Scattering Theory: Born Series

Stefan Blügel

This document has been published in

Manuel Angst, Thomas Brückel, Dieter Richter, Reiner Zorn (Eds.):
Scattering Methods for Condensed Matter Research: Towards Novel Applications at
Future Sources
Lecture Notes of the 43rd IFF Spring School 2012
Schriften des Forschungszentrums Jülich / Reihe Schlüsseltechnologien / Key Tech-
nologies, Vol. 33
JCNS, PGI, ICS, IAS
Forschungszentrum Jülich GmbH, JCNS, PGI, ICS, IAS, 2012
ISBN: 978-3-89336-759-7
All rights reserved.
A2 Scattering Theory: Born Series 1

Stefan Blügel
Peter Grünberg Institut and
Institute for Advanced Simulation
Forschungszentrum Jülich GmbH

Contents
1 Introduction 2

2 The Scattering Problem 2

2.1 The Experimental Situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Description of Scattering Experiment . . . . . . . . . . . . . . . . . . . . . . 4
2.3 Coherence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 The Cross Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Lippmann Schwinger Equation 9

4 Born Approximation 11
4.1 Example of Born Approximation: Central Potential . . . . . . . . . . . . . . . 12
4.2 Example of Born Approximation: Square Well Potential . . . . . . . . . . . . 13
4.3 Validity of first Born Approximation . . . . . . . . . . . . . . . . . . . . . . . 14
4.4 Distorted-Wave Born Approximation (DWBA) . . . . . . . . . . . . . . . . . 15

5 Method of Partial Wave Expansion 17

5.1 The Born Approximation for Partial Waves . . . . . . . . . . . . . . . . . . . 20
5.2 Low Energy Scattering: Scattering Phases and Scattering Length . . . . . . . . 21
5.3 S-Wave Scattering at Square Well Potential . . . . . . . . . . . . . . . . . . . 22
5.4 Nuclear Scattering Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

6 Scattering from a Collection of Scatterers 25

Lecture Notes of the 43rd IFF Spring School “Scattering Methods for Condensed Matter Research: Towards
1

Novel Applications at Future Sources” (Forschungszentrum Jülich, 2012). All rights reserved.
A2.2 Stefan Blügel

1 Introduction
Since Rutherford’s surprise at finding that atoms have their mass and positive charge concen-
trated in almost point-like nuclei, scattering methods are of extreme importance for studying
the properties of condensed matter at the atomic scale. Electromagnetic waves and particle
radiation are used as microscopic probes to study a rich variety of structural and dynamical
properties of solids and liquids. Atomistic processes in condensed matter take place at length
scales on the order of an Ångström (1Å= 10−10 m) and an energy scale between a meV and a
few eV. Obviously, detailed information concerning atomic systems require measurements re-
lated to their behavior at very small separations. Such measurements are in general not possible
unless the de Broglie wavelength (λ = hp = mv h
) of the relative motion of the probing particle is
comparable to these distances. This makes x-ray scattering and neutron scattering, in addition
to electron scattering and to a certain extent also Helium scattering, to the outstanding micro-
scopic “measurement instruments” for studying condensed matter. To push electromagnetic
waves in this area one uses either x-rays with wavelengths of a few Ångströms, but in the keV
energy range, or light with energies in the eV range, but wavelengths of some 1000 Å. Neutrons
(and Helium atoms) make it possible to match energy and wavelengths simultaneously to the
typical atomic spacings and excitation energies of solids (solid surfaces). Thus, a simultaneous
spatial and temporal resolution of atomistic or magnetic processes is possible. In addition, the
photon, neutron, electron and under certain conditions also Helium possess internal degrees of
freedom such as a polarization vector or a spin with which the probes couple to core and va-
lence electrons. The photon, neutron and Helium result in only week interaction with matter,
which simplifies considerably the analysis and interpretation of experiments as multiple scatter-
ing processes are frequently of minor importance and can often be ignored completely, which
makes an interpretation of the scattering results valid within a kinematic scattering theory.
In this Chapter, we will provide a brief introduction to the elementary concepts and methodol-
ogy of scattering theory. The focus lies on the introduction of the description of the scattering
process in terms of the Hamiltonian of the scattering projectile at a finite range interaction poten-
tial of a single site target by the method of partial waves, i.e. using differential equation methods
and the Lippmann-Schwinger equation, i.e. using an integral equation formulation of scattering
that leads then to the first Born approximation of scattering and the distored wave Born ap-
proximation. The former is the approximation of choice if multiple scattering is unimportant
and the latter is applied in the analysis of grazing-incidence small-angle scattering experiments
discussed in more details chapter D2. The lecture closes with the discussion of the scattering
on the lattice rather than a single site target resulting briefly the Bragg scattering, that will be
discussed in more detail in the lecture Scattering Theory: Dynamical Theory (A3). The subject
is typically part of an Advanced Quantum Mechanics curriculum and is therefore elaborated at
textbooks on quantum mechanics. A selection is given as references [1, 2, 3, 4].

2 The Scattering Problem

In a scattering experiment a beam of particles is allowed to strike a target,2 and the particles that
emerge from the target area or scattering volume, respectively, are observed.
2
In the language of elementary scattering theory one frequently refers to a target although we keep in mind that
in the language of condensed matter, it is referred to as the sample.
Basic Scattering Theory A2.3

Detector

Θ
Collimator z-axis
Source Target

Fig. 1: A set-up of a typical diffraction experiment consisting of a particle source, a scattering

target and a particle detector. The beam passes through a collimator with an beam opening ∆r.

2.1 The Experimental Situation

A schematic representation of a standard scattering experiment appears in Fig. 1. Each scat-
tering experiment consists of three indispensable elements: (i) The source of incident beam of
particles or electromagnetic wave, to propagate with wave vector k = 2π λ
k̂ of wavelength λ
along the direction k̂, which we assume without loss of generally to be the ẑ-direction (the axis
of the collimator). (ii) The target, that we consider stationary, a reasonable assumption in con-
densed matter physics and (iii) the detector, whose function is to simply count the number of
particles of a particular type that arrive at its position r with the coordinate r = (x, y, z) in real
space along the direction r̂ at the angle (θ, ϕ) with respect to the axis of the propagating incident
beam. Ideally it may be set to count only particles of a given energy, spin or polarization vector,
respectively. We assume throughout that the source and detector are classical objects that have
a clearly defined, precisely controllable effect on the scattering process. The detector will be
assumed to be 100% efficient and to have no effect on the scattered particle prior to the time it
enters the detector.
The result of the scattering experiment will vary with the energy E of the incident beam. In or-
der to simplify the analysis of the experimental data, the energy spectrum of the incident beam
should be sharply peaked so that the experiment may be considered to take place at a unique en-
ergy eigenvalue E. To this end, in most experiments care is taken to achieve a monochromatic
incident beam characterized by the wave vector k. We shall assume here that the beam emerg-
ing from the collimator is both perfectly monochromatic and perfectly collimated, as well. Of
course, according to the uncertainty principle, a beam of finite cross section (of the size of the
collimator opening ∆r) cannot be perfectly monochromatic (∆k > 0) and perfectly collimated
as well. We may, however, assume the beam to be sufficiently well collimated that the angular
divergence may be ignored in an actual experiment. In that case we must necessarily have not a
monochromatic beam represented by a plane wave, but rather a beam describable as a superpo-
sition of such waves. In this respect the collimator can be considered the fourth indispensable
element of a scattering experiment. It shields the detector from the incoming beam to the ideal
extent that no count is measured in the detector without target, and produces a small beam of
monochromatic energy. We further assume that the detector has a small opening angle dΩ and
is positioned at large distance from the target. Under these conditions, the scattered beam can be
characterized at the position of the detector by the wave vector k0 and energy E 0 . Summarizing,
in a scattering experiment a wave packet of incident particles characterized by the initial state
(k, E, e), denoting the polarization vector e of an x-ray beam as a representative of an internal
A2.4 Stefan Blügel

x-axis

Detector

k0
φ scattered beam
(E 0 , k0 , e0 )
incident beam
(E, k, e) r̂ θ r2 dΩ

k Target z-axis

y-axis

Fig. 2: The geometry of the scattering experiment.

degree of freedom of the particle, is scattered into the final state:

scattering
(k, E, e) −→ (k0 , E 0 , e0 ). (1)

The scattering process is characterized by the scattering vector

Q = k0 − k (2)

and the energy transition

~ω = E 0 − E. (3)
~Q represents the momentum transfer during scattering, since according to de Broglie, the mo-
mentum of the particle corresponding to the wave with wave vector k is given by p = ~k. For
elastic scattering (diffraction), it holds that E 0 = E and |k0 | = |k| and all possible scattering
vectors are located on a sphere, called Ewald-sphere. Structural investigations are always car-
ried out by elastic scattering. The magnitude Q of the scattering vector can be calculated from
wavelength λ and scattering angle θ as follows
r
√ p θ θ 4π θ
Q = k + k − 2kk cos θ = 2k (1 − cos θ) = k 2(1 − cos2 + sin2 ) =
2 02 0 2 sin
2 2 λ 2
(4)

2.2 Description of Scattering Experiment

After the beam of particles is emitted from the collimated source, the experimenter has no
control over the particles until they have reached his detector. During that time, the propagation
is controlled solely by the laws of quantum mechanics and the Hamiltonian of the projectile-
target system. We restrain our description to the nonrelativistic domain and may thus formulate
the physical situation in terms of the solution of a Schrödinger equation using an appropriate
Basic Scattering Theory A2.5

Hamiltonian and suitable boundary conditions. The occurring phenomena can be very complex.
We assume in the following that the incident particles do not interact with each other during
the time of flight, rather that the particles fly one by one, and that incident particles and the
target particles do not change their internal structures or states, but only scatter off each other.
Internal excitations, rearrangements, charge or spin exchange are also excluded. In fact, internal
degrees of freedom such as the spin or polarization vector are currently completely neglected.
That means we shall consider only the purely elastic scattering. We further neglect the multiple
scattering in the target and consider at first only the interaction of an incident particle with one
target particle, the state of both is described by a two-particle wave function Ψ(rP , rT ) and the
interaction is described by a potential V (r) which depends only on the relative distance r = rT −
rP , where the subscripts T and P denote the target and the incoming particle, respectively. We
shall confine ourselves in this article to scattering processes in which only short-range central
forces are present. In the presence of such potentials, the particle is not under influence of the
target potential when they are emitted from the source or when they enter the detector. Since
for the two-body problem the motion of the center of mass R can be separated out, the problem
reduces to the scattering of a particle with the reduced mass m1 = m1T + m1P at the potential
V (r). Since the potential does not depend explictly on the coordinate of the center of mass R,
the two-particle wave function can be expressed in terms of a product of single-particle wave
functions Ψ(rP , rT ) = φ(R)ψ(r),3 both solutions of two separate Schrödinger equations. The
elementary two-body scattering process could be intrinsically elastic, but recoil of the target
particle might lead to a transfer of energy to the target. In the present context elastic scattering
specifically excludes such effects. Considering a solid as target, depending on the energy of
the projectile and the interaction of the constituent atoms in a solid, this can be a very good
assumption, as for favorable circumstances all atoms contribute to the scattering mass of the
solid (of course, in other chapters it becomes clear that atoms in a solid vibrate and a scattering
event may cause inelastic excitations of phonons in the vicinity of the elastic energy). To think
that the target particle or a target solid is infinitely heavy relative to the mass of the incident
particle simplifies our thinking further. In this case the center of mass of the target remains
stationary at the position of the target particle throughout the scattering process. Under these
circumstances the relative coordinate r represents the actual laboratory coordinate of the light
particle and the mass m is then the mass of the incident particle. These parameters enter the
time-dependent Schrödinger equation
~2 2
 
∂ψ
i~ = − ∇ + V (r) ψ with V (r) = 0 except r ∈ target region T (5)
∂t 2m
to be solved.
The time-dependent Schrödinger equation seems a natural starting point for the description
of a scattering event as it is not a stationary process but involves individual discrete particles
as projectiles, but as we see a bit later the good news is, that under reasonable assumptions
that are fulfilled in typical experimental situations, the same results are obtained using a time-
independent description applying the stationary Schrödinger equation.
At the vicinity of the collimator and detector, the solution of the potential-free Schrödinger
equation (5) is analytically known as the free-particle wave packet:
Z
1 3 ikr −i ~k
2
t
ψ(r, t) = d k A(k) ψk (r, t) with ψ k (r, t) = e e 2m . (6)
(2π)3
3
This does not hold if target and projectile are identical particles. Identical articles scattering about angle θ and
π − θ cannot be distinguished.
A2.6 Stefan Blügel

The wave packet is expressed as a superposition of a complete set of stationary-state solutions

ψk (r) of this Schrödinger equation, which are plane waves ψk (r) = eikr . The energy eigenvalue
~2 2
corresponding to the eigenfunction ψk is simply Ek = 2m k . For convenience the target is
placed at the origin of the coordinate system (r = 0). The coefficient A(k) is the probability
amplitude for finding the wave number k, or momentum ~k in the initial state. We assume
that the properties of the source are such that the wave packet is close to monochromatic and
that the amplitude function A(k) peaks about the average momentum ~k = ~k◦ with a spread
in the wave number ∆k that is small compared to k◦ (∆k  k◦ ), related to the opening of
the collimator ∆r by Heisenberg’s uncertainty principle (∆k∆r ' 1). If we finally impose
2
the condition (∆k)
k◦
L  1 (equivalent to the condition ∆r λ◦
 ∆r L
) the packet does not spread
appreciably during the course of the experiment with the set-up of length L, and we can finally
show that a stationary description of the scattering problem is sufficient. Under these conditions
the energy-dependent phase factor of ψk (r, t) can be conveniently be approximated about the
median energy k 2 ' −k◦2 + 2k · k◦ and Eq. (6) will become
Z
~k2
i 2m◦ t 1 3 ik·(r−v◦ t)
~k2
i 2m◦ t
ψ(r, t) = e d k A(k) e = e ψ(r − v◦ t, 0), (7)
(2π)3

i.e. a wave packet ψ(r, t) centered about the origin r = 0 at t = 0 moves at the classical
velocity v◦ = ~k m
◦
and the packet at time t > 0 will have exactly the same shape, but centered
about r = v◦ t. Thus, the initial state limt→−∞ ψ(r, t) and the final state limt→+∞ ψ(r, t) can
be expressed by Eq. (7), but with the coefficients A(k) in the final state having been modified
compared to the ones in the initial state due to the scattering, as we shall discuss immediately.
A solution of Eq. (5) requires, however, the specification of boundary conditions imposed on
the solution that reflect the physical situation in the laboratory as discussed in section 2.1. The
proper boundary condition is a condition on the wave function when the particle and target are
far apart. It can be motivated from Huygens’ principle [5] who proposed that every point which
a luminous disturbance reaches becomes a source of a spherical wave, and the sum of these
secondary waves determines the form of the wave at any subsequent time. For a single target
scatterer we express the wave function

r→∞ 1 ikr
ψk (r) −→ eikr + e fk (θ, φ) ∀k and t > 0 (8)
r
in terms of a superposition for the incoming wave plus an outgoing scattered wave emanating
from the target, removing some of the incoming particles from the incident primary beam.
f (θ, φ), f (r̂) or f (k̂0 ), respectively, denotes the scattering amplitude. This form of the wave
function is motivated by the fact that we expect, after scattering, an outgoing spherical wave,
modified by the scattering amplitude, interfering with the incoming wave; we will later show a
more rigorous justification of this expression. Consistent to the lab schematics in Fig. 1, k◦ kẑ
and the azimuthal and polar scattering angle (θ, φ) are given by the projection of the direction of
the wavevector k̂ of the scattered wave, e.g. into the detector at direction r̂ and the ẑ direction.
If we replace ψk (r, t) in (6) by its asymptotic form given in (8), when the packet is far from
the target, the wave function ψ 0 (r, t) (where we use a prime to denote the wave function after
scattering) breaks up into two terms

ψ 0 (r, t) = ψk (r, t) + ψsc (r, t), with ψsc (r, t) = 0 for t<0 (9)

with the incident beam ψk (r, t) identical to Eq. (7) and the scattered wave ψsc (r, t) according
Basic Scattering Theory A2.7

to Eq. (6) Z
1 1 ~k2
ψsc (r, t) = d3 k fk (r̂)A(k) eikr e−i 2m t . (10)
r (2π)3
If we assume now that the scattering amplitude is slowly varying over the spread of the wave
numbers ∆k, and thus approximate fk (r̂) ' fk◦ (r̂) as well as making use of the approximations

k ' k̂◦ · k and k 2 ' −k◦2 + 2k · k◦ for (∆k)2  k◦2 (11)

(k = |k| = |k◦ +∆k| = [k◦2 +2k◦ ·∆k+(∆k)2 ]1/2 ' k◦ [1+2k◦ ·∆k/k◦2 ]1/2 ' k◦ + k̂◦ ·(k−k◦ ))
in which k is equal to the projection along k◦ we obtain
Z
−i 2m◦ t 1 1 −i 2m◦ t 1
~k2 ~k2
3 ik·(k̂◦ r−v◦ t)
ψsc (r, t) = e fk◦ (r̂) d k A(k) e = e fk (r̂) ψ(k̂◦ r−v◦ t, 0) .
r (2π)3 r ◦
(12)
Thus, after the incident packet has passed the target a spherical scattered wave shell of thickness
∆r, equal to the size of the packet, centered on the origin and having a radius r = v◦ t emerges
from the target. One finds further that the incoming wave packet given in (7) and the scattered
wave, Eq. (12), share absolutely the same time dependence and are the same Rfor all k. The
solution is actually a superposition of all available wave numbers according to d3 kA(k) . . . .
Since there is no mode-mode coupling such as k → k1 + k2 , it is totally sufficient to solve
the problem in terms of a scattering problem of ψk (r) on the basis of a stationary Schrödinger
equation for all relevant wave vectors k, which will be pursued during the rest of the manuscript,
and keep thereby in mind that a wave packet is formed with a certain probability amplitude. This
stationary problem with plane waves as incident beam simplifies the description of scattering
significantly.

2.3 Coherence
The formation of a wave packet bears, however, a consequence on which we shall briefly touch
upon: The scattering pattern or diffraction pattern, respectively, will be a superposition of pat-
terns for different incident wave vectors (k, k + ∆k) and the question arises, which information
is lost due to these non-ideal conditions. This “instrumental resolution” is intimately connected
with the “coherence” of the beam and the size of the scattering volume in comparison to the tar-
get volume. Coherence is needed, so that the interference pattern is not significantly destroyed.
Coherence requires a phase relation between the different components of the beam. A measure
for the coherence length l is given by the distance, at which two components of the beam be-
come fully out of phase, i.e. when one wave train at position r exhibits a maximum, meats a
wave train exhibiting a minimum, thus experiencing a phase difference of λ/2. If the coherence
length lk is determined by the wavelength spread, λ and λ + ∆λ one refers to the temporal or
longitudinal coherence. The condition lk = nλ = (n − 21 )(λ + ∆λ) translates then into

1 λ2
lk = for longitudinal coherence
2 ∆λ

and

1 λ
l⊥ = for transversal coherence.
2 ∆θ
A2.8 Stefan Blügel

Analogously one obtains the transversal coherence length l⊥ shown in above equation due to
the divergence of the beam ∆θ that results from the finite transverse beam size due to the
finite extension of the source. In many instruments, the vertical and horizontal collimations are
different and the vertical one can even be different along different spatial directions.
Together, the longitudinal and the two transversal coherence lengths define a coherence volume.
This is a measure for a volume within the sample, in which the amplitudes of all scattered
waves superimpose to produce an interference pattern. Normally, the coherence volume is
significantly smaller than the sample size, typically a few 100 Å for neutron scattering, up to µm
for synchrotron radiation. Scattering between different coherence volumes within the sample
is no longer coherent, i.e. instead of the amplitudes, the intensities of the contributions to the
scattering pattern have to be added. This limits the spatial resolution of a scattering experiment
to the extension of the coherence volume.

2.4 The Cross Section

dσ
A general measure of the scattered intensity I(Ω) is the differential cross section ( dΩ ). It is
defined by the number of particles dN counted per unit time dt scattered into a cone of solid
angle dΩ = sin θ dθ dφ in the detector located at the distance r along a ray specified by the
direction r̂, at angle (θ, φ) or the solid angle Ω, respectively, normalized to the current of the
incoming particles jin
   
1 dN dσ dσ 1
= dΩ = dA. (13)
jin dt dΩ dΩ r2
dσ describes a cross-sectional area with a surface normal parallel to k◦ , through which the
number of particles dN that get scattered into the angle Ω flow per unit time. The total cross
section
Z 4π  
dσ
σtot = dΩ (14)
0 dΩ
is the total effective geometrical cross-sectional area of the incident beam that is intercepted and
the particles therein deflected by the target object.
From Eq. (13) the scattered current density is jsc = r12 jin ( dΩdσ
). On the other hand jsc can be
calculated directly employing the expression of the probability current density given by
 
~ ∗ ∗ 1 1
jsc (r) = −i [ψsc (r)∇ψsc (r) − ψsc (r)∇ψsc (r))] ' jin 2 |f (Ω)|2 r̂ + O ... (15)
2m r r3

where as ψsc = 1r eikr f (Ω) is the asymptotic scattering wave, Eq. (8). We explicitly inserted
here the current density jin = ~k
m
to the incoming plane wave ψin . Equating the two expressions
gives the relation
 
dσ
I(Ω) ∝ = |f (Ω)|2 (16)
dΩ
for the differential cross section. This expression relates the experimental quantity, the differ-
ential cross section, to the scattering amplitude, which characterizes the wave function at large
distances from the target. It is the fundamental relation between scattering theory and scattering
experiments.
Basic Scattering Theory A2.9

3 Lippmann Schwinger Equation

Having established the basic concepts for the scattering problem, we turn now to the illustration
of the physical ideas that underlie the scattering analysis using integral equation methods. We
recall that we are looking for the solution of the stationary Schrödinger equation

~2 2
 
− ∇ + V (r) ψk (r) = Eψk (r) with V (r) = 0 except r ∈ target region T,
2m
(17)
ikr
that is consistent with the boundary condition (8) of an incident plane wave ψk (r) = e and
an emanating scattered wave. The energy E is determined by the energy of the incident plane
~2 2
wave Ek = 2m k . By introducing the Green function G◦ ,

~2 2
 
∇ + E G◦ (r, r0 |E) = δ(r − r0 ), (18)
2m

for the potential-free Schrödinger equation, the Schrödinger equation for ψk (r),
 2 
~ 2
∇ + E ψk (r) = V (r) ψk (r), (19)
2m

can be transformed into an integral equation

Z
ψk (r) = ψk (r) + d3 r0 G◦ (r, r0 |E) V (r0 ) ψk0 (r0 ),
0
(20)
T

in which the formal expression V (r) ψk0 (r) is conceived as inhomogeneity of the differential
equation (18). This integral equation is called the Lippmann-Schwinger equation. Hereby,
ψk (r) is the above cited plane-wave solution of the potential-free Schrödinger equation. The
index k in ψ 0 expresses the fact that this state has evolved from one that in the remote past was a
plane wave of the particular wavevector k. Obviously, in the limit of zero potential, V (r) → 0,
the scattered and the incident wave are identical, ψk0 (r) = ψk (r).
The Green function G◦ (r, r0 |E) is not uniquely determined by the Schrödinger equation (18).
Also here the unique solution requires a boundary condition, which is chosen such, that the
solution ψk0 (r) describes outgoing scattered waves. The Green function G◦ (r, r0 |E),
0
r
0 2m 1 eik|r−r | 2m
G◦ (r, r |E) = − 2 0
with k= E, (21)
~ 4π |r − r | ~2

describes then the stationary radiation of a particle of energy E, that is generated at r0 , by a

spherical wave outgoing from the target. In other words, the Green function G◦ (r, r0 |E) gives
the amplitude of this wave at location r due to its generation by the source at r0 , under the
condition that the wave is not further scattered during its propagation from r0 to r. By the
Lippmann-Schwinger equation, the incident wave ψk (r) is superimposed with spherical waves
emitted from scattering at position r0 in the target. The amplitude of these scattered waves is
proportional to the interaction potential V (r0 ) and the amplitude of the total wave field ψ 0 (r0 ) at
that point.
Recalling our experimental set-up that the distance between target and detector is significantly
larger than the size of the sample, for large distances between r and the scattering center r0
A2.10 Stefan Blügel

0
it is useful to expand the Green function G◦ in powers of rr  1 assuming that the extent
of r0 is restricted to the space of a small target or scattering volume, respectively, r0 ∈ T.
Approximating for r0  r
 
1 1 1 r
0
= +O 2 and |r − r0 | ≈ r − r̂ · r0 with r̂ = (22)
|r − r | r r r

and inserting this into the relation (21) one obtains the asymptotic form, or far-field limit, re-
spectively, of the Green function G◦ ,

2m 1 eikr −ikr̂·r0
 
0 1
G◦ (r, r |E) = − 2 e +O . (23)
~ 4π r r2

Inserting this expression into the Lippmann-Schwinger equation (20) one obtains the asymptotic
solution of the wave function ψk0 (r) for large distances r

1 ikr
ψk0 (r) ' eikr + e fk (r̂), (24)
r
which is exactly the boundary condition (8) we conjectured from Huygens’ principle, whereas
the scattering amplitude f (r̂) = f (θ, φ) is given by the integral,

~2
Z
2m 1 0 0
fk (r̂) = − 2 d3 r0 e−ik r V (r0 ) ψk0 (r0 ) = −4π T (k0 , k) (25)
~ 4π 2m

that can be interpreted as a transition-matrix element from the scattering state described by
ψk0 (r0 ) to the scattered state at far distances, which is a plane-wave state described by k0 =
k · r̂, the wave vector of the scattered wave in the direction of the detector, which is known
in the experiment. T (k0 , k) is referred to as the T matrix or transition amplitude, a quantity
proportional to the scattering amplitude. Due to the far-field approximation (22) the scattering
pattern fk (r̂) is independent of the distance between target and detector, depending only on the
angles to the detector from the target. In optics this is known as the Fraunhofer diffraction and
in this context approximation (23) is also referred to as the Fraunhofer approximation of the
Green function.

Detector
r − r0

k0 kr
r0

k scattering volume

Fig. 3: Scattering geometry for the calculation of the far-field limit at the detector. In the
Fraunhofer approximation, we assume that |r|  |r0 |.
Basic Scattering Theory A2.11

4 Born Approximation
Note that in the Lippmann-Schwinger equation (20) the wave function ψ 0 (k) appears both on
the left and right hand side. In a general case, there is no simple way to find exact solutions of
the Lippmann-Schwinger equation. The form of the Lippmann-Schwinger equation provides a
natural but approximate means that can be used for any potential, under the proper conditions,
to proceed by an iterative procedure. At zeroth order in V , the scattering wave function is
specified by the unperturbed incident plane wave,
0(0)
ψk (r) = eikr . (26)

Then one can iterate the Lippmann-Schwinger equation (20) according to the rule
Z
0(n+1) 0(n)
ψk (r) = e + d3 r0 G◦ (r, r0 |E) V (r0 ) ψk (r0 )
ikr
(27)

that results in the Born expansion of the wave function in powers of the interaction potential V
written here in a symbolic form4
0(0) (1) (2) (3)
ψk0 = ψk + ψk + ψk + ψk + · · · + (28)
0(0) 0(0) 0(0) 0(0)
= ψk + G◦ V ψk + G◦ V G◦ V ψk + ··· +
+ G◦ V G◦ V G◦ V ψk (29)
0(0) 1
= (1 + G◦ T ) ψk with T = V + G◦ V + · · · + = (30)
1 − V G◦

k k k

+ + + ...
0 0 0
k k k

A term-by-term convergence of this series is in general not guaranteed and depends on the
potential and the energy of the incident particle, even though the final expression is always
valid. Physically, an incoming particle undergoes a sequence of multiple scattering
events from the potential. The first term in the series expansion (29) de-
scribes single scattering processes of the incident wave, while the follow-
ing terms describe then scattering processes of successively higher order. k0
k
Rarely are higher-order terms calculated analytically, since the complica-
tions then become so great that one might as well use a numerical method
to obtain the exact solution if this is possible at all. Thus, only the first
iteration of the series is taken into account, i.e. only single scattering, and
the T matrix is approximated by the potential matrix V (k0 , k),

T (k0 , k) ' V (k0 , k). (31)

This first order term, in which the exact wave function ψk0 (r0 ) in the integral kernel is replaced
0
by the plane wave eikr is the first Born approximation and typically abbreviated as the Born
approximation.5 This approximation is most useful when calculating the scattering amplitude.
(n) 0(0) 0(n)
4
Please note that ψk = (G◦ V )n ψk . This is different from definition ψk (r) in (27).
5
It should not be confused with the Born-Oppenheimer approximation.
A2.12 Stefan Blügel

In first Born approximation the general equation for the scattering amplitude (25) reads then
Z
(1) 2m 1 0 0 0 2m 1
fk (r̂) = − 2 d3 r0 e−ik r V (r0 ) eikr = − 2 V (Q) with Q = k − k0 , (32)
~ 4π ~ 4π
with V (Q) denoting the Fourier transform of the potential with the momentum transfer Q.6
V (Q) can be interpreted as a transition-matrix describing the transition from the incoming
plane-wave of state k into the outgoing plane-wave state k0 due to the action of the potential
expressed in the reciprocal space at scattering angle Q. From (16) follows then the differential
cross section
(1) 
 2
dσ 2m π
= |V (Q)|2 . (33)
dΩ ~2 2
The physics behind the 1st Born approximation is provided by the assumption that the incoming
wave scatters only once inside the target potential before forming the scattered wave ψ (1) . This
is the concept behind the kinematic theory of scattering, that simplifies the interpretation of
the scattering experiment substantially. For example, for the case of elastic scattering that we
assumed all the time during the derivations, energy is conserved |k|2 = |k0 |2 , all possible
scattering vectors are placed on the so-called Ewald sphere with radius |k|. The length of
the scattering vector Q is then given by
1 4π 1
Q(Ω) = |Q| = 2 |k| sin Ω = sin Ω with Ω = (θ, φ)^(k0 , k). (34)
2 λ 2

Note, and this is the essence of the Ewald-sphere, that this shows
Ewald sphere
that the differential cross section (33) does not depend on scat- 0
tering angle and beam energy independently, but on a single pa- k0 Q |k| = |k |
rameter through the combination Q = 2k sin 12 Ω. By using a
range of energies, k, for the incoming particles, this dependence θ k
can be used to test whether experimental data can be well de-
scribed by the Born approximation. A very common use of the
dσ Fig. 4: The Ewald sphere.
Born approximation is, of course, in reverse. Having found dΩ ,
experimentally, a reverse Fourier transform can be used to obtain the form of the potential.

4.1 Example of Born Approximation: Central Potential

For a centrally symmetry potential, V (r) = V (r), we can make some progress with the matrix
element integral (32) if we choose a polar coordinate system with Q along the z-axis, so that
Q · r = Qr cos θ. Then, the scattering amplitude in Born approximation f (1) (θ) is written after
some manipulations in the form
2m 1 ∞
Z
(1)
fk (θ) = − 2 V (r) r sin Qr dr (35)
~ Q 0
and is seen to be independent of φ due to the cylindrical symmetry of the problem at hand and
all scattering vectors are placed on an Ewald-circle. An example, is the Rutherford scattering
or Coulomb scattering, respectively, where a charged particle with charge Z1 e impinges on an
6
It would be mathematically more correct to denote the Fourier transformation F of V (r) by a different function
name e.g. Ve (Q) = F[V (r)]. To avoid incomprehension of reading due to unduly complicated notation we replace
Ve (Q) by V (Q).
Basic Scattering Theory A2.13

other charged particle with charge Z2 e under the action of a Coulomb potential, which results
into the scattering amplitude

(1) 2m Z1 Z2 2 1 Z1 Z2 2 1
fk (θ) = − e =− e (36)
2
~ Q 2 4 sin2 12 θ E

and the differential cross section,

 (1) 2
dσ 1 Z1 Z2 2 1
= e (37)
dθ 4 sin2 21 θ E2

known as the Rutherford formula. Due to the long-range nature of the Coulomb scattering
potential, the boundary condition on the scattering wave function does not apply. We can, how-
ever, address the problem by working with the screened (Yukawa) potential, V (r) = Z1rZ2 e−κr ,
1
leading to f (1) ∝ Q2 +κ2 and taking κ → 0, which leads then to the Rutherford formula (37). Ac-

cidentally, the first Born approximation gives the correct result of the differential cross section
for the Coulomb potential.

4.2 Example of Born Approximation: Square Well Potential

Consider scattering of particles interacting via a spherical three dimensional (3D) square well
potential V (r) = V◦ for r ≤ R◦ and zero outside (V (r) = 0 for r > R◦ ). The integral (35) for
the scattering amplitude required here is then
R◦ R
sin Qr − Qr cos Qr ◦
Z 
(1) 2m 1 2m 1
fk (θ) = 2 V◦ r sin Qr dr = 2 V◦ (38)
~ Q 0 ~ Q Q2 0

7
and whence to the differential cross section
 (1)  2 2 ( 1
1 − 15 Q2 R◦2


dσ 2m V◦ 2m V ◦ kR◦ < 1for low E,
= R◦2 j12 (QR◦ ) ' 9
R◦2 .
dθ ~2 Q ~2 Q Q2
kR◦ > 1for high E,
(39)
From integrating over θ and φ the low and high energy limits for the total cross section are
2 2
V◦ R◦3
  
2m 8 2 2 2 4 4
σ(E → ∞) = π σ(E → 0) = σ(E → ∞) k R◦ − k R◦ + · · · .
~2 kR◦ 9 5
(40)
The two examples illustrate some general features of scattering in the Born approximation:
(i) Born approximation is based on perturbation theory, so it works best for high energy parti-
cles.
(ii) At high energy, the scattering amplitude and the cross section are inversely proportional
to the energy (E = ~2 k 2 /2m). E.g. both become smaller and the scattering weaker with
increasing energy. This is a general phenomenon, if no bound states appear in the vicinity
of the energy. This can be seen best by inspecting the Fourier transformed Green function
2 2
G◦ (k|E) ∝ 1/(E − h2mk ) that is inverse proportional to the energy.
(iii) Scattering depends on square of the interaction potential, e.g. V◦2 , so both attractive and
7
j0 (Qr) = sin Qr/Qr is the spherical Bessel function for angular momentum ` = 0. Radial integration leads
to Bessel function j1 (Qr).
A2.14 Stefan Blügel

repulsive potentials behave the same.

(iv) The dependence on the energy of the incident beam k and scattering angle θ arises only
through the combination Q = 2k sin 2θ . Thus as energy increases, the scattering angle θ is re-
duced and the scattered beam becomes more peaked in the forward direction.
(v) Angular dependence depends on the range of the potential R◦ but not on the strength V◦ .
(vi) The total cross section depends on both range R◦ and depth V◦ of the potential.

4.3 Validity of first Born Approximation

This raises the practical questions (i) under which conditions the Born expansion converges and
(ii) whether the first term is a good approximation. In the Born approximation the T matrix is
approximated by the potential matrix V . This will not work if the denominator |1 − V G◦ | in
(30) is small or zero. This is the situation at low energy, when the energy of the incoming beam
coincides with bound states of the potential. Then, the Born approximation is invalid and the
Born expansion will not converge. The solution to this problem is provided by the dynamical
scattering theory discussed in Chapter A3. According to (30) the Born approximation T ' V
is equivalent to the condition
ZZ
d3 r d3 r0 V (r)G◦ (r, r0 |E)V (r0 )  1 . (41)
T

At the same time this condition determines the radius of convergence of the Born series with
respect to the strength of the potential. This condition means that the first Born approximation
is valid and the Born series converges if the potential is sufficiently weak and the approximation
improves as the energy is increased. Concerning the question whether the first term is itself a
good approximation to the wave function, a convenient, although nonrigorous, criterion can be
obtained by requiring that the first-order correction to the wave function be small compared to
0(1) 0(0)
the incident wave in the region of the potential, i.e. |ψk (r)|  |ψk (r)| which results to
Z
2m 1 0 0

2
d3 r0 e−ikr V (r0 ) e−ikr  1. (42)
~ 4π T
For the above introduced spherical 3D square well potential V (r ≤ R◦ ) = V◦ and V (r > R◦ ) =
0, this implies
mV◦ ikR◦

e sin kR ◦ − kR ◦  1. (43)
~2 k 2
or
m
|V◦ |R◦2  1 for low energies kR◦ < 1 (44)
~2
m 1
2
|V◦ |R◦  1 for high energies kR◦ > 1. (45)
~ k
Since a bound state for this potential exists when ~m2 |V◦ |R◦2 & 1, as said above, the Born ap-
proximation will not be valid at low energies if the potential is so strong that it has a bound
state. On the other hand criterion (45) can be satisfied for any potential by going to sufficiently
high energy. When we square criterion (45) and multiply it by the geometrical cross section
σgeo = πR◦2 , criterion (45) reads
2  2
V◦ R◦3

2m
π 2
 πR◦2 ⇐⇒ σtot  σgeo . (46)
~ kR◦
Basic Scattering Theory A2.15

and provides a hand-waving criterion when the potential is sufficiently weak so that the Born
approximation gives reliable results: If the ratio of the scattering cross section and the geomet-
rical extension of the potential is small, u := σσgeo
tot
 1, the Born approximation can be used.
For x-ray and neutron scattering, the scattering cross sections amount to a few 10−24 cm2 , the
cross-sectional area per atom is of the order of several 10−16 cm2 . This results indeed in a very
small potential strength of u ∼ 10−8 ÷ 10−7 for scattering on different atoms: that means, the
Born approximation is justified and the easy-to-interpret kinematic interpretation of scattering
results is sufficient. The arguments become invalid for the nuclear scattering of neutrons by
individual nuclei as the cross-sectional area of a nucleus is eight orders of magnitude smaller
and the scattering cross section and the geometrical gross section can be of similar size and the
potential strength u can be even larger than 1, u > 1. Due to the strong Coulomb interaction
potential, the probability for multiple scattering processes of electrons in solids is extremely
high, making the interpretation of electron diffraction experiments very difficult. Although in
neutron and x-ray scattering, the first Born approximation is almost always adequate, even for
neutrons and x-rays, the kinematic scattering theory can break down, for example in the case of
Bragg scattering from large nearly perfect single crystals. In this case as in the case of electron
scattering the wave equation has to be solved exactly under the boundary conditions given by
the crystal geometry. This is then called the dynamic scattering theory discussed in Chapter A3.
For simple geometries, analytical solutions can be obtained. Other examples where the Born
series do not converge are neutron optical phenomena like internal total reflection in a neutron
guide, or grazing-incidence small-angle neutron scattering (GISANS). The same holds for x-
ray scattering for example in combination with grazing-incidence small-angle x-ray scattering
(GISAXS) experiments. The grazing-incidence small-angle scattering (GISAS) techniques and
their application will be discussed in Chapter D2. The theoretical analysis makes use of the
distorted-wave Born approximation (DWBA).

4.4 Distorted-Wave Born Approximation (DWBA)

In the previous Section we discussed that the Born approximation is accurate if the scattered
field is small, compared to the incident field, in the scatterer. The scatterer is treated as a pertur-
bation to free space or to a homogeneous medium, and the incident wave is a plane wave. When
this smallness criteria is not met, it is often possible to generalize the idea of the Born approxi-
mation, which is frequently referred to as the distorted-wave Born approximation (DWBA). In
generalization to the Born approximation, the free space zero-potential, V◦ (r) = 0, is replaced
by a non-trivial reference potential V1 (r) to which the scattered field ψk01 is known analytically,
numerically, e.g. due to the solution of the Lippmann-Schwinger equation (20),
Z
ψk (r) = e + d3 r0 G◦ (r, r0 |E) V1 (r0 ) ψk01 (r0 ),
01 ikr
(47)
T

or experimentally. The interaction of interest V

V (r) = V1 (r) + δV (r) with |δV |  |V1 | (48)

is treated as a perturbation δV to the reference system V1 . In the distorted-wave Born approx-

imation, the scattering field ψk0 (r) due to the potential V is then determined applying the Born
approximation Z
ψk0 (r) = ψk01 (r) + d3 r0 G1 (r, r0 |E) δV (r0 ) ψk01 (r0 ) (49)
A2.16 Stefan Blügel

to the description of the scattering of the incident wave ψk01 (r), the so-called “distorted” wave,
due the perturbative potential δV (r). The “distorted” incident wave, is the outgoing-wave solu-
tion of  2 
~
∇ − V1 (r) + E ψk01 (r) = 0,
2
(50)
2m
that is supposed to be known, and G1 (r, r0 |E) is the corresponding Green function with the
outgoing boundary condition for the same potential,
 2 
~
∇ − V1 (r) + E G1 (r, r0 |E) = δ(r − r0 ).
2
(51)
2m
In analogy to the potential-free case (19), the difference to the reference system that appears in
the Schrödinger equation, δV (r)ψk01 (r), can be considered as inhomogeneity that constitutes a
Lippmann-Schwinger equation with ψk01 (r) as homogeneous solution. The Born approximation
to this equation is given by Eq. (49).
To satisfy the boundary conditions we must also require that the “distorted” wave function
behaves in the asymptotic limit as plane wave plus an outgoing wave
r→∞ 1 ikr 1
ψk01 (r) −→ eikr + e fk (θ, φ), (52)
r
where, as in (25) Z
2m 1 0 0
fk1 (θ, φ)
=− 2 d3 r0 e−ik r V1 (r0 ) ψk01 (r0 ). (53)
~ 4π
This is simply the scattering amplitude for the potential V1 (r), as if it were the only potential
present, assumed to be known. The total scattering amplitude fk (θ, φ) is

fk (θ, φ) = fk1 (θ, φ) + δfk (θ, φ) (54)

where δfk (θ, φ) is calculated in the Born approximation (ψk0 (r) ' ψk01 (r))
Z
2m 1 01(−)∗
δfk (θ, φ) ' − 2 d3 r0 ψk0 (r0 )δV (r0 ) ψk01 (r0 ). (55)
~ 4π
The scattering amplitude describes the scattering strength of an outgoing spherical wave. By
inspection of Eq. (53) one finds that the first wave function of the integrand is a plane wave
0 0
e−ik r , whose negative sign in the exponent represents an incoming plane wave. According of
0 0 (−)∗
the standard definition of plane waves we can write e−ik r = ψk , where (−) denotes the
01(−)∗
incoming boundary condition. Quite in the same way ψk0 (r0 ) is the known incoming wave
function corresponding to the reference potential V1 .
Clearly Eq. (55) will be a good approximation if δV (r) is sufficiently small, so that the ad-
ditional scattering that is generated does not significantly modify the wave function. Some
example in which this method is useful include scattering in which δV (r) may be the spin-orbit
interaction or a perturbation due to many-particle excitations, atomic scattering where δV (r)
may be a deviation from the Coulomb potential or from a Hartree average potential, or in case
of scattering at a magnetic superlattice where V1 (r) contains the scattering at the nuclei or elec-
tron charge distribution plus the interaction to an average magnetization, and δV (r) describes
the interaction to the modulated magnetic structure of the superlattice. The DWBA is at place
analyzing grazing-incidence small-angle scattering (GISAS) experiments to resolve the mag-
netic structure of superlattices [6].
Basic Scattering Theory A2.17

5 Method of Partial Wave Expansion

The differential equation formulation of scattering provides additional insights that are not read-
ily apparent from the integral equation discussed in the previous section. Many potentials in
nature are spherically symmetric, or nearly so, and thus for simplicity, here we will focus on the
properties of a centrally symmetric potential, V (r), where the scattering wave function, ψ 0 (r)
(and indeed that scattering amplitudes, f (θ)) must be symmetrical about the axis of incidence,
and hence independent of the azimuthal angle, φ. The method of partial wave expansion is
inspired by the observation that a plane wave ψk = eikr can actually be written as a sum over
spherical waves
∞
X
ψk = eikr = eikr cos θ = (2` + 1) i` j` (kr)P` (cos θ) , (56)
`=0

known as the Rayleigh expansion. As we shall discuss in more detail below, the real function
is a standing wave, made up of incoming and outgoing waves of equal amplitude. The radial
functions j` (kr) appearing in the above expansion of a plane wave in its spherical components
are the spherical Bessel functions, discussed below.
Generalizing this concept, if we define the direction of the incident wave k to lie along the
z-axis, and θ denotes the scattering angle to the detector, θ = ^(k, r), then the azimuthal rota-
tional symmetry of plane waves and the spherical potential around the direction of the ingoing
wave ensures that the wave function can be expanded in a series
∞
X
ψ(r) = ψ(r, θ) = (2` + 1)i` R` (r) P` (cos θ) (57)
`=0
q
4π
of Legendre polynomials P` (cos θ) = 2`+1 Y`0 (θ), where Y`m denotes the spherical harmon-
ics. Each term in the series is known as a partial wave, and is a simultaneous eigenfunction of
the angular momentum operators L2 and Lz having eigenvalue ~2 `(` + 1), and 0, respectively.
Following standard spectroscopic notation, ` = 0, 1, 2, · · · are referred to as s, p, d, · · · waves.
The partial wave amplitudes, f` are determined by the radial functions, R` (r), defined by
 2 
d 2 d `(` + 1) 2
+ − − v(r) + k R` (r, E) = 0 with V (r) = 0 except r ∈ RT,
dr2 r dr r2
(58)
2m 2
where v(r) = ~2 V (r) represents the effective potential and k refers to the energy of the incom-
ing beam k 2 = 2m E . The energy Ek can be chosen positive and equal to the kinetic energy of
~2 k
the projectile when it is far from the scattering center. The potential V (r) or v(r), respectively,
will be assumed to vanish sufficiently rapidly with increasing r that it may be neglected beyond
some finite radius, that defines the radial target region RT or scattering volume, respectively.
We are looking for the solution of the stationary Schrödinger equation that is consistent with
the boundary condition (8) of an incident plane wave ψk (r) = eikr and an emanating spherical
scattered wave. Beyond the range of the potential, i.e. r outside the radial target region RT,
the R` (r, E) may be expressed in terms of the solutions of the potential free radial differential
equation  2 
d 2 d `(` + 1) 2
+ − + k R` (r, E) = 0. (59)
dr2 r dr r2
A2.18 Stefan Blügel

This is a differential equation of 2nd order which has two linearly independent solutions at each
energy E, known as the spherical Bessel function

rk1 z` rk1 1
 π
R` (r, E) = j` (kr) with j` (kr) −→ and j` (kr) −→ sin kr − `
(2l + 1)!! kr 2
(60)
and the spherical Neumann function
rk1 (2` − 1)!! rk1 1 π

R` (r, E) = n` (kr) with n` (kr) −→ and n` (kr) −→ cos kr − ` ,
z `+1 kr 2
(61)
whereas j` and n` show a regular and irregular solutions, respectively, in the origin r = 0
and n!! = n(n − 2)(n − 4) · · · 1. That means any solution R` (r) of the radial Schrödinger
equation (59) can be expressed at a given energy E for r outside RT as linear combination of
j` and n` or in the form spherical Hankel functions
(±) rk1 1 ±i(kr−` π2 )
h` (kr) = n` (kr) ± i j` (kr) −→ e , (62)
kr
a different set of independent solutions that correspond to incident (−) and emanating (+) radial
waves at large distances r. This holds also for the wave function of the incident beam before
scattering expressed in terms of a plane wave
∞ ∞
X i X 
(−) (+)

ψk = eikr = (2`+1) i` j` (kr)P` (cos θ) = (2`+1) i` h` (kr) − h` (kr) P` (cos θ),
`=0
2 `=0
(63)
that can be recast according to Rayleigh into incoming and outgoing spherical Hankel functions.
(−)
After scattering, the incoming spherical wave h` is unaffected by the scattering process, while
(+)
the outgoing wave h` is modified by a herewith introduced quantity,

S` (k) or S` (E) = ei2δ` (E) , (64)

the partial wave scattering matrix, subject to the constraint |S` (k)| = 1 following from the
conservation of particle flux (current density times area). δ` (E) is the phase shift (the name
becomes clear below as the phase difference between incoming and outgoing wave). For scat-
tering processes where the net flux of particles is zero, the phase shift is real, and thus only the
phase and not the amplitude of the outgoing spherical wave is affected but the presence of the
potential. The wave after scattering ψ 0 (r) reads then
∞
iX 
(−) (+)

ψk0 (r) = ψk0 (r, cos θ) = (2` + 1)i` h` (kr) − S` (k) h` (kr) P` (cos θ) (65)
2 `=0
X ∞  
(+)
= (2` + 1)i` j` (kr) + T` (k) h` (kr) P` (cos θ) r ∈
/ RT.(66)
`=0

The first term in the parenthesis proportional to j` sums up according to the Rayleigh expan-
sion (56) to the incoming plane wave, the second describes the outgoing spherical wave multi-
plied by a partial wave scattering amplitude f` (k) or the partial wave transition matrix element
T` (k)
1 1
T` (k) = (S` (k) − 1) = eiδ` (k) sin δ` (k) = = kf` (k) (67)
2i cot δ` − i
Basic Scattering Theory A2.19

due to the presence of the interaction potential. The wave function after scattering takes the
asymptotic form
∞
1 π 1
(2` + 1)i` T` (k) ei(kr−` 2 ) P` (cos θ) = eikr + eikr fk (θ).
X
ψk0 (r) ' eikr + (68)
`=0
kr r

consistent with the scattering boundary condition (8) where the scattering amplitude fk (θ) can
be related to the partial wave scattering amplitude and the phase shift as
∞
X
fk (θ) = (2` + 1)f` (k)P` (cos θ) . (69)
`=0

4π
R
Making use of the identity dΩP` (cos θ)P`0 (cos θ) = 2`+1 δ``0 and the definition of the total
cross section (14) one obtains
Z ∞ ∞ ∞
2
X 4π X 2 4π X
σtot (k) = |fk (θ)| dΩ = σ` (k) = 2 (2` + 1)|T` (k)| = 2 (2` + 1) sin2 δ` (k) .
`=0
k `=0
k `=0
(70)
The total cross section is additive in the contribution of the σ` (k) of each partial wave. This
does not mean, though, that the differential cross-section for scattering into a given solid angle
is a sum over separate ` values, no the different components interfere. It is only when all
angles are integrated over, that the orthogonality of the Legendre polynomials guarantees that
the cross-terms vanish.
Notice that the scattering cross-section for particles in angular momentum state ` is upper
bounded by
4π
σ` (k) ≤ 2 (2` + 1) , (71)
k
which is four times the classical cross section for that partial wave impinging on, e.g. a hard
sphere: Imagine semi-classically particles in an annular area, with the angular momentum L =
rp, but L = ~` and p = ~k so ` = rk. Therefore, the annular area corresponding to angular
momentum between ` and ` + 1 has inner and outer radii, `/k and (` + 1)/k, respectively, and
therefore the area kπ2 (2` + 1). The quantum result is essentially a diffractive effect.
The maximal contribution is obtained for the phase shifts δ` (k) = (n + 21 )π , with n =
0, ±1, ±2, · · · . For these energies Ek , resonant scattering occurs if in addition δ` (k) changes
rapidly. On the other hand, for energies leading to phase shifts δ` (k) = nπ with n = 0, ±1, ±2, · · · ,
the scattering amplitude and the cross section vanish.
Since for the imaginary part of the partial wave scattering amplitude (67) holds
1 2 1
=f` (k) = sin δ` (k) = k|f` (k)|2 or more simply = = −k (72)
k f` (k)

and the Legendre polynomial at unity are always unity, P` (1) = 1 for ∀`, and apply this to
equation (69) we find that
k
=fk (0) = σtot (k) , (73)
4π
a relation known as the optical theorem. It is a direct consequence of the flux conversation
in elastic scattering and says for example that the scattering amplitudes are complex valued
quantities.
A2.20 Stefan Blügel

Comparing equation (57) with equation (66) and replacing the definition of the partial wave
transition matrix T` by the phase shift given in (67) we can write the radial wave function
R` (r, E) after scattering outside the target region, r ∈
/ RT, as
(+)
R` (r, E) = j` (kr) + h` (kr) eiδ` (k) sin δ` (k) for r ∈ / RT (74)
iδ` (k)
= e (cos δ` (k)j` (kr) + sin δ` (k)n` (kr)) (75)
1   π   π 
' eiδ` (k) cos δ` (k) sin kr − ` + sin δ` (k) cos kr − ` (76)
kr 2 2
1  π 
' eiδ` (k) sin kr − ` + δ` (k) for kr  1 . (77)
kr 2
1
sin kr − ` π2 and the scattered


In the asymptotic limit, the radial incoming wave j` (kr) ' kr
wave differ by just a phase δ` (k) known as the scattering phase, which gives δ` (k) the name
phase shift, as well as a phase phase factor eiδ` (k) .
I would like to end this section with remarking that the scattering and transition matrices S,
T , respectively, describe the scattering at different boundary condition. The scattering matrix
describes the scattering from the incoming spherical wave into an outgoing spherical wave,
while the transition matrix describes scattering from an incoming plane wave into an emanating
spherical wave. The scattering matrix contains all the scattered and the unscattered states and
the matrix elements are unity without scattering. The T matrix contains only the scattered states
and it has only zero valued matrix elements in the absence of scattering.

5.1 The Born Approximation for Partial Waves

From the boundary condition (8) and the solution of the Lippmann-Schwinger equation (see
Section 3) in far-field limit
Z
0 2m 1 0 0
ikr
ψk (r) ' e − 2 d3 r0 e−ik r V (r0 ) ψk0 (r0 ) (78)
~ 4π
we obtained the respective definition of scattering amplitude f (θ) (25). On inserting expression
(56) and (57) for the plane wave and the wave function after scattering, respectively, and inte-
grating over the angle dΩ0 one yields the radial Lippmann-Schwinger equation for the far field
limit, Z
R` (r, k) ' j` (kr) − r02 dr0 j` (kr0 ) v(r0 ) R` (r0 , k) (79)
RT
and an explicit formulation of the partial scattering amplitudes
Z
1 iδ` (k)
f` (θ) = e sin δ` (k) = − r02 dr0 j` (kr0 ) v(r0 ) R` (r0 , k) (80)
k RT

which provides an elegant procedure to calculate the phase shift. We recall that in the first Born
Approximation the exact wave function R` (r0 , k) in the integral kernel is replaced by the plane
wave represented by the Bessel function j` (kr0 ) and the partial-wave Born approximation of the
scattering matrix and the transition matrix, respectively, reads
Z
(1) 1 (1) 1 iδ(1) (k) (1) 1 (1)
f` (θ) = T` (k) = e ` sin δ` (k) = − r02 dr0 j` (kr0 ) v(r0 ) j` (kr0 ) ≈ δ` (k)
k k RT k
(81)
Basic Scattering Theory A2.21

(1)
with an approximate expression for the phase shift in Born approximation δ` (k) valid for small
phase shifts (the only place where the Born approximation is valid).
Concerning the Distorted-Wave Born Approxmiation (DWBA) one can also perform a partial-
wave analysis of Eq. (54) to obtain an approximate expression for the phase shift. This result
is Z
(1)
2
(1) iδ`1 (k)
e iδ` (k)
sin δ` (k) = e 1
sin δ` (k) − r02 dr0 R`1 (kr0 ) v(r0 ) , (82)
RT

where δ`1 (k)

are the phase shifts to the nontrivial reference potential V1 , and the Bessel function
representing the plane waves are replaced by the exact radial scattering solution R`1 (kr).

5.2 Low Energy Scattering: Scattering Phases and Scattering Length

From (81) follows that the sign of δ` (k) is determined by the sign of the potential. For an
attractive potential, the phase shift δ` > 0 is positive and the phase shift is negative, δ` < 0,

1 π
for a repulsive potential. At large
1 π

distances r, the zeros of R` (r, k) ' kr sin kr − ` 2 + δ` (k)
are at r0 = k nπ + ` 2 − δ` (k) . Positive (negative) δ` (k) relate to an inward (outward) shift
of the nodes. That means, for an attractive potential, the probability of a particle to stay in the
potential range becomes greater, so that the wave function is drawn into the potential range: the
nodes shift inwards, i.e. δ` > 0. For a repulsive potential the wave function is squeezed out of
the potential range. In consequence the nodes move to the outside, i.e. δ` > 0.
If kr0  1 or λ  r0 , respectively, we shall be able to approximate the Bessel function j` (kr0 ) '
1
(2`+1)!!
(kr0 )` and one obtains the simple estimate
 2 Z
1
δ` (k) ≈ k 2`+1
dr0 r0 2`+2 v(r) (83)
(2` + 1)!! RT

for the scattering phase. For low energies and high angular momenta the scattering phases δ` (k)
behave proportional to δ` (k) ∝ k 2`+1 . In particular one expects that only s-wave scattering
(` = 0) survives for k → 0 since the cross section scales as
4π
σ` (k) = 2
(2` + 1) sin2 δ` (k) ∝ k 4` . (84)
k
When a slow particle scatters off a short ranged scatterer it cannot resolve the structure of the
object since its de Broglie wavelength λ is very long, larger than the scatterer. The idea is that
then it should not be important what precise potential V (r) one scatters off, but only how the
potential looks at long length scales. At very low energy the incoming particle does not see
any structure, therefore to lowest order one has only a spherical symmetric outgoing wave, the
so called s-wave scattering (angular momentum ` = 0). At higher energies one also needs to
consider p and d-wave (l = 1,2) scattering and so on.
Although exact at all energies, the partial wave method is most useful for dealing with scat-
tering of low energy particles. This is because for slow moving particles to have large angular
momentum (~kb) they must have large impact radii b. Classically, particles with impact radius
larger than the range of the potential miss the potential. Thus, for scattering of slow-moving
particles we need only to consider a few partial waves, all the others are unaffected by the po-
tential (δ` ≈ 0). Thus at a given incoming momentum, ~k, we can determine how many terms
in the partial wave expansion to consider from ~kbmax ≈ `max ~, where bmax is the maximum
impact parameter for classical collision, i.e. the range of the potential RT. Since the angular
A2.22 Stefan Blügel

variation of the Legendre polynomial for the angular momentum ` = 0 is P`=0 (cos θ) = 1, the
s-wave scattering is isotropic, consistent with the thought of averaging over the potential. Since
in practical applications the expansion into Legendre polynomials has to be truncated for higher
` values, since otherwise the effort becomes too large, the partial wave analysis is primarily a
method of approximation for low energies. It generally requires an exact (numerical) solution
of the radial equations, since the Born approximation fails at low energies in general. Thus
partial waves and the Born approximation are complementary methods, good for slow and fast
particles, respectively.
At energy E → 0, the radial Schrödinger equation for rR`=0 away from the potential becomes
d2
dr2
rR`=0 = 0 with a straight line solution rR`=0 = (r − as ).8 For the s-wave solution the ap-
k→0

becomes exact and the radial wave function rR0 (r, k) = sin (kr + δ0 (k)) −→
proximation (77)
k r + k1 δ` (k) can only become a straight line in r, if δ0 (k) is itself linear in k for sufficiently
small k. Then δ0 (k) = −kas , as being the point at which the extrapolated external wave func-
tion intersects the axis (maybe at negative r). So, as k goes to zero, the term

1
lim k cot δ0 (k) = − (85)
k→0 as

dominates in the denominator of expression (67) where the scattering amplitude and the cross
section take the expression

f`=0 (k → 0) = −as and σ`=0 (k → 0) = 4πa2s . (86)

The parameter as of dimension length is called the scattering length. At low energies it deter-
mines solely the elastic cross section. This is a nontrivial construction from the potential itself
and the wave function of the state. We see that for momenta much less than the inverse radius
of the potential the scattering length is sufficient to describe all of the interactions. It is clear
that by measuring the scattering length of a system alone we cannot reconstruct the potential
uniquely. There are infinitely many different shapes, depths and ranges of potentials that will
reproduce a single scattering length.

5.3 S-Wave Scattering at Square Well Potential

The properties of scattering phases are studied for the problem of quantum scattering from an at-
~2
tractive spherically symmetric three-dimensional (3D) square well potential V (r) = − 2m v0 Θ(R◦ −
r). For convenience we write v0 = kv2 . The continuity condition of the wave function R`

R` (r, k) = A` (κ) j` (κr) with κ2 = k 2 + kv2 for r ≤ R◦ (87)

R` (r, k) = eiδ` (k) (cos δ` (k) j` (kr) + sin δ` (k) n` (kr)) for r ≥ R◦ (88)

and its derivative R`0 = dr

d
R` at the boundary, r = R◦ , where A` (κ) is a normalization constant.
The wave function outside of R◦ is normalized to to incoming plane wave normalized to unity.

A` (κ) j` (κR◦ ) = eiδ` (k) (cos δ` (k) j` (kR◦ ) + sin δ` (k) n` (kR◦ )) (89)
κ A` (κ) j`0 (κR◦ ) = k eiδ` (k) (cos δ` (k) j`0 (kR◦ ) + sin δ` (k) n0` (kR◦ )) (90)
8
Normalization constant A0 is neglected for simplicity.
Basic Scattering Theory A2.23

translates to the following relation for the phase shifts

k j`0 (kR◦ ) − L` (E) j` (kR◦ ) 1 j`0 (κR◦ )

tan δ` (k) = − 0
= with L` (E) = κ ,
k n` (kR◦ ) − L` (E) n` (kR◦ ) cot δ` (k) j` (κR◦ )
(91)
where L` is the logarithmic derivative of the wave function at the potential boundary. Here
j`0 (x) = dx
d
j` (x) and similarly for n0` (x).

Hard Sphere Potential

The simplest case is the scattering at a hard sphere potential:

v◦ (r) = ∞ for r < R◦ and v◦ (r) = 0 for r ≥ R◦ . (92)

Since then the wave function at the boundary r = R◦ vanishes, it follows that L` (v◦ → ∞) →
∞, and the phase shift reduces to

j` (kR◦ ) (sin kR◦ )/kR◦

tan δ` (k) = − s-wave : tan δ0 (k) = − = − tan kR◦ (93)
n` (kR◦ ) (cos kR◦ )/kR◦

so that δ0 (k) = kR◦ . Thus, the s-wave radial wave function for r > R◦ takes the form
 
0 −ikR◦ sin kr cos kr 1
R0 (r) = e cos kR◦ − sin kR◦ = e−ikR◦ sin k(r − R◦ ). (94)
kr kr kr
1
The corresponding radial wave-function for the incident wave takes the form R0 (r) = kr sin kr.
It is clear that the actual ` = 0 radial wave function is similar to the incident wave function,
except that it is phase-shifted by kR◦ . According to (86) the total s-wave cross-section of the
∞
hard wall potential yields then σ`=0 = 4πR◦2 , four times the geometric cross-section σgeo = πR◦2
(i.e., the cross-section for classical particles bouncing off a hard sphere of radius R◦ ). However,
low energy scattering implies relatively long wave-lengths, so we do not necessarily expect to
obtain the classical result in this limit. Recall that the s-wave scattering is a good approximation
to the low-energy scattering.
Consider the high energy limit kR◦  1. At high energies, all partial waves up to `max = kR◦
contribute significantly to the scattering cross-section. With so many ` values contributing,
it is legitimate to replace sin2 δ` in the expression (70) for the partial wave cross section by its
average value 1/2 and thus (for comparison we include also the low energy result, i.e. kR◦  1)
∞ ∞ ∞
σtot (kR◦  1) ' σ`=0 = 4πR◦2 and σtot (kR◦  1) ' 2πR◦2 . (95)

This is twice the classical result σgeo = πR◦2 , which is somewhat surprising, since we might
expect to obtain the classical result in the short wave-length limit. For hard sphere scattering,
incident waves with impact parameters less than R◦ must be deflected. However, in order to
produce a “shadow” behind the sphere, there must be scattering in the forward direction (re-
call the optical theorem) to produce destructive interference with the incident plane-wave. In
fact, the interference is not completely destructive, and the shadow has a bright spot in the for-
ward direction. The effective cross-section associated with this bright spot is πR◦2 which, when
combined with the cross-section for classical reflection, πR◦2 , gives the actual cross-section of
2σgeo .
A2.24 Stefan Blügel

Fig. 5: Radial scattering wave function, rR0 (r), for three-dimensional square well potential of
radius R◦ for kR◦ = 0.1 and γ = kv R◦ = 1 (left), π/2 (middle) and 2 (right). Note that the
scattering length, a0 changes from negative to positive as system passes through bound state.

Soft Sphere Potential

2
For a potential with finite scattering strength V (r) = − 2m
~
v◦ (r) the logarithmic derivative of the
scattering potential L` is finite. Analogously to (93) for s-wave scattering, Eq. (93) is simplified
to
k tan(κR◦ ) − κ tan(kR◦ )
tan δ0 (k) = . (96)
κ + k tan(kR◦ ) tan(κR◦ )

 unless tan(κR
Then,  ◦ ) = ∞, an expansion at low energy (small k) yields δ0 (k) ' kR◦ ×
tan(κR◦ )
× κR◦
− 1 , and the ` = 0 partial cross section,
 
4π 4π 1 4π tan(κR◦ )
σ`=0 (k) = 2 sin2 δ0 (k) = 2 2
' 2 δ02 (k) = 4πR◦2 −1 (97)
k k 1 + cot δ0 (k) k κR◦
tan(κR◦ )
From this result we find that, when κR◦
= 1, the scattering cross-section vanishes. An
expansion in small k obtains,
1 1
k cot δ0 (k) = − + r0 k 2 + · · · , (98)
a0 2
 
where a0 = 1 − tan(k v R◦ )
k v R◦
R◦ , defines the scattering length a0 or as , respectively, and r0 is the
effective range of the interaction that is obtained from the Taylor expansion of (96) for small
kR◦ . At low energies, k → 0, the scattering cross section, σ0 = 4πa20 (see above) is fixed by the
scattering length alone. If |kv R◦ |  1, a0 is negative. As kv R◦ is increased, when kv R◦ = π/2,
both a0 and σ0 diverge there is said to be a zero energy resonance. This condition corresponds
to a potential well that is just able to support an s-wave bound state at zero energy. If kv R◦
is further increased, a0 turns positive as it would be for an effective repulsive interaction until
kv R◦ = π when σ0 = 0 and the process is repeated with the appearance of a second bound state
at kv R◦ = 3/2, and so on. Since the scattering state must be orthogonal to all bounded states
of the potential V (r), the radial wave function of the scattering state must be orthogonal to all
radial wave functions of bound states at equal angular momentum `.9 Consider the situation of
a potential that supports at a given energy two bound states of s character. Then the scattering
9
The orthogonality to states of different ` is automatically taken care of by orthogonality conditions of the
angular part of the wave function.
Basic Scattering Theory A2.25

wave should also have two additional nodes, which moves the scattering phase δ` about 2π.
This is quantitatively expressed by the Levinson Theorem

δ` (k = 0) = N` π (99)

where N` is the number of bound states at given angular momentum `. If δ` (k) is increasing
rapidly through an odd multiple of π/2, sin2 δ` = 1 the `-th partial cross-section takes its
maximum value and the cross-section exhibits a narrow peak as a function of energy and there
is said to be a resonance. The analysis leads to the Breit-Wigner formula that goes beyond the
scope of this lecture.

5.4 Nuclear Scattering Length

Two fundamental interactions govern the scattering of neutrons by an atomic system and define
the neutron scattering cross-section measured in an experiment. The residual strong interaction,
also known as the nuclear force, gives rise to scattering by the atomic nuclei (nuclear scattering).
The electromagnetic interaction of the neutron’s magnetic moment with the sample’s internal
magnetic fields gives rise to magnetic scattering. The latter is neglected in this chapter. The
nuclear force is not weak as it is responsible for holding together protons and neutrons in the
nucleus. However, it has extremely short range, 10−13 cm to 10−12 cm, comparable with the size
of the nuclei, and much smaller than the typical distances in solids and much smaller than the
typical neutron’s wavelength. Consequently, away from the conditions of the resonance neutron
capture, the probability of a neutron being scattered by an individual nucleus is very small. To
describe the neutron’s interaction with the atomic system in which the typical distances are
about 1 Å (10−8 cm), the nuclear scattering length operator bN can be effectively treated as a
δ-function in the coordinate representation

bN = bN δ(r − RN ) , (100)

where r is a coordinate of a neutron and RN is that of a nucleus. Alternatively, in the momentum

representation it is just a number (for the nucleus fixed at the origin), bN (Q) = bN , independent
of the incident neutron’s wave vector and of the wave-vector transfer, Q. This again indicates
that the applicability of such treatment is limited to neutrons whose wavelength is large enough
compared to the size of the nuclei. In the Born approximation, Eq. (100) for the scattering
length would correspond to the neutron-nucleus interaction,

~2
V (r, RN ) = −4π bN δ(r − RN ) (101)
2mn

generally known as the Fermi pseudopotential [7, 8]. In Eqs. (100) and (101), the scattering
length refers to the fixed nucleus. Usually, it is treated as a phenomenological parameter that is
determined experimentally [9].

6 Scattering from a Collection of Scatterers

Finally, after having considered so far only the scattering at a single site with the target potential
V (r) placed at position Rτ = 0, prior to closing this chapter we shall relate these derivations
A2.26 Stefan Blügel

to the scattering phenomena in solid state systems with the potential composed of an assembly
of targets X
V (r) = vτ (r − Rτ ) (102)
τ
centered at a collection of sites Rτ . Inserting this into the Lippmann-Schwinger equation (20),
replacing the integration variable r0 by a vector r0τ ∈ Tτ within the target τ and the center-of-
gravity-vector Rτ , r0τ + Rτ , and taking into consideration that the free-space Green function
G◦ (r, r0 |E) depends only on r − r0 , the Lippmann-Schwinger equation for many potentials can
be written as
XZ
0
ψk (r) = ψk (r) + d3 rτ0 G◦ (r − Rτ , r0τ |E) V (r0τ ) ψk0 (r0τ + Rτ ). (103)
τ Tτ

Approximating the Green function by its far-field asymptotic form (23) and considering that
in the far-field solution ψk0 (r0τ + Rτ ) ' eikRτ ψk0 (r0τ ), behind which is the Huygens’ principle
where the wave function generated at different sites share a phase relation, which becomes exact
in the limit of the Born approximation, one obtains the asymptotic solution of the wave function
ψk0 (r) for large distances r
X 1
ψk0 (r) ' eikr + eik|r−Rτ | eikRτ fk (r\
− Rτ ) (104)
τ
|r − R τ |

Since in the far-field approximation (22), r  Rτ , k|r − Rτ | ' k(r − r̂ · Rτ ) = kr − k0 · Rτ ,

we shall rewrite the asymptotic solution of the wave function for scattering at many potentials
as
1 X
ψk0 (r) ' eikr + eikr F (Q) with F (Q) = Pτ (Q) fτ k (r̂τ ) and Q = k0 −k .
r τ
(105)
where the static structure factor (or structure factor for short) F (Q) describes the way in which
an incident beam is scattered by the atoms of a solid state system, taking into account the
different scattering power of the elements through the term fτ k (r̂τ ) also called atomic form
factor
Z
2m 1 0 0
fτ k (r̂τ ) = − 2 d3 rτ0 e−ik rτ V (r0τ ) ψk0 (r0τ ) with r τ = r − Rτ (106)
~ 4π Tτ
that depends only on the potential and not on the position of the atom. The atomic form factor,
or scattering power, of an element depends on the type of radiation considered. Since the atoms
are spatially distributed, there will be a difference in phase when considering the scattered
amplitude from two atoms. This phase difference is taken into account by the phase factor
Pτ (Q) = eiQ·Rτ , (107)
which depends only on the position of the atoms and is completely independent of the scattering
potential. So in total the structure factor separates the interference effects from the scattering
within an target from the interference effects arising from scattering from different targets.
Thus, the scattering intensity and the differential cross section are proportional to the square of
the structure factor
  2
dσ 2
X
I(Q) ∝ = |F (Q| = Pτ (Q) fτ k (r̂τ ) . (108)
dΩ τ
Basic Scattering Theory A2.27

If we ignore spin degrees of freedom, so that we do not have to worry whether an electron
does or does not flip its spin during the scattering process, then at low energies the scattering
amplitude f (θ) of particles from a cluster of atoms or a crystal becomes independent of angle
(s-wave) and maybe described by the scattering length bτ for atom τ , i.e. fτ k (r̂τ ) = bτ . Then,
the differential cross section simplifies to
  2
dσ X
= Pτ (Q) bτ . (109)
dΩ τ

If we consider scattering from a periodic crystal lattice, all atoms are same, i.e. have the same
nuclear number (and we consider here also all nuclei as identical), thus bτ = b for all atoms τ .
Then, we are left with the differential cross section
  2
dσ 2 1 X iQRτ
= N |b| S(Q) with S(Q) = e . (110)
dΩ N τ

S(Q) is the form factor of the lattice, a quantity closely q q q q q q q q q

related to the pair density of a solid. For simplicity          
we consider scattering at a finite Bravais lattice. The q q q q q q q q q 
        N a
lattice points are spanned by the lattice vector Rτ = q q q q q q q q q 2 2
Rm = Am, where A is the Bravais matrix consistent ~a2 q ~a-1 q q q q q q q q 
        
with the primitive vectors (a1 , a2 , a3 ) of the three di-  N1 a1 -
mensional lattice, and m = (m1 , m2 , m3 ) ∈ N3 with
0 ≤ mi ≤ (Ni − 1). The reciprocal lattice is defined
by the matrix B, that is orthogonal to A, B T A = 2π1. The reciprocal lattice vectors are given
as Gh = Bh with h = (h1 , h2 , h3 ) ∈ Z3 . The transfer of scattering momentum vector is
expressed by Q = (Q1 , Q2 , Q3 ) = Bκ, with κ = (κ1 , κ2 , κ3 ) ∈ R3 . The scattering amplitude
is calculated analytically summing up the geometrical series

X X N
X 1 −1 N
X2 −1 N
X3 −1
iQRm i2πκm
e = e = ei2πκ1 m1 ei2πκ2 m2 ei2πκ3 m3
m m m1 =0 m2 =0 m3 =0
i2πκ1 N1 i2πκ2 N2
1−e 1−e 1 − ei2πκ3 N3
= · ·
1 − ei2πκ1 1 − ei2πκ2 1 − ei2πκ3
sin N1 πκ1 sin N2 πκ2 sin N3 πκ3
= ei2π(κ1 (N1 −1)/2+κ2 (N2 −1)/2+κ3 (N3 −1)/2) · · ·
sin πκ1 sin πκ2 sin πκ3
giving the scattered intensity

sin2 12 N1 Q1 a1 sin2 12 N2 Q2 a2 sin2 21 N3 Q3 a3

 
dσ
I(Q) ∝ = |b|2 · · . (111)
dΩ sin2 12 Q1 a1 sin2 12 Q2 a2 sin2 21 Q3 a3

taking into account that 2πκi = Qi ai , for i = 1, 2, 3. The dependence of the scattering intensity
on the scattering vector Q is given by the so-called Laue function, which separates according to
the three Bravais vectors. One factor along one lattice direction a is plotted in Fig. 6.
The main peaks are the Bragg reflections. They occur at integer κ, κ = (κ1 , κ2 , κ3 ) ∈ Z3 , i.e.
at reciprocal lattice vectors Q = Gn . At points of Bragg reflection the coherent interference
of scattering waves of all atoms add up constructively so that the maximum intensity scales
A2.28 Stefan Blügel

Fig. 6: Laue function along the lattice direction a Fig. 7: Three-dimensional render-
for a lattice with N = 5 and N = 10 periods. ing of x-ray diffraction data obtained
from over 15 000 single nanocrys-
tal diffraction snapshots of a protein
complex [10].

with the square of the number of periods N 2 . This high intensity is the reason why the Born
approximation can in general not be used to describe the scattering at Bragg peaks. At small
deviations ∆κ from the exact Bragg condition the intensity drops fast,10 so that the total intensity
integrated over a small ∆κ region, ∆κ ' 1/N , varies only ' N . The half width is given
approximately by ∆Q = 2π 1
a N
. The more periods contribute to coherent scattering, the sharper
and higher are the main peaks. Between the main peaks, there are N − 2 side maxima. With
increasing number of periods N , their intensity becomes rapidly negligible compared to the
intensity of the main peaks. From the position of these Bragg peaks in momentum space, the
metric of the unit cell can be deduced (lattice constants a1 , a2 , a3 in the three Bravais vector
directions and unit cell angles α, β, γ). The width of the Bragg peaks is determined by the size
of the coherently scattering volume N = N1 N2 N3 and experimental factors (resolution) as well
as details of the sample (size of crystallite, mosaic distribution, internal strains, etc.). For large
N the form factor approaches a δ-function
 
dσ X
= N |b|2 VBZ δ(Q − Gh ) , (112)
dΩ h

where VBZ is the Brillouin-zone volume.

I shall conclude this chapter by mentioning that in 1914 the Nobel Prize in Physics was awarded
to Max von Laue “for his discovery of the diffraction of x-rays by crystals” and in 1915 the No-
bel Prize in Physics was awarded to Sir William Henry Bragg and William Lawrence Bragg
“for their services in the analysis of crystal structure by means of x-rays”. Since the pioneering

2
sin2 N πκ
R∞ 2 R∞ 2
10
For κ  1 it follows that sin N πκ
sin2 πκ
≈ π 2 κ2 =⇒ dκ sin N πκ
sin2 πκ
≈ N
π dx sinx2 x = N =⇒ κ̄ ' 1/N .
−∞ −∞
Basic Scattering Theory A2.29

days of von Laue, Ewald, Knipping, Friedrich, the Bragg’s, Compton, etc., diffraction experi-
ments went a long way deciphering today the atomic arrangement of noncrystalline solids such
as viruses as shown in a recent experiment [10] carried out at the Linac Coherent Light Source
(LCLS), at SLAC National Accelerator Laboratory in California, USA) as shown in the diffrac-
tion image Fig. 7. Physical principles established 100 years ago and subsequent theoretical and
experimental methods developed, reinvented and constantly brought to perfection contribute to
day and will contribute in the future to the welfare of mankind.

Acknowledgements
I thank Dr. Phivos Mavropoulos and Prof. Peter Dederichs for discussions and Benedikt Schwe-
flinghaus for his assistance in preparing the figures.

References
[1] J.J. Sakurai, Modern Quantum Theory, (Addison Wesley, 1984).

[2] Wu and Ohmura, Quantum Theory of Scattering, (Prentice Hall, 1962).

[3] A. S. Davidov, Quantum Mechanics, (Pergamon Press, 1991)

[4] N. W. Ashcroft and N. D. Mermin, Solid State Physics, (Brooks Cole, 1976)

[5] Chr. Huygens, Trait de la Lumiere (completed in 1678, published in Leyden in 1690)

[6] Sabrina Disch, Erik Wetterskog, Raphaël P. Hermann, German Salazar-Alvarez, Peter
Busch, Thomas Brückel, Lennart Bergström, and Saeed Kamali, Nano Lett. 11, 1651
(2011).

[7] G. L. Squires, Introduction to the theory of thermal neutron scattering, (Cambridge: Cam-
bridge University Press, 1978); (New York: Dover Publications, 1996).

[8] E. Fermi, Ric. Sci. 7, 13 (1936).

[9] E. Fermi, and L. Marshall, Phys. Rev. 71, 666 (1947); Phys. Rev. 72, 408 (1947).

[10] Henry N. Chapman et al., Nature 470, 73 (2011).