Lecture 13

Cross Sections and Decay Rates

Now that we know how to compute amplitudes for given processes, we would like to make
contact with observables such as cross sections and decay rates based on those amplitudes.
This will complete the path from computing correlation functions and then amplitudes,
which can be easily obtained by using the derived Feynman rules of a given theory.

13.1 The S Matrix

We will state the amplitude in the language of the S matrix. Let us consider a scattering
process with a given initial state and a final state. We define the asymptotic states by

|i, ini for t → −∞

|f, outi for t → +∞ , (13.1)

where the states labeled “in” are those asymptotic states created by creation operators
evaluated at times −∞, e.g. a† (−∞), etc; and the states labeled “out” are those created
by creation operators evaluated at times +∞, such as a† (+∞). These two distinct sets
of asymptotic states are the ones we have used up until now to write down the desired
amplitude

hf, out|i, ini . (13.2)

The “in” and “out” asymptotic states are however isomorphic, i.e. there are the same set
of states but labeled differently. We can define a unitary transformation S such that

|i, ini = S|i, outi , (13.3)

1
2 LECTURE 13. CROSS SECTIONS AND DECAY RATES

in such a way that we can rewrite (13.2) in terms of either both “in” or“out” states.

hf, out|i, ini = hf, in|S|i, ini = hf, out|S|i, outi ≡ hf |S|ii . (13.4)

The last equality stems from the fact that we can equally express the amplitude in terms
of the “in” or the “out” states as long as is an element of the S matrix. The S operator
can be written as

S ≡ 1 + iT , (13.5)

where we defined the T matrix elements. The identity in the first term in (13.5) reflects
the fact that the amplitude must include the possibility of no interaction. But in order to
compute a cross section we are only concerned with the part of the amplitude that allows
for interactions, i.e. the second term in (13.5). Schematically, we can express this as

hf |S|ii = disconnected diagrams + LSZ formula , (13.6)

where the contributions of disconnected diagrams comes from the identity in (13.5). Thus,
the LSZ formula will give the contribution of the T matrix to a given amplitude.

13.2 From Scattering Amplitudes to Cross Sections

In general we want to compute the transition probability from an initial state to a final
state. In practice, we are mainly interested in two cases: the decay of a particle to two
or more particles, and the scattering of two particles in the initial state into two or more
particles in the final state.
We start with the scattering process 2 → n. The transition amplitude is given by

hp1 . . . pn | iT |pA pB i ≡ (2π)4 δ (4) (PA + PB − P1 − · · · − Pn ) i A , (13.7)

where we have defined the amplitude A as the transition amplitude with the overall mo-
mentum conservation delta function already factored out. In order to obtain a probability,
we will define it as the squared of the transition amplitude appropriately normalized.

|hp1 . . . pn | iT |pA pB i|2

P ≡ , (13.8)
hp1 . . . pn |p1 . . . pn ihpA pB |pA pB i

where the denominator corresponds to the normalization of the initial and final states.
13.2. FROM SCATTERING AMPLITUDES TO CROSS SECTIONS 3

We start by considering the numerator of (13.8). This is

n
!2
X
|hp1 . . . pn | iT |pA pB i|2 = (2π)4 δ (4) (PA + PB − Pf ) |A|2
f =1
n
X
= (2π)4 δ (4) (PA + PB − Pf ) (2π)4 δ (4) (0) |A|2 , (13.9)
f =1

where f = 1, . . . , n labels the final state momenta. However, we can write

Z
(4) (3) 1
δ (0) = δ(0)δ (0) = d4 x ei0·x . (13.10)
(2π)4

If we consider for a moment a finite volume V and a finite time T , the integral in (13.10)
results in

(2π)4 δ (4) (0) = V T . (13.11)

For the denominator, we consider the asymptotic momentum eigenstates normalized ac-
cording to

2Ep a†p |0i ,

p
|pi = (13.12)

such that the normalization of an eigenstate of momentum p is given by

hp|pi = 2Ep h0|ap a†p |0i

= 2Ep (2π)3 δ (3) (p − p) = 2 Ep V , (13.13)

where in the last equality we used (13.10). Then, the two factors in the denominator of
(13.8) are

hpA pB |pA pB i = 2EA 2EB V 2

Y
hp1 . . . pn |p1 . . . pn i = 2E1 . . . 2En V n = (2Ef V ) . (13.14)
f

Replacing (13.10) and (13.14) into (13.8) and dividing by T , we obtain the probability of
transition for unit time
4 LECTURE 13. CROSS SECTIONS AND DECAY RATES

(2π)2 δ (4) (PA − PB − f Pf ) V |A|2

P
P
= Q . (13.15)
T 2EA 2EB V 2 f (2Ef V )

But this probability requires that we have precise knowledge of all final state momenta.
Often times we will need to either partially or totally integrate over the phase space of
the final states. For this we need to know the probability that a given final state particle
has momentum in the interval

(pf , pf + d3 pf ) , (13.16)

where d3 pf contains information about the momentum vector. We would like then to
convert (13.15) into the differential probability that the final states are in a region of the
final state phase space defined by (13.16). In order to obtain this we need to multiply
(13.15) by the number of states in each interval defined by (13.16) for each final state
particle. Given that we are using a finite volume V , the momentum of each final state
particle obeys the quantization rule

2π
p= (n1 , n2 , n3 ) , (13.17)
L

where L3 = V , and the ni with i = 1, 2, 3 refer to the number of states in each spatial
direction. Then, the number of states inside the interval (13.16) of size d3 p is

Ldpx Ldpy Ldpz

n1 n2 n3 =
2π 2π 2π
V d3 p
= (13.18)
(2π)3

Putting all these together we obtain the differential probability per unit time

n 
(2π)2 δ (4) (PA + PB − f Pf ) |A|2 Y
P
d3 pf

dP
= , (13.19)
T 2EA 2EB V f =1
(2π)3 2Ef

Finally, in order to convert this into a differential cross section we need to account for
the incident flux. In other words, we are interested in the differential probability per unit
time and per unit of initial flux so that we obtain a probability that depends intrinsically
on the amplitude A and the final state phase space, not on how intense our beams of
A and B particles were. The flux is the number of particles per unit volume times the
relative velocity of the particles. For instance, for a typical head on collision
13.2. FROM SCATTERING AMPLITUDES TO CROSS SECTIONS 5

Figure 13.1: Head on collision. pB = −pA .

the initial flux “seen” by either the A or the B particle is given by

|vAz − vBz |
. (13.20)
V

So dividing (13.19) by the flux in (13.20) we obtain

1 1 X Y  d3 pf 
4 (4) 2
dσ = (2π) δ (PA + PB − Pf ) |A| , (13.21)
2EA 2EB |vA − vB | f f
(2π)3 2Ef

which is the differential cross section for the scattering of the two initial particles with
momenta PA and PB going into an n-particle final state.
At this point we will make some comments:

• We can define the final state phase space by

n 
d3 pf
Z Z Y  X
dΠn ≡ (2π)4 δ (4) (PA + PB − Pf ) . (13.22)
f =1
(2π)3 2Ef f

It is separately Lorentz invariant.

• The amplitude squared |A|2 is also Lorentz invariant by itself.

• The factor

1
, (13.23)
EA EB |vAz − vBz |

is not Lorentz invariant, but it is invariant under boosts in the z direction.

6 LECTURE 13. CROSS SECTIONS AND DECAY RATES

13.3 Two-particle Final State

A very paradigmatic example is the scattering of two particles in the initial state into two
particles in the final state. We first compute the two-particle phase space for A+B → 1+2.
We will use the center of momentum frame. From (13.22) we have

d3 p1 1
Z Z Z 3
d p2 1
dΠ2 = 3
(2π)4 δ (4) (PA + PB − P1 − P2 )
(2π) 2E1 (2π)3 2E2
Z 3
d p1 1
= 2π δ(EA + EB − E1 − E2 ) , (13.24)
(2π)3 4E1 E2

where the second line is obtained by using the spatial delta function to perform the d3 p2
integral. The final momentum differential is

d3 p1 = p21 dp1 dΩ1 = p21 dp1 d cos θ1 dφ1 , (13.25)

with θ1 the angle of p1 with respect to the direction of the incoming momentum pA , and
φ1 the corresponding azimuthal angle. There is typically no azimuthal angle dependence
in |A|2 , so we can integrate over φ1 obtaining a factor of 2π. Then (13.24) now reads

p21 dp1
Z Z  q q 
dΠ2 = (2π d cos θ1 ) 2π δ EA + EB − p21 + m21 − p21 + m22 ,
(2π)3 4 E1 E2
(13.26)
where we have used that p1 = −p2 in the delta function, wich stems from the fact that
we have used the spatial delta function in the center of momentum frame. We are now in
a position to perform the integral in the absolute value of the spatial momentum of the
particle 1, p1 , by using the delta function. Restoring the differential solid angle to have a
more general expression, we have

p21
Z Z
dΩ1
dΠ2 =
(2π) 4E1 E2 | E1 + Ep12 |
2 p1
Z
1 p1
= 2
dΩ1 . (13.27)
16π E1 + E2

But, since E1 + E2 = ECM then we obtain

Z Z
1 p1
dΠ2 = 2
dΩ1 . (13.28)
16π ECM
13.4. DECAY RATE OF AN UNSTABLE PARTICLE 7

Let us compute now the cross section in the CM frame. It is

dσ 1 1 p1
= z z 2
|A|2 , (13.29)
dΩ 2EA 2EB |vA − vB | 16π ECM

where the solid angle refers to the final states particles, and z is the direction of the
incoming A particle.
If we now consider the relative velocity we have

pzA pz
|vAz − vBz | = − B . (13.30)
EA EB

If we now consider the simplified case mA = mB = m1 = m2 = m, we have

2 4pA 4p1
|vAz − vBz | = |pzA − (−pzA )| = = , (13.31)
ECM ECM ECM

Then, we arrive at a final expression for the angular distribution for scattering in the CM
of two particles into two particles, all of the same mass m:

 
dσ 1 1
= 2
|A|2 . (13.32)
dΩ CM 64π 2 ECM

13.4 Decay Rate of an Unstable Particle

If instead of considering the transition probability per unit time from a two-particle initial
state we start with a state of one particle, we are computing the decay rate for the process
A → 1 . . . n, for the decay of a particle A to n particles in the final state. The derivation
is just straightforward and the result is the differential decay probability per unit time
given by

n 
!
1 Y d3 pf 1
 X
4 (4)
dΓ = (2π) δ P A − Pf |A|2 , (13.33)
2mA f =1 (2π)3 2Ef f

where the factor of 2mA comes from using 2EA in the rest frame of the decaying particle,
and A is the amplitude for the decay process. For a given decay channel (i.e. a given
final state), the integral gives the so-called partial width of A into that channel
8 LECTURE 13. CROSS SECTIONS AND DECAY RATES

Z
Γ(A → f1 ) = dΓ(A → f1 ) . (13.34)

The total width of A is a property of the particle and corresponds to the sum of the
partial widths into all the available channels into which A can possibly decay

X
ΓA ≡ Γ(A → fi ) . (13.35)
i

The lifetime of tha particle is then the inverse of the total decay rate or total width.
Decay rates have units of energy, thus if we want the lifetime in seconds we can use

h̄
τA = . (13.36)
ΓA

For instance, if we initially have a given number of particles of type A, at a later time t
we have

N (t) = N (0) e−t/τA . (13.37)

The lifetime also determines the typical displacement of a particle produced before it
decays. This is

c τA γ , (13.38)

where c is the speed of light, and γ is the relativistic factor.

Finally, the propagation of an unstable particle is affected by its decays. We will show
later in the course that the propagator of a particle with open decay channels gets modified
to be

i
, (13.39)
p2 − m2A − iΓA mA

where we considered a scalar propagator and p is the four-momentum of A. We will

derive (13.39) in the context of renormalization and see that the new term appears as a
consequence of an imaginary shift in the pole of the propagator that arises due to the
existence of open decay channels for A. As a result, unstable particles appear in cross
sections for processes that are mediated by them as resonances of widths characterized by
ΓA . This is the reason why these particles are called resonances, and also why the total
decay rate ΓA is called the particle width.
13.4. DECAY RATE OF AN UNSTABLE PARTICLE 9

Additional suggested readings

• Quantum Field Theory, by C. Itzykson and J. Zuber, Chapter 5.1.

• Quantum Field Theory , by M. Srednicki, Chapter 5.

• The Quantum Theory of Fields, Vol. I, by S. Weinberg, Section 10.3.