PH 303 Supervised Learning Project

Department of Physics
Indian Institute of Technology Bombay

Quantum Electrodynamics

Name : Gopal Parwani

Roll No. : 200260018
Supervisor : Prof. Asmita Mukherjee

1
Contents

1 The Klein-Gordon Equation 4

1.1 Non-relativistic Quantum Mechanics . . . . . . . . . . . . . . 4
1.2 Four-Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 The Klein-Gordon Equation . . . . . . . . . . . . . . . . . . 6
1.4 Interpretation of the Negative Energy Solutions . . . . . . . . 7
1.5 Non-relativistic Perturbation Theory . . . . . . . . . . . . . . 9
1.6 Rules to Calculate Scattering Amplitudes . . . . . . . . . . . 12

2 QED of Spinless Particles 13

2.1 ‘Spinless’ Electron in an Electromagnetic Field . . . . . . . . 13
2.2 ‘Spinless’ Electron-Muon Scattering . . . . . . . . . . . . . . 14
2.3 More Rules for Feynman Diagrams . . . . . . . . . . . . . . . 15
2.4 Scattering Cross-section . . . . . . . . . . . . . . . . . . . . . 15
2.5 The Decay Rate in Terms of Invariant Amplitude . . . . . . . 17
2.6 ‘Spinless’ Electron-Electron Scattering . . . . . . . . . . . . . 18
2.7 ‘Spinless’ Electron-Positron Scattering . . . . . . . . . . . . . 18
2.8 Mandelstam Variables and Crossing . . . . . . . . . . . . . . 19

3 The Dirac Equation 20

3.1 Dirac Equation in Covariant Form . . . . . . . . . . . . . . . 21
3.2 Continuity Equation . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Free-Particle Spinors . . . . . . . . . . . . . . . . . . . . . . 22
3.4 Intrinsic Angular Momentum . . . . . . . . . . . . . . . . . . 24
3.5 Antiparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.6 Bilinear Covariants . . . . . . . . . . . . . . . . . . . . . . . 26

4 QED of Spin- 12 Particles 28

4.1 Electron in an Electromagnetic Field . . . . . . . . . . . . . . 28
4.2 Møller Scattering . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3 Electron-Muon Scattering and Crossing . . . . . . . . . . . . 32
4.4 Propagators . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2
 4.4.1 The Propagator for Spinless Particles . . . . . . . . . 33
4.4.2 The Electron Propagator . . . . . . . . . . . . . . . . 34
4.4.3 The Photon Propagator . . . . . . . . . . . . . . . . . 34
4.4.4 Propagator for Massive Vector Particles . . . . . . . . 35
4.5 Summary of the Feynman Rules for QED . . . . . . . . . . . 36

3
Chapter 1

The Klein-Gordon Equation

1.1 Non-relativistic Quantum Mechanics

Being inspired by the work of de Broglie on the matter waves, Schrödinger,
in 1926, gave the propagation equation of these waves using the Hamiltonian
principle and developed a new theory, which was called ‘Wave Mechanics’.
Heisenberg, independently, in 1925, developed a more abstract theory, which
he called ‘Quantum Mechanics’. Later, Schrödinger showed that both theo-
ries are just a mathematical transcription of one another.
Heuristically, the Schrödinger equation can be derived from the non-
relativistic energy-momentum relation,

p2
E= (1.1)
2m
by substituting corresponding operators,
∂
E −→ iℏ , and p −→ −iℏ∇ (1.2)
∂t
to obtain an equation, the solution of which is ψ(x, t), the wavefunction,
(ℏ ≡ 1 from now onwards)
∂ψ 1 2
i + ∇ ψ=0 (1.3)
∂t 2m
The wavefunction ψ(x, t) is a mathematical description of the particle that
assigns a complex number to each point in spacetime. The interpretation of
this complex wavefunction, due to Born, is that the probability of finding
the particle in a volume element d3 x is |ψ|2 d3 x. Therefore, the probability
density of finding the particle at (x, t) is given as,

ρ = |ψ|2 (1.4)

4
The decrease in the number of particles per unit time in a volume V is equal
to the total number of particles going out of the volume V per unit time,
i.e., Z Z Z
∂
− ρ dV = j · n̂ dS = ∇ · j dV (1.5)
∂t V S V
where S is the surface enclosing the volume V and n̂ is the unit vector normal
to surface area element dS in the outward direction. Therefore,
∂ρ
+∇·j =0 (1.6)
∂t
This is a continuity equation. On multiplying (1.3) by −iψ ∗ , taking the
conjugate of (1.3) and multiplying it by −iψ, and subtracting the latter
from the former we get,
∂ρ i
− (ψ ∗ ∇2 ψ − ψ∇2 ψ ∗ ) = 0 (1.7)
∂t 2m
On comparing (1.6) and (1.7),
i
j=− (ψ ∗ ∇ψ − ψ∇ψ ∗ ) (1.8)
2m
A solution of (1.3), using the separation of variables, can be given by
ψ = N ei(p·x−Et) (1.9)
This is the wavefunction of a free particle having energy E and momentum
p with
p
probability density, ρ = |N |2 , and current density, j = |N |2 (1.10)
m

1.2 Four-Vectors
A four-component quantity that transforms between two inertial frames,
according to Lorentz transformation, is called a four-vector.


Contravariant four-vectors are given as Aµ ≡ A0 , A
.
Covariant four vectors are given as Aµ ≡ A0 , −A = gµν Aµ where gµν is
the metric tensor defined as
g00 = 1, gµµ = −1, for µ = 1, 2, 3, and the other components are zero
(1.11)
The scalar product of two four-vectors, A · B is a Lorentz invariant quantity,
i.e., equal in all inertial frames.
A · B ≡ A0 B 0 − A · B (1.12)
5
 A · B = Aµ B µ = Aµ Bµ = gµν Aµ B ν = g µν Aµ Bν (1.13)
The contravariant form of important four-vectors that we are going to use are

Position four-vector, (ct, x) = (x0 , x1 , x2 , x3 ) ≡ xµ (1.14)

 
E
Momentum four-vector, , p = (p0 , p1 , p2 , p3 ) ≡ pµ (1.15)
c
 
∂
Gradient four-vector, , −∇ ≡ ∂ µ (1.16)
∂t
Note that spatial vectors will be bold, and four-vectors won’t be in normal
font.
We also have D’Alembertian operator
□2 ≡ ∂µ ∂ µ (1.17)

1.3 The Klein-Gordon Equation

The Schrödinger equation in (1.3) corresponds to the classical energy-momentum
relation, which doesn’t obey Lorentz covariance. When we use relativistic
energy-momentum relation instead,
E 2 = p2 + m2 (1.18)
then, after substituting in operator form from (1.2), we get
∂ 2ϕ
− 2
+ ∇2 ϕ = m2 ϕ (1.19)
∂t
The equation (1.19) is the Klein-Gordon equation.
On multiplying (1.19) by −iϕ∗ and its complex conjugate by −iϕ, and sub-
tracting one from another gives the relativistic analog of the continuity equa-
tion (1.7)
∗
  
∂ ∂ϕ ∂ϕ
i ϕ∗ −ϕ +∇ · [−i (ϕ∗ ∇ϕ − ϕ∇ϕ∗ )] = 0. (1.20)
∂t ∂t ∂t | {z }
| {z } j
ρ

Identification of the probability and the flux densities by comparing them

with the general continuity equation can be made.
For a free particle (with E and p as the energy and momentum, respectively),
when we plug the solution of the Klein-Gordon equation (again by variable
separable method)
ϕ = N eip·x−iEt (1.21)
6
then,
ρ = i(−2iE)|N |2 = 2E|N |2 (1.22)
j = −i(2ip)|N |2 = 2p|N |2 (1.23)
When we write the results in four-vector notation, then the Klein-Gordon
equation becomes
□2 + m2 ϕ = 0


(1.24)
The probability and the flux densities form a four-vector.

j µ = (ρ, j) = i (ϕ∗ ∂ µ ϕ − ϕ∂ µ ϕ∗ ) (1.25)

and satisfy the covariant continuity relation

∂µ j µ = 0 (1.26)

The free particle solution can be written as

ϕ = N e−ip·x (1.27)

and
j µ = 2pµ |N |2 (1.28)
When we substitute the solution (1.27) in the Klein-Gordon equation, we
get the energy eigenvalues for a free particle as

1/2
E = ± p2 + m2 (1.29)

We have negative energy solutions along with positive ones as well. The
consequences of this negative energy solution are:
1. If the free particle is in some energy state, it can always go to lower
energy states.
2. For negative energy solutions, the probability density is negative from
(1.22)
We cannot ignore these negative energy solutions by considering them inad-
missible because we want a set of solution states, which is complete.

1.4 Interpretation of the Negative Energy Solutions

We can see that we are getting these negative energy solutions because we
have the square of energy operator or square of ∂/∂t in the Klein-Gordon
equation. Dirac, in 1927, gave a relativistic wave equation, which is linear
7
in ∂/∂t and ∇. He got rid of the negative probability densities but not the
negative energy solutions. In addition to this, the equation Dirac derived
was able to explain spin- 12 , which is regarded as the biggest triumph of the
‘Dirac Equation’, which we are going to consider later. Dirac postulated that
all the negative energy states are filled, and the vacuum is an infinite sea of
negative energy electrons, so the positive energy electrons cannot transit to
the negative energy states according to the exclusion principle. It’s possible
to create ‘holes’ in the sea by exciting an electron from negative energy (−E)
to positive energy (E ′ ). The absence of charge −e and energy −E of the
electron can be regarded as the presence of charge +e and energy +E due
to a particle that is antiparticle to an electron, which was predicted by Paul
Dirac in 1928 and was discovered by Carl Anderson in 1932.

Fig. 1.1. Dirac’s interpretation of negative energy electrons

This excitation leads to pair production of an electron and its antiparticle

positron.
e− (E ′ ) + e+ (E)
In 1934, Pauli and Weisskopf introduced −e into, the four-vector current j µ
and interpreted it as the charge-current density of the electron, thus reviving
the Klein-Gordon equation.

j µ = −ie (ϕ∗ ∂ µ ϕ − ϕ∂ µ ϕ∗ ) (1.30)

Now, ρ = j 0 is charge density, not a probability density, and it can be

accepted to have negative values. For E > 0, ρ < 0 represents charge
density corresponding to electron, and for E < 0, ρ > 0 corresponds to
positron. This idea can be applied to bosons as well because no appeal to
the exclusion principle has been made here.
The Feynman-Stückelberg Interpretation of E < 0 Solutions: Stückelberg
(1941) and Feynman (1948) proposed that negative energy solutions describe
8
particles that propagate backward in time or positive energy antiparticles
that propagate forward in time. For an electron with energy E, p, and
charge −e, a four-vector current is

j µ e− = −2e|N |2 (E, p)


(1.31)

For its antiparticle, a positron, with the same E, p,

j µ e+ = +2e|N |2 (E, p)


(1.32)
= −2e|N |2 (−E, −p), (1.33)

which is same as the four-current j µ for an electron with −E, −p. This
interpretation comes due to the following

e−i(−E)(−t) = e−iEt (1.34)

and will be considered while dealing with Feynman diagrams.

1.5 Non-relativistic Perturbation Theory

Often, a perturbative approach is used while dealing with the scattering of
particles. Given a free-particle Schrödinger equation
Z
H0 ϕn = En ϕn with ϕ∗m ϕn d3 x = δmn (1.35)
V

where H0 , the Hamiltonian, is time-independent, and the solutions are nor-

malized over volume V . The task is to solve Schrödinger’s equation when
the particle interacts with a potential V (x, t).
∂ψ
(H0 + V (x, t)) ψ = i (1.36)
∂t

Solutions of (1.36) can be written as

X
ψ= an (t)ϕn (x)e−iEn t (1.37)
n

We substitute ψ into (1.36) to find the coefficients an (t). On substituting,

we get
X dan X
−iEn t
i ϕn (x)e = V (x, t)an ϕn (x)e−iEn t (1.38)
n
dt n
9
 Multiplying (1.38) by ϕ∗f , integrating over whole volume of interest, and
using orthogonality relation, gives
Z
daf X
= −i an (t) ϕ∗f V ϕn d3 xei(Ef −En )t (1.39)
dt n

Assuming that before the particle interacts with potential V it is in an

eigenstate i of the unperturbed Hamiltonian. Therefore, at time t = −T /2

ai (−T /2) = 1, (1.40)

an (−T /2) = 0 for n ̸= i (1.41)

where T is the total time of interaction with potential with the particle.
Then, Z
daf
= −i d3 xϕ∗f V ϕi ei(Ef −Ei )t (1.42)
dt
Provided the potential is small and transient, these initial conditions can be
applied during the whole course of interaction, up to the first approximation,
and Z t Z
′
af (t) = −i dt′ d3 xϕ∗f V ϕi ei(Ef −Ei )t (1.43)
−T /2
after the interaction is complete,
Z T /2 Z
∗
dt d3 x ϕf (x)e−iEf t V (x, t) ϕi (x)e−iEi t
  
Tf i ≡ af (T /2) = −i
−T /2
(1.44)
= Probability of transition i → f (1.45)

In the covariant form

Z
Tf i = −i d4 xϕ∗f (x)V (x)ϕi (x) (1.46)

This is a valid approximation of Tf i only if af (t) ≪ 1. When we have a

time-independent perturbation, then
Z ∞
Tfi = −iVfi dte(Ef −Ei )t (1.47)
−∞
= −2πiVf i δ (Ef − Ei ) (1.48)

where Z
Vf i ≡ d3 xϕ∗f (x)V (x)ϕi (x). (1.49)
10
The δ function in (1.48) means that it will be non-zero only when the final
state is reached, so Tf i is not a useful quantity and so define W as

|Tf i |2
W = lim (1.50)
T →∞ T

which is the transition probability per unit time, Squaring (1.48), we can get

|Vf i |2
Z +T /2
W = lim 2π δ (Ef − Ei ) dtei(Ef −Et )t (1.51)
T →∞ T −T /2
2 Z +T /2
|Vf i |
= lim 2π δ (Ef − Ei ) dt (1.52)
T →∞ T −T /2
= 2π |Vf i |2 δ (Ef − Ei ) . (1.53)

Starting from a given initial state, the interaction ends in a set of final states.
If ρ (Ef ) is the density of final states, then integrating over this density and
imposing energy conservation, we get
Z
Wf i = 2π dEf ρ (Ef ) |Vfi |2 δ (Ef − El ) (1.54)

= 2π |Vfi |2 ρ (Ei ) . (1.55)

This is known as Fermi’s Golden Rule.

By inserting the result for an (t), from (1.43),in (1.39), we can better the
approximation
 
Z t
daf X ′
= . . . + (−i)2  Vni dt′ ei(En −Ei )t  Vf n ei(Ef −En )t (1.56)
dt −T /2
n̸=i

Dots represent the first-order result. Therefore,

Z ∞ Z t
′
X
Tf i = . . . − Vf n Vni dtei(Ef −En )t dt′ ei(En −Ei )t (1.57)
n̸=i −∞ −∞

t
ei(En −Ei −iε)t
Z
′ i(En −Ei −iε)t
dt e =i (1.58)
−∞ Ei − En + iε
where we have introduced ε so that the integrand doesn’t blow up and after
integration we put ε to 0. Then, second-order correction to Tf i is
X Vf n Vni
Tf i = · · · − 2πi δ (Ef − Ei ) (1.59)
Ei − En + iε
n̸=i
11
The perturbation series for the amplitude with terms to first, second, . . .
order in V is
X 1
Vf i → Vf i + Vf n Vni + · · · (1.60)
Ei − En + iε
n̸=i

Fig. 1.2. Feynman diagrams of first and second-order contributions to the

i → f transition

1.6 Rules to Calculate Scattering Amplitudes

RThe∗ rule is to simply4 evaluate the matrix elements ⟨ϕoutgoing |Tf i |ϕingoing ⟩ ∝
ϕoutgoing V ϕingoing d x The Feynman diagram of the interaction of a positron
with an electromagnetic potential V , which is proportional to e−iωt where ω
is the energy of the incoming photon, can be given as

Time −→
Fig. 1.3. e Scattering, pictured as negative-energy e− scattering backward
+

in time.
The transition amplitude Tf i for the interaction of in positron states is pro-
portional to
Z

∗
e−iEf t e−iωt e−iEi t dt = 2πδ (Ef − ω − Ei ) (1.61)

thus, Ef = Ei + ω.
We can also write it terms of electron states alone as
Z  ∗
−i(−Ei )t
e e−iωt e−i(−Ef )t dt = 2πδ (−Ei − ω + Ef ) (1.62)

12
Chapter 2

QED of Spinless Particles

The spin of a particle is not implied by the solutions of the Klein-Gordon

equation, and we will be considering the interaction of spinless particles in
this chapter, which is a hypothetical case.

2.1 ‘Spinless’ Electron in an Electromagnetic Field

The Klein-Gordon equation satisfied by a free particle is
∂µ ∂ µ + m2 ϕ = 0


(2.1)


In the case of an electromagnetic potential Aµ = A0 , A , the four-momentum
of an electron changes as
pµ → pµ + eAµ (2.2)
In operator form, the motion can be given by the following substitution,
i∂ µ → i∂ µ + eAµ (2.3)
After substitution, we get
∂µ ∂ µ + m2 ϕ = −V ϕ,


(2.4)
where
V = −ie (∂µ Aµ + Aµ ∂µ ) − e2 A2 (2.5)
is the electromagnetic perturbation. For small perturbations, we neglect the
e2 A2 term in (2.5). The amplitude, (1.46), for the scattering from ϕi to ϕf
due to this potential V can be given by
Z
Tf i = −i ϕ∗f (x)V (x)ϕi (x)d4 x (2.6)
Z
= i ϕ∗f ie (Aµ ∂µ + ∂µ Aµ ) ϕi d4 x (2.7)

13
Using integration by parts on second term, and taking Aµ → 0 as |x|, t →
±∞ we get
Z
Tf i = i ieAµ ϕ∗f ∂µ ϕi − ϕi ∂µ ϕ∗f d4 x


(2.8)
Z
= −i jµf i Aµ d4 x (2.9)

Fig. 2.1. A ‘spinless’ electron interacting with Aµ

where
jµf i (x) ≡ −ie ϕ∗f ∂µ ϕi − ϕi ∂µ ϕ∗f


(2.10)
is the electromagnetic current for the i → f electron transition. Substituting

ϕi (x) = Ni e−ipi ·x and ϕf (x) = Nf e−ipf ·x (2.11)

we get,
jµf i = −eNi Nf (pi + pf )µ ei(pf −pi )·x (2.12)

2.2 ‘Spinless’ Electron-Muon Scattering

This problem can be solved by taking muon as the source of electromagnetic
potential Aµ . Using Maxwell’s equations,
µ
□2 Aµ = j(2) (2.13)

Fig. 2.2. Electron-muon scattering

14
The muon-current can be written using (2.12)
µ
j(2) = −eNB ND (pD + pB )µ ei(pD −pB )·x , (2.14)

□2 eiq·x = −q 2 eiq·x , (2.15)

Solution of (2.13) can be given by
1 µ
Aµ = − j with q = pD − pB (2.16)
q 2 (2)
After the substitution of these quantities in (2.9), the lowest-order transition
amplitude for e− µ− scattering can be given by
Z  
1 µ
Tf i = −i jµ(1) (x) − 2 j(2) (x)d4 x (2.17)
q
(1) (2)
Substituting for electron current jµ from (2.12), muon currentjµ , from
(2.14), and integrating we get
Tf i = −iNA NB NC ND (2π)4 δ (4) (pD + pC − pB − pA ) ℜ (2.18)
where  
gµν
−iℜ = (ie (pA + pC ) ) −i 2 (ie (pB + pD )ν )
µ
(2.19)
q
and ℜ is known as the invariant amplitude. The Kronecker-delta encodes
the energy-momentum conservation in the interaction.

2.3 More Rules for Feynman Diagrams

−igµν /q 2 is called the photon propagator, where q is the four-momentum of
the photon. Each vertex factor has an electromagnetic coupling e multiplied
by a four-momentum. If more vertices or propagators are present in an in-
teraction, the corresponding factors multiply to give the invariant amplitude
ℜ.

2.4 Scattering Cross-section

Our free particle wavefunction is
ϕ = N e−ip·x (2.20)
The probability density ρ of particles can be given by
ρ = 2E|N |2 (2.21)
15
We fix the number of particles in a volume V to 2E, i.e.,
Z
ρdV = 2E (2.22)
V
Therefore,
1
N=√ (2.23)
V

Fig. 2.3. The vertex factors and propagator for ‘spinless’ electron-muon
scattering
For A + B → C + D, transition rate per unit volume is
|Tf i |2
Wf i = (2.24)
TV
with T and Tf i being the total time of the interaction, and the transition
amplitude, respectively. On substituting Tf i from (2.18),N from (2.23), and
integrating out one delta function, we get
(4)
4δ (pC + pD − pA − pB ) |ℜ|2
Wf i = (2π) (2.25)
V4
Experimentalists like to quote a process A + B → C + D using a quantity
called ‘cross-section’, defined as
Wf i
Cross-section = (number of final states), (2.26)
(initial flux)
No. of final states for one particle in volume V with momenta in element
d p = V d3 p/(2π)3 . For 2E particles in volume V
3

V d3 p
No. of final states per particle = (2.27)
(2π)3 2E
For C and D into momentum elements d3 pC , and d3 pD ,
V d3 pC V d3 pD
No. of final states = (2.28)
(2π)3 2EC (2π)3 2ED
16
Initial flux can be given as the no. of beam particles passing through a unit
area per unit time |vA | 2EA /V multiplied by the no. of target particles per
unit volume 2EB /V .
2EA 2EB
Initial flux = |vA | (2.29)
V V
After appropriate substitutions, we find differential cross-section dσ for scat-
tering into d3 pC , d3 pD as
V2 1 4
2 (2π) (4) d3 pC d3 pD 2
dσ = |ℜ| δ (pC + pD − pA − pB ) V
|vA | 2EA 2EB V 4 (2π)6 2EC 2ED
(2.30)
The cross-section has the units of area and represents an intrinsic scattering
probability or strength of A + B → C + D interaction. We can write
differential cross-section as
|ℜ|2
dσ = dQ (2.31)
F
where dQ is the Lorentz invariant phase space factor (dLips)
4 (4) d 3 pC d 3 pD
dQ = (2π) δ (pC + pD − pA − pB ) (2.32)
(2π)3 2EC (2π)3 2ED
and incident flux
F = |vA − vB | × 2EA × 2EB (2.33)
= 4 (|pA | EB + |pB | EA ) (2.34)
 1/2
2 2 2
= 4 (pA · pB ) − mA mB , (2.35)
which is invariant as well.

2.5 The Decay Rate in Terms of Invariant Amplitude

The differential rate for the decay A → 1+2+· · · n into momentum elements
d3 p1 , . . . , d3 pn of the final state particles can be given as

1 2 d 3 p1 d 3 pn
dΓ = |ℜ| ... (2π)4 δ (4) (pA − p1 − · · · pn ) . (2.36)
2EA (2π)3 2E1 (2π)3 2En
The total decay rate, is
dNA
Γ=− /NA (2.37)
dt
NA (t) = NA (0)e−Γt (2.38)
and Γ−1 is called the lifetime of particle A.
17
2.6 ‘Spinless’ Electron-Electron Scattering
Due to the indistinguishability of electrons, there are two Feynman diagrams
for this interaction for the lowest-order amplitude
e2 (pA + pC )µ (pB + pD )µ e2 (pA + pD )µ (pB + pC )µ
!
−iℜe− e− = −i − −
(pD − pB )2 (pC − pB )2
(2.39)

Fig. 2.4. The two (lowest-order) Feynman diagrams for electron-electron

scattering
The invariant amplitude ℜ is symmetric under pA ↔ pB , as it’s symmetric
in pC ↔ pD

2.7 ‘Spinless’ Electron-Positron Scattering

There will be two Feynman diagrams here as well, which we can translate to
electron-only diagrams to calculate the (lowest-order) e− e+ → e− e+ amplitude
µ µ
!
(p A + p C )µ (−p D − p B ) (p A − p B )µ (−p D + p C )
−iRe− e+ = −i −e2 2 − e2
(pD − ṗB ) (pC + pD )2
(2.40)

Fig. 2.5. The two (lowest-order) Feynman diagrams, (b) and (d), for
spinless e− e+ → e− e+ scattering.
ℜ is symmetric under pC ↔ −pB , that is, under the interchange of the
outgoing electrons.
18
 To calculate the invariant amplitude Re− e+ , of this interaction, we "cross"
the result for Re− e− to get

Re− e+ →e− e+ (pA , pB , pC , pD ) = Re− e− →e− e− (pA , −pD , pC , −pB ) (2.41)

2.8 Mandelstam Variables and Crossing

The incident energy and the scattering angle are two independent variables
for a scattering A + B → C + D. However, we can write ℜ in terms of
Lorentz invariants called Mandelstam variables: s, t, and u, which can be
given as
s = (pA + pB )2 (2.42)
t = (pA − pC )2 (2.43)
u = (pA − pD )2 (2.44)
They can be related as follows
s + t + u = m2A + m2B + m2C + m2D (2.45)
where mi is the rest mass of particle i. A process is classified as a f -channel
process when the square of the total center of mass energy or q 2 in the
propagator term is equal to f , where f is one of the Mandelstam variables.
Following is an example of crossing where a s-channel process is converted
to a u-channel process.

Fig. 2.6. The crossing (or interchange) of particles B and D

For ‘spinless’ electron-positron invariant amplitude as given in (2.40) can be
written as  
2 s−u t−u
Re− e+ (s, t, u) = e + (2.46)
t s
And the crossing done to electron states in (2.41) can be written as
Re− e+ (s, t, u) = Re− e− (u, t, s) (2.47)
19
Chapter 3

The Dirac Equation

Dirac gave the following equation, which, unlike the Klein-Gordon equation,
is linear in ∂/∂t which in order to be covariant, required to be linear in ∇
as well
Hψ = (α · P + βm)ψ. (3.1)
Coefficients β and αi (i = 1, 2, 3) are determined by imposing the condition
that this must satisfy relativistic energy-momentum relation,

H 2 ψ = P2 + m2 ψ


(3.2)

H 2 ψ = (αi Pi + βm) (αj Pj + βm) ψ

= (αi2 Pi2 + (αi αj + αj αi ) Pi Pj + (αi β + βαi ) Pi m + β 2 m2 )ψ (3.3)
↓ | {z } | {z }
0 0
where i > j. Comparing with (3.2), we conclude that
1. α1 , α2 , α3 , β all anti-commute with each other =⇒ they are not numbers
but should be matrices operating on a 4 component column vector Dirac
spinor, but not four-vector as it doesn’t transform according to Lorentz
transformation in Lorentz frames.
2. α12 = α22 = α32 = β 2 = 1.
αi and β are hermitian (as Hamiltonian must be hermitian), traceless ma-
trices of even dimensionality, with eigenvalues ±1.

4 × 4 is the minimum dimension for these matrices to satisfy all these

conditions. Also, they are not unique. In Dirac-Pauli representation they
are given as:    
0 σ I 0
α= , β=
σ 0 0 −I
where I is 2 × 2 identity matrix and where σ are the Pauli matrices:
20
      
0 1 0 −i 1 0
σ1 = , σ2 = , σ3 =
1 0 i 0 0 −1

3.1 Dirac Equation in Covariant Form

On multiplying Dirac equation, (3.1), by β from left, we get
∂ψ
iβ = −iβα · ∇ψ + mψ
∂t
(iγ µ ∂µ − m) ψ = 0 (3.4)
where
γ µ ≡ (β, βα) (3.5)
also called Dirac-γ-matrices. This is the covariant form of the Dirac equation.
Dirac-γ-matrices satisfy the following properties:
1.
γ µ γ ν + γ ν γ µ = 2g µν (3.6)
2.
2
γ 0† = γ 0 , γ0 =I (3.7)
3. )
k †
k†

k k
γ = βα = α β = −γ

2 k = 1, 2, 3 (3.8)
γ k = βαk βαk = −I
4. Hermitian conjugation properties can be encoded as
γ µ† = γ 0 γ µ γ 0 (3.9)

3.2 Continuity Equation

The hermitian conjugate of Dirac’s equation,
∂ψ ∂ψ
iγ 0
+ iγ k k − mψ = 0 (3.10)
∂t ∂x
where k = 1, 2, 3, can be given by

∂ψ † 0 ∂ψ †
γ − i k −γ k − mψ † = 0


−i (3.11)
∂t ∂x
(3.11) can be written in covariant form by multiplying it with γ 0 from the
right, using the relation γ 0 γ k = −γ k γ 0 , and introducing adjoint spinor
ψ̄ ≡ ψ † γ 0 (3.12)
21
we get,
i∂µ ψ̄γ µ + mψ̄ = 0 (3.13)
By multiplying (3.4) from the left by ψ̄ and (3.13) from the right by ψ,
and adding, we get
ψ̄γ µ ∂µ ψ + ∂µ ψ̄ γ µ ψ = ∂µ ψ̄γ µ ψ = 0



(3.14)
where j µ = ψ̄γ µ ψ satisfies the continuity equation. The probability density
can be written as
4
X
0 0
ρ ≡ j = ψ̄γ ψ = ψ ψ = †
|ψi |2 (3.15)
i=1

is positive definite.
According to the Pauli-Weisskopf prescription, we insert the charge −e in
j µ,
j µ = −eψ̄γ µ ψ (3.16)
and from now j µ can be regarded as electron (four-vector) current density.

3.3 Free-Particle Spinors

For Dirac’s equation, we put the following spinor (a vector with complex
entries) as our ansatz
ψ = u(p)e−ip·x (3.17)
where u is a four-component spinor in (3.4), to get
(γ µ pµ − m) u(p) = 0 (3.18)
/ ≡ γ µ Aµ for any four-vector Aµ , we obtain
Introducing notation A
(p/ − m)u = 0. (3.19)
(3.19) can also be written as
Hu = (α · p + βm)u = Eu (3.20)
There are four independent solutions of this equation, two with E > 0
and two with E < 0. This is particularly easy to see. First, for the sake
of simplicity, we take the particle at rest, i.e., p = 0, in the Dirac-Pauli
representation of α and β, we have
 
mI 0
Hu = βmu = u (3.21)
0 −mI
22
 Eigenvalues of E are m, m, −m, −m, and eigenvectors are
       
1 0 0 0
0 1
 ,  0 ,  0 .
   
 ,
0 0 1 0
0 0 0 1
−e, is regarded as the particle. Therefore, E > 0 solutions describes
an electron, while E < 0 solutions describe an antiparticle (positron). For
p ̸= 0, we have
    
m σ·p uA uA
Hu = =E
σ · p −m uB uB
where u is written in terms of two two-component spinors, uA and uB .

σ · puB = (E − m)uA , (3.22)

σ · puA = (E + m)uB (3.23)
(σ · p)2 uA = (E 2 − m2 )uA (3.24)

(σ · p)2 = p2 = E 2 − m2 and for the two E > 0 solutions, we are free to

choose any two eigenvectors which span the space of all 2 × 1 vectors. So we
(s)
choose uA = χ(s) , where
   
1 0
χ(1) = , χ(2) =
0 1
and by (3.23),
(s) σ · p (s)
uB = χ (3.25)
E+m
χ(s)
 
(s)
u =N σ·p (s) , E>0
E+m χ
√
where s = 1, 2, and N = E + m is the normalization constant. For the
(s)
E < 0 solutions, we put uB = χ(s) , and
(s) σ · p (s) σ · p (s)
uA = uB = − χ (3.26)
E−m |E| + m
Then, we get  −σ·p (s) 
|E|+m χ
u(s+2) = N (s) , E < 0.
χ
For a given momentum p, we have four orthogonal solutions for an electron
u(1,2) , with positive energy, and u(3,4) , with negative energy.
23
 This extra two-fold degeneracy for each positive as well as negative energy
implies there is an observable and an operator that commutes with H and
p, whose eigenvalues can be used to label the states. The choice of such an
operator is not unique. One such operator is the helicity operator 12 Σ · p̂,
 1 
1 σ · p̂ 0 1
h = Σ · p̂ ≡ 2 1 = (σ · p̂) ⊗ I
2 0 2 σ · p̂ 2
Helicity is defined as the spin in the direction of momentum or eigenvalues
of 21 σ · p̂ , λ

Fig. 3.1. Interpretation of helicity

Note that tensor product with identity will just give the degeneracy in eigen-
values which just means that two states for negative energy solutions also
exist.

3.4 Intrinsic Angular Momentum

Orbital angular momentum, L = r × P doesn’t commute with Hamiltonian
of equation (3.1), as
[H, L] = −i(α × P) (3.27)
The operators of conserved quantities commute with the Hamiltonian. This
means L is not conserved, and there must be some other angular momentum
which, on adding to L gives the total angular momentum, which commutes
with the Hamiltonian in order to make the total angular momentum a con-
served quantity. We note that
 
σ 0
[H, Σ] = +2i(α × P), where Σ ≡ (3.28)
0 σ
and the quantity
1
J=L+ Σ (3.29)
2
commutes with the Hamiltonian,

[H, J] = 0. (3.30)

24
Therefore, J is the total angular momentum and 12 Σ denotes the intrinsic an-
gular momentum operator. Associated intrinsic (or spin) magnetic moment
is
e e
µ=− σ ≡ −g S (3.31)
2m 2m
where the gyromagnetic ratio g is 2 (2.00232 experimentally) and the spin
angular momentum operator S is 21 σ.

3.5 Antiparticles
The Dirac equation for an electron (charge −e ) in an electromagnetic field
is (2.3)
[γ µ (i∂µ + eAµ ) − m] ψ = 0 (3.32)
An equivalent Dirac equation for the positron (+e) can be given as
[γ µ (i∂µ − eAµ ) − m] ψC = 0. (3.33)
Taking the complex conjugate of (3.32), we get
[−γ µ∗ (i∂µ − eAµ ) − m] ψ ∗ = 0. (3.34)


We can find a matrix, Cγ 0 , for which
− Cγ 0 γ µ∗ = γ µ Cγ 0



(3.35)
then multiplying this matrix to the left of (3.34) we get
[γ µ (i∂µ − eAµ ) − m] Cγ 0 ψ ∗ = 0


(3.36)
Comparing (3.33) and(3.36), we can say
ψC = Cγ 0 ψ ∗ = C ψ̄ T (3.37)
A possible C which satisfies our requirements is
 
1
−1
Cγ 0 = iγ 2 = 
 

 −1 
1
which has the following properties
C −1 γ µ C = (−γ µ )T , (3.38)
C = −C −1 = −C † = −C T (3.39)
ψ̄C = −ψ T C −1 . (3.40)
25
The electron current is
j µ = −eψ̄γ µ ψ
The current associated with the charge conjugate field can be given by

jCµ = −eψ̄C γ µ ψC (3.41)

= +eψ T C −1 γ µ C ψ̄ T (3.42)
= −eψ T (γ µ )T ψ̄ T (3.43)
= −(−)eψ̄γ µ ψ (3.44)

There is a minus sign in (3.41) because in single-particle (electron) theory,

positrons are not there; so, C changes a positive energy electron state into a
negative energy electron state. This is the reason that we insert a negative
sign by hand for every final negative energy electron.

3.6 Bilinear Covariants

We want to construct the most general currents that obey Lorentz covari-
ance (and consider rotations, boots, and parity), we enlist quantities of the
following form
(ψ̄)(4 × 4matrix)(ψ) (3.45)
Introduce γ 5 ≡ iγ 0 γ 1 γ 2 γ 3 (3.46)
and we have
2
γ 5† = γ 5 , γ5 = I, γ 5γ µ + γ µγ 5 = 0 (3.47)
In the Dirac-Pauli representation,(3),
     
0 I 0 0 σ 5 0 I
γ = , γ= , γ =
0 −I −σ 0 I 0
All such matrices are enlisted here
No. of Matrices Space Inversion, P
Scalar ψ̄ψ 1 + under P
µ
Vector ψ̄γ ψ 4 Space compts.: - under P
µν
Tensor ψ̄σ ψ 6
Axial Vector ψ̄γ 5 γ µ ψ 4 Space compts.: + under P
5
Pseudoscalar ψ̄γ ψ 1 - under P
where
i µ ν
σ µν = (γ γ − γ ν γ µ ) (3.48)
2
26
 Dirac equation in frames x and x′
∂ψ(x)
iγ µ − mψ(x) = 0, (3.49)
∂xµ
′ ′
µ ∂ψ (x )
iγ ′µ
− mψ ′ (x′ ) = 0, (3.50)
∂x
where x′ = Λx and
Substituting
ψ ′ (x′ ) = Sψ(x) (3.51)
into (3.50) and comparing with (3.49), we get
S −1 γ µ S = Λµ ν γ ν , (3.52)
Dirac’s γ-matrices, spinors, and bilinear covariants follow the following
transformation properties with matrix S (L and P in subscript represent
transformation corresponding to proper Lorentz and parity, respectively):
1. γ-matrices
SL−1 = γ 0 SL† γ 0 (3.53)
γ 5 SL = SL γ 5 (3.54)
SP−1 γ 0 SP = γ 0 , (3.55)
SP−1 γ k SP = −γ k for k = 1, 2, 3, (3.56)
γ 5 SP = −SP γ 5 (3.57)

This is satisfied by
SP = γ 0 (3.58)
Therefore, under parity
′ ′
ψ1,2 = ψ1,2 and ψ3,4 = −ψ3,4 (3.59)

2. Spinors
ψ̄ ′ = ψ † γ 0 = ψ † S † γ 0 = ψ † γ 0 S −1 (3.60)
= ψ̄S −1 (3.61)

3. Bilinear covariants
ψ̄ ′ γ µ ψ ′ = ψ̄SL−1 γ µ SL ψ = Λµν ψ̄γ ν ψ ,


(3.62)

ψ̄γ 0 ψ

ψ̄ ′ γ µ ψ ′ = ψ̄SP−1 γ µ SP ψ = (3.63)
−ψ̄γ k ψ
i.e., they transform like four-vectors.
27
Chapter 4

QED of Spin- 12 Particles

In this chapter, we will revisit the concepts discussed in Chapter 2, but with
the prescription of the Dirac equation, which takes care of the spin of the
particles as well.

4.1 Electron in an Electromagnetic Field

The solution of the Dirac equation for free particle (3.18),
(γµ pµ − m) ψ = 0
can be given by
ψ = u(p)e−ip·x
In an electromagnetic field, Aµ , the equation the corresponding equation can
be obtained by taking
pµ → pµ + eAµ , (4.1)
to get
(γµ pµ − m) ψ = γ 0 V ψ, (4.2)
where
γ 0 V = −eγµ Aµ (4.3)
is the electromagnetic perturbation.
The transition amplitude from ψi to ψf for an electron, up to first order
can be given by
Z
Tf i = −i ψf† (x)V (x)ψi (x)d4 x (4.4)
Z
= ie ψ̄f γµ Aµ ψi d4 x (4.5)
Z
= −i jµf i Aµ d4 x (4.6)

28
where

jµf i ≡ −eψ̄f γµ ψi (4.7)

= −eūf γµ ui ei(pf −pi )·x (4.8)

is the electromagnetic transition current. Comparing it with the ‘spinless’

electron case,
jµf i = −e (pf + pi )µ ei(pf −pt )·x
which gave the rules for Feynman diagrams, the corresponding rules now are:
the vertex factor is now a 4×4 matrix in spin space. It is sandwiched between
column u(s) (pi ) and row ū(r) (pf ) spinors in spin states s, r, respectively.
1
ūf γ µ ui = ūf (pf + pi )µ + iσ µν (pf − pi )v ui ,


(4.9)
2m
which means that the physical spin- 21 electron interacts via both its charge
and its magnetic moment, unlike ‘spinless’ electron, which gives coupling
(pf + pi ) as it interacts with charge only.

(a) ‘Spinless’ electron

(b) Spin 1/2 electron

Fig. 4.1. Factors for the Feynman rules

Identification of magnetic moment:

Z
Tf i = −i2πδ (Ef − Ei ) jµf i Aµ d3 x (4.10)

Z h Z
e ν
i
µ 3 f†
 e 
− ψ̄f iσµν (pf − pi ) ψi A d x = ψA σ · B ψAi d3 x (4.11)
2m 2m
where ψA are the upper two components of ψ.

29
4.2 Møller Scattering
The transition amplitude for Møller scattering, e− e− → e− e− . is
Z  
1 µ
Tf i = −i jµ(1) (x) − 2 j(2) (x)d4 x
q
 
1
= −i (−eūC γµ uA ) − 2 (−eūD γ µ uB ) (2π)4 δ (4) (pA + pB − pC − pD )
q
where q = pA − pC ,

(a)

(b)
Fig. 4.2. Feynman diagrams for Møller scattering
Invariant amplitude for 4.2(a) is
 
−igµν
−iℜ = (ieūC γ µ uA ) (ieūD γ ν uB ) (4.12)
q2
Similarly, we can calculate for 4.2(b) to give the total lowest-order ampli-
tude for Møller scattering as
µ µ
2 (ūC γ uA ) (ūD γµ uB ) 2 (ūD γ uA ) (ūC γµ uB )
R = −e +e (4.13)
(pA − pC )2 (pA − pD )2
If no measurement of the spin of electrons is done, we need to work
with an unpolarized cross section for which we need to make the following
replacement
30
 1 X
|ℜ|2 → |ℜ|2 ≡ |R|2 (4.14)
(2sA + 1) (2sB + 1)
all final spin
states
with sA , sB being spins of the incoming particles. Thus, averaging over
the spins of incoming particles is done. To make life simple, we calculate it
in the non-relativistic limit, i.e., |p| → 0. Putting |p| → 0 in (3.3), we get

√ χ(s)
 
For ingoing electron: u(s) = 2m
0
√ †
 
χ(s)
For outgoing electron: ū(s) = 2m
0
where s = 1, 2 implies to spin up and down, respectively. Using Dirac
representation of γ-matrices, we get
(
2m if µ = 0,
ū(s) γ µ u(s) = (4.15)
0 ̸ 0,
if µ =
′
ū(s) γ µ u(s ) = 0 for all µ, if s ̸= s′ . (4.16)
It means that it is only at the higher energies the magnetic field interaction
takes place which can flip the spin. Substituting (4.15) into (4.13) gives
 
1 1
ℜ(↑↑→↑↑) = ℜ(↓↓→↓↓) = −e2 4m2 −
t u
1
ℜ(↑↓→↑↓) = ℜ(↓↑→↓↑) = −e2 4m2
t
1
ℜ(↑↓→↓↑) = ℜ(↓↑→↑↓) = e2 4m2 ,
u
where s, t and u are the Mandelstam variables.

Summing over all spins we get,

" 2 #
1 2 2 1 1 1 1
4m2 e


|ℜ|2 = 2 − + + (4.17)
4 t u t2 u2
In the center-of-mass frame, where θ is the scattering angle
θ
t = −2p2 (1 − cos θ) = −4p2 sin2 (4.18)
2
θ
u = −2p2 (1 + cos θ) = −4p2 cos2 (4.19)
2
31
and p = |pı | with i = A, B, C, D. The differential cross-section in the
nonrelativistic limit can be given as
!
2 2
dσ mα 1 1 1
= + − (4.20)
dΩ cm 16p4 sin4 2θ cos4 2θ sin2 2θ cos2 2θ

where α ≡ e2 /4π, and we have put s ≃ 4m2

4.3 Electron-Muon Scattering and Crossing

Fig. 4.3. Feynman diagram for electron-muon scattering

The invariant amplitude can be written as
1
ℜ = −e2 ū (k ′ ) γ µ u(k) 2 ū (p′ ) γµ u(p) (4.21)
q

8e4 ′ ′
= 4 [(k · p ) (k · p) + (k ′ · p) (k · p′ ) −m2 p′ · p − M 2 k ′ · k + 2m2 M 2

|ℜ|2
q
(4.22)
where m is the mass of electron and M is the mass of muon. At high en-
ergies, we neglect the terms involving m2 and M 2 to get invariant amplitude
as
8e4 ′ ′ ′ ′
|ℜ|2 = 4 [(k · p ) (k · p) + (k · p) (k · p )] (4.23)
′
(k − k )
and Mandelstam variables as
s ≡ (k + p)2 ≃ 2k · p ≃ 2k ′ · p′ (4.24)
2
t ≡ (k − k ′ ) ≃ −2k · k ′ ≃ −2p · p′ (4.25)
2
u ≡ (k − p′ ) ≃ −2k · p′ ≃ −2k ′ · p (4.26)
Therefore, at high energies
2
+ u2
4s
|R|2
= 2e (4.27)
t2
And by crossing, we can get the invariant amplitude of e− e+ → µ+ µ− by
interchanging k ′ ↔ −p, that is, s ↔ t in (4.27),
32
 Fig. 4.4. The Feynman diagram for e− e+ → µ+ µ−

2
+ u2
4t
|R|2= 2e (4.28)
s2
where the s-channel process is e− e+ → µ+ µ− .

4.4 Propagators
Perturbative expansion of the transition amplitude can be given as
 
X 1
Tf i = −i2πδ (Ef − Ei ) ⟨f |V |i⟩ + ⟨f |V |n⟩ ⟨n|V |i⟩ + · · · 
Ei − En
n̸=i
(4.29)
⟨f |v|n⟩ are regarded as vertices and the associated propagator for interaction
is 1/ (Ei − En ). Also,
H0 |n⟩ = En |n⟩
where H0 is the unperturbed Hamiltonian. Therefore, transition amplitude
can be written as
 
i
Tf i = 2πδ (Ef − Ei ) f (−iV ) + (−iV ) (−iV ) + · · · i (4.30)
Ei − H0
Now, −iV is the vertex factor, hence the associated propagator will be the
inverse of the −i(Ei − H0 ), and this quantity can be identified by comparing
with
−i (Ei − H0 ) ψ = −iV ψ, (4.31)
which acts on the intermediate state.

4.4.1 The Propagator for Spinless Particles

Comparing Klein-Gordon equation
i □2 + m2 ϕ = −iV ϕ,


(4.32)
to the form in (4.31), the propagator for an intermediate state of momentum
p is
1 i
= (4.33)
i (−p2 + m2 ) p2 − m2
33
4.4.2 The Electron Propagator
In an electromagnetic field, [(4.2) and (4.3)], an electron obeys the equation
(p/ − m)ψ = −eγ µ Aµ ψ, (4.34)
After multiplying both sides by −i, we can see the vertex factor is ieγ µ .
Again by comparing with the general form in (4.31), the electron propagator
is
i(p/ + m)
P
1 i i s uū
= = 2 = 2 (4.35)
−i(p/ − m) p/ − m p − m2 p − m2
where the completeness relation (5.47). The numerator is the sum of the
spin states of virtual electron. For a virtual particle, the general form of
propagator is P
i spins
.
p2 − m 2

4.4.3 The Photon Propagator

Due to the gauge freedom in the potential Aµ , the propagator for a photon
is not unique, The wave equation for a photon is
g νλ □2 − ∂ ν ∂ λ Aλ = j ν ,


(4.36)
Using the Feynman gauge condition, ∂λ Aλ = 0, which we have considered
so far, (4.36) can be written as
g νλ □2 Aλ = j ν . (4.37)
And we know that
gµν g νλ = δµλ (4.38)
where δµλ equals 1 if λ = µ, and 0 otherwise. The Feynman propagator,
which is the inverse of momentum space operator multiplied by −i, can be
given by
gµν
−i 2 (4.39)
q
If we don’t take the Feynman gauge, the wave equation becomes
   
1
g νλ □2 − 1 − ∂ ν ∂ λ Aλ = j ν (4.40)
ξ
where ξ encodes the gauge freedom (ξ = 1 corresponds to Feynman gauge)
and the photon propagator can be given as
 
i qµ qν
−gµν + (1 − ξ) 2 (4.41)
q2 q
34
4.4.4 Propagator for Massive Vector Particles
Substituting □2 → □2 + M 2 in the wave equation of photon in (4.36) gives
the wave equation of a spin-1 particle of mass M,

g νλ □2 + M 2 − ∂ ν ∂ λ Bλ = 0



(4.42)

where Bλ is the wavefunction of free particle. Then we determine the inverse

of the momentum space operator

−1
g νλ −p2 + M 2 + pν pλ = δλµ (Agµν + Bpµ pν )


(4.43)

by solving for A and B. The propagator is the quantity inside brackets in

the RHS of (4.43) multiplied by i,


i −g µν + pµ pν /M 2
(4.44)
p2 − M 2

35
4.5 Summary of the Feynman Rules for QED
The rules of evaluating invariant amplitude ℜ can be obtained from following
table

36
Bibliography

1. F. Halzen and A.D. Martin, Quarks and Leptons: An Introductory

Course in Modern Particle Physics (Wiley, New York, NY, 1984).
All the images for Feynman diagrams have been taken from the book.
2. A.I.J. Rhind and H.A.J. G., Gauge Theories in Particle Physics: A
Practical Introduction (CRC Press, Boca Raton (FL), 2013).
3. IITB logo from:
https://www.iitb.ac.in/sites/www.iitb.ac.in/themes/touchm/logo.png
Fair use, https://en.wikipedia.org/w/index.php?curid=66409590

37