Scattering
Scattering
Scattering
Contents
1 Scattering theory 2
1.1 Scattering amplitudes . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 The Born approximation . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Virtual Particles 5
Appendices 10
.A Beyond Born§ 10
1
Non-relativistic scattering
1 Scattering theory
We are interested in a theory that can describe the scattering of a particle from a
potential V (x). Our Hamiltonian is
H = H0 + V.
p2
H0 = .
2m
In the absence of the potential V the solutions of the Hamiltonian could be written
as the free-particle states satisfying
H0 |φi = E|φi.
We define the eigenstates of H such that in the limit where the potential disappears
(V →0), we have |ψi→|φi, where |φi and |ψi are states with the same energy
eigenvalue. (We are able to do this since the spectra of both H and H + V are
continuous.)
1
|ψ (±) i = |φi + V |ψ (±) i . (2)
E − H0 ± i
To calculate scattering amplitudes we are going to have to use both the position
and the momentum basis, the incoming beam is (almost) a momentum eigenstate,
and V is a function of position x. If |φi stands for a plane wave with momentum
~k then the wavefunction can be written
eik·x
hx|φi = 3 .
(2π) 2
We can express (2) inR the position basis by bra-ing through with hx| and inserting
the identity operator d3 x0 |x0 ihx0 |
Z D 1 E
hx|ψ i = hx|φi + d3 x0 x
(±)
x0 hx0 |V |ψ (±) i. (3)
E − H0 ± i
The wave function (4) is a sum of two terms. The first is the incoming plane wave.
For large r = |x| the spatial dependence of the second term is e±ikr /r. We can now
understand the physical meaning of the |ψ (±) i states; they represent outgoing (+)
and incoming (−) spherical waves respectively. We are interested in the outgoing
(+) spherical waves – the ones which have been scattered from the potential.
We want to know the amplitude of the outgoing wave at a point x. For practical
experiments the detector must be far from the scattering centre, so we may assume
|x| |x0 |. x
k0 = kr̂.
It’s safe to replace the |x − x0 | in the denominator in the integrand of (4) with just
r, but the phase term will need to be replaced by r − r̂ · x0 . So we can simplify the
wave function to
1 2m eikr
Z
r large 0 0
hx|ψ (+) i −−−−→ hx|ki − d3 x0 e−ik ·x V (x0 )hx0 |ψ (+) i
4π ~ r2
This makes it clear that we have a sum of an incoming plane wave and an outgoing
spherical wave with amplitude f (k0 , k) given by
1 2m
f (k0 , k) = − (2π)3 2 hk0 |V |ψ (±) i. (5)
4π ~
We will ignore the interference between the first term which represents the orig-
inal ‘plane’ wave and the second term which represents the outgoing ‘scattered’
wave, which is equivalent to assuming that the incoming beam of particles is only
approximately a plane wave over a region of dimension much smaller than r.
We then find that the partial cross-section dσ — the number of particles scattered
into a particular region of solid angle per unit time divided by the incident flux2 —
is given by
r2 |jscatt |
dσ = dΩ = |f (k0 , k)|2 dΩ.
|jincid |
This means that the differential cross section is given by the simple result
dσ
= |f (k0 , k)|2 .
dΩ
The differential cross section is simply the mod-squared value of the scattering
amplitude.
2 Remember that the flux density is given by j = ~
2im
[ψ ∗ ∇ψ − ψ∇ψ ∗ ].
If the potential is weak we can assume that the eigenstates are only slightly modified
by V , and so we can replace |ψ (±) i in (5) by |ki.
1 2m
f (1) (k0 , k) = − (2π)3 2 hk0 |V |ki. (6)
4π ~
This is known as the Born approximation. Within this approximation we have the
simple result that
f (1) (k0 , k) ∝ hk0 |V |ki.
Up to some constant factors, the scattering amplitude is found by squeezing the
perturbing potential V between incoming and the outgoing momentum eigenstates
of the free-particle Hamiltonian.
Expanding out (6) Rin the position representation (by insertion of a couple of com-
pleteness relations d3 x0 |x0 ihx0 |) we can write
Z
1 2m 0 0
f (1) (k0 , k) = − 2
d3 x0 ei(k−k )·x V (x0 ).
4π ~
This result is telling us that scattering amplitude is proportional to the 3d Fourier
transform of the potential. By scattering particles from targets we can measure
dΩ , and hence infer the functional form of V (r). This result is used, for example,
dσ
2 Virtual Particles
One of the insights of subatomic physics is that at the microscopic level forces are
caused by the exchange of force-carrying particles. For example the Coulomb force
between two electrons is mediated by excitations of the electromagnetic field – i.e.
photons. There is no real ‘action at a distance’. Instead the force is transmitted
between the two scattering particles by the exchange of some unobserved photon or
photons. The mediating photons are emitted by one electron and absorbed by the
other. It’s generally not possible to tell which electron emitted and which absorbed
the mediating photons – all one can observe is the net effect on the electrons.
e−
Other forces are mediated by other force-carrying particles. In each case the mes-
senger particles are known as virtual particles. Virtual particles are not directly e−
γ
observed, and have properties different from ‘real particles’ which are free to prop-
agate. e−
To illustrate why virtual particles have unusual properties, consider the elastic scat- e−
tering of an electron from a nucleus, mediated by a single virtual photon. We can
assume the nucleus to be much more massive than the electron so that it is approx-
imately stationary. Let the incoming electron have momentum p and the outgoing,
scattered electron have momentum p0 . For elastic scattering, the energy of the
Eγ = 0 e− e−
pγ = ∆p = p − p.0 γ
The exchanged photon carries momentum, but no energy. This sounds odd, but is
nevertheless correct. What we have found is that for this virtual photon, Eγ2 6= p2γ .
The particular value Eγ = 0 is special to the case we have chosen, but the general
result is that for any virtual particle there is an energy-momentum invariant3 which
is not equal to the square of its mass
P · P = E 2 − p · p 6= m2 .
Such virtual particles do not satisfy the usual energy-momentum invariant and are
said to be ‘off mass shell’.
Note that we would not have been able to escape this conclusion if we had taken the
alternative viewpoint that the electron had emitted the photon and the nucleus had
absorbed it. In that case the photon’s momentum would have been pγ = −∆p.
The square of the momentum would be the same, and the photon’s energy would
still have been zero.
These exchanged, virtual, photons are an equally valid solution to the (quantum)
field equations as are the more familiar travelling-wave solutions of ‘real’ on-mass-
shell photons. It is interesting to realise that all of classical electromagnetism is
actually the result of very many photons being exchanged.
g 2 e−µr
V (r) = . (7)
4π r
This is known as the Yukawa potential. The constant g 2 tells us about the depth
of the potential, or the size of the force. When µ = 0 (7) has the familiar 1/r
dependence of the electrostatic and gravitational potentials. When µ is non-zero,
the potential also falls off exponentially with r, with a characteristic length of 1/µ.
∂
2
− ∇ 2
+ µ2
ϕ(r, t) = 0. (8)
∂t2
3 We used sans serif capitals P to indicate Lorentz four-vectors P = (E, p , p , p ). The dot
x y z
product of two Lorentz vectors A · B = a0 b0 − a1 b1 − a2 b2 − a3 b3 is a Lorentz invariant scalar.
This is the relativistic wave equation for spin-0 particles. The plane-wave solutions
to (8) are
We can go some way towards interpreting this result as the exchange of a virtual
particle as follows. We justify the two factors of g as coming from the points where
a virtual photon is either created or annihilated. This vertex factor g is a measure
of the interaction or ‘coupling’ of the exchanged particle with the other objects.
There is one factor of g the point of creation of the virtual particle, and another
one at the point where it is absorbed.
The other important factor in the scattering amplitude (9) is associated with the
momentum and mass of the exchanged particle:
1
−
µ2 + |∆k|2
1
(10)
P · P − µ2
(Q1 e)(Q2 e)
VEM = .
4π0 r
For scattering from a Coulomb potential we can therefore use the Yukawa result
(9) by making the substitution
g2 Q1 Q2 e2
⇒ .
4π 4π0
This identification shows that the vertex factors g are just dimensionless measures
of the charges of the particle. The vertex factor for a charge Qe is QgEM where
2
gEM 1
= αEM ≈ .
4π 137
P · P − µ2 = E 2 − p2 − µ2
= 0 − |∆k|2 − µ2
= − µ2 + |∆k|2
Note that the propagator (13) becomes singular as the particle gets close to its
mass shell. i.e. as P · P → µ2 . It is only because the exchanged particles are off
their mass-shells that the result is finite.
The identification of the vertex factors and propagators will turn out to be very useful
when we later try to construct more complicated scattering processes. In those cases
we will be able to construct the most important features of the scattering amplitude
by writing down:
1
|ψ (±) i = |φi + V |ψ (±) i
E − H0 ± i
dσ
= |f (k0 , k)|2
dΩ
P · P = E 2 − p · p 6= m2
• The scattering amplitude contains a vertex factors g for any point where
particles are created or annihilated
1
P · P − µ2
Non examinable
To see how things develop if we don’t want to rashly assume that |ψ ± i ≈ |φi it is
useful to define a transition operator T such that
V |ψ (+) i = T |φi
The operator
1
(13)
E − H0 + i k V
When is this condition likely to hold? Remember that the Yukawa potential was
proportional to the square of a dimensionless coupling constant ∝ g 2 . If g 2 1
then successive applications of V introducing higher and higher powers of g and can
g2 e2 1
=α= ≈ .
4π 4π0 ~c 137
Since α 1, we can usually get away with just the first term of (12) for electric
interactions (i.e. we can use the Born approximation).