10

Perturbation theory
Quantum mechanical systems may be exposed to perturbations including external electric fields,
magnetic fields, or electromagnetic radiation. Due to such perturbations, the quantum system
considered here is stimulated and, as a consequence, changes its state. This change of state may
include changes in the shape of wave functions, state energies, and occupation probability of
states. This is what perturbation theory is all about. Perturbation theory is one of the most
important methods for obtaining approximate solutions to Schrdingers equation.
10.1 First-order time-independent perturbation theory
This section covers first-order perturbation calculation of a stationary, non-degenerate quantum
state. Suppose a quantum mechanical system whose eigenstate energies and wave functions are
known. Suppose that the unperturbed system is described by the hamiltonian operator H0, the
eigenstate energies En0, and the wave functions n0. Then the Schrdinger equation of the
unperturbed system is given by

H 0 0n

En0 0n .

(10.1)

Here, the superscript 0 is used for energies, wave functions and the hamiltonian operator of the
unperturbed system. If the system is subjected to a small perturbation, then perturbation theory
allows one to determine the modifications of the eigenstate energies, wave functions, and
occupation probabilities. It may seem that these are very special circumstances; however, it will
become clear, that perturbation theory is of great practical importance.
The hamiltonian operator of a perturbed system is given by

H0 + H

(10.2)

where H0 is the hamiltonian operator of the unperturbed system and H is called the
perturbation term in the hamiltonian. The parameter allows us to turn the perturbation on
( = 1) and off ( = 0). The parameter further indicates the smallness of the perturbation. That
is, the system described by the hamiltonian H has experienced only a small perturbation when
compared to the unperturbed system. The parameter can have the value of unity, ( = 1),
without loss of general validity of the perturbation theory. The Schrdinger equation of the
perturbed system is given by

H n

(H 0

+ H n

En n .

(10.3)

It is evident that the perturbed system merges with the unperturbed system if approaches zero,
i. e.

lim En

En0

(10.4)

and
E. F. Schubert

85

lim n

= 0n .

(10.5)

To obtain a solution of the perturbed problem, an expansion of En and n in a power series in

En

En0 + En + 2 En + L

(10.6)

= 0n + n + 2 n + L

(10.7)

where En = dEn / d and n = dn / d. The first three terms of this power series are illustrated
in Fig. 10.1, where the unperturbed values of En and n are displayed together with their firstorder correction term (En and n) and their second-order correction term (2En and 2n).
It is the goal of first-order perturbation theory to find the values of En and n. Correspondingly,
it is the aim of second-order perturbation theory to find the values of En and n.
Substitution of Eqs. (10.6) and (10.7) into the perturbed Schrdinger equation (Eq. 10.3)
yields

(
+ 1 (H 0 n
+ 2 (H 0 n

0 H 0 0n En0 0n

+ H 0n En0 n En 0n

+ H n En0 n En n En 0n

(10.8)

+ 3 ( L )
+ 4 ( L )

+ L = 0.

E. F. Schubert

86

If the perturbation is neglected ( = 0), one obtains the original eigenvalues En0 and
eigenfunctions n0. We next consider the case 0. The sum in Eq. (10.8) equals zero, only if
each summand is zero, that is
0th order terms:

H 0 0n

1st order terms:

H 0 n + H 0n

En0 n + En 0n

(10.10)

2nd order terms:

H 0 n + H n

En0 n + En n + En 0n

(10.11)

En0 0n

(10.9)

The first of these three equation is the unperturbed Schrdinger equation. The second equation
contains only first-order terms and it will be used to derive first-order perturbation results. The
third equation contains only first-order and second-order terms and it will be used to derive
second-order perturbation results. Note that in the above three equations, H0, n0, En0, and H are
known.
It is the purpose of first-order perturbation theory to find solutions for En and n (En and
n will be determined by second-order perturbation theory which is discussed in the subsequent
section). Because n is an unknown wave function, we will try to express n as a series of a
complete set of orthogonal eigenfunctions

a j 0j

(10.12)

The wave functions j0 represent the complete orthonormal set of wave functions of the system.
Nearly any wave function can be synthesized from this orthonormal set. The particular wave
function n is one specific wave function of the complete set. In order to determine the perturbed
wave function n, the coefficients aj must be determined. Substitution of Eq. (10.12) into
Eq. (10.10) yields
H0

a j 0j

+ H 0n

E n0

a j 0j

+ E n 0n .

(10.13)

Using
H0

a j 0j

a j E 0j 0j

(10.14)

one obtains

a j E 0j 0j

+ H 0n

En0

a j 0j

+ En 0n .

(10.15)

Consider next a wave function m0 which is the mth wave function of the orthonormal set of
wave functions j0 given in Eq. (10.12). Pre-multiplication of Eq. (10.15) with m0*, and
integration over position space yields
0
am Em
+ 0m H 0n

= am En0 + En mn .

(10.16)

Here, we have used the orthogonality of the set j0, that is m0j0 = 0 for m j and

E. F. Schubert

87

m0j0 = 1 for m = j. This can be expressed by the Kronecker delta m0j0 = mj or

En

m = n:

0n H 0n

(10.17)

which is the first-order correction term to the energy. Hence, the energy of the nth state of a
system, subjected to the perturbation hamiltonian H , calculated by first-order perturbation
theory, is given by

En

En0 + 0n H 0n

(10.18)

Furthermore, one obtains for

m n:

am

0m H 0n

(10.19)

0
En0 Em

This equation can be used to calculate all am , except the value of am=n . The value of am=n can be
calculated by requiring that the first-order corrected wave function is normalized, i. e.
0n + n | 0n + n = 1 . This condition yields for
*

m = n:

n + am 0m

*
*
n + am 0m dx = 1 + am + am
+ 2 am am
= 1

m
m

(10.20)

A simple solution of this equation is am = am* = an = an* = 0. Hence, the wave function of the
nth state of a system, subjected to the perturbation hamiltonian H , calculated by first-order
perturbation theory, is given by

0n

mn

0m H 0n
En0

0
Em

0m

(10.21)

The sum in this equation is carried out for all values of m except the value m = n. This equation
shows, that the wave functions of all other (unperturbed) states have to be known, to calculate
the perturbed wave function of the nth state. The influence of other wave functions decreases, as
the energy-separation increases, since En0 Em0 is in the denominator of the expression.
10.2 Second-order time-independent perturbation theory
The energy and the wave function of a perturbed state can be expressed in terms of the expansion
of Eqs. (10.6) and (10.7). It is the aim of the second-order perturbation calculation to find
formulas for En and n. In analogy to the first-order calculation, n is expressed in terms of
the complete set of orthonormal wave functions of the unperturbed system, i. e.

E. F. Schubert

88

b j 0j

(10.22)

Inserting Eq. (10.22) and Eq. (10.12) into Eq. (10.11) yields

j b j E 0j 0j

+ H j a j 0j

j b j En0 0j

a j En j + En 0n . (10.23)

Similar to the previous section, we consider one specific wave function m0 of the complete
orthonormal set of wave functions of the unperturbed system. Pre-multiplication with m0*,
integration over all configuration space, and recalling that m0 j0 = mj yields
0
bm Em
+

a j 0m H 0j

= bm En0 + En am + En nm .

(10.24)

In this equation, it is either m = n or m n. With m = n, one obtains the second-order correction

En

m = n:

jn a j

a j 0n H 0j

En an

0n H 0j

(10.25)

+ an 0n H 0n

En an .

(10.26)

In Eq. (10.25), the sum is carried out for all values of j, whereas in Eq. (10.26), the sum is
carried out for all values of j except the value j = n. Using the result of first-order perturbation
theory for En (see Eq. 10.17), then the last two terms of Eq. (10.26) cancel. Using the first-order
perturbation result for aj (Eq. 10.19), one obtains

En

0n H 0j

(10.27)

En0 E 0j

jn

which is the second-order correction to the energy of the nth state of a system subjected to the
perturbation hamiltonian H. Note that the second-order term given in the above equation
increases drastically if two energy-levels are closely spaced, i. e. for small En0 Ej0. The secondorder correction term becomes large for a small separation of two energy levels. It is therefore
frequently said that energy levels repel each other. States energetically distant from the state of
interest may be neglected in practical calculations.
For m n in Eq. (10.24), using that an = 0, and with Eqs. (10.17) and (10.19), one obtains

m n:

bm

jn

0j H 0n

)(

0m H 0j

0
En0 E 0j En0 Em

0n H 0n

0m H 0n

0 2
En0 Em

. (10.28)

Finally bm = bn (m = n) must be determined, which can again be achieved with the normalization
condition.

E. F. Schubert

89

 0n + a j 0j + 2 b j 0j 0n + a j 0j + 2 b j 0j

m = n:

=1.

(10.29)

Evaluation of this integral yields the value of bm=n = bn

bn

1
2

aj

0j H 0n

2 jn

2
En0 E 0j

(10.30)

Using these values of bm, the second-order correction term of the wave function (see Eq. 10.22)
is given by

0j H 0n 0m H 0j

n =

0
m n j n E n0 E 0j E n0 E m

)(

0m H 0n
0n H 0n 0m H 0n

0
0n
m

0 2
0 2

E n0 E m
2 E n0 Em

(10.31)
The second order correction terms for the energy and the wave function have now been obtained.
For convenience, the results of first-order and of second-order perturbation theory for the energy
and wave function of the nth state are summarized:
First- and second-order correction to the energy of the nth state:

En

En0 + 0n H 0n

+ 2

0n H 0j

(10.32)

En0 E 0j

jn

First- and second-order correction to the wave function of the nth state:
n

0n +

mn

0m H 0n
0
En0 Em

0j H 0n 0m H 0j

0
m n j n En0 E 0j En0 Em

)(

0m +

0m H 0n
0
m
2
0
0
En Em

0n H 0n

0n
0 2

2 En0 Em

0m H 0n

(10.33)
10.3 Example for first-order perturbation calculation
A simple example for a first-order perturbation calculation is illustrated in Fig. 10.2. The
unperturbed potential is a square-shaped quantum well. The lowest eigenstate energy En=00 and
wave function n=00 are shown as well. The system is now subjected to a potential perturbation,
and it is the purpose of this example to calculate the perturbed ground-state energy

E. F. Schubert

90

En=0 = En=00 + En. The perturbation used in this example is a potential energy perturbation
given by
U p ( x) = 0

(x < b)

(10.34)

U p ( x) = U 0

(b x c)

(10.35)

U p ( x) = 0

(x > c).

(10.36)

Since the perturbation is purely a potential energy perturbation, the hamiltonian operator is given
by
H

= H 0 + H = H 0 + U p ( x) .

(10.37)

The solution of the Schrdinger equation for the unperturbed problem (which is the finite squarewell potential) is given by
I ( x) =

A1 e x

(10.38)

II ( x) =

e a
cos kx
A1
cos ka

(10.39)

III ( x) =

A1 e x

(10.40)

where A1 is a normalization constant and

E. F. Schubert

91

1
h

2 m U 0 E00

1
h

2 m E00

(10.41)

(10.42)

where En=00 = E00 is the lowest eigenstate energy of the unperturbed system.
The change of the lowest eigenstate energy is now calculated by using to Eq. (10.18).
Furthermore, we use = 1 and obtain
E0

E00 +

0 U p ( x) 0

dx .

(10.43)

Since Up(x) = 0 outside the interval between b and c, the integration can be limited to this
interval
E0

E00 +

2 2
b ( U 0 ) A1 e
c

dx

E00

U 0 A12 2 b
e
e 2 c .
2

(10.44)

The equation reveals that the perturbed energy of the state, E0, is lower than the unperturbed
value of the energy, E00. Thus the eigenstate energy decreases upon perturbation.

Exercise: Quantum well electroabsorption. The band diagram of a quantum well structure is
shown in Fig. 10.3. Consider an electron with effective mass m* in a symmetric quantum well
with thickness LQW, clad by infinitely high barriers. Assume that the center of the quantum well
is located at the origin of the coordinate system, i. e. at x = 0. The quantum well is now subjected
to a constant electric field E, so that the potential energy created by the electric field is given by
U(x) = e E x. Upon application of an electric field, the quantum well potential is perturbed. The
perturbation hamiltonian for a constant electric field is given by

H = e E x .

(10.45)

(a) What is the energy of the lowest eigenstate in the well for E = 0?
(b) Calculate the change in lowest eigenstate energy due to the electric field using first-order
perturbation theory and show that first-order perturbation theory yields no change in the
energy of the ground state.
(c) Calculate the change in lowest eigenstate energy due to the electric field using second-order
perturbation theory. Use reasonable approximations and explain them. Show that second
order perturbation theory yields a decrease in the ground-state energy.
(d) Show that the energy between the highest valence band state and the lowest conduction band
state (i. e. the absorption edge) decreases upon application of an electric field.
The concept of shifting the absorption edge by an electric field is used in electroabsorption
quantum-well modulators. In these modulators, quantum wells are placed in the depletion region
of a reverse-biased pn junction. Such electroabsorption modulators can be modulated at a much
higher bit rate than would be possible by direct current modulation of a semiconductor laser. The
speed advantage is due to the much smaller depletion capacitance of the reverse-biased

E. F. Schubert

92

modulator junction as compared to the diffusion capacitance of the forward-biased laser

Exercise: Coupled quantum wells. If the states of two identical quantum wells are perturbed by
decreasing the distance between the quantum wells (QWs), the state will split into two states, a
bonding state and an anti-bonding state. The energy split E increases as the separation of the
QWs decreases. Coupled QWs and the schematic dependence of E on the distance between the
two QWs are shown in Fig. 10.4. The energy splitting is given by:

(10.46)

Show the dependence of the ground-state energy levels of two identical atoms as a function of
the distance between the two atoms. Explain the difference between the coupling of two QWs
and the coupling of two atoms.

E. F. Schubert

93