Sie sind auf Seite 1von 34

4

Particles with spin one


Particles with spin have extra degrees of freedom which transform nontrivially
under rotations. Such particles are therefore characterized by the value of their
momentum and by the orientation of their spin. For spin-1 this orientation
is described by a three-dimensional spin vector. Quantum-mechanically, for
given momentum, there are three independent states distinguishable by the
value of the spin projected along a certain axis (in this case these values are 1
and 0 in units of ). However, massless particles with spin are special, because
their spin can only be directed parallel or anti-parallel to their direction of
motion. Hence, massless particles with spin have only two dierent spin states,
irrespective of the value of their total spin. This chapter deals with spin-1
particles, also called vector particles. Well-known examples of massive spin-1
particles are the mesons , , J/ and , and the weak intermediate vector
bosons W

and Z. The only known massless spin-1 particle is the photon.


Free massive spin-1 elds are described by the Proca Lagrangian, and massless
spin-1 elds by the Maxwell Lagrangian. We shall discuss four typical particle
reactions that involve photons, namely electromagnetic scattering of pions,
pion-Compton scattering, the decay
0
, and the radiative decay K
0
S

.
4.1. Massive spin-1 particles
The standard relativistic treatment of spin-1 particles is in terms of a four-
vector eld V

(x), whose free Lagrangian is the Proca Lagrangian,


L =
1
4
(

)
2

1
2
m
2
V
2

=
1
2
(

)
2
+
1
2

1
2
m
2
V
2

. (4.1)
The rst and the third term are a straightforward generalization of the Klein-
Gordon Lagrangian for scalar elds; the relevance of the second term will
become clear shortly. The eld equations corresponding to (4.1) follow from
Hamiltons principle, according to which we require that the action be sta-
tionary under small variations V

+ V

, where V

, vanishes at the
boundary of the integration domain of the action integral. Hence we consider,
S =
_
d
4
x
__

+m
2
V

(x)
_
= 0 . (4.2)
121
122 Particles with spin one [Ch.4
After integration by parts, using the boundary conditions on V

, this reads
S =
_
d
4
xV

_
V

m
2
V

= 0 , (4.3)
which implies the Proca eld equation
(m
2
)V

= 0 . (4.4)
For m
2
= 0 contraction with one more derivative gives

= 0, so that (4.4)
is equivalent to the following two equations,
(m
2
)V

= 0 ,

= 0 . (4.5)
The reason for the second term in (4.1) is now clear; its coecient was chosen
such that one obtains a Klein-Gordon equation for each of the components of
V

separately, together with a subsidiary restriction. The latter restricts the


number of independent plane-wave solutions of (4.5) to three, as is appropriate
for a spin-1 particle, i.e.,
V

(x) =

(k) e
ikx
, with k
2
= m
2
) , k (k) = 0 . (4.6)
It is straightforward to dene a set of independent polarization vectors

(k). In the rest frame k

= (0, m), so k (k) = 0 implies that the fourth


component of

vanishes. In a general frame one distinguishes two dier-


ent types of polarizations: two independent transversal polarization vectors
which are orthogonal to both k and k, and a longitudinal polarization vec-
tor whose spatial components are taken along the direction of k. Hence, for
k

= (k, (k)), we have two transverse polarizations,

(k; ) = (

, 0) , k

= 0 , ( = 1, 2) , (4.7)
and one longitudinal polarization,

(k; 0) =
_
(k)k
m|k|
,
|k|
m
_
, (4.8)
where we can adopt the convenient normalization
(k; ) (k;

) =

, ,

= 1, 2, 0 . (4.9)
Note that the distinction between transversal and longitudinal polarization
vectors is not a Lorentz invariant notion, since transversal and longitudinal
polarization vectors rotate into each other by a Lorentz transformation.
We have assumed above that the polarization vectors are real ((

).
The corresponding plane waves (4.6) are then linearly polarized. Another
4.1] Massive spin-1 particles 123
choice for the polarization vectors, which is often convenient, involves helic-
ity eigenvectors. Helicity measures the spin of the particle along its direction
of motion (in units of ). To measure the spin along k one applies a rota-
tion around k to the eld. For k pointing in the positive z-direction such a
(clockwise) rotation takes the form
V

(k) V

(k) = L

(k) ,
with
L

=
_
_
_
_
_
cos sin
sin cos
1
1
_
_
_
_
_
. (4.10)
A polarization vector with helicity is an eigenstate of the rotation matrix
with eigenvalue exp(i). Obviously, the longitudinal polarization vector (4.8)
is invariant under the rotation, and has thus zero helicity. The transverse
polarization vectors (4.7) now decompose into two helicity eigenvectors with
= 1, which are complex, namely (with k in the positive z-direction),

(k; ) =
1
2

2(1 , i , 0 , 0) (4.11)
while the zero-helicity vector is the longitudinal one given by (4.8). The po-
larization vectors (4.8) and (4.11) characterize incoming particles of corre-
sponding helicities. For outgoing particles one must use the conjugate vectors
polarization vectors (k; ). For complex polarization vectors we have the
(Lorentz invariant) orthonormality conditions,
(k, ) (k,

) =

, (,

= +, , 0) (4.12)
corresponding to (4.9).
Let us now turn to the Feynman rules. For vector elds the construction of
Feynman diagrams proceeds along the same lines as for scalar elds. Because
a vector eld consists of four separate components, the propagator takes the
form of a 4 4 matrix. Its derivation follows directly from the prescription
given in section 2.4. First we construct the Fourier transform of the action
corresponding to (4.1)
S[V

] =
1
2
(2)
4
_
d
4
k

V

(k)
_
(k
2
+m
2
)

(k) , (4.13)
where

V

(k) = V

(k) because V

(x) is a real eld. The propagator is the


inverse of the matrix in the integrand, viz.

(k) =
1
i(2)
2
_
(k
2
+m
2
)

1
. (4.14)
124 Particles with spin one [Ch.4
In order to calculate the inverse of this matrix it is convenient to rst decom-
pose

according to

(k) = A(k
2
)

+B(k
2
)k

, after which one nds


A(k
2
) and B(k
2
) by requiring that
i(2)
4

(k) [(k
2
+m
2
)

] =

.
The result is

(k) =
1
i(2)
4
1
k
2
+m
2
_

+
k

m
2
_
. (4.15)
As we have been emphasizing in chapter 2 the propagator poles at k
2
= m
2
are associated with physical particles. Therefore, at rst sight, it may seem
that (4.15) describes four rather than three physical degrees of freedom with
mass m, because of the overall factor (k
2
+m
2
)
1
. However, this is no the case.
The three polarization vectors (4.8) and (4.8) have zero innder product with
k

, and are thus eigenvectors of (4.15) with equal eigenvalues proportional


to (k
2
+ m
2
)
1
. The remaining eigenvector is proportional to k

, and its
eigenvalue has no pole at k
2
= m
2
. Indeed, one easily veries that

(k)k

=
k

i(2)
4
m
2
. (4.16)
Figure 4.1: Graphical representation of the propagator

(k).
The endpoints of the propagator lines carry a four-vector index, as is indi-
cated in g. 4.1. To dene the probability amplitude one follows the prescrip-
tion given in section 3.3. This leads to a residue matrix (

+m
2
k

) for
every external line. However, physical particles are characterized by polariza-
tion vectors

that are orthogonal to k

, so that the k

term drops out.


On the other hand, it is not really necessary to impose (??) as an independent
restriction; on the mass shell (k
2
= m
2
) the polarization vector

,
does not contribute by virtue of (4.16). By the same arguments wave functions
are decomposed as (cf. (3.3),
f

(x) =
1
(2)
3/2
_
d
3
k
_
2(k)
3

=1
f
()
(k)

(k, ) e
ikxi(k)t
,
4.2] Massive spin-1 particles 125
where one includes only the physical polarization vectors.
The invariant amplitude for a process involving n vector particles charac-
terized by polarization vectors

(k, ) takes the form


M

n
= M

1
(k
1
;
1
)

n
(k
n
;
n
) , (4.17)
where we have now dropped the k

-terms of the propagator residues for the


external lines. In the denition of the probability amplitude one now encoun-
ters the same normalization factors [(2)
3
2(k)]
1/2
as for scalar particles.
The transition probability for a given process is proportional to the square of
the invariant amplitude. When the details of the nal state are not of interest
one sums over all possible spin orientations of the outgoing particles. Fur-
thermore, as particle beams are often unpolarized, the spin of the incoming
particles may be unknown, so one averages over all the possible spins of the
incoming particles. Let us consider just one of the polarization vectors in the
invariant amplitude in order to show how the summation over spins is done.
Hence we write
M

= M

(k; ) . (4.18)
Remember that the spin of incoming particles is characterized by

(k, ),
while for outgoing particles one must take the conjugate vector

(k, ). The
square of (4.18) is
|M

|
2
=

(k, )

(k; ) M

(4.19)
Summing (4.19) over gives

,
_

(k; )

(k; )
_
M

. (4.20)
One way to evaluate the polarization sum makes use of explicit expressions
for the polarization vectors. For orthonormal vectors one always nds,

(k; )

(k; ) =

+
k

m
2
. (4.21)
This result does not come as a surprise, because the vectors

(k; ) span
the three-dimensional subspace that is orthogonal to the vector k

with k
2
=
m
2
; indeed, multiplying the right-hand side of (4.21) with k

gives zero on
the mass-shell.
126 Particles with spin one [Ch.4
4.2. Massless spin-1 particles
One might expect that the description of massless spin-1 particles should
follow directly from the theory of massive spin-1 particles, but we shall see
that this not the case. This is for instance obvious from the fact that the
expression for the longitudinal polarization vector (4.8) and the propagator
(4.15) are singular in the limit m 0. A primary reason for this is that the
massless limit of (4.1),
L =
1
4
(

)
2
, (4.22)
is invariant under local gauge transformations
A

(x) A

(x) +

(x). (4.23)
This transformation is familiar from Maxwells theory of electromagnetism
where the vector potential is subject to the same transformations (see sec-
tion 1.3). Therefore we have changed notation in this section and denote the
massless vector eld by A

. The electromagnetic eld strength equals


F

, (4.24)
and the particles described by this theory are called photons.
One consequence of an invariance under local gauge transformations is that
the theory depends on a smaller number of elds. Correspondingly the number
of plane wave solutions is also reduced in comparison to the massive case. To
see this explicitly, consider the eld equation following from (4.22) and inves-
tigate its possible plane-wave solutions. The eld equation is just Maxwells
equation,

) = 0 . (4.25)
In order to examine plane-wave solutions of this equation we consider the
Fourier transform of A

(x)
A

(k) = (2)
4
_
d
4
xA

(x) e
ikx
. (4.26)
Under gauge transformations A

(k) changes by a vector proportional to k

(k) A

(k) = A

(k) +(k) k

. (4.27)
The eld equation (4.25) now takes the form
k
2
A

(k) k

(k) = 0 , (4.28)
4.2] Massless spin-1 particles 127
which is manifestly invariant under the transformation (4.27). Decomposing
A

(k) into four independent vectors,

(k, ), k

and

k

, dened by
k

(k; ) = 0 ,
0
(k; ) = 0 , ( = 1, 2) ,
k

= (k, k
0
) ,

k

= (k, k
0
) , (4.29)
we decompose,
A

(k) = a

(k)

(k, ) +b(k)

+c(k) k

. (4.30)
The eld equation (4.28) now corresponds to
k
2
a

(k)

(k, ) +b(k) [k
2

k

(k

k) k

] = 0 , (4.31)
from which we infer for the coecient functions (note that k

k is positive),
k
2
a

(k) = 0 , ( = 1, 2) b(k) = 0 . (4.32)


Obviously the eld equation does not lead to any restriction on c(k). This
should not come as a surprise because c(k) can be changed arbitrarily by a
gauge transformation, whereas the eld equation is gauge invariant. Conse-
quently the eld equation cannot x the value of c(k). By means of a gauge
transformation we may adjust c(k) to zero, which shows that c(k) has no phys-
ical meaning. We thus nd that there are only two independent plane wave
solutions characterized by lightlike momenta (k
2
= 0) and the two transverse
polarization vectors.
The fact that we have two rather than three solutions indicates that a direct
generalization of the arguments of the previous section runs into diculties.
One such diculty is that the spin of a massless particle cannot be dened
by referring to its rest frame. Therefore, the three-dimensional rotation group
no longer plays the decisive role in order to characterize the spin, but rather
the group of two-dimensional rotations around the three-momentum k of the
particle. This complication is also apparent from the singular dependence
of the longitudinal polarization vector (4.8) on the mass. One may wonder
whether a characterization of the spin in terms of only transverse polarizations
will not lead to a violation of Lorentz invariance in view of the fact that
transversality is not preserved by Lorentz transformations. However, it turns
out that Lorentz invariance is preserved provided that the photon eld couples
to a conserved current (i.e.

= 0). As is well-known from electrodynamics


this is indeed the case (see, again, section 1.3. Conserved currents can be
viewed as a consequence of gauge invariance.
There is a further diculty when attempting to calculate Feynman diagrams
for massless spin-1 particles, again related to the aforementioned invariance
128 Particles with spin one [Ch.4
under gauge transformations. To show this consider the Fourier transform of
the action corresponding to (21.34), viz.
S[A

] =
1
2
(2)
4
_
d
4
k

A

(k)
_
k
2

(k) . (4.33)
According to the general prescription the propagator is proportional to the
inverse of [k
2

]. In this case, however, the inverse does not exist


because the matrix has null vector, as we see from
_
k
2

= 0 . (4.34)
The existence of the null vector is a direct consequence of the gauge invariance
of the theory. Gauge invariance implies that the theory contains fewer degrees
of freedom; this fact must reect itself in the presence of zero eigenvalues in
the quadratic part of the Lagrangian. Indeed, the null vector proportional to
k

is directly associated with the gauge transformation (4.27) in momentum


space. Obviously, the degree of freedom that is absent in (4.33) should not
reappear through the interactions. One can show that this is ensured provided
that the photon couples to a conserved current (see problem 4.5).
The standard way to circumvent the singular propagator problem is to make
use of a gauge condition. A convenient procedure is based on introducing the
missing (gauge) degrees of freedom, which formally spoils the gauge invariance.
However, the degrees of freedom are introduced only in order to make the
propagator well-dened and they will not aect the interactions of the theory.
Therefore, the eect of this procedure can still be separated from the true
gauge invariant part of the theory, and the physical consequences remain
unchanged. It is a rather subtle matter to prove that this is indeed the case.
In this section we only present the prescription for dening the propagator.
To do that one introduces a so-called gauge-xing term to the Lagrangian.
The most convenient choice is to add to (4.22)
L
g.f.
=
1
2
(

)
2
, (4.35)
where is an arbitrary parameter. Because of this term the Fourier transform
of the action corresponding to the combined Lagrangian becomes
S[A

] =
1
2
(2)
4
_
d
4
k

A

(k)
_
k
2

+
2
k

(k) , (4.36)
so that for = 0 the propagator is equal to

(k) =
1
i(2)
4
_
k
2

(1
2
)k

1
=
1
i(2)
4
1
k
2
_

(1
2
)
k

k
2
_
. (4.37)
4.3] Electromagnetic scattering of pions 129
Clearly the propagator has more poles at k
2
= 0 than there are physical pho-
tons (characterized by transversal polarizations). However, one must realize
that by making the above modication we have somewhat obscured the re-
lation between propagator poles and physical particles. In order to extract
the physical content of the theory one should contract the amplitude with
only transversal polarization vectors. This requirement forms an essential in-
gredient of the proof that physical results do not depend on the parameter
.
Using the propagator (4.37) one can now construct Feynman diagrams and
corresponding scattering and decay amplitudes for photons in the standard
fashion. The -dependent k

-term of the propagator residue vanishes when


contracting the invariant amplitude with transversal polarization vectors. In
order to sum over photon polarizations one may use (for orthonormal polar-
ization vectors)

=1,2

(k; )

(k; ) =

+

k

k
, (k
2
= 0) , (4.38)
where the vector

k

was dened in (4.29). In order to verify this equation we


note that the right-hand side of (4.38) is equal to

=1,2

(k; )

(k; ) =
_
_
_

|k|
2
for , = 1, 2, 3;
0 for and/or = 0.
(4.39)
Obviously (4.38)-(4.39) are not manifestly Lorentz covariant, which is related
to the fact that the transversality condition k (k, ) = 0 is not Lorentz
invariant. However, if the photons couple to conserved currents, such that the
amplitude vanishes when contracted with the photon momentum,
k

= 0, (4.40)
then the non-covariant terms in (4.38) may be dropped. Consequently, when
summing |M

(k; )|
2
over the transverse polarizations, one has

=1,2
|M

(k; )|
2
= M

, (4.41)
which is manifestly Lorentz invariant.
4.3. Electromagnetic scattering of pions
In order to illustrate the use of Feynman diagrams for spin-1 particles consider
the reaction caused by virtual photon exchange. This
+

process
130 Particles with spin one [Ch.4
is not of direct physical relevance. Pion targets or colliding pion beams are not
available, and even if they were it would be almost impossible to distinguish
the electromagnetic contributions to this scattering process from those of the
strong interactions, which are dominated by -meson exchange. Nevertheless it
is of interest to give the relevant expression for pure electromagnetic scattering
of pointlike spinless particles, in order to appreciate the more complicated but
analogous results for the scattering of pointlike spin-
1
2
fermions. The latter
will be discussed extensively in chapter 6..
The pion-photon coupling follows from the minimal substitution

ieA

in the free Klein-Gordon Lagrangian, where ( is a complex


scalar eld (e is the
+
charge). Combining this with Maxwells Lagrangian
gives
L =
1
4
F

m
2

ieA

) (

)] e
2
A
2

. (4.42)
The electromagnetic couplings are the same as used in section 2.3 to study
the scattering of pions by an external electromagnetic potential. An important
property of the Lagrangian (4.42), is its invariance under the combined gauge
transformations
A

(x) A

(x) +

(x) , (x) e
ie(x)
(x) . (4.43)
The propagators and vertices implied by (4.42) are shown in table 4.1. The
arrow on the pion line indicates the ow of (positive) charge rather than of the
momentum. Just as in section 2.3 we choose conventions such that an outgoing
arrow on an external line indicates the emission of a
+
or the absorption of
a

.
Table 4.1: Feynman rules corresponding to the Lagrangian (4.45)
a
.
Two diagrams contribute in lowest order to the amplitude for
+

. They are given in g. 4.2 where the momentum assignments are de-
ned. We distinguish a scattering and an annihilation diagram, which are
related by an interchange of the momenta (p
2
q
1
). Note that p
1
, p
2
, q
1
and q
2
denote the particle momenta, so that the momentum ow and the
charge ow in the diagrams do not always coincide. As usual we extract an
overall factor of i(2)
4
and a momentum-conserving -function. The invariant
4.3] Electromagnetic scattering of pions 131
Figure 4.2: Tree diagrams contributing to the electromagnetic scattering of
charged pions.
amplitude is given by
M=
e
2
(p
1
+p
2
)

(q
1
q
2
)

(p
1
p
2
)
2
_

(1
2
)
(p
1
p
2
)

(q
2
q
1
)

(p
1
p
2
)
2
_
+
e
2
(p
1
q
1
)

(p
2
q
2
)

(p
1
+q
1
)
2
_

(1
2
)
(p
1
+q
1
)

(p
2
+q
2
)

(p
1
+q
1
)
2
_
,
(4.44)
where we have set the photon momentum equal to k

= (p
1
p
2
)

= (q
2
q
1
)

in the scattering diagram, and to k

= (p
1
+q
1
)

= (p
2
+q
1
)

in the annihila-
tion diagram. A rst important observation is that the gauge-dependent part
of the photon propagator vanishes when the pions are taken on the mass shell,
because (p
1
+p
2
) (p
1
p
2
) = p
2
1
p
2
2
= 0 and (p
1
q
1
) (p
1
+q
1
) = p
2
1
q
2
1
= 0.
This conrms that the physical consequences of the theory have not been af-
fected by introducing the gauge-xing term (4.35) into the Lagrangian.
Introducing Mandelstam variables
s = (p
1
+q
1
)
2
, t = (p
1
p
2
)
2
, u = (p
1
q
2
)
2
,
the amplitude can be written in a simple form
M= e
2
_
u s
t
+
u t
s
_
. (4.45)
This result remains relevant when the four external particles are not all
of the same type. For instance, if we assume that
+
and

are not each


132 Particles with spin one [Ch.4
others antiparticle, but unrelated positively and negatively charged particles of
dierent mass, then the annihilation diagram is not possible and the scattering
diagram gives
M
scatt
= e
2
u s
t
. (4.46)
On the other hand, if the incoming and outgoing particles are of dierent type,
then the scattering diagram is not possible and the annihilation diagram gives
M
ann
= e
2
u t
s
. (4.47)
The above results allow us to discuss a few typical reactions. First consider
the case described by (4.45), which in the analogous case of electron-positron
scattering is called Bhabha scattering. In the centre-of-mass frame t and u are
expressed in terms of s and the scattering angle between p
1
and p
2
t = (s 4m
2
) sin
2 1
2

CM
,
u = (s 4m
2
) cos
2 1
2

CM
. (4.48)
The dierential cross section follows directly from (3.51)
d
d
CM
=

2
4s
3
1
sin
4 1
2

CM
_
s
2
s 4m
2
+s(1 2 sin
2 1
2

CM
+ 2 sin
4 1
2

CM
)
+4m
2
sin
2 1
2

CM
(1 2 sin
2 1
2

CM
)
_
2
, (4.49)
where is the ne-structure constant (
s
= e
2
/4). Obviously the complexity
of this expression is caused by the fact that there are two diagrams involved.
At high energy (4.49) simplies to
d
d
CM
=

2
s
(1 sin
2 1
2

CM
+ sin
4 1
2

CM
)
2
sin
4 1
2

CM
. (4.50)
If there is only one diagram then the result is much simpler. For instance,
if we have a pure annihilation reaction, say of
+

into another charged


particle-antiparticle pair (with mass M), the cross-section reads (using 4.47)
d
d
CM
=

2
4s
_
s 4m
2
s
_
1/2
_
s 4M
2
s
_
3/2
cos
2

CM
, (4.51)
or, for s m
2
, M
2
,
d
d
CM
=

2
4s
cos
2

CM
. (4.52)
4.4] Electromagnetic scattering of pions 133
The scattering amplitude for

scattering o some positively charged particle


with mass M follows from (4.46). Substituting this result into (3.55) we obtain
d
dt
=
2
s
_
u s
t
_
2

1
(s, M
2
, m
2
) . (4.53)
Let us convert (4.53) into the laboratory frame, where the energies of the
incoming and outgoing pions are E and E

, respectively, and the outgoing


pion is deected over an angle . After the scattering has taken place the
target particle is no longer at rest and has an energy E
R
(cf. 3.56). A general
formula for the dierential cross section in the laboratory system is given in
(3.60). If we assume that the pion mass may be neglected it is easy to express
E

in terms of . Using t = 2EE

(1 cos ) = 2M(M E
R
) = 2M(E

E)
we nd
E

=
E
1 + 2E/M sin
2 1
2

,
t =
4E
2
sin
2 1
2

1 + 2E/M sin
2 1
2

,
s = M
2
+ 2ME ,
u = M
2
2ME t . (4.54)
Combining (3.60), (4.53) and (4.54) then leads to
d
d
lab
=

2
s
4E
2
sin
4 1
2

_
1 +E/M sin
2 1
2

1 + 2E/M sin
2 1
2

_
2
. (4.55)
In the limit that M we obtain the celebrated Rutherford cross section.
The extra factor on the right-hand side of (4.55) represents the recoil correc-
tion for the target particle. As we shall see in chapter 6 such recoil corrections
depend on the spin of the target particle.
The observant reader should have noticed that the dierential cross sections
(4.49) and (4.55) diverge in the forward direction (i.e. t or 0). Two
comments will help to clarify this phenomenon. One is that a dierential cross
section can never be measured at = 0 as the scattered particles cannot be
separated from the unscattered beam. Hence measurements of the dierential
cross sections are made in a nite range of > 0. The second comment is
theoretical. Since the range of the electromagnetic interaction is innite, soft
photons will always be radiated in the collisions of charged particles. Therefore
higher-order corrections must be taken into account before it is possible to
dene a total cross section. In section 4.6 we shall discuss a similar problem
in the denition of the decay rate for K
0
S

+
.
134 Particles with spin one [Ch.4
4.4. Pion-Compton scattering
A reaction that involves real photons is pion-Compton scattering, i.e. (k) +

+
(p) (k

) +
+
(p

), where the particle momenta are indicated in paren-


theses. Information on this reaction has been obtained from a study of pion
dissociation in the Coulomb eld of a heavy nucleus (i.e.
+
+ nucleus

+
+ + nucleus). However, for our immediate purpose the experimental as-
pects are not terribly relevant, and we can always take a limit of low-energy
photon scattering, in which the spin of the target plays no role, and thereby
single out nonrelativistic Compton scattering, a process for which there is
adequate experimental data. First some kinematics: in the laboratory frame,
with m the pion mass, ,

the incoming and outgoing photon energies,


respectively, and the scattering angle, we can write the four momenta as
p = m(0, 0, 0, i) ,
k = (0, 0, 1, i) ,
k

(sin , 0, cos , i) ,
p

= (

sin , 0,

cos , i(m+

)). (4.56)
Energy momentum conservation tells us that (p+k)
2
= (p

+k

)
2
or pk = p

leading to the famous relation

=

1 +/m(1 cos )
, (4.57)
which coincides with the rst equation in (4.54). Hence the two-particle kine-
matics does not allow us to let

0 without 0 at the same time. For


later use we list the Mandelstam variables (which may again be compared to
results in (4.54))
s = (p +k)
2
= m
2
+ 2m ,
t = (k k

)
2
= 2

(1 cos ) =
2
2
(1 cos )
1 +/m(1 cos )
. (4.58)
The lowest-order diagrams for pion-Compton scattering follow from the La-
grangian (4.42). There are three diagrams shown in g. 4.3. The third diagram
requires a combinatorial factor of 2 to account for the two possible ways of
attaching the photon lines. The corresponding invariant amplitude contracted
with the photon polarization vectors (k

) and (k) is
M= e
2

(k

)T

(k

, k)

(k) , (4.59)
with
T

=
(2p

+k

(2p +k)

(p +k)
2
+m
2
+
(2p k

(2p

k)

(p k

)
2
+m
2
2

. (4.60)
4.4] Pion-Compton scattering 135
Figure 4.3: Tree diagrams contributing to pion Compton scattering.
Note that T

is transverse for on-shell pions; for instance consider (p


2
=
p
2
= m
2
)
k

= (2p

+k

2p k +k
2
2p k +k
2
+ (2p k

2p

k k
2
2p k

+k
2
2k

. (4.61)
Using
k
2
2k p

= (k p

)
2
p
2
= (k

p)
2
+m
2
= k
2
2k

p,
one nds straightforwardly
k

= 2p

+k

2p

+k

2k

= 0. (4.62)
Likewise one may verify that
k

= 0. (4.63)
Now we have to calculate the square of the amplitude and sum (average) over
nal (initial) photon polarizations. The most straightforward way to do this
is to use (4.41), since we veried that (4.62) and (4.63) are satised. After
some algebra one nds
1
2

pol
|M|
2
=
1
2
e
4

pol
[

(k

)T

(k)][

(k

)T

(k)]

=
1
2
e
4
T

= 2e
4
_
m
4
_
1
p k

1
p k

_
2
2m
2
_
1
p k

1
p k

_
+ 2
_
, (4.64)
where we have used k
2
= k
2
= 0 and p
2
= m
2
. In the laboratory frame this
answer reduces to the simple expression 2e
4
(1 +cos
2
) because
1

1
=
136 Particles with spin one [Ch.4
m
1
(1 cos ). Although at rst sight the formulae in (4.60) and (4.64)
seem to diverge as or

0, the answer thus remains nite (innities


that usually arise in this limit are called infrared divergences). The absence
of the divergence can be understood from the fact that the virtual pion in the
rst two diagrams of g. 4.3 cannot approach its mass shell without violating
angular momentum conservation.
Now that we have the result above, we should comment that there is a
faster way to do the calculation. Begin in the laboratory frame and choose
the following set of independent real polarization vectors

(k, 1) = (1, 0, 0, 0) ,

(k, 2) = (0, 1, 0, 0) ,

(k

, 1) = (cos , 0, sin , 0) ,

(k

, 2) = (0, 1, 0, 0) , (4.65)
which satisfy p = p

= k = k

= 0. Therefore the amplitude (4.59)


reduces to M = 2e
2
(k)

(k

). If we then perform the explicit sum over


the (real) polarization vectors (4.65) the result is
1
2

pol
|M|
2
= 2e
4

=1,2

(k, )

(k

(k, )

(k

)
= 2e
4
(1 + cos
2
) . (4.66)
The dierential cross section in the laboratory frame follows from (3.55),
d
dt
=
1
16
1
(s m
2
)
2
_
1
2

pol
|M|
2
_
, (4.67)
and (3.60), which reduces to
d
d
lab
=

2

d
dt
(4.68)
Comparing the above formulae one nds
d
d
lab
=

2
2m
2
1 + cos
2

(1 +/m(1 cos ))
2
, (4.69)
where is the ne-structure constant. In the limit 0 we may compare
(4.69) to the result for scattering of classical electromagnetic radiation
d
d
lab
=

2
2m
2
(1 + cos
2
) . (4.70)
The corresponding total cross section,
=
8
2
3m
2
, (4.71)
4.5] Decay rate for
0
137
is named after Thomson, and provides a means to measure the ratio of the
charge to the mass of a particle. In units where = c = 1, 1 GeV
2
= 389 b
so the cross section is 8.8b. For photon-electron scattering the Thomson cross
section is much larger, since one replaces m by the (much smaller) electron
mass. The cross section for radiation on an atomic electron is equal to 0.66 b.
4.5. Decay rate for
0

The lifetime of the neutral pion is

0 = 0.83 10
16
s. The particle decays
primarily into two photons, and has small branching ratios into the modes
e
+
e

and e
+
e

e
+
e

. In this section we calculate the decay rate for


0

starting from the interaction Lagrangian


L
int
= iC

. (4.72)
In (4.72) C is a coupling constant with dimension [mass]
1
, is the ne-
structure constant ( = e
2
/4), F

is the eld-strength tensor dened in


(4.24), and is the eld associated with the neutral pion (

is a completely
antisymmetric tensor normalized such that
1234
= 1). We should point out
that the Lagrangian (4.72) is only introduced here as a useful mnemonic for
writing down the amplitude, without pretending that it describes some fun-
damental theory. In this context one often uses the term phenomenological
or eective Lagrangian. If the photons and the pion are on their respective
mass shells C is just a constant, because all Lorentz-invariant combinations
of the external momenta in a three-point vertex can be expressed in terms
of the masses of the incoming and outgoing particles. Therefore, taking the
lowest-order diagram corresponding to (4.72) will describe the full result for

0
in terms of an unknown parameter C. We shall argue later that it
is possible to construct models in which one can calculate C.
The eective Lagrangian (4.72) can be motivated as follows. Since the La-
grangian must be gauge invariant the photon eld must be expressed in terms
of the gauge-invariant tensor F

(we have two tensors F

because we need
a three-point vertex with two photon elds and one pion eld). The ne-
structure constant has been introduced since we expect that each photon cou-
ples with a strength proportional to the elementary charge e. The pion has
negative intrinsic parity which means that the eld changes under parity
according to
(x, t)
P

p
(x, t) = (x, t). (4.73)
Under parity reversal F

transforms as
F

(x, t)
P
F
p

(x, t) =
_
F

(x, t) if or = 0 ,
+F

(x, t) if or = 0 .
(4.74)
138 Particles with spin one [Ch.4
Figure 4.4: Diagram describing the decay
0
.
Therefore we have introduced the -tensor because

has negative
intrinsic parity. This can be seen more easily by expressing F

in terms of
the magnetic and electric elds;

is proportional to E B, and
as is well-known E transforms as a vector and B as an axial vector under
x x. Because the -symbol forces precisely one of the indices , , or
to be equal to 4, and F
4
is imaginary in our convention,

is
purely imaginary. Therefore we have introduced a factor i in (4.72) so that C
is real.
In order to calculate the decay rate for
0
(Q) (p) + (q) we must rst
nd the invariant amplitude which corresponds to the diagram in g. 4.4.
After contraction with transverse polarization vectors
1
(p) and
2
(q) for the
outgoing photons we have
M(
0
) = 2iC

(2ip


1
(p))(2iq


2
(q)) , (4.75)
where the overall factor of 2 arises because the external photons can be hooked
to the vertex in two dierent ways. The terms 2ip


1
(p) and 2iq


2
(q)
arise from each of the eld strength tensors, where we have dened the mo-
menta of the photons as outgoing; the factor of two in both expressions orig-
inates from the fact that F

contains two terms (cf. 4.24) which add as a


result of the antisymmetry of the -symbol.
To calculate the decay rate one must rst square the amplitude (4.40):
|M(
0
)|
2
= (8C)
2
|


1
(p)
2
(q)|
2
. (4.76)
We rst note that


1
(p)
2
(q) is purely imaginary because the -
symbol forces precisely one of the indices to be equal to 4. Therefore we may
write
|


1
(p)
2
(q)|
2
=
=


1
(p)
1
(p)
2
(q)
2
(q) . (4.77)
4.5] Decay rate for
0
139
Now we recall the following property of the -symbol:

perm
(1)
P

, (4.78)
where we sum over all 24 permutations of the indices , , , , and ()
P
is
minus (plus) if the permutation is odd (even). Equation (4.78) is based on the
observation that the set of indices [] as well as [

] must cover all


index values in some permutation, so that each of the [] should coincide
with precisely one of the [

] indices (see appendix B).


Only 5 permutations in (4.78) contribute, because of the relations k
2
=
k = 0 for photons. The result can thus be cast in the form
|M(
0
)|
2
= (8C)
2
{
1
4
m
4

[(
1
(p)
1
(p))(
2
(q)
2
(q)) |
1
(p)
2
(q)|
2
]

1
2
m
2

[(Q
1
(p))(Q
2
(q))(
1
(p)
2
(q))
+(Q
1
(p))(Q
2
(q))(
1
(p)
2
(q))]
+|Q
1
(p)|
2
|Q
2
(q)|
2
} , (4.79)
where Q

is the
0
momentum and we have used
p
2
(q) = Q
2
(q) ,
q
1
(p) = Q
1
(p) ,
p q =
1
2
Q
2

1
2
p
2

1
2
q
2
=
1
2
m
2

. (4.80)
If we consider (4.79) in the rest frame, where Q
1
(p) = Q
2
(q) = 0,
we nd that |M|
2
is proportional to ((
1

1
)(
2

2
) |
1

2
|
2
). This may
be compared to the analogous result for the two-photon decay of a scalar
meson, since an even-parity interaction F

leads to |M|
2
proportional to
|
1

2
|
2
. Conservation of angular momentum implies that two spin-1 particles
produced by the decay of a spinless particle should have equal helicities in
the rest frame of this particle (see g. 4.5). Equal helicity means that
1
=
2
(as the spin-1 particles move in opposite directions), opposite helicity implies

1
=
2
. Since helicity vectors satisfy = 1 and = 0 (cf. 4.11a) one
readily veries that both ((
1

1
)(
2

2
) |
1

2
|
2
) and |
1

2
|
2
vanish
for
1
=
2
, and are equal to unity for
1
=
2
. Consequently the restrictions
imposed by angular momentum conservation are satised in both cases.
However, one may also analyze the decay in terms of linear photon polar-
izations ( = , as shown by 4.8a). In that case one readily concludes that
(
1

1
)(
2

2
) |
1

2
|
2
) vanishes if the plane polarizations of the two
photons are parallel, whereas |
1

2
|
2
. vanishes if the plane polarizations are
perpendicular. In principle this feature can be used to determine the intrin-
sic parity of the decaying spinless particle. For example, if the photon were
an unstable particle, its two-body decay would dene a plane whose orienta-
tion would be correlated to the polarization of the photon. If both photons
140 Particles with spin one [Ch.4
Figure 4.5: Photon helicities in the rest frame of the decaying
0
. Opposite
helicity photons are forbidden by angular momentum conservation.
decayed one could thus measure the number of events as a function of the
angle between the two decay planes. For pseudoscalar mesons one expects
a sin
2
=
1
2
(1 cos 2) distribution whereas scalar mesons give rise to a
cos
2
=
1
2
(1 + cos 2) distribution. The photon, however, is a stable particle
so that this direct verication is not possible. Nevertheless, the polarization
may still be measured if the photons produce an e
+
e

pair in the electric eld


of a nucleus, a process called external conversion. For example, if the pho-
tons go through a thin slab of material some of them may be converted into
e
+
e

pairs. Unfortunately the opening angle between the e


+
and the e

is too
small in order to reliably measure the decay planes. Furthermore, the photon
polarization and the decay plane are less correlated in the conversion so that
the -distribution is changed into 1
2
cos 2 where
2
is a positive number
which is much smaller than unity. The eect therefore is less pronounced, so
this experiment has never been performed. If, however, we consider internal
conversion, where the
0
decays into virtual photons which then immediately
decay into e
+
e

pairs, the distribution can be measured. The experimen-


tal result conrms the pseudoscalar nature of the pion and agrees with the
theoretical prediction
2
= 0.18.
In order to determine the total decay rate one must sum (4.82) over the
polarizations for each of the photons. At this point it is not allowed to use
the covariant polarization sum (4.41), because we have already dropped cer-
tain terms in (4.79) by virtue of the transversality of the photon polarization
vectors. Hence, either we sum (4.79) directly over the physical photon po-
larizations, preferably in the
0
rest frame, or we base ourselves on (4.77)
and use the covariant result (4.41); the latter amounts to replacing the sum
of
1

1
, and
2

2
, by

and

respectively. One may then use the


identity (see appendix B)

= 2(

) ,
4.5] Decay rate for
0
141
which follows from (4.78). The nal answer is

pol
|M|
2
=
1
2
(8C)
2
m
4

. (4.81)
The decay rate in the rest frame follows now directly from substituting
(4.81) into (3.82), and we nd
(
0
) =
N
16

1/2
(m
2

, 0, 0)
m
3

1
2
(8C)
2
m
4

.
Because we have two identical particles in the nal state the combinatorial
factor N is equal to 1/2. Using the explicit form of the function the decay
rate is thus equal to
(
0
) =

2

m
3

C
2
. (4.82)
The only unknown in this equation is the constant C, which can be deter-
mined from the experimental lifetime, yielding C = 4.3 10
4
MeV
1
. Its
theoretical value depends on the mechanism responsible for the interaction de-
scribed by (4.72). If we suppose that this interaction arises from the pion rst
dissociating into a virtual charged fermion-antifermion pair then the decay is
described by the Feynman diagrams given in g. 4.6. Here the (anti)fermion
emits a photon and then annihilates with its partner into a second photon. A
calculation gives
C =
g
8M
, m

M , (4.83)
where g is the fermion-pion coupling constant associated with the correspond-
ing vertices in the diagrams of g. 4.6, and M is the fermion mass (in section
7.1 we will present the calculation leading to (4.83) in considerable detail).
When more than one fermion is involved then
C =
1
8

i
g
i
M
i
e
2
i
, m

M, . (4.84)
where e
i
is the fermion charge in units of the elementary charge. We may now
invoke the Goldberger-Treiman relation, which is based on chiral symmetry
arguments and implies that the ratio g
i
/M
i
does not depend on the fermion
mass and is equal to 2g
i
A
f
1

; f

is the pion-decay constant that governs


the decays

+
e
and

, while g
i
A
is the axial-vector
coupling constant which is (approximately) equal to the value of the third
isospin component (for nucleons g
A
is measured in neutron decay: n
p + +e). Experimentally f

130 MeV, and for nucleons we have |g| 10,


142 Particles with spin one [Ch.4
Figure 4.6: One-loop diagrams contributing to the decay
0
.
|g
A
| 1/2 and M 939 MeV/c
2
, so that the Goldberger-Treiman relation is
true within 10%.
Assuming that the above arguments hold for all fermions contributing to
the diagrams of g. 4.6 we can write the decay width in the following way
(
0
) =

2
16
3
m
3

f
2

S , (4.85)
where
S =
_

i
g
i
A
e
2
i
_
2
. (4.86)
If one assumes that only the proton contributes to (4.86) we have
S =
1
4
, (4.87)
which is consistent with the experimental value S = 0.27. If, on the other
hand, one uses the quark model then there are two kinds of quarks that
will contribute, namely u- and d-quarks with charges
2
3
and -
1
3
, respectively
(although we do not really know what to use as a mass value for the constituent
quarks we have assumed here that they are heavier than pions). These quarks
form an isospin doublet, so their values for g
i
A
are equal to
1
2
.
S =
_
1
2
(
2
3
)
2

1
2
(
1
3
)
2
_
2
=
1
36
. (4.88)
This result is nine times smaller than (4.87), which suggests that one should
have three times as many quarks contributing to the diagrams of g. 4.6,
in agreement with current theoretical ideas, according to which each quark
exists in three dierent varieties with equal mass and charge. These varieties
are called quark colours, and they play a crucial role in the standard model
of strong interactions called quantum chromodynamics (QCD ).
4.6] Bremsstrahlung in the decay K
0
S

+

143
4.6. Bremsstrahlung in the decay K
0
S

+

The energy of a photon produced in the radiative decay K


0
S

+

is
restricted by energy-momentum conservation to be less than
1
2
M
1
(M
2

4m
2
) 160 MeV (here M and m denote the mass of the K and mesons,
respectively). For these small photon energies it is possible to relate the
K
0
S

+

amplitude to the amplitude of the nonradiative K


0
S

+

decay mode on the basis of the Born approximation diagrams shown in g.


4.7. In these diagrams the photon line is hooked onto one of the external pion
lines. The K vertex is represented by a function F(P
2
, q
2
+
, q
2

), where P, q
+
and q

are the momenta associated with the K- and -meson lines (here we
have made use of the fact that all Lorentz invariants formed from three mo-
menta satisfying energy momentum conservation, can be expressed in terms
of the three invariants P
2
, q
2
+
and q
2

). In the limit that the photon mo-


mentum k tends to zero the momentum of the internal pion line approaches
one of the external pion momenta which are on the mass shell. Consequently
the propagator of the internal pion line diverges. This is the characteristic
feature of Born approximation diagrams, which therefore give the dominant
contribution for small photon energies.
Figure 4.7: Born approximation diagrams corresponding to (4.92).
Expanding the vertex function F(P
2
, q
2
+
, q
2

) about q
2
+
= q
2

= m
2
leads
to the following result for the amplitude M
B

from the graphs of g. 4.7 (for


the moment we keep the photon o shell):
M
B

=e
_
(2Q
+
+k)

k
2
+ 2k Q
+
F(M
2
, m
2
, m
2
)

(2Q

+k)

k
2
+ 2k Q

F(M
2
, m
2
, m
2
)
_
, (4.89)
where Q
+
and Q

are the momenta of the outgoing


+
and

, respectively.
For the -vertex we made use of the Feynman rules listed in table 4.1. It
144 Particles with spin one [Ch.4
is important to observe that the amplitude (4.89) is conserved, because
k

M
B

= e
_
(2k Q
+
+k
2
)
k
2
+ 2k Q
+

(2k Q

+k
2
)
k
2
+ 2k Q

_
F(M
2
, m
2
, m
2
) = 0 .
(4.90)
One may deduce that the remaining contributions to the full amplitude must
be linear in the photon momentum. To see this one notes that a conserved
amplitude depending on the three independent momenta in this process, Q
+
,
Q

and k, can be parametrized as a sum of two terms, namely


M

= A(k Q

Q
+
k Q
+
Q

) +B

Q
+
Q

. (4.91)
Since (4.89) is conserved and represents all contributions to the full amplitude
that are singular in the k 0 limit the remaining terms in the amplitude must
take the form of (4.91) with coecients A and B that are nite in this limit.
Because (4.91) is already linear in k the full amplitude must satisfy
M

= e
_
(2Q
+
+k)

k
2
+ 2k Q
+

(2Q

+k)

k
2
+ 2k Q

_
F(M
2
, m
2
, m
2
) +O(k) .
(4.92)
Setting also the photon on the mass shell we may now conclude that the
invariant amplitudes for the radiative and nonradiative decay are related by

(K
0
S

+

) = e
_
Q

k Q

Q
+
k Q
+
_
M(K
0
S

+

) +O(k) ,
(4.93)
where M(K
0
S

+

) = F(M
2
, m
2
, m
2
) does not depend on the exter-
nal momenta. Squaring (4.93) and summing over photon polarizations leads
to

pol
|

(K
0
S

+

)|
2
= e
2
_
Q

k Q

Q
+
k Q
+
_
2
|M(K
0
S

+

)|
2
+O(k
0
)
= e
2
_
2Q
+
Q

(k Q
+
)(k Q

)

m
2
(k Q
+
)
2

m
2
(k Q

)
2
_
|M

(K
0
S

+

)|
2
+O(k
0
) , (4.94)
where we have used (4.41). For the moment we ignore the fact that (4.94)
diverges for small photon energies. This so-called infrared divergence is caused
4.6] Bremsstrahlung in the decay K
0
S

+

145
by the fact that the photon is massless. We return to this aspect at the end
of this section.
The rate for the nonradiative decay follows from (3.82). The result is
(K
0
S

+

) =

0
16M
|M(K
0
S

+

)|
2
, (4.95)
where
0
is the pion velocity in the rest frame of the K meson,

0
=
_
1
4m
2
M
2
, (4.96)
which equals the maximal possible pion velocity for the radiative decay.
For the radiative decay we use the expression (3.105) for the phase space
integral, where m
1
= 0 and m
2
= m
3
= m. In the parametrization (3.101)
=
1
denotes the photon energy and
2
and
3
the pion energies in the rest
frame of K
0
S
. Using momentum conservation it is easy to show that
2k Q
+
= M( + 2x) , 2k Q

= M( 2x) ,
2Q
+
Q

= M
2
+ 2M + 2m
2
. (4.97)
Since the amplitude does not depend on the orientation of the decay plane,
we may integrate (3.105) over the corresponding three angles. This leads to a
factor 8
2
. According to the denition (3.79) we must divide (3.105) by 2M
in order to obtain the decay rate. Substituting the combined results of (4.94)
and (4.97) then gives (dropping the unknown O(k
0
) terms)
(K
0
S

+

) =
e
2
16
3
M
3
|M(K
0
S

+

)|
2

_

m
0
d
_
x
+
x

dx
_
M
2
2M 2m
2
( + 2x)( 2x)

m
2
( + 2x)
2

m
2
( 2x)
2
_
, (4.98)
where according to (3.104) and (3.102)
x

=
1
2

_

m

m
+ 2m
2
/M
,

m
=
1
2
M
2
0
. (4.99)
It is convenient to introduce a parameter
=
_
M
2
2M 4m
2
M
2
2M
, (4.100)
which corresponds to the pion velocity measured in the rest frame of the two
pions (see problem 4.8) (in the limit that 0, this frame coincides with
146 Particles with spin one [Ch.4
the rest frame of the K meson and
0
). The parameters x

now take the


form
x

=
1
2
. (4.101)
The necessary x-integrals for (4.98) are simple and are given by
_
x
+
x

dx
1
( + 2x)( 2x)
=
1

ln
+ 2x
2x

x
+
x

=
1
4
ln
1 +
1
,
_
x
+
x

dx
1
( 2x)
2
=
1
2
1
2x

x
+
x

=

(1
2
)
. (4.102)
Hence
(K
0
S

+

)
(K
0
S

+

)
=

_
1
2
M
2
0
0
d

_
1 +
2

ln
1 +
1
2
_

0
_
1
2
M
_
.
(4.103)
where is the ne-structure constant.
As anticipated previously, (4.103) diverges if the photon energy tends
to zero. Nevertheless, this result can be compared to experiment, because in
an experimental situation there is always a minimal photon energy
0
below
which it is impossible to detect the photons emitted. Therefore the relevant
radiative decay rate is
(K
0
S

+

; >
0
)
(K
0
S

+

)
=

_
1
2
M
2
0
0
d

_
1 +
2

ln
1 +
1
2
_

0
_
1
2
M
_
. (4.104)
Figure 4.8 shows the measured photon spectrum in the range 25 MeV< <
175 MeV. For >
0
= 50 MeV the data show an
1
behaviour consistent
with the
1
behaviour of the integrand in (4.104). Note that the shape of the
histogram for <
0
reects the photon trigger eciency. For
0
= 50 MeV
the integral in (4.104) gives the numerical value of 2.5410
3
, in excellent
agreement with the result of the same experiment which quotes
(K
0
S

+

; > 50MeV)
(K
0
S

+

)
= (2.68 0.15) 10
3
. (4.105)
But how do we interpret the part of the integral (4.103) where <
0
? Since
the photons are not observed in this energy range the corresponding events
would be counted as genuine K
0
S

+

decays. Consequently the relevant


4.6] Bremsstrahlung in the decay K
0
S

+

147
denition of the decay rate for K
0
S

+

must be changed in order to


account for unobservable soft photons. Hence to order we have
(K
0
S

+

) =

16M
|F(M
2
, m
2
, m
2
)|
2

_
1 +

_
1 +
2
0

0
ln
1 +
0
1
0
2
_
_

0
0
d

_
, (4.106)
where we have reintroduced the vertex function F and approximated the
integral (4.103) for small values of . Of course, the minimal observable photon
energy
0
depends on the experiment under consideration. The extra terms
in (4.109) are called radiative corrections.
Figure 4.8: Distribution of events with photon energy in the K
0
rest frame.
The events above
0
= 50 MeV reect the
1
behaviour as given in (4.104),
whereas the shape of the histogram below
0
reects the eciency of the
photon trigger. [H. Taureg, G. Zech, F. Dydak, F. L. Navarria, P. Steen, J.
Steinberger, H. Wahl, E. G. H. Williams, C. Geweniger and K. Kleinknecht,
Phys. Lett. 65B (1976) 92.]
The divergent part of the integral (4.103) thus appears in the nonradiative
decay rate as a correction of order . However, it is important to observe that
there are other corrections to this process in the same order of . These cor-
rections originate from virtual photons, and also lead to infrared divergences.
148 Particles with spin one [Ch.4
The calculation of the nonradiative decay rate now involves additional graphs
as is schematically shown in g. 4.9. At this point it is convenient to introduce
a small photon mass to give a more precise denition of the infrared diver-
gences. The integral in (4.106) from the bremsstrahlung graphs then gives a
divergent term proportional to ln
0
/. The corrections from the virtual pho-
ton diagrams lead to terms proportional to ln /M. Hence both contributions
diverge logarithmically when 0, but, surprisingly enough, these two di-
vergences cancel in the decay rate, so that the answer contains only a term
ln
0
/M, where
0
is a nite number that may change from experiment to ex-
periment. The procedure by which these infrared divergences cancel was rst
described by Bloch and Nordsieck. Later it was shown that this cancellation
holds to all orders in (higher-order terms usually generate an exponential
series). In section 8.6 we will explicitly establish the cancellation of infrared
divergences to lowest nontrivial order in a dierent process.
Figure 4.9: Order diagrams contributing to the decay rate for K
0
S

+

.
Thus we see that the emission of photons from charged particles gives rise to
either radiative processes, in which the photon is properly identied, or radia-
tive corrections to nonradiative processes in which the photon identication is
not possible. We should add that the virtual photon corrections may lead to
infrared divergences when the photon momentum is small and to ultraviolet
divergences when the momentum is large (we will learn how to interpret the
latter divergences in due course). After both divergences have been eliminated
order by order in perturbation theory the remaining nite terms should agree
with experimental results. The evaluation of these nite corrections is rather
complicated. For that reason we have relegated a technical discussion of some
of the necessary integrals to appendix D.
Problems 149
Problems
4.1. Consider the decay of a vector particle V with momentum P into two scalars
S
1
, and S
2
with momenta p
1
; p
2
(p
2
1
= m
2
1
, p
2
2
= m
2
2
, P
2
= M
2
). Show that the
decay amplitude for the process generally takes the form
M

(p
1
, p
2
) = i(p
1
+p
2
)

f(P
2
, p
2
1
, p
2
2
) + i(p
1
p
2
)

g(P
2
, p
2
1
, p
2
2
) (1)
by exploiting Lorentz invariance. Argue that f does not contribute to the decay
rate, so that the decay is precisely described by an eective Lagrangian
L = gV

(
1

2
) (2)
Consider the decay VS
1
S2 with S
1
emitted in the x z plane at an angle with
respect to the z-axis. Show that the dierential decay rates for polarization states
of the V with spin S
z
= 1 and 0 along the z-direction are proportional to sin
2

and cos
2
respectively. Integrate these expressions over to nd the decay rate,
=
g
2
48

3/2
(M
2
, m
2
1
, m
2
2
)
M
5
(3)
for each spin state. Note that the total decay rate for an unpolarized V is also given
by (3).
As an example consider the decay of rho mesons into pions. Just as the pion elds
can be written as real isospin vectors
a
(cf. 2.54) the rho elds can be treated
in the same way, i.e.
0
=
3
,

=
1
i
2
. Argue that the isospin invariant
generalization of (2) is
L = g
ijk
V
i

(
j

k
) . (4)
Write this Lagrangian in terms of the

,
0
,

, and
0
elds. Calculate the decay
rates for
+

+

0
,
0

+

,
0

0

0
and show that they are in the ratio
1:1:0.
4.2. Consider a massive vector boson coupled to a conserved current so that the
amplitude equals M =

(k)M

(k) where k

= 0. Show that, although the


longitudinal polarization vector is singular in the limit m 0 (cf. 4.11b), the am-
plitude vanishes for longitudinal vector bosons in this limit. This property makes it
hard to distinguish between an almost massless and a massless photon. [For bounds
on the photon mass see A. Goldhaber and M. M. Nieto, Rev. Mod. Phys. 43 (1971)
227.] The smooth decoupling of the longitudinal polarization in the massless limit
turns out to be a special property of neutral vector bosons. For instance it does not
hold for gravitons (the spin-2 quanta of gravitation) where there is a discontinuity
at m = 0 [H. van Dam and M. Veltman, Nucl. Phys. B22 (1970) 397; see also M.
Veltman, in: Methods in Field Theory, Les Houches 1975, eds. R. Balian and J.
Zinn-Justin (North-Holland/World Scientic, 1976) p. 266.]
4.3. Using the Lagrangian of problem 4.1, calculate the amplitude for the reaction

+
+


+
+

in tree approximation.
150 Particles with spin one [Ch.4
When s = M
2
one should take into account that the
0
is not a stable particle.
Therefore the propagator factor s M
2
becomes s M
2
+ iM where is the
total decay rate in the rest frame. Now evaluate the dierential cross section in the
centre-of-mass frame, which in the limit M 0 and 0 can be compared to
(4.51). Integrate this result to nd the total cross section as a function of s, and
compare your answer with (3.92). Observe the factor of 3 and verify that this agrees
with the discussion in the text after (3.92).
4.4. To describe the interactions of photons with charged massive vector particles,
consider (4.1) for two real elds W
1

and W
2

, change to a complex basis described


by

2W

= W
1

iW
2

2

W

= W
1

+iW
2

and apply the minimal electromagnetic


substitution as in section 4.3. Note that

W

(1 2
4
)W

transforms as a vector
under Lorentz transformations (cf A.24). Show that the resulting Lagrangian is
L
1
=
1
2
(

)(

) M
2

ieA

[

W

) (

)W

]
e
2
(A

) . (1)
Verify the invariance of (1) under electromagnetic gauge transformations (cf. 4.43):
A

(x) A

(x) +

(x) , W

(x) e
ie(x)
W

(x) . (2)
It is possible to add to (1) a term that is separately gauge invariant, namely
L = ie(

)

W

, (3)
which in the particle rest frame describes an additional coupling with the magnetic
eld. Therefore (3) is called the anomalous magnetic moment term.
Using k, p and r as incoming particle momenta for the elds A

, W

and

W

,
respectively, show that the three-point vertex corresponding to (1) and (3) is
V

(k, p, r) = e[(k p)

+ (p r)

+ (r k)

] , (4)
where we have suppressed the usual factor i(2)
4

4
(k +p +r).
Construct the invariant amplitude for a virtual photon interacting with an incoming
and outgoing W
+
,
M

(p
2
, p
1
) = e

(p
2
)[(p
1
+p
2
)

+(1+)((p
2
p
1
)

(p
2
p
1
)

(p
1
) ,
(5)
where

(p
1
) and p
1
(

(p
2
) and p
2
) are the polarization vector and momentum
of the incoming (outgoing) W
+
. The two terms in (5) characterize the coupling of
the photon to the charge and the magnetic moment of the vector particle. Observe
that the magnetic moment coupling is proportional to 1 +. Prove that the photon
couples to a conserved current, i.e. (p
2
p
1
)

(p
2
, p
1
) = 0. Compare (5) to the
corresponding expression for the coupling of a virtual photon to an incoming and
Problems 151
outgoing
+
. Show that the square of (5), averaged over incoming W polarizations
and summed over outgoing W polarizations, equals
W


1
3

spins
M

(p
2
, p
1
)

M

(p
2
, p
1
)
=
1
3
e
2
_
2(1 +)
2
_
1 +
Q
2
4M
2
_
(Q
2

)
+
_
3 + (1 +
2
)
Q
2
2M
2
+
2
Q
4
4M
4
_
P

_
, (6)
where Q = p
2
p
1
, P = p
2
+ p
1
(so that P Q = 0). Note that (6) satises
Q

= W

= 0.
Repeat this analysis for the pion vertex using Q = q
1
q
2
, R = q
1
+ q
2
(so that
Q R = 0) to nd the corresponding tensor
L

= e
2
R

, (7)
which satises Q

= L

= 0.
Write the square of the amplitude for the electromagnetic scattering process

+
(q
1
) +W
+
(p
1
)
+
(q
2
) +W
+
(p
2
)
in terms of (6) and (7). Using s = (p
1
+q
1
)
2
, t = (p
1
p
2
)
2
and u = (p
1
q
2
)
2
,
show that the invariant dierential cross section is
d
dt
=

2
(s, M
2
, m
2
)
1
3t
2

__
3 (1 +
2
)
t
2M
2
+
2
t
2
4M
4
_
(u s)
2
2(1 +)
2
t(t 4m
2
)
_
1
t
4M
2
__
. (8)
Express this result in terms of the laboratory scattering angle using (4.54) and
compare the answer with (4.55).
Now consider the invariant amplitude for Compton scattering (i.e. + W
+
+
W
+
). Include the contribution from the four-point vertex corresponding to (1)
V

(k, q, p, r) = e
2
[2

] , (9)
where k, q, p and r are the incoming momenta for the elds A

, A

, W

and

W

.
Keep the photons o-shell and show that the Compton amplitude is transverse (cf.
section 4.4).
4.5. To illustrate that the gauge-xing term (4.35) only introduces an extra degree
of freedom into the theory which does not interfere with the interactions (cf., the
discussion leading to 4.35), consider the Maxwell theory coupled to a conserved
source, i.e.
L =
1
4
F

+J

(1)
152 Particles with spin one [Ch.4
with

= 0. After the addition of (4.35) the eld equation for A

is

+J

+
2

( A) = 0 . (2)
Contracting this with another derivative leads to the free eld equation for A i.e.,
( A) = 0. (3)
To be more specic one can take the theory given by (4.42) for charged pions coupled
to photons. In this model the current corresponds to
J

= ie

2e
2
A

||
2
. (4)
Show that this current is conserved as a result of the pion equations of motion, so
that (3) will be satised in the presence of the gauge-xing term (4.35).
4.6. From (4.64) derive the following formula for pion Compton scattering, where

CM
is the photon scattering angle in the centre-of-mass system
d
d
CM
=

2
s
(s
2
+m
4
)(1 + cos
2

CM
) + 2(s
2
m
4
) cos
CM
[s +m
2
+ (s m
2
) cos
CM
]
2
.
Integrate this equation to nd the total cross section,
=
4
2
s
_
s +m
2
s m
2
__
1
2m
2
s
(s +m
2
)(s m
2
)
ln
s
m
2
_
.
4.7. Consider the interaction Lagrangian involving the photon eld A, the neutral
pion eld and a massive charged spinless eld associated with a nucleus N of
charge Ze:
L = iC

iZeA

)] . (1)
Write down the order Ze
3
amplitude for the photoproduction process (k
1
) +
N(P
1
)
0
(k
2
) + N(P
2
), often called the Primako process. Use invariants
s = (k
1
+P
1
)
2
, t = (k
1
k
2
)
2
and u = (k
1
P
2
)
2
and calculate the dierential
cross section in the centre-of-mass frame,
d
d
CM
=
Z
2
(
0
)
2sm
3

t
2
_
(s, m
2

, M
2
)
(s, 0, M
2
)
[(t 4M
2
)(m
2

t)
2
t(us)
2
] . (2)
Now take the limit that M (static nucleus) and show that
d
d
lab
=
8Z
2
(
0
)
m
3

3
_

2
t
_
2
sin
2
. (3)
where =
_
1 m
2

/
2
, is the photon energy, and the laboratory angle of the
outgoing
0
. For a measurement of (
0
) using this process see A. Browman,
References 153
J. DeWire, B. Gittelman, K. M. Hanson, D. Larson, E. Loh and R. Lewis, Phys.
Rev. Lett. 33 (1974) 1400.
4.8. Consider the kinematics of the reaction K
0
S
(P)
+
(Q
+
) +

(Q

) + (k).
Dene the invariant mass

s of the
+

system by s = (Q
+
+ Q

)
2
and show
that s = M
2
2M. Since s is a Lorentz-invariant quantity one can go to the
centre-of-mass frame of the
+

system and calculate the pion velocities in terms


of s. Use this method to show that the velocity is equal to given in (4.100).
4.9. The K
0
L
-meson has a small CP-violating decay rate into
+

,so the the


corresponding bremsstrahlung amplitude for the decay K
0
L

+

is suppressed.
However, there is also a direct (magnetic dipole) transition which contributes to
the decay rate. The corresponding interaction Lagrangian expressed in terms of the
elds A

, ,

and
K
for ,
+
,

and K
0
L
, respectively, is
L
int
=
eg
M
3

K
,
where the K mass is denoted by M and g is a dimensionless constant. Show that
the photon spectrum resulting from this interaction is given by
d(K
0
L

+

)
d
=
g
2
24
2
M
3

3
_
1
2
M
_
,
where is given by (4.100) and is the photon energy in the K
0
L
rest frame. Note
that this result remains nite when 0. For a measurement of this spectrum see
A. S. Carroll, I.-H. Chiang, T. F. Kycia, K. K. Li, L. Littenberg, M. Marx, P. O.
Mazur, J. P. de Brion and W. C. Carithers, Phys. Rev. Lett. 44 (1980) 529.
References
The eld equation for massive vector mesons was introduced by A. Proca, J.
Phys. Radium 9 (1938) 61. The original references on Rutherford, Compton and
Thomson scattering E. Rutherford, Philos. Mag. 21 (1911) 669.
A. H. Compton, Phys. Rev. 21 (1923) 207, 483.
J. J. Thomson, Conduction of Electricity through Gases, 3rd Ed., Vol. 11 (Cambridge
University Press, 1933) p. 256. For the latest data on Compton scattering on pions
M. Zielinski et al., Phys. Rev. D29 (1984) 2633. For a review of Compton scattering
on nucleons H. Rollnick and P. Stichel. Compton Scattering,
Elementary Particle Physics Series, Vol. 7 Springer, New York, 1976). References
for our discussion of the decay
0
J. Steinberger, Phys. Rev. 76 (1949)
1180.
C. N. Yang, Phys. Rev. 77 (1949) 242.
E. Karlson, Ark. Fys. 13 (1958) 1.
154 Particles with spin one [Ch.4
N. P. Samios, R. Plano, A. Prodell, M. Schwartz and J. Steinberger, Phys. Rev. 126
(1962) 1844.
T. Mijazaki and E. Takasugi, Phys. Rev. D8 (1973) 2051. For the Goldherger-
Treiman relation M. L. Goldberger and S. B. Treiman, Phys. Rev. 110 (1958)
1178; 111 (1958) 354.
S. L. Adler and R. F. Dashen, Current Algebras (Benjamin, New York, 1968).
Theoretical papers on the treatment of infrared divergences F. Bloch and A.
Nordsieck, Phys. Rev. 52 (1937) 54, 59.
D. R. Yennie, S. C. Frautschi and H. Suura, Ann. Phys. 13 (1961) 379.
G. Grammer and D. R. Yennie, Phys. Rev. D8 (1973) 4332.

Das könnte Ihnen auch gefallen