1.

2 Scattering 5

The numerical relation1 between wavelength λ in Å and photon energy E in keV is

h i hc 12.398
λ Å = = (1.1)
E E [keV]
An X-ray photon interacts with an atom in one of two ways: it can be scattered or it can be absorbed,
and we shall discuss these processes in turn. When X-rays interact with a dense medium consisting of
a very large number of atoms or molecules it is sometimes more convenient to treat the material as
a continuum, with an interface to the surrounding vacuum (or air). At the interface the X-ray beam
is refracted and reflected, and this is an alternative way in which the interaction may be discussed.
The scattering and refraction descriptions are of course equivalent. In Chapter 3 we derive the X-ray
reflectivity equations, and exploit this equivalence to relate the reflectivity to the microscopic properties
of the medium of interest.

1.2 Scattering
To start with we shall consider the scattering of an X-ray by a single electron. In the classical description
of the scattering event the electric field of the incident X-ray exerts a force on the electronic charge,
which then accelerates and radiates the scattered wave. Classically, the wavelength of the scattered
wave is the same as that of the incident one, and the scattering is necessarily elastic. This is not true in
general in a quantum mechanical description, where the incident X-ray photon has a momentum of ~k
and an energy of ~ω. Energy may be transferred to the electron with the result that the scattered photon
has a lower frequency relative to that of the incident one. This inelastic scattering process is known as
the Compton effect, and is discussed at the end of this section. However, the elastic scattering of X-rays
is the main process that is exploited in investigations of the structure of materials, and in this case it
suffices to adopt what is essentially a classical approach.

One electron
The most elementary scattering object that we shall consider is a single, free electron. The ability of an
electron to scatter an X-ray is expressed in terms of a scattering length, which we shall now derive.
Figure 1.5 shows a schematic of a generic scattering experiment. The fundamental quantity
determined in such an experiment is the differential scattering cross-section (dσ/dΩ) which is defined
by !
dσ I sc
= (1.2)
dΩ Φ0 ∆Ω
The strength of the incident beam is given by the flux, Φ0 , which is simply the number of photons
passing through unit area per second. The incident beam interacts with the scattering object and is
scattered. The number of scattered photons recorded per second in a detector is I sc , where the detector
is positioned a distance R away from the object and subtends a solid angle ∆Ω. The differential cross-
section is thus a measure of the efficiency of the scattering process where details of the experiment,
1 In this book we shall mostly limit the wavelength band to 0.1 − 2 Å corresponding to the energy band 120 − 6 keV. The
first limit, 0.1 Å or 120 keV, ensures that relativistic effects are negligible since the X-ray energy is considerably lower than the
rest mass of the electron, mc2 = 511 keV. The second limit, 2 Å or 6 keV, is a practical limit ensuring that the X-rays have a
high penetration power through light materials, such as beryllium. In many X-ray tubes, and in synchrotron radiation beam lines,
the X-rays must be transmitted through a Be window, and above 6 keV the transmission of a 0.5 mm Be window exceeds 90%.
Lower energy X-rays are called soft X-rays and will not be dealt with in this book.
6 X-rays and their interaction with matter

Fig. 1.5 Schematic layout of a generic scattering experiment used to determine the differential cross-section (dσ/dΩ): see
Eq. (1.2). The incident beam flux Φ0 is the number of particles per second per unit area. For an electromagnetic wave this is
proportional to |Ein |2 times the velocity of light, c. The incident beam interacts with the target object to produce the scattered
beam. A detector records the scattered intensity, Isc , defined as the number of counts recorded per second, which is proportional
to |Erad |2 times the area of the detector and the velocity of light. The detector is located a distance R from the target object, and
subtends a solid angle of ∆Ω.

specifically the flux of the incident beam and size of the detector, have been normalized away. (See
Appendix A for a more complete discussion.)
For the particular case of the scattering of an electromagnetic wave indicated in Fig. 1.5 an
expression for Φ0 can be obtained in terms of the electric field Ein of the incident beam. Since the energy
density is proportional to |Ein |2 , the number density of photons is proportional to |Ein |2 /~ω, while the
flux Φ0 is the number density multiplied by speed of light, c. (This follows from the realisation that
in one second a beam of area A sweeps out a volume equal to A c.) A similar argument applies to
the intensity I sc of the scattered beam. In this case the number density is proportional to the modulus
squared of the radiated electric field, |Erad |2 . This quantity then must be multiplied by the area of
the detector, R2 ∆Ω, and c to yield and expression for I sc . With these considerations, the differential
cross-section is given by
|Erad |2 R2
!
dσ
= (1.3)
dΩ |Ein |2
In a classical description of the scattering process an electron will be forced to vibrate when placed
in the electric field of an incident X-ray beam, as illustrated in Fig. 1.6(a). A vibrating electron acts
as a source, and radiates a spherical wave Erad ∝ ε̂′ eikR /R. The problem then is to evaluate the
radiated field at an observation point X. This calculation is performed in Appendix B starting from
 1.2 Scattering 7

Fig. 1.6 The classical description of the scattering of an X-ray by an electron. (a) The electric field of an incident plane wave
sets an electron in oscillation which then radiates a spherical wave. (For clarity the radiated wave is shown for positive values of
y only, and for the simplest case of an isotropic spherical wave the phase and amplitude are constant on spherical surfaces.) The
incident wave propagates along the z axis and has its electric field polarized along x. The wave crests of the incident wave lie
in between those of the scattered spherical wave because of the 180◦ phase shift in Thomson scattering. In the text the radiated
field at an observation point X is calculated. Point X lies in the plane spanned by the polarization vector and the propagation
direction of the incident wave, and the observed acceleration has to be multiplied by a factor of sin Ψ. (b) From geometry,
sin Ψ = −ε̂ · ε̂′ where ε̂ (ε̂′ ) represents the polarization of the incident (scattered) beam. The effect of this factor on the radiated
wave is illustrated by plotting surfaces of constant amplitude.
8 X-rays and their interaction with matter

Maxwell’s equations. Here, an heuristic argument is outlined. Initially we consider the situation where
the observation point X lies in the plane spanned by the polarization vector and the propagation direction
of the incident wave, and at an angle 90◦ − Ψ with respect to the direction of propagation of the incident
beam (Fig. 1.6(a)).
The radiated field is proportional to the charge of the electron, −e, and to the acceleration, aX (t′ ),
evaluated at a time t′ earlier than the observation time t due to finite speed c at which the radiation
propagates. The radiated field is thus expected to be of the form
−e
Erad (R, t) ∝ a (t′ ) sin Ψ (1.4)
R X
where t′ = t − R/c. The total energy flow through a spherical shell of radius R is the energy density,
2
proportional to Erad , multiplied by the surface area, proportional to R2 , so with Erad ∝ R−1 the total
energy flow becomes independent of R, as it must2 . A further factor of sin Ψ has been included to allow
for the variation of the acceleration with observation angle. For an observer at point X in the x −z plane,
the acceleration observed is zero for Ψ = 0◦ , and a maximum for Ψ = 90◦ . Therefore, the acceleration
observed is the full acceleration multiplied by sin Ψ.
To proceed we evaluate the full acceleration from the force on the electron divided by its mass,
which yields
′
−e E0 e−iωt −e −e
aX (t′ ) = = Ein eiω(R/c) = E eikR
m m m in
where Ein = E0 e−i ωt is the electric field of the incident wave. Hence Eq. (1.4) can be rearranged to read

Erad (R, t) e2 eikR

!
∝ sin Ψ (1.5)
Ein m R

For an observation point at an arbitrary angle with respect to the polarization of the incident beam,
the factor of sin Ψ must be reevaluated. If ε̂ is the polarization of the incident field, and ε̂′ that of the
radiated field, then from Fig. 1.5(b), ε̂ · ε̂′ = cos(90◦ + Ψ) = − sin(Ψ). The advantage of writing the
trigonometric factor for the apparent acceleration in this way is that it is valid for all possible angles
of observation. This is ensured by the azimuthal symmetry evident in Fig. 1.6(b) of the radiated field
around the x axis.
To complete the derivation of the differential cross-section it is necessary to check whether we have
the correct units. Clearly, the ratio of electric fields given in Eq. (1.5) is dimensionless. This requires
that whatever the factor is multiplying the spherical wave form eikR /R, it must have units of length. The
appropriate length can be found by noting that in SI units the Coulomb energy at distance r from a point
charge −e is e2 /(4πǫ0r), while dimensionally the energy is also of the form mc2 . Thus equating these
two expressions for energy and rearranging provides an expression for the fundamental length scale in
the problem, namely

e2
!
r0 = = 2.82 × 10−5 Å (1.6)
4πǫ0 mc2

This is referred to as the Thomson scattering length, or classical radius, of the electron. While these
arguments fix the magnitude of the scattering length, they do not fix its phase. As shown in Appendix
2 Equation (1.4) represents the electric field radiated by an oscillating dipole in the far-field limit.
 1.2 Scattering 9

B, the scattering amplitude from a single electron is in fact equal to −r0 ε̂ · ε̂′ . Physically, the factor
of −1 represents the fact that there is a 180◦ phase shift between the incident and scattered waves. This
phase shift also has consequences for the refractive index n which in the X-ray region is less than unity,
as discussed in Section 1.4 and Chapter 3.
The ratio of radiated to incident electric fields is therefore
Erad (R, t) eikR
= −r0 ε̂ · ε̂′ (1.7)
Ein R

and from Eq. (1.3) the differential cross-section becomes

!
dσ 2
= r02 ε̂ · ε̂′ (1.8)
dΩ

This equation describes the Thomson differential scattering cross-section of an electromagnetic wave
by a free electron.
2
The factor of ε̂ · ε̂′ has important implications for the choice of optimal geometry for different
types of X-ray experiments. For example, synchrotron sources naturally produce X-rays which are
linearly polarized in the horizontal plane of the synchrotron. It follows that scattering experiments are
2
best performed in a vertical scattering plane as then ε̂ · ε̂′ = 1 independent of the scattering angle,
ψ = 90◦ −Ψ. Conversely, if one wants to study fluorescence from a sample, then it is possible to suppress
2
the scattering by working in the horizontal plane at ψ = 90◦ since ε̂ · ε̂′ = 0. These considerations
lead us to define P, the polarization factor for scattering, which depends on the X-ray source:



 1 synchrotron: vertical scattering plane
2

 2
′
cos ψ (1.9)

P = ε̂ · ε̂ =  synchrotron: horizontal scattering plane
1 2
  
 1 + cos ψ unpolarized source


2

The total cross-section for Thomson scattering is found by integrating the differential cross-section
over all possible scattering angles. Exploiting the rotational symmetry of the radiated field around ε̂,
it can be shown that the average value of < ε̂ · ε̂′ 2 > over the unit sphere is (2/3). Thus the total
"

cross-section σT is equal to 4πr02 × (2/3) = 8πr02 /3 = 0.665×10−24 cm2 =0.665 barn. It is evident that
the classical cross-section, both the differential and total, for the scattering of an electromagnetic wave
by a free electron is a constant, independent of energy. This result is particularly relevant to the X-ray
part of the electromagnetic spectrum, as here a photon is energetic enough that even atomic electrons
respond to a good approximation as if they are free. Where it breaks down entirely is at low energies in
the optical part of the spectrum, or when the energy of photon passes a threshold for resonantly exciting
electrons from deeply bound atomic states, as outlined later in this section. In Chapter 8 we discuss the
origin and consequences of such resonant scattering processes.
Finally in this section we note that the classical derivation of the scattering of a photon by a free
electron given here yields the same result as the full quantum mechanical derivation given in Appendix
C.
10 X-rays and their interaction with matter

(a) One atom

k k´

r
! !

(b) One molecule

Q=k-k´
k k´

(c) A crystal
Rn
d

Fig. 1.7 (a) Scattering from an atom. An X-ray with a wavevector k scatters from an atom to the direction specified by k′ . The
scattering is assumed to be elastic, i.e. |k| = |k′ | = 2π/λ. The difference in phase between a wave scattered at the origin and
one at a position r is (k − k′ ) · r = Q · r. This defines the wavevector transfer Q. (b) The scattering from a molecule. Here the
scattering triangle is shown which relates k, k′ and Q. (c) Scattering from a molecular crystal. The molecules are organized on
a lattice with position vectors Rn , and a lattice plane spacing of d.
 1.2 Scattering 11

One atom
Let us now proceed from the scattering by a single electron to consider the elastic scattering from an
atom with Z electrons.
To start with a purely classical description will be used, so that the electron distribution is specified
by a number density, ρ(r). The scattered radiation field is a superposition of contributions from different
volume elements of this charge distribution. In order to evaluate this superposition one must keep track
of the phase of the incident wave as it interacts with the volume element at the origin and the one at
position r, as shown in Fig. 1.7(a). The phase difference between two successive crests is 2π. The
phase difference between the two volume elements is 2π multiplied by the ratio of r, projected onto
the incident direction, and the wavelength. This is nothing other than the scalar product of the two
vectors k and r. The simplicity of this expression is one of the reasons why it is so convenient to use
the wavevector k to describe the incident wave. In the vicinity of the observation point X in Fig. 1.6,
the scattered wave is locally like a plane wave with wavevector k′ . The phase difference, between the
scattered wave from a volume element around the origin and one around r is −k′ · r. The resulting
phase difference is thus
∆φ(r) = (k − k′ ) · r = Q · r
where
Q = k − k′ (1.10)

Q = k − k′ is known as the wavevector transfer or scattering vector. The scattering events depicted
in Fig. 1.7 are elastic, with |k|= |k′ |, so that from the scattering triangle we have |Q| = 2|k| sin θ =
(4π/λ) sin θ. As we shall see, Q is the natural variable to describe elastic scattering processes and is
usually expressed in units of Å−1 .
Thus a volume element dr at r will contribute an amount −r0 ρ(r)dr to the scattered field with a
phase factor of eiQ·r . The total scattering length of the atom is

Z
−r0 f 0 (Q) = −r0 ρ(r) eiQ·r dr (1.11)

where f 0 (Q) is known as the atomic form factor. In the limit that Q → 0 all of the different volume
elements scatter in phase so that f 0 (Q = 0) = Z, the number of electrons in the atom. As Q increases
from zero the different volume elements start to scatter out of phase and consequently f 0 (Q → ∞) = 0.
The right hand side of Eq. (1.11) is recognizable as a Fourier transform. Indeed one of the recurrent
themes of this book is that the scattering length may be calculated from the Fourier transform of the
distribution of electrons in the sample3 . It should be clear that to calculate the scattered intensity we
have to evaluate Eq. (1.11) and multiply by its complex conjugate (see Eq. (1.2) and accompanying
discussion).
Atomic electrons are of course governed by quantum mechanics, and have discrete energy levels.
The most tightly bound electrons are those in the K shell, which have energies comparable to those of
a typical X-ray photon. If the X-ray photon has an energy much less than the binding energy of the K
shell, the response of these electrons to an external driving field is reduced by virtue of the fact that
they are bound. Electrons in shells that are less tightly bound (L, M, etc.) will be able to respond to the
3
The reader is reminded of the definition and properties of Fourier transforms in Appendix E.