12 Scattering in three dimensions

12.1 Cross sections and geometry

Most experiments in physics consist of sending one particle to collide with another, and looking
at what comes out.
The quantity we can usually measure is the scattering cross section: by analogy with classical
scattering of hard spheres, we assuming that scattering occurs if the particles ‘hit’ each other. The
cross section is the apparent ‘target area’. The total scattering cross section can be determined by
the reduction in intensity of a beam of particles passing through a region on ‘targets’, while the
differential scattering cross section requires detecting the scattered particles at different angles.
We will use spherical polar coordinates, with the scattering potential located at the origin and the
plane wave incident flux parallel to the z direction. In this coordinate system, scattering processes
dσ
are symmetric about φ, so dΩ will be independent of φ.
We will also use a purely classical concept, the impact parameter b which is defined as the distance
of the incident particle from the z-axis prior to scattering.
S(k)

δΩ

I(k)
θ z
φ

Figure 11: Standard spherical coordinate geometry for scattering

12.2 The Born Approximation

We can use time-dependent perturbation theory to do an approximate calculation of the cross-

section. Provided that the interaction between particle and scattering centre is localised to the
region around r = 0, we can regard the incident and scattered particles as free when they are far
from the scattering centre. We just need the result that we obtained for a constant perturbation,
Fermi’s Golden Rule, to compute the rate of transitions between the initial state (free particle of
momentum p) to the final state (free particle of momentum p0 ).
The Hamiltonian for a single particle being scattered by a fixed potential as
p̂2
Ĥ = Ĥ0 + V̂ (r) where Ĥ0 = , the kinetic energy operator
2m
and treat the potential energy operator, V̂ (r), as the perturbation inducing transitions between
the eigenstates of Ĥ0 , which are plane waves.
If we label the initial and final plane-wave states Φin = exp(ik.r−iωt) and Φscat = exp(ik0 .r−iω 0 t)
by their respective wave-vectors, then Fermi’s Golden Rule for the rate of transitions is
2π 0
R= |hk |V̂ |ki|2 g(Ek )
h̄
where g(Ek ) is the density of final states; g(Ek )dEk is the number of final states with energy in
the range Ek → Ek + dEk .
The quantity hk0 |V̂ |ki is known as the matrix element of the perturbation and is usually abbrevi-
ated thus Z Z Z
0
Vk0 k ≡ hk |V̂ |ki = u∗k0 (r) V (r)uk (r) dτ.
The time variation has been suppressed here. For constant potential, the only non-zero terms
come from ω = ω 0 : elastic scattering. For a time oscillating potential (e.g. V (r) sin ω0 t) the
non-zero contribution comes from ω = ω 0 ± ω0 : inelastic scattering where the scattered particle
gains/loses a quantum of energy from/to the system providing the potential.

12.3 Box Normalisation and Density of Final States

Plane-wave states have wavefunctions of the form: uk,ω (r) = C exp(i(k.r − ωt)) with C a normal-
isation constant. Because plane-wave states are not properly normalisable we employ the trick
of normalising them in a large (relative to potential range) cubic box of side L with periodic
boundary conditions. We then take the limit L → ∞ at the end of the calculation.
Thus we require that
Z Z Z Z Z Z
u∗k,ω (r)uk,ω (r) dτ = |C|2 dτ = |C|2 L3 = 1
box box

giving for the normalised eigenfunctions: uk,ω (r) = L−3/2 exp(ik.r − ωt)
Of course, enclosing the system in a finite box has the consequence that the allowed momentum
eigenvalues are no longer continuous but discrete. With periodic boundary conditions
L L
u(− , y, z) = u( , y, z), etc.
2 2
the momentum eigenvalues are forced to be of the form
2πh̄
p ≡ h̄k = (nx , ny , nz ), with nx , ny , nz = 0, ±1, ±2, . . .
L
For sufficiently large L, we can approximate the continuous spectrum arbitrarily closely.
Any possible final-state wave-vector, k, corresponds to a point in wave-vector space with coordi-
nates (kx , ky , kz ). The points form a cubic lattice with lattice spacing 2π/L. Thus the volume of
k–space per lattice point is (2π/L)3 , and the number of states in a volume element d3 k is
µ ¶3 µ ¶3
L L
d3 k0 = k 2 dk dΩ
2π 2π

We require g(Ek ), the density of states per unit energy, where: Ek = h̄2 k 2 /2m is the energy
corresponding to wave-vector k0 . Now, the wave-vectors in the range k0 → k0 + d3 k0 correspond
to the energy range Ek → Ek + dEk , so that
µ ¶3
L
g(Ek ) dEk = k 2 dk dΩ
2π
is the number of states with energy in the desired interval and with wave-vector, k0 , pointing into
the solid angle dΩ about the direction (θ, φ). Noting that dEk = (h̄2 k/m) dk yields the final result
for the density of states,

L3 mk
g(Ek ) = dΩ
8π 3 h̄2
12.4 Incident and Scattered Flux

The box normalisation corresponds to one particle per volume L3 , so that the number of particles
crossing unit area perpendicular to the beam per unit time is just given by the magnitude of the
incident velocity divided by L3 :
|p|/m h̄k
incident flux = 3
=
L mL3

Using the Golden Rule, we have that the rate of transitions between the initial state of wave-vector
k and final states whose wave-vectors k0 lie in the element of solid angle dΩ about the direction
(θ, φ) of the wave-vector k0 , is given by
2π L3 mk
R= |Vk0 k |2 3 2 dΩ
h̄ 8π h̄
but this is just the number of particles scattered into dΩ per unit time. To get the scattered flux
we simply divide by dΩ to get the number per unit time per unit solid angle.

12.5 The Differential Cross-Section

We now have all the ingredients, the scattered flux and the incident flux, to compute the cross-
section:
dσ scattered flux mL3 2π 3
2 L mk
≡ = |Vk 0k|
dΩ incident flux h̄k 0 h̄ 8π 3 h̄2
Noting that, for elastic scattering, k 0 = k, we obtain finally the so-called Born approximation for
the differential cross-section:
dσ m2 ¯ ¯2
¯ ¯
= 2 4 L6 ¯hk0 |V̂ |ki¯
dΩ 4π h̄
where the matrix element Vk0 k ≡ hk0 |V̂ |ki is given by
1 Z Z Z
hk0 |V̂ |ki = V (r) exp (−iχ.r) dτ
L3
with χ ≡ k0 − k, the so-called wave-vector transfer . Thus the required matrix element in the Born
approximation is just the 3-dimensional Fourier transform of the potential energy function. The
total scattering cross section is simply:
Z Z Z
dσ dσ
σT = dΩ = sin θ dθdφ
dΩ dΩ
Observe that the final result for the differential cross-section is independent of the box size, L,
which we used to normalise the plane-wave states.

12.6 Further Simplification to 1D for Conservative, Central Potential

Consider a central potential V (r) = V (|r|) where energy is conserved |k0 |2 = |k|2 . Here χ is a
vector of length 2k sin 2θ where θ is the scattering angle.
We can make some progress with the matrix element integral if we choose a polar coordinate
system with χ along the z-axis, so that χ.r = χr cos θ. Since we are trying to integrate over all
space this change does not affect the limits of the integral.
Z 2π Z +1 Z ∞
Vk0 k = dφ V (r)e−iχr cos θ r2 drd(cos θ)
0 −1 0

Z ∞ −iχr
e − eiχr 4π Z ∞
= 2π V (r)r2 dr = rV (r) sin(χr)dr
0 −iχr χ 0
But since |k| = |k0 |, |χ| = 2k sin 2θ , Whence we obtain the most useful form of the Born approxi-
mation:
¯Z ¯2
dσ m2 ¯ ∞ θ ¯¯
¯
= ¯ rV (r) sin(2kr sin )dr¯
dΩ (k sin 2θ )2 h̄4 ¯ 0 2 ¯

Thus the scattering cross-section is independent of φ (due to cylindrical symmetry of the problem).
Note that this shows that the differential cross section does not depend on scattering angle and
beam energy independently, but on a single parameter χ. By using a range of energies for the
incoming particles, k, this dependence can be used to test whether experimental data can be well
described by the Born Approximation.
dσ
The most common use of the 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.

12.7 Example of Born Approximation

Consider scattering of particles interacting via a 3D square well potential: V (r < a) = V0 ;

V (r > a) = 0.
The integral required here is then (with χ = 2k sin 2θ ):
Z a " #a
sin(χr) − χr cos(χr)
rV0 sin(χr)dr =
0 χ2 0

whence:
" #2 " #2
dσ 2µV0 sin(χa) − χa cos(χa)
=
dΩ χh̄2 χ2

Using a Maclaurin expansion, the low energy limit is:

" #2 · ¸
dσ 2µV0 1 1
= 1 − χ2 a2
dΩ χh̄2 9 5

From integrating over θ and φ the low and high energy limits for the total cross section are
· ¸2 " #2 · ¸2 " #2
µ V0 a3 µ V0 a3 8 2 2 2 4 4
σ(E → ∞) = 2π 2 σ(E → 0) = 2π 2 (k a − k a + ...)
h̄ ka h̄ ka 9 5

12.8 General Notes on Scattering in the Born Approximation

The square well illustrates some general feature of scattering in the Born approximation:

• Born approximation is based on perturbation theory, so works best for high energy particles.
• Scattering depends on V02 , so both attractive and repulsive potentials behave the same.
• At high energy, cross section is inversely proportional to the energy (E = h̄2 k 2 /2m)
• Dependence on k and θ arises only through the combination χ = 2k sin 2θ . Thus as energy
increases, the scattering angle θ is reduced and the scattered beam becomes more peaked in
the ‘straight on’ direction.
• Angular dependence depends on the range of the potential a but not on the strength V0 .
• Total cross section depends on both range a and depth V0 of the potential.