Sei sulla pagina 1di 30

C H A P T E R

9
Time-Independent Perturbation
Theory
In theory, the Schrdinger equation allows us to solve any quantum mechanical
systemexactly. We simply insert the potential V and solve for the wave function
and the energy E. Unfortunately, there are very few potentials, such as the innite
square well or the Coulomb potential of the hydrogen atom, for which a simple
exact solution exists. In order to make any further progress, we need to develop
some techniques for nding approximate solutions to the Schrdinger equation.
This chapter and Chapter 11 are devoted to a very important set of these techniques
called perturbation theory.
The basic idea of perturbation theory rests on a simple general argument. Sup-
pose we begin with a potential for which we can solve the Schrdinger equation
exactly, such as the innite square well of width a; recall from Chapter 4 that the
ground-state energy and wave function are given by =

2/a sin(x/a) and
E =

h
2

2
/2ma
2
. Now suppose we make a tiny change in V such as a small notch
in the center of the potential (Figure 9.1). We cannot solve the Schrdinger equa-
tion for this new potential, but intuition suggests that a small change in V ought
to produce a small change in and in E. This intuition is correct. The reason is
that the Schrdinger equation is a special kind of differential equation: it is linear,
i.e., and its derivatives are taken only to the rst power. Linear differential equa-
tions like the Schrdinger equation have the property that small changes in the
parameters produce small changes in the solution. The fact that a small change in
V produces a small change in and E is the fundamental idea of perturbation
theory.
9.1 DERIVATION OF TIME-INDEPENDENT PERTURBATIONTHEORY
Now we will calculate mathematically the change in E produced by an arbitrary
small change in the Hamiltonian H. [Although we talk generically about a change
in the Hamiltonian, this usually amounts to a change in the potential.] Assume we
have a Hamiltonian H for which we can solve the Schrdinger equation exactly.
We need to consider two possibilities for a small change in H: either the change
in H is constant in time, or it varies as a function of time. If the change in H is
constant in time, we are dealing with time-independent perturbation theory, which
is the subject of this chapter. If, on the other hand, the change in H varies with time,
we have time-dependent perturbation theory, which is discussed in Chapter 11.
195
196 Chapter 9 Time-Independent Perturbation Theory
0 a

E
0 a
'
E'
FIGURE 9.1 The innite square well on the left has the ground-state wave function and
energy =

2/a sin(x/a) and E =

h
2

2
/2ma
2
. Asmall change in the potential, shown
on the right, produces a small change in and E.
We will now derive what happens if we have a small, time-independent change
in the Hamiltonian. Assume that we have a Hamiltonian H
0
, for which we can nd
all of the eigenstates |
n
with energies E
n
:
H
0
|
n
= E
n
|
n

Note that this is just shorthand for an innite set of equations: H


0
|
1
= E
1
|
1
,
H
0
|
2
= E
2
|
2
, and so on. For example, the |
n
could correspond to the hy-
drogen wave functions, the spin eigenfunctions of an electron in a magnetic eld,
or any other set of wave functions that are exact solutions of the Schrdinger
equation. Now add a small perturbation H

to the Hamiltonian:
H = H
0
+H

(9.1)
Here is taken to be a dimensionless small number, 1, so that the perturbation
H

is small compared to the original Hamiltonian H


0
. We would like to solve the
new Schrdinger equation:
H| = E| (9.2)
In this equation H is the new (perturbed) Hamiltonian of Equation (9.1), | is
the new wave function after we have added the perturbation to the Hamiltonian,
and E is the new energy. Of course, we cannot solve this equation exactly (or
we wouldnt be bothering to develop perturbation theory), but we can use some
mathematical techniques to see how the change in the Hamiltonian changes the
wave functions and energies.
The rst step is to write the new energy and wave function in Equation (9.2) as
a power series in the small number that appears in Equation (9.1):
E = E
n
+E
[1]
+
2
E
[2]
+ (9.3)
| = |
n
+|
1
+
2
|
2
+ (9.4)
In these equations, |
n
and E
n
are the original eigenfunction and energy before
we apply the perturbation; since Equation (9.2) has an innite number of solu-
tions (corresponding to a small perturbation applied to any of the eigenfunctions
9.1 Derivation of Time-Independent Perturbation Theory 197
of H
0
) we have to pick a particular eigenfunction |
n
to perturb. The energies
E
[1]
, E
[2]
, . . . and the wave functions |
1
, |
2
are unknown; our goal is to solve
for them.
Recall that we are interested in the small change in E which results from our
small change H

in the Hamiltonian. The rst term in Equation (9.3) gives us


the energy before we apply the small perturbation. The rest of the terms give us
the small change in the energy due to the small change in H. But if is tiny,
only the rst of these terms really matters. For instance, if we take = 10
6
, then

2
= 10
12
,
3
= 10
18
, and so on. So the third term in Equation (9.3) is tiny
compared to the second term, and the fourth term is tiny compared to the third
term. That means the change in E is essentially E
[1]
, and we can ignore all of the
other terms in Equation (9.3). [Exception: if, for a particular perturbation, E
[1]
is
exactly zero, we will have to go further and solve for
2
E
[2]
.]
Substituting the expressions for E and | from Equations (9.3) and (9.4) into
the Schrdinger equation given by Equation (9.2) gives the following rather messy
result:
(H
0
+H

)(|
n
+|
1
+
2
|
2
+ )
= (E
n
+E
[1]
+
2
E
[2]
+ )(|
n
+|
1
+
2
|
2
+ )
(9.5)
Although it looks like this is only making matters worse, we can now apply two
ideas to simplify this equation. First, recall that if is small, then is much larger
than
2
, which is much larger then
3
, and so on. This means that for very small ,
the terms in Equation (9.5) with different powers of do not affect each other, so
Equation (9.5) must be satised separately for each individual power of . Equating
powers of in this equation gives

0
H
0
|
n
= E
n
|
n
(9.6)

1
H

|
n
+H
0
|
1
= E
n
|
1
+E
[1]
|
n
(9.7)

2
H
0
|
2
+
2
H

|
1
=
2
E
[1]
|
1
+
2
E
[2]
|
n
+
2
E
n
|
2
(9.8)
Equation (9.6) is just the original unperturbed Schrdinger equation, which is
reassuring, but we cannot make further progress with Equations (9.7) and (9.8)
unless we know what happens when H
0
operates on |
1
and |
2
. However, we
dont even know what |
1
and |
2
are! Nonetheless, we can solve this problem
by applying a second idea. Recall that for any Hamiltonian H
0
, we can nd a set
of eigenfunctions |
m
which form an orthonormal basis, i.e., any wave function
| can be written as
| = c
1
|
1
+c
2
|
2
+ +c
m
|
m
+
So even though we dont know |
1
and |
2
in Equations (9.7) and (9.8), we can
expand them out as linear combinations of eigenfunctions of H
0
, and we do know
198 Chapter 9 Time-Independent Perturbation Theory
what happens when we apply H
0
to these eigenfunctions. We write
|
1
= c
1
|
1
+c
2
|
2
+ +c
m
|
m
+ =

i
c
i
|
i
(9.9)
and
|
2
=

i
d
i
|
i
(9.10)
Then when H
0
is applied to the sums in Equations (9.9) and (9.10), it simply pulls
out the appropriate energy in front of each term:
H
0
|
1
= H
0
_

i
c
i
|
i

_
=

i
E
i
c
i
|
i
(9.11)
and
H
0
|
2
=

i
E
i
d
i
|
i
(9.12)
When H
0
is applied to |
1
in Equation (9.7), it produces the sumgiven in Equation
(9.11). Then we get
H

|
n
+

i
E
i
c
i
|
i
= E
n

i
c
i
|
i
+E
[1]
|
n
(9.13)
We would like to solve this equation to nd E
[1]
. We take the inner product of
n
|
with both sides of the equation, recalling that
n
|
n
= 1:

n
|H

|
n
+

i
E
i
c
i

n
|
i
= E
n

i
c
i

n
|
i
+E
[1]
(9.14)
Note that |
n
is the original eigenfunction which satises the unperturbed
Schrdinger equation, while the |
i
s are the complete set of such eigenfunc-
tions, including |
n
as a particular case. Hence, when we take the inner product
of
n
| with |
i
, we get zero for i = n and one for i = n. This selects out i = n
from the sum in Equation (9.14) giving

n
|H

|
n
+E
n
c
n
= E
n
c
n
+E
[1]
so
E
[1]
=
n
|H

|
n
(9.15)
where E
[1]
is the dominant or lowest order (i.e., lowest power of ) change in the
energy due to the small change H

in the Hamiltonian.
We can now calculate the second-order change in E, i.e., the term in Equation
(9.3) which is proportional to
2
. Of course, it is reasonable to ask why we would
9.1 Derivation of Time-Independent Perturbation Theory 199
ever want to know this, since E
[1]
is much larger than
2
E
[2]
. Normally, the
second order perturbation is irrelevant except for one very important case: if E
[1]
is exactly zero, then
2
E
[2]
gives the dominant change in the energy. To nd

2
E
[2]
, we substitute both Equations (9.9) and (9.10) into Equation (9.8). When
H
0
operates on the sum of eigenvectors as in Equation (9.12), we get

i
d
i
E
i
|
i
+
2
H

i
c
i
|
i
=
2
E
[1]

i
c
i
|
i
+
2
E
[2]
|
n
+
2
E
n

i
d
i
|
i

As before, we take the inner product with


n
|, and now we substitute
n
|H

|
n

for E
[1]
. Solving for
2
E
[2]
, we get

2
E
[2]
=
2

i
c
i

n
|H

|
i

2
c
n

n
|H

|
n

The right-hand side is the sum over all eigenfunctions |


i
minus the particular
eignfunction |
n
, so it can be written as

2
E
[2]
=
2

i=n
c
i

n
|H

|
i
(9.16)
Nowall we have to do is to nd the coefcients c
i
which rst appeared in Equation
(9.9). We go back to Equation (9.13), but now instead of applying
n
| to both
sides, we apply an arbitrary eigenfunction
m
|, where
m
| =
n
|. This gives

m
|H

|
n
+E
m
c
m
= E
n
c
m
+0
so that
c
m
=

m
|H

|
n

E
n
E
m
(9.17)
Substituting this expression for the c
i
s in Equation (9.16) gives the nal expression
for
2
E
[2]
:

2
E
[2]
=

i=n

n
|H

|
i

i
|H

|
n

E
n
E
i
=

i=n
|
n
|H

|
i
|
2
E
n
E
i
(9.18)
To summarize: if we start out with some Hamiltonian H
0
for which we can
solve the Schrdinger equation exactly, and we begin in an eigenstate |
n
with
energy E
n
, then after we change the Hamiltonian by the small amount H

, the
dominant change in the energy, proportional to , will be given by Equation (9.15),
while the next largest change, proportional to
2
, will be given by Equation (9.18).
200 Chapter 9 Time-Independent Perturbation Theory
While we have used to remember which terms are larger than others, we can
now simplify our expressions by writing the change in H as
H = H
0
+H
1
where
H
1
= H

is very small compared to the unperturbed Hamiltonian H


0
. Then we write the rst
and second order changes in the energy as
E
(1)
= E
[1]
and
E
(2)
=
2
E
[2]
These changes in the energy are given by
E
(1)
=
n
|H
1
|
n
(9.19)
and
E
(2)
=

i=n
|
n
|H
1
|
i
|
2
E
n
E
i
(9.20)
where the |
i
s which appear in Equation (9.20) are all of the other eigenfunctions
of H
0
aside fromthe one being perturbed. Equations (9.19) and (9.20) are the main
result of this section; they give the rst-order and second-order perturbations to
the energy from an arbitrary perturbation to the Hamiltonian. As usual, the inner
products which appear in Equations (9.19) and (9.20) can represent a variety of
different mathematical possibilities. If the wave functions and the perturbation
are functions of position (as in Example 9.1), then these inner products will be
integrals. If the eigenstates are spin states and the perturbation is a function of spin
operators (as in Example 9.2), then these inner products will be matrix products.
There is one case in which the entire argument falls apart. If the original
eigenfunction |
n
is degenerate with some other eigenfunction |
m
of H
0
, i.e.,
E
n
= E
m
, then the argument fails. This can be seen from the fact that both Equa-
tions (9.17) and (9.20) blow up in this case with zero in the denominator. Note
that Equation (9.19) might be completely well behaved in this case, and it is a very
common mistake to use Equation (9.19) in the case of this kind of degeneracy.
DONT DO IT! Since the second-order term in this case is innite, the entire per-
turbation expansion becomes inapplicable for the case of degeneracy. Perturbation
9.1 Derivation of Time-Independent Perturbation Theory 201
theory applied to degenerate eigenfunctions requires some further mathematical
machinery (degenerate perturbation theory) which is beyond the scope of this text.
Note, however, that there is one special case (which we will encounter frequently)
in which we can use Equations (9.19) and (9.20) with degenerate wave functions:
if
n
|H
1
|
i
= 0 whenever E
n
= E
i
, then our expressions will be well behaved.
Although the change in the energy is usually the quantity that can be most easily
measured directly, it is also possible to calculate the change in the wave function
due to the perturbation. Returning to Equation (9.4), we see that the lowest-order
change to |
n
is given by |
1
, and |
1
has already been expressed as a sum
over the unperturbed eigenfunctions in Equation (9.9) with the c
i
s that appear in
this equation given by Equation (9.17). Substituting these values for the c
i
s from
Equation (9.17) into Equation (9.9), we obtain
|
1
=

i=n

i
|H

|
n

E
n
E
i
|
i

i=n

i
|H
1
|
n

E
n
E
i
|
i

Note that we have dropped the i = n term from the sum in Equation (9.9). We
have the freedom to do this because this term is just proportional to the original
unperturbed eigenfunction |
n
. Hence, in the original expansion of the wave
function (Equation 9.4), this termcan be removed fromthe |
1
termand absorbed
into the |
n
term. Then to rst order, the new wave function in the presence of
the perturbation is
| = |
n
+

i=n

i
|H
1
|
n

E
n
E
i
|
i

Thus, the effect of the perturbation is to mix together all of the other eigenfunc-
tions in the new perturbed wave function.
Example 9.1. The Anharmonic Oscillator.
In Chapter 4, we derived the solutions for the one-dimensional harmonic oscillator
potential,
V(x) =
1
2
Kx
2
The energies are
E =
_
n +
1
2
_

h, n = 0, 1, . . .
202 Chapter 9 Time-Independent Perturbation Theory
V(x)
x
FIGURE 9.2 Solid curve: the unperturbed harmonic oscillator potential V = (1/2)Kx
2
.
Dashed curve: the perturbed potential V = (1/2)Kx
2
+x
4
.
where =

K/m is the classical oscillation frequency, and the corresponding
wave functions are

0
(s) =
1

1/4
e
s
2
/2

1
(s) =

1/4
se
s
2
/2
.
.
.
with s = [(Km)
1/4
/

h
1/2
]x.
Suppose we begin in the rst excited state
1
with energy E = (3/2)

h. We
will calculate what happens to the energy of this state if we add a small anharmonic
term to the potential (Figure 9.2):
V(x) =
1
2
Kx
2
+x
4
From Equation (9.19), the rst-order perturbation is
E
(1)
= n = 1|x
4
|n = 1
9.1 Derivation of Time-Independent Perturbation Theory 203
Expressing the perturbation in terms of s, we get
E
(1)
=
_

s=
_

2

1/4
se
s
2
/2
_
_


h
2
Km
s
4
_
_

2

1/4
se
s
2
/2
_
ds
=
2


h
2
Km
_

s=
s
6
e
s
2
ds
=
15
4


h
2
Km
With =

K/m, the total energy including the perturbation is
E =

h
_
3
2
+
15
4


h
K
2
_
Example 9.1 shows how to apply perturbation theory when the wave function
is a function of position. However, perturbation theory can also be applied to the
matrix representation of spin states.
Example 9.2. Spins in a Magnetic Field.
Suppose we begin with an electron having spin magnetic moment
s
in a strong
magnetic eld B
z
in the z direction (Figure 9.3). Recall from Chapter 8 that the
potential for an electron with spin magnetic moment
s
in a magnetic eld B is
V =
s
B
where

s
=
g
S

h
S
Hence, the Hamiltonian is
H
0
=
g
S

h
B S
=
g
S
2

B
B
and g
s
= 2 for the electron. The eigenstates of H
0
are just the spin up and spin
down states, | and | , with energies E
+
= +
B
B
z
and E

=
B
B
z
.
Suppose we are in the spin up state, | , and we add the small magnetic eld
B
x
x with B
x
B
z
(Figure 9.3). What is the change in the energy of the electron
due to this new magnetic eld? Our perturbing potential is
V
1
=
g
S
2

B
B
x

x
204 Chapter 9 Time-Independent Perturbation Theory
B
z
z
^
B
x
x
^
FIGURE 9.3 We begin with an electron in the spin up direction in the magnetic eld B
z
z,
and we add the small perturbation B
x
x.
with g
s
= 2, and the rst-order change in the energy is
E
(1)
=
n
|H
1
|
n

= |
B
B
x

x
|
=
B
B
x
( 1 0 )
_
0 1
1 0
__
1
0
_
=
B
B
x
( 1 0 )
_
0
1
_
= 0
So the rst-order perturbation, proportional to B
x
, is equal to zero. The second-
order perturbation is
E
(2)
=

i=n
|
n
|H
1
|
i
|
2
E
n
E
i
=
| |
B
B
x

x
| |
2
E
+
E

=
[
B
B
x
]
2
2
B
B
z

( 1 0 )
_
0 1
1 0
__
0
1
_

2
=

B
B
2
x
2B
z
As expected, E
(2)
is proportional to B
2
x
. In fact, this problemcan be solved exactly,
and it is instructive to see how the exact solution compares with the perturbation
theory result (see Exercise 9.1).
9.2 PERTURBATIONS TOTHE ATOMIC ENERGY LEVELS
In Chapter 6 we developed an elegant model for the hydrogen atom. The energy
levels were determined by the principal quantum number n, while the other quan-
tum numbers l, m
l
, and m
s
had no effect on the energy. As is often the case in
9.2 Perturbations to the Atomic Energy Levels 205
E
n
1
E
n
2
FIGURE9.4 Agiven spectral line corresponds to a single transition between two different
energy levels. If two supposedly degenerate levels are slightly separated in energy, a double
line will be produced.
physics, the elegant theory is extremely accurate, but it is not exact. There are
small corrections to the theory due to internal interactions in the hydrogen atom.
With perturbation theory, we now have the tools to derive these corrections.
Fine Structure
Recall that for hydrogen, the energy is given by
E
n
= (13.6 eV)
1
n
2
In hot hydrogen gas, a series of spectral lines are observed, each one correspond-
ing to a particular transition with h = E
n
1
E
n
2
. However, upon examining the
spectrum closely, it is observed that some spectral lines are not really single lines
but are closely spaced double lines. This feature is called ne structure. There is
an obvious way to get such closely spaced spectral lines: they will be observed if
the degenerate energy levels are not truly degenerate but separated in energy by a
very small amount (Figure 9.4). Recall that a given energy level E
n
corresponds
to 2n
2
different states. Apparently, some sort of interaction, which we have not yet
accounted for, splits some of these states apart in energy.
Our simple model for the hydrogen atom considered only the Coulomb interac-
tion between the proton and the electron. What we have neglected are the various
magnetic elds produced in the atom. The orbital motion of the electron sets up a
magnetic eld with magnetic moment given by

l
=
g
l
e
2m
e
L, g
l
= 1 (9.21)
while the spin of the electron produces a spin magnetic moment equal to

s
=
g
s
e
2m
e
S, g
s
2
So the orbital motion of the electron produces a magnetic eld, and the electron
itself acts like a small magnet embedded in this magnetic eld (Figure 9.5). The
electron will prefer to line up with its spin magnetic eld in the opposite direction
to the orbital magnetic eld. Hence, the spin up and spin down states of the
electron will have different energies. This is the basis of ne structure: it arises
from the spin-orbit coupling of the electron.
206 Chapter 9 Time-Independent Perturbation Theory

s
B
orbital +

FIGURE 9.5 The ne-structure splitting is produced by the interaction between the spin
magnetic eld of the electron and the magnetic eld produced by its orbital motion.
More precisely, the interaction energy between the spin magnetic moment
s
and the external magnetic eld B (produced by the orbital motion of the electron)
is
H
1
=
s
B (9.22)
To nd B imagine that we are sitting in the rest frame of the electron watching the
proton orbit around us. In the rest frame of the electron, the electric eld of the
proton,
E =
1
4
0
e
r
3
r
transforms into a magnetic eld, B = v E/c
2
. Using L = r p, we obtain,
for the magnetic eld induced by the orbital motion of the electron,
B =
_
1
4
0
_
e
m
e
c
2
r
3
L
Substituting our expressions for
s
and B into Equation (9.22), we obtain, for the
energy of the spin-orbit interaction,
H
1
=
e
2
4
0
m
2
e
c
2
r
3
S L
Unfortunately, this expression is wrong. The problem arises because we did the
calculation in the rest frame of the electron, which is an accelerating frame of
reference, so we cannot simply transform back into the rest frame of the proton
and expect to get the right answer. When this error is corrected, an additional factor
of 1/2 is obtained (this correction is called Thomas precession after L.H. Thomas,
who explained this effect in 1926). The corrected expression is
H
1
=
e
2
8
0
m
2
e
c
2
r
3
S L (9.23)
9.2 Perturbations to the Atomic Energy Levels 207

S
L

+
FIGURE 9.6 The spin-orbit interaction Hamiltonian is proportional to S L. Classically,
S L = SLcos .
From a classical point of view, this expression makes sense, since we expect the
interaction energy of the two magnetic elds to depend on the alignment of the S
and L vectors (Figure 9.6).
We can get a crude order of magnitude estimate of the size of this interaction
energy by taking S

h, L

h, and r 10
10
m. Then Equation (9.23) gives a
perturbation energy of the order of 10
4
eVcompared to a hydrogen binding energy
of 13.6 eV. It is therefore a good approximation to treat the spin-orbit interaction
as a small perturbation.
In our expression for H
1
, the interaction is proportional to S L = S
x
L
x
+
S
y
L
y
+S
z
L
z
. However, this expansion is essentially useless, since the electron
is never in a state which is a simultaneous eigenstate of all three components of
angular momentum. Instead, we use the standard procedure from Chapter 8 for
dealing with dot products of operators. We dene the total angular momentum
operator J to be
J = L +S
Then
J
2
= L
2
+S
2
+2L S
which implies
L S =
1
2
(J
2
L
2
S
2
)
208 Chapter 9 Time-Independent Perturbation Theory
and the spin-orbit perturbation becomes
H
1
=
e
2
16
0
m
2
e
c
2
r
3
(J
2
L
2
S
2
)
In Chapter 6 we wrote the hydrogen wave functions in terms of the quantum
numbers n, l, and m
l
, and in Chapter 8 we added the spin quantum number m
s
,
but now we want to express the hydrogen wave functions as eigenfunctions of J
and J
z
, i.e., in the form |n l j m
j
, where
J
2
|n l j m
j
=

h
2
j (j +1)|n l j m
j

J
z
|n l j m
j
=

hm
j
|n l j m
j

Then the perturbation to the energy levels due to spin-orbit coupling is


E
(1)
spin-orbit
= n l j m
j
|H
1
|n l j m
j

= n l j m
j
|
e
2
16
0
m
2
e
c
2
r
3
(J
2
L
2
S
2
)|n l j m
j

=

h
2
[j (j +1) l(l +1) s(s +1)]n l j m
j
|
e
2
16
0
m
2
e
c
2
r
3
|n l j m
j

What is the value of n l j m


j
|(e
2
/16
0
m
2
e
c
2
r
3
)|n l j m
j
? Note that this is
just an integral over the radial wave function, which is a function only of n and l;
hence, we can write
n l j m
j
|(e
2
/16
0
m
2
e
c
2
r
3
)|n l j m
j
= f
nl
where f
nl
is a function only of n and l. Then our expression for the change in E
due to the spin-orbit interaction (taking s = 1/2 for the electron) becomes
E
(1)
spin-orbit
=

h
2
f
nl
_
j (j +1) l(l +1)
3
4
_
(9.24)
From Chapter 9 recall that for l = 0, j has two possible values, either l 1/2
or l +1/2, while for l = 0 we have only j = 1/2. FromEquation (9.24), it is clear
that the spin-orbit coupling splits each state with l = 0 into two different states with
j = l +1/2 having the higher energy and j = l 1/2 having the lower energy.
The expression for f
nl
can be evaluated to yield a nal expression for the change
in E from spin-orbit coupling:
E
(1)
spin-orbit
= |E
n
|
2
1
2n
[j (j +1) l(l +1) 3/4]
l(l +1/2)(l +1)
(9.25)
9.2 Perturbations to the Atomic Energy Levels 209
where
=
e
2
4
0
hc
Note that is a dimensionless number with a value of roughly 1/137. Because
of its origin in this calculation, is called the ne-structure constant, although it
crops up in many other areas of physics. Recall that the hydrogen energy levels
E
n
are all negative, so we take the absolute value of E
n
in Equation (9.25) and
in Equations (9.28)(9.29) to avoid any confusion over the sign of the change in
energy.
However, this is not the full story of the ne-structure splitting. We have applied
our standard nonrelativistic treatment to the electron, and this is an excellent ap-
proximation, since the electron in the hydrogen atom is not highly relativistic (its
kinetic energy is much smaller than its rest energy; classically, this corresponds to
an electron velocity v much smaller than the speed of light). However, now that
we are working in the realm of tiny changes in the energy levels, we have to take
into account small corrections due to relativistic effects.
In relativistic classical mechanics, the total energy of a particle with rest mass
m and momentum p is
E =
_
p
2
c
2
+m
2
c
4
(9.26)
For now we are interested only in the case where the particle is only slightly
relativistic so that p mc. (We will relax this restriction in Chapter 15 when we
discuss relativistic quantummechanics in more detail.) In this limit we can expand
the square root in Equation (9.26) to obtain
E mc
2
+
p
2
2m

p
4
8m
3
c
2
+ (9.27)
In the limit of small p, each term in this expression is small compared to the
preceding one. In relativity, the rst term in Equation (9.27) is interpreted as the
rest energy of the particle, while the remainder of the expression corresponds to
the energy of motion. But what is the correct energy to use in the Schrdinger
equation? In the standard nonrelativistic Schrdinger equation, the Hamiltonian
operator corresponds to the kinetic energy plus the potential energy, and the rest
energy plays no role. Hence, in writing the Hamiltonian, we use the second term
in Equation (9.27) to give us the unperturbed Hamilton, while the third term gives
the lowest-order perturbation due to relativistic effects.
Then we have, for our perturbation,
H
1
=
p
4
8m
3
e
c
2
210 Chapter 9 Time-Independent Perturbation Theory
and the lowest-order change in the hydrogen energy levels is
E
(1)
relativistic
=
1
8m
3
e
c
2
n l j m
j
|p
4
|n l j m
j

This expression can be evaluated for the hydrogen wave functions, yielding a nal
result of
E
(1)
relativistic
= |E
n
|
2
1
n
2
_
2n
2l +1

3
4
_
(9.28)
Since E
(1)
relativistic
is a function of n and l but not a function of j, the relativistic
correction does not contribute anything to the splitting of the energy levels, which
is determined entirely by the spin-orbit interaction, but it does change the overall
dependence of the energy levels on n and l. Since l n 1, the termin brackets in
Equation (9.28) is always positive, so the relativistic correction always decreases
the energy levels.
Note that E
(1)
spin-orbit
and E
(1)
relativistic
are roughly equal in magnitude; both of them
are approximately
2
E
n
. Hence, neither contribution to the ne structure can be
neglected. We therefore add Equations (9.25) and (9.28) to get the total change in
energy due to both the spin-orbit coupling and relativistic effects:
E
(1)
ne structure
= E
(1)
spin-orbit
+E
(1)
relativistic
= |E
n
|
2
1
n
2
_
3
4

n
j +1/2
_
(9.29)
The dependence on l has cancelled out, so that the total change in energy is a
function only of j and n. The net effect of the ne structure is to split the l 1/2 and
l +1/2 states (due to the spin-orbit coupling) and to decrease the energies of both
states relative to the unperturbed hydrogen energy levels (see Exercise 9.10). The
ne-structure perturbation (both the decrease in energy relative to the unperturbed
energy levels and the splitting between the j = l +1/2 and the j = l 1/2 states)
is of order
2
E
n
10
4
E
n
, since the other factors in Equation (9.29) are all of
order unity.
We nowintroduce a standardif somewhat arcane notationtodescribe the angular
momentum states of the hydrogen atom. In this notation, the different l states of
the hydrogen atom are written with different capital letters: the l = 0 state is
called the S state, the l = 1 state is called the P state, the l = 2 state is called
the D state, the l = 3 state is called the F state, and then the sequence continues
alphabetically (G, H, I, . . .). (The origin of these abbreviations is buried in the
history of spectroscopy; there is nothing particularly logical about them.) The
standard way of writing the various j states is to indicate the value of j as a
subscript, e.g., P
1/2
is the notation for l = 1, j = 1/2. [We will see an additional
twist to this notation in Chapter 13.] The different hydrogen energy levels written
9.2 Perturbations to the Atomic Energy Levels 211
n = 1
n = 2
n = 3
S
1/2
l = 0
S
1/2
P
1/2
P
3/2
l = 1
P
1/2
P
3/2
D
5/2
D
3/2
l = 2
S
1/2
FIGURE 9.7 The energy levels of hydrogen showing the ne structure (not drawn to
scale).
S
p
S
e

+
FIGURE 9.8 The spin-spin interaction between the proton and electron produces the
hyperne splitting.
in this notation are shown in Figure 9.7. The S states (l = 0) do not split since they
correspond to a single value of j, while the l = 0 states split into the j = l +1/2
state and the j = l 1/2 state, with the former having higher energy than the
latter.
Finally, note that we have blithely applied rst-order perturbation theory
to degenerate states, ignoring the warning in the previous section. However,
it is all right to use nondegenerate perturbation theory in this case, since
n l m
l
m
s
|H
1
|n l

l
m

s
= 0 if l = l

or m
l
= m

l
or m
s
= m

s
.
There is a second internal magnetic interaction in the hydrogen atom with a
much smaller effect. The proton has a spin magnetic moment given by

p
=
g
p
e
2m
p
S (9.30)
with g
p
5.6. So the electron also feels this magnetic eld and is perturbed by
it (Figure 9.8). However, a comparison of Equation (9.30) with Equation (9.21)
shows that the ratio of the spin magnetic eld of the proton to the magnetic eld
produced by the orbital motion of the electron is roughly m
e
/m
p
6 10
4
.
Hence, we expect the splitting from the spin-spin interaction to be much smaller
than the effect of the spin-orbit interaction.
212 Chapter 9 Time-Independent Perturbation Theory
Nonetheless, this spin-spin interaction does produce a splitting in energies.
Since it is so much smaller than the ne structure, it is called hyperne splitting.
In the ground state of hydrogen, for example, the triplet (S = 1) state has a higher
energythanthe singlet (S = 0) state; the energydifference is E = 5.9 10
6
eV.
Although this energy difference is tiny, hyperne splitting has an importance out of
proportion to its magnitude. The universe contains clouds of neutral hydrogen gas;
this gas radiates by dropping from the triplet into the singlet state. The frequency
of this radiation is = E/h = 1420 MHz, corresponding to a wavelength of
= 21 cm: the famous 21-centimeter line.
The Lamb Shift
The ne-structure calculations in the previous section predict that the hydrogen
energy levels do not depend on l. Hence, two states with the same n and j quantum
numbers but different values of l should be degenerate in energy. In 1947 Willis
Lamb and his student, R.C. Retherford, showed experimentally that this was not
the case. Specically, they measured a splitting between the n = 2, S
1/2
state and
the n = 2, P
1/2
state.
This splitting, now called the Lamb shift, cannot be explained in the context
of quantum mechanics, but arises from the more esoteric area of quantum eld
theory (which was, in part, motivated by Lambs experimental result). Quantum
eld theory is beyond the scope of this book; here we will simply use one of the
predictions of the theory.
In quantum eld theory, the vacuum is no longer simply empty space; it is
literally seething with activity. Virtual particles, such as electron-positron pairs,
can pop into existence and disappear. As long as the energy of the particles E and
their lifetime t satisfy Et <

h/2, then these particle-antiparticle pairs cannot be
detected directly. [This is a rather crude explanation, which is made much more
precise within the framework of quantum eld theory.]
These particle-antiparticle pairs produce an effect called vacuum polarization.
Consider a dielectric surrounding a point positive charge. The point charge polar-
izes the dielectric, attracting negative charge inward and repelling positive charge
outward. This tends to cancel the electric eld produced by the point charge, lead-
ing to a reduced electric eld inside the dielectric (Figure 9.9).
Now consider the same positive charge in a vacuum. The production of virtual
electron-positron pairs tends to cancel the charge, just as in a physical dielectric.
However, unlike a dielectric, we can never remove the positive charge from the
vacuumpolarization to measure its true charge: the charge we measure has already
been cancelled by the effect of the vacuumpolarization. This means that the bare
charge, which cannot be measured directly, is much larger than we thought; in fact,
it is mathematically innite!
The upshot of all of this is that the electric eld of a point charge must be
modied at the origin (where the charge is innite), but everywhere else in
9.2 Perturbations to the Atomic Energy Levels 213
+
+

+
+
+ +
Dielectric Vacuum
e
+
e

e
+
e

e
+
FIGURE 9.9 In a dielectric, polarization reduces the electric eld produced by a point
charge. Vacuum polarization produces the same effect in a vacuum.
space the charge has already been cancelled by the effects of vacuum polarization,
so the electric eld is unchanged. The result for V(r), derived from quantum eld
theory, is
V(r) =
e
2
4
0
1
r

e
2
15
0

h
2
m
2
e
c
2

3
(r) (9.31)
where
3
(r) is the three-dimensional Dirac delta function discussed in Chapter 7.
The second term in Equation (9.31) is the perturbation to the Hamiltonian, so the
rst-order shift in the energy is
E
(1)
= n l m
l
m
s
|H
1
|n l m
l
m
s

=
_
d
3
r

nlm
l
(r)
_

e
2
15
0

h
2
m
2
e
c
2

3
(r)
_

nlm
l
(r)
=
e
2
15
0

h
2
m
2
e
c
2
|
nlm
l
(0)|
2
Recall that the hydrogen wave functions are all identically zero at the origin, except
for the l = 0 states. Thus, the effect of the Lamb shift is to reduce the energy of the
l = 0 states relative to the corresponding l = 0 states. The effect is smaller than
the ne-structure splitting, e.g., for the n = 2 states, the splitting between the l = 0
and l = 1 state is about 10
7
eV. As bizarre as all of this sounds, it is important to
remember that it is based on solid experimental evidence.
214 Chapter 9 Time-Independent Perturbation Theory
9.3 THE ATOM IN EXTERNAL ELECTRIC OR MAGNETIC FIELDS
In the previous section, we discussed perturbations which are intrinsic to the atom.
We will now examine what happens when the atom is placed in an external elec-
tromagnetic eld. Since the atom consists of charged particles, and the electons
produce both a spin and orbital magnetic moment, any external electric or magnetic
eld will perturb the energy levels of the atom. The effect produced by an external
electric eld is called the Stark effect, while the effect of an external magnetic eld
is the Zeeman effect.
The Atom in an Electric Field: The Stark Effect
We will rst examine the effect of a uniform electric eld with magnitude E on
the ground state of hydrogen. Recall that the ground-state wave function is

100
=
1

a
3
0
e
r/a
0
where the 100 subscript denotes the n l m
l
quantum numbers. We can ignore the
spin state of the electron, since the spin interacts only with magnetic elds through
the electrons spin magnetic moment. (Of course, we will have to consider spin in
the next section when we discuss external magnetic elds.) We take the electric
eld to be uniform, static, and pointing in the z direction.
Since the groundstate of hydrogenis nondegenerate, we canuse the perturbation
theory expressions from Section 9.1. Classically, the potential energy of a charge
e in an electric eld E is V = eEz, so the perturbation H
1
produced by the electric
eld is
H
1
= eEz
and the rst-order change in the energy of the hydrogen atom is, from Equation
(9.19),
E
(1)
=
100
|eEz|
100
(9.32)
Taking z = r cos and writing Equation (9.32) in spherical coordinates, we get
E
(1)
=
_
_
1
_
a
3
0
e
r/a
0
_
(eEr cos )
_
1
_
a
3
0
e
r/a
0
_
r
2
dr sin d d
But the integral over vanishes:
_

=0
cos sin d =
1
2
sin
2
|

0
= 0
so E
(1)
= 0.
9.3 The Atom in External Electric or Magnetic Fields 215
Since the rst-order perturbation vanishes, we must use second-order perturba-
tion theory to calculate the change in the energy due to the external electric eld.
Equation (9.20) gives
E
(2)
=

n,l,m
l
|
100
|eEz|
nlm
l
|
2
E
1
E
n
(9.33)
where E
n
= 13.6 eV/n
2
. Recall fromChapter 6 that every hydrogen wave function
can be written as the product of a radial wave function R
nl
(r) and the spherical
harmonic Y
m
l
(, ). Then the inner product which appears in Equation (9.33) can
be written in the form

100
|eEz|
nlm
l
= eE
_
R

10
(r)Y
0
0
(, ) r cos R
nl
(r)Y
m
l
(, ) d
3
r (9.34)
But nowrecall that Y
0
0
= 1/

4, andY
0
1
=

3/4 cos , soY
0
0
cos = (1/

3)Y
0
1
.
This allows us to write Equation (9.34) as

100
|eEz|
nlm
l
=
eE

3
_
R

10
(r)R
nl
(r)r
3
dr
_
Y
0
1
(, )Y
m
l
(, ) sin d d
But the Y
m
l
s are orthonormal, so
_
Y
0
1
(, )Y
m
l
(, ) sin d d = 1 (l = 1, m = 0)
= 0 (l = 1 or m = 0)
Hence, in the sum in Equation (9.33), only the l = 1, m = 0 terms are nonzero
giving
E
(2)
=

n
|(eE/

3)
_
R

10
(r)R
n1
(r)r
3
dr|
2
E
1
E
n
(9.35)
The integral under the sum in Equation (9.35) can be evaluated exactly for all
values of n, and the terms in the series decrease rapidly with n:
E
(2)
= (4
0
)a
3
0
E
2
(1.48 +0.20 +0.066 + ) =
9
4
(4
0
)a
3
0
E
2
The change in energy is negative, since the hydrogen atom becomes polarized and
aligns itself so as to partially cancel the external electric eld (Figure 9.10; see
also Exercise 9.2). Since the change in energy is proportional to the square of the
applied electric eld, this effect is called the quadratic Stark effect.
Our use of nondegenerate perturbation theory breaks down for the excited states
of hydrogen, since these states are degenerate. Using degenerate perturbation the-
ory, it is possible to show that the change in energy for these excited states is
proportional to E rather than E
2
. Hence, the change in energy when an electric
eld is applied to the excited states of hydrogen is called the linear Stark effect.
216 Chapter 9 Time-Independent Perturbation Theory
atomic
external

+
FIGURE 9.10 A classical picture of the quadratic Stark effect: the hydrogen atom is
polarized by the external electric eld, and the eld produced by the polarized atom is in
the opposite direction to the external eld.
The Atom in a Magnetic Field: The Zeeman Effect
Now consider what happens when we apply an external magnetic eld B to the
hydrogen atom. Assume that the magnetic eld has magnitude B and is pointing
in the z direction so that
B = B z
9.3 The Atom in External Electric or Magnetic Fields 217
S
S
J
L

FIGURE 9.11 The magnetic moment of the hydrogen atom is proportional to L +2S,
while the total angular momentum J is proportional to L +S, so and J are not, in general,
parallel.
The potential energy of a magnetic dipole in a magnetic eld B is just V =
B, so the perturbation produced by the magnetic eld is
H
1
= B (9.36)
In Equation (9.36), there are two contributions to the atomic magnetic moment: the
contribution from the orbital magnetic moment
l
and the contribution from the
spin magnetic moment of the electron
s
. (In principle, we should also include the
spin magnetic moment of the proton, but this is much smaller and can be ignored.)
Hence, the total magnetic moment is
=
l
+
s
=
g
l

h
L
g
s

h
S
Recall that g
l
= 1 and g
s
2, so the expression for becomes
=

h
[L +2S] (9.37)
Note that the magnetic moment of the atom is not proportional to the total angular
momentum operator J, which is L +S. In classical terms, the angular momentum
vector J and magnetic moment vector are not parallel (Figure 9.11). This has
important consequences for the Zeeman effect. (You are already familiar with a
much larger classical systemin which the angular momentumand magnetic dipole
are not parallel: the earth!)
218 Chapter 9 Time-Independent Perturbation Theory
We can use Equation (9.37) to rewrite the perturbation in Equation (9.36) as
H
1
= B

h
[L
z
+2S
z
]
Applying this perturbation to the hydrogen state |n l m
l
m
s
gives the rst-order
change in energy,
E
(1)
= n l m
l
m
s
|(B
B
/

h)(L
z
+2S
z
)|n l m
l
m
s

= B
B
(m
l
+2m
s
) (9.38)
The problem with this result is that it requires the atom to be in a state of denite
m
l
and m
s
(or, equivalently, an eigenstate of S
z
and L
z
). However, we saw in our
discussion of ne structure in Section 9.2 that the spin-orbit coupling drives the
atom into an eigenstate of J
2
, which does not commute with S
z
and L
z
. Hence,
the atom is in a state of denite j and m
j
rather than m
l
and m
s
, so our argument
would appear to be invalid.
To clarify this situation, we can write the full Hamiltonian as
H = H
0
+
e
2
8
0
m
2
e
c
2
r
3
S L +B

h
[L
z
+2S
z
] (9.39)
where the second term is the perturbation due to the spin-orbit interaction (given
in Equation 9.23), and the third termis the perturbation fromthe external magnetic
eld.
Nowconsider two possible cases: for very strong magnetic elds (B 1T), the
third term in Equation (9.39) dominates the second term, while for weak magnetic
elds (B 1 T), the second term dominates the third. Consider the case of strong
magnetic elds rst. For this case we simply ignore the effect of the spin-orbit
coupling; the strong magnetic eld overwhelms the spin-orbit coupling and drives
the atom back into a state of denite m
l
and m
s
. Therefore, for the strong magnetic
eld case, the expression we derived for E
(1)
in Equation (9.38) is correct:
E
(1)
= B
B
(m
l
+2m
s
)
This regime of the Zeeman effect is called the strong-eld Zeeman effect or the
Paschen-Back effect. An illustration of this perturbation in the energy levels is
shown in Figure 9.12 for the case l = 1.
Now consider the opposite regime in which spin-orbit coupling dominates the
effect of the external magnetic eld. In this case the atom is in a state of denite
j and m
j
rather than m
l
and m
s
, and the perturbation must be written as
E
(1)
= n l j m
j
|(B
B
/

h)(L
z
+2S
z
)|n l j m
j

9.3 The Atom in External Electric or Magnetic Fields 219


l = 1
m
l
= 1, 0, 1
m
s
= 1/2, +1/2
m
l
+ 2m
s
2
1
0
1
2
m
l
= 1, m
s
= 1/2
m
l
= 0, m
s
= 1/2
m
l
= 1, m
s
= 1/2
m
l
= 1, m
s
= 1/2
m
l
= 0, m
s
= 1/2
m
l
= 1, m
s
= 1/2
FIGURE 9.12 The strong-eld Zeeman effect for the energy levels of an l = 1 state in
hydrogen.
This can be partially simplied by using the fact that J
z
= L
z
+S
z
:
E
(1)
= n l j m
j
|(B
B
/

h)(J
z
+S
z
)|n l j m
j

= B
B
m
j
+
B
B

h
n l j m
j
|S
z
|n l j m
j
(9.40)
In order to further simplify this expression, the state |n l j m
j
must be written as a
linear combination of the |n l m
l
m
s
states. FromChapter 8, we knowthat s = 1/2
and a given value of l can couple to give either j = l +1/2 or j = l 1/2, while
m
j
= m
l
+m
s
. The actual linear combination is
|j = l +1/2, m
j
=
_
l +1/2 +m
j
2l +1
_
1/2
|m
l
= m
j
1/2, m
s
= 1/2
+
_
l +1/2 m
j
2l +1
_
1/2
|m
l
= m
j
+1/2, m
s
= 1/2
|j = l 1/2, m
j
=
_
l +1/2 m
j
2l +1
_
1/2
|m
l
= m
j
1/2, m
s
= 1/2

_
l +1/2 +m
j
2l +1
_
1/2
|m
l
= m
j
+1/2, m
s
= 1/2
We can use these equations to solve for n l j m
j
|S
z
|n l j m
j
. For j = l +1/2,
220 Chapter 9 Time-Independent Perturbation Theory
we get
n l j m
j
|S
z
|n l j m
j
=
_
_
l +1/2 +m
j
2l +1
_
1/2
m
l
= m
j
1/2, m
s
= 1/2|
+
_
l +1/2 m
j
2l +1
_
1/2
m
l
= m
j
+1/2, m
s
= 1/2|
_
S
z
_
_
l +1/2 +m
j
2l +1
_
1/2
|m
l
= m
j
1/2, m
s
= 1/2
+
_
l +1/2 m
j
2l +1
_
1/2
|m
l
= m
j
+1/2, m
s
= 1/2
_
=

h
2
_
l +1/2 +m
j
2l +1
_


h
2
_
l +1/2 m
j
2l +1
_
=
m
j
h
2l +1
Similarly, for j = l 1/2, we obtain
n l j m
j
|S
z
|n l j m
j
=
m
j
h
2l +1
Combining the results for j = l +1/2 and j = l 1/2, we get
n l j m
j
|S
z
|n l j m
j
=
m
j
h
2l +1
2(j l)
and substituting this result into Equation (9.40) yields
E
(1)
= B
B
m
j
_
2j +1
2l +1
_
In analogy with the g
s
factor for the electron spin and g
l
for the orbital angular
momentum, we can write
g =
2j +1
2l +1
where this g is called the Land g factor. In terms of the Land g factor, the energy
shift becomes
E
(1)
= gB
B
m
j
In contrast to g
s
and g
l
which are constant, the Land g factor is not constant,
but rather is a function of j and l. The reason for this is the fact, already alluded
to, that is not parallel to J, since the operator which determines is L +2S
Exercises 221
m
j
+3/2
+1/2
+1/2
1/2
1/2
3/2
P
3/2
(l = 1, j = 3/2)
P
1/2
(l = 1, j = 1/2)
g = 4/3
g = 2/3
FIGURE 9.13 The splitting of the P
3/2
and P
1/2
states in the weak-eld Zeeman effect.
while J = L +S. Hence, the ratio between and J can depend on the relative
orientation of L and S (Figure 9.11), so is not a xed multiple of m
j
. The effect
of the weak-eld Zeeman effect is to split the energies of the individual m
j
levels,
with a magnitude which depends on both the magnitude of the magnetic eld and
the value of the Land g factor (Figure 9.13).
To summarize, for weak magnetic elds, the hydrogen atom can be taken to
be in a state of denite j and m
j
, and the magnetic eld separates the energies of
the individual m
j
states. As the magnetic eld is increased, it eventually becomes
stronger than the internal magnetic elds of the atom. In this limit, the magnetic
elddrives the hydrogenatomintoa state of denite m
l
andm
s
, andthe perturbation
in energy is just proportional to m
l
+2m
s
.
EXERCISES
9.1 (a) In Example 9.2, the energy of the system can be calculated exactly. Take B =
B
x
x +B
z
z, and calculate the exact energies. [Hint: Feel free to use a different
coordinate system; the energy levels cannot depend on the choice of the coordinate
system].
(b) Take the answer in part (a) and expand it out in powers of B
x
, remembering
that B
x
B
z
. Show that the terms proportional to B
x
and B
2
x
correspond to the
answers derived in Example 9.2.
222 Chapter 9 Time-Independent Perturbation Theory
9.2 A particle is in a potential V
0
in its ground state |
0
. A small perturbation H
1
is
applied to the particle. Suppose that the rst order perturbation to the energy is zero:
E
(1)
=
0
|H
1
|
0
= 0. Show that the lowest-order effect of H
1
is to decrease the
energy of the ground state.
9.3 Aparticle of mass mis conned to move in a one-dimensional square well with innite
potential barriers at x = 0 and x = a, with V = 0 for 0 x a. The particle is in the
ground state. A perturbation H
1
= (x a/2) is added, where is a small constant.
(a) What units does have?
(b) Calculate the rst-order perturbation E
(1)
due to H
1
.
(c) Calculate the second-order perturbation E
(2)
. The answer may be expressed as
an innite series.
9.4 A particle of mass m is conned to move in a narrow, straight tube of length a which
is sealed at both ends with V = 0 inside the tube. Treat the tube as a one-dimensional
innite square well. The tube is placed at an angle relative to the surface of the
earth. The particle experiences the usual gravitational potential V = mgh. Calculate
the lowest-order change in the energy of the ground state due to the gravitational
potential.
9.5 A particle of mass m is in the ground state in the harmonic oscillator potential
V(x) =
1
2
Kx
2
A small perturbation x
6
is added to this potential.
(a) What are the units of ?
(b) How small must be in order for perturbation theory to be valid?
(c) Calculate the rst-order change in the energy of the particle.
9.6 In the hydrogen atom, the proton is not really a point charge but has a nite size.
Assume that the proton behaves as a uniformly-charged sphere of radius R = 10
15
m.
Calculate the shift this produces in the ground-state energy of hydrogen.
9.7 The photon is normally assumed to have zero rest mass. If the photon had a small mass,
this would alter the potential energy which the electron experiences in the electric eld
of the proton. Instead of
V(r) =
e
2
4
0
1
r
(9.41)
we would have
V(r) =
e
2
4
0
e
r/r
0
r
(9.42)
where r
0
is a constant with units of length. Assume r
0
is large compared to the size
of the hydrogen atom, so the potential energy given in Equation (9.42) differs only
slightly from the standard one given by Equation (9.41) in the vicinity of the electron.
Calculate the change in the ground state energy of hydrogen if the correct potential is
given by Equation (9.42) instead of Equation (9.41).
Exercises 223
9.8 Suppose that that the proton had spin 0 instead of spin 1/2.
(a) Howwould this alter the ne structure of the energy levels of the hydrogen atom?
(b) How would this alter the hyperne structure of the energy levels of the hydrogen
atom?
9.9 We have seen that the spin-orbit interaction splits the l = 0 states in the hydrogen
atom into j = l +1/2 states (with slightly higher energy) and j = l 1/2 states
(with slightly lower energy). Suppose that the electron had spin 1. Howmany different
energy levels would the spin-orbit interaction produce, and what would their relative
energies be? Be sure to consider how the answer would depend on the value of l.
9.10 Equation (9.29) gives the ne-structure energy shift.
(a) Show that the j = l +1/2 state has a higher energy than the j = l 1/2 state.
(b) Show that the change in energy, E
(1)
f ine st ruct ure
, is always negative.
9.11 An electron is in the ground state in a three-dimensional rectangular box given by
0 x a, 0 y b, and 0 z c, where V = 0 inside the box, and there are
innite potential barriers at all of the walls. A homogeneous, static electric eld with
magnitude E is applied in the x direction. What is the lowest-order change in the
energy of the electron?
9.12 A hydrogen atom in its ground state is placed in a homogeneous, static electric eld
with magnitude E in the x direction.
(a) Show that the rst-order perturbation E
(1)
is 0.
(b) Show that the second-order perturbation E
(2)
is the same as if the eld was
pointing in the z direction. [This is obvious from symmetry, but calculate E
(2)
using perturbation theory and show it explicitly.]
9.13 A hydrogen atom is in its ground state. A proton is xed in space a distance R from
the nucleus of the hydrogen atom, where R a
0
. Calculate the perturbation to the
energy of the hydrogen atom due to the electric eld of this proton.
9.14 The electron in a hydrogen atom is in a D state. A homogenous, static magnetic eld
is applied in the z direction.
(a) Draw a diagram showing the splitting of the energy levels in the weak-eld limit.
Calculate the value of g for each energy level.
(b) Drawa diagramshowing the splitting of the energy levels in the strong-eld limit.
9.15 (a) A particle is in a state | which is an eigenfunction of the Hamiltonian H
0
with
energy E. Aperturbation H
1
is applied such that H
1
| = 0. Showthat the energy
of the system is completely unchanged by this perturbation.
(b) In the ground state of the helium atom, both electrons are in the l = 0 state, and
the spin wave function for the two electrons is the singlet spin state (s = 0 and
m
s
= 0). [This is a consequence of the Pauli exclusion principle, which will be
discussed in Chapter 13.] A homogeneous, static magnetic eld is applied in the
z direction. Show that the energy of the ground state of helium is completely
unaffected by this magnetic eld. [Ignore the magnetic moment of the nucleus.]
What is the physical reason for this?

Potrebbero piacerti anche