 m - 11}, each lie in 2m -dimensional Hilbert
space. Here one gains an inference of the potential strength of quantum comput-
ing . In one operation the computer yields a multiple of 2m values contained in the
state

If} = 10}lf(0)} + II}lf(l)} . . . 12m - I}lf(2 m - I)} (l 6.lOd)

However, no quantum measurement can tell you what this state If} is. Never-
theless, methods are available to provide limited features of the output state. For
example, there are measurements that give information about output values of
f(x) . This feature of quantum computing is an example of quantum parallelisrrr'
(i.e., a process unqiue to quantum computers) as in (l6.l2d). In describing the
3See, for example, A. Ekert, Quantum Computation, in AlP Conf. Proc. 323, eds. D. J. Wineland, C.
E. Wieman, and S. J. Smith (1994). A. O. Pittrenger, An Introduction to Quantum Computing and
Algorithm s, Birkhauser, Boston (2000).
836 Chapter 16 Quantum Computing

quantum factoring of an integer, it is necessary first to describe two preliminaries

described below.

Euclid's Algorithm
Lemma: If c divides a and b then c divides gcd(a, b) . (gcd == greatest common
divisor). Note that gcd(a, b) = ax + by, for some integers (x, y) . Since c divides
a and b, it must also divide ax + by .
The first part of Euclid's algorithm is the following theorem: Let a > band k
be positive integers such that a = kb + r, r < a . That is, r is the remainder of
a/b . Then,

gcd(a , b) = gcd(b, r) (l6.l1a)

This theorem is demonstrated by showing that each side of the preceding ex-
pression divides the other. First note that r = a - kb, for some integer k. Since
gcd(a, b) divides both a and b and their linear combinations, it also divides r .
By the preceding lemma, gcd(a, b) divides gcd(b, r) . To prove the reverse divi-
sion, first note that gcd(b, r) divides b. As a is a linear combination of band r,
gcd(b, r) also divides a . By the preceding lemma, gcd(b, r) divides gcd(a , b) .
Euclid's algorithm then proceeds as follows : Again let a > b. We rewrite our
starting equation as

(l6.11b)

With (l6.11a), gcd(a, b) = gcd(b,rl). In the next step, b -+ a and q -+ b to

give

(l6.llc)

which, again with (16.11a), gives

gcd(a , b) = gcd(b , rl) = gcdtr}, r2) (l6.l1d)

In the third step, rl -+ a and ri -+ b to give

(l6.11e)

which gives

gcd(a, b) = gcd(b, q) = gcdo-,, r2) = gcd(r2, r3) (l6.11f)

The general equation in this sequence is

rm = km+2 rm+l + rm+2 (l6.11g)

An extended equality of gcd's is obtained at each step. The algorithm ceases when
the last remainder is zero, that is, when Tm = km+2rm+l for some integer m. In
16.4 Qubits and Quantum Logic Gates 837

this case,

gcd(a, b) = gcd(rm, rm+l) = rm+1 (16.11h)

For example, consider the pair (91, 28): gcd(91, 28) = gcd(28, 7) = 7. So the
gcd(91, 28) = 7 (in two steps).

Periodicity of a Sequence
We wish to find an algorithm for a quantum computer that finds the period of a
long sequence. Consider, for example, the sequence of distinct values that repeat
after an unknown period.

fo, Ii. .. . ,fq-I (16.12a)

where q == 2k , and k is an integer. We begin with the quantum state of a set

of independent qubits, which are grouped into two sets, corresponding to two
quantum registers, as follows:

10; O} = 10, 0, . .. 0; 0, 0, . .. O} (16.12b)

The first set of qubits is described by k bits. Operating on each bit of the first
register with U- n /2 [recall (16.8a)] yields a superposition state of all possible
k-length bit strings of the form

1 q-I
U- n /21 0 ; O} = - L la ; O} (16.12c)
.j(j a=O
The next step is designed to map the state la; O} to the state la; fa} for any input a.
This is accomplished by transforming the sequence {fa} into a set of one-qubit
and two-qubit unitary operations. The number of bits required for this second
register is at least sufficient to store the longest construct of {fa} in any of these
computations . There results (an example of quantum parallelism)

1 q-I
U- n /21 0; O} = - L la; fa} (16.12d)
.j(j a=O
where U represents unitary operations. In the next step, one performs a discrete
Fourier transform on the first register.

1 q-I
(FT)la) =- L exp(2rrias /q)ls} (16.12e)
.j(j s=O
Where (FT) denotes Fourier transform. The form (16.12e) is reversible though
the inverse transform and is unitary. When this Fourier transform is applied to the
838 Chapter 16 Quantum Computing

superposition (l6.l2d), there results

. I q-I q-I
(FT)U-rr j21 0 ; 0) = - L L exp(2niaslq)ls; fa) (l6.l2f)
q a=Os=O

The computation is complete .

Consider the output. Suppose {fa} has period r. It follows that fa+r = fa .
The sum over a in (l6.12f) gives constructive interference only when s l q is an
integer multiple of the reciprocal period 11 r. In the ideal case, the probability dis-
tribution of the s-values with respect to the ratio s1q in the first register, is peaked
at integer s-values. All other s-values produce destructive interference . Thus, one
obtains a random multiple of inverse periods . To obtain the period itself, it is nec-
essary to repeat the quantum calculation roughly [log log(r 1k)] times to produce a
multiple that is relatively prime to the period r, uniquely determining it. Note the
following: A classical computer takes roughly N 10g(N) = n2 n steps to trans-
form N = 2n numbers. A quantum computer, employing a component of the
preceding analysis, can perform this transform in approximately log2 N = n 2
steps. Forming the ratio of these expressions gives (log N) 1N = 2n 1n, so that the
classical method takes roughly an exponentially longer number of steps to per-
form this operation. In this process, the main feature is the speed of the Fourier
transform .

Factoring an Integer"
Shor's algorithm for factoring a number N on a quantum computer does not work
for all N. However, it succeeds 2/3 of the time so we know that the problem
of factoring is no more difficult than the complexity class BQP. The algorithm
begins with the question of the period of the sequence

I« = x n (modN) (l6.13a)

where x and N are co-prime. Consider the sequence {2n (mod9)}.

n =0, 1,2,3,4,5,6 ...

2n (mod9) = 1,2,4,8,7,5,1,2,4,8,7,5, . . .

This sequence has the period 6. The period r of the sequence (l6.13a) is the first
nontrivial power of the form x r == 1(modN). Thus, for the preceding sequence,

2r == 1(mod9) = 1 + k9 = 1 + 7 . 9 = 64 = 26
in agreement with the observed value. Employing standard algorithms, this pe-
riod relation is not readily amenable for a lengthy sequence. However, with the
procedure of the previous section, this period may be evaluated efficiently. In
this technique , for the Fourier transform (l6.l2f) to give sufficient resolution,
4p. w. Shor, Algorithms for Quantum Computation, AT&T Bell Labs Report, April (1994).
16.4 Qubits and Quantum Logic Gates 839

N 2 .::: q = 2k transform steps are required . In the algorithm considered, this pe-
riod must be even. If not, we can consider this value N to be among the 1/3 values
the algorithm does not solve.
Having found a value of x that gives an even period of the sequence

one rewrites the expression x r == 1(modN) as the difference

(x r / 2 )2 - 1 == O(modN) (16.l3b)

Equivalently,

[(x r / 2 ) - 1][(x r / 2 ) + 1] == O(modN) = 0 + sN (16.13c)

where s is an integer. This expression tells us that the product on the left of the
preceding expression is a multiple of the number N, so that at least one of these
terms has a factor in common with N. In the final step of the algorithm, one em-
ploys Euclid's method (described above) for calculating the gcd of these factors
individually, with N . Any nontrivial common divisor is a factor of N . Call this
factor g . Then if N / g is a nonprime integer, one may start the procedure again to
find the next factor. If not, N has only two factors and the problem is solved. It has
been noted" that for the class of languages that can be computed in polynomial
time, one may expect a correction factor 2/3 of the time.

Quantum Search AlgorithmS

Another significant property of quantum computers relates to the quantum search
problem: Given a search space of size N and no prior knowledge regarding the
information in this space, how long does it take to find an element in this space
satisfying a given property? Classically, this search requires approximately N op-
erations . The quantum search algorithm involves approximately -IN operations.
Here is an heuristic description of Grover's algorithm for this problem . We
wish to search through a space of N elements. First we index these elements with
numbers from 0 to N - 1 and set N = 2n , thereby permitting the index to be
stored in n bits. The search problem has M solutions, where 1 .::: M .::: N . We
introduce a function A(X), where 0.::: x .::: N - 1. By definition, A(X) = 1, if x is
a solution to the search problem , and A(x) = 0, if x is not a solution to the search
problem . Now introduce a unitary operator, A, with the property

Alx)lq) = Ix)lq EB A(X)) (16.14a)

where the qubit Iq) flips if A(X) = 1 and remains the same otherwise. It is conve-
nient to introduce the state

18) == (10) -Il))/h (16.l4b)

If x is not a solution to the search problem then

5L. K. Grover, Phys. Rev. Lett, 79, 325 (1997) .
840 Chapter 16 Quantum Computing

Alx}IG} = Ix}IG} (16.l4c)

However, if x is a solution to the search problem, then

Alx}IG} = -lx}IG} (16.14d)

These properties may be combined to give
Alx}IG} = (-l) A(x)lx}IG) (16.l4e)

The operator A registers a solution to the search problem by shifting the phase of
the solution. For a search of N items with M solutions, one finds that 0 (J N / M)
applications of A suffices to obtain a solution. To illustrate, consider that M = I,
and that the corresponding solution is the N-dimensional state vector,

Ixo} = (0,0,0, . . . , 1,0,0) (16.15a)

Let Ix *} be perpendicular to Ixo} and let Is} be the normalized vector

I
Is} = ../N (1, I, I, . . . ) (16.15b)

component Is} in the direction of Ix *}.

*
Consider that Is} makes the angle a with Ixo} so that cos a = l /../N. Now the
action of A on Is} is to reflect Is} through the Ix O} axis and thus preserve the
We introduce the operator W that pre-
serves the component of any state vector along Is} but changes the sign of its
component orthogonal to Is}. (This requires many I and 2 qubit gates.) The trans-
formation W has the projection structure [recall (11.126) et seq.]

W = 2Is}(sl - I (16.15c)

Grover's algorithm consists of repeatedly applying the product W A to the input

register, starting with the initial state, Is}.
To better picture the effect of W A on a vector in the (Ixo), Ix *}) plane, con-
sider that it acts on Ix *}.The action of A on [x *}
leaves it invariant. Subse-
quent application of W reflects [x *} through the line that passes through the
origin and along the direction of Is}. If this is the case then the inner product
(six*) = (slWx *).That is,

(slx*) = (sIWx*) = 2(sls}(slx*} - (slx*) = (slx*) (16.l6a)

To establish that W A [x*} doubles the angle between Is} and Ix *}, we must show
further that the vectors A Ix *} = Ix *} and W A Ix *} have the same 'lengths,'
namely, that
.L .L .L.L
(W X 0 IW x o) = (x 0 Ix 0 ) (16.16b)

This equality follows from the unitary property of W. One may conclude that
W A is a rotation. Namely, if the angle between Ix *}
and Is} is f3, then the angle
16.4 Qubits and Quantum Logic Gates 841

FIGURE 16.12 The angles Q and f3 in Grover's algorithm.

between [x *) and WA Ix *) is 2f3. This property implies that the result of oper-
ating on any other vector in the (Ixo), Ix *)) plane is to rotate that vector through
*)
2f3 in the direction from [x to Is). As ex and f3 are complementary, with f3 « 1,

f3 ~ sinf3 = sin(rr/2 - ex) = cos ex = 1/.JN (l6.16c)

In one step of Grover's algorithm, the Is) vector rotates through 2f3 ~ 2/.;N. It
follows that approximately .;Nrr/4 steps bring Is) close to the solution Ixo). (See
Fig. 16.12). Thus we find that a search problem that requires N operations for a
classical solution, with Grover's quantum computing algorithm requires approxi-
mately .;N operations.
We return to quantum parallelism, which is a basic property of quantum com-
puters. Roughly speaking, this feature allows quantum computers to evaluate a
function I(x) for many different values of x simultaneously. Suppose the func-
tion I(x) has a single qubit domain . To compute this function we introduce the
quantum state [x, y), which, with appropriate quantum logic gates, may be trans-
formed to [x, y EB I (x)) . Let the unitary operator that effects this transformation
be labeled U so that

Ulx, y) = [r , y EB I(x)) (l6.17a)

Consider the action of this operator on the single-qubit state (10) + 11))/ J2.
To make this calculation we operate with U on the equivalent two-qubit form
(10,0) + 11,0)) / J2 to obtain the state

(10,/(0)) + 11, l(l)))/h (l6.17b)

842 Chapter 16 Quantum Computing

The two terms in this expression contain information about both f(O) and f(l)
and is an example of quantum parallelism. Equation (l6.12d) is likewise an ex-
ample of quantum parallelism.

PROBLEMS

16.16 Does the sequence {3n (mod9)} come into play in finding the factors of 9? Explain
your answer.
16.17 What are the respective periods of the sequences (a) (2n(mod7)} , (b) {3n(modlO)}?
16.18 What are the values of (kn(modn)} where k and n are integers ?
16.19 What are the values of {2n (mod2q)} where nand q are integers ?
16.20 (a) What is the period of the sequence {(I3)n(modI5)} ? (b) Use this answer to
factor 15 in the technique discussed in the text.

Answer
(a) The resulting sequence is, 13,4,7, I , 13,4,7 , I, . ... and has the period 4. (b)
It follows that for some finite integer k, (132 + 1)(132 - 1) = 168 x 170 =
kI5 . Forming the gcd of the related integer pairs gives the results, gcd(168 , 15) =
3, gcd(170 , 15) = 5.
16.21 Construct , diagrammatically, a quantum AND logic gate.
16.22 Construct , diagrammatically, a quantum OR logic gate.
16.23 Describe two distinctions between a quantum and classical computer.

Answer
(a) The operations in a quantum computer are reversible. (b) The input data in a
quantum computer is maintained in a given operation .
16.24 A problem cannot be solved in polynomial time on a classical computer. Under
what additional condition s would this problem be an element of the NP complexity
class?

Answer
If the solutions to this problem are readily verified on a classical computer.
16.25 A decision problem may be solved on a Turing machine running in exponential
time. To what complexity class does this problem belong?
16.26 Employing Euclid's algorithm , find the gcd of (6825, 1430).
3
16.27 Consider a computer that takes (a) 0(2n ) seconds and, (b) O(3n) seconds, in n
steps of a given process. Are these respective time intervals polynomial or expo-
nential?
16.28 (a) Write down the truth table of an XOR gate.
(b) Consider the two-particle superposition (single qubit) states: a = all ) + ,810);
b = yll) + <510), where both states are normalized , y2 + <5 2 = 1, a 2 +,82 = 1.
What is the output a', b' of an XOR gate for the input a, b?
Problems 843

(c) Write down four binary sequence s, each with four elements, that are equivalent
to your answer s to (b).
(d) Show that output states are normalized.
Answers
(a) The truth table is given by

a b a' b'
0 0 0 0
0 1 0 I
I 0 I I
1 I 1 0

(b) First we note that, say, a 2 , represents the probability that the qubit a is in the
state II), etc. With the truth table we see that b' change s only when a = I. As
a' = a for all cases, we need only address the form of b' and write

b' = kll ) +qIO)

The factor k 2 represents the probability of finding the qubit b' in the state 11 ),
for which there are two possibilities : First, all ) in a and 810) in b contribute
to kll ) in b' , Thu s (a8)2 is the probability that all ) in a goes to kll ) in b' ,
Secondly , the state .610) in a and Y II) in b likewise contribute to k II) in b' , We
obtain (.6y)2 for the probability that .610) in a goes to k II ) in b' , There results,
2
k = a 282 + .62 y 2
For the contributions to q 10) we note first that a II ) in a and y II ) in b contribute
to qlO) in b' with the probability term (ay)2 . Secondly, .610) in a and 810) in b
likewise contribute to qlO) in b' with the probability term (.68)2. We obtain

q2 = a 2 y2 + .6282
There results

This finding is an example of "quantum entanglement," in the sense that the

starting states are entangled in the output states .
(c) (0101) , (lOll) kll ); (0000) , (1l1O) qlO )
(d) Set a = sin 1:1 , .6 = cos 1:1 ; y = sin A; 8 = cos A. As with the starting state, both
outputs a' and b' are normalized .

16.29 (a) Consider a set of N elements. What is the expression for the number of subsets
of this set?
(b) What is the binary representation of this number ?
Answers
(a) 2N
(b) All zeros except 1 in the Nth space to the left of the starting value in the binary
representation.
 List of Symbols

ao Bohr radius
a Scattering length, edge length
A At
a,a Annihilation and creation operators
A Vector potential
Ai(x) Airy function
Alb Blk Einstein A and B coefficients
b(k) Momentum probability amplitude
:B Magnetic field (also :B in figures)
~ Basis
c Speed of light
et(t) Transition probability amplitude
C m ,m2 Clebsch-Gordon coefficient
:J) Coefficient matrix
iJ Displacement operator
D Derivative operator (a /ax)
d Electric dipole moment
DC Denotes constant in time
8 Electric field (also 8 in figures)
e Unit vector
e Charge
IE Photoelectric energy
IE Perturbation energy
E Energy
EF Fermi energy
8 Electric field
Inn' Oscillator strengths
1(8) Scattering amplitude
9 Acceleration due to gravity ; Lande 9 factor
9(E) ,9(V) Density of states
9 Degeneracy, density of states
H Hamiltonian
H' Perturbation Hamiltonian

845
846 List of Symbols

Sj Hilbert space
Jfn(~) Hermite polynomial
IHI r-dependent perturbation Hamiltonian
h,1i Planck's constant
i Identity operator
j Total angular momentum quantum number
J Spin, orbital, or total angular momentum
r Current density
kB Boltzmann's constant
k,K Wavenumbers
K2 Eigenvector of j2 /1i2
I Orbital angular momentum quantum number
L Orbital angular momentum
.£ Orbital-angular-momentum-term notation symbol
L Edge length
L~(x) Laguerre polynomial
m Mass
m* Effective mass
M Mass of center of mass
JfE,Jfk Number of states in an energy band
N Total number of particles
P Momentum
fir Unsymmetrized radial momentum
Pr Radial momentum
PI2 1\vo-particle joint probability density
P P dr is probability
Pr P, dr is probability
P Nondimensional Pr
P Radiated power
9l Momentum of center of mass
P Permutation operator
IPI Order of permutation
IP' Parity operator
Plk Transition probability
[ilk Probability relevant to transition to an energy band
Pl(cosB) Legendre polynomial
Pt (COS B) Associated Legendre polynomial
PnQM , pn CL Quantum and classical probability densities
q Charge
q Coordinate
List of Symbols 847

r Integral in WKB analysis

r Radius vector
:R Radius to center of mass
R Reflection coefficient
R Radial wavefunction
Ii Nondimensionalized radial wavefunction
~ Rydberg constant
Rot/> Rotation operator
S Spin angular momentum
S Transfer matrix
s Spin angular momentum quantum number
T Transmission coefficient
T Temperature
Tc Critical temperature
T Kinetic energy
u Radial wavefunction (u = r R)
u(v) Energy density per frequency interval
U Energy density
'U Column vector
VF Fermi velocity
V Potential energy
V Average potential
V Column vector
Wlk Transition probability rate
Wlk Probability rate for transition to an energy band
Wlk Total atomic transition rate
i Exchange operator
yt(B, ¢) Spherical harmonic
a Fine-structure constant
a Polarizability
a 4 x 4 matrix
a,f3 Spin eigenstates
f3 Harmonic oscillator wavenumber
f3 Speed nondimensionalized with respect to the speed of light
f3 4 x 4 matrix
XS ,XA Symmetric and antisymmetric wavefunctions
B Energy
E Mass ratio, dielectric constant
,., Integral in WKB analysis
r State of a system ; 4 x 4 matrix
y 4 x 4 matrix
K Spring constant
848 List of Symbols

Wavelength
Parameter of smallness
Characteristic lengths in p-n junction analysis
Angular momentum parameter; scattering centers per unit
area
/L Reduced mass
/L Chemical potential
It Magnetic moment
v Frequency
a Stefan-Boltzmann constant
a Total scattering cross section
ir Pauli spin operator
da Differential scattering cross section
w Angular frequency
n Solid angle, spin
c , n /2 Cyclotron and Larmor frequencies
<I> Work function
<I> Electric potential
<l>m (¢) Eigenfunction of i,
1{! Wavefunction
p Nondimensional radius in hydrogen wavefunction
p Density matrix
p(x) Particle density
Gl(e, ¢) Eigenfunction of L2
cp Time-independent wavefunction
~ Nondimensional displacement
~ Spin state
~ ,17 Nondimensional wavenumbers
Units
A Ampere
em , /Lm, m Centimeter, micron, meter
C Coulomb
s, ms, /LS, ns Second, millisecond, microsecond, nanosecond
V, eV, meV, keV, Volt, electron volt, milli electron volt, kilo electron volt,
MeV mega electron volt
W,kW,MW Watt, kilowatt, megawatt
EEl Addition mod(2)
 Appendixes

A. Additional Remarks on the x and p Representations

B. Spin and Statistics
C. Representations of the Delta Function
D. Differential Vector Relations
E. Harmonic Oscillator in Spherical Coordinates
F. Physical Constants and Equivalence Relations
APPENDIX

x
A Additional Remarks on the
p
and Representations

Let [x' ) represent an eigenstate of x. Then

xix') = x'l x') (A.I)

These eigenstates obey the orthononnality condition

(x'lx) = 8(x - x') (A.2)

The matrix elements of x are then given by

(x IxIx') = x ' (x [x' ) = x' 8 (x - x') (A.3)

This is a continuous matrix with nonzero entries only on the diagonal x = x' ,
As remarked in the text, summations over continuous matrices are replaced by
x
integrations . For example, the multiplication of the matrix by the column state
vector 11ft (x) ) gives

f dx'I1ft(x'»)(x'lxl x ) = x f dx 'I1ft(x'») 8(x ' - x) = xl1ft(x»)

x
In the x representation, operating on a state has the effect of multiplying the
state by the scalar x .
The projection of 11ft ) onto the basis vector [x') is the coordinate representation
of 11ft).

(x'I1ft) =f (x' lx )(x l1ft ) dx = 1ft(x') (A A)

Here we have employed the spectral resolution of unity,

j = fiX) dx (xl (A.5)

Note in particular that the coordinate representation of Ix) is the delta function,
8(x-x') , as given by (A.2) . This identification permits one to write the eigenvalue
x
equation for in the form (3.26).

851
852 Appendix A Additional Remarks on the xand p Representations
If Ip} represents an eigenstate of P, then
(A6)

The matrix of Pin the coordinate representation is given by

(xlplx' ) = -ili~8(x
ax -x') (A7)

This relation allows us to obtain an explicit form for the transfer matrix (xlp).

p(xlp} = (xlplp) = ! dx '(xlplx')(x'lp} (A8)

= -iii! dx'~8(x
ax -x')(x'lp}
. a
= -11i-(xlp}
ax
The solution to this differential equation is

(xlp) = _l_eipx jn (A9)

v'2n Ii
The normalization ensures the unitarity of the continuous matrix (xlp ). To see
this, we first recall the condition for unitarity,

I: (plx} *(plx') dp = 8(x - x ') (A 10)

With the representation (A9) for (xlp) and using the property (plx)* = (xlp),
we find that

LHS(AlO) = 1 00

-00
(xlp)(plx') dp = _1_1
2nli
00

-00
e ipx jne-ipx' jn dp (All)

Setting p / h. == y reduces the right-hand side of the latter equation to

-I
2n
1 00

-00
. ( x-x ')
ely dy = 8(x - x ') (A12)

which establishes the unitarity of (xlp) . Note that the projection (A .9) gives either
the coordinate representation of the eigenstates of p or the momentum represen-
tation of the eigenstates of x.
Let us see how the form (A9) allows one to reconstruct the matrix for P as
given by (A7). In the p representation, we have

(plplp') = p' 8(p - p') (AI3)

Appendix A Additional Remarks on the x and pRepresentations 853

Using (A.9) together with the last equation gives

(x!filx ') = -1- foo foo dp dp' p' 8(p - p')e IPX
. /ne- lp
. ,X'/n (A.14)
2rrn -00 -00

= _1_ foo dp peip(x-x')/n

2rrn -00

= -i n~ _1_ foo eip(x-x')/n dp

ax 2rrn -00
a ,
= -in-8(x -x)
ax
which agrees with (A.7). We may use this relation to calculate the coordinate
representation of fil1/J).

(xlfil1/J) = foo dx' (xlfilx')(x'I1/J) = -infoo dx' ~ 8(x - x')1/J(x') (A.15)

-00 -00 ax
a
= -in-1/J(x)
ax
This has the same effect as simply operating on the state 1/J with the differential
operator -i n.a/ ax.
As a simple example of these concepts,' consider the problem of finding the
matrix of (x fi - fix) in the x representation. Let us first examine the term

III
00

(xlxfilx') = dx" dp' dp(xlxlx")(x"lp')(p'lfilp)(plx')

-00

= _1_
2rrn
II 00

dx" dpx" 8(x - x")peip(x"-x')/n

-00

= ~ foo dp peip(x-x')/n
2rrn -00

= ~ (-in~) foo dpeiP(x-x') /n

2rr n ax_ oo
a ,
= -i!i.x- 8(x - x)
ax
In like manner we find that

-(xlfixlx') = ihx' ~
ax
8(x - x')

1For further development of these topics, see W. Louisell , Radiation and Noise in Quantum Electron-
ics, McGraw-Hill, New York, 1964.
854 Appendix A Additional Remarks on the xand p Representations
Combining these results gives

(xl xp - Pilx
l
) = -ih(x - x
l
) -
o 8(x - x ') = +ih8(x - x ') (A.I6)
ox
In concluding this discussion we note the following . Suppose that a complete
set of commuting observables is diagonalized by the ket vectors I~} . Then the
coordinate and momentum representations of these states are (xl~) and (pl~),
respectively. For example, consider the eigenvectors In} that simultaneously diag-
aa
onalize the number operator N = t and the Hamiltonian H = nwo (N + !),
appropriate to the harmonic oscillator (Section 7.2).

Hln} =!lwo (n + ~) In}

(A. 17)
Nln} = nln}
No information is revealed by these equations other than the fact that In} is an
eigenvector of Hand N with respective eigenvalues as shown. If, for example, one
wishes the coordinate representation of these states, one must form the projections
(xln). These are the weighted Hermite polynomials (7.58).
In a similar vein the coordinate representations of the eigenvectors lim} of the
operators f2 and i, are the projections (fJ¢lim) [i.e., the spherical harmonics,
yt(fJ, ¢)].
As a further case of this formalism, consider the following example . We wish
to show that the coordinate representation of the Schri:idingerequation in abstract
ketspace

ol1Jr}
ih = HI1Jr}
ot
is the standard relation (3.45) .
Without loss in generality we work in one dimension . Multiplying the given
relation from the left by (x], we obtain

O A f dx '(xIHlx '}(x'I1Jr}
ih-(xl1Jr) = (xIHI1Jr) =
A

ot
= f dx' 8(x -x')H(x')(x'I1Jr}

where H(x) denotes Hin the coordinate representation. The further identification
(xl1Jr) = 1Jr(x)

gives the desired relation .

APPENDIX

B Spin and Statistics

In this appendix we wish to offer a brief elementary outline of the argument con-
necting spin and the exclusion principle. As described in Chapter 12, particles
with integral spin do not obey the exclusion principle, whereas those with half-
integral spin do obey the exclusion principle.
The particle quality of a field may be described in second quantization,
wherein, in accord with Problems 13.37 and 13.38, the state of the system is
written In I, nz, ... } . In this notation n ; represents the number of particles in the
ith state.
There are two prescriptions for the quantization of a field. The first is given by
the Jordan-Wigner anticommutation rules (Problem 7.32),

A am
{an, A} = {an
A tAt}
,am =0
(B.1)

The second is given by the Bose commutation rules,

(B.2)

As established in Problem 7.32, particles such as electrons, which obey the

Jordan-Wigner anticommutation rules (B.I), exist in accordance with the exclu-
sion principle. Number eigenvalues n, are either 0 or 1 (ni z = ni) . Particles, such
as photons, which obey the Bose commutation rules (B.2) do not adhere to the
exclusion principle.
In his argument relating spin and exclusion, Pauli I imposed the following re-
quirements on physical systems.

1. Let A be an observable pertaining to the space-time point ri , fl, and let

B be an observable at rz , fZ . Consequently, if Irl - r zl/lfl - fzl > c, then
A and B commute. The rationale behind this stipulation is as follows. In

1W. Pauli, Phys. Rev. 58, 716 (1940). See also R. Streater and A. S. Wightman , peT, Spin and Statis -
tics. and All That , W. A. Benjamin, New York, 1964.

855
856 Appendix B Spin and Statistics

that these space-time points are separated by speeds greater than that of
light, relativity (or causality) specifies that measurement of A can in no
way interfere with measurement of B. Equivalently, we may say that A and
B commute .
2. The total (relativistic) energy of the system is greater than or equal to zero.
What Pauli then showed is that
(a) Quantization of integral spin fields according to Jordan-Wigner anti-
commutation rules (B.1), corresponding to exclusion, violates the first
postulate .
(b) Quantization of half-integral spin fields according to Bose commutation
quantization rules (8.2) violates the second postulate.

The distinction between half-integral spin fields and integral spin fields en-
ters the argument through the manner in which these fields transform under a
Lorentz transformation. The Lorentz transformation relates observation of prop-
erties (fields, mass, length, etc.) in one inertial frame to observation in another
inertial frame of these same properties . The corresponding matrix is orthogonal
(see Table 11.1) and represents a rotation in four-dimensional space. A somewhat
similar distinction between integral and half-integral spin states evidenced under
ordinary rotation of axes in 3-space, such as described in the discussion on the
rotation operator in Section 11.5, is found to persist under Lorentz transformation
of spin fields.?

Statistics
The property that particles have of either obeying or not obeying the exclusion
principle has direct consequence in the distributions in energy that aggregates of
particles have in equilibrium at a temperature T . Thus fermions (particles with
half-integral spin) satisfy Fermi-Dirac statistics . A collection of such noninter-
acting particles at the temperature T has the energy distribution

1
fFD = e(Ej-/1-)/kBT +1
This expression gives the average number of particles per state at the energy Ei.
The parameter JL represents chemical potential, which at 0 K reduces to the Fermi
energy, E F . At this temperature , no states of energy greater than E F are occupied
(see Fig. 2.5).
Bosons (particles with integral spin) satisfy Bose-Einstein statistics. A collec-
tion of noninteracting bosons at the temperature T has the energy distribution

1
fBE = e(Ej-/1-)/kBT _ 1

2For further discu ssion of the distinction between integral spin and half-integral spin fields, see H. Yil-
maz, The Theory of Relativity and the Principles ofModem Physics, Blaisdell , New York, 1965.
Appendix B Spin and Statistics 857

Here again, f represents the average number of particles per state at the energy E,
and JJ., is written for the chemical potential. This distribution appears in the Planck
radiation formula (2.3) relevant to a photon gas in equilibrium at the temperature
T, for which case JL = O.
APPENDIX

c Representations of the
Delta Function!

Cartesian Coordinates

2Jro(x - x ') = i: e ik (x-x') dk (C.l)

tt 0(x - x ') = 1 00

cos k(x - x ') dk (C.2)

00

2Jro(x - x') = L exp[in(x - x')]

-00

00

2Jr0(x - x') = 1 + L 2 cos n (x - x ')

1

it oO(X -x ' ) = l'1m sin 1J(x - x ') (C.S)

1)-->00 x - x'
- (X- X')2 / E2
o(x - x ') = lim e c (C.6)
E--> O Ey Jr

O( ') li 1 - cos 1J(x - x ')

Jro X - X = Hfl -----=--- (C.?)
1)--> 00 1J(x - x ' )2
2[1J(x
O( ') l' 2sin - x ')/2]
Jro x- X = 1m (C.S)
1)-->00 1J(x - x ' )2

Jro(x - x ') = lim E = lim 1m - - - - (C.9)

E-->O (x - x ' )2 + E 2 E-->O (x - x ') - i E

Let H, (x) be the nth-order Hermite polynomial. Then

(C.lO)

1Additional properties of the delta function may be found in Problem 3.6.

2Domain of validity : x' - n :s: x :s: x' + tt .

859
860 Appendix C Representations of the Delta Function

i:
All the above representations obey the normalization

8(x -x /)dx ' = 1

In three dimensions, with r = (x , y , z), one has

8(r - r ') = _1_

(2iT)3
Iff
00

eik'(r-r') dk dk dk
x y Z
-00
(C.11)
00

fff
-00
8(r - r ') dx ' dy ' dz' = 1

The Unit Step Function

Let S(x) denote the unit step function:

S(x)=I, x :::O

S(x) = 0, x < 0

Then
dS(x)
~ =8(x) ,

Note also that

dlxl
- = S(x) - S(-x)
dx
It follows that

Spherical Coordinates
Let P/(JL) be the lth-order Legendre polynomial. Then

/ ~21+1 /
8(JL - JL) = L..- - - P I (JL) PI(JL )
1=0 2
(C.12)

1 1
-1
8(JL - JL /) dJL' = 1

The delta function over solid angle may be written in terms of the yt (B, ¢) spher-
ical harmonics:
8(B B/) 8 (A. A./)
8(0 - 0 /) = - . 'I' - 'I' = L L
00 1
[yt(B, ¢)]*yt(B', ¢ /) (C.13)
smB 1=0 m=-I
Appendix C Representations of the Delta Function 861

The directional coordinates of 0 are 0 and ¢ . Normalizations are given by

l 1r 8(0 - 0') dO' = 1, 121r 8(¢ - ¢ ') d¢' = 1, [

141r
d(O - 0 ') dO = 1

dO = sin OdO d¢

In three dimensions one obtains the representation

8(r - r') = 8(0 _ 0 ') 8(r ~ r') (C.l4)

r

wherej[(kr) is the lth-order spherical Bessel function.

1 o
00
j[(kr)j[(kr')k 2 dk
Jr
= 2 8(r - r '),
2r
1 00

8(r - r ') dr' = 1 (C. IS)

We note also the differential relations

(C.16)

= -4Jr8(r)

Cylindrical Coordinates
Let Jm (x) be the mth integral -order Bessel function . Then

8(p - p') = [00 Jm(kp)Jm(kp')kdk

p 10 (C.l7)
1 00

8(p - p')dp = 1

With kjPo denoting the zeros of Jo(x), that is,

one has the representation

2 ~ Jo(kjP)
JrPo 8(p) = ~ [J (k ' )]2
J=\ \ JPo
(C.18)

l Po
2Jr8(p)pdp = 1
862 Appendix C Representations of the Delta Function

Three other important normalizations of Jm(x) are

k 100 Jm(kp)dp =1 (C.19)

Jm(kp)
= -1
1
00
- - dp (m > 0) (C.20)
o P m
PO

lo
Jo(kjp) JO(klP) dp
1 2 2
= -PO
2
J1 (kjPO) Ojl (C.2I)
APPENDIX

D Differential Vector Relations

The line element in orthogonal coordinates is

(D .1)

The V operations are

1 8l/t 1 8l/t 1 8l/t

Vl/t = - - a l +- -a2 +- -a3 (D.2)
hi 8xI h2 8X2 h3 8X2

V· A = _1_
hlh2h3
[~(h2h3A()
8x!
+ ~(hlh3A2) + ~(hlh2A3)]
8X2 8X3
(D.3)

V x A = _1_ [~(h3A3) - ~(h2A2)] ai (D.4)

h2h3 8X2 8X3

+ _1_ [~(hIAj) _ _8_(h 3A3)] a2

hlh3 8X3 8xI

+ _1_ [~(h2A2) - ~(h!Aj)] a3

hlh2 8x! 8X2

2 1 [8 (h3h2 8l/t) 8 (h3hl 8l/t) 8 (h!h2 8l/t)]

V l/t = hi h2h3 8x! ~ 8x! + 8X2 ---,:;:; 8X2 + 8X3 ----,;; 8X3
(D.5)

Cartesian coordinates :

X!=X, X2=Y, X3=Z

Cylindrical coordinates:

XI = p, X2 = ¢, X3 =Z
Spherical coordinates:

hi = 1, h2 = rsin8, h3 = r; XI = r, X2 = ¢, X3 =8
863
864 Appendix D Differential Vector Relations

Differential operator relations in three coordinate frames

Cartesian Coordinates Cylindrical Coordinates Spherical Coordinates

dX ,dy,dz dp,pd¢,dz dr, r d« , r sinO d¢

Unit vector and elementary cross relations
ax , ay , az 3p , 3q" 3z a- , ae , 3q,
X = pcos¢ x = rsinOcos¢
y = p sin ¢ y = r sinOsin e
z= z z = r cos s
pZ = xZ + y Z r Z = x Z + yZ + zZ
tan e = y /x cos s = z/ J '"'
xZ
' +
- y""
Z-+- z""
Z
z=z ¢ = tan-I (y /x)
3p = cos¢ 3x + sin¢ 3y 3r = sinOcos¢ 3x + sinOsin¢ 3y
+ cos e 3 z
3q, = - sin¢ 3x + cos¢ 3y 3e = cos s cos¢ 3x + cos e sin¢3y
- sinO 3z
3Z = 3z 3q, = -sin¢3x +COS¢3y
3x = COS¢3p - sin¢3q, 3x = sinO COS¢3 r + cos e cos¢ ae
- sin¢3q,
3y = sin¢3p + COS¢3q, 3 y = sinO sin¢ 3r + cos e sin¢ae
+COS¢3q,
3Z = 3z 3z = cos 0 3r - sin 0 3e
The gradient of w, \11fr
o1fr o1fr o1fr
(\11frh = -
ax
(\11fr)p = &P (\11fr)r = -
or
o1fr 1 o1fr 1 o1fr
(\11fr)y = -
oy (\11fr)q, = P o¢ (\11fr)e = - -
r 00
o1fr 1 o1fr
(\11fr)z = - (\11fr) z = o1fr (\11fr)q,=-.--
oz oz r smO o¢
The divergence of A, V . A

oAx oA y oA 1 o(pA p) I 0Aq, OA 1 o(rZA r) 1 o(sinOAe)

-
ax
+-oy
-z
+oz -
p
- +oz-z
-op- +p-o¢ ;:z -o-r- + rsin 0 00

1 0Aq,
+rsinO
- - -o¢-

The curl of A, V x A

OAZ-oA
(VxAh= ( -
oy
-
oz
y)
(V x A)p = (
1 OAz
P &i - az
0Aq,) 1
(V x A)r = - -
(O(SinOAq,)
rsin O
OAe)
- -
00 o¢

OA x - OAz)
(VxA)y= ( -
oz
-
ax
(V x A)q, = ( OA
p
az- a;;
OAz) oAr11 o(r Aq,)
(V x A)e = - - - - - - - -
rsinO o¢ r or

oA y-OAx) 1 (O(PAq,) OAp)

(VxA}z= ( -
ax
-
oy
(V x A}z = P ---a;;- - &;;; (V x A)q, =! (o(rAe) _ OAr)
r or 00
Appendix D Differential Vector Relations 865

Formulas of Vector Analysis and Vector Calculus

v . (A x B) = B • V x A - A • V x B (D.6)

V(A • B) = (A· V)B + (B • V)A + A x (V x B) +B x (V x A) (D .?)

V • (V x A) = 0 (D .8)

V x (V x A) = V(V. A) - (V· V)A (D.9)

V x (A x B) = (B. V)A + A(V. B) - (A . V)B - B(V. A) (D .IO)

A x (B x C) = B(A . C) - C(A . B) (D.ll)

A x (B x C) +B x (C x A) +C x (A x B) =0 (D.12)

Ax Ay Az
A • B x C = B, By B, (D.13)
c, c, c,
A· B x C = B· C x A = C· A x B (D.14)
(A x B) . (C x D) = (A · C)(B • D) - (A . D)(B . C) (D.IS)

(A x B) x (C x D) = (D· A x B)C - (C· A x B)D (D.16)

V(¢1fr) = ¢V1fr + 1frV¢ (D .l?)

V· (¢A) = A· V¢ + ¢V. A (D.18)
V x (¢A) = ¢V x A - A x V¢ (D .19)

If r is the radius vector from the origin and A is a con stant vector, then

V.r =3 (D.20)

Vx r = 0 (D.21)

Vr = r/r (D.22)
Vr- I = -r/r 3 (D.23)
V . (rr- 3 ) = - V 2r- 1 = 4rro(r) (D .24)

V · (Ar- I
) = A· (Vr- ) = -(A· r)r-
I 3
(D.25)
V x [A x (r/r 3 ) ) = -V(A. r/r 3 ) for r =/= 0 (D.26)
V 2 Ar- I = A V 2r- 1 = 0 for r =/= 0 (D .2?)

ei kr coskr
(V 2 + k2 ) -
r
= (V 2 + k2 ) - r
- = -4rr 0 (r) (D .28)
866 Appendix D Differential Vector Relations

Properties of Integrals over Vector Functions

The Line Integral

(D.29)

If A = V ¢, or V x A = 0, then the line integral

(D.30)

is independent of the path of integration from rl to r2. If the path of integration is

a closed curve, then

(D.3!)

Gauss 's Theorem Let S be a surface bounding a region of volume V. Then ,

Iv V · AdV Is A · dS
= (D.32)

Stokes's Theorem Let S be an open surface bounded by the closed, nonintersect-

ing curve C . Then

LA. dl = Is (V x A) • dS

Green's First Identity

Iv (¢V
2l/t
+ v¢. Vl/t) dV = Is (¢Vl/t) • dS (D.33)

Green's Second Identity

(D.34)

Other Identities

Iv V x AdV Is dS x A= (D.35)

L dl Is dS x
¢ = V¢ (D.36)
APPENDIX

E Harmonic Oscillator in
Spherical Coordinates

We wish to find the solution to the Schrodinger equation for a particle of mass m
in the harmonic potential

(E.!)

in spherical coordinates. (The analysis follows closely that of Section l 0.6 in

derivation of eigenproperties of the Hydrogen atom.) The variable r is the dis-
placement of the particle from the origin and w is a characteristic frequency. The
related radial equation is given by

(E.2)

With the substitution u = Rr and the relations

J2mE mw
k=-li- ' 'A=- (E.3)
Ii
there results

(E.4)

To examine the asymptotic behavior of u at r ~ 0 and r ~ 00 we introduce the

transformation

u = r l + ! (exp -!'Ar 2) w (E.5)

Substituting this equation into (E.4) gives

w" +2 C: I - 'Ar ) w' - [2'A(1 +~) - k

2
] W =0 (E.6)

With the new independent variable y = 'Ar 2 , the preceding becomes

y w" + [(1 + ~) - y] w' + [! (l + ~) - !,8] w = 0 (E.?)

867
868 Appendix E Harmonic Oscillator in Spherical Coordinates

where

The solution to (E.7) is the confluent hypergeometric function

w = F [ (l + ! ~ - (3), l + ~ ; y] (E.8a)

The condition that R ~ 0 as r ~ 00 gives

n=-!(l+~-f3); n=O,1,2, . . . (E.8b)

which returns the eigenenergies,

En ,l = liw(l + 2n + ~) (E.9a)

and corresponding eigenfunction

(E.9b)

The relation of the function F to Hermite polynomials (Sec. 7.3) is given by,

_ (-1)P(2p)! 1. 2
Hn(x) - F( - p, 2' x ), n =2p (E.10a)
p!
_
Hn(x) - (-1) 2
P (2p + 1)! xF ( - 3. 2),
p, 2'x n = 2p +1 (E . lOb)
p!
APPENDIX

F Physical Constants and

Equivalence Relations

Velocity of light in vacuum e 2.9979 x 108 m/s

2.9979 x 1010 cm/s
Planck's constant h 6.6261 x 10-34 J s
6.6261 x 10-27 erg s
4.1357 x 10- 15 eV s
1.0546 x 10-34 J s
1.0546 X 10-27 erg s
6.5821 x 10- 16 eV s
Avogadro's number No 6.0221 x 1023 atoms/mole
Boltzmann's constant kB 1.3807 x 10-23 J/K
1.3807 x 10- 16 erg/K
8.6174 x 10-5 eV/K
Room temperature 300 K = 0.02585 eV
Gas constant R = NokB 8.3145 J/mole K
8.3145 x 107 erg/mole K
1.9870 cal/mole K
Volume of 1 mol 22.421 liters
of perfect gas, at
normal temperature
and pressure
Electron charge e 1.6022 x 10- 19 C
4.8032 x 10- 10 esu
Electron rest mass m 9.1094 x 10-31 kg
9.1094 x 10-28 g
0.511 MeV
Electron classical radius ro = e2/me2 2.818 x 10- 13 cm
Electron magnetic moment /Le 1.001/Lb
Fine-structure constant ex= e2 / IU: 7.297 x 10-3 = 1/137.04
Compton wavelength AC = hf mc 2.426 x 10- 10 em
Xc = Ii/me 3.862 x 10- 11 cm
Lyman ex 0.1216/Lm

869
870 Appendix F Physical Constant s and Equivalen ce Relations

Gravitati onal constant G 6.672 x 10- 8 dyne-em/ j g2

Proton rest mass Mp 1.6726 x 10- 27 kg
1.6726 X 10- 24 g
1.0073 amu
938 .27 MeV
Neutron rest mas s Mn 1.675 X 10- 27 kg
1.675 X 10- 24 g
939.57 MeV
Deuteron rest mass Md 1.876 GeV
eli
Bohr magneton /-Lb = - 9.27 x 10- 21 erg gauss" !
2mc
Ratio of proton mass Mp
1836.1
to electron mass m
Charge-to-m ass ratio e 1.7588 x 1011 C/kg
of electron m 5.2730 x 10 17 esu/g

Stephan-Boltzmann 2) 4
(rr ( kB 5.6697 X 10-5 erg cm- 2 s- 1 K-4
constant a = 60 1i3 c2 )

m e4 1i2
Rydberg constant JR = - 2 = - - 109,737 .32 cm'
21i 2mao2
e 2 a 2mc 2
= -=-- 13.6 eV
2ao 2
1i2
Bohr radius ao=-2 0.52918 A
me
1i2
JRa 2 = -
o 2m
Triple point of water 273 .16 K
Atomic mass unit 1amu 1.6605 x 10- 24 g
931.5 MeV
1
Mas s ofC l2 x -
12
Wien's displacement AmaxT 0.290 cm K
law con stant
Resistivity unit aolij e2 21.7 /-L Q em
Useful Conversion Constants and Units
Constants, MKS unit s EO 107 j 4rrc2 F/m
= 8.8542 x 10- 12 F/m
/-LO 4rr x 10-7 Him
1j 4rr Eo 9 x 109 mIF
1 j EO/-LO c2
Appendix F Physical Constants and Equivalence Relations 871

§ 120JtQ = 377Q
YEO
Energy in wavenumbers E=hd- I
he = 1.240 x 10-4 eV em
= 1.24 eV lIm
Useful relations fijm = 1.16 cm 2 js
emje 2 = 118 sjcm 2
1 electron volt eV 10- 6 MeV
1.6022 x 10- 12 erg
1.6022 x 10- 19 J = [e, C] J
3.829 x 10- 20 Cal
11,605 K
1 MeV 1.6022 x 10-6 erg
1.6022 x 10- 13 J
1 coulomb 0.1 abcoulomb (emu)
(ejlO) statcoulomb (esu) » 1 esu
1 emu 1 esu/c
IV 360 statvolts = 360 erg/esu
1 dyne 10-5 N
1 erg 10-7 J
0.624 x 106 MeV
E (eV) 1.24jA (in lIm)
1 Wjm 2 = 1 T 104 gauss
(l gauss cm 2 = 1 T m2 )
1 Cal 4.186J
4.186 x 107 ergs
kBNO 2CaVK

eB] -+ [eB] ,B in Gauss

[me esu m emu

( p _ ~A)2
e esu
-+ (p _ eA)2
emu
Coulomb's Law
cgs MKS (SI)
F = qlq2 F = qlq2
c(cgs)r 2 4Jtc(MKS)r 2
ql = qi = 1 esu, at 1-cm separation, ql = q: = J4Jtco C, at 1-m
in vacuum , gives 1 dyne of force separation, in vacuum, gives
1 N of force
c(MKS)
C(cgs) = co(MKS) == crelative
c(cgs)vacuum = 1 c(MKS)yacuum = cO
Istatvoltjem = Idynejesu IVjm = INjC
lO nem = lOn-2 m Ijem 3 = 106 jm 3
872 Appendix F Physical Constants and Equivalence Relations

Resistance and Resistivity

[Pegs] = stat - ohm - cm(s) [PMKS] = Ohm - m
[Regs] = stat - ohm(sjcm) [RMKs] = Ohm
 Index

action integral , 16,254,268,401 Berestetskii, V. B., 813 Born approximation, 777, 782p
adiabatic invariant, 737p Bernstein, R. B., 731 standard form of, 780
adiabatic perturbation, 727, 737p Bessel functions, see spherical Born postulate, 55, 74
adiabatic process, 63p Bessel functions Born-Fock theorem, 729
adiabatic theorem, 727 Bessel's equation, 424 , 445p Born-Oppenheimer
alkali metals, 608 annihilation and creation approxiamtion, 623
alkaline earths, 608 operators for generation of, Borowitz, S., 80
AND gate , 824 475p Bose commutation rules , 851
angular momentum Bessels's inequality, 110p Bose-Einstein condensation, 630
addition of, 386ff., 398ff. beta decay, 177, 735p Bose-Einstein distribution, 724p,
classical, 6ff., 26p Bethe, H. A., 59 , 747 852
commutator relations for, 350 bifurcation of energies, 702 boson , 467 ,600
eigenvalues of, 358ff. binary arithmetic, 819ff. composite, 568p
Hermiticity of, 357p binary numbers, 817ff. box
matrices for, 504ff. bit, 818 , 820 , 822, 832 energy representation for, 499
orbital plus spin, 579ff. Bjorken, J. D., 813 one-dimensional, 90ff., 687p
quantum, 349ff. blackbody radiation, 32ff., 63p three-dimensional (table) , 450
and rotation, 368, 383,410 Bloch wavefunctions, 291ff., two-dimensional, 317ff.
uncertainty relations and, 353, 303p,305p Bragg reflection, 299ff.
382p Bloch's theorem, 292 Brillouin, L., 243
anharmonic oscillator, 688p Bogoliubov, N. N., 638 broken symmetry, 177
annihilation operator, 193ff., 500 Bohm correspondence, 567p built-in potential, 658
antibonding, 625 Bohm, D., 59 , 567 byte, 822
antisymmetric operator, 489 Bohr atomic model , 39ff., 461p,
antis ymmetric state , 319, 600 526p,623p calcium energy levels , 621
anyons,600 Bohr frequency, 465 calcium P states, 618
Ashcroft, N. w., 633 Bohr magneton, 519, 521 canonical transformation, 503p,
Bohr radius, 41, 447 , 460p 549p
Baker-Hausdorf lemma, 146p Bohr-Sommerfeld quantization canonical variables, 16, 503p
Balmer series, 41 rule, 253 capacitance, 666
band gap, see energ y bands Boltzmann distribution, 6Op, 720 , Cauchy sequence, 99, 1l0p
band structure, see energy bands 722 Cauchy-Schwartz inequality, 110p
Bardeen, J., 632 Boltzmann statistics, 638p causality, 852
basis , 98, 481ff. bonding, 625 center of mass, 436 , 444p
Baym, G., 608 bonding and antibonding levels , center-of-mass frame, 774 , 777p
Belinfante, F., 59 335,625 central field approximation, 602
Bell, J. S., 559 Boolean postulates, 827 central potential, 26p, 439
Bell's theorem, 559 Boolean variable, 823 charge conjugation, 806

873
874 Index

charge density, 454, 703p coordinates quantum, 641

charge variable, 675 canonical,4, 16 symmetry, 321, 326
chemical potential, 657 cyclic , II degenerate material , 647
chopped wave, 157 generalized, 3 degenerate perturbation theory,
Church-Turing theorem, 831 good ,3 692,696
circulation, quantum, 639p correspondence principle, 94, 165, degrees of freedom, 4
classical trajectory, 182p 202,477p de Haas-van Alphen effect, 435
Clauser, J., 59, 565 angular momentum, 402p , 428p delta function, see Dirac delta
Clebsch-Gordan coefficients, 390, harmonic oscillator, 202ff. function
401p, 587 hydrogen atom , 462p densit y of states , 61p, 64p, 724p
coherent photons, 722 radiation law, 721 in various dimensions, 336ff.
column-vector representation, 481 rigid-sphere scattering, 773 density operator, 544
commutator relations, 130ff., 146p WKB method, 243ff. and equation of motion, 545 ,
and uncertainty principle, Cotten , F. A., 702 738p
140ff., 143p Coulomb gauge , 740 depletion-zone approximation,
dyadic form of, 149p Coulomb interaction energy, 617ff. 653,658
compatible operators, 140, 503p Coulomb potential, shielded, 781 depletion-zone width , 659
complement, 820 coupled representation, 388, 393, determinism, 53
complementary variables , 54 536,539,54Ip deuterium molecule, 385p , 623,
complete sequence, I lOp, 487p covalent bond , 625 629p
complete set of commuting covariant descriptions, 81Off. nuclear component for, 628
operators, 214,486 creation operator, I93ff., 500 rotational modes of, 626
complete sets of commuting cross section deuteron, 286p, 445p
operators, I43ff. differential, 764 ground state, 672ff.
complete space, l l Op total,764 DeWitt , B., 74
complex variables, 26p current densit y, 217ff. , 221p, 414p diagonalization, 483, 504p
complexity classes, 830ff. divergence of, 414p of submatrix, 693
Compton effect, 51p cyclotron frequency, 431 diagonal matrices, 483 , 497p
and wavelength , 52p cylindrical quantum well, 423 diamond, 314
and wavelength , 747 Dicke, R. H., 548, 590
computing functions, 835 d'Espagnat, B., 551 , 565 diffraction, 48, 57
conduction in solids, 313ff. Davisson-Germer expenniment, diffusion , quantum mechanical,
conservation principle 47 167p
of angular momentum, 171 Davydov, A. S., 813 dipole radiation, 464
of energy, 171 de Broglie hypothesis, 46 Dirac delta function , 74ff., 163
of momentum, 171 de Broglie wavelength, 47ff. in Cartesian coordinates, 855
of parity, 176 de Broglie wavelength and WKB in cylindrical coordinates, 857
Constantinescu, E , 747, 813 method ,244 in spherical coordinates, 856
constants of the motion, 8, 169, de Broglie wavelength, electron, Dirac equation , 165,800,810
172 50p Dirac notation, 97ff.
constants, physical , 865ff. de Broglie wavelength , proton, dispersion relations, 46, 166p
contact potential, 38 50p distribution function , 631
continuity conditions, 76, 443, Debye potential, 781 Doppler shift, 635, 637p
771 decimal numbers , 817 Drell , S. D., 813
continuity equation , 217 degeneracy, 137ff., 143, 343p , drift velocity, see velocity, drift
Cooper, L. N., 632 694, 702 , 705 dumbbell, rigid , see rotator , rigid
Cooper pairs, 632 accidental, 321, 324p
coordinate representation, 74, 214, exchange, 319 effective density of states, 647
391,847 of hydrogen energies, 449, 461P effective mass , 294
 Index 875

effective potential, 440, 441 Feynman, R., 80, 522 , 636 Green's identities, 862
Ehrenfest's principle, 59, 169, path integral formulation, 16, Green's theorem, 415p
185p 270 group property, 491 p
Einstein A and B Coefficients, 719 fine structure, see hydrogen atom Grover's algorithm, 839, 840
Einstein relations, 655 fine-structure constant, 43p, 52p,
Einstein, photoelectric theory of, 589 , 747 Hadamard gate matrix, 834
36ff. Fliigge, S., 747,813 halogens, 608, 639p
Einstein-Podolsky-Rosen Fleishmann, H. A., 415 Hamilton's principal, 16
paradox, 59, 551p, 558 Fock, V., 729, 757 Hamiltonian
Eisenbud, L., 673 forbidden domain, 25 classical, 6ff., 17, 51P
electron radius, classical, 150p central potential of, 443 perturbation, 682
electron-positron creation, 806 four vector, 567p, 795 quantum, 70, 419 , 435p, 499
energy bands, 296, 306p, 610p Fourier transform, 837 harmonic oscillator, 190ff., 469p
energy gaps , 289ff., 296 , 646, 649, Fowler-Nordheim equation, 260p electric field in, 735p
805 Fraga, S., 608 energy representation, in, 499
energy operator, 70, 106, 499 Frauenfelder, H., 177 excited states of, 197
ensemble, 115 free particle ground states of, 199
average, 115,545 in Cartesian coordinates, 404ff. matrices of, 499
entropy, 63p, 630 in spherical coordinates, 410ff. momentum space, in, 211ff.
equipartition hypothesis, 61p , 63p free -particle propagator, 161ff., representation of radiation field
Euclid's algorithm, 836 166p in,725p
Ewen, H. I., 679 Freedman, S., 59 spectrum of, 197
exchange binding, 623 Frohlich, H., 632 two-dimensional, 324ff., 696
exchange energy, 619 function space, 98 harmonic perturbation, 712
exchange operator, 538, 597 , 602p complete, 99 , 110p and Born approximation, 777
exchange symmetry, 599 linear, 98 Harris, E. G., 464, 739, 806
excitations in liquid helium, 633 Hartree, D. a., 757
exclusion principle, 37, 600 , 604 , gamma matrices, 815p Hartree-Fock approximations, 758
617,630 Gamow's formula, 262p Hartree-Fock model, 757ff.
expectation value, 76ff. Gasiorowicz, S., 623 heat capacity of liquid helium, 633
time development of, 168ff. Gaussian distribution, 79 heavy hydrogen, 460p, (see also
exponential time, 831 Gaussian state, 185p deuterium molecule)460
extended states, 296 Gaussian wave packet, 160, 170p Heisenberg model for
gcd,836 ferromagnetism, 639p
factoring an integer, 838 germanium, 314 Heisenberg picture, 553ff.
Fermi energy, 37, 64p , 315 Goertzel, G o, 416, 421 Heisenberg's uncertainty relation,
extrinsic, 649 , 654 Goffman, C., 100 ssrr., 140ff., 142p
intrinsic, 648, 654 golden rule, 716ff. for energy-time, 714
of copper, 64p and Born approximation, 777 for temperature-displacement,
of sodium, 64p Goldstein, n., 16,269,403 631
Fermi sea, 38, 805 good quantum numbers, 144,387, Heitler, w., 678
Fermi's golden rule , see golden 503p,541p helium, 615ff.
rule good variables, 144 ionization energy of, 618 ,619
Fermi, E., 717 Goodstein, Do L., 636 isotope of, 636
Fermi-Dirac distribution, 645, 852 Gordon, J. P., 722 singly ionized, 460p
ferromagnetism, see Heisenberg Gottfried, K., 74 , 370 Helium II, 633, 639p
model for Goudsmit, S., 590 Henley, H., 177
Feshbach, H., 673, 708 Gradshteyn, I. S., 402 , 451 Hermite polynomials, 200 , 206p,
Feynman diagram, 807 gravitation, 26p 325
876 Index

Hermite's equ ation , 201 , 209p interlaced zeros of wavefunctions, Landau levels , 431
Hermitian adjoint, 104ff., 489 184p degeneracy, 432
Hermitian operators, 104, 106, invariance and Fermi energy, 433
108ff., 483 , 489 of determinant, 497p Landau theory of superfiuidity,
heterojunction, 342 of eigenvalues, 494 633
hidden variables, 59, 558 of trace , 497p Lande g factor, 530, 595p
high-frequency limit, 715 inverse of an operator, 73p , 488 Laplacian operator
Hilbert space, 98ff., 376 , 382p invert bubble, 824 in Cartesian coordinates, 404
superposition principle and , inverter, 824 in cylindrical coordinates, 423
119ff. ionic bond, 625 in spherical coordinates, 413
hole, 643 , 647 isospin (or isotopic spin), 669ff. Larmorfrequency, 528, 595p
hole density, 647, 659 isotropic scattering, 764 precessional, 531
Horn, M . A., 565 Larmor rotation, 535p
Huang, K., 549 Jackiw, R. w., 59 laser, 722
Hund's rules, 610p Jackson, J. D., 465 , 468 and electric field , 724p
hybridization, 455 Jauch, J., 74 lattice constant, 292
hydrogen atom, 8, 446ff., 735p jellium, 65p least action, principle of, 268ff.
eigenenergies of, 40 , 446, 452 J -J coupling, 580 Lee, D. M., 636
fine-structure, 589ff. joint probability, 67p Legendre polynomials, 371ff.
ground state, 453 Jordan, T., 100 associated, 372
radial wavefunction of, 443, 454 Jordan-Wigner anticommutation and polar plots, 375
hydrogen chloride molecule rules, 215p, 852 tables of properties of, 373-375
rotational modes of, 365p Judge, D., 382 length contraction, 797
vibrational modes of, 197, 365p Lengyel, B. L., 722
hydrogen molecule, 623ff. Kapitza, P. L., 633 level repelling, 702
nuclear component Karwowski, J., 608 Levine, R. D., 731
wavefunction of, 628 Kemble, E., 74 Liboff, R. L., 415,556,567,631
potential of, 623 kinetic-energy operator, 415p Lifshitz, E. M ., 294 , 633, 675 ,
rotational states of, 630p Kittel, C., 306, 435 767 ,813
hydrogenic atoms, 446 Klein-Gordon equation, 640p, 798 light cone, 793
hyperfine levels, 679 Knoll, G. E, 653 line-shape factor, 784
Kramers, H., 243 linear combination of atomic
ideal gas , 62p Kronig-Penney, 706 orbitals (LCAD), 331ff.
impact parameter, 768 Kronig-Penney eigenenergies, linear operator, 73p
index of refraction, 329p 299,305p linearly independent functions,
indistinguishable particles, 597 Kron ig-Penney Hamiltonian, 294 , 133ff.
induced emission, see stimulated 297 Lippmann-Schwinger equation,
emission Kronig-Penney potential, 290, 298 785
inel astic scattering, 768 liquid helium, 633, 634
Infeld, L., 475 lab frame, 774 lithium, 641p
inference in quantum mechanics, ladder operators, 358ff. locality, principle of, 564
150p Lagrangian, 16, 268p logarithmic derivative, 443
information me asure, 822 Laguerre polynomials, 449 logic gates, 823ff.
infrared radiation, 32 relation of, to other notations Lorentz group, 797
ingoing wave , 420 (table), 450 Lorentz invariant, 797
initial-value problem, 84, 87p , 152 relation of, to other Lorentz transformation, 567p, 796,
intensity ratios for H2 and D2, 628 notations(table),451 852
interaction picture, 556ff. Landau, L., 294 , 633, 675, 767 Lorentzian line-shape factor, 784
interference, 43 and critical velocity, 633 Loudon, R., 747
 Index 877

Louisell, W., 382, 678, 849 momentum representation, 212, and exchange operator, 602p
low-bias approximation, 662ff. 214, 487p , 850 and spherical harmonics, 400p
low-pass energy filter, 66p Morse, P. M., 708 Park, D., 373
L-S coupling, 580ff., 594p multiplet separation, 616 partial waves, 762ff.
Lyman series, 41 mutually exclusive events, 67p phase shift of, 766
Pauli equation, 738p
magnetic field, 430, 718p, 734p NAND gate, 826 Pauli principle, 37, 597ff., 600,
dipole, 519 near-classical domain, 246-247 614,676
and orbital motion , 585 nearly free electron model, 703ff. Pauli spin matrices, 512, 801
magnetic moment, 519ff., 535 Nebenzahl, 1.,415 Pauli, W.,478, 552, 851
of deuteron, 669ff. neon,64lp Pauling , L., 451
of electron, 526p , 55Op, 586, Neumann-Liouville expansion, Pedrick, G., 100
806ff. 750p periodic lattice, 289ff.
of neutron, 535 neutrino, 553p periodic table , 602ff., 609
of proton, 534 neutron star, 638p perturbation theory, 681ff.
magnetic resonance, 530ff. neutron, lifetime of, 143p time-dependent, 709, 734
Magyari, E., 747 ,813 noble elements, 608 time-independent, 681, 692
majority carriers, 650 nondegenerate semiconductor, phase factor, 44, 94
maser, 722 647 phase space , 254
Mathieu equation, 708p nonintegrable systems, 702 Phillips, J. C., 645
Mathieu functions, 708p NOR gate, 825 phonons,632,635,636
matrix mechanics, 480ff. norm of a function, 99, 376 photoelectric effect , 31, 36ff.
matrix properties, 488ff. NOT operation, 823, 824 photon, 33ff.,37,62p
matrix representation of operator, n-type semiconductor, 642 Pitaevskii, L. P., 675, 813
482 nuclear magnetic resonance, Planck radiation law, 31, 6Op,
Matthews , P. T., 487 530ff., 536p 719ff.
McCullen, J. D., 551 nuclear reactor, 65p plane waves, 406, 407, 421, 427p
measurement, 74ff., 116, 120, 124, number operator, 502 plasma frequency, 166p
126, 145 p-n junction, 653ff.
on free particle, 419 observables, 68ff. biased, 661
of L z and L 2 , 421 one-component plasma, 65p unbiased, 656
Meissner effect, 632, 637p Onnes , K. H., 632 Poisson bracket, 28p
Melia, T. B., 722 optical pumping, 723 polynomial time, 831
Merman, N. D., 633 optical theorem, 768 population inversion , 723
Merzbacher, E., 212, 371,451, OR gate , 825 position operator, 72, 500
577, 785 orbital quantum number, 356, 449 positron, 550p, 623p , 805
meson,460,513,640p ortho states of D2, 627 positronium, 623p , 753p
mesonium, 460p ortho states of H2, 627 postulates of quantum mechanics,
Messiah, A., 165,451 orthohelium, 620 68ff.
metastable state, 723 oscillator strength, 746ff. potassium P states , 590
methane, 458 Osheroff, D. D., 636 potential barrier, 241
microscopic reversibility, 740 outgoing wave, 420 , 764, 766 potential well
microwaves, 32, 436p, 535p oxidation-reduction, 608 finite, bound states of, 278ff.,
minority carriers, 650 287p
mixed states , 543 parahelium, 620 finite, scattering from, 241 ff.,
mobility, 317p paralellogram law, 109p 770ff.
mod operation, 819 para states of D2, 628 infinite, bound states of, 90ff.,
momentum operator, 69,106,500 parity, 21,181, 182p,207p,288p, 216
curvilinear coordinates, 478p 310 , 471p , 676 two-dimensional, 699p
878 Index

precession in magnetic field, 527ff. Ramsauer effect , 235ff., 771 scattering amplitude, 272p, 764,
pressure, 62p Ramsey, N., 565 779 , 787
primeval fireball , 36p, 63p random phases, 546 Born approximation of, 777
principal quantum number, 448 , 449 Rayleigh-Jeans formul a, 32, 33p, scattering length, 782p
probability, 55, 88p 61p, 721 Schiff, L. I., 451, 510, 592, 696,
projection operators, 181p rectangular potential barrier, 773 ,801
projection representation, 547 scattering from, 228ff. Schlipp, P. A., 59
propagator rectangular potential well, Schmidt orthogonalization
free-particle, 161ff. scattering from , 223ff. procedure, 695, 750p
and particle in a potential field, reduced-zone, 299 Schrieffer, J. R., 632
183p reflection coefficient, 219, 227p , Schrodinger equation, 55
proper time, 795, 797 236 general properties of, 187ff.
p-type impurity, 643ff. two-dimensional, 328p time-dependent, 80, 85
Purcell, E. M., 679 Reif, F., 720 time-independent, 71
pure states, 543 relativistic corrections, 586, 590 Schrodinger picture, 553ff.
relativity, 50, 51p, 852 Schwinger, J., 673
quadrupole moment, deuteron, 673 repre sentations, 20, 211ff., 481 ff., Scully, M. 0 ., 551
quantum defect, 593p 853 second quantization, 215p, 725p
quantum dot, 429p repulsive sphere, 772 secular equation, 695
quantum entanglement, 845 resonances in transmission, (see selection rules , 465, 467
quantum logic gate, 832 also Ramsauer effect) 239, semiconductor
quantum numbers, good , 21 239p acceptor-type, 643
quantum parallelism, 835, 837, resonant absorption, 716 donor-type, 642
841 rest mass , 795 extrinsic, 314, 641
quantum search algorithm, 839 Richardson, R. C., 636 intrinsic, 314, 641, 649
quantum versus classical physics, Robertson-Schrodinger relation, n-type ,649
20ff. 141 p-type,651
quantum well , 342 Rodrigues formula, 372, 450 sequence, periodicity, 837
quantum wire, 429p Rojansky, V., 303 Shea, R., 551
quasi-momentum, 293ff. rotational modes , see deuterium Shimony, A., 565
qubit,832 molecule, see hydrogen Shor's algorithm, 838
chloride, see hydrogen short-time approximation, 715
radial equation molecule Slater determinant, 612ff.
for central potential, 439ff. rotator, rigid, 4, 363 Smith, P. w., 747
free-particle, 419ff. in collision, 470p Snell's law, 329p
for hydrogen atom, 446 rotons,636 solid angle, 369
radial momentum, 411 Runge-Lenz vector , 469 Sommerfeld model, 37
radial probability, 442 Rutherford atom, 31, 762 source-free wavefunction, 413p
nondimensional, 454, 460p Rutherford cross section, 781 spectral resolution of unity, 487p
radial wavefunction Rydberg constant, 40 , 43p , 447 spectrum
for central potential, 440 Ryzhik, I. M., 402 , 451 emis sion , for hydrogen, 41
continuity conditions of, 443 energ y levels for calcium, of,
free-particle, 415ff. Saleh , B. E. A., 678 621
radiation Salpeter, E. E., 747 excitation energy for helium,
elementary theory of, 463ff. Saxena,K.,608 619
power formula for, 466 , 467 Saxon, D., 243 particle in a large box , of, 432
radiation field, number-operator scattering permitted k values, 293
representation of, 725p three-dimensional, 762ff. vibrational diatomic molcules
radio waves, 32 one-dimensional, 222ff. of, 197
 Index 879

spherical Bessel equation, 445p Stem-Gerlach experiment, 523ff. Thomson's "plum pudding"
spherical Bessel functions, 416ff. stimulated emission, 712, 719, 723 model, 762
annihilation and creation Streater, R., 851 threshold frequency, 38, 39p
operators for, 475p strong nuclear forces , 670 tight-binding approximation,
equation of, 417 , 426 Sturm comparison theorem, 185p 305p,331,707
plots of, 417 subsets , relation to binary code, 842p time development of
table of, 418 sudden perturbation, 731 wavefunction, see initial
spherical energy operator, 461p domain of validity, 730 value problem
spherical harmonics , 367ff., 400p summational probability, 67p time dilation, 797
addition theorem, 390, 427p superconducting ground state, 633 time reversibility, 273p, 572
spherical quantum well, 422 superfluidity, 630, 633 time-ordering operator, 751p
spin angular momentum, 356 superlattice, 342 tin
eigenstates of, 513 superposition principle , 115ff. resistance curve of, 632
matrix representation of, 513ff. and uncertainty principle, 124ff. superconductivity of, 632
operators of, 512ff. superposition state, 548p Tornonaga, S., 451
vs. orbital, 356 supersyrnrnetry, 330p torque , 9
spin wavefunction surface waves, 166p Townes, C. H., 722
for two electrons, 536, 539 S-wave resonances, 771 trace, 489
superposition, 525 S-wave scattering, 770ff. Tralli, N., 416, 421
spin, addition of, 536 symmetric operator, 489 transfer matrix , 571ff.
for three electrons, 540 symmetric state, 319 transformation of basis, 493
spin-dependent force, 673 attraction of particles in, 322, transformation of state vector, 493
spin-orbit coupling, 584ff. 323p transition probabilities, 730, 734
spin-statistics relation , 600, 630, symmetrization requirements on transmission coefficient , 219ff.,
851 wavefunctions, 319, 598, 236ff., 238p
spinor, 511, 515p 624 two-dimensional, 329p
spontaneous emission, 718, 723, symmetry breaking , 702 WKB estimate and, 256ff.
743 transpose of an operator, 489
spring constant, 191 Teich, M. C., 678 triangle inequality, 1lOp
square waveform, 121 temperature, kinetic definition of, tritium, 735p
square well, see potential well 630 truth table, 828
squeezed states, 678p Term diagram tunneling, 64, 228ff.
standing waves, 153 for calcium, 621 Turing machine , 830
at band edges, 307ff., 706 for helium , 620 turning points , 25, 188
Stark effect, 700ff. for rigid rotator, 364 twenty one-ern line, 679p
state Term notation , 364, 450-451, 581 two's complement, 820
classical, 19 thermal speed, 166p two-particle problem , 436ff.
quantum , 20 thermodynamics two-photon decay, 559
stationary phase , 242 and adiabatic process, 63p
stationary state, 86, 153 first law of, 63p Uhlenbeck, G. E., 590
statistical mechanics , quantum , second law of, 63p ultraviolet spectrum, 32
146p, 544, 630ff., 852 temperature relation, 630 unbound states, 216ff.
Stefan-Boltzmann constant, 35p Thomas factor, 586 uncertainty relation , see
Stefan 's law, 63p Thomas-Fermi Heisenberg's uncertainty
Stehle , P., 371 equation, 474 relation, see
step potential , scattering from, model, 472ff. Robertson-Schrodinger
222ff. potential, 473, 781 relation
Stem-Gerlach analyzer, 526p, Thomas-Reiche-Kuhn sum rule, uncertainty relation for
718p 746 number-ph ase, 678p
880 Index

uncoupled repre sent ation, 388, virial theorem, 207p , 461p Wittke, J. P., 548 , 590
393,536 viscosity, 634 WKB approximation, 243ff.
unitary operator, 49Op, 492, 497p, visible spectrum, 32 of bound states, 247ff.
832 Von Buttlar, u, 671 connecting formulas in,
unitary-similarity transformation, 248ff.
492 Wang, So, 653 eigenenergies in, 255ff.
Watson, GoN., 708 WKB wavefunctions
vacuum fluctuations, 743 wave packet, 46 , 83, 156ff., 17Op, and transmission coefficient,
vacuum state , 215p 222p 256ff.
variational principle, III p minimum uncertainty, 263p work function , 37
vector potential, 430 scattering of, 241ff. world line, 793
vector space , 98 wavefunction properties, see Wronskian, 148p, 185p
complete, normed, linear, 98, 99 continuity conditions,
velocity interlaced zeros x rays, 51p , 535p
drift,317p waveguide, 166p XOR gate, 826 , 834
group, 83 Weissbluth, M., 607
phase , 157 Wentzel, G., 243 Yariv, A., 723
wave, 82 Whittaker, E. T., 708 Yarnell, J. L., 635
Venn diagram, 828 Wien's displacement law, Yilmaz , H., 852
vibrational modes 35p
of carbon monoxide molecule, Wien's law, 33, 35p Zeeman effect, 595p
198p Wightman, A. S., 851 Zeiger, H. J., 722
of deuterium molecule, 626, Wigner crystallization, 65p zeros of Bessel functions, 425
629p Wigner, E. P., 673 Zurek, W. H., 559
of radiation field, 61p Wilson, E. B., 451