PHYSICAL
REVIEW
Single
D
VOLUME
quark
polarization
53, NUMBER
in quantum
of Phy sics,
1 FEBRUARY
chromodynamics
W. G. D. Dharmaratna
Department
3
1996
subprocesses
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK
University of Ruhuna,
M&ma,
Sti Lanka
Gary R. Goldstein
Department
of Phy sics,
Tufts University , M edford, M assachusetts
02155
(Received 29 August 1995)
It is well known that the single-polarization asymmetries vanish in QCD with massless quarks.
But, heavy quarks with a nonzero mass should be transversely polarized due to the breakdown of the
helicity conservation. In this paper we give the exact fourth-order pertwbative QCD predictions for
the transverse polarization from all QCD subprocesses, qfq’ -+ q+q’, q+g + q+g, gfg + q+b and
q + r$+ q’ + q’, which are significant for heavy quark production, with a description of the method
of calculation. The kinematical dependence of the polarization is discussed. Top quark polarization
from gluon fusion and quark annihilation processes, which are the important subprocesses at high
energies, is estimated and its significance is discussed.
PACS number(s): 12.38.Bx, 13.88.+e
It is well known that in QCD with massless quarks (and
order in a,,
the strong fine-structure
constant.
Those
gluons) the quark helicity is conserved at each vertex, and
so there can be no single-quark polarization in any per-
imaginary parts all involve similar integrations using the
Cutkosky rules. The imaginary contiibutions to the he-
twbatively
licity amplitudes
calculated
nomenologically
scattering
significant
@recess. This is a phe-
statement
for any hadronic
are then projected
out and inserted into
the relevant expression for the polarization.
The one-loop
process in which the relevant energies and transverse mo-
calculations
menta are large compared to quark “mass” scales. What
sets such mass scales is a long-standing question in QCD
applications. It is foreshadowed by the distinctions made
manipulation software, but are quite well defined.
To make the procedure more concrete we will sketch
the calculation of the QCD analogue of electron-muon
between current and constituent quark masses in preQCD quark and p&on models. Whatever the appropri-
scattering, p + q’ + Q + q’. The lowest-order diagram is
just one-gluon exchange as shown in Fig. l(a).
The M
ate definition
matrix for that diagram can be written as
of the quark mass, whether current, con-
are lengthy and are facilitated
by symbolic
stituent, effective, or running mass, once this parameter
is nonzero the strict helicity conservation breaks down.
And with the possibility of helicity flip vertices, singlequark polarization
can occur in scattering processes.
The observation of single-quark polarization, if it were
possible in spite of confinement, would signal the importance of quark masses as well as higher-order QCD carrections. The latter is necessitated by the dependence of
single-quark
polarization
on an interference between op-
posite helicities, which requires, in turn, imaginary parts
of bilinear products of amplitudes. Imaginary or complex
amplitudes
level.
arise from loop diagrams, i.e., beyond the tree
While
it is clear that direct observation
of quark
polarization is not possible, there are many indirect measures of the putative quantity, such as the polarization
or density matrix elements of leading hadrons [l] or the
distribution
of hadrons in jet fragmentation
[2].
Given the fundamental nature of QCD and quark spin,
we have been collecting together QCD perturbation theory predictions for single massive quark’ polarization in
some two-body processes. Below we present the results
for various flavor combinations
in quark-quark scattering,
quark-gluon scattering, quark-antiquark
annihilation and
gluon fusion. The technique for obtaining these results
is straightforward.
We first calculate the lowest-order
Feynman diagrams for the particular process and project
out the helicity amplitudes.
diagrams
that contribute
Then
we calculate all the
an imaginary
0556-2821/96/53(3~/1073(14~/$06.M)
part in second
53
(b)
FIG. 1. Lowest-order (a) and fourth-order (b) Feynman diagrams for the q+q’ + q+q’ scattering process. In the fourth
order only the diagrams which contribute to the polarization
are shown.
1073
zyxwvutsrqp
01996 The American Physical Society
W. G. D. DHARMARATNA
AND GARY R. GOLDSTEIN
1074 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
53
(1)
M=
where the indices and momenta are indicated in Fig. l(a)
and q2 = (pl - p3)‘. The resultant six independent helicity flip amplitudes,
*I(++
-+ ++)
I
a?(++ -+ --)
,
The f-‘s
are given by
fi = 8p2 - 4(p2 - E1E2) co2 ; ,
fi = -4mlmz
a$(+-
+ +-)
,
a%(+-
+ -+)
,
f3 = 4(p2
a%(++
+ -+)
a% (+-
- 3 ++)
2 ’
fs = -4mlEz
sin’ i,
+ E1E2) cd
8
f4 = 4mlmz sin’ -
i,
fe =
-4m&
03si sin i ,
cm i sin i ,
where the symbols E1, Ea, p, and 0 defined in the centerof-mass frame are the energy of Q, the energy of q’. the
, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
momentum and the scattering angle, respectively. The
next order imaginary contribution is shown in Fig. l(b).
There
is only one Feynman diagram that contributes to
,
the imaginary part in the s-channel physical region, the
box diagram. The M matrix for that diagram is
can be expressed in the form
(3)
which contains the integration over the internal four-momentum k of the loop, and with S,, a gluon mass cutoff
that protects against infrared divergences, which will be taken to zero (since such divergences cancel out of the final
physical result). First, the k-dependent terms can be separated as
MC&,
(XaX’);j(~a~“)~~{4~~
.p2(7” @7v)b - 2[(Y7”$,
@ yv) - (7” @ .YvYli# l)]b, - (y ” y T
@ w f’~,Jbnp}
,
(4)
where the notation (A @ B) = [f&)A(p,)][~@ &% (p)2)]
is used and
d% (l; k,; k,k,)
(b;bx;bxJ
=q
(k~- 6,$J((k- q)~- 6,z,J((pl
- k)Z
- mT)((pz+k)2
- +)
(5)
.
The imaginary part (or the discontinuity across the unitarity cut in the physical s channel) is obtained from the
Cutkosky rules by replacing the two direct channel propagators with on-shell 6 functions: zyxwvutsrqponmlkjihgfedcbaZYXWVUT
(p1 - ?%;2 - r nt
(pz + !$
- m;
*
2?ri6((p1 - k)Z - mf)O((pl
==3 27w(p,
+ !q
- k)O) )
- m?J6y(pz - k)O) )
where the 0 functions take the positive-energy parts. This replacement puts the particles corresponding
intermediate states on their positive-energy mass shell. Equation (5) becomes
6; b,; b,,),ut
= i(274’
/
wl;kw~((Pl
- kY - rnfyq(Pl
(@ - 6L)((k
These integrations have been performed for electron-electron
- rc)“ )s((Pz + k)Z - m;)e((pz
- q)Z - &)
to the
- k)O)
scattering in quantum electrodynamics
[3], but in the
SINGLE QUARK POLARIZATION IN QUANTUM zyxwvutsrqponmlkjihgfedcbaZYXWVUTSR
53 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
1075
large-s limit. Here these integrals are calculated
The integrati&s
exactly, up to terms that vanish as S, goes to 0.
are handled by using cylindrical
k E (ko, kl
cos4,kl
coordinates
sin4, k,)
for zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIH
k in the center-of-mass frame: i.e.,
and d4k E dkok,.dkld4dk,
First the integration over 4 is performed and then one of the 8 functions is used to do the ki integration.
The
remaining 6 function becomes a constraint on ko (in this particular case ko = 0), so only the integration over k,
remains to be done. For the simplest case (b,,,) this takes the form
where F = 2k,p and S = (pl + pz)’ is the square of the
total center-of-mass energy of the subprocess. The integration over 5 is done analytically neglecting the terms
that vanish in the zero-gluon
mass limit. Equation
Imb== = (5)
(6) is
&!!$*(!&
related to the imaginary parts of the integrals via
2Wb;
b,; bp)
The resulting imaginary
Imb=
(5)
Imb, =
= (b; b,; b&t
(5)
+ y]
,
(8)
parts of the integrals are
-+I($)
[&l*(s)
Imb,,=
($)
[$I($)-~]
,
Imbw=
(5)
[q&p2ln($)]
Integrals
with other combinations
,
sin(g)
,
of indices all vanish
because of the symmetry properties of the integrand.
Note that the masses never enter explicitly in any of
these results. Masses are imbedded in the kinematic definitions of B,.p, and 4’. Note also that there are only
1
-~
l*
( )I
Q2
logarithmic terms in the scalar and vector
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
-q=
q2+4p2
ing nonlogarithmic
Imb,,
= Imb,,
terms to the polarization.
The next step requires the integrated
(9)
= (5)
[~sinOln(~)
expressions,
but there are some nonlogarithmic terms in the tensor
expressions. These latter will be significant in contribut-
’
-isin@]
,
terms be folded
in with the Dirac matrices and s&nom in Eq. (3). This is
performed in the helicity basis by using MACSYMA, and
the results, the imaginary parts of the second-order six
independent helicity amplitudes (4;) can be written as
4; =
Here fa’s
are d&ned
(10)
earlier for the lowest-order
p2i + mfmi
91 = 16
gz = -16m1mz
In sin g + p’(p? - El&)
co? i
K
diagram and ga’s are
>
sin’ i ((J&L&
- p”) In sin i + p’)
1
sin’ % ~~,
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJ
,
16
g3 = cos2(0/2)
In sin i + p2(p2 + EIE2)
[(
sin’ i co2 i
1 (11)
,
1076
W. G. D. DHARMARATNA
g4 = 16mlmz sin’ z
(&&
9
+ p’) In sin 5 - p2
AND GARY R. GOLDSTEIN
53
,
sin(fl/2)
p2& + Elm; cosz
Q5 = -16m1cos(e/2) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
g6 = -16mzp
sin(B/2)
cos(6/2)
p2& + Ezmf cos2
Each amplitude has the expected dependence on quark masses and angular variables. Note that the infrared divergent
term which appears only in Eq. (10) is proportional to the lowest-order amplitudes as it should be for cancellation.
Having obtained the lowest-order helicity amplitudes and the imaginary parts of the second-order h&city flip
amplitudes, the next step is to calculate the leading contribution to the transverse polarization. The transverse
polarization (‘P) of a particle in a scattering process can be defined in terms of cross sections measured with different
spin configurations of that particle: namely,
where the arrows denote spatial spin (up, down) configurations of the particle with respect to the scattering plane.
Here spin up (down) direction is chosen to be parallel (antiparallel) to the vector A = pa x p~/Ip, x FBI.
When expressed in terms of helicity amplitudes, Eq. (12) takes the following form for the polarization of particle
C of the process considered:
Finally, the substitution of helicity amplitudes into the above equation leads to the following relatively simple expression for the polarization of the scattered quark [4]:
P=
a.(-l/3)mlq2sin(8/2){(&
2p~cos(e/2){2p4cos4(~/2)
+ 2El)ms ln[sin(6’/2)] + E~p2cos2(O/2)}
+ [(mi + mf + 2E1Ez)p2 - 2p4] cosz(8/2) + 2p4 + mfm;}
where the factor (-l/3)
is the SU(3) factor for color
triplets scattering on &or triplets aAd a. is the strongcoupling constant, which depends on the logarithm of
the momentum transfer in the usual treatment of QCD.
Note that the polarization is independent of infrared divergences as expected.
Note also that this exact, expression contains a logarithtiic term in the numerator,
ln[sin(0/2)], and an additional contribution, - [cos(0/2)],
that would not be present in a “leading log” expansion.
The general form of the color factor for s - 2~scattering
can be written as
where C!, = 4/3 and C, = 10/9 for a color triplet source
and C, = 3 and C, = 0 for a color octet source.
For U(1) or QED the color factor and the coupling constant have to be replaced by {l} and a. Otherwise the
formula is the same for electron (e-)-muon
(II-) scattering and is the same but with opposite sign for e--p+
scattering. Note that the overall sign of the expression
depends on whether it is a particle-particle interaction
or particle-antiparticle interaction in the case of QED
(in addition, the color factor is also different in the case
of QCD). The sign change is a consequence in ordering
fermion-antifermion fields. In the limit that the mass
’
(14)
(mz) of p+ becomes very large, compared to ml and p,
this reduces to the polarization of electrons scattered off
a spin-i nucleus (2 = l), the Mott asymmetry [5]:
mlp z&(0/2)
’
= 2aEf cos(0/2)[1-
ln[sin(0/2)]
(pZ/Ef) sin2(Cl/2)]
(16)
That neglects any form factors, so is relevant for low q’.
Hence, Eq. (14), with the indicated group and coupling
factors altered so that {‘}a. + -laZ,
provides the recoil
correction to the Mott asymmetry formula for electronnucleus scattering. Whether or not this explains slight
discrepancies between the Mott formula and data [6] for
scattering from ,~Au remains to be checked.
Having calculated the asymmetry for s - u scattering
and following the above arguments, it is easy to show
that the transverse polarization of the s quark from s-c
scattering can be obtained from Eq. (14) by changing
the overall sign and replacing the color factor by {7/6}.
Therefore the final result gets the same sign for both
processes and is equivalent to a scattering of an s quark
on an effective color potential, which is attractive, m the
limit m2 is very large.
The other three subprocesses can be calculated by following a similar procedure. However, there is a number
53 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
SINGLE QUARK POLARIZATION
IN QUANTUM zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQP
1077
of fourth-order Feynman diagrams that contribute to the.
gluon fusion, and quark annihilation-are
given in the
imaginary part of the amplitudes of,each of these subprofollowing sections.
:
cesses. Particularly, quark-gluon elastic scattering and
QUARK+GLUON
SCATTERING
gluon fusion processes involve not only a large number of
F:ynman diagrams but also a large number of additional
The lowest-order and the fourth-order Feynman diaintegrations zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
(b,,,,b.,,,
etc.) due to the triple-gluon cougrams that contribute to the imaginary amplitudes of
pling vertex which makes the calculation more difficult.
this subprocess are shown in Figs. 2(a) and 2(b), respecFeynman diagrams that contribute to the imaginary part
tively. Adding the contributions from all Feynman diof amplitudes and the resultant expressions for the poagrams the following lengthy expression is obtained for
larieation of the other processes-quark-gluon
scattering,
the polarization of the scattered quark of mass m [4]:
where
1
F=2 sin(e/2)JEZ
NL =
ln
- p2 cos2(e/2)
2 a4
;(P - E)p2 + cos2 ;(4Ep2
N2 = l6 cos6 ;(P - E)p2 + 7 Cm4 ;(E2
N3 = -2p cos i
E2 -pzcose
-p2)p
+ 2p sin(S/2)JE2
-p2cosz(B/2)
- Ezp + 3E3) _ E2(p + SE) ,
+ 3 cos’ ;(Is~~
- 14~~ + 3E2)E - (47p2 + 28Ep + 45Ez)E
sin’ i (9p + E)p - llp2 - 12Ep + 7E*
16p sin’ i - 96
>(
D = [4P2cos2, +p(p + 9E)cosB
’
+ 4p2 + 9E(E +p)][cos6’(p2
,
,
>
+ 2Ep + 4E’)p
+p3 co? 0 + cos2 e(E3 - p3 + E=p + 2Ep2) + p3 + 2Ep2 + E=p + Es] ,
where E, p, and B are the energy of the quark, the momentum, and the scattering angle, respectively. All quantities
are defined in the center-of-mass frame of the subprocess. The color factors are not separately shown in the expression
as in the previous case because the addition of the contributions from different diagrams with different color factors is
done. The delicate cancellation of infrared divergences in the imaginary parts of loop diagrams, that must be obtained
in the polarization formula, is a good test of the correctness of such a lengthy calculation. The above expression,
the polarization of the scattered quark off a gluon, is distinctly different from the analogous QED expression, the
polarization of the scattered electron off a photon-Compton
scattering. This is mainly due to the existenw of the
triple:gluon coupling vertex in &CD. In the lowest order the g-g-9 vertex produces a t-channel exchange and in the
fourth order several new diagrams contribute as shown in Fig. 2.
By replacing the non-Abelian color factors appropriately, the corresponding scattered electron polarization in Compton scattering is obtained (41:
p = am(E + p cos 0) s&(0/2)
pcos(8/2)D’
e
E-P
2~
,
FN:+N:l*E+p ---N;cos~~G zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPON
>
(
D’ = (1 + cm2 e)[p ‘3 cos e + 2Ep2 + E’p + E3] + p3 sin2 e + 2Ep(p + 2E) cos e ,
N: = 2P2(P - E) ox* ; + E(4p2 - Ep + 3E2) &
N; = (P - E)
N
(
P cos2 ; - E
; - E2(p + 5~) ,
cm2 ; - E(p + SE) ,
>
= EP ~0s’ ; + 2(pz + Ep + E2)
(18)
W. G. D, DHARMARATNA AND GARY R. GOLDSTEIN
1078 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
However, the magnitude of the polarization
For example, the largest polarization
of
electron (mass - 0.5 MeV) off a photon,
when 0 N 0.85 rad and p - 1.1 MeV, is
is quite small.
the scattered
which occurs
about 0.06%.
(a)
53
For p (mass N 105.7 MeV)
and 7 (mass -
the magnitude of the polarization
at
about the same as it should be due
could be written as a functmn of m/p.
tude of the polarization is mainly due
1784 MeV)
the peak remains
to Eq. (18) which
The small magnito the small value
(a) zyxwvutsrqponmlkjihgfedcbaZY
(b)
FIG. 2. Lowest-order (a) and fourth-order (b) Feynman
diagrams for the g+g + pfg scattering process. In the fourth
order only the diagrams which contribute to the polarization
are shown.
FIG. 3. Lowest-order (a) and fourth-order (b) Feynman
diagrams for the glum fusion, 9 f 9 --t 4 + q, process. In
the fourth order only the diagrams which contribute to the
polarization are shown.
SINGLE QUARK POLARIZATION IN QUANTUM..
3 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
of the coupling constant
(a) in QED. This small,
01 effect would be very difficult to measure.
GLUON
order
FUSION
The Feynman diagrams that contribute
to the polarization through gluon fusion are shown in Figs. 3(a) (lowest order) and 3(b) (fourth order).
Note that the existence of the triple-gluon
coupling vertex produces an schannel exchange in the lowest order. In one-loop level
several new diagrams
contribute
to the complex am-
1079
plitude which include radiative correction
to the gluon
propagator-“vacuum
polarization.”
Particularly,
the radiative corrections to the gluon propagator
contribute
to
the polarization
as a result of the interference
with lowest order t-channel
and u-channel
amplitudes.
This did
not happen in s-g scattering
since the va.cuum polarization corrections to the gluon propagator
are pure real in
the cross channel kinematic
region.
Only the radiative
corrections
to the quark propagator
contributed
to the
polarization
of the s-quark from s-u scattering.
The resulting quark polarization
for 9 + 9 + p + Q is
given by [7] zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
- k2cos*B)
p = o18m(p2
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
(~~+~~)Y++(~~-12’2)Y-+~~ln~+~~+18k3sin2RcosB(C1+Cz)
24k sin OD
where
D=(9k2cos20+7p2)(k4cos40- 2k2m2sin20- p4),
Nl = Qk2Cos30(~= + 2k2) + 6kp cosO(27p2 + llkp - 27k=) + 27kacos0
Nz = kp cos20(llp2
,
+ 76k2) - 162p2m2 + 33k3p ,
Ns = P COS0[243m=p zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
CoS4
0 - cos’ O(324m2p - 54k3) + 22kp2 - 243m=p + 164k3] ,
N4 = Gksin2
0 ~0~0172 cos2 6’(27p3 - 18k2p + k3) + 27(Q6kzp - 24p3) _ 8(22kpz + 45k~)1 ,
cl
= ;
+ rnf)O(&‘zTni)
) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJ
c
&=?(2P3p2
y*
cp*y4*>
=I* zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
This QCD polarization
formula has an interesting
structure,
mainly due to the contribution
from diagrams with
internal fermion loops. rn; is the mass of the internal fermion and the summation
runs over all possible intermediate
quark masses. For strange quaik production,
as an example, rni takes up, down, and strange quark masses at energies
below the threshold for charm production.
In other words an extra contribution
is added to the polarization
at each
threshold when the center of mass energy is increased.
In the case of pair annihilation,
the QED analogue of gluon fusion, the corresponding
expression for the polarization
of the produced lepton can be obtained by setting the QCD structure
constant
(f&
to zero and by replacing the
other color factors by 1. The resulting expression, which is distinctly
different from that of &CD, takes the form
p=,m(p2- k2cos20)
4p sin OD’
where
N+y +
+ N- Y-
+ 2 cos0
(p” + 5k2) ln;G
+ 2kp cosO)]
,
1080
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
W. G. D. DHARMARATNA
AND GARY R. GOLDSTEIN zyxwvutsrqponmlkjihgfedcbaZYXWVUTS
D’ = k4 ax4 0 + 2k=p= cos= 0 - 2k4 cm= 0 - p4 - Zk=p= + 2k4 ,
N* = +[(p* + 2k2) cos’,
The center-of-mass
+ 3k2 z&6kp COST]
variables and Y* are the same as above.
QUARK
PAIR
ANNIHILATION
Here, only the cake with different flavors in the initial and in the final states is considered.
vacuum polarization
diagrams
appear in the s-channel exchange and add to the imaginary
not contribute to the polarization
lowest-order and the fourth-order
a.mz
ss is
Nl + E4kzmf(56X-
24p2k3 sin t’D
But, they do
in this case, since they are just overall factors to the lowest-order amplitude. The
Feynman diagrams that contribute to the polarization are shown in Figs. 4(a) and
4(b), respectively.
Polarization of the s quark from uti +
p=-
Similar to gluon fusion,
amplitude.
+ 16X+) + Nz In (~)+W+)]
(21)
,
where
D = 4E4 + 2E2(mf + rnz - E’) sin’ 6 + 2~~~77~~
co? 6 ,
Nl = -p3k sin’ .9[(4k3 + 72Emz + 36E3)p cm 0 + 40E3k] ,
NZ = E’p[p’mi:
sin’ 0(54p cos 6 + 20k) + k’mf(36p
N3 = 4E2pk2m;(18k
x* =
and p(k)
cos B - 5~) ,
(E2 f kp cos S)= - mfm;
p(3kp cos 6 f p2 f 2k=)
2EJ(E2
cos 0 - 40k)] ,
f kp c0s0)~ - m$n;
(E2 f kp cost?)= - mfm;
’
state,
and the other symbols are the same as defined in Eq. (21).
The experimental verification of this formula is ex-
respectively, and E is the energy of the quark (or antiquark) which is the same for both initial and final states.
For the QED analogue of quark annihilation,
e-
tremely difficult because of the small magnitude of the
polarization.
For example, the highest polarization of
the produced p from e- + e+ annihilation is of the or-
+e+
der of 0.5%, which is approximately
and ml(mz)
are the center-of-mass
and the mass of the particles
momenta
in the initial (final)
--t p7 + JL+, the polarization
of the produced
CL- is
the same for heav-
obtained by neglecting the diagram with the triple-gluon
ier fermion (7) production
vertex and changing the color factors appropriately
higher momentum.
The suppression due to the small QED
p=-
L+mz
p2k2 sin6’D
N; + 2E4kmf(X-
- X,)
[4]:
except the peak appears at a
stant does not OCCUI‘in the polarization
massive quark from e-+e+
contribution
*
coupling con-
of the produced
~q. In this case the leading
to the polarization
can be calculated
rela-
tively easily, since it involves only one Feynman diagram
in the one-loop level. The polarization arises from the
-2E2p2mf
(E-p
)I’
gluon radiative correction to the photon-quark vertex as
E+P zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
shown in Fig. 4(c). Hence, the resultant polarization is
In zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
(24
proportional to the strong-coupling constant a,. Contributions from all the other diagrams are suppressed by a
where
factor of ~QED.
Ni = p3k sin’ 6’(k2p cos B - 4E3] ,
The resultant polarization
duced quark, in the coordinate
for the pro-
system considered,
has
the simple form
mzk sin 9 cos 6
N; = 2E2p[m:(E2
- p2 cos* 0) - mf(E2 + k2)] ,
= -as{4’31
[2k2(l + co9 0) + 4m;] ’
(23)
SINGLE QUARK POLARIZATION
IN QUANTUM
53 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
1081
where mz and zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
k are the mass and the center-of-mass moquarks, as given in [2], for example,
mentum of the quark, respectively.
{4/3}
is the resultant
SU [3] color factor from quark vertices. This has been derived previously [S] in the large-Q’ limit.
It is clear from Eq. (23) that the polarization of the
quark can be written as a function of 8 and mJk.
The polarization
mz/k
N
0.87.
peaks when 0 =
Therefore
the peak occurs above
of their momentum
direction, has been suggested
ful description of fragmentation
less than 5% asymmetry
KINEMATICAL
can be used to ob-
However,
Eq. (23) predicts
at the best.
DEPENDENCE
POLARIZATION
OF THE
the
able quantity, assuming that the particles in the jet could
remember the direction of the polarization and preferhave a component
quantities.
x/4 (or 3?r/4) and
threshold but well below the scaling region. An interesting way to relate the quark polarization to an observ-
entially
tain observable
The magnitude
variables.
of the polarization
The scattering
depends on several
angle, 6’, and the momentum
or the energy in the center-of-mass
frame can be varied
in that
independently. Hence, we need to study the variation of
the polarization with these two variables. Furthermore,
IS]. Also, a more careto estimate the polarization from the full QCD formula,
of transversely polarized zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
one needs quark masses and the strong-coupling
aa. What quark mass-current,
constituent,
a larger effective mass incorporating
constant
or perhaps
confinement effects
at lower energies where the perturbative QCD is partially valid-should
be inserted? Spontaneous symmetry breaking of chiral invariance, incorporating
instanton
effects in the vacuum, have been evoked to give quark
propagators mass corrections 191.The effective mass that
arises for one quark flavor is momentum dependent, as
(a) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
anticipated, and gives expected “constituent” values at
low momentum.
It is an interesting
ization
expressions
point to mention
can be written
that the polar-
as functions
of a.,
8, and m;/p, eliminating the other parameters by using
the kinematical relations. This shows that the properties of polarization
can be discussed for any given quark
mass; for a higher mass the same properties will appear
at a higher momentum.
Hence, for simplicity, the constituent quark masses, 0.3, 0.5, 1.5, and 4.5 GeV/c’ for
up (or down), strange, charm, and bottom quarks, respectively, are used when necessary. Also, a constant
value, a, = 0.4, is used for the strong-coupling constant
which is an overall factor in the polarization.
The kinematical dependence of the polarization
outgoing
strange quark in the center-of-mass
of the
frame is
shown in Figs. 5(a)-5(d)
for the four subprocesses. The
polarization as a function of 0 is shown for different values
(b)
of center-of-mass
momentum
(the region of the momen-
tum is chosen to study the polarization at the peak). For
large momentum relative to quark masses the polarization still becomes
negligible
as expected.
However,
for
relatively low momenta there is a considerable structure.
The first two subprocesses, s+u and s+g scattering, produce smaller polarization
glu0.l
*
change in the polarization
Cc)
FIG. 4. Lowest-order (a) and fourth-order (b) Feynman
diagrams for the annihilation, r~+ rJ + q’ + rj’, process. In
the fourth order only the diagrams which contribute to the
polarization are shown. (c) The most significant Feynman
diagram for the polarization of massive quark through e++eannihilation.
compared
with the other two.
It should be pointed out that the full expression for the
polarization is significantly distinct from that of the leading logarithmic approximation,
as an example the sign
of the additional
from s + u is a consequence
contribution
from the nonlogarithmic
term in the expression.
Another interesting thing to study is the quark mass
dependence of the polarization. For s + u scattering this
is done by expressing
the polarization
fl, C.T= ma/p, and y = mu/p.
as a function
of
Figure 6(a) shows the 3:
and I/ dependence of the polarization at the peak which
occurs at 0 N 130’. Although the parameters + and y
zyxwvuts
W . G. D. DHARM ARATNA
AND GARY R. GOLDSTEIN
1082 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
are expressed in terms of strange and up quark masses,
the result could be read for the scattering
of any two
quarks of different flavor. The sign of the polarization
is determined
by the parameter
y. For small values of y
the polarization
is positive and it peaks when y = 0 and
cc N 1.5. For large y, the sign changes and the largest
negative polarization
(- - 5.5%) occurs when y -+ 03 at
z N 0.4, i.e., the polarization
gets its largest negative
value if the s quark is scattering
off a static color source.
For a color octet static source the polarization
is still
negative and it is - -12% at the peak.
For the s+g scattering,
most of the features, peaks, and
the sign changes, are very similar to the previous case.
The major difference is that the largest negative polarization occurs at moderate momenta
(p - 1600 MeV/c)
and the positive peak exists at smaller values of p N 200
MeV/c. Therefore the kinematic regions in which the two
different subprocesses
change sign and peak do not coincide. However, those regions could overlap when smeared
by relevant quark and gluon structure
functions.
The
quark mass dependence
of the polarization
is easy to
study here, since this process involves only one quark.
The polarization
as a function of the center-of-mass
scat-
FIG. 5. Polarization
53
tering angle (B) and m/p is shown in Fig. 6(b). Although
there are several bumps and dips, the overall magnitude
is smaller than in the previous case. Again, the magnitude is largest when 0 N 130°, i.e., in the backward
scattering
and the sign changes with m/p. The sign at
the peak changes from negative to positive when m/p increases from - 0.35 to - 2.0. For larger m/p it becomes
negative again and peaks at B N 85O.
The quark polarization
from gluon fusion is more interesting than the two subprocesses
discussed above; the
magnitude
is comparatively
larger and it remains significant for larger values of center-of-mass
momenta.
The
estimation
of the s-quark polarization
involves not only
the strange quark mass but also the other quark masses
which could be excited depending
on the available energy. The polarization
vanishes [Fig. 5(c)] at 0 = s/2 and
the sign changes leaving the magnitude
the same under
B + ?r - 6 as one expects from constraints
on helicity amplitudes under identical particle interchange.
The quark
mass dependence
of the polarization
at the peak, i.e., at
0 N 60”, is shown in Fig. 6(c). The variation with the
final-state
quark mass (m) and with the gluon centerof-mass momentum
(p) is given. The constituent
quark
of outgoing s quark in the subprocess center-of-mass frame as a function of the scattering angle with
me, = 4.5 G&‘/c’.
(a) q+q’ -i q+q’, (b) q+g + q+g,
a, = 0.4, m, = md = 0.3 GeVjc=, m, = 0.5 GeVJ c=, m, = 1.5 G&/c’,
(c) g + g --t q + q, and (d) q + q + q’ + 4’.
SINGLE zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
QUARK POLARIZATION IN QUANTUM
53 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
masses are used for all internal fermion loops and the
numerical result is almost the s&me if they are replaced
by current quark masses., Note that the magnitude of
the highest polarization remains the same while the peak
gets broader when m increases. The peak occurs when
mfp N 0.3. More interesting is the momentum structure
of the polarization. For small momenta, polarization increases rapidly with p until the peak value is reached and
then decreases slowly, a slower rate than one would expect fxom a leading logarithmic approximation in
The new terms that are added to the polarization at each
threshold of pair production, when p increases, are partially responsible for that effect.
It is an interesting point to mention, even though an
m/E.
1083
individual Feynmati diagram is not a gauge-invariant entity, that the dominant contribution to the polarization
comes from one particular diagram, namely from the radiative corrections to the gluon propagator--“vacuum
polarization.” This appears in the s-channel exchange
and contributes 4o the imaginary parts of amplitudes in
gluon fusion, in contrast to the gluon propagator corrections in the t-channel exchange for s + u and s + 9 scattering subprocesses with no contribution to the imaginary amplitudes. That imaginary amplitude contributes
to the polarization as a result of the interference with
the lowest order t- and u-channel amplitudes. The existence of the diagram with an s-channel intermediate
gluon, for the subprocesses considered, is a result of the
FIG. 6. Mass dependence of the polarization of the outgoing s quark. AU parameters are the same as in Fig. 5. (a)
q+q”q+q’,(b)q+g~q+g,(c)g+g’q+4,and(d)q+~--tq’+6.
W. G. D. DHARMARATNA
1084
AND GARY R. GOLDSTEIN
non-Abelian character of the theory which makes the polarization large and distinct from the analogous QED expression.
The polarization of the s quark produced from the
quark annihilation [Fig. 5(d)] has interesting characteristics which are similar to those f?om gluon fusion; Namely,
the magnitude is large and the highest polarization occurs at reasonably larger center-of-mass, momenta. Interestingly,, the region in the parametric space, p and 8, at
which the polarization peaks is the same for these two
subprocesses. This will certainly help to minimize the
dilution of the polarization in the addition of the two
contributions within the convolution integrals. For the
values of parameters chosen the polarization is always
positive and it peaks at p N 1.5 and 6’ N 80’. The magnitude is slightly larger than all the other processes. More
interesting is the quark mass dependence of the polarization which can be parametrized as a function of z = m,/p
and y = mJp, where m, and m. are the initial-state
(u-quark) and the final-state (s-quark) quark masses. I
and y dependences of the polarization at the peak, i.e.,
at 0 N SO’, is shown in Fig. 6(d). Note that the kinematical constraint for pair production, S > 4mi, leads
to 1 + y2 > 1’. While the magnitude is quite small and
the z dependence is insignificant for large y, there is a
reasonable structure for small y which is the region that
produces heavier quarks from lighter quark annihilation.
The largest polarization (10.5%) occtirs at z N 0.5 and
smallest possible y. i.e., y = 0 or for massless quarks in
the initial state.
HEAVY-QUARK
POLARIZATION
It is interesting to pursue the possibility of heavy
quarks, charm, or bottom quarks, becoming polarized
in the production processes. It is well known [lo] that
the heavy-quark production cross section is dominated
by the gluon fusion subprocess, especially at large transverse momenta and small Feynman z, while the annihilation processes give the next significant contribution.
This general rule is true as long as the beam or the target
does not carry any valence heavy quarks, because, then
the heavy-quark contribution to the structure functions
is negligible. Therefore, the first two subprocesses, the
flavor excitation processes, are not considered for heavyquark production.
The polarization of down, strange, charm, and bottom
quarks from gluon fusion are compared in Fig. 7(a). Here,
the center-of-mass momentum of gluons is chosen (p = 13
GeV/c) to obtain the largest polarization for the bottom
quark. Aspcan be seen by comparing with Fig. 5(c) the
overall magnitude of the polarization has not increased
for heavy quarks, but at a given momentum the heaviest quark acquires the largest polarization. This is always
true if the center-of-mass energy is well above the’threshold for pair creation. For heavy-quark production from
u + ti annihilation, the result is very similar and is shown
in Fig. 7(b). Here, the center-of-mass momentum of the
u quark is chosen to be p = 0.9 GeV/c to see the peak
of bottom quark polarization. Again, the heaviest quark
FIG. 7. Polarization of up, strange, charm, and bottom
quarks at the subprocess CM momentum of (a) 13 G&‘/c
for glum fusion and (b) 9 GeV/c for annihilation. Other
parameters are identical to Fig. 5.
gets the largest polarization at a given momentum as in
the previous case. In addition, the magnitude at the peak
gets larger for heavier quarks. For example, the peak polarization for bottom quarks is about - 10% while that
for strange quarks is about 7%. This is due to the fact
that for heavier quarks the peak polarization occurs at a
higher momentum which lowers the value of the quantity
y in Fig. 6(d) moving the kinematical region towards the
peak of the polarization. However, once we go to the
higher momentum the effects of the decreasing coupling
constant have to be taken into account, and that is an
overall factor.
TOP QUARK
POLARIZATION
The study of the polarization properties of the top
quark,, when its mass is near or above 120 GeV, is more
interesting for several reasons and worth giving special
consideration. As has been already discussed in detail
[11,12] the physics of the top quark has distinctly different features from other quarks, especially in its polarization phenomena. The main interesting feature for the
SINGLE QUARK POLARIZATION
IN QUANTUM
s3 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
1085
zyxwvuts
spin physics of the top quark is the rapidity of its decay
process. First, the depolarization mechanism &ective for
other quarks does not play a role for the heavy top quark
and, second, it decays as a free quark with no hadronization or recombination
So, any polarization
effects due to its rapid decay [ll].
information
top quark should be directly
products
with neither dilution
thermore,
carried by the produced
transferred
The
Fur-
the dominant decay process of the top quark,
the electroweak two-body decay t +
ideal analyzer of its own polarization.
gluon
into its decay
nor enhancement.
transverse
fusion
polarization
and quark
bW+,
of the top
annihilation
will be an
quark
subprocesses
from
are
shown in Figs. S(a) and S(b) for a top quark of IWSS
140 GeV/c’.
The constituent quark masses given earlier are used for the other quarks.
quark polarization
Note
that the top
from gluon fusion depends on all the
other quark masses due to the contribution
from the vac-
uum polarization diagram. The approximate value of the
running coupling constant, a. = 0.15, is used in the estimation of the magnitude.
Although the magnitude is
small, - 2% from gluon fusion and N 4% from annihilation at the peak with the above value of the a,, it is not
iero even in the kinematic region where the perturbative
QCD is expected to be valid with no doubt. This is an
even more interesting result to verify experimentally, if
it is possible, since the predicted polarization
arises from
the one-loop level but not as a correction to the leadingorder predictions. This is a direct test of the fourth-order
prediction from perturbative &CD.
By using the polarization
formula for gluon fusion,
published previously 171, an interesting method to test
the top quark polarization
Collider
(LHC)
at the CERN
Large Hadron
has been suggested in [12]. At the LHC
energies, the top quark production is dominated by the
gluon fusion subprocess and the next significant contribution comes from the annihilation subprocess [12]. Having obtained the polarization from annihilation processes,
one can include both subprocesses’ polarization in the
convolution with the quark and gluon distribution functions to obtain the resultant polarization
for the appro-
priate hadron collider, which is neglected in [12].
Since the top quark has been observed at Fermilab, it
FIG. 8. Polarization of the top quark of mass 140 GeV/c’ ,
(a) from gluon fusion and (b) from annihilation. Other parameters are identical to Fig. 5 except a. = 0.15.
incorporated
into new physics.
It is an interesting
point to mention
that the large
transverse polarization in inclusive hyperon production
processes, which involve the strange quark production,
observed previously
(see references in 171) will not occw
here since the top quark is produced at very high energy and it decays very rapidly-the
polarization of the
top quark is preserved
as discussed above.
Therefore,
may be possible to collect a few hundred top events (if the
expected luminosity is reached) before the LHC to study
any observation of top quark polarization larger than the
predictions given here will be an indication of some new
the physics of the top quark.
or nonstandard
energies the annihilation
At the Fermilab
Tevatron
subprocess plays the dominant
model interactions,
which could happen
at high energies where the top quark is produced.
This
role in the production of the heavy top quark. Therefore,
the polarization of the top quark through the annihilation
makes the measurement of the top quark polarization
very important and interesting experimentally as well as
process is more significant here.
theoretically.
As a result, the resul-
tant polarization of the top quark at Tevatron energies
is slightly larger than that of LHC. The significance of
the top quark polarization at the Fermilab Tevatron has
been discussed already [13]. However, in the discussion of
the single-spin asymmetry
ing plane, the contribution
perpendicular
to the scatter-
from qg subprocesses, which
is given here, should be included in order to obtain the
upper bound for observable effects due to the standard
theory.
Then, any sizable asymmetries
zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK
Note added in proof. An error in N4 of Eq. (19) was
brought to our attention by A. Brandenberg 1141to whom
we are grateful.
observed can be
ACKNOWLEDGMENT
This work was supported
U.S. Department
of Energy.
in part by a grant from the
W. G. D. DHARMARATNA
[l] R. H. Dalitz, G. Ft. Goldstein, and R. Marshall, Phys.
Lett. B 215, 783 (1988); G. Ft. Goldstein, in High Energy Spin Physics: Eighth Internnational Symposium, Proceedings, Minneapolis, Minnesota, 1988, edited by K. .I.
He&r, AIP Conf. Proc. No. 187 (AIP, New York, 1989).
[Z] J. Collins, Nucl. Phys. B396, 161 (1993), and references
contained therein.
[3] R. N. Cahn and Y. S. Tsai, Phys. Rev. D 2, 870 (1970).
[4] W. G. D. Dharmaratna, G. R. Goldstein, and G. A. Ringland, Z. Phys. C 41, 673 (1989). (Note: The sign error
has been corrected here.)
[5] W. A. Mckinley and H. Feshbach, Phys. Rev. 74, 1759
(1948); R. H. Dali& Proc. R. Sot. London A206, 509
(1951); A. I. Akhieaer and V. B. Berestetskii, Quantum
Electrodynamics,
3rd ed. (Nauka, Moscow, 1969).
AND GARY R. GOLDSTEIN
53
[6] P. E. Spivak et al., Sov. Phys. JETP 14, 759 (1962).
[7] W. G. D. Dharmaratna and G. R. Goldstein, Phys. Rev.
D 41, 1731 (1990).
[8] G. L. Kane et al., Phys. Rev. Lett. 41, 1889 (1978).
[9] D. I. Dyakonov,and V. Yu. Petrov, Nucl. Phys. B272,
457 (1986).
[lo] P. D. B. Collins and A. D. Martin, Hadron Interactions
(Hilger, London, 1985).
[ll] R. H. Dalits and G. R. Goldstein, Phys. Rev. D 45, 1531
(1992).
[12] G. L. Kane, G. A. Ladinsky, and C. -P. Yuan, Phys. Rev.
D 45, 124 (1992).
[13] G. A. Ladinsky, Phys. Rev. D 46, 3789 (1992).
[14] W. Berm&her,
A. Brandenberg, and P. Uwer, Aachen
Report No. PITHA 95/26, 1995 (unpublished).