Y. D. Chong Ch.

1: Scattering Theory | Graduate Quantum Mechanics

Chapter 1: Scattering Theory

1.1. SCATTERING EXPERIMENTS ON QUANTUM PARTICLES

Quantum particles exhibit a feature known as wave-particle duality, which can be

summarized in the quantum double-slit thought experiment. As illustrated below, a
source emits electrons of energy E toward a screen with a pair of slits. A movable detector,
placed on the far side, measures the arrival rates at different positions.

The experiment reveals the following: (i) the electrons arrive in discrete units, like classical
particles; (ii) the statistical distribution of electron arrivals forms an interference pattern,
like a diffracted wave. In the non-relativistic limit, the inferred wavelength λ is related to
the electron energy E by
2π ℏ2 k 2
λ= , E= , (1.1)
k 2m
where ℏ = h/2π is Dirac’s constant, and m is the electron mass. (Thus, if E is known, we
can deduce the spacing of the slits from the diffraction pattern.)
In this chapter, we will study a generalization of the double-slit experiment called a scat-
tering experiment. The idea is to shoot quantum particles at a target, called a scatterer,
and measure the distribution of scattered particles. Just as the double-slit interference pat-
tern can be used to deduce the slit spacing, a scattering experiment can be used to deduce
various facts about the scatterer. Scattering experiments constitute a large proportion of
the methods used to probe the quantum world, from electron- and photon-based microscopy
to accelerator experiments in high-energy physics.
We will focus on the non-relativistic scattering experiment depicted below:

An unbounded d-dimensional space is described by coordinates r, with a scatterer near

r = 0. An incoming quantum particle, with energy E, is governed by the Hamiltonian

Ĥ = Ĥ0 + V (r̂), (1.2)

where
p̂2
Ĥ0 = (1.3)
2m

1
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

describes the particle’s kinetic energy, m is the particle’s mass, r̂ and p̂ are the position and
momentum operators, and V is a scattering potential describing how the scatterer acts
upon the particle. We assume this potential vanishes far from the scatterer:
|r|→∞
V (r) −→ 0. (1.4)

How is the particle scattered by the potential? In order to study this, we formulate the
above description as the following mathematical problem:
1. A particle state |ψ⟩ obeys the time-independent Schrödinger equation
Ĥ|ψ⟩ = E|ψ⟩, (1.5)
where E is the incoming particle energy.
2. The state |ψ⟩ can be decomposed into two pieces,
|ψ⟩ = |ψi ⟩ + |ψs ⟩, (1.6)
where |ψi ⟩ is called the incident state and |ψs ⟩ is called the scattered state.
3. The incident state is described by a plane wave, which is a simultaneous eigenstate of
Ĥ0 (with energy E) and p̂ (with momentum pi ):
(
Ĥ0 |ψi ⟩ = E|ψi ⟩, |pi |2
where E = . (1.7)
p̂|ψi ⟩ = pi |ψi ⟩, 2m

4. The scattered state is an “outgoing” state. What this means will be explained later.
We can interpret the above statements as follows. Statement 1 says the scattering is elastic:
the scatterer is described by a potential V (r), so its interaction with the particle is conserva-
tive and E is fixed. Statement 2 says that the wavefunction ψ(r) = ⟨r|ψ⟩ is a superposition
of an incoming wave and a scattered wave. Statement 3 defines the incoming wave as a
plane wave with wavelength determined by E. Statement 4 says the scattered wave moves
outward to infinity, away from the scatterer.
Note that Statements 1–4 do not define an eigenproblem! Usually, we deal with the time-
independent Schrödinger equation by finding eigenvalues and eigenstates. But here we are
given |ψi ⟩, E, and V (r), and want |ψs ⟩. In particular, E is an input to the calculation (a
continuously variable parameter), not an output (an eigenvalue).

1.2. RECAP: POSITION AND MOMENTUM STATES

Before proceeding, let us review the properties of a quantum particle in free space. In a
d-dimensional space, a coordinate vector r is a real vector with d components. We suppose
there is a uncountably infinite set of position eigenstates, {|r⟩}, where each |r⟩ describes a
particle with a definite position r. The position eigenstates are assumed to span the state
space, such that the identity operator can be resolved as
Z
ˆ
I = dd r |r⟩⟨r|, (1.8)

2
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

with the integral is taken over all allowed r (e.g., r ∈ Rd ). It follows that
⟨r|r′ ⟩ = δ d (r − r′ ). (1.9)
Thus, we say that the position eigenstates are delta-normalized, rather than being nor-
malized to unity. Here δ d (· · · ) denotes the d-dimensional delta function; e.g., in 2D,
⟨x, y | x′ , y ′ ⟩ = δ(x − x′ ) δ(y − y ′ ).
Moreover, the position operator r̂ is defined such that r̂|r⟩ = r |r⟩.
Momentum eigenstates are constructed from position eigenstates via Fourier transforms.
First, let us suppose r runs over a finite box of length L on each side, with periodic boundary
conditions. For plane waves of the form exp(ik·r), where k = (k1 , . . . , kd ) is the wave-vector,
the periodic boundary conditions are satisfied if and only if
kj = 2πm/L for m ∈ Z. (1.10)
Note that the set of allowed wave-vectors is discrete (i.e., countable), with a discretization
of ∆k = 2π/L in each direction. Next, define
Z
1
|k⟩ = d/2 dd r eik·r |r⟩, (1.11)
L
where the integral is taken over the box. We identify |k⟩ as a momentum eigenstate, i.e., a
state with definite momentum ℏk. Using Eqs. (1.8)–(1.9) and Eq. (1.11), we can show that
1 ik·r X
⟨k|k′ ⟩ = δk,k′ , ⟨r|k⟩ = e , I= |k⟩ ⟨k|. (1.12)
Ld/2 k

In particular, note that each momentum eigenstate is normalizable to unity.

Now let us take L → ∞. In this limit, the wave-vector discretization ∆k = 2π/L goes to
zero, so the momentum eigenvalues coalesce into a continuum. To handle this, we redefine
the momentum eigenstates as
 d/2
(new) L
|k⟩ = |k⟩(old) . (1.13)
2π
Then, by using the formula
Z ∞
dx exp(ikx) = 2π δ(k), (1.14)
−∞

we can show that the new momentum eigenstates are related to the position eigenstates by

Z
1
|k⟩ = dd r eik·r |r⟩, (1.15)
(2π)d/2
Z
1
|r⟩ = dd k e−ik·r |k⟩. (1.16)
(2π)d/2

3
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

Position and momentum are now treated on similar footing, with the integrals over r and k
both taken over infinite spaces. Moreover, the momentum eigenstates now satisfy
Z
I = dd k |k⟩ ⟨k|, (1.17)

⟨k|k′ ⟩ = δ d (k − k′ ), (1.18)
eik·r
⟨r|k⟩ = . (1.19)
(2π)d/2
The first two equations are analogous to their position eigenstate counterparts, Eq. (1.8) and
(1.9). In particular, Eq. (1.18) says the new |k⟩’s are delta-normalized. The last equation,
(1.19), can be interpreted as the wavefunction of the delta-normalized momentum eigenstate.
The momentum operator p̂ is defined using the |k⟩’s as the eigenbasis:

p̂|k⟩ = ℏk |k⟩. (1.20)

For any quantum state |ψ⟩, whose wavefunction is ψ(r) = ⟨r|ψ⟩, we can use Eqs. (1.16) and
(1.20) to show that the momentum operator acts on wavefunctions in the following way:
Z
1
⟨r|p̂|ψ⟩ = d/2
dd k eik·r ⟨k|p̂|ψ⟩
(2π)
Z
1
= d/2
dd k ℏk eik·r ⟨k|ψ⟩
(2π) (1.21)
 Z 
1
= −iℏ∇ d/2
dd k eik·r ⟨k|ψ⟩
(2π)
= −iℏ∇ψ(r).

1.3. SCATTERING FROM A 1D DELTA FUNCTION POTENTIAL

We are now ready to solve a simple scattering problem. Consider a 1D space with spatial
coordinate x, and a scattering potential consisting of a “spike” at x = 0:
ℏ2 γ
V (x) = δ(x). (1.22)
2m
The form of the prefactor ℏ2 γ/2m is chosen for later convenience; the parameter γ, which
has units of [1/x], controls the strength of the scattering potential.

The particle wavefunction ψ(x) satisfies the 1D Schrödinger wave equation

ℏ2 d2
 
− + V (x) ψ(x) = Eψ(x). (1.23)
2m dx2

4
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

As this is a second-order differential equation, we normally require ψ(x) to be continuous and

to have well-defined first and second derivatives. However, the delta function potential (1.22)
relaxes these requirements in a peculiar way: at x = 0, ψ(x) must still be continuous, but
its first derivative may now be discontinuous (and hence, its second derivative is singular).
To see this, integrate Eq. (1.23) over an infinitesimal range around x = 0:
Z +ε Z +ε
ℏ2 d2 ℏ2 γ
 
lim dx − + δ(x) ψ(x) = lim+ dx Eψ(x)
ε→0+ −ε 2m dx2 2m ε→0 −ε
 +ε ! (1.24)
ℏ2 dψ ℏ2 γ
= lim+ − + ψ(0) = 0.
ε→0 2m dx −ε 2m

Therefore,  
dψ dψ
lim − = γ ψ(0). (1.25)
ε→0+ dx x=+ε dx x=−ε

In accordance our previously-discussed formulation of the scattering problem, and specif-

ically Eq. (1.6), the total wavefunction is split into
ψ(x) = ψi (x) + ψs (x). (1.26)
Moreover, the incident state should be a momentum eigenstate. But we will not let it be
delta-normalized, as momentum eigenstates normally are (see Sec. 1.2). Instead, we write
r
ikx 2mE
ψi (x) = Ψi e , k = > 0. (1.27)
ℏ2
The complex coefficient Ψi is called the incident amplitude. Introducing Ψi in this way
emphasizes that the magnitude and phase of the incident wavefunction are, in principle,
adjustable parameters in the scattering experiment. In terms of the usual delta-normalized
momentum eigenstates, √
|ψi ⟩ = 2π Ψi |k⟩. (1.28)
With this, Eq. (1.23) becomes
ℏ2 d2 ℏ2 γ
    
ikx ikx
− + δ(x) Ψi e + ψs (x) = E Ψi e + ψs (x) . (1.29)
2m dx2 2m
Taking E = ℏ2 k 2 /2m, and doing a bit of algebra, simplifies this to
 2 
d 2

ikx

+ k ψs (x) = γδ(x) Ψi e + ψs (x) . (1.30)
dx2

We can construct the solution by separately considering the regions x < 0 and x > 0.
Within each half-space, δ(x) vanishes so Eq. (1.30) reduces to the Helmholtz equation
 2 
d 2
+ k ψs (x) = 0. (1.31)
dx2
This has the general solution
ψs (x) = Ψi fR eikx + fL e−ikx ,


(1.32)

5
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

which is a superposition of right-moving (eikx ) and left-moving (e−ikx ) waves. The complex
parameters f1 and f2 can have different values in the x < 0 and x > 0 regions.
We want the scattered wave to be an outgoing wave. Thus, for x < 0, ψs is purely
left-moving, so fR = 0; and for x > 0, ψs is purely right-moving, so fL = 0. Therefore,
(
f− e−ikx , x < 0
ψs (x) = Ψi × (1.33)
f+ eikx , x > 0.

The complex numbers f− and f+ are called scattering amplitudes. They parameterize
the magnitude and phase of the scattered wavefunction moving outward from the scatterer.
Recall from the discussion at the beginning of this section that ψ(x) must be continuous
at x = 0. This implies that ψs (x) is also continuous, and therefore that

f− = f+ ⇒ ψs (x) = Ψi f± eik|x| . (1.34)

Plugging Eqs. (1.27) and (1.34) into Eq. (1.25) yields

γ
f± = − . (1.35)
γ − 2ik

For now, let us focus on the magnitude of the scattering amplitude (in the next chapter,
we will see that the phase also contains useful information). The quantity |f± |2 describes
the overall strength of the scattering process:
 −1
2 8mE
|f± | = 1 + . (1.36)
(ℏγ)2
Its dependence on E is plotted below:

The behavior matches our intuition. For a given particle energy E, the scattering grows
stronger as the potential strength parameter |γ| increases. On the other hand, for fixed γ, a
higher-energy particle undergoes less scattering. Note also that attractive potentials (γ < 0)
and repulsive potentials (γ > 0) are equally effective at scattering.

1.4. SCATTERING AMPLITUDE AND SCATTERING CROSS SECTION

In the 1D scattering problem of the previous section, the particle can only scatter in two
directions: forward or backward. In 2D and higher spatial dimensions, there is an important

6
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

complication: the particle can also go “sideways”, and in fact there is a continuous set of
directions for it to scatter into. To deal with this, physicists have developed a systematic
formalism for expressing the outcomes of scattering experiments using direction-dependent
“scattering amplitudes”, which we will now describe.
To begin, let us define the incident wavefunction as a plane wave in d dimensions, by
analogy with the 1D case [Eq. (1.27)]:
r
iki ·r 2mE
ψi (r) = Ψi e , |ki | = k = , (1.37)
ℏ2
where ki is the incident wave-vector and Ψi ∈ C is the incident amplitude.
The incident wavefunction generates a scattered wavefunction ψs (r). We can express the
r-dependence of this wavefunction using a coordinate system of the form (r, Ω), where r is
the distance from the origin and Ω denotes the other “directional” coordinates. Specifically,

• For d = 2, we use polar coordinates (r, ϕ), i.e. Ω ≡ ϕ.

• For d = 3, we use spherical coordinates (r, θ, ϕ), i.e. Ω ≡ (θ, ϕ).

We can also wrangle the d = 1 case into this framework by letting Ω ∈ {+, −}, where +
denotes the forward (+x) direction and − denotes the backward (−x) direction.
In a typical scattering experiment, the scattered wavefunction is detected at r → ∞; i.e.,
at distances much longer than the size of the scatterer, the free-space wavelength, and any
other relevant length scales. In this regime, we assert that the scattered wavefunction must
reduce to the form 1−d
r→∞
ψs (r) −→ Ψi r 2 eikr f (Ω). (1.38)
To understand the reasoning behind this, let us go through the individual multiplicative
factors on the right side of Eq. (1.38).
The first factor Ψi accounts for the fact that the Schrödinger wave equation is linear. If
we vary the incident amplitude, the scattered wavefunction must vary proportionally.
1−d
The next two factors, r 2 eikr , give the r-dependence of a wave expanding radially from
the origin in d dimensions. The eikr factor captures its outgoing nature (the wavefunc-
tion’s phase increases along the direction of propagation, i.e., the direction of increasing r).
The other factor provides the r-scaling needed for probability conservation. Note that the
probability current density associated with the scattered wavefunction is
ℏ  ∗ 
Js = Im ψs ∇ψs . (1.39)
m
Thus, its radial component is
 
r ℏ ∗ ∂
Js = Im ψs ψs
m ∂r
 ∗ ∂  1−d 
r→∞ 1−d
ikr ikr


−→ · · · Im r 2 e r 2 e
∂r (1.40)
 ∗  1−d 

1−d
ikr ikr 1 − d −1−d ikr
= · · · Im r 2 e ikr 2 e + r 2 e
2
= · · · r−(d−1) + O(r−d ),

7
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

where (· · · ) denotes r-independent terms. In d dimensions, the area of the surface at distance
r from the origin scales as rd−1 . Hence, Jsr scales inversely with area, and the total flux
obtained by integrating Jsr over the area is independent of r.
The final factor in Eq. (1.38), f (Ω), is called the scattering amplitude. This complex
function is typically the main avenue by which a scattering experiment reveals information
about the scatterer. As noted in the previous paragraphs, in the large-r limit, the other
factors are determined by linearity, outgoing-ness, and flux conservation. Hence, only f is
sensitive to the details of the scattering potential.
Note that the scattering amplitude is a function of Ω, i.e., the direction of r. Sometimes,
it is written using the alternative notation
r
f (ki → kf ), kf = k . (1.41)
|r|
This emphasizes that the particle is initially incident with wave-vector ki , then gets scattered
elastically into a wave-vector kf pointing in the direction of r.
From the scattering amplitude, we can define two other important quantities:

dσ 2
= f (Ω) (the differential scattering cross section) (1.42)
dΩ Z
2
σ = dΩ f (Ω) (the total scattering cross section). (1.43)

R
In Eq. (1.43), dΩ denotes the integral(s) over all the angle coordinates. For 1D, this is
replaced by a discrete sum over the forward and backward directions (see above).
The term “cross section” comes from an analogy with the scattering of classical particles.
Imagine a stream of classical particles incident on a “hard-body” scatterer:

Let the incident classical particles have the same flux as the incident wavefunction (1.37),
ℏk
Ji = |Ψi |2 . (1.44)
m
Note that this has units of [x1−d t−1 ] (i.e., number per unit time per unit area in d dimensions).
We assume the scatterer only interacts with the particles striking it directly, and the rate
at which these strikes occur is
Is = Ji σ, (1.45)
where σ is the exposed cross-sectional area of the scatterer, as shown in the figure. This is
the total rate at which scattering occurs, regardless of where the particles scatter to.

8
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

Let us compare this to the quantum mechanical scattering rate. Using Eq. (1.38), we can
repeat the calculation in (1.40) for the radial component of the probability current density:

r→∞ ℏk
Jsr −→ |Ψi |2 |f (Ω)|2 r1−d + O(r−d ). (1.46)
m
Hence, the total flux of outgoing probability is
Z   Z
d−1 ℏk 2 2
Is = dΩ r Js,r = |Ψi | dΩ f (Ω) . (1.47)
m

On the right side, the quantity in parentheses is the incident flux, Eq. (1.44). Comparing
Eq. (1.47) to Eq. (1.45), we see that
Z
2
σ ≡ dΩ f (Ω)

is the quantity analogous to the exposed cross-sectional area of the classical hard-body. We
therefore call this the total scattering cross section.
Moreover, the integrand |f |2 represents the rate, per unit of solid angle, at which particles
are scattered in a given direction. We call this the differential scattering cross section.
The total and differential scattering cross sections are the principal observable quantities
typically obtained in scattering experiments.

1.5. THE GREEN’S FUNCTION

The scattering amplitude f (Ω) can be calculated using a variety of analytical and nu-
merical methods. We will discuss one particularly important approach, based on a quantum
variant of the Green’s function technique for solving inhomogenous differential equations.
Let us return to the previously-discussed formulation of the scattering problem:

Ĥ = Ĥ0 + V̂
Ĥ|ψ⟩ = E|ψ⟩
(1.48)
|ψ⟩ = |ψi ⟩ + |ψs ⟩
Ĥ0 |ψi ⟩ = E|ψi ⟩.

These equations can be combined as follows:

 
Ĥ0 + V̂ |ψi ⟩ + Ĥ|ψs ⟩ = E (|ψi ⟩ + |ψs ⟩) (1.49)
⇒ V̂ |ψi ⟩ + Ĥ|ψs ⟩ = E|ψs ⟩ (1.50)
 
⇒ E − Ĥ |ψs ⟩ = V̂ |ψi ⟩ (1.51)

To proceed, we define an operator called the Green’s function, which is the inverse of the
operator on the left side of Eq. (1.51):

−1
Ĝ(E) = E − Ĥ . (1.52)

9
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

Note that Ĝ(E) depends parametrically on E. We can now convert Eq. (1.51) into

|ψs ⟩ = ĜV̂ |ψi ⟩. (1.53)

It will also be useful to define the Green’s function for a free particle,

−1
Ĝ0 (E) = E − Ĥ0 . (1.54)

As we shall see, Ĝ0 often can be calculated exactly, whereas Ĝ typically has no closed-form
analytic expression. Moreover, we can relate G and G0 as follows:

Ĝ(E − Ĥ0 − V̂ ) = I and (E − Ĥ0 − V̂ )Ĝ = I

(1.55)
⇔ ĜĜ−1
0 − ĜV̂ = I and Ĝ−1
0 Ĝ − V̂ Ĝ = I.

Upon respectively right-multiplying and left-multiplying these equations by Ĝ0 , we arrive

at the following pair of equations, called Dyson’s equations:

Ĝ = Ĝ0 + ĜV̂ Ĝ0 (1.56)

Ĝ = Ĝ0 + Ĝ0 V̂ Ĝ (1.57)

These equations are “implicit”, as the unknown Ĝ appears on both the left and right sides.
Applying the second Dyson equation, Eq. (1.57), to the scattering problem (1.53) gives
 
|ψs ⟩ = Ĝ0 + Ĝ0 V̂ Ĝ V̂ |ψi ⟩ (1.58)
 
= Ĝ0 V̂ Iˆ + ĜV̂ |ψi ⟩. (1.59)

Using Eq. (1.53) again, we get

|ψs ⟩ = Ĝ0 V̂ |ψ⟩. (1.60)
This remains an implicit equation, but the unknown is |ψ⟩ rather than Ĝ. Note that the
results up to this point have been approximation-free.
One approach to solving Eq. (1.60) is to repeatedly plug its right side back into itself:
 
|ψs ⟩ = Ĝ0 V̂ |ψi ⟩ + Ĝ0 V̂ |ψ⟩
.. (1.61)
= .
h i
= Ĝ0 V̂ + (Ĝ0 V̂ )2 + (Ĝ0 V̂ )3 + · · · |ψi ⟩.

Or, equivalently, h i
|ψ⟩ = Iˆ + Ĝ0 V̂ + (Ĝ0 V̂ )2 + (Ĝ0 V̂ )3 + · · · |ψi ⟩. (1.62)
This infinite series is called the Born series.

10
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

To interpret this result, let us go to the position basis:

Z
ψ(r) = ψi (r) + dd r′ ⟨r|Ĝ0 |r′ ⟩ V (r′ )ψi (r′ )
Z (1.63)
+ dd r′ dd r′′ ⟨r|Ĝ0 |r′ ⟩ V (r′ ) ⟨r′ |Ĝ0 |r′′ ⟩ V (r′′ )ψi (r′′ ) + · · ·

This can be regarded as a description of multiple scattering. The wavefunction is a

superposition of terms involving zero, one, two, or more scattering events, as shown below:

Each successive term in the Born series involves more scattering events, i.e., higher multiples
of V̂ . For example, the second-order term is
Z
dd r′ dd r′′ ⟨r|Ĝ0 |r′ ⟩ V (r′ ) ⟨r′ |Ĝ0 |r′′ ⟩ V (r′′ )ψi (r′′ ),

which describes the particle undergoing (i) scattering of the incident particle at point r′′ ,
(ii) propagation from r′′ to r′ , (iii) scattering again at point r′ , and (iv) propagation from r′
to r. Note that r′ and r′′ are integrated over all possible positions. Since the integrals are
weighted by V , the positions with the strongest scattering potential contribute the most.
For a sufficiently weak scatterer, it is often a good approximation to retain just the first
few terms in the Born series. The question of when V̂ is “sufficiently weak”—i.e., when the
Born series converges—is a complex topic beyond the scope of our present discussion.

1.6. GREEN’S FUNCTION FOR A FREE PARTICLE

In the position basis representation of the Born series, Eq. (1.63), a crucial role is played
by the matrix elements ⟨r|Ĝ0 |r′ ⟩. We call this the propagator, as it describes how the
particle propagates between discrete scattering events.
Starting from the definition of Ĝ0 in Eq. (1.54), we can evaluate it in the position basis:

ˆ ′⟩
⟨r| E − Ĥ0 Ĝ0 |r′ ⟩ = ⟨r|I|r

ℏ2 2
 
= E+ ∇ ⟨r|Ĝ0 |r′ ⟩ = δ d (r − r′ )
2m

Hence,
2m
∇2 + k 2 ⟨r|Ĝ0 |r′ ⟩ = 2 δ d (r − r′ ),


(1.64)
ℏ
where
2mE
k2 = . (1.65)
ℏ2

11
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

The left side of this equation is the same as the d-dimensional Helmholtz equation, with
the ∇2 acting on the r coordinates (not r′ ). On the right side, we have a term proportional
to a d-dimensional delta function at r = r′ . The propagator thus describes a wave at each
position r, emitted from a point source at r′ .
Eq. (1.64) can be solved analytically for different values of the spatial dimension d. In
the following, we will work through the derivation for d = 3. The derivations for d = 1 and
d = 2 are left as exercises.
To derive the 3D propagator, we first re-express it using the momentum eigenbasis:
Z 
′
⟨r|Ĝ0 |r ⟩ = ⟨r|Ĝ0 d3 k ′ |k′ ⟩⟨k′ | |r′ ⟩
Z
1
= d3 k ′ ⟨r|k′ ⟩ ℏ2 |k′ |2
⟨k′ |r′ ⟩
E − 2m
′ ′
3 ′ exp [ik · (r − r )]
Z
2m 1
= 2 d k . (1.66)
ℏ (2π)3 k 2 − |k′ |2
Next, we express k′ using spherical coordinates (k ′ , θ, ϕ). We are free to choose the θ = 0
axis so that it points in the direction of r − r′ , as illustrated below:

Then,
′ ′
3 ′ exp [ik · (r − r )]
Z
′ 2m 1
⟨r|Ĝ0 |r ⟩ = 2 d k
ℏ (2π)3 k 2 − |k′ |2
Z ∞
exp (ik ′ |r − r′ | cos θ)
Z π Z 2π
2m 1 ′ ′2
= 2 dk dθ dϕ k sin θ
ℏ (2π)3 0 0 0 k2 − k′2
Z ∞ Z 1 ′ ′
2m 1 ′ ′ 2 exp (ik |r − r |µ)
= 2 dk dµ k (letting µ = cos θ)
ℏ (2π)2 0 −1 k2 − k′2
k ′ 2 exp (ik ′ |r − r′ |) − exp (−ik ′ |r − r′ |)
Z ∞
2m 1 ′
= 2 dk
ℏ (2π)2 0 k2 − k′2 ik ′ |r − r′ |
Z ∞ ′ ′ ′
2m 1 i ′ k exp (ik |r − r |)
= 2 dk . (1.67)
ℏ (2π)2 |r − r′ | −∞ (k ′ − k)(k ′ + k)
This looks like something we can handle via contour integration. But there’s a snag: the
integration contour runs over the real k ′ line, but since k ∈ R+ the poles at ±k lie on the
contour. The integral, as currently defined, is singular.

12
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

To get a finite result, we must “regularize” the integral by adjusting its definition. One
way is to perturb the integrand and shift its poles infinitesimally in the complex plane,
moving them off the contour. We have a choice of whether to move each pole infinitesimally
up or down. It turns out that the appropriate choice is to shift the pole at −k down, and
shift the pole at +k up, as illustrated by the red dots in the following figure:

This alters the denominator in the integrand of Eq. (1.67) as follows:

2
(k ′ − k)(k ′ + k) → (k ′ − k − iϵ)(k ′ + k + iϵ) = k ′ − (k + iϵ)2 , (1.68)

where ϵ is a positive infinitesimal. The implications of this adjustment will be discussed

further below. For now, let us proceed to do the integral, which is no longer divergent:
Z ∞ ′ ′ ′ Z ∞
′ k exp (ik |r − r |) ′ k ′ exp (ik ′ |r − r′ |)
dk → lim dk (regularize)
−∞ (k ′ − k)(k ′ + k) ε→0+ −∞ (k ′ − k − iε)(k ′ + k + iε)
k ′ exp (ik ′ |r − r′ |)
Z
= lim+ dk ′ ′ (close contour above)
ε→0 C (k − k − iε)(k ′ + k + iε)
k ′ exp (ik ′ |r − r′ |)
 
′ +
= 2πi lim+ Res , k = k + iε
ε→0 (k ′ − k − iε)(k ′ + k + iε)
= πi exp (ik|r − r′ |) .

Finally, plugging this into Eq. (1.67) gives the result

2m exp (ik|r − r′ |)
⟨r|Ĝ0 |r′ ⟩ = − · . (1.69)
ℏ2 4π|r − r′ |

The 3D propagator (1.69) describes a spherical wave centered at the source r′ . The wave
is isotropic (i.e., independent of the direction of r − r′ ), consistent with the fact that the
source, a delta function, has no preferred direction. The amplitude of the wave decreases
inversely with ∆r = |r − r′ |, consistent with flux conservation in 3D (see Sec. 1.4).
Most importantly, the wave propagates outward from r′ (since the phase of the wave-
function increases with ∆r). Such a propagator is said to be causal, a word based on the
term “cause and effect”—the wave travels from the source at r′ (the cause) to r (the effect).
We can trace this behavior back to the regularization in Eq. (1.68); other regularization
choices can be shown to yield propagators with different properties (see exercises). Our
regularization (1.68) can also be expressed as an infinitesimal shift in E, via Eq. (1.65):

ℏ2 k 2 ℏ2 (k + iϵ)2 ℏ2
k 2 + 2ikϵ + · · · = E + iε.


=E ⇒ = (1.70)
2m 2m 2m

13
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

In other words, we have swapped the original definition of the Green’s function, Eq. (1.54),
for a causal Green’s function

−1
Ĝ0 = lim+ E − Ĥ0 + iε . (1.71)
ε→0

Based on Eq. (1.71), causal propagators can also be derived for other dimensions (see
exercises). The results for d = 1, 2, and 3 are summarized below:


1
exp (ik|x − x′ |) , d = 1



2ik






′ 2m 1
⟨r|Ĝ0 |r ⟩ = 2 × H0+ (k|r − r′ |), d=2 (1.72)
ℏ 
 4i

exp (ik|r − r′ |)



− , d = 3.


4π|r − r′ |



For each d, the causal propagator describes a wave propagating outward isotropically from
r′ . In 1D, this has the form of a plane wave on each side. In 2D, it is a circular wave (H0+
denotes a Hankel function; see Appendix A), and in 3D it is a spherical wave.

1.7. SCATTERING AMPLITUDES IN 3D

We are now in a position to generate explicit solutions for the scattering problem by
combining the results of Sections 1.4–1.6. For ease of presentation, we will focus on the 3D
case; the 1D and 2D cases are handled in a similar way.
Using Eqs. (1.38) and (1.59), we can write the scattered wavefunction as

r→∞ eikr
ψs (r) = ⟨r|Ĝ0 V̂ Iˆ + ĜV̂ |ψi ⟩


−→ Ψi f (ki → kr̂). (1.73)
| {z } r
=|ψ⟩

We want to use this to find the scattering amplitude f , which encodes the results of the
scattering experiment in the large-r limit (see Sec. 1.4).
Let us work on the expression on the left first. Using the position representation,
Z
ψs (r) = d3 r′ ⟨r|Ĝ0 |r′ ⟩ V (r′ ) ⟨r′ | Iˆ + ĜV̂ |ψi ⟩.


(1.74)

From Eq. (1.72), the causal propagator in 3D is

2m exp (ik|r − r′ |)
⟨r|Ĝ0 |r′ ⟩ = − . (1.75)
ℏ2 4π|r − r′ |

14
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

In the r → ∞ limit, this can be simplified using the Taylor expansion

|r − r′ | = r − r̂ · r′ + · · · , (1.76)

where r̂ is the unit vector pointing parallel to r. (This is the same large-r expansion used
to derive the electric dipole moment in classical electromagnetism.) Hence, to lowest order,

r→∞ 2m eikr
⟨r|Ĝ0 |r′ ⟩ −→ − exp (−ik r̂ · r′ ) . (1.77)
ℏ2 4πr
Applying this to Eq. (1.74) gives

eikr
Z
r→∞ 2m
d3 r′ exp (−ik r̂ · r′ ) V (r′ ) ⟨r′ | Iˆ + ĜV̂ |ψi ⟩


ψs (r) −→ − 2
ℏ 4πr
r (1.78)
2m eikr π
= − 2 kf V̂ + V̂ ĜV̂ ψi .
ℏ r 2

On the second line, we simplified the expression by introducing the momentum eigenstate
with kf ≡ kr̂. This corresponds to the “final” momentum of the scattered particle, measured
when the particle is picked up by a detector placed in the r̂ direction relative to the origin.
Note that |kf | = |ki | = k, consistent with elastic scattering. Similarly, we can express the
incident state in terms of a momentum eigenstate, by comparing Eqs. (1.19) to (1.37):

|ψi ⟩ = Ψi (2π)3/2 |ki ⟩. (1.79)

Thus, Eq. (1.78) becomes

r→∞2m eikr
ψs (r) −→ − 2 Ψi 2π 2 kf V̂ + V̂ ĜV̂ ki . (1.80)
ℏ r
Comparing this to the right side of Eq. (1.73), we can read off the scattering amplitude:

2m
f (ki → kf ) = − · 2π 2 kf V̂ + V̂ ĜV̂ ki
ℏ2 (1.81)
2m
= − 2 · 2π 2 kf V̂ + V̂ Ĝ0 V̂ + V̂ Ĝ0 V̂ Ĝ0 V̂ + · · · ki .
ℏ

As noted above, this result is subject to the elasticity constraint |ki | = |kf |. To get the
infinite series on the second line of Eq. (1.81), we have used Eq. (1.62).
Eq. (1.81) is the culmination of numerous definitions and derivations from the preceding
sections. On the left side is the scattering amplitude, the fundamental quantity of interest in
a scattering experiment. The right side contains the various inputs to the scattering problem:
the initial and final momenta, the scattering potential, and the Green’s function. Although
this result was derived for the 3D case, very similar formulas hold for other dimensions, with
the 2π 2 factor replaced with other numerical factors.

15
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

1.8. EXAMPLE: UNIFORM SPHERICAL WELL IN 3D

Let us test the Born series with a simple scatterer. Consider a scattering potential that
consists of a uniform spherical well of depth U > 0 and radius R:
(
−U, |r| ≤ R
V (r) = (1.82)
0, |r| > 0.

As it turns out, this particular scattering problem can be solved exactly by exploiting
the spherical symmetry of V (r). The solution method, called partial wave analysis, is
explained in Appendix A. Its result is
∞
1 X 2iδℓ


f (ki → kf ) = e − 1 2ℓ + 1 Pℓ (k̂i · k̂f ),
2ik ℓ=0
π h
+′ + ′
i
where δℓ = + arg khℓ (kR) jℓ (qR) − qhℓ (kR) jℓ (qR) , (1.83)
2
p
q = 2m(E + U )/ℏ2
k ≡ |ki | = |kf |.
The solution is expressed in terms of the special functions jℓ (spherical Bessel function of the
first kind), h+
ℓ (spherical Hankel function of the first kind), and Pℓ (Legendre polynomial).
We will pit this against the Born series, which can be regarded as a more general method
since it does not assume spherical symmetry. According to Eq. (1.81), the Born series gives
2m h i
f (ki → kf ) = − 2 · 2π 2 kf V̂ |ki + kf V̂ Ĝ0 V̂ ki + · · · . (1.84)
ℏ
Evaluating the full infinite Born series is generally an intractable problem, but we can
truncate it to get an approximation for f . For example, if we keep only the first term, the
result is called the first Born approximation; if we keep only the first two terms, we
obtain the second Born approximation. To evaluate the various bra-kets, we can use
the position representation to express them as explicit integrals. For example,
exp(−ikf · r1 ) exp(iki · r1 )
Z
kf V̂ |ki = d3 r1 3/2
V (r1 )
(2π) (2π)3/2
Z
U
=− 3 d3 r1 exp [i(ki − kf ) · r1 ] (1.85)
8π |r1 |≤R
exp(−ikf · r2 ) exp(iki · r1 )
Z Z
kf V̂ Ĝ0 V̂ ki = d r1 d3 r2
3
3/2
V (r2 ) ⟨r2 |Ĝ0 |r1 ⟩ V (r1 )
(2π) (2π)3/2
U 2 2m exp [i(k|r1 − r2 | − kf · r2 + ki · r1 )]
Z Z
=− 4 2
d3 r1 d3 r2 .
32π ℏ |r1 |<R |r2 |<R |r1 − r2 |
(1.86)

To evaluate the integrals, an expedient approach is Monte Carlo integration. For instance,
the integral (1.85) has the form
Z
I= d3 r F (r). (1.87)
|r|<R

16
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

To compute this, we randomly sample N points within a cube of volume (2R)3 centered at
the origin. Note that this cube encloses the radius-R sphere over which is the integration
region. For each sampled point, r(n) , we take
(
F (r(n) ), |r(n) | < R
Fn = (1.88)
0, otherwise,

which drops contributions from points outside the sphere. Then our estimate for I is
N
3 (2R)3 X
I ≈ (2R) ⟨Fn ⟩ = Fn . (1.89)
N n=1

The estimate converges to the true value as N → ∞. In practice, N ∼ 104 yields a good
result for typical 3D integrals, and can be calculated in seconds on a modern computer.
Similarly, to calculate a double integral like Eq. (1.86), we sample pairs of points,
(n) (n)
(r1 , r2 ), and replace the volume factor (2R)3 in Eq. (1.89) with (2R)6 . The rest of the
estimation procedure is the same.
In the figure below, we compare the Born approximation to the exact solution (1.83).
We plot |f |2 versus the energy E at a 90◦ scattering angle (i.e., kf perpendicular to ki ), for
fixed well radius R = 1 and two well depths, U = 0.1 and U = 1. We adopt computational
units ℏ = m = 1, and compute each Monte Carlo integral using 3 × 104 samples.

The first thing to notice is that |f |2 diminishes to zero for large E. This makes sense:
the scattering potential has some energy scale ∼ U , so a particle with energy E ≫ U should
simply zoom through, with little chance of being deflected.
Looking more closely, we see that for the shallower well (U = 0.1), the first Born ap-
proximation already agrees quite well with the exact results. However, the second Born
approximation is significantly more accurate, particularly for small E.
For the deeper well (U = 1), neither the first nor second Born approximations do a good
job. Roughly speaking, for the deep potential well, we cannot neglect multiple-scattering
events represented by higher terms in the Born series, whereby the particle bounces around
many times before leaving the scatterer (see Sec. 1.5).
In fact, for very strong scattering potentials, even taking the Born approximation to high
orders may not work, as the Born series may become non-convergent. In such cases, different
methods must be brought to bear. We will see an example in the next chapter.

17
Y. D. Chong Ch. 1: Scattering Theory | Graduate Quantum Mechanics

EXERCISES

1. In Sec. 1.2, we derived the eigenstates of a particle in an empty infinite space by

considering a box of length L on each side, applying periodic boundary conditions,
and taking L → ∞. Suppose we instead use Dirichlet boundary conditions (i.e., the
wavefunction vanishes on the walls of the box). Show that this gives rise to the same
set of momentum eigenstates in the L → ∞ limit.
2. Using the results for the 1D delta-function scattering problem described in Section 1.3,
calculate the probability current
dψ ∗
 
ℏ ∗ dψ
J(x) = ψ −ψ , (1.90)
2mi dx dx
where ψ(x) is the total (incident + scattered) wavefunction. Explain the relationship
between the values of J on the left and right side of the scatterer.
3. Derive the causal 1D propagator:
2m 1
⟨x|Ĝ0 |x′ ⟩ = · exp (iki |x − x′ |) . (1.91)
ℏ2 2iki
4. Derive the causal 2D propagator:
2m 1 +
⟨r|Ĝ0 |r′ ⟩ =
2
H0 (k|r − r′ |). (1.92)
ℏ 4i
You should be able to adapt many of the steps in the 3D propagator derivation from
Section 1.6. When evaluating the polar integral, you may want to refer to the discus-
sion of Bessel and Hankel functions in Appendix A. You may also need the following
mathematical identity between the Hankel functions:
+ −
Hm (z) = −(−1)m Hm (−z). (1.93)
5. In place of the causal Green’s function defined in Eq. (1.71), consider an infinitesimal
shift of the opposite sign:

−1
Ĝ0 = lim+ E − Ĥ0 − iε . (1.94)
ε→0

(a) Whereas the causal propagators in Eq. (1.72) represent outgoing waves, explain
why Eq. (1.94) will give rise to propagators representing incoming waves.
(b) In the derivation of the 3D propagator, Eqs. (1.66)–(1.69), show mathematically
how Eq. (1.94) changes the contour integration, resulting in an incoming-wave
solution.
6. In Section 1.7, the scattering amplitude f (k → k′ ) for the 3D scattering problem was
derived using the Born series. Derive the corresponding expressions for 1D and 2D.

FURTHER READING

[1] Bransden & Joachain, §13.1—13.3 and §13.5—13.6.

[2] Sakurai, §7.1–7.3, 7.5–7.6

18