Scattering: Theory and Experiment Topic 5

Topic 5. Scattering: Theory and Experiment

Roger D. Johnson

Learning outcomes

• Understand the geometry of scattering, including the scattering triangle,

scattering vector, momentum transfer and energy transfer for both elastic
and inelastic scattering processes.

• Be familiar with the Ewald sphere so as to be able to visualise scattering

events in reciprocal space.

• Understand the Laue condition and be able to demonstrate its equivalence

with Braggs Law.

• Appreciate the basic properties of X-rays and neutrons relevant to their ap-
plication in scattering experiments.

• Understand the scattering of X-rays by electrons and the scattering of neu-

trons by nuclei and magnetic moments.

• Be able to obtain expressions for scattering amplitudes, form factors, unit

cell structure factor and differential scattering cross-section for elastic Bragg
scattering of both X-rays and neutrons from a crystal.

• Appreciate the relationship between magnetic order and magnetic diffrac-

tion.

PHAS0075 Page 1
Scattering: Theory and Experiment Topic 5

• Be familiar with experimental techniques for determining crystal structures

such as powder and single crystal diffraction.

• Be able to describe, compare and contrast different X-ray and neutron scat-
tering facilities.

• Understand the mechanisms behind the production of neutrons in reactor

and spallation sources, and the principles of the instrumentation required to
perform elastic and inelastic scattering experiments.

• Understand the production of X-rays from a lab-based and synchrotron source,

and how the properties of synchrotron radiation has revolutionised X-ray sci-
ence.

• Be familiar with the generic layout of a neutron or X-ray scattering instrument

and understand the roles of source, monochromator, analyser and detector.

• Be able to compare and contrast X-ray and neutron scattering for determin-
ing order in the solid state.

Acknowledgement: Contributions from Prof. S. T. Bramwell

PHAS0075 Page 2
Scattering: Theory and Experiment Topic 5

Contents

5.1. Mathematical Background 6

5.1.1 Fourier transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

5.1.2 Fourier transform theorems . . . . . . . . . . . . . . . . . . . . . . . . . . 6

5.1.3 Fourier transform of a lattice . . . . . . . . . . . . . . . . . . . . . . . . . 7

5.2. The Scattering Process 7

5.2.1 Properties of neutrons and X-rays . . . . . . . . . . . . . . . . . . . . . . 7

5.2.2 The scattering cross section . . . . . . . . . . . . . . . . . . . . . . . . . 8

5.2.3 General expression for scattering . . . . . . . . . . . . . . . . . . . . . . 9

5.3. Atomic scattering processes 10

5.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

5.3.2 Neutron nuclear scattering . . . . . . . . . . . . . . . . . . . . . . . . . . 10

5.3.3 Neutron magnetic scattering . . . . . . . . . . . . . . . . . . . . . . . . 11

5.3.4 X-ray Thomson scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

5.3.5 Thomson scattering from an atom . . . . . . . . . . . . . . . . . . . . . 13

5.4. Elastic scattering from groups of atoms 14

5.4.1 Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5.4.2 Structure factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5.5. Diffraction conditions and the Ewald sphere 16

5.5.1. The Laue diffraction condition . . . . . . . . . . . . . . . . . . . . . . . 16

PHAS0075 Page 3
Scattering: Theory and Experiment Topic 5

5.5.2. The Ewald sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.5.3. Bragg’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5.6. Experimental Facilities 20

5.6.1 Laboratory X-ray source . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5.6.2 Synchrotron X-ray source . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5.6.3 X-ray detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5.6.4 X-ray optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5.6.5 Neutron reactor source . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

5.6.6 Neutron spallation (synchrotron) source . . . . . . . . . . . . . . . . . . 23

5.6.7 Neutron detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5.6.8 Neutron optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5.7. Single crystal diffraction 24

5.7.1. Reflection geometry and ‘scans’ . . . . . . . . . . . . . . . . . . . . . . 25

5.7.2. Reciprocal space access in reflection . . . . . . . . . . . . . . . . . . . 26

5.7.3. Transmission geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

5.7.4. Triple-Axis Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

5.8. Powder diffraction 29

5.8.1. Powder diffraction geometry . . . . . . . . . . . . . . . . . . . . . . . . 29

5.8.2. Multiplicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5.8.3. Debye-Scherrer cones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

PHAS0075 Page 4
Scattering: Theory and Experiment Topic 5

Appendix 32

A.1 Derivation of the general expression for scattering . . . . . . . . . . . . 32

A.2 Scattering from molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

A.3 Coherent and incoherent scattering . . . . . . . . . . . . . . . . . . . . . 34

A.4 Thermal vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

PHAS0075 Page 5
Scattering: Theory and Experiment Topic 5

5.1. Mathematical Background

5.1.1 Fourier transforms

The Fourier transform of a function f (x) or f (r) gives another function fˆ(Q) or fˆ(Q).
Hence functions come in pairs, for example;

f (r) = δ 3 (r) ↔ fˆ(Q) = 1.

1 −κr 1
f (r) = e ↔ fˆ(Q) = ,
r κ2 + Q2
and in general a Gaussian of width σ transforms to a Gaussian of width 1/σ.

Here we’ll define the Fourier transform by

Z
F{f (r)} = fˆ(Q) = f (r)e−iQ·r d3 r.

(integral over all space). Alternative conventions include 2π factors: for a discus-
sion, see http://mathworld.wolfram.com/FourierTransform.html.

5.1.2 Fourier transform theorems

Fourier transforms obey certain theorems, for example the addition theorem:

F{f + g} = fˆ + ĝ,

the shift theorem:

F{f (r − r0 )} = fˆ eiQ·r0 ,

and the convolution theorem:

F{f ∗ g} = fˆ × ĝ.

Here the convolution of two functions is defined by the integral over all space:
Z
f (r) ∗ g(r) = f (r′ )g(r − r′ )d3 r′

PHAS0075 Page 6
Scattering: Theory and Experiment Topic 5

which is a function of r. Note that a convolution – a complicated integral – Fourier

transforms to a simple product.

5.1.3 Fourier transform of a lattice

It will be useful to obtain the Fourier transform of a lattice, which we define as a

periodic array of delta functions. Here we derive the result for a one dimensional
lattice. The lattice function is periodic with period a so we can write:
X
L(x) = δ(x − R)
R

where R = na. The one dimensional Fourier transform is:

Z X X X
ˆ
Fx {f (x)} = f (Q) = δ(x − R)e−iQx dx = e−iQR ∝ δ(Q − G)
R R G

where Q · G = 2πm, m integer. This defines the reciprocal lattice, so the reciprocal
lattice is the Fourier transform of the direct lattice. This result can be generalised
to three dimensions.

5.2. The Scattering Process

5.2.1 Properties of neutrons and X-rays

Neutrons are treated as free quantum particles, momentum p = h̄k with k = 2π/λ
and energy E = h̄2 k 2 /2m. X-rays are electromagnetic radiation, again with mo-
mentum p = h̄k and k = 2π/λ, but with energy E = h̄ck. The wavelengths λ are
typically chosen to be a few tenths of nanometers to match interatomic spacings
in condensed matter. For λ = 0.3 nm, the energies are En ≈ 100 K and EX ≈ 50 M
K. Energy transfers to and from thermal excitations in solids will be of order Kelvins
to hundreds of Kelvins, so neutrons are useful for resolving these, while X-rays are

PHAS0075 Page 7
Scattering: Theory and Experiment Topic 5

generally not. Hence X-ray scattering will be treated as elastic, neutron scattering
can be either elastic or inelastic.

5.2.2 The scattering cross section

Scattering is treated in the first Born approximation, a first order dynamical per-
turbation theory of the scattering process that amounts to neglect of multiple
scattering. This approximation is valid for a ‘sufficiently small’ sample or scattering
volume, and it is only because of this that neutron and x-ray scattering are quan-
titative probes of condensed matter. Consider a ‘unit flux’ of incoming radiation
with wavefunction ψi = eiki ·r , then we can think of a single point scatterer within
the sample as emitting a spherical wave with wavefunction:

ρ ikf ·r
ψf = e .
r

where ρ is a complex scattering amplitude. It has dimensions of length so is often

called the scattering length. At large enough ‘observation distance’ R we can
treat ψf as a plane wave:
ρ ikf ·r
ψf ≈ e .
R

We define a total scattering cross-section, σ, such that

Z
σ = |ψf |2 dA

where dA is an infinitesimally small area on the surface of a sphere of radius R. dA

subtends a solid angle, dΩ, such that dA = R2 dΩ. Hence, we can write
Z Z
σ = |ψf | dA = ρ2 dΩ = 4πρ2 .
2

We now define a differential cross-section

dσ
= ρ2
dΩ

PHAS0075 Page 8
Scattering: Theory and Experiment Topic 5

which, with a view to an experimental measurement, can be written

dσ N′
=
dΩ ΦdΩ

where Φ is the incident flux (units of neutrons or photons per second per unit area)
and N ′ is the number of neutrons or photons per second scattered into a detector
with solid angle dΩ. The total cross section is recovered by integration:
Z 4π
dσ N
σ= dΩ =
0 dΩ Φ

where N is the number of neutrons per second scattered by the sample.

5.2.3 General expression for scattering

We now ask the question, what is N ′ ? It can be shown that for elastic scattering,
in general,
 −1
V dEf
N = ′
2
(kf )2 |⟨kf αf |Hin |ki αi ⟩|2
h̄(2π) dkf
where αi and αf are the initial and final quantum states of the scatterer, Hin is the
interaction Hamiltonian, and V is the volume of the box containing the radiation.
In the Born approximation we have αf = αi , and can write

⟨kf αf |Hin |ki αi ⟩ = ⟨kf |Hin |ki ⟩⟨αf |αi ⟩ = ⟨kf |Hin |ki ⟩.

If we assume a general interaction potential, U (r) and normalised incident and

scattered particle wavefunctions, ψi = √1 exp(iki · r), and ψf = √1 exp(ikf · r), we
V V

can rewrite the above matrix element as follows;

Z Z Z
∗ 1 −ikf ·r iki ·r 1
⟨kf |Hin |ki ⟩ = ψf U (r)ψi dr = U (r)e e dr = U (r)e−iQ·r dr
V V

where the scattering vector Q = kf − ki . Hence,

 −1 Z 2
′ 1 dEf −iQ·r
N = (kf )2 U (r)e dr
V h̄(2π)2 dkf

where the integral is the Fourier transform of the interaction potential.

PHAS0075 Page 9
Scattering: Theory and Experiment Topic 5

5.3. Atomic scattering processes

5.3.1 Introduction

Condensed matter consists of atoms, each of which contains nuclei, core elec-
trons and valence electrons. Three types of scattering are important in the present
context (Fig. 1) : (i) X-ray Thomson scattering from (mainly) the core electrons, (ii)
neutron scattering from the nuclei and (iii) magnetic neutron scattering from any
unpaired electron density in the valence shell.

Figure 1: Atomic scattering processes: X-ray

Thomson, neutron nuclear and neutron mag-
netic.

5.3.2 Neutron nuclear scattering

Neutron elastic scattering from a single fixed nucleus is the easiest type of scatter-
ing to describe theoretically. We start with the general expression for the differen-
tial scattering cross section derived above. Referring to that equation, we have
the following: (1) For neutrons dEf /dkf = h̄2 kf /mn , where mn is the neutron mass
and (2) The flux Φ = h̄ki /mn V , where V the system volume. Plugging this into our
general expression gives:
 2 Z 2
dσ mn
= U (r)e−iQ·r dr ,
dΩ 2πh̄2

The simplest approximation for U (r) is the Fermi pseudopotential:

2πh̄2
 
dσ
U (r) = b δ(r) =⇒ = b2 =⇒ σ = 4πb2 ,
m dΩ

PHAS0075 Page 10
Scattering: Theory and Experiment Topic 5

where b is the nuclear scattering length (the bound scattering length, since the
nucleus is fixed). A more detailed analysis shows that both potential and reso-
nance scattering may occur. The potential scattering goes as the nuclear ‘size’:
b ≈ rnuc where rnuc ≈ 1.35A1/3 fm (A is the nucleon number), but resonance scatter-
ing causes large and haphazard deviations from this curve: see Fig. 2. It can even
result in a π phase shift, reversing the sign of b. For example, while the experimen-
tal value for lead (A = 207) is close to the theoretical value of 8, the experimental
value for manganese (A = 55) is −4 as opposed to the theoretical +5. The exis-
tence of negative scattering lengths is very useful, as discussed later.

5.3.3 Neutron magnetic scattering

Neutrons (n) with S = 1/2 are magnetic with moment µn , so are scattered by the
magnetic fields of unpaired valence electrons, B(r). The interaction potential is
then based on the dipole-dipole interaction:

U (r) = −µn · B(r).

One can show that the Fourier transform of the potential is given by
2πh̄2
  
γr0
U (Q) = 2 sn · M⊥ (Q)
mn 2µB
where sn is the neutron’s spin angular momentum and M⊥ (Q) is the Fourier trans-
form of the crystal’s magnetisation, projected into the plane perpendicular to Q.

Figure 2: How neutron scattering lengths depend

on atomic number.

PHAS0075 Page 11
Scattering: Theory and Experiment Topic 5

This arises because the Fourier transformed magnetic field is purely transverse to
Q, a consequence of Gauss’ law for magnetism;

∇ · B(r) = 0 =⇒ iQ · B(Q) = 0

For unpolarised neutron scattering the neutron magnetic moment components

are summed over and drop out the expression. The electronic magnetic moments
that contribute to the magnetisation may be expressed in terms of spin operators
and integrated over the magnetic electron density in the atom. Orbital mag-
netism in many cases may be accounted for by projecting onto a compounded
spin operator with effective g factor. If the atom or ion has spherical symmetry (a
close approximation in most cases) then the effective spin (operator) is located at
the centre of the spherical charge distribution by the Wigner-Eckart theorem. The
final cross section for magnetic neutron scattering is then:
 2
dσ γr0 2
= fm (Q)s⊥
dΩ 2µB

where M⊥ (Q) has be written as the product of a magnetic form factor fm (Q), the
Fourier transform of the unpaired electron density, and the perpendicular total
electron spin, s⊥ . The form factor will fall off more quickly with Q than the X-ray form
factor because of the greater spatial extent of the magnetic valence electrons
(Fig. 3). Note also, that for the reasons identified above, it is only spin components
perpendicular to the scattering vector that magnetically scatter neutrons.

5.3.4 X-ray Thomson scattering

Thomson scattering is X-ray elastic scattering from a free electron. The electric
vector of the incoming ray excites oscillations of the electron, which then re-emits
dipole radiation. The dipole character means that the scattering depends on
the initial and final polarisation state ϵ̂i , ϵ̂f respectively, where ϵ̂ is polarisation unit

PHAS0075 Page 12
Scattering: Theory and Experiment Topic 5

Figure 3: Fall off with Q for scattering

from the Er3+ ion (curves scaled to-
gether at zero).

vector. A field-theoretic treatment establishes the following cross section:

dσ
= r02 (ϵ̂i · ϵ̂f )2
dΩ
where the ϵ̂’s are polarisation unit vectors and r0 is the Thomson scattering length:
e2
 
r0 = 2
= 2.82 × 10−6 nm.
4πϵ0 me c
(me is the electron mass).

5.3.5 Thomson scattering from an atom

We now have to integrate over all the electrons in the atom, most of which are
tightly bound in the atomic core. Introducing the electron density n(r) we approx-
imate the matrix element as
Z
n(r) e−iQ·r d3 r = f (Q),

where spherical symmetry means the integral only depends on the amplitude Q
(Figure 4).

The scattering amplitude f (Q) is the Fourier transform of the electron density and
is called the atomic form factor. If we resolve the incoming ray into two orthogo-

PHAS0075 Page 13
Scattering: Theory and Experiment Topic 5

Figure 4: The atomimc form factor as the Fourier

transform of a spherical electron density

nal polarisation components, then the polarisation factor (ϵ̂i · ϵ̂f )2 ensures that an
incoming beam only scatters into its own polarisation state. The cross section is:

dσ
= r02 (ϵ̂i · ϵ̂f )2 |f (Q)|2
dΩ

5.4. Elastic scattering from groups of atoms

5.4.1 Crystals

We can generalise this calculation for any number of atoms simply by writing an
P
appropriate function L = i δ(r − Ri ), where the lattice {R} is a set of points

and on each point is placed the basis which is an atom or group of atoms. The
electron density of the crystal is the convolution of that of the basis with the lattice
function. Then it follows that

The scattering amplitude of the crystal is the product of the Fourier transform of
the basis with that of the lattice.

Here the Fourier transform of the basis can be written in the same way as for a
molecule:
X
fbasis = fj (Q)e−iQ·rj
j

(the basis is not necessarily a molecule, but there is nothing in the above expres-
sion to imply bonding of the atoms). The Fourier transform of the lattice may be

PHAS0075 Page 14
Scattering: Theory and Experiment Topic 5

obtained using the result proved in the previous section providing a mathematical
background:

The Fourier transform of the direct lattice {R} is the reciprocal lattice {G}.

It follows that the scattering amplitude for a crystal is

X
fcrystal = fbasis δ(Q − G).
G

Calculating the cross section, which is proportional to the square modulus of the
scattering amplitude, involves squaring a set of delta functions. If we consider the
delta function as a box function of height ∼ N and width ∼ 1/N , we can see that
δ 2 ∼ N δ. Hence,
2
X X
δ(Q − G) ∝N δ(Q − G),
G G

where N is the number of unit cells in the crystal and V0 is the volume of the unit
cell. Finally, the Thomson X-ray scattering cross section is:

dσ X
= N r02 (ϵ̂i · ϵ̂f )2 |fbasis (Q)|2 δ(Q − G).
dΩ G

5.4.2 Structure factors

The above expression describes elastic scattering from a static structure. The delta
function is a statement of the Laue condition Q = G while fbasis is the structure
factor, often denoted F . Hence, for a monatomic solid, for example, there is a
very sharp Bragg peak at each reciprocal lattice point, with the intensity falling
off as the square modulus of the atomic form factor.

If r is expressed in terms of a, b, c and G is expressed in terms of a∗ , b∗ , c∗ then

Q · ri = 2π(hxi + kyi + lzi ) in the usual crystallographic notation (xyz are fractional
co-ordinates, hkl Miller indices). The X-ray structure factor becomes:
X
X
Fhkl = fj (Q)e−2πi(hxj +kyj +lzj ) .
j

PHAS0075 Page 15
Scattering: Theory and Experiment Topic 5

Similarly the neutron nuclear (NN) structure factor is:

X
NN
Fhkl = bj e−2πi(hxj +kyj +lzj ) ,
j

and the neutron magnetic structure factor is:

X
NM
Fhkl = fm (Q)j s⊥
j e
−2πi(hxj +kyj +lzj )

We note that in the magnetic case the structure factor is a vector quantity, and
if the sj are not parallel (e.g. in an antiferromagnet), then they scatter differently,
so must must be considered to be different atoms when constructing the lattice,
basis, unit cell etc...

5.5. Diffraction conditions and the Ewald sphere

5.5.1. The Laue diffraction condition

The Laue diffraction condition states that diffraction occurs at the nodal points of
the reciprocal lattice:
Q=G

Q is the so-called scattering vector;

Q = kf − ki

where kf and ki are the scattered and incident wavevectors, respectively. An

equivalent description of the Laue diffraction condition is obtained using the fact
that ui · u∗ j = 2πδij ;

Q · a = 2πh, Q · b = 2πk, Q · c = 2πl

where h, k, and l are integers.

PHAS0075 Page 16
Scattering: Theory and Experiment Topic 5

Figure 5: The scattering triangle.

In the case of elastic scattering (|kf | = |ki |), Q, kf , and ki form three sides of an
isosceles triangle known as the scattering triangle. The angle between kf , and ki
is known as the scattering angle, 2θ.

For inelastic scattering, conservation of momentum maintains the scattering trian-

gle, but it is no longer isosceles (see next section).

5.5.2. The Ewald sphere

The locus of possible scattering vectors, Q, is defined by the surface of a sphere

with diameter |kf | + |ki |. This is known as the Ewald sphere.

It is informative to draw the scattering triangle superimposed onto the intersec-

tion of the Ewald sphere with a 2D plane of the reciprocal lattice whose origin is
placed at the bottom of Q, as shown in Figure 6 for elastic scattering. Diffraction
occurs when a non-extinct reciprocal lattice point, G, lies on the circumference
of the circle (in 2D) or the surface of the sphere (in 3D). For a given reciprocal
lattice point this condition is achieved for a specific orientation of the reciprocal
lattice with respect to the incident wavevector, ki , and at a specific value of 2θ.

The Ewald sphere construction holds for inelastic scattering. Consider the case
where a neutron or X-ray interacts with an excitation of the crystal, for example a
phonon or a magnon. The neutron can loose energy to the crystal and create an
excitation (energy loss scattering) or it can gain energy from the crystal through

PHAS0075 Page 17
Scattering: Theory and Experiment Topic 5

Figure 6: The Ewald sphere in 2D.

annihilating an excitation (energy gain scattering).

kexcitation = ki − kf (momentum transfer)

Ephonon = Ei − Ef (energy loss scattering)

Ephonon = Ef − Ei (energy gain scattering)

For consistency, we define the scattering vector, Q, to be opposite to the momen-

tum transfer, kphonon :

Q = −kphonon = kf − ki (scattering vector)

Reminder: The excitation dispersion is periodic under the translational symmetry of

the reciprocal lattice, hence the momenta of equivalent excitations are related
by a reciprocal lattice vector, G:

k′excitation = kexcitation + G

PHAS0075 Page 18
Scattering: Theory and Experiment Topic 5

In energy loss inelastic scattering |kf | is smaller than |ki |, as Ef < Ei , and

h̄2 |ki,f |2
Ei,f =
2mn

where mn = 1.675 × 10−27 kg is the mass of the neutron (see Figure 7).

Figure 7: The Ewald sphere in 2D for energy loss inelastic scattering.

In energy gain inelastic scattering |kf | is larger than |ki |, as Ef > Ei (see Figure 8).

5.5.3. Bragg’s Law

Using the scattering triangle, we can identify a relationship between the scatter-
ing angle 2θ and the length of Q by geometry:

λ|Q|
sin(θ) =
4π

where λ is the wavelength of the elastically scattered particles such that |kf | =
|ki | = 2π/λ. If we define the spacing between planes of atoms, dhkl , that are in
phase with the scattering vector Q(hkl) as

2π
dhkl =
|Q|

PHAS0075 Page 19
Scattering: Theory and Experiment Topic 5

Figure 8: The Ewald sphere in 2D for energy gain inelastic scattering.

and substitute |Q| into the above we obtain Bragg’s law:

λ = 2dhkl sin(θ)

5.6. Experimental Facilities

This section treats two types of X-ray source (laboratory and synchrotron) and two
types of neutron source (reactor and spallation). In each case, the generation
of the radiation, its delivery to the sample and its ultimate detection will be de-
scribed. One important point to note immediately is the large variation of relative
intensities of the sources. Laboratory X-ray sources typically produce 109 pho-
tons per second, synchrotron X-ray sources 1012 photons per second and neutron

PHAS0075 Page 20
Scattering: Theory and Experiment Topic 5

sources only 107 neutrons per second.

5.6.1 Laboratory X-ray source

The basic laboratory source is based on the Coolidge Tube (Coolidge, 1913)
where thermionically emitted electrons in a ‘cathode ray tube’ are accelerated
by a potential difference to bombard a water-cooled metal anode (typically cop-
per or molybdenum). The bombarding electrons dislodge electrons from inner
atomic shells and then subsequent transitions back to these shells give X-radiation.
The most used transitions are CuKα, CuKβ, MoKα, MoKβ. The spectrum of X-rays
produced is accompanied by a broad background of bremsstralung radiation
coming from decelerating electrons. To select a single wavelength (e.g. CuKα
rather than CuKβ), Bragg reflection(s) from a monochromator crystal (often made
of germanium) may be used.

5.6.2 Synchrotron X-ray source

These are central facilities costing many millions of pounds and typically invested
in by multiple countries. Examples are the UK’s Diamond Light Source (DLS) and
the European Synchrotron Radiation Facility (ESRF). Electrons in a circular storage
ring are maintained near light speeds and accelerated (their path is bent) to
produce tangential X-radiation. Typically the X-rays come in pulses on the ns to ps
time scale and are of very high intensity (hence very dangerous to humans). The
details of the light-producing acceleration and hence X-ray spectrum are tuned
by insertion devices, arrays of magnets called bending magnets, undulators and
wigglers. The X-ray experiment takes place on a beamline consisting, typically, of
an optics cabin, an experimental cabin and a control cabin. Strict protocols are
enforced to ensure there is no (likely fatal) human contact with the X-ray beam.

PHAS0075 Page 21
Scattering: Theory and Experiment Topic 5

5.6.3 X-ray detectors

After scattering from a sample of interest, X-ray detection is based on two princi-
ples: (i) phosphors (X-rays → light of a different wavelength, and (ii) the photoelec-
tric effect (X-rays → electrons). The light or electrons so produced may then be
detected by photographic or electronic methods respectively. Devices typically
used include the charge coupled device (CCD) or the complementary metal ox-
ide semiconductor (CMOS), which convert visible light to electrons and holes and
then count, or the hybrid pixel array detector (HPAD) which uses p-n junctions
to convert X-rays to electrons and holes. Phosphor-based image plates are also
used.

5.6.4 X-ray optics

Optical effects (diffraction, refraction, reflection, absorption) are generally used

to prepare the X-ray beam, i.e. to monochromate, focus, or collimate it, before it
impinges upon the sample (collimate means to reduce divergence). Monochro-
mators are well-defined single crystals (e.g. Si, Ge) which are used to select a
single wavelength by Bragg reflection (with a small intrinsic bandwidth). Double-
bounce monochromators may be used to avoid beam re-direction. Note that
the X-ray refractive index is (slightly) < 1, hence there is total external reflection
at small angle (∼0.5 degrees), which can be exploited to design highly polished
X-ray mirrors for focusing. Multilayers of different materials e.g. heavy metals, or
light elements like C, Si, give many options, e.g for manipulating diffraction wave-
lengths. Zone-plate type focusing optics (focussing by diffraction) are also used.

PHAS0075 Page 22
Scattering: Theory and Experiment Topic 5

5.6.5 Neutron reactor source

In contrast to the case of X-rays, there is no usable laboratory source of neutrons

for neutron scattering. All neutron sources are central facilities. The first, and older,
type is a nuclear reactor source, such as that at the Institut Laue Langevin (ILL),
Grenoble, France, which is jointly funded largely by the UK, France and Germany.
Here neutrons are produced by nuclear fission where neutrons split target heavy
metal nuclei to produce yet more neutrons and a consequent chain reaction.
Holes in the side of the reactor then release a continuous source of neutrons which
are slowed to ‘thermal’ speeds by passing through moderators made out of light
elements e.g. H2 , D2 , H2 O, D2 O, Be or C held at fixed temperature between 20
K and 2000 K. Light elements tend to have low absorption and high scattering
cross sections allowing for efficient thermalisation of the neutrons, which produces
a Maxwellian flux distribution. The thermal neutrons so produced are then used
directly or passed down guide tubes (see below) and are usually reflected from a
monochromator crystal before use.

5.6.6 Neutron spallation (synchrotron) source

Driven in part by public concern about the dangers of nuclear fission technology,
the more modern neutron sources are so-called spallation sources, based on syn-
chrotron technology. Example include the ISIS source in Southern England and the
under-construction European Spallation Source (ESS) in Lund, Sweden. At these
sources, protons produced in a synchrotron accelerator impinge on (typically
tungsten) targets and neutrons are produced by the spallation reaction, where
neutrons are spalled (chipped off) the heavy metal nuclei. Spallation sources are
pulsed sources. For example, the UK’s ISIS source produces 50 pulses of neutrons
per second. The neutrons produced are moderated and used on instruments,
where the pulsed nature of the beam may be exploited as a useful property of
the experimental probe.

PHAS0075 Page 23
Scattering: Theory and Experiment Topic 5

5.6.7 Neutron detectors

Neutrons scattered from the sample of interest are generally detected by gas
filled devices such as helium-3 proportional detectors. These utilise the nuclear
reaction: 3 He + n →3 H + p + 765 keV, where the energy is dissipated by ionization
of a (CH4 ) proportional counting gas. Also commonly used are BF3 proportional
detectors which use the reaction 10
B + n →7 Li∗ + 4 He + 2310 keV; 7 Li∗ →7 Li + 480
keV. Position sensitive detectors may use multiple anode/cathode wires or time
delay/coincidence methods. All these detectors are relatively large, giving much
lower resolution than is possible in X-ray detection.

5.6.8 Neutron optics

In neutron scattering, the complete (going beyond Born approximation) theory

reveals the usual array of wave-like optical effects: reflection, refraction as well as
diffraction. Like X-rays, the neutron refractive index is slightly less than unity, giv-
ing total external reflection at glancing angles. This allows a major application in
neutron guides – tubes that are many metres long and often slightly curved. These
transport neutrons to instruments far from the source, hence reducing background
and allowing more space for experiments etc. Multilayers may be utilised to cre-
ate diffracting mirrors, so-called ‘supermirrors’ with a modified critical angle. Bent
crystals (or crystal mosaics) are used to make focusing monochromators.

5.7. Single crystal diffraction

In diffraction experiments we measure intensities, I(hkl), where hkl are the Miller
indices that index specific diffraction reflections. The intensity is proportional to the

PHAS0075 Page 24
Scattering: Theory and Experiment Topic 5

differential scattering cross section evaluated at the respective hkl:

dσ N′ Ihkl
= =
dΩ ΦdΩ Φ

where we have assumed the detector of solid angle dΩ integrates the full diffrac-
tion peak in 3D space (in practice this is done by ’scanning’, as discussed below).

In most experiments we can only measure relative intensities, hence it is sufficient

to use the following expression (up to non-examinable experimental corrections
such as absorption)
dσ
Ihkl ∝ ∝ |Fhkl |2
dΩ

5.7.1. Reflection geometry and ‘scans’

Figure 9: The reflection geometry.

In a single crystal diffraction experiment in reflection geometry individual diffrac-

tion intensities with a chosen (h, k, l) are ‘scanned’. Scans are performed by either
rotating the sample by an angle, ω, with the detector at fixed 2θ (Figure 10) or by
scanning the detector angle, 2θ, whilst keeping the sample at fixed ω (Figure 11).

PHAS0075 Page 25
Scattering: Theory and Experiment Topic 5

Figure 10: An ω scan, performed by rotating the sample.

Figure 11: A 2θ scan, performed by rotating the detector.

5.7.2. Reciprocal space access in reflection

The above motions provide access to all accessible reciprocal lattice points in the
diffraction plane above the sample surface and encompassed by a semicircle
with radius |Q|max . Reflection geometry imposes two further constraints on ω. Firstly,
ω cannot be negative, as this would require the incident x-ray beam to hit the
back of the sample. Second, ω cannot be larger than 2θ, as this would result in
x-rays being diffracted into the sample. The accessible region of reciprocal space
in the diffraction plane is summarised in Figure 12.

PHAS0075 Page 26
Scattering: Theory and Experiment Topic 5

Figure 12: The accessible region of reciprocal space within the diffraction plane for single
crystal diffraction in the reflection geometry (ω ≤ 0 corresponds to the incident beam
hitting the back of the sample, and ω ≥ 0 corresponds to a scattered beam below the
surface of the sample.

5.7.3. Transmission geometry

In a transmission experiment the above restrictions on ω do not apply. We measure

a full circle of data (Figure 13), or in 3D - a full sphere. Typically 2D detectors are
employed, positioned over a range of 2θ values whilst the crystal is rotated by the
sample stage into all orientations such that all diffraction conditions are met within
the accessible range of |Q|. It is important to minimise absorption by the sample.
Therefore small samples are typically measured in conjunction with higher energy
x-rays.

5.7.4. Triple-Axis Spectrometer

In the case of inelastic scattering we want to measure not just the direction but
also the magnitude of kf and hence the energy and momentum transfer. This
is most clearly realised using a triple-axis spectrometer. The three axes referred
to are those of (1) a monochromator crystal placed before the sample, (2) the
single crystal sample itself, and (3) an analyser crystal placed after the sample

PHAS0075 Page 27
Scattering: Theory and Experiment Topic 5

Figure 13: The accessible region of reciprocal space within the diffraction plane for elastic
single crystal diffraction in the transmission geometry.

but before the detector. The monochromator provides a known k. The scattering
angle at the sample determines the direction of k′ but not its magnitude. Bragg
reflection by the analyser crystal then selects a particular wavelength, hence fixes
k ′ and the energy. By scanning the angle of the analyser crystal the distribution of
momentum transfer h̄Q and energy can be measured.

Unlike elastic diffraction in transmission, it is not possible to access all regions of

reciprocal space within a sphere of radius |Q|max (Figure 14). In inelastic scattering
we can access the range |kf − ki | ≤ |Q| ≤ |kf + ki |:

Often one is forced to measure points at larger Q, related by the translational

symmetry of the reciprocal lattice to points within the first Brillouin zone. Take,
for example, the case of a zone-boundary acoustic phonon, kphonon = ( πa , 0, 0),
with energy Ephonon = 10 meV, to be measured by energy loss inelastic neutron
scattering (a = 0.5 nm). If the neutron beam has an incident energy of Ei = 25 meV,
we can determine which kphonon + G points are accessible:

PHAS0075 Page 28
Scattering: Theory and Experiment Topic 5

Figure 14: The accessible region of reciprocal space within the diffraction plane for inelas-
tic single crystal diffraction in the transmission geometry.

r r
2mn Ei 2mn (Ei − Ephonon )
|ki | = = 34.7 nm−1 and |kf | = = 26.9 nm−1
h̄2 h̄2
therefore

|Q|min = 34.7 − 26.9 = 7.8 nm−1 and |Q|max = 34.7 + 26.9 = 61.6 nm−1

|kphonon | = π
a
= 6.3 nm−1 , |kphonon + (1, 0, 0)| = 3π
a
= 18.8 nm−1 , |kphonon + (2, 0, 0)| = 5π
a
=
31.4 nm−1 etc...

So the boundary of the first Brillouin Zone cannot be accessed, but points further
out in reciprocal space can.

5.8. Powder diffraction

5.8.1. Powder diffraction geometry

A powder sample is a collection of randomly oriented small single crystals of suffi-

cient number such that, to a good approximation, every crystal orientation exists

PHAS0075 Page 29
Scattering: Theory and Experiment Topic 5

Figure 15: Scattering triangle construction illustrating limits to reciprocal space access in
inelastic scattering.

within the sample. In the single crystal experiment the detector and sample must
be positioned correctly in order to achieve a given diffraction condition. Owing to
the random orientation of crystals in a powder, the sample orientation is no longer
relevant — so long as the detector is set at the correct value of 2θ, diffraction will
occur. As such, we are only concerned with the value of |Q|:

Figure 16: Relationship between single crystal and powder diffraction.

PHAS0075 Page 30
Scattering: Theory and Experiment Topic 5

5.8.2. Multiplicity

All symmetry equivalent reciprocal lattice points contribute to the same powder
diffraction intensity. When calculating the powder diffraction intensity we can
consider single diffraction peaks, and then multiply the structure factor squared
(or the intensity) by the multiplicity of the peak:

Figure 17: Peak multiplicity in powder diffraction.

Note that in some cases powder diffraction peaks are accidentally degenerate
beyond symmetry equivalence. For example, in the case of the tetragonal crystal
p
system |Q| = 2π (h2 + k 2 )/a2 + l2 /c2 , hence the (5,0,0) and (4,3,0) reflections are
degenerate in |Q|, despite being inequivalent by symmetry.

5.8.3. Debye-Scherrer cones

The 2θ angle is fixed for a given diffraction condition. For a randomly oriented
powder, the locus of all possible scattered beams is a cone around the direction
of the incident x-ray beam. These cones are known as Debye-Scherrer cones
(see Figure 18). The intensity around the cones is integrated over the width of
the detector, producing 1D data of x-ray intensity as a function of |Q|. As the
detector has finite width the fraction of the cone measured is angle dependent,
with a maximum at 2θ = 0 and 180, and a minimum at 2θ = 90, which requires an
instrument specific correction to the diffraction data.

PHAS0075 Page 31
Scattering: Theory and Experiment Topic 5

Figure 18: The Debye-Scherrer cone.

PHAS0075 Page 32