ANNIHILATION

In these notes I explain the e+ e annihilation process. At the tree level of QED,
there are two diagrams related by interchanging of the two photons in the final state:
1

e+

e+

(1)

The net amplitude due to these diagrams is

M = e (k1 , 1 ) e (k2 , 2 ) M ,

M = M
1 + M2 ,
i
(ie )u(e ),
iM
= v(e+ )(ie )
1
6q m
i
(ie )u(e ),
iM
= v(e+ )(ie )
2
6 q m

(2)

where q = p k1 = k2 p+ and q = p k2 = k1 p+ . Note the opposite orders of the

and vertices in the M1 and the M2 amplitudes. We may re-write these amplitudes
without matrix denominators using
1
6q + m
6q + m
= 2
=
2
6q m
q m
t m2

and

1
6 q + m
6 q + m
= 2
=
.
2
6 q m
q m
u m2

(3)

Consequently,
e2
v (6 q + m) u,
t m2
e2
=
v (6 q + m) u.
u m2

M
=
1
M
2

(4)

Ward Identity
Before we go any further, lets check the Ward identities for the annihilation amplitude:
for the first photon we should have k1 M = 0, and for the second photon k2 M = 0.
Lets start with the first photon and the first diagram. Multiplying the second factor in the
first eq. (4) by k1 , we have
v (6 q + m) u k1 = v (6 p 6 k1 + m) 6 k1 u
= v (6 p + m) 6 k1 u hhbecause 6 k1 6 k1 = k12 = 0ii



= v 2(p k1 ) 6 k1 (6 p m) u

(5)

= 2(p k1 ) v u hhbecause (6 p m)u = 0ii

= (m2 t) v u
and consequently
2
u .
M
1 k1 = +e v

(6)

Note the non-zero right hand side the first diagram does not satisfy the Ward identity all
by itself. As for the second diagram, we have
v (6 q + m) u k1 = v 6 k1 (6 k1 6 p+ + m) u
= v 6 k1 ( 6 p+ + m) u hhbecause 6 k1 6 k1 = k12 = 0ii


= v 2(p+ k1 ) + (6 p+ + m) 6 k1 u

(7)

= 2(p+ k1 ) v u hhbecause v(6 p+ + m) = 0ii

= (m2 u) v u

and consequently
2
M
u .
2 k1 = e v

(8)

Again we have a non-zero result the second diagram also does not satisfy the Ward identity
all by itself. However, the right hand sides of eqs. (6) and (8) cancel each other, so together,
2

the two diagrams do satisfy the Ward identity:

M k1 = M
1 k1 + M2 k1 = 0.

(9)

This is an example of a general rule: The Ward Identity does not work diagram by diagram
but only for entire amplitudes, or for partial sums of all diagrams related by permutations
of photonic vertices on the same fermionic line.
The Ward identity M k2 = 0 for the second photon works similarly to the first, and
I see no point in repeating the argument. Indeed, it would be an exactly similar argument
because the net annihilation amplitude is symmetric with respect to the two photons.
Summing over the Spins and Polarizations
Earlier in class I explained how to use Ward identities to sum |M|2 over polarizations
of the two photons:
X

|M|2 = +M M .

(10)

1 ,2

Combining the two diagrams, we have

X

|M|2 = +M M = M
1 M1 + M2 M2 + 2 Re M1 M2 .

(11)

1 ,2

Note that this formula works despite the fact that M

1 and M2 do not satisfy the Ward

identities by themselves its enough that the sum M

1 + M2 satisfies the identities.

Thus, in light of eqs. (4),

X

1 ,2

|M|2 =

e4
v (6 q + m) u u (6 q + m) v
(t m2 )2
e4
v (6 q + m) u u (6 q + m) v
(u m2 )2


2e4

Re
v

(6
q
+
m)
u

(6
q

+
m)
v
.
+

(t m2 )(u m2 )
+

(12)

This takes care of the photon polarizations. The next step is to average over spins of
3

the initial electron and positron. Proceeding is usual, we have

X X
|M|2 41
|M|2
s ,s+ 1 ,2

e4
2e4
e4

A
+

A
+
Re A12 ,
=
11
22
(t m2 )2
(u m2 )2
(t m2 )(u m2 )

(13)

where


Tr (6 p+ m) (6 q + m) (6 p + m) (6 q + m) ,


= 14 Tr (6 p+ m) (6 q + m) (6 p + m) (6 q + m) ,


= 14 Tr (6 p+ m) (6 q + m) (6 p + m) (6 q + m) ,

A11 =
A22
A12

1
4

(14)

Traceology 1
Our next task is to evaluate the traces (14). Lets start with the A11 .
Back in homework set #5, you saw that = 4 and 6 p = 26 p. Applying these
formulae to the expression inside the trace in A11 , we have
(6 p + m) = 2(6 p 2m),

(6 p+ m) = 2(6 p+ + 2m),

(15)

and consequently


A11 = Tr (6 p+ + 2m)(6 q + m)(6 p 2m)(6 q + m) .

(16)

Next, we expand the parentheses inside this trace and throw away terms with odd numbers
of momenta 6 p or 6 q. This gives us
A11 = Tr(6 p+ 6 q 6 p 6 q) + m2 Tr(6 p+ 6 p ) 4m2 Tr(6 q 6 q)
+ 2 2m2 Tr(6 p 6 q) 2 2m2 Tr(6 p+ 6 q) 4m4 Tr(1)
= 8(p+ q) (p q) 4(p+ p ) q 2 + 4m2 (p+ p )
16m2 q 2 + 16m2 (p q) 16m2 (p+ q) 16m4 .


= 8(p+ q)(p q) 4(q 2 m2 ) (p+ p ) 16m2 q 2 (p q) + (p+ q) + m2 .
(17)
We may further simplify this formula by expressing all the momenta products in terms
4

of the Mandelstams variables s, t, and u. Using p2 = p2+ = m2 and k12 = k22 = 0, we have
q 2 = (p k1 )2 = t,
qp = (p k1 )p = m2 p k1 = m2 + 21 (t m2 ) = + 12 (m2 + t),
qp+ = (k2 p+ )p+ = p+ k2 m2 = 21 (t m2 ) m2 = 12 (t + m2 ),
p p+ =

(18)

2
1
2 (s 2m ).

Consequently, on the last line of eq. (17), the last term vanishes
q 2 (p q) + (p+ q) + m2 = t 12 (t + m2 ) 21 (t + m2 ) + m2 = 0

(19)

while the remaining terms add up to

A11 = 8(p+ q)(p q) 4(q 2 m2 ) (p+ p )
= 2(t + m2 )2 2(t m2 ) (s 2m2 = t u)
= 2t2 4t m2 2m4 + 2t2 + 2tu 2t m2 2u m2

(20)

= 2tu 6tm2 2um2 2m4

= 2(t m2 ) (u 3m2 ) 8m4 .
This completes our evaluation of the first trace.
Now consider the second trace A22 . Instead of working through the calculation, we may
use the photon exchange / crossing symmetry between the two diagrams (1). This symmetry
exchanges t u and also A11 A22 , thus
A22 = 2(u m2 ) (t 3m2 ) 8m4 .

(21)

Traceology 2
Now we need to evaluate the third trace A12 which accounts for the interference between
the two diagrams (1). This trace is more complicated, so lets start by simplifying the
5

 part. Back in homework #5, we had

6 a = 2 6 a,

6 a 6 b = 4(ab),

6 a 6 b 6 c = 2 6 c 6 b 6 a,

(22)

which now gives us

(6 q + m) (6 p + m) = 2m2 + 4m(q + p ) 2 6 p 6 q .

(23)

Plugging this formula into eq. (14) for the A12 and remembering that we need an even
number of slash-momentum or factors inside the trace, we obtain


Tr (6 q + m) (6 p + m) (6 q + m) (6 p+ m)
h
i

i h


= Tr m(q + p ) 21 m2 + 6 p 6 q m 6 p+ 6 q + 6 q 6 p+ m2 )







= m2 (q + p ) Tr 6 p+ 6 q 21 Tr m2 + 6 p 6 q 6 q 6 p+ m2

A12 =

1
4

= m(q + p ) 4m(p+ q)


12 Tr 6 p 6 q 6 q 6 p+ m2 6 p 6 q + m2 6 q 6 p+ m4

= 4m2 (q + p ) (p+ q)


2
2
4
1
2 Tr 4(q q) 6 p 6 p+ + 2m 6 p 6 q 2m 6 q6 p+ 4m


= 4m2 (q q) + (qp+ ) (
qp ) + (p p+ )
8(q q)(p p+ ) 4m2 (p q) + 4m2 (
q p+ ) + 8m4 .

(24)
To simplify this rather messy formula, we need to work out the kinematics. Besides
eqs. (18), we have
qp = (p k2 )p = m2 k2 p = m2 + 12 (u m2 ) = + 21 (u + m2 ),
qp+ = (k1 p+ )p+ = k1 p+ m2 = 21 (u m2 ) m2 = 12 (u + m2 ),
qq = (p k2 )(p k1 ) = p2 p (k1 + k2 = p + p+ ) + k1 k2
= k 1 k 2 p p+ =

1
2s

21 (s 2m2 ) = m2 .
6

(25)

Consequently,
A12



2
2
2
2
1
1
1
= 4m m 2 (t + m ) 2 (u + m ) + 2 (s 2m )
2

4m2 (s 2m2 ) 2m2 (t + m2 ) 2m2 (u + m2 ) + 8m4



= 4m2 +m2 (t + m2 ) (u + m2 ) 12 (s 2m2 )

(26)

= 2m2 (2t + 2u + s)

= 2m2 (t + u + 2m2 )
= 2m2 (t m2 ) 2m2 (u m2 ) 8m4 .

Annihilation Summary
Having worked out the traces, lets plug them into eq. (13):


e4
2
2
4

2(t

m
)(u

3m
)

8m
(t m2 )2


e4
2
2
4
+
2(u m )(t 3m ) 8m
(u m2 )2


2e4
2
2
2
2
4

2m
(t

m
)

2m
(u

m
)

8m
+
(t m2 )(u m2 )


2
t 3m2
2m2
2m2
4 u 3m
= 2e
+

t m2
u m2
u m2
t m2
2

1
1
4 4
+
8e m
t m2
u m2



2
t m2
1
1
4 um
2
= 2e
+
4m
+
t m2
u m2
t m2
u m2
2 

1
1
4
+
,
4m
t m2
u m2

|M|2 =

(27)

or more compactly
|M|2 = 2e4

"

t m2
u m2
+
+1
t m2
u m2

This is our final result; the rest is kinematics.

7


1+

2m2
2m2
+
t m2
u m2

2 #

(28)

Annihilation Kinematics
In the center of mass frame, p = (E, p) where E = +
where = |k| = E. Consequently,

p2 + m2 , and k1,2
= (, k)

s = 4E 2 ,
t = (p k)2 = p2 E 2 + 2|p|E cos ,
u = (p + k)2 = p2 E 2 2|p|E cos ,

(29)

t m2 = 2E(E |p| cos ),

u m2 = 2E(E + |p| cos ),
and therefore
u m2
t m2
E + |p| cos
E |p| cos
+
+1 =
+
+1
2
2
tm
um
E |p| cos
E + |p| cos
=

3E 2 + p2 cos2
E 2 p2 cos2

3m2 + p2 (3 + cos2 )
,
m2 + p2 sin2


1
1
1
=
+
2E E |p| cos
E + |p| cos

=
1
1
+
2
tm
u m2

1+

(30)

1
2E
1
,
2
=
2
2
2E E p cos
m2 + p2 sin2

2m2
p2 sin2 m2
2m2
+
=
.
t m2
u m2
p2 sin2 + m2

Thus
|M|2 = 2e4

"

3m2

+ p2 (3 + cos2 )

m2

+ p2 sin2

p2 sin2
p2 sin2

2
m2

+ m2

(31)

and finally the partial cross section of annihilation

"
2 #
 2 2
d(e+ e )
p sin m2
|k| |M|2
2
3m2 + p2 (3 + cos2 )
.

=
=

dc.m.
|p| 64 2 s
8E|p|
m2 + p2 sin2
p2 sin2 + m2
(32)
8

For the non-relativistic electron and positron with |p| m, the expression in the square
brackets becomes 3 (1)2 = 2, hence isotropic partial cross section
d(slow e+ e )
2
=
.
dc.m.
4m|p|

(33)

And the total cross section in this limit is

tot (slow e+ e ) =

2
2
4

=
,
2
4m|p|
2m|p|

(34)

where total solid angle is 4/2 because of 2 identical photons in the final state.
In the opposite limit of ultra-relativistic e and e+ with |p| E m, we have
h i
3 + cos2
2(1 + cos2 )

1
=
sin2
sin2

(35)

and hence highly un-isotropic cross section

d(fast e+ e )
2
1 + cos2
.

dc.m.
4E 2
sin2

(36)

Note how this cross-section is strongly peaked in the forward direction = 0 where one
photon continues the electrons motion while the other continues the positrons motion.
According to eq. (36), the total annihilation cross-section
Z/2
d
tot (fast e+ e ) = 2 d sin
dcm

(37)

diverges at small angles, but thats an artefact of the approximation (35) becoming inaccurate
9

2
at small angles where p2 sin2 <
m . Instead, for small angles we have

h i
=

4p2
+ O(1)
m2 + p2 2

(38)

and consequently
d(fast e+ e )
2
2p2

.
dc.m.
4E 2 m2 + p2 2

(39)

This cross-section is strongly peaked in the forward direction, but it does not diverge. Instead,


2E
1
2
.
log

tot (fast e e ) =
E2
m
2
+

(40)

Compton Scattering
Compton scattering of an electron and a photon e e is related by crossing
symmetry to the e e+ annihilation. Indeed, at the tree level there are two diagrams

(41)

which are obviously related by s t crossing to the annihilation diagrams (1). Hence,
given eq. (28) for the annihilation, we may immediately write down a similar formula for the
Compton scattering without doing any work. All we need is to exchange s t in eq. (28)
and change the overall sign because we cross one fermion, thus
#
"

2
2
2 2
2
s

m
2m
2m
u

m
.

1 + 1+
+
|MCompton |2 = 2e4
s m2
u m2
s m2
u m2
This is it; all we need to do now is kinematics.
10

(42)

Compton scattering is usually studied in the lab frame where the initial electron is at
rest, p = (m, 0). In this frame, the initial and the final photon energies and are related
to photons scattering angle via Comptons formula
1
1 cos
1
=
+
,

me

(43)

originally written by Arthur Compton in terms of the photons wavelengths as

=

h
(1 cos ).
me c

(44)

According to this formula, there is an upper limit on the energy of the final photon for
any fixed 6= 0: regardless of the initial energy , the final energy can never exceed
me /(1 cos ).
The Comptons formula follows from the energymomentum conservation
+ E = + m and k + p = k + 0,

(45)

p2 = (k k )2 = k2 + k2 2k k = 2 + 2 2 cos

(46)

which imply

while
p2 + m2 = E 2 = ( + m )2 = 2 + 2 2 + 2m 2 m + m2 .

(47)

Subtracting these two formulae and canceling similar terms gives us

2m = 2 m + 2 (1 cos )
and hence eq. (43).
11

(48)

The Mandelstam variables s and u in the lab frame are

s (k + p)2 = ( + m)2 (k + 0)2 = 2m + m2 ,
u (k p)2 = ( m)2 (k 0)2 = 2 m + m2 ,

(49)

and hence
s m2 = +2m,

u m2 = 2m .

(50)

Plugging these values into eq. (42), we have

u m2
s m2

1
=
+
+ 1,
2
2
sm
um

2
2
2m
2m
m
m
1+
+
= 1+

2
2
sm
um

= cos

where the last equality follows from eq. (43), and therefore



2
4
Compton
2
+ 1 + cos .
|M
| = 2e

(51)

(52)

Finally, we need the phase space factor for the lab frame. For a generic 2 2 scattering
process,
d = |M|2 dP,

where

1
d3 p3
d3 p1

(2)4 (4) (p1 + p2 p1 p2 )

2E1 2E2 v (2)3 2E1

(2)2 2E2

1
3

d
p
(E
+
E

E
)
1
2
1
1
2
p2 =p1 +p2 p1
64 2 E1 E2 E1 E2 v
!1

+ E ) 
p2
d(E
d1
1
1
2 

.
=
64 2 E1 E2 E1 E2 v
d|p1 | p =p1 +p2 p

dP =

(53)

Specializing to the Compton scattering and the lab frame for initial electron, we immediately
obtain
d
2

dP =
64 2 m E

!1

dE 
.
1+
d p =kk

(54)

The only non-trivial issue here is the derivative in the parentheses. This derivative
should be taken for a fixed photon angle and before applying the energy conservation rule
12

Ee = + m . Instead, we use the momentum conservation p = k k , hence eq. (46)

for the p2 and consequently
E 2 = p2 + m2 = 2 + 2 2 cos + m2 .

(55)

For fixed and ,

2E dE = 2|p | d|p | = 2( cos ) d ,

(56)

dE
cos
=
.
d
E

(57)

and hence

Once we have taken this derivative, me may now use energy conservation, thus

1 +

dE
E + cos
m + cos
m
=
=
= ,

d
E
E
E

(58)

where the last equality follows from the Compton formula (43). Plugging the derivative (58)
into eq. (54), we arrive at
dP =

d
2

64 2 m2 2

(59)

and hence the KleinNishina formula for the partial cross-section:

2
d Compton
2

=
dlab
2m2e
2

+ sin2

(60)

where is given by eq. (43).

For low photon energies me , the Comptons formula gives , and the Klein
Nishina cross-section (60) becomes the good old Thompson cross-section
d Compton
d Thompson
2
(2 sin2 = 1 + cos2 ),

=
2
dlab
dlab
2me
13

(61)

and the total cross-section is

Thompson
total
=

8 2
0.663 barn.
3 m2e

(62)

On the other hand, for very high photon energies me and 6 0, we have

+ sin2 ,

(63)

and the KleinNishina formula becomes

d Compton

2
1
2

dlab
2m2e

2me 1 cos

(64)

p
< 2me / for which 6 , so the
This approximation is not accurate at small angles
cross section does not really diverge for 0. Instead, at small angles we have large but
finite partial cross-section
4 22 (2me /) + 2(2me /)2
2
d Compton

dlab
me
(2 + (2me /))3

(65)

and hence finite total cross-section

Compton
total



2
1
2
.
log
+

me
me
2

14

(66)