Klein-Gordon Equation

● The existence of plane waves

φ(r, t) ∝ exp(ik · r − iωt)

satisfying de Broglie and Einstein relations

p = !k , E = !ω

implies the quantum operator interpretation

∂
p → −i!∇ , E → i! .
∂t

● Then the relativistic energy-momentum equation

E 2 = p2 c2 + m2 c4

implies the Klein-Gordon equation

2
2∂ φ
−! 2
= −! 2 2 2
c ∇ φ + m2 4
c φ
∂t
1
● In covariant notation (see handout)
! " mc #2 $
∂µ ∂ µ + φ=0
!
where
1 ∂ 2
∂µ ∂ µ = 2 2 − ∇2
c ∂t

● KG wave function φ is a Lorentz-invariant (scalar) function; Lorentz

transformation r, t → r ! , t! implies φ → φ! where

φ! (r! , t! ) = φ(r, t) .

Hence it must represent a spin-zero particle (no orientation).

● Since |φ|2 is also invariant, this cannot represent a probability density. A
density transforms as time-like (0-th) component of a 4-vector, due to Lorentz
contraction of volume element.

2
● Correct definition of density follows from the continuity equation:
∂ρ
= −∇ · J
∂t
(J = corresponding current vector), i.e.
∂µ J µ = 0
where J µ = (cρ, J) is the 4-current.
● We can obtain an equation of this form from the KG equations for φ and φ∗ ,
% 2 ∗
&
∂ 2
φ ∂ φ ' ∗ 2 (
i! φ ∗
−φ 2 = i!c φ ∇ φ − φ∇ φ ,
2 2 ∗
∂t2 ∂t
% ∗
&
∂ ∗ ∂φ ∂φ
i! φ −φ = i!c2 ∇ · (φ∗ ∇φ − φ∇φ∗ ) .
∂t ∂t ∂t
Hence
% ∗
&
∂φ ∂φ
ρ = i! φ∗ −φ
∂t ∂t
J = −i!c2 (φ∗ ∇φ − φ∇φ∗ )
i.e. J µ = i!c2 (φ∗ ∂ µ φ − φ∂ µ φ∗ ).
3
● Normalization is such that the energy eigenstate φ = Φ(r)e−iEt/! has
ρ = 2E|Φ|2 . Thus |Φ| = 1 corresponds to 2E particles per unit volume
(relativistic normalization).
● Compare with Schrödinger current
i! 1
J =−
S
(φ ∇φ − φ∇φ ) =
∗ ∗
J KG
2m 2mc2
which thus has in fact E/mc2 particles per unit volume.

4
 Problems with Klein-Gordon Equation
1. Density ρ is not necessarily positive (unlike |φ|2 ) ⇒ equation was rejected
initially.
2. Equation is second-order in t ⇒ need to know both φ and ∂φ ∂t at t = 0 in order
to solve for φ at t > 0. Thus there is an extra degree of freedom, not present in
the Schrödinger equation.
3. The equation on which it is based (E 2 = p2 c2 + m2 c4 ) has both positive and
negative solutions for E.
● Actually these problems are all related, since a solution φ = Φ(r)e∓iEt/! has
ρ = ±2E|Φ|2 , and so for the general solution

φ = Φ+ (r)e−iEt/! + Φ− (r)e+iEt/!
)
∂φ )
both φ(t = 0) = Φ+ + Φ− and i!
E ∂t ) = Φ+ − Φ− are needed in order to
t=0
specify Φ+ and Φ− .

5
 Electromagnetic Waves
● In units where %0 = µ0 = c = 1 (‘Heaviside-Lorentz’) Maxwell’s equations are
∂B
∇ · E = ρem , ∇×E =−
∂t
∂E
∇ · B = 0 , ∇ × B = J em +
∂t
where (ρem , J em ) = Jem
µ
is the electromagnetic 4-current.
● In terms of the scalar and vector potentials V and A,
∂A
E=− − ∇V , B = ∇×A.
∂t
So we find
∂2A ∂V
∇ × (∇ × A) ≡ ∇(∇ · A) − ∇ A = J em −
2
−∇
∂t2 ∂t

● In terms of the 4-potential Aµ = (V, A)

(∂ν ∂ ν )Aµ − ∂ µ (∂ν Aν ) ≡ ∂ν F νµ = Jem

µ

6
 where the electromagnetic field-strength tensor is

F νµ = ∂ ν Aµ − ∂ µ Aν = −F µν .

● E and B, and hence Maxwell’s equations, are invariant under gauge

transformations
Aµ → A!µ = Aµ + ∂ µ χ
where χ(r, t) is an arbitrary scalar function.
● Therefore we can always choose Aµ such that ∂µ Aµ = 0 (Lorenz gauge). If
∂µ Aµ = f (= 0, we can change to A!µ = Aµ + ∂ µ χ where ∂µ ∂ µ χ = −f .
● Then in free space (J µ = 0) we have ∂ν ∂ ν Aµ = 0.
❖ Massless KG equation for each component of Aµ
❖ Aµ is ‘wave function’ of photon
❖ Aµ is a 4-vector ⇒ photon has spin 1.
● Plane wave solutions Aµ = εµ exp(ik · r − iωt) ≡ εµ e−ik·x where εµ =
polarization 4-vector, k · x ≡ k µ xµ , k µ = (ω, k) = wave 4-vector.
7
● From wave equation k · k = 0 hence ω 2 = k2 , i.e. E 2 = p2 c2 (massless
photons).
● From Lorenz gauge condition ε · k = 0 ⇒ ε0 = ε · k/ω.
● Polarization 4-vector ε!µ = εµ + ak µ is equivalent to εµ for any constant a.
Hence we can always choose ε0 = 0. Then Lorenz condition becomes
transversity condition: ε · k = 0.
● E.g. for k along z-axis we can express εµ in terms of plane polarization states

εµx = (0, 1, 0, 0) , εµy = (0, 0, 1, 0) ,

µ √
or circular polarization states εR,L = (0, 1, ±i, 0)/ 2.
N.B. only 2 polarization states for real photons.

8
 Electromagnetic Interactions
● As in classical (and non-relativistic quantum) physics, we introduce e.m.
interactions via the minimal substitution in the equations of motion:

E → E − eV , p → p − eA

i.e.
pµ → pµ − eAµ , ∂ µ → ∂ µ + ieAµ

● The Klein-Gordon equation becomes

(∂µ + ieAµ )(∂ µ + ieAµ )φ + m2 φ = 0 ,

(∂µ ∂ µ + m2 )φ = −ie[∂µ (Aµ φ) + Aµ (∂ µ φ)] + e2 Aµ Aµ φ

The conserved current is now (! = c = 1)

J µ = i(φ∗ ∂ µ φ − φ ∂ µ φ∗ ) − 2eAµ φ∗ φ

9
 Klein Paradox
● Consider KG plane waves incident on electrostatic barrier, height V , width a
V
e ipx−iEt A e ip’x−iEt T e ipx−iEt

R e −ipx−iEt B e −ip’x−iEt

x=0 x=a
KG equation for x < 0, x > a gives E 2 = p2 + m2
*
⇒ p = + E 2 − m2

(sign from B.C.).

● In 0 < x < a, Aµ = (V, 0) and so (E − eV )2 = p!2 + m2
*
⇒ p = + (E − eV − m)(E − eV + m)
!

(sign choice is arbitrary since we include ±p! ).

10
● Matching φ and ∂φ/∂x at x = 0 and a gives (as for Schrödinger equation)
) % & )−2
) i p p ! )
2 )
|T | = )cos p a −
!
+ ! )
sin p a)
2 p ! p

● Now consider behaviour as V is increased:

❖ eV < E − m: p! is real, |T | < 1 (|T | = 1 when p! a = nπ).
❖ E − m < eV < E + m: p! is imaginary, |T | < 1, transmission by tunnelling.
❖ eV > E + m: p! is real again! |T | = 1 when p! a = nπ!?
● Note that when eV > E + m density inside barrier is negative:

ρ! = 2(E − eV )|φ|2 < −2m|φ|2

● Meanwhile, the current inside remains positive, Jx! = 2p|T |2 (current

conservation). Hence when eV > E + m there is a negative density flowing
from right to left, giving a positive current. We interpret this as a flow of
antiparticles: Jem
µ
= eJ µ always.
11
● When eV > E + m and |T | = 1, antiparticles created at the back of the barrier
(x = a) travel to x = 0 and annihilate the incident particles. At the same time,
particles created at x = a travel to x > a, replacing the incident beam.
E E=0 E

π+ π π+

● Antiparticles are trapped inside the barrier, but field is zero there, so there can
be perfect transmission for any thickness.

12
● Antiparticles are like particles propagating backwards in time
π+
t t

π+

π+ π+ π− π+

x x
eV < E−m eV > E+m

13
 Charge Conjugation
● If φ is a negative-energy plane-wave solution of the KG equation, with
momentum p, φ = exp(ip · r + iEt) (E > 0), then φ∗ = exp(−ip · r − iEt) is a
positive-energy wave with momentum −p. Furthermore, in e.m. fields, φ∗
behaves as a particle of charge −e:

(∂µ + ieAµ )(∂ µ + ieAµ )φ + m2 φ = 0

⇒ (∂µ − ieAµ )(∂ µ − ieAµ )φ∗ + m2 φ∗ = 0

● Thus if φ is a negative-energy solution, we take it to represent an antiparticle

with wave function φ∗ (and hence positive energy, opposite charge and
momentum).
● Correspondingly, KG equation is invariant w.r.t. φ → φ∗ , e → −e. This is
called charge conjugation, C.
N.B. Under C, J µ → −J µ as expected.

14
 Electromagnetic Scattering
● We assume (for the moment) the same formula as in NRQM for the scattering
amplitude in terms of the first-order perturbation due to e.m. field:
+
Af i = −i φ∗f {ie[∂µ (Aµ φi ) + Aµ (∂ µ φi )]}d4 x
+
by parts = e Aµ [φ∗f (∂ µ φi ) − (∂ µ φ∗f )φi ]d4 x
+
= −ie Aµ Jfµi d4 x

where Jfµi = i[φ∗f (∂ µ φi ) − (∂ µ φ∗f )φi ] is generalization of J µ to φf (= φi

(transition current). Note that to get Af i to order e1 we only need Jfµi to
order e0 . Similarly, for Aµ we can use the free-field form
Aµ = εµ e−ik·x

● For plane waves, φf,i = e−ipf,i ·x ,

Jfµi = (pi + pf )µ ei(pf −pi )·x
15
 Hence
+
Af i = −ieεµ (pi + pf )µ ei(pf −pi −k)·x d4 x

= −ie(2π)4 ε · (pi + pf ) δ 4 (pf − pi − k)

● This corresponds to the Feynman rules for the diagram

p
i

p
k f

❖ An overall factor of (2π)4 δ 4 (pf − pi − k) (momentum conservation)

❖ εµ for an external photon line
❖ −ie(pi + pf )µ for a vertex involving a spin-0 boson of charge e.
N.B. 4-momentum cannot be conserved in this process for free particles! But
we shall see that it can occur as part of a more complicated process, e.g.
particle-particle scattering by photon exchange.
16
● We shall consider process ab → ab as scattering of a in e.m. field of b (both
spin-0). +
Af i = −iea Aµ Jaµ! a d4 x

p pa’
a

q=p ’−p
b b
p p’
b b
● Then (in Lorenz gauge) Aµ satisfies

∂ν ∂ ν Aµ = eb Jbµ! b

N.B. We assume correct source current is

! µ i(p!b −pb )·x
Jbµ! b = (pb + pb ) e

17
● Solution for 4-vector potential is then
1
Aµ = − 2
eb (p b + p b) e
! µ iq·x
q
where q = p!b − pb and q 2 = q · q.
● Hence
+
iea eb i(p!a +p!b −pa −pb )·x 4
Af i = (pa + pa ) · (pb + pb ) e
! !
d x
q2
! $
−igµν
= [−ie(pa + p!a )µ ] 2
[−ie(p b + p b) ]
! ν
q
×(2π)4 δ 4 (p!a + p!b − pa − pb )

N.B. symmetry in a, b.
● Thus we have the additional Feynman rule:
❖ −igµν /q 2 for an internal photon line.

18
● In processes involving antiparticles, remember we use particles with opposite
energy and momentum; pµ = −p̄µ .
p
a

q
−
pa
“pi ”=pa , “pf ”=−p̄a , “k”=−q,
Af i = −iea (2π)4 ε · (pa − p̄a )δ 4 (q − pa − p̄a )

pb

q
−
pb
“pi ”=−p̄b , “pf ”=pb , “k”=q,
Af i = −ieb (2π)4 ε · (pb − p̄b )δ 4 (pb + p̄b − q)
19
● Annihilation process
p pb
a

q
−
pa p−
b

ea eb
Af i = i 2 (pa − p̄a ) · (pb − p̄b )(2π)4 δ 4 (pb + p̄b − q)
q
where q = pa + p̄a = pb + p̄b .
● Since we have already normalized to 2E particles per unit volume, we have
, ,
Af i = Mf i (2π) δ (
4 4
pf − pi )

where Mf i is the invariant matrix element (see handout).

● Thus e.g. for annihilation process
ea eb
Mf i = i 2 (pa − p̄a ) · (pb − p̄b )
q

20
● In terms of the Mandelstam variables

s = (pa + p̄a )2 = q 2
t = (pb − pa )2 = (p̄a − p̄b )2
u = (pa − p̄b )2 = (p̄a − pb )2

we get
ea eb
Mf i = i (u − t)
s
and hence the invariant differential cross section is
dσ e2a e2b (u − t)2
=
dt 64πs3 (p∗a )2
*
where p∗a = s/4 − m2a = c.m. momentum of a.

21
 Dirac Equation
● Historically, Dirac (1928) was looking for a covariant wave equation that was
first-order in time, to avoid the above ‘problems’ of the Klein-Gordon equation:
∂ψ
i! = βmc2 ψ − i!c α · ∇ψ ≡ HDirac ψ
∂t

● We want ψ also to satisfy KG equation ⇒ β, αx , αy , αz are matrices. Setting

! = c = 1:
∂2ψ ∂ψ ∂ψ
− 2 = βmi +α·∇
∂t ∂t ∂t
= β 2 m2 ψ − im(βα + αβ) · ∇ψ − (α · ∇)2 ψ
= m2 ψ − ∇2 ψ (KG equation)

Hence β 2 = αx2 = αy2 = αz2 = 1 and βαj + αj β = αj αk + αk αj = 0 for all

j (= k = x, y, z. This means that β, αx , αy , αz are (at least) 4 × 4 matrices.

22
● A suitable representation is
 
1 0 0 0  
 
 0 1 0 0  I 0
   
β =   ≡
 0 0 −1 0  0 −I
 
0 0 0 −1
 
0 σj
αj =  
σj 0

where σj are the Pauli matrices:

     
0 1 0 −i 1 0
σx =   , σy =   , σz =  
1 0 i 0 0 −1

23
● Then ψ is represented by a 4-component object called a spinor (not a 4-vector!)
 
ψ1
 
 ψ 
 2 
ψ= 
 ψ3 
 
ψ4

N.B. Each component ψ1,2,3,4 satisfies the KG equation.

● For a particle at rest, ψ = φ exp(−imc2 t/!), Dirac equation ⇒ φ = βφ, and so
 
φ1
 
 φ 
 2 
φ= 
 0 
 
0

where φ1,2 tell us the spin orientation.

24
 +imc2 t/!
● For antiparticle at rest, ψ = φe ⇒ φ = −βφ, so
 
0
 
 0 
 
φ= 
 φ3 
 
φ4

where φ3,4 now give spin orientation.

25
 Spin of Dirac Particles
● How do we prove that Dirac equation corresponds to spin one-half? We must
show that there exists an operator S such that J = L + S is a constant of
motion, and (! = 1) S 2 = S(S + 1) = 34 I.
● Note first that L = r × p is not a constant of motion:
H = βm + α · p
[Lz , H] = [x, H]py − [y, H]px
= iαx py − iαy px .
In general, [L, H] = iα × p (= 0.
● Thus we need [S, H] = −iα × p.
This is true if S = 12 Σ where
 
σj 0
Σj =   = −iαx αy αz α .
0 σj

Then S 2 = 14 (Σ2x + Σ2y + Σ2z ) = 34 I, proving that S = 12 .

26
 Magnetic Moment
● In an electromagnetic field we make the usual minimal substitutions:

H → H − eV , p → p − eA

in the Dirac equation, to obtain

H = α · (p − eA) + βm + eV

● Note that we no longer get the KG equation when we “square”:

,
(H − eV ) =
2
αj αk (pj − eAj )(pk − eAk ) + m2
j,k
,
= (p − eA) + m − e
2 2
(αj αk pj Ak + αj αk Aj pk )
j,k

Now for j (= k,

αj αk = iεjkl Σl , pj Ak = Ak pj − i∇j Ak
εjkl Σl ∇j Ak = Σ · (∇ × A) = Σ · B
27
 Hence

(H − eV )2 = (p − eA)2 + m2 − eΣ · B
1 e
H − eV * m+ (p − eA)2 − Σ·B
2m 2m
● This corresponds to a magnetic moment
e " e #
µ = S = ge S
m 2m
where ge = 2 (experiment ⇒ 2.0023193. . . ).

28
 Dirac Density and Current
● Write Dirac equation as
∂ψ
= −imβψ − α · (∇ψ)
∂t

● Transpose and complex conjugate:

∂ψ †
= +imψ † β − (∇ψ † ) · α
∂t
N.B. β, α are hermitian. Hence
∂ †
(ψ ψ) = −∇(ψ † αψ)
∂t

● Thus we can take

ρ = ψ † ψ ≡ |ψ1 |2 + |ψ2 |2 + |ψ3 |2 + |ψ4 |2

J = ψ † αψ
29
 N.B. Density ρ is positive definite! This is what Dirac wanted, but it is really a
problem – what about antiparticles?!
● Answer will not come until we learn some quantum field theory.

30
 Covariant Notation
● Nobody uses α and β any more. Instead we define γ-matrices:

γ0 = β , γ j = βαj (j = 1, 2, 3)

⇒ γ µ γ ν + γ ν γ µ ≡ {γ µ , γ ν } = 2g µν . Also define

ψ̄ ≡ ψ † β = (ψ1∗ , ψ2∗ , −ψ3∗ , −ψ4∗ )

in usual (‘Bjorken and Drell’) representation. Then

ρ = ψ † ψ = ψ † β 2 ψ = ψ̄γ 0 ψ
J = ψ † αψ = ψ † β 2 αψ = ψ̄γψ

and J µ is a 4-vector:
J µ = (ρ, J ) = ψ̄γ µ ψ

31
● We can also show that

ψ̄ψ = |ψ1 |2 + |ψ2 |2 − |ψ3 |2 − |ψ4 |2

transforms like a scalar (invariant) under Lorentz transformations.

● Multiplying through by β, Dirac equation becomes
∂ψ
iγ 0 = mψ − iγ j ∇j ψ
∂t
Hence

(γ µ ∂µ + im)ψ = 0
(γ µ pµ − m)ψ = 0

32
 Free-Particle Spinors
● A positive-energy plane wave

ψ = u(E, p) exp(ip · r − iEt)

satisfies (γ µ pµ − m)u = 0. Writing

 
  φ1
 
φ  φ2 
 
u= = 
χ  χ 
 1 
χ2

this means that   

E−m −σ · p φ
  =0
+σ · p −E − m χ

σ·p
Thus χ= φ
E+m

33
● Remember that  
σ 0
S= Σ= 1 1  
2 2
0 σ
Hence
 
1
φ = N  for spin up (along z-axis)
0
 
0
= N  for spin down
1

We have also
       
1 pz 0 px − ipy
σ · p =  , σ · p = 
0 px + ipy 1 −pz

34
 Thus    
1 0
   
 0   1 
   
u↑ = N   , u↓ = N   .
 pz   px −ipy 
 E+m   E+m 
px +ipy −pz
E+m E+m

❖ Normalization is as usual ρ = ψ † ψ = u† u = 2E particles per unit volume.

This gives 3 4
px + py + pz
2 2 2
N 1+
2
= 2E
(E + m)2
√
Using p = E − m gives N = E + m.
2 2 2

❖ Notice that the ‘small’ (3,4) components are O(v/c) relative to ‘large’ ones
(1,2).
● For antiparticle of 4-momentum (E, p) we need solution with pµ → (−E, −p):

ψ = v(E, p) exp(−ip · r + iEt)

35
 From the Dirac equation we now find
  
−E − m σ·p φ
  =0
−σ · p E − m χ

σ·p
Thus φ= χ
E+m

● Like 4-momentum, spin must be reversed, so

   
px −ipy pz
E+m E+m
   
 −pz   px +ipy 
   
v↑ = N  E+m
 , v↓ = N  E+m
 .
 0   1 
   
1 0

36
 Charge Conjugation
● Like the KG equation, the Dirac equation has charge conjugation symmetry. If
ψ is a negative-energy solution, there is a transformation

ψ → ψ c = Cψ ∗

such that ψ c is a positive-energy solution for charge −e. To find C:

γ µ (∂µ + ieAµ )ψ + imψ = 0

⇒ γ ∗µ (∂µ − ieAµ )ψ ∗ − imψ ∗ = 0
⇒ −Cγ ∗µ C −1 (∂µ − ieAµ )ψ c + imψ c = 0 .

Hence we need Cγ ∗µ C −1 = −γ µ , i.e. γ µ C = −Cγ ∗µ .

37
● Since all γ µ are real except γ 2 (which is pure imaginary) in our standard
representation, we can take
 
0 0 0 1
 
 0 0 −1 0 
 
C = iγ 2 =  
 0 −1 0 0 
 
1 0 0 0

Explicitly, for free particles, v ↑c = u↑ , v ↓c = −u↓

38
 Parity Invariance
● Similarly if ψ(r, t) is a solution of the Dirac equation, there exists a
transformation
ψ(r, t) → ψ P (r, t) = P ψ(−r, t)
such that ψ P is also a solution. Now
% &
∂
γ 0 − γ · ∇ + im ψ(r, t) = 0
∂t
% &
∂
⇒ γ 0 + γ · ∇ + im ψ(−r, t) = 0
∂t
% &
∂
⇒ P γ 0 P −1 + P γP −1 · ∇ + im ψ P (r, t) = 0 .
∂t
Hence we need P γ 0 P −1 = γ 0 , P γP −1 = −γ ,
i.e P γ 0 = γ 0 P , P γ j = −γ j P (j = 1, 2, 3)

39
● These relations are satisfied by
 
1 0 0 0
 
 0 1 0 0 
 
P = γ0 =  
 0 0 −1 0 
 
0 0 0 −1

● For a particle at rest,

ψ = u(m, 0) e−imt , ψ P = +ψ

but for an antiparticle at rest

ψ = v(m, 0) e+imt , ψ P = −ψ

Thus particle and antiparticle have opposite intrinsic parity.

● Notice that for KG equation the parity transformation is simply

φ(r, t) → φP (r, t) = φ(−r, t)

40
 i.e. φ is a true scalar function, since
% 2 &
∂
2
− ∇2
+ m2
φ(r, t) = 0
∂t
% 2 &
∂
⇒ 2
− ∇2
+ m2
φ(−r, t) = 0
∂t

● For the Dirac equation, the scalar is not ψ but

Φ = ψ̄ψ = ψ † γ 0 ψ

Check:

Φ(r, t) = ψ † (r, t)γ 0 ψ(r, t)

ΦP (r, t) = ψ † (−r, t)γ 0† γ 0 γ 0 ψ(−r, t)
= ψ † (−r, t)γ 0 ψ(−r, t)
= Φ(−r, t)

41
● Similarly, J µ is a true vector:

J µ (r, t) = ψ † (r, t)γ 0 γ µ ψ(r, t)

J P µ (r, t) = ψ † (−r, t)γ 0† γ 0 γ µ γ 0 ψ(−r, t)

But γ 0† γ 0 γ µ γ 0 = γ µ γ 0 = γ 0 γ µ for µ = 0, = −γ 0 γ µ for µ = 1, 2, 3. Hence, as

expected for a true vector,

J P 0 (r, t) = J 0 (−r, t) , J P (r, t) = −J (−r, t) .

● Weak interactions involve the axial current

µ
JA = ψ̄γ µ γ 5 ψ

where  
0 I
γ = iγ γ γ γ = 
5 0 1 2 3 
I 0
in our standard representation.
42
 µ
● Under parity transformations JA is an axial vector:
Pµ
JA (r, t) = ψ † (−r, t)γ µ γ 5 γ 0 ψ(−r, t)

Now γ 5 γ 0 = −γ 0 γ 5 (actually γ 5 γ µ = −γ µ γ 5 for all µ = 0, 1, 2, 3), so

P0
JA (r, t) = −J 0 (−r, t) , A (r, t) = J (−r, t)
JP

as expected for an axial vector.

● Similarly ΦP = ψ̄γ 5 ψ is a pseudoscalar

ΦP
P (r, t) = ψ (−r, t)γ γ ψ(−r, t)
† 5 0

= −ψ̄(−r, t)γ 5 ψ(−r, t)

= −ΦP (−r, t)

43
 Massless Dirac Particles
● For m = 0 the positive-energy free particle solutions are

ψ = u(E, p) exp(ip · r − iEt)

 
φ
where E = |p| and so u =   gives
χ
  
|p| −σ · p φ
  =0
σ·p −|p| χ

Hence χ = Λφ where Λ = σ · p/|p| is the helicity operator: Λ = ±1 for spin

aligned along/against direction of p (‘right/left-handed’)
● Note that if ψ represents a massless particle then
 
Λφ
γ ψ=
5   = Λψ (Λ2 = 1)
φ

44
● Hence γ 5 is the helicity operator for massless particles (minus helicity for
massless antiparticles).
● Weak interactions are ‘V–A’, i.e. they involve the current
µ
(J µ − JA )f i = ψ̄f γ µ (1 − γ 5 )ψi

If i is a massless particle, then (1 − γ 5 )ψi vanishes for helicity +1, i.e. only
left-handed states interact. The same applies to particle f , since
† 0 5 6† 0 µ
ψ̄f γ (1 − γ )ψi = ψf γ (1 + γ )γ ψi = (1 − γ )ψf γ γ )ψi
µ 5 5 µ 5

❖ Thus if neutrinos are massless, only left-handed neutrinos (right-handed

antineutrinos) interact.
❖ In the Standard Model, neutrinos are massless and right-handed neutrinos
do not exist.
❖ This is consistent with relativity, because helicity is frame-independent for
massless particles.
❖ In reality neutrinos do have mass, so both helicities must exist, but the
right-handed states interact more weakly (as for electrons).
45
 Electromagnetic Interactions
● We already saw that in an e.m. field Dirac Hamiltonian is
H = α · (p − eA) + βm + eV
= H0 + eγ 0 γ µ Aµ
where H0 is the free-particle Hamiltonian.
● Hence first-order perturbation theory gives a transition amplitude
+
Af i = −i ψf† (eγ 0 γ µ Aµ )ψi d4 x
+
= −ie Jfµi Aµ d4 x

where Jfµi = ψ̄f γ µ ψi .

● For plane waves, ψf,i = uf,i e−ip·x , and so the only difference from the KG
(spin zero) case is that we need a vertex factor of
−ieūf γ µ ui
for spin one-half instead of −ie(pf + pi )µ for spin 0.
46
● Invariant matrix element is
iea eb
Mf i = (ū a ! γ ua )(ūb! γµ ub )
µ
q2

p pa’
a

q=p ’−p
b b
p p’
b b
Hence
2 2
ea eb µν b
|Mf i | = 2 La Lµν
2
t
where

Lµν
a = (ūa! γ µ ua )(ūa! γ ν ua )∗
Lbµν = (ūb! γµ ub )(ūb! γν ub )∗

47
● For given spin states of a, b, a! and b! , we can evaluate these tensors explicitly
using the above expressions for free-particle spinors. However, we often
consider unpolarized scattering, when we have to average over initial and sum
over final spin states. Then
,
La = 2
µν 1
(ūa! γ µ ua )(ūa! γ ν ua )∗
spins

and similarly for Lbµν . This can be evaluated using the algebra of the
γ-matrices.

48
 Gamma Matrix Algebra
● The tensor ,
Lµν
a = 1
2
(ūa! γ µ ua )(ūa! γ ν ua )∗
spins
can be expressed in terms of traces of products of γ-matrices, using
,
uū = u↑ ū↑ + u↓ ū↓ = γ µ pµ + m
spins

We also have
(ūa! γ ν ua )∗ = (u†a! γ 0 γ ν ua )∗ = u†a γ ν† γ 0 ua! = ūa γ ν ua!
since γ ν† γ 0 = γ 0 γ ν .
● Thus ,
Lµν
a = 1
2
ūa! γ µ (( pa + ma )γ ν ua!
a! spins
where we use Feynman’s notation ( p = γ µ pµ . Putting in Dirac matrix indices,
ūα Γαβ uβ = Tr (uūΓ). Hence
Lµν
a = 1
2
Tr {(( p!a + ma )γ µ (( pa + ma )γ ν }
49
 kµ kν! Lµν
a = 1
2
Tr {(( p!a + ma )( k (( pa + ma )( k ! }
= 1
2
Tr {( p!a( k( pa( k ! } + 12 m2a Tr {( k( k ! }

= 2(p!a · kpa · k ! + pa · kp!a · k ! − pa · p!a k · k ! + m2a k · k ! )

(see examples sheet for last step).
● Removing the arbitrary vectors kµ and kν! ,
5 µ !ν 6
La = 2 pa pa + pa pa − (pa · pa − ma )g
µν !µ ν ! 2 µν

and similarly
5 6
Lbµν = 2 pbµ pbν + pbµ pbν − (pb · pb − mb )gµν
! ! ! 2

so

Lµν
a Lb
µν = 8(p a · p b p !
a · p !
b + p a · p ! !
p
b a · p b − m2
a p b · p !
b − m2
b p a · p !
a + 2ma mb )
2 2

50
● Expressing this in terms of the Mandelstam invariants s, t and u, we find an
invariant differential cross section
dσ e2a e2b 5 2 6
= s + u − 4(ma + mb )(s + u) + 6(ma + mb )
2 2 2 2 2 2
dt 32πst (pa )
2 ∗ 2

● For processes involving Dirac antiparticles, we should use the v-spinors in

place of u’s:
p
a

q
−
pa
“ui ”=ua , “ūf ”=v̄a! ⇒ vertex factor −iev̄a! γ µ ua .

p−a −
pa’
q

“ui ”=va! , “ūf ”=v̄a ⇒ vertex factor −iev̄a γ µ va! .

51
● We also need
,
vv̄ = v ↑ v̄ ↑ + v ↓ v̄ ↓ = ( p − m
spins

N.B. Different sign of m!

● Note, however, that the tensor Lµν a only involves m a . Replacing a by ā
2

reverses sign of ma , which does not affect the (unpolarized) scattering cross
section. Hence ab, āb, ab̄ and āb̄ scattering (by single photon exchange) are all
the same.

52
 Compton Scattering
ε ε’ ε ε’
k k’ k k’
+
p+k p−k ’
p p’ p p’
● In the Compton scattering process γ + e → γ + e, we need the propagator
factor for a virtual Dirac particle. This is
i , i(( q + m)
uū =
q 2 − m2 spins q 2 − m2

● Compare with photon propagator

i , i(−gµν )
ε ε∗
µ ν “=”
q 2 spins q2

53
 Thus the two graphs give Mf i = M1 + M2 where
i(( p + ( k + m)
M1 = ε!ν ū! (−ieγ ν ) (−ieγ µ
)uεµ
(p + k)2 − m2
µ i(( p − ( k + m)
!
M2 = εµ ū (−ieγ )
!
(−ieγ ν
)uε!
(p − k ! )2 − m2 ν

● The relative phase is +1 because the graphs differ by exchange of identical

bosons.
● For the unpolarized case, we want to average over initial spin states and sum
over final ones. We know how to do this for the electrons. For the photons,
consider the incoming polarization εµ . We can write schematically
, ,
|M1 + M2 | =
2
εµ ε∗λ M µλ
spins ε=εx ,εy

where the tensor M µλ is to be determined. However, we know that it must

have the properties kµ M µλ = kλ M µλ = 0 to ensure gauge invariance, which
allows us to replace εµ → εµ + akµ for any a.

54
● Choose z-axis along k: kµ = |k|(1, 0, 0, −1). Then above property implies
M 00 = M 03 = M 30 = M 33 , while
,
εµ ε∗λ M µλ = M 11 + M 22
ε=εx ,εy

= M 11 + M 22 + M 33 − M 00
= −Mµµ = −gµλ M µλ

● Thus, due to gauge invariance, we can replace photon polarization sum by

−gµλ .
● Applying the same trick to the outgoing photon polarization (ε!ν ) sum, we find
a contribution from the first diagram of
1 , e4
|M1 | =
2
Tr {(( p !
+ m)γ ν
(( p + ( k + m)γ µ
(( p + m)γµ (( p + ( k + m)γν }
4 spins 4(s − m2 )2

55
● In the extreme relativistic limit (s, |t|, |u| + m2 ), this becomes (using results
on examples sheet)
e4 e4
2
Tr {( p (( p + ( k )( p !
(( p + ( k )} = 8 2 (p · k)(p! · k)
s s
4u
= −2e
s
● Other diagram and interference terms are left as an exercise.

56