e+e-→ hadrons at higher orders
THE DIAGRAMS AND RELEVANT PIECES OF THE AMPLITUDES
Let us return to the total hadronic cross section in electron-positron annihilations.
As has been discussed before, at leading order in perturbation theory the total cross section for
the production of a massless quark-antiquark pair q reads
Here Nc=3 is the number of colours, s is the Mandelstam variable, i.e. the c.m. energy squared,
eq is the fractional charge of the quark in question, and α is the fine structure
constant. Of course, the total hadronic cross section then is given by the sum of all quarks that
are kinematically accessible, i.e with mass smaller than half of the c.m. energy of the colliding
electron-positron pair. In this reasoning, and in the following, quark mass effects have been and will
be neglected in both the matrix element and the phase space integration.
When considering perturbative corrections of the first order in the strong coupling constant, two
types of diagrams contribute:
-
Corrections due to the emission of a virtual gluon, namely vertex correction
and quark self energy
- Also, real corrections play a role, i.e. the emission of an extra gluon:
Here, two important points should be stressed:
- Namely, first, that in all amplitudes above, only the relevant QCD part is shown, the
electron current, coupling to the photon, has been ignored: it is the same as in the
leading order case;
-
and that, second, the amplitudes have to be added and squared. This sounds trivial but
there is some catch. The catch is easy to understand pictorially:
where the "2" signify that there are two diagrams each of the corresponding kind,
i.e. two self-energies and two real emission, connected to both the quark and the antiquark line,
respectively. Squaring this expression, the LO contribution is readily recovered namely
It is also clear that only those diagrams can be multiplied with an identical number of ingoing and
outgoing legs. Therefore, at 1st order in the strong coupling constant α the virtual and the
real pieces neatly separate. The virtual pieces are multiplied with the LO contribution,
whereas the real pieces square directly:
DIVERGENCE STRUCTURE
It is clear that the virtual diagrams may contain ultraviolet divergences, that is divergences
for the loop momentum k→ infinity, that need to be regularised and renormalised. However,
a little bit of thought shows that both the real and the virtual diagrams also contain a new
class of divergence, called infrared divergence. This new type of divergent structure appears
for either k→ 0 or when the gluon momentum becomes collinear with one of the quark
momenta. In order to see this better, consider first the virtual amplitudes. Applying the
anticommutation rule of the γ-matrices and the E.o.M. for the spinors on the vertex correction
term results in
Clearly, at the level of approximation considered here, i.e. focusing on the most infrared-singular
terms, the effect of the correction for k→ 0 results in some overall correction on
the Born term. Similar reasoning also applies for the self-energy like diagrams. As a result,
the terms that are most singular in the infrared region can be combined to yield
Therefore the product of Born and virtual correction amplitude in this limit is proportional to
the leading order amplitude squared (and thus the differential cross section) up to the
loop factor
Applying the same reasoning to the real emission pieces, a few things change a little bit: First of
all, the unconstrained loop integral over k is replaced with a phase space integration over k, fixing
the invariant mass of the gluon. Hence, in this case, is fixed by a δ function such that
k² = 0. Second, in this case the correction term stems from both diagram pieces, cf. the
scetch above. However, following the same steps as above, shows that the real correction in the infrared
limit merely amounts to a corrections factor to the LO cross section, this time given by
CANCELLATION OF INFRARED DIVERGENCES
At this point it should be simple to see that the infrared divergencies cancel exactly. This in fact
should better be the case, since in any particle physics experiment as in any other real situation,
charges are accelerated leading to the emission of (electromagnetic or strong) radiation through
Bremsstrahlung. This radiation is predominantly soft; taking a closer look, it turns out that the
corresponding Fock state of, e.g. an electron, is populated with infinitely many photons with zero
energy. In order to ensure that physical quantities such as cross sections remain finite in the presence
of such soft Bremsstrahlung - as they have to, since cross sections can be measured and are finite! -
these infrared divergencies have to cancel. This is the essence of two theorems, namely the
Kinoshita-Lee-Nauenberg and the Bloch-Nordsieck theorems.
At this point it is worth noting that the reasoning above in fact generalises to all orders, i.e. to
arbitrarily many photons or gluons. This can be seen straightforwardly: The trick above (commuting
the slashed momenta and the γ-matrices before emplyoing the E.o.M.) just needs to be repeated
consecutively.
TOTAL AND DIFFERENTIAL CROSS SECTIONS FOR ee→qq AND ee→qqg
When doing the full calculation, the total hadronic cross section reads
Employing this correction, with α taken at the c.m. energy and adjusting the number of
flavours appropriately, there is perfect agreement with data.
So far, only the soft/collinear limit of real gluon emission has been dealt with. Let us now turn to
a discussion of the "full truth" of extra gluon emission in quark-antiquark production in
electron-positron annihilations. From the reasoning above it is clear that there are two
diagrams, related to the gluon emitted by either the quark or the antiquark line. The two Feynman
amplitudes read
Of course they need to be summed and squared. In so doing, the lepton and the quark line can be treated
independently. The total expression thus can be written as
where
Using completeness relations for the gluon, i.e. the sum over all polarisations, described by the
product of ε and its complex conjugate, and by making suitable replacements of the p+k terms
the quark piece can be written as
This is a lengthy expression, that needs to be contracted with the lepton piece. In order to come to
a more compact expression, it is worthwhile to introduce the dimensionless energy fractions
where i=1,2,3 stands for the quark, the antiquark and the gluon, respectively. Obviously, the corresponding
energies have to add up to the total c.m. energy, given by E² = s. Therefore, the sum of the three
x equals 2. In addition, they satisfy the following relations
In this terms, the three real contributions can be expressed as
This has to be integrated over the three-particle phase space. After expressing the three (massless) momenta
through the x, the three particle phase space reads
Taking everything together, the differential cross section for the emission of a real gluon together
with the original quark antiquark pair therefore reads
This expression clearly exhibits divergencies when either of the energy fractions goes to one.
The origin of this is clear, it has been noticed already before. Consider for example the case of
x1→ 1:
thus for this case either one of the two energies approaches zero (the soft divergence) or the opening
angle between antiquark and gluon (the collinear divergence) vanishes, resulting in a vanishing propagator
term and thus in a divergence of the emission cross section. Of course, these divergences, when
combined with virtual corrections vanish in the total cross section. Nevertheless, as soon as the gluon
is to be resolved, i.e. as soon as we're not talking inclusive total cross sections but exclusive
"gluon production" cross sections, we are left with divergencies when the resolution parameter is pushed to
zero. This, however, is an unphysical concept. There is just no detector with angular or energy resolution
down to zero. On top of this, the lightest QCD particle that can be measured is a hadron: the pion, with a mass
of roughly 140 MeV. At least this "minimal" mass shields the soft and collinear region, after all,
in order to be distinguished both the antiquark and the gluon must give rise to independent, distinct QCD
particles.