Research: Science and Education

edited by
Advanced Chemistry Classroom and Laboratory Joseph J. BelBruno
Dartmouth College
Hanover, NH 03755

On the Solution of the Quantum Rigid Rotor W

Cory C. Pye
Department of Chemistry, Saint Mary’s University, Halifax, NS Canada B3H 3C3; [email protected]

The author has been developing a one-semester honors allows eq 3 to be separated into two ordinary differential
quantum chemistry course to follow courses in thermody- equations
namics and kinetics. Students will have had two years of cal- 2
culus but generally little experience with differential equations h 2 1 d    h 2m 2
    (5)
or linear algebra. The “laboratory” component of the course 2I  d2 2I
is composed of a series of formal mathematical derivations
of results assumed in the lecture part of the course, designed h2 1 d d m2
to improve and teach new mathematics skills and, at the same  sin        E     (6)
time, allow students to demonstrate for themselves why a
2I sin  d d sin2 
particular result is valid. A general overview of the pedagogi-
cal approach can be found in ref 1, where the solution of the where  is the separation constant, which can be alternatively
harmonic oscillator problem is discussed and an intuitive so- expressed in terms of some number m defined in eq 5. Equa-
lution for the behavior at infinity is presented. tion 5 is easily solved to give
This article presents some finer details of Legendre’s e im 
equation as applied to the solution of the rigid rotor prob-     (7)
lem not usually covered in most texts. The solution to the 2
quantum rigid-rotor problem is widely discussed in elemen- with ()  (  2) ensuring that m is an integer, whereas
tary quantum mechanics texts (2–9) and in this Journal (10). eq 6 is much more difficult to solve. Some texts simply state
The Laplacian operator expressed in spherical polar coordi- the solutions to this equation, the associated Legendre func-
nates (r, , ) is tions (3, 5, 6, 9). Other texts continue towards the solution
of this equation by making the change of variable z  cos
1  2  1 1   and then stating the solution (7). Equation 6 becomes
2  r  sin 
r 2 r r r 2 sin    
d m2
1 1  2 (1)
dz
1  z 2  ddz   
1  z2
  0 (8)

r sin  2
2 2
where
The laborious but straightforward derivation of this equa- 2 IE
tion has been designed as lab; however, it is not the focus of   (9)
h2
this article. For a rigid rotor, r is fixed and the first term of
eq 1 applied to a function vanishes. The external potential At this point, texts containing more in-depth treatments
V(, ) is zero. The Schrödinger equation, start to diverge. Some first partially solve the special case of
m  0 by the method of power series, deriving a two-term
h2 2 recursion formula, assume that it diverges for z  1, and
    ,    V  ,     ,    E   ,   (2) mention the relationship between the Legendre polynomials
2

and the associated Legendre functions (4), as does Dushman
becomes in his classic article in this Journal (10). Other texts (2, 8),
and Dushman (10) introduce a change of dependent vari-
able
h2 1 1   1 2
 sin     ,   m
 r 2 sin  
2  sin 2  2
(3) z  1  z 2
  2 Q z (10)
 E   ,  
without any justification whatsoever as to why this form was
h
where is Planck’s constant divided by 2,  the reduced
⁄ chosen. Some texts in elementary differential equations prove
mass, E the energy, and  the wavefunction. Noting that the the existence of solutions via singular Sturm–Liouville theory
moment of inertia is I  r2 and assuming that the wave- (11, 12). Consideration of the singularities at z  1 via the
function is separable method of Frobenius for series solutions near a regular sin-
gular point can give a form like eq 10, but a similar form
  ,            (4) had been previously assumed (2). It is worth mentioning that

460 Journal of Chemical Education • Vol. 83 No. 3 March 2006 • www.JCE.DivCHED.org

 Research: Science and Education

Whippman has examined the convergence of the power se- diverges. Now
ries solution in some depth (13). The approach taken here
and used in the quantum chemistry course is to develop the 2k  2k  1  
a 2k  2  a (17)
students’ mathematical “intuition” to dispel the notion that  2 k  2   2 k  1 2 k
the instructor is “pulling a rabbit out of a hat”.
and thus
Convergence of the Special Case, m = 0
2 k  2 k  1  
bk 1  b (18)
When m  0, eq 8 simplifies to  2k  2   2 k  1 k
d Because
dz
1  z 2  ddzP  P
bk
lim  1
2 (11) k  bk 1 (19)
2 d P dP

 1  z  dz 2  2z
dz
 P  0
the ratio test is indeterminate (14). We rewrite the ratio as
where we have replaced (z) with P(z) to emphasize the dif-
ferent solution. Standard application of the power series so-
lution about z  0,
bk  2k  2   2k  1 4k 2  6k  2
 
bk 1 2k  2k  1   4k 2  2k  

P (z )   ak z k (12)  1 
4k    2
k 0
4k 2  2k  
to eq 11 gives the two term recursion relationship 1 k  
 1  1 2 (20)
k 4k  2 k  
k  k  1  
ak  2  a (13)
 k  2   k  1 k  1
 1  o
k k
with arbitrary a0, a1. If we assume that, for all integers k,
k(k 1) 0, then the solution is not a polynomial but a
true infinite series. The radius of convergence of this infinite where o represents the remaining terms and characterizes
series about z  0 is one, which suggests that for all points the convergence of the series if it is greater than or less than
|z| < 1 the series converges, but there may possibly be a con- 1. With  1, Raabe’s test is indeterminate (14). Fortunately,
vergence problem at z  1. Because of how z was obtained, Gauss’s test can be applied (14) if eq 20 is expressed as fol-
we need not consider the case |z| > 1. lows
To use the standard theorems of convergence involving
ratios of successive terms, we need to recast the sequence {ak} bk  1
 1   O  1 (21)
in terms of two new sequences for the even and odd terms: bk  1 k k
bk  a 2 k ; ck  a 2 k  1 (14) where 1 and 1. Because  > 0 the series properly
If both of the associated series converge, then their sum con- diverges.
verges. If one diverges and the other converges, then the sum We now demonstrate that the solution of eq 11 is not
diverges. If both diverge, then there are two cases to con- square integrable at z1. For z1, the second derivative
sider. For k > [(  0.25)1!2  0.5] , the terms of {bk} and {ck} multiplier 1  z2 0, making z1 a regular singular point.
no longer alternate in sign. If after this point, they are both Using the method of Frobenius, we define a new power se-
of the same sign, then the series must diverge. If after this ries expanded about z1 as
point, they are of opposite sign, then this series in principle 
ks
could converge, but then the case for z  !1 must diverge. R z    g k  z  1 (22)
The net result is that we only need to show that one of these k 0
series
for some real number s and g0  0, to generate the recur-
 
rence relation. Substituting R(z) back into eq 11, and defin-
bk or  ck (15) ing g!1  0, gives
k 0 k 0

diverges to show that at least one of 

  g k  k  s   k  s  1
 
k k k  1 (23)
 ak 1 or  ak  1 (16)  2 g k  1  k  s  1
2
 z  1k  s  0
k 0 k 0

www.JCE.DivCHED.org • Vol. 83 No. 3 March 2006 • Journal of Chemical Education 461

 Research: Science and Education

The recurrence relation is therefore gk(k  s)(k  s  1)  where P (n) = dnp!dzn and we note that
2gk+1(k  s  1)2  0. The indicial equation is the special case
for k  !1 that gives the quadratic equation 2g0s2  0, which j j
dP  
d2P   j  2 j 1
has a double root of zero. The theory of differential equa-  P ;  P  (29)
tions (11) gives two general solutions of eq 11 of the form dz 2 dz

 The solutions of the slightly more complicated eq 28 are the

n
y1  z   1  d n  z  1 derivatives of the solutions of eq 11.
n 1 At this point we focus our attention on eq 8. We can
 (24) assume a trial solution
n
y 2  z   y1  z  ln z  1   fn  z  1 k
n 1 z  1  z2   u z  (30)

where the two series are convergent. For values of z  1   where k is some constant that we are free to choose. If we
close to 1, y1(1  )  1 and y2(1  )  ln||. We know substitute eq 30 into eq 8, and collect terms of like powers
therefore that the solution must diverge as ln||. The area of 1  z2, we will obtain for k  1 (the case for k  1 is sim-
under the square of this curve encloses the family of rect- pler),
angles bounded by the coordinate axes and the axes passing
through the vertex (, (ln)2). Because d 2u k 1 du
2 2
1  z2   4kz  2 z
lim   ln     dz dz
 
(25)
k
   2k u 1  z 2
the function is not square integrable, and the wavefunction (31)
is not acceptable. Therefore our assumption that k(k  1)  2 2 2 2 k 1
 4z k  m u 1  z  0
  0 must be false, and, for an acceptable wavefunction,
there must exist some integer k for which k(k  1). We
call this special integer l, so that l(l  1)and therefore which can be rearranged to
obtain our energy quantization condition
h 2 l  l  1 d2 u k 1 du
1  z2  4kz  2 z
E  (26) dz 2
dz
2I k

   2k  4k 2 u  1  z 2
and our Legendre polynomial solutions (32)
l 2 k 1
 4k 2  m 2 u 1  z  0
Pl z    ak z k (27)
k 0
Dividing through by (1  z2)k will give
Relating the Legendre Polynomials d2u du
to the Associated Legendre functions 1  z2 2
 2 z 2k  1    2k 1  2k u
dz dz
1 (33)
If we compare the simple differential eq 11 to the more  4k 2  m 2 u 1  z 2  0
complicated differential eq 8, we notice some features. The
integer m appears only as a square, so the solution should
depend on |m|. We also note the prominence of the form If we set our free constant k  |m|!2, we get rid of the singu-
1  z2, which suggests that it has some additional relevance larity of the last term to obtain
in the solution of eq 8.
d2 u du
Eyring (2) gives an intuitive account of the relationship 1  z2  2z m  1
between the Legendre polynomials, the associated Legendre dz2 dz
polynomials, and the Rodriguez formulas for these. One can (34)
obtain these relationships in a different way by exploring these    m m 1 u  0
differential equations, which has been designed as a lab. By
differentiating eq 11 j times, one can show by mathematical Comparing eq 34 with eq 28, and equating j  |m| and u(z)
induction that  P ( j)(z) completes the result that
m
j j
d 2P   dP   m
2
1  z  dz 2  2  j  1 z
dz
z  1  z2  2 Pl z  (35)

(28)
j
   j  j  1 P    0 We note that because Pl (z) is a polynomial of order l, we
must have |m| < l for a nontrivial solution.

462 Journal of Chemical Education • Vol. 83 No. 3 March 2006 • www.JCE.DivCHED.org

 Research: Science and Education

Acknowledgments 5. Dykstra, C. E. Introduction to Quantum Chemistry; Prentice

Hall: Englewood Cliffs, NJ, 1994; pp 100–110.
I would like to thank the Chemistry 412 (Quantum
6. Griffiths, D. J. Introduction to Quantum Mechanics; Prentice
Chemistry) students of 2001–2004 at Saint Mary’s Univer-
Hall: Upper Saddle River, NJ, 1995; pp 124–128.
sity for working through the associated Legendre equation
7. House, J. E. Fundamentals of Quantum Mechanics; Academic
lab and correcting many errors in the early versions.
Press: San Diego, CA, 1998; pp 63–64.
8. Levine, I. N. Quantum Chemistry, 5th ed.; Prentice Hall:
W
Supplemental Material
Upper Saddle River, NJ, 2000; pp 108–115, 123–136.
Two sample labs, The Laplacian Operator in Spherical 9. Ratner, M. A.; Schatz, G. C. Introduction to Quantum Me-
Polar Coordinates and The Solution of the Associated Leg- chanics in Chemistry; Prentice Hall: Upper Saddle River, NJ,
endre Equation, are available in this issue of JCE Online. 2001; p 56–59.
10. Dushman, S. J. Chem. Educ. 1935, 12, 436–444.
Literature Cited 11. Boyce, W. E.; DiPrima, R. C. Elementary Differential Equa-
1. Pye, C. C. J. Chem. Educ. 2004, 81, 830–831. tions and Boundary Value Problems, 4th ed.; Wiley: New York,
2. Eyring, H.; Walter, J.; Kimball, G. E. Quantum Chemistry; 1986.
Wiley: New York, 1944; pp 52–60, 72–75. 12. Andrews, L. C. Elementary Partial Differential Equations with
3. McQuarrie, D. A. Quantum Chemistry; University Science Boundary Value Problems; Academic Press: Orlando, FL,
Books: Mill Valley, CA, 1983; pp 207–209. 1986.
4. Eisberg, R.; Resnick, R. Quantum Physics of Atoms, Molecules, 13. Whippman, M. L. Amer. J. Phys. 1966, 34, 656–659.
Solids, Nuclei, and Particles, 2nd ed.; Wiley: New York, 1985; 14. Hyslop, J. M. Infinite Series, 5th ed.; Oliver and Boyd:
pp N1–3. Edinburgh, 1959; pp 43–50.

www.JCE.DivCHED.org • Vol. 83 No. 3 March 2006 • Journal of Chemical Education 463