Chapter 4: One-Dimensional Potentials

The Potential Well

Bound states in a potential well ( −V0  E  0 ) :

The case E  0 corresponds to a particle which is confined in a bound state. When well
is deep enough, there are bound states with discrete energies.

 0 x  -a region I

V( x ) = − V0 - a  x  a region II
 0 x  a region III


Classically, when E  0 the particle is completely confined to the region −a  x  a ; it

will bounce back and forth between x = − a and x = a with constant momentum
p = 2m(V0 − E ) .

Quantum mechanically, the solutions are particularly interesting for they are expected to
yield a discrete energy spectrum and wave functions that decay in two regions x  − a
and x  a , but oscillate in −a  x  a .

If the potential well is attractive, there are also solutions for E  0 when E  V0 (no
meaning for E  −V0 ). The negative energy states have discrete. E = − E is called the
binding energy of the particle.

Let us write the Schrodinger equation in these three regions.

2
d 2 u I (x)
− + 0u I (x) = − Eu I (x) ; x  -a
2m dx 2
2
d 2 u II (x)
− − V0 u II (x) = − Eu II (x) ; - a  x  a
2m dx 2
2
d 2 u III (x)
− + 0u III (x) = − Eu III (x) ; xa
2m dx 2

The general solutions of Schrodinger equation for region x  -a becomes

d 2u I ( x ) 2m E
− u I ( x) = 0  u I ( x) = C1e x + C1'e − x ; x  -a
dx 2 2

1
 2m E
where  2 = 2
. Square integrable property (finite property) of the wave function
requires that
−a

 u ( x) dx   
2
I
  C1 = 0 .
'
−

lim e
− x
→ 
x →− 

Therefore the wave function in this region ( x  -a ) becomes uI ( x) = C1e x .

The general solutions of Schrodinger equation for region x  -a becomes

d 2uII ( x) 2m(V0 − E )
2
+ 2
uII ( x) = 0  u II ( x) = A1eiqx + B1e −iqx ; -a  x  a
dx

2m(V0 − E )
where q 2 = 2
. uII ( x) is written in terms of sine and cosine as

uII ( x) = A1 eiqx + B1 e−iqx = ( A1 + B1 ) cos qx + i ( A1 − B1 )sinqx

cos qx +i sin qx cos qx −i sin qx
A B

uII ( x) = A cos qx + B sin qx

This satisfies the requirements of the wave function in the region -a  x  a .

The general solutions of Schrodinger equation for region x  a becomes

d 2uIII ( x) 2m E
− uIII ( x) = 0  u III ( x) = C2e − x + C2' e x ; xa
dx 2 2

Square integrable property (finite property) of the wave function requires that



u
( x) dx   
2
III
  C2 = 0 .
'
a
x
lim e →  
x → 
Therefore the wave function in this region ( x  a ) becomes uIII ( x) = C2e− x .

Thus the physically acceptable solutions are found as

uI ( x) = C1e x ; x  -a
uII ( x) = A cos qx + B sin qx ; -a  x  a
uIII ( x) = Te − x ; xa

2
In region -a  x  a , the well is centered about the origin. Therefore the solutions will be
either even or odd functions (even solution is given by cosqx and the odd solution is
given by sinqx). Discrete spectrum in an attractive potential well becomes

At the point x = − a , the continuity property of the wave functions requires

u I (x)|x =-a = u I (x)|x =-a  C1e- a = A cos qa − B sin qa

du I (x) du (x)
|x =− a = II |x =− a   C1e- a = −qA sin qa + qB cos qa
dx dx
Using these two expressions we obtain 
C1e- a = A cos qa − B sin qa   C1e- a −qA sin qa + qB cos qa
 =
 C1e- a = −qA sin qa + qB cos qa  C1e- a A cos qa − B sin qa
q( A sin qa + B cos qa)
 =
A cos qa − B sin qa

At the point x = a , the continuity property of the wave functions requires

u II (x)|x =a = u III (x)|x =a  A cos qa + B sin qa = C2e- a

du II (x) du (x)
|x = a = III |x = a  −qA sin qa + qB cos qa = − C2e- a
dx dx

Using these two expressions we obtain  again

A cos qa + B sin qa = C2e- a  − C2 e- a −qA sin qa + qB cos qa

 =
−qA sin qa + qB cos qa = − C2e- a  C2 e- a A cos qa + B sin qa
q( A sin qa − B cos qa)
 =
A cos qa + B sin qa

Let us equal these two 

3
 q ( A sin qa + B cos qa ) q ( A sin qa − B cos qa )
=
A cos qa − B sin qa A cos qa + B sin qa
A2 sin qa cos qa + AB sin 2 qa + AB cos 2 qa + B 2 cos qa sin qa =
A2 sin qa cos qa − AB sin 2 qa - AB cos 2 qa + B 2 cos qa sin qa
AB(sin 2 qa + cos 2 qa ) = − AB (sin 2 qa + cos 2 qa )  2 AB = 0  AB = 0
1 1
As a result,

uI ( x) = C1e x ; x  -a
uII ( x) = A cos qx + B sin qx ; -a  x  a
uIII ( x) = Te − x ; xa

 A  0 and B = 0 : even solutions u II ( x ) = A cos qx  α=qtanqa

AB = 0  
 A = 0 and B  0 : odd solutions uII ( x) = B sin qx  α= − qcotgqa

where
 q ( A sin qa − B cos qa ) 
 A 0 =   α=qtanqa
B =0
 A cos qa + B sin qa  BA=00
 q ( A sin qa − B cos qa ) 
 A= 0 =   α= − qcotgqa
B 0
 A cos qa + B sin qa  BA=00

The transcendental equations ( α=qtanqa and α= − qcotgqa ) cannot be solved directly.

They can be solved numerically on a computer, or graphically.

For the graphical solutions, we have

The even solutions:

2mE
The transcendental equation for even solution is given by  = qtanqa where  = 2

2m(V0 − E ) 2mV0
and q = 2
. Let us change of variable as  = 2
− q 2 . So the equation

becomes
2mV0
= 2
− q 2 = q tan qa

or dimensionless variable is taken as

2ma 2V0
a = 2
− (qa) 2 = qa tan qa

2mV0 a 2
Define y  qa and  = 2
, we then have

 a =  − y 2 = y tan y
4
The solution of this transcendental equation for even solution can be obtained by
a a  − y2
graphical method. Let us plot = tan y and =
as a function of y. The
y y y
points of intersections determine the energy eigenvalues (bound states) for even solution.
These form a discrete set. The functions  a has zeros at  a = 0,  , 2 ,... and has
asymptotic behavior at y =  a =  / 2,3 / 2,5 / 2,...

Note:
a) There is always at least one (for small λ ).
b) For large λ there are more bound sates.
c) For large λ the intersection points is given approximately
1
y  (n + )π , n = 0,1,2,...
2
1
y qa n
2
2
2 1 2
2m(V0 E) / n
2
2 2
En V0 2
(2n 1) 2 0
8ma

d) This is familiar the spectrum of even solutions in the infinite box.

5
The odd solutions:

The transcendental equation for odd solution is given by  = −q cot qa where

2mE 2m(V0 − E ) 2mV0
= 2
and q = 2
. We can write as  = 2
− q 2 . So the equation

becomes

2mV0
= 2
− q 2 = −q cot qa

or dimensionless variable is taken as

2ma 2V0
a = 2
− (qa)2 = −qa cot qa

2mV0 a 2
Define y  qa and  = 2
, we then have

 a =  − y 2 = − y cot y

a a  − y2
Let us plot = − cot y and =as a function of y. The intersections of these
y y y
functions give the energy values for odd solution. The function cot y has asymptotic
behavior at y =  a =  , 2 ,3 ,...

6
Note:
a) The large λ behavior is
y  n π , n = 1,2,...
π 22
y = qa  n π  E = -V0 + ( 2n ) 2  0
8ma 2
This is familiar the spectrum of odd solutions in the infinite box.

b) There will be an intersection only is

2
π
2mV0a 2
λ − (π/2)  0 
2
 
2
2

The binding energy of deuteron can be found by using this result.

Problem:

Consider an electron of mass m in a bound stationary state of an atom. Suppose that

Coulomb potential binding the electron is a finite square well of width 2a 0.4nm and of
25 2 2
depth V0 60eV , where 1 nm = 10-9 m . With the notation
8ma2

2 m V0 a 2 2m( V0 − E ) 2m E
= , y = qa , q= and  = , V(x)
2 2 2
-a +a
the eigenvalue conditions may be written in the form
x
2
 − y2 −y I II III
= tan y and = − cot y for the even and odd
y y
solutions, respectively, whose graphical solutions are given by −V0
the figures below.

a) How many bound states do the potential energy in the question support? Explain
what the parity of each is.
b) Calculate the energy eigenvalues of the ground state and the first excited state.
2 2
(Hint: Use 2.4 eV )
8ma2
7
A Second way to calculate the energy eigenvalues is possible. For this, a relation between
 a and qa is considered as

 = 2mE / 2  2mV
 ( a ) + ( qa ) = 2 0 a 2  R 2
2 2

q = 2m(V0 - E ) / 2 

The right side of the above equation defines a circle of radius R. The solutions are
determined by the points where the circle intersects the tangent function,  a = qa tan qa
and the cotangent function  a = −qa cot qa .

8
Example:

Consider the one-dimensional problem of a particle of mass m in a potential given by


V (x)

  x0 region I
a 
0 V ( x) =  0 0 xa region II
x −V xa
 0 region III
-E
−V0

a) Obtain the wave functions and show that the bound state energies ( −V0  E  0 )
are given by
 2m(V0 − E )  V −E

tan a =− 0
  E
 
b) Find the minimum value of V0 for one bound sate.
c) Find the minimum value of V0 for two bound sates.
d) Find the lowest value of V0 giving n bound state.

Note that since all odd solutions vanish at x = 0 , the odd solutions will be the same as for
the potential well.

The Schrodinger equation for different regions are

For 0  x  a :
2
d 2 II d 2 II
− 2
− V 
0 II = − E  II  2
+ k12 II = 0
2m dx dx
 II ( x) = A sin k1 x + B cos k1 x
where k1 = 2m(V0 − E ) / . The wave function must be continuous at x = a boundary,
 I ( x) x =0 =  II ( x) x =0  0 = A sin 0 + B cos 0  B = 0
The wave function for 0  x  a becomes  I ( x) = A sin k1x

9
For x  a :
d 2 III
2
d 2 III
− 2
+ 0 III = − E  III  2
− k22 III = 0
2m dx dx
 III ( x) = Ce− k2 x + Dek2 x

where k2 = 2mE / . Finite property of the wave function requires that

lim III ( x) → finite . To be a wave function lim III ( x) → 0  D = 0 . The wave
x → x →
− k2 x
function for x  a becomes  III ( x) = Ce . The physically acceptable solutions
becomes

 I (r ) = 0 ; x0

 II (r ) = A sin k1 x ; 0 xa

 III (r ) = Ce xa
- k2 x
;

a) The requirement that wave functions and their first order derivatives are continuous at
x = a.

 II ( x) x =a =  III ( x) x =a  A sink1 a = Ce − k2a 

 k1
d d   tank a = −
 A k1 cosk1 a = − k 2 Ce − k2a 
1
 II ( x) =  III ( x) x =a k2
dx x=a dx 

k1 k2
tank1 a = − or cot k1 a = −
k2 k1
2m(V0 − E ) /
(
tan a 2m(V0 − E ) / )=− 2mE /
=−
V0 − E
E

Dividing these two terms we have

k1a cot k1a = −k2 a .

We have also

( k1a ) 2
+ ( k2 a ) =
2 (
2ma 2 V0 − E ) + 2ma2 E =
2ma 2V0
 R2
2 2 2

These two transcendental equations can be solved graphically and energy eigenvalues are
given by the intersection of equations.

10
  3
b) If Rthere is only one bound state for R (n=1 ground state). Thus, the
2 2
minimum value of V0 for one bound sate
 3 2ma 2V0    2 2
2
R  R2 = =    V0 =
2 2 2
2 8ma 2

c) Similarly, for two bound states the value of R must be in the interval
3 5
R . The minimum value of V0 for two bound sates becomes
2 2

3 5 2ma 2V0  3  9 2 2
2
R  R2 = =    V0 =
2 2 2
 2  8ma 2

 
d) We extend these result to a general one. If n −  R  n +
, there are n
2 2
bound state and then the lowest value of V0 giving n bound state becomes
   
2
2ma 2V0 
2 2

2(
n −  R  n + =  n −   V0 = 2n − 1)
2
R =
2
2 2 2
 2 8ma

11
Results:

We see even solutions at intersections of blue and red curves (always at least one) and
odd solutions at intersections of blue and green curves.

• There is tunneling of particle into classical forbidden region in which V0  E .

• Amount of tunneling depends exponentially on V0 − E .
• There are finite number of solutions. The number of bound states depends on the depth
of the well, but there is always at least one (even parity) state.
• Potential is even, so wave functions must be even or odd.
• Limit as V0 →  , we recover the infinite well solutions.

Example: The quantum well

Quantum well is a “sandwich” made of two different semiconductors in which the

energy of the electrons is different, and whose atomic spacings are so similar that
they can be grown together without an appreciable density of defects:

Quantum well is used in many electronic devices (some transistors, diodes, solid-state
lasers)

12
Delta Function Potentials

Single delta function potential

Consider a potential where spatial behavior is given by  ( x) . For an attractive force, V(x)
is

2λ
V(x) = − δ(x)
2ma

Where a is an arbitrary quantity with dimension of length, and  is an dimensionless

quantity that characterizes the strength of potential.

The Schrodinger equation for this potential when E  0 becomes

2
d 2u ( x ) 2

− 2
−  ( x)u ( x) = − E u ( x)
2m dx 2ma
d 2u ( x ) 2 m E 
2
− 2 u ( x) +  ( x)u ( x) = 0
dx a
2m E
where E = − E . If  2 = 2
, we then have

d 2u ( x ) 
2
−  2u ( x) +  ( x)u ( x) = 0
dx a

The solutions everywhere, except x = 0 , must satisfy the equation

d 2u ( x )
−  2u ( x ) = 0
dx 2
Two independent solutions of this second order linear differential equation with constant
coefficient are e x and e− x . The finite property of the wave functions requires
lim u(x) → 0 . Thus, for x  0 , the solution becomes
x →

u I ( x) = Ae x for x  0

u ( x) = 
− x
u II ( x) = Ae for x  0

where lim u I (x)= lim Ae x → 0 and lim u II (x)= lim Ae − x → 0 . Even with a delta
x →− x →− x → x →

function potential, continuity of the wave function is required but its first derivative is
discontinuous at x=0. At x=0, amplitudes must be the same. The normalization is used to
find A,

13
 0− 
2 2
 uI ( x) dx +  uII ( x) dx = 1
− 0 +
0− 
x 2 2
 Ae dx +  Ae− x dx = 1
− 0 +
0− 
2 x −2 x
dx = 1  A = 
2
A  e dx +  e
− 0 +
1 2 x 0 − 1 1 −2 x  1
e = − e =
2 − 2 2 0 + 2

Thus, the normalized wave functions are obtained as

u I ( x) =  e x for x  0
u ( x) = 
− x
u II ( x) =  e for x  0

Let us return to the full equation

d 2u ( x ) 
−  2
u ( x ) +  ( x)u ( x) = 0
dx 2 a

Integrating both sides over x from − to + where  is an arbitrary small positive

number, we get

d 2u ( x ) 
2
−  2u ( x) +  ( x)u ( x) = 0
dx a

du ( x)   
−  2  u ( x)dx +   ( x)u ( x)dx = 0
dx − − a −
u (0)
 2
d u ( x)  
 dx −  2
 u ( x )dx + u (0) = 0
− dx 2 − a

where u(0) =  . As  → 0 ,
  

lim  u ( x)dx  lim  u (0)dx =  lim  dx = lim  = 0 .

 →0  →0  →0  →0
− − −
2

We then get

du ( x) du ( x) 
− + u (0) = 0
dx x = 0+ dx x = 0− a
14
Let us use the above solutions for x  0 and x  0 ,

duII ( x) d
=  e− x = − 
dx x =0+ dx x = 0+
duI ( x) d
=  e x = 
dx x =0− dx x = 0−

Thus, the eigenvalue condition becomes

 
−  −   + u (0) = 0   
a   −  −   +  = 0 =
 a 2a
u (0) =  

λ
The eigenvalue condition is κ = . The binding energy can be found
2a


 
2 2 2 2 2 2
2m E
 =
2
E=− =−   =−
2
2m 2m  2a  8ma 2

This is unique solution, no excited bound state are expected.

The probability of finding the particle in the interval 0  x  x 0

x0 x0 2 x0
 e− x dx =   e −2 x dx
2
P(0  x  x0 ) =  uII ( x) dx = 
0 0 0
1 −2 x x0
− e
2 0

1
P(0  x  x0 ) = (1 − e−2 x0 )
2

The probability of finding the particle in the interval x 0  x  0

0 x0 2 x0
 e x dx =   e 2 x dx
2
P( x0  x  0) =  u I ( x) dx = 
− x0 − x0 − x0

1 2 x 0
e
2 − x0

1
P( x0  x  0) = (1 − e −2 x0 )
2

Total probability at x0 will be

15
 P ( x0 ) = P (0  x  x0 ) + P ( x0  x  0)
1 1
= (1 − e −2 x0 ) + (1 − e −2 x0 )
2 2
−2 x0
= 1− e

Note that
0 
2 2
P(−  x  −) =  u I ( x) dx +  u II ( x) dx
− 0
0 2  2
=   e x dx +   e− x dx
− 0
0  1 1
2 x −2 x
=  e dx +  e dx =  ( + ) =1
− 0 2 2
1 2 x 0  1
− e−2 x =
1 1
e =
2 − 2 2 0 2

Example:
1
At which point x0 does the probability become ?
2
1 1 1
1 − e−2 x0 =  −e−2 x0 = − 1  e−2 x0 =
2 2 2
1 1
ln(e−2 x0 ) = ln( )  −2 x0 = ln1 − ln 2  x0 = ln 2
2 =0 2
2m E 1
where  = and x0 = ln 2 = ln 2 .
2 2 2m E

16
Example:

Consider a particle of mass m in one dimension subject to the potential

2

V ( x) = −  ( x)
2m a
2

where V0   0 . The wave function for E  0 is found as
2m a
− x
u I ( x) =  e x for x  0
u ( x) =  e 
− x
u II ( x) =  e for x  0


2
2m E
where  = 2
  .
2
 2a 
a) Find the binding energy of the ground state. Are there excited states?
b) What is the probability that the particle remains bound when suddenly the
strength of the potential  becomes (i)  / 2 and (ii) 4 ?

a. Binding energy of the ground sate is obtained as

  
2 2 2 2 2 2
2m E
2 = 2  E = − =−   = − .
2m 2m  2a  8ma 2
There is no excited bound state.

b.


− x  − x
u ( x) =  e = e 2a
2a

' − x
 →  '  u '( x) = e 2a
2a
2 2
2  '   − 1 (  +  ') x
P = u '( x) u ( x) =  u ' ( x)u ( x)dx =
*

2 e 2a dx
− 2a 2a 0

2
'   − 1 (  +  ') x '  4a 2 4 '
= 4  e 2a dx = 4 (0 − 1) =
2a 2a 0 2a 2a ( +  ') 2
( +  ') 2

1 2
2a − (  +  ') x
e 2a
(  +  ')
0

 1 2 2 8
 ' =   P = 2 =  89%
4 '  2 9 / 4 9
P= 
( +  ') 2
 ' = 4  P = 16 = 16  64%
2

 25 2 25

17
Example:

Consider a particle of mass m in one dimension subject to the potential

2

V ( x) = −  ( x)
2m a
2

where V0   0 . For the scattering case, the energy of particle is E  0 .
2m a
a) Find the wave function for E  0
b) Calculate the reflection and transmission coefficients as a function of the wave
2mE
number k = 2
.

a. The Schrodinger equation for this potential when E  0 becomes

2
d 2u ( x ) 2

− 2
−  ( x)u ( x) = Eu ( x)
2m dx 2ma
d 2u ( x) 2mE 
2
+ 2 u ( x) +  ( x)u ( x) = 0
dx a
2mE
If k 2 = 2 , we then have

d 2u ( x ) 
2
+ k 2u ( x) +  ( x)u ( x) = 0
dx a
The solutions everywhere, except x = 0 , must satisfy the equation

d 2u ( x )
2
+ k 2u ( x ) = 0
dx

Two independent solutions of this second order linear differential equation with constant
coefficient are eikx and e−ikx . The acceptable functions becomes

u I ( x) = Aeikx +Be−ikx ; -  x  0

u ( x) = 
u II ( x) = Ce 0 x
ikx
;

The continuity property of wave functions at x=0 requires A+B=C. But their first
derivatives are discontinuous at x=0.

Let us return to the full equation

d 2u ( x ) 
2
+ k 2u ( x) +  ( x)u ( x) = 0
dx a

Integrating both sides over x from − to + where  is an arbitrary small positive

number, we get

18
 d 2u ( x ) 
2
+ k 2u ( x) +  ( x)u ( x) = 0
dx a

du ( x)   
+ k 2  u ( x)dx +   ( x)u ( x)dx = 0
dx − − a −
=0 u (0) =C
where u (0) = C and as  → 0 ,
  

lim  u ( x)dx  lim  u (0)dx =  C lim  dx = lim C = 0 .

 →0  →0  →0  →0
− − −
2
We then get

du ( x) du ( x) 
− + u (0) = 0
dx x =0+ dx x =0− a
Let us use the above solutions for x  0 and x  0 ,

duII ( x) d
= Ceikx = ikC
dx x =0+ dx x = 0+
duI ( x) d
= ( Aeikx + Be−ikx ) = ik ( A − B)
dx x =0− dx x = 0−

Thus, the eigenvalue condition becomes

  2mV 
ikC − ik ( A − B ) + C = 0   2ik − 2 0  C = 2ikA
C−A a  
2 mV0 / 2
Transmission coefficient
2
  
C 2ik
2
 −2ik  2ik  4k 2
T= = =   = 2
2mV  −2ik − 2mV0   2ik − 2mV0  2  2mV0 
A 2ik − 2 0
 2  2  4k +  2 
   
1 1 1
T= 2
= 2 2
=
1  2mV0  mV mV02
1+ 1 + 2 04 1 +
4k 2  2  k 2E 2
2
B
The reflection coefficient R = becomes
A
1 k2 4
R = 1− T = 1− = 1− 2 4
m 2V02 k + m 2V02
1+ 2 4
k
k 2 4
+ m V0 − k 2 4
2 2
m 2V02 1 1
R= = = 2 4 =
k 2 4
+ m V0
2 2
k 2 4
+ m V0
2 2
k 2E 2
+ 1 1 +
m 2V02 mV02
19
Double delta function potential

The double delta function potential is more intensity because it gives a quick way to
study the properties of a narrow well. A particle of mass m, moving in one dimension, is
confined by a double delta function potential defined by

2λ
V(x) = − δ(x - a) + δ(x + a)
2ma

where   0 . Since the potential is an even function, bound states have even or odd parity
and therefore any solution can be expressed as a linear combination of even and odd
solutions.

The Schrodinger equation for this double delta function potential is

d 2 u(x) 2mE λ
− 2 u(x) + δ(x - a) + δ(x + a) u(x) = 0
dx 2
 a

where E =− E . The solutions everywhere, except x =  a , are found as

 κ e − κx ;x a
2
d u(x) 
− κ u(x) = 0  u( x ) = Acosh( κx ) + B sinh( κx ) ; - a  x  a
2

dx 2  κx
 κ e ; x  -a

Even solutions

 κ e − κx ;x a

u( x ) = Acosh( κx ) ; - a  x  a
 κx
 κ e ; x  -a
λ
tanh y = − 1
y

20
At the intersection point, the eigenvalue condition becomes
λ λ
κ
a 2a
When we compare this result with the single delta function potential, the energy for the
double potential is lower than the energy in the single potential
2 2
2  λ  2  λ  1
−    E even  −  
2m  a  2m  a  4

Example:
The energy of an electron bound to two protons is lower than an electron bound to one
proton.

Odd solutions

 κ e − κx ;x  a

u( x ) =  Bsinh( κx ) ; - a  x  a
 κx
 κ e ; x  -a
−1
λ λ 
cot ghy = − 1  tanh y =  − 1
y y 

At the intersection point, the eigenvalue condition becomes

2
λ 2  λ  1
κ  E odd  −  
2a 2m  a  4

As a result

E even  E odd  | E even |  | E odd |

The odd solutions (if there are a bound state) is less strongly bound than even solution.

Example
Metals generally have a crystalline structure. The Kronig-Penney model is the theory of
metals, insulators, and semiconductors. In this model, a series of repulsive delta-function
potentials is used as
2
 
V (x) =   ( x − na)
2m a n=−

21
The Harmonic Oscillator

We have discussed several potentials which are discontinuous functions of position with
constant values (for example; potential step, barrier, etc.). The harmonic oscillator
potential is continuous function of position (another example for continuous function is
the Coulomb potential VC (r ) 1/ r ).

Consider one-dimensional simple harmonic motion of a particle of mass m under an

attractive force F = −kx (Hook’s law) with angular frequency  = k / m . The potential
energy produced by this force becomes
xf xf
1 2
V f − Vi = −  Fdx  V = −  (−kx)dx = kx
xi xi
2
Such a parabolic potential (the quantum harmonic oscillator) is a very important problem
in quantum mechanics as well as classical physics. For example, it serves as a first-order
approximation for the bond vibrational problem in diatomic and (with coupling)
polyatomic molecules. Both the vibration of the atoms in a solid and of the
electromagnetic field itself can be modeled by independent linear harmonic oscillators.
Simple harmonic oscillator is prototype for any system involving oscillations:

• The vibration of atoms in diatomic molecules

• The acoustic and thermal properties of solids (arise from atomic vibrations)
• Magnetic properties of solids (vibrations in the orientation of nuclei)
• Electrodynamics of quantum systems (the vibration of electromagnetic waves)
• Etc.

The simple harmonic oscillator can be used to describe almost the system in which there
is small vibrations about a point of stable equilibrium. Consider a Taylor series of a
continuous function W(x) about x=a,
dW 1 d 2W
W ( x) = W (a ) + ( x − a ) + ( x − a) 2 + ...
dx x =a 2 dx 2 x =a

dW d 2W
Since W (x) has a minimum at x = a we have = 0 and  0 . Choose a
dx x =a dx 2 x =a
is the origin of coordinates and W(a) as the origin of the energy. Thus, the harmonic
1 2 d 2W
oscillator potential V (x) = kx (with k = ) is the first approximation to W (x)
2 dx 2 x =a

22
The harmonic oscillator model is very important in physics and it occurs widely in nature
and is utilized in many manmade devices, such as clocks and radio circuits. Simple
harmonic oscillator can be used to describe almost any system in which an entity is
executing small vibrations about a point of stable equilibrium. Classically, Newton’s
second law gives

d 2x
F = −kx = m  x(t ) = A cos(t +  )
dt 2
where A is the amplitude and  is angular frequency. The system has continuous energy
starting from zero (energy can have any value) since A is arbitrary. However, quantum
mechanics predicts that the total energy can assume only a discrete set of values because
the particle is bound by the potential to a region of finite extent. To find the allowed
energy values predicted by Schrodinger quantum mechanics, the time-independent
Schrodinger equation for the simple harmonic oscillator potential must be solved.

There are essentially two different methods to solve for simple harmonic oscillator

1. Polynomial method
2. Operator method (Chapter 6)

23
Polynomial method

The classical Hamiltonian is of the form

p2 1 2
H= + kx
2m 2
The Schrodinger equation for the harmonic oscillator potential is
2
d 2u ( x ) 1 2
− + kx u ( x) = Eu ( x)
2m dx 2 2
Where the frequency of the oscillator is  = k / m . Let us introduce a dimensionless
mw 2E
variables y = x and  = . We then get the simple form of this equation
w
d 2u (y)
2
+ ( − y 2 )u ( y ) = 0
dy

The solution u ( y ) and its first derivative must be single valued, continuous, and finite for
all y from − to + . To solve this differential equation, let us look at the asymptotic
behavior of u ( y ) as y →  .

For any finite eigenvalue of energy, the term involving  is negligible ( y 2 completely

dominates over  ) in the limit y 2 →  . In this case, the differential equation (for very large y
we have an asymptotic equation) is approximately given by
d 2u (y)
2
− y 2u ( y )  0
dy
e y
2
/2
which has the asymptotic solutions u ( y ) . The general solution becomes
u ( y ) = Ae − y + Be + y
2 2
/2 /2

where A and B are arbitrary constants. The solution e + y

2
/2
is not an acceptable solution
(physical solution) since lim e+ y →  that requires to set B=0. This result suggests
2
/2

y →
2
that we look for a solution of differential equation having the form u ( y ) ey /2
for very
large y . Note that lim e− y → 0 satisfy finite property of solution. This allows only the
2
/2

y →

asymptotic solution with the negative sign in the exponent. This suggests looking for
solutions having in the form

u ( y ) = Ae − y /2 H ( y )
2

24
This solution are to valid for all y. To evaluate H(y), we consider its derivatives with
respect to y

du ( y ) dH ( y )
= − Aye − y /2 H ( y ) + Ae − y /2
2 2

dy dy
d 2u ( y ) − y 2 /2 2 − y 2 /2 − y 2 /2 dH ( y ) − y 2 /2 dH ( y )
2
− y 2 /2 d H ( y )
= − Ae H ( y ) + Ay e H ( y ) − Aye − Aye + Ae
dy 2 dy dy dy 2
− y 2 /2  dH ( y ) d 2 H ( y ) 
= Ae  − H ( y ) + y H ( y ) − 2 y dy + dy 2 
2

 

d 2u ( y )
We then substitute u ( y ) and into the full equation,
dy 2
d 2u (y)
+ ( − y 2 )u ( y ) = 0
dy 2
 dH ( y ) d 2 H ( y ) 
Ae− y − + − + + ( − y 2 ) Ae − y /2 H ( y ) = 0
2 2
/2 2
 H ( y ) y H ( y ) 2 y 2 
 dy dy 
Dividing by Ae− y /2 , we have
2

d 2 H ( y) dH ( y )
2
− 2y + ( − 1) H ( y ) = 0 “Hermite equation”
dy dy

This differential equation determines the functions H(y). This equation is solvable by
means of the power series technique,


H ( y ) =  am y m
m=0

dH ( y ) 
=  mam y m −1
dy m=0

d 2 H ( y)  

dy 2
= 
m=0
m ( m − 1) am y m−2
= 
m=0
(m + 1)(m + 1)am + 2 y m

0+ 0+  m ( m −1) am y m−2
m= 2

 m→m+ 2
m−2 ⎯⎯⎯⎯⎯  
where we use  m( m−1) am y →  ( m+ 2)( m+ 2−1)am+ 2 y m+ 2−2 =  (m+ 2)(m+1)am+ 2 y m .
m= 2 m =0 m =0
Substituting them into the differential equation, we obtain

25
   

 (m + 1)(m + 2)a
m=0
m + 2 y − 2 y  mam y
m

m=0
m −1
+ ( − 1)  am y m = 0
m=0


 mam y m−1+1
m =0

 (m + 1)(m + 2)a

m=0
m+ 2 − 2mam + ( − 1)am  y m = 0

This equation can only be true for every possible value of y if the term inside the square
brackets is itself 0 for every m. The coefficients of y m satisfy the recursion relation as
2m − ( − 1)
(m + 1)(m + 2)am + 2 − 2mam + ( − 1)am = 0  am + 2 = am
(m + 1)(m + 2)

where m=0,1,2,… Because the index jumps by 2 we may select both coefficients a0 and
a1 independently. For given a0 and a1 , the even and odd series can be generated
separately. For an arbitrary value of  both even and odd series will contain an infinite
number of terms. This will not lead to acceptable eigenfunctions. To see this, consider
either series and evaluate the ratio of coefficients,

am + 2 2m −  + 1
=
am (m + 1)(m + 2)

For large m,
am + 2 2m −  + 1 2m + ... 2
=
am (m + 1)(m + 2) m 2 + ... m

2
Let us compare it with the power series expansion of e y
1 1 1 1 1
e y =  (y 2 ) m = 1 + y 2 + y 4 + y 6 + .... + ym + y m+2
2

m=0 m ! 2! 3! ( m / 2)! ( m / 2 + 1)!

1
(m / 2 + 1)! (m / 2)! 1 2
for large m, = =
1 (m / 2 + 1)! ( m / 2 + 1) m
(m / 2)!
As a result, for large values of m
am+ 2 2 
→ 
am m 
  H ( y) e
y2

1 1 a 2
e y =  (y 2 ) m   m+1 → 
2

m=0 m ! m! am m 

2
If the series of H(y) contain an infinite number of terms, the behavior of the solution e y
becomes
u ( y ) = Ae − y /2 H ( y ) → Ae − y /2e y
2 2 2

26
however, lim Ae− y /2e y = lim Ae y →  which is not acceptable (not square-integrable)
2 2 2
/2

y → y →

for an eigenfunction. The only way to avoid this divergence of u(y) at large y is require
that the series must terminate, which means that H(y) must be a polynomial in the
variable y 2 . To obtain a finite wave function, the expansion for a given value of m must
terminate at some value of m which we call mmax = n , where the quantum number
n = 0,1, 2,3,... is a positive integer or zero. The coefficient becomes an  0 , while an+1
must vanish. The cut off condition on the series also serves to determine the energy
values as

2n −  + 1
mmax = n  an + 2 = an = 0  2 n −  + 1 = 0
(n + 1)(n + 2)
2n −  + 1 = 0 

  En = ( n + 2 ) w, n = 0,1, 2,...
1
2E
= 
w

There are eigenvalues of the simple harmonic oscillator potential, expressed in terms of
its classical oscillatory frequency w. Energy eigenvalues E n depend only on the quantum
number n. The n constant is the number of energy quanta in the oscillator. We find the
possible energy values discrete. The ground sate energy (the lowest energy or the zero
1 1
point energy of this oscillator) is obtained as E0 = w = h .
2 2

In this case the energy levels are equidistant, and the E difference of consecutive levels
is
E = En+1 − En = w = h

That is the energy of the linear harmonic oscillator may only change by an integer
multiple of the energy quantum h .

Let us now turn to the wave function u ( y ) = Ae − y /2 H ( y ) where H ( y ) are polynomials

2

of order n (called Hermite polynomials), and satisfy the Hermite equation

d 2 H ( y) dH ( y )
2
− 2y + ( − 1) H ( y ) = 0
dy dy
Hermite polynomials, which are the solution of Hermite equation, are consistent with

d n − y2 d
H n ( y ) = (−1) e y2
e = e y /2 ( y − ) n e − y /2
2 2
n
n
dy dy

27
The first few Hermite polynomials are found as

H 0 ( y) = 1
H1 ( y ) = 2 y
H 2 ( y) = 4 y 2 − 2
H 3 ( y ) = 8 y 3 − 12 y
H 4 ( y ) = 16 y 4 − 48 y 2 + 12

The wave functions for the linear harmonic oscillator becomes as

u n ( x ) = N n e − H n ( x)
2 2
x /2

mw
where  = and y =  x . Normalization condition is used to find N n
 2 
Nn 
 u ( x)  e H (y) dy = 1  N n =
− y2
dx =
2 2

−
n
 −
n
 2n n !
 2n n!

The solution of the differential equation gives the following harmonic oscillator
eigenfunctions in terms of Hermite polynomials H n (α x) and the energy eigenvalues

un ( x ) = e − H n ( x)
2 2
x /2

 2 n!
n

En = ( n + 12 ) w, n = 0,1, 2,...


 e H n (y) H m (y)dy = 0 for n  m gives the

−y2
The property of Hermite polynomials
−
normalized eigenfunctions u(y) satisfy the orthogonality condition as


 u ( y)u ( y )dy =  mn
*
n m
−
This means that any small peroration of a system from its equilibrium state will give rise
to small oscillations which are decomposable into normal modes, independent oscillators.

As a result, the eigenfunctios un ( y) and the energy eigenvalues En have the following
properties:

• The eigenfunctions u n (x) are orthogonal functions.

• Energy eigenvalues are quantized.
• The energy levels are equidistant ( E = En+1 − En = w ).
• E 0  0 is completely quantum mechanical effect, and can be interpreted in terms
of uncertainty principle.

28
 1/4
 α2 
• u 0 (x) =   e − α x /2 is a Gauss packet with a maximum at x = 0.
2 2

 π 
u n (x) is even for even n
• u n (-x) = (-1)n u n (x)  
u n (x) is odd for odd n

The probability densities for quantum harmonic oscillators,  n ( y ) , are plotted as solid
2

lines for the quantum numbers n = 0, 2, 5, 20. The probability densities of the classical
harmonic oscillators having the same energies as the quantum oscillators are plotted as
dashed lines. The classical limits of motion are shown by the vertical lines. The ground
state is plotted in the upper left panel of Figure. Contrast this behavior with the classical
harmonic oscillator, which has a minimum in the probability at x = 0 and maxima at the
turning points. Probability for n=20 oscillator shows a good agreement between classical
curve and the average value of quantum mechanically probability (correspondence
principle).

29
This chapter describes one of the most powerful predictions of quantum mechanics. We
have seen that material particles can penetrate into classically forbidden regions due to
their wave nature.

Important concepts in this chapter are

• The potential step
• The potential well problem
• The potential barrier
• The harmonic oscillator

30
Some relations:

 ( n + 1)( n + 2 )
 1 n +1  for m = n + 2
 for m = n + 1  2 2
  2  2n + 1
 1 n  for m = n
un x um =  for m = n − 1 un x 2 um =  2 2
 2 
 0 for m  n  1  n ( n − 1) for m = n − 2
  2 2
 
0 else

 2
 n +1  ( n + 1)( n + 2 ) for m = n + 2
 for m = n + 1  2
 2
 2

u = 2 (
 n 2 − 2n + 1) for m = n
un um = − for m = n − 1 un
x 2 m
 2 x  2
 0 for m  n  1  n ( n − 1) for m = n − 2
  2
 
 0 else

31
Example:
1
At t=0 a particle in a harmonic oscillator potential V = kx 2 has a wave function
2
2 i
ψ(x) = u 0 (x) + u 2 (x)
5 5
where u 0 (x) and u 2 (x) are real, normalized and orthogonal eigenfunctions for the
ground and second excited states of the harmonic oscillator.
a) Find the expectation value of kinetic energy.
2
d2
 T  =−  
2m dx 2
 
 
2  2 * * 2 
 2 d2  2   i  d2  i   2  d 2
i d 2

=−  u0 u +     0
u u2 +     u2 u0 + u2 u2 
2 0 2 2 2
2m  5 dx  5  5 dx  5  5 dx 5 dx 
  2 2
 2 2
 2 5 2 
 − − 
 2 2 2 2 
2 2 2 2 
  2  2  i   −i   2  i 
=−  − +    2 +   2 − 5
2m 2 

5  5  5   5  5  5 

2
m 1  4 5  9
=− − −  = 
2m 2  5 5  20

b) Calculate the expectation value of potential energy.

1
 V  = m 2  x 2 
2
 
 
 2 * * 2 
1 2 2  2   i   i   2  i 
= m  2
u0 x u0 +    
2
u0 x u2 +    
2
u2 x u0 + u2 x 2 u2 
2  5  5  5  5  5 5 
20 +1 ( + )( + ) ( − ) 22 +1
 0 1 0 2 2 2 1 
 2 2 2 2 2 2 2 2 

1  2
2 2 
1  2  i   −i  2  i 
= m 2  +   2 +    2 + 5
2 2  5
2
 5  5   5  5  5 

1 4 5 9
= m 2  + = 
2 2m  5 5  20

c) Find the expectation values of total energy.

2 2
2 i 4 1 1 1  4 5 9
 E = E0 + E2 =  0 +   +  2 +   =  +  = 
5 5 5 2 5 2  10 10  10
or
9 9 9
 E =  V  +  K = + = 
20 20 10

32
Example:

a) Find the expectation value of p ˆ x for a harmonic oscillator in the superposition

1
state  ( x, t ) =   0 ( x, t ) +  1 ( x, t ) 
2
where  n ( x, t ) = e−iEnt / u n ( x ) is the time-dependent and u n (x) is the time-
independent eigenfunctions of the harmonic oscillator.

b) If the particle has a wave function at t = 0 given by

i 3
 ( x) = u1 ( x) + u3 ( x) , what is the expectation value of total energy
10 10
of this particle? Write your result in terms of ω . (Hint:
H | un = En | un  ( n + 12 )  | un , n = 0,1,2,... )

a.
pˆ x   ( x, t ) pˆ x  ( x, t )
1 1
=   0 ( x, t ) +  1 ( x, t )  pˆ x   0 ( x, t ) +  1 ( x, t ) 
2 2
1
=   0 ( x, t ) pˆ x  0 ( x, t ) +  0 ( x, t ) pˆ x  1 ( x, t ) +  1 ( x, t ) pˆ x  0 ( x, t ) +  1 ( x, t ) pˆ x  1 ( x, t ) 
2
 0 ( x, t ) = e−iE0t / u 0 ( x ) →  0 ( x, t ) = e + iE0t / u 0 ( x )
 1 ( x, t ) = e−iE t / u1 ( x ) →  1 ( x, t ) = e + iE t / u1 ( x )
1 1

 
 
pˆ x = (−i )  u0 ( x) 
1 d d d d
u0 ( x) + e − i ( E1 − E0 ) t/ u0 ( x) u1 ( x ) + ei ( E1 − E0 ) t/ u1 ( x ) u0 ( x ) + u1 ( x ) u1 ( x )
2  dx dx dx dx 
 
m =1 m = 0
0 n = 0  m = n +! n =1  m = n −1 0
 
1  0 + 1 i ( E1 − E0 ) t/ 1  1 1 i ( E1 − E0 ) t/
= ( −i ) e − i ( E1 − E0 ) t/  +e (− )  = (+i ) e − e − i ( E1 − E0 ) t/ 
2  2 2  2 2
(E −E ) t  1 0
2 i sin  
 

mw 1  ( E − E0 ) t  mw  ( E − E0 ) t 
= i2 s in  1  =− sin  1 
2   2  
mw
−
2

b.
i 3
H  ( x) = H u1 ( x) + H u3 ( x )
10 10
E1 u1 ( x ) E3 u3 ( x )

i 3
= E1 u1 ( x ) + E3 u3 ( x )
10 10
2 2
i 3 1 9  3 63  23
E = E1 + E3 = E1 + E3 = w  +  = w
10 10 10 3 10 7  20 20  10
w w
2 2

33