C

Quantization of the electromagnetic field

The cross-section for either scattering or absorption is evaluated from time-dependent perturbation
theory, as outlined in Appendix A. In any perturbation problem it is of course first necessary to specify
completely the non-interacting Hamiltonian, H0 , of the system, before the effect of the perturbing
Hamiltonian, HI , may be calculated. For the scattering or absorption of an X-ray this amounts to
establishing a quantum mechanical description of both the electromagnetic field and the sample. The
former may well be unfamiliar to many readers and here we explain briefly how this is achieved.
The starting point in quantizing the electromagnetic field is the classical expression for its energy
in terms of the electric and magnetic fields, both of which may be derived from the vector potential A
(Appendix B). When seeking a quantum mechanical description of the electromagnetic field it would
therefore seem natural to focus on A. Indeed quantizing the electromagnetic field amounts to quantizing
the vector potential. It also transpires that the Hamiltonian, HI , that describes the interaction of the
X-ray and the sample, is a simple function of A. As a consequence the matrix elements of HI that
enter into the perturbation theory may be calculated readily, and in the last section we work through the
example of the Thomson cross-section.

Classical energy density of the radiation field

The total energy of the electromagnetic field in free space is
Z h
1 i
Erad = ǫ0 hE2 i + µ0 hH2 i dV
2
Z V
= ǫ0 hE2 i dV
V

Here it is assumed that the field is confined to some volume V, and ǫ0 hE2 i =µ0 hH2 i, where the brackets
h· · · i indicate a temporal average. The E field is related to vector potential A through

∂A
E=−
∂t
Elements of Modern X-ray Physics, Second Edition. Jens Als-Nielsen and Des McMorrow
© 2011 John Wiley & Sons, Ltd. Published 2011 by John Wiley & Sons, Ltd.
356 Quantization of the electromagnetic field

The most general approach for dealing with A would be to write it as a Fourier sum of plane waves.
For clarity we shall consider just one term in this series and write the vector potential as
h i
A(r, t) = ε̂A0 ak ei(k·r−ωt) + a∗k e−i(k·r−ωt) (C.1)

The direction of A is specified by the polarization unit vector ε̂, and in addition to the amplitude
coefficients ak we have introduced a normalization factor A0 . The electric field is
h i
E = ε̂A0 (iω)ak ei(k·r−ωt) − (iω)a∗k e−i(k·r−ωt)

and its modulus squared is

E2 = E · E = 4ω2 A20 a∗k ak cos2 (k · r − ωt)

The temporal average of the modulus squared of the field is

hE2 i = 2ω2 A20 a∗k ak

since hcos2 (k · r − ωt)i= 12 . The total energy of the electromagnetic field is therefore equal to

Erad = ǫ0 2ω2 A20 a∗k ak V

= ~ωa∗k ak (C.2)

where A0 has been chosen to be equal to

s
~
A0 =
2ǫ0 Vω

It should be emphasized that so far we have only considered one particular k and polarization state, and
in general we would need to sum over these quantities to obtain the total energy.

Quantization of the vector potential, A

The normalization constant A0 in the last section was chosen to reveal the formal equivalence of the
Hamiltonian of the electromagnetic field to that of the harmonic oscillator. The quantum mechanical
Hamiltonian of the latter is usually written in the form
 
1
Hosc = ~ω a† a + (C.3)
2

A direct comparison of this expression with Eq. (C.2) should at least make this equivalence plausible
(apart from the additive term of 12 which we shall return to later). The reason for this equivalence is
that when our form for A (Eq. (C.1)) is substituted into the wave equation, the coefficients ak obey the
equation of motion of the harmonic oscillator.
The operators a and a† appearing in Eq. (C.3) are known as the annihilation and creation operators,
since they have the properties √
a|ni = n|n − 1i (C.4)
 357

and √
a† |ni = n + 1|n + 1i (C.5)
where |ni is an eigenfunction of Hosc with an eigenvalue
1
En = ~ω(n + )
2
and n is an integer 0, 1, 2, · · · . We thus quantize the electromagnetic field by requiring that the coeffi-
cients ak in Eq. (C.2) become operators that obey the same commutation relations as the annihilation
and creation operators of the harmonic oscillator. Here we must extend our notation to allow for the
different possible polarization states of the photon, so that the commutation relations read
h i
auk , a†vk′ = δkk′ δuv
h i h i
auk , avk′ = a†uk , a†vk′ = 0

where the first subscript, u or v, refers to the polarization state.

The Hamiltonian of the radiation field is thus given by
XX
Hrad = ~ωk a†uk auk
u k

For a given value of k and polarization u, the eigenfunctions of Hrad are |nuk i, where nuk is the number
of photons in that state. The nuk ’s are sometimes referred to as the occupation numbers. It follows that
a general state of the field, involving photons with different wavevectors and polarizations is a product
of such states, since they are all independent. In writing down Hrad we have followed convention and
P P
set the energy of the vacuum state (all nuk =0) equal to 21 u k ~ωk .
The operator form of the vector potential is

s
XX ~ h i
A(r, t) = ε̂u auk ei(k·r−ωt) + a†uk e−i(k·r−ωt) (C.6)
u k
2ǫ0 Vωk

The interaction Hamiltonian, HI

In the absence of any interaction between the photon field of the X-ray and the electrons in the sample
the Hamiltonian is
H0 = He + Hrad
where He refers to the electrons and Hrad is given above. The eigenfunctions of H0 are a product of
the eigenfunctions of He and Hrad .
Classically, it can be shown1 that the interactions between an electromagnetic field and a charge
q may be allowed for by replacing the momentum p by p − qA. For simplicity we shall consider the
1 The substitution of p by p − qA can be shown to produce the correct equation of motion of a charged particle in an

electromagnetic field.
358 Quantization of the electromagnetic field

case of a free electron for which He = p2 /2m. This allows us to write down the Hamiltonian of the
interacting system as
(p + eA)2
H= + Hrad
2m
p2 eA · p e2 A2
= + + + Hrad
2m m 2m
= He + HI + Hrad
where HI is the interacting Hamiltonian

eA · p e2 A2
HI = + (C.7)
m 2m

The first term is linear in A and gives rise to absorption of the X-ray, whereas the second is quadratic
in A and gives rise to scattering, as we shall now explain.
The operator for A is linear in the annihilation and creation operators. Hence when it acts on a state
|nuk i it can either destroy or create a photon in that state. Absorption corresponds to the former, and it is
clear that the first term in HI results in absorption. Scattering on the other hand involves the destruction
of a photon in one state (labelled by k say), and the creation of a new photon in a state (labelled by
k′ ). This process then requires a combination of operators of the form a†k′ ak to act on the product states
|nk i|nk′ i that are the eigenfunctions of Hrad . Such combinations of operators can arise only from a term
in the Hamiltonian that is quadratic in A. In the next section we explicitly calculate the cross-section
arising from the second term in HI , and show that it is equivalent to the classical Thomson scattering
of an X-ray by an electron. The absorption cross-section is derived from the first term in HI in Chapter
7.

Thomson scattering cross-section

We imagine that an X-ray photon of wavevector k and polarization ε̂u is scattered by an electron to k′
and ε̂v . The scattering is restricted to be elastic, i.e. ~ω = ~ω′ , which is equivalent to assuming that the
energy of the X-ray is large compared to the binding energy of the electron. As the scattering is elastic
the electron can be assumed to remain in its ground state |pi. The eigenfunction of the photon field on
the other hand changes so that one photon is removed from the state |nuk i, and one photon added to the
state |nvk′ i. Before the scattering event we have nuk = 1 and nvk′ = 0, and after nuk = 0 and nvk′ = 1. We
therefore write the initial and final eignestates of H0 as |ii=|pi|u1, v0i and | f i=|pi|u0, v1i respectively.
The scattering cross-section is calculated by taking the matrix element of the second term in Eq.
(C.7), and using Eq. (A.7). The matrix element we need to evaluate is
e2 2
Mi f = hu0, v1|hp| A |pi|u1, v0i
2m
When we form the square of A from Eq. (C.6) there will be cross terms in the annihilation and creation
operators such as a†ikk a jkl , which can destroy a photon in state |n jkl i, and create one in state |nikk i. These
terms give rise to scattering and, as there are two such terms, the matrix element becomes
 
e2 ~ ε̂u · ε̂v ′ ′
Mi f = hp|ei (ω−ω )t ei (k−k )·r |pi
2mǫ0 V (ωω′ ) 21
 359

The cross-section given in Eq. (A.7) involves an integral with respect to the final X-ray energy, E f ≡
~ω′ , whereas the matrix element is given in terms of ω. We therefore rewrite Eq. (A.7) as
!  2 Z
dσ V 1
= |Mi f |2 ω′2 δ(ω − ω′ ) dω′
dΩ 2π ~2 c4

Taking the square of the matrix element and inserting it into the above gives

! !2
dσ e2
ε̂u · ε̂v 2 | f (Q)|2
 
=
dΩ 4πǫ0 mc2

This is the Thomson scattering cross-section, and should be compared with the classical result given
in Chapter 1. The two descriptions are entirely equivalent, with the polarization factor given by P =
ε̂u · ε̂v 2 , and the form factor by
 
f (Q) = hp|ei Q·r |pi

Further reading
Quantum Field Theory, F. Mandl and G. Shaw (John Wiley & Sons, 1996)