Fa oltà diS ienze Matemati he, Fisi hee Naturali
Dottorato di Ri er a in Fisi a
Study of double harmonium
produ tion around
√
s
= 10.58 GeVwith the BaBar dete tor
Dr. Annalisa Ce hi
Advisor:
Prof. Diego Bettoni
Coordinator:
Prof. Filippo Frontera
Referees:
Prof. Antimo Palano
Prof. Dr. Klaus Peters
Introdu tion 5
1 Charmonium physi s 7
1.1 Potential models . . . 7
1.2 Quantum Chromodynami s . . . 12
1.3 Ee tive FieldTheories . . . 14
1.4 Nonrelativisti QCD . . . 16
1.4.1 The NRQCDlagrangian . . . 16
1.5 Experimental studyof harmonium . . . 19
1.5.1 Quarkoniumprodu tion . . . 19
1.5.2 Charmonium spe trum . . . 20
1.5.3 New harmonia . . . 22
2 Double harmonium produ tion 25 2.1 Introdu tion . . . 25
2.2 Cross se tion . . . 26
2.2.1 Color-singletmodel al ulation . . . 27
2.2.2 Cal ulationof the ross se tions . . . 29
2.2.3 Cross se tion for the produ tion of pseudos alar and ve tor doubleheavymesons . . . 32
2.2.4 Light one formalism . . . 33
2.3 Dis repan ybetween theory and experiment . . . 36
2.4 Previous resultson
e
+
e
−
→ J/ψ + X
. . . 362.4.1 X(3940) interpretation . . . 38
2.5 Double
cc
produ tion viaγ
∗
γ
∗
. . . 383 The BABAR experiment 41 3.1 The PEP-II asymmetri ollider . . . 42
3.1.1 Luminosity . . . 44
3.1.2 Ma hine ba kground . . . 45
3.3 Tra kingSystem . . . 48
3.3.1 Sili onVertexDete tor . . . 48
3.3.2 Drift Chamber . . . 48
3.4 Cherenkov dete tor . . . 52
3.5 Ele tromagneti alorimeter . . . 56
3.6 Instrumented uxreturn . . . 59
3.7 Trigger,Data A quisitionand Re onstru tion . . . 65
4 Analysis 67 4.1 Analysis strategy . . . 67
4.2 Data samples and presele tion . . . 67
4.2.1 Datapro essing inBABAR . . . 67
4.2.2 Dataand Monte arlo samples . . . 68
4.2.3 Ba kground . . . 69
4.3 Sele tion strategy . . . 70
4.4 Presele tion . . . 70
4.5 Sele tion uts . . . 72
4.5.1 Numberof hargedtra ks . . . 72
4.5.2 Momentum of the
J/ψ
inthe CM frame . . . 734.6 Multivariate Analysis(MVA) . . . 75 4.6.1 Prin ipleof the MVA . . . 75 4.6.2 Appli ation . . . 76 4.6.3 Cal ulationof
N
S
exp
andN
B
exp
. . . 81 4.6.4 Optimization onsigni an e . . . 84 4.6.5 Summaryonsele tion . . . 85 4.7 Fit to data. . . 874.7.1 Fit and toys validation on the signal MC . . . 87
4.7.2 Fit and toys validation on the ba kground . . . 94
4.7.3 Globalt with signal and ba kgroundembedded . . . . 96
4.7.4 Simultaneous t . . . 104
5 Results 109 5.1 Unblind up to 3.8 GeV . . . 109
5.2 Outlook . . . 110
The goal of high energy physi s is to identify the elementary onstituents
of matter and to understand their fundamental intera tions. Over the last
twentyyears,thisendeavorhas beenextraordinarilysu essful. Agauge
the-ory alled Standard Model provides a satisfa tory des ription of the strong,
weak,andele tromagneti intera tionsofalltheknownelementaryparti les.
There are very few dis repan ies between theory and experiment, and most
of them are atthe level of a few standard deviationsor less. However there
are pro esses for whi h experimental results have diered from theoreti al
predi tions by orders of magnitude: soe of these studies are related to the
produ tionof harmonium. This dramati oni t between experimentand
theory presents a unique opportunity to make a signi ant step forward in
our understanding of heavy quarkonium physi s.
Quarkonia play an important role in several high energy experiments.
The diversity, quantity and a ura y of the data still under analysis and
urrently being olle ted inmanyhigh energyexperimentsaroundtheworld
is impressive.
These data ome from experiments of quarkonium formation (BES at
the Beijing Ele tron Positron Collider, E835 at Fermilab, and CLEO atthe
Cornell Ele tronStorage Ring), lean samples of harmonia produ ed in
B-de ays,inphoton-photonfusionandininitialstateradiation,attheB-meson
fa tories (BaBar at PEP-II and Belle at KEKB), in luding the unexpe ted
observation of large amounts of asso iated
(c¯
c)(c¯
c)
produ tion and theob-servation of new and possibly exoti s quarkonia states. The CDF and D0
experiments at Fermilab measuring heavy quarkonia prodution from
gluon-gluonfusion in
p¯
p
annihilationsat2TeV; ZEUSandH1,atDESY, studyingharmoniaprodu tioninphoton-gluonfusion;PHENIXandSTAR,atRHIC,
and NA60, at CERN, studying harmonia produ tion, and suppression, in
heavy-ion ollisions [7℄.
This has led tothe dis overy of newstates, new produ tionme hanisms,
newde ays andtransitions, andingeneraltothe olle tionof highstatisti s
expe ted from the BES-III upgraded experiment, while the B fa tories and
the Fermilab Tevatron will ontinue to supply valuable data for few years.
Lateron,newexperimentsatnewfa ilitieswillbe omeoperational(theLHC
experimentsatCERN,PANDAatGSI,hopefullyaSuper-Bfa tory,aLinear
Collider, et .) oering fantasti hallenges and opportunities inthis eld.
Inthisthesistheanalysisondouble harmoniumprodu tionattheenergy
ofthe
Υ
(4S),withtheBABARdataisdo umented. Theaimofthisanalysisistounderstandthe me hanismofprodu tionofdouble harmoniumstates fro
e
+
e
−
annihilation,inparti ularafterthedis repan ieswhi hatthebeginning
of these studiesappeared.
With su essive studies, these dis repan ies have been almost solved.
This analysis was already performed by BABAR ollaboration [28℄, and in
this thesis wewant toupdate that work, with a luminosity early fourtimes
higher (468
f b
−1
. In the mean time, also Belle ollaboration published on
this analysis [16 ℄, obtaining results ompatible with BABAR and ndingout
anew harmonium state, namedX(3940). Weaim here also to onrmthis
state.
This thesis is omposed by ve hapters. Chapter I is an introdu tion
to harmonium spe tros opy, with a des ription of the NRQCD, whi h the
theorethi al framework of this analysi, then the potential models that have
been developed todes ribe the mass spe trum.
The theories related to the double harmonium produ tion me hanisms
are presented inChapter II:in parti ularthe al ulationof the ross se tion
and the di repan ies between theory and experiment.
Then inChapter III the BABAR dete tor isdes ribed.
In Chapter IV there is a des ription on how has been performed the
analysis: theanalysisstrategy,sele tionand utoptimization,andvalidation
of the tare do umented, before the unblind of the interested region.
Charmonium physi s
Until 1974 all the known hadrons were omposed by three quark avours:
theup (
u
),down (d
)andstrange (s
). Themassesof thesestateswere rathersmall: afew MeV for
u
andd
, and 100-200 MeV fors
.InNovemberof1974,aremarkablymassiveandnarrowresonan e,named
J, was dis overed [1℄ with a mass of 3.1
GeV/c
2
, de aying toe
+
e
−
, in the rea tionp + Be → e
+
e
−
+ X
. Simultaneously,the resonan ewasdis overed
[2℄ in the dire t hannel
e
+
e
−
→ hadrons
(also to
e
+
e
−
, µ
+
µ
−
), and was
named the
ψ
. The dual nameJ/ψ
has afterwards persisted.Withthedis overyofthe
J/ψ
,the existen eofanewquarkavour alledharm (
c
), with a mass of the order of 1 GeV, as well as the existen e of afamily of states alled harmonia wasdemonstrated.
The
J/ψ
is a member of this family, that is omposed by the boundstates of harm quark and antiquark (
cc
). The harmonium is the mostwidely studied heavy quarkonium system, and the goal of this hapter is to
givethe theoreti altoolsne essarytofa ethe quarkonium,andinparti ular
harmonium, physi s.
1.1 Potential models
Whentwoparti lesformaboundstate,the attra tivepotential an be
stud-ied measuring the energy spe trum of the system. In atomi physi s, the
binding energy of the ele tron-nu leus system depends on the orbital
angu-lar momentum(
L
),spin(S
)and totalangularmomentum(J = L + S
)state(negle ting the nu leus angularmomentum
I
). To lassify the energy levelsof the system the spe tros opi notation
n
2S+1
L
J
is used. A similarpattern of energy levels is present in positronium (thee
+
e
−
bound state); this has
Thesame on ept anbeappliedalsotothemesons,whi harethe
quark-antiquark (
n
2S+1
L
J
forthe lassi ationof the mesonsis used.The intrinsi parity
P
and harge onjugationC
of a harmonium stateare relatedto the angular momentum by the relations:
P = (−1)
L+1
,
C = (−1)
L+S
.
Also the
J
P C
notation an be used to lassify the
cc
states.Quantum Chromodynami s (QCD) is the modern theory of the strong
intera tions. The non perturbative features of QCD prevent the possibility
of des ribing it on the basis of the fundamental theory of the intera tion.
Forthis reasonthenaturalapproa hto harmoniumspe tros opyistobuild
an ee tive potential model des ribing the observed mass spe trum. This
approximation allows to integrate out many fundamental ee ts like gluon
emissionorlightquark pairs andto dealwith anee tive potential whi h is
theresult ofthe
This potential should nevertheless reprodu e the two main features of the
boundquarkstates inthetwolimitsof smalland largedistan e: asymptoti
freedom and onnement.
The
cc
system an bedes ribedwith a S hroedingerequation:HΨ(x) = EΨ(x),
(1.1)where the hamiltonian forthe
cc
system an be writtenas:H = H
0
+ H
′
.
(1.2)H
0
an be expressedasa freeparti lehamiltonian plusa non-relativisti potentialV (r)
:H
0
= 2m
c
+
p
2
m
c
+ V (r),
(1.3)where
m
c
is the harm quark mass andp
itsmomentum.V (r)
an bebuilt takingintoa ountthe properties ofstrongintera tionin the limit of small and large distan es. At small distan e the potential
between the quarks for a quark-antiquark pair bound in a olor singlet, is
oulomb-like:
V (r) ∼
4
3
α
s
r
(r)
,
where
r
is the distan e between the quarks,α
s
is the strong oupling onstant and the fa tor 4/3 omes from the group theory of SU(3), related0
0.1
0.2
0.3
1
10
10
2
µ GeV
α
s
(µ
)
Figure 9.2:
Summary of the values of
( ) at the values of
where they are
Figure 1.1: Summary of the values of
α
s
at the values ofµ
where they are measured[5℄.The value of the running oupling onstant
α
s
depends on the energy s ale of the intera tion in the way shown in Fig. 1.1 , where is lear thede rease of
α
s
with in reasingµ
. At the leading order in the inverse power ofln(µ
2
/Λ
2
)
,α
s
isdes ribedby:α
s
(µ) =
4π
β
0
ln(µ
2
/Λ
2
)
,
(1.4)β
0
= 11 −
2
3
n
f
where
Λ ≃ 0.2 GeV
is the non-perturbative s ale of QCD (the energywhere(1.4)diverges)and
n
f
isthenumberofquarkslighter thanthe energy s ale . It is lear from equation 1.4 that, as the energy s ale of a strongpro ess de reases and be omes loser to
Λ
,α
s
in reases and the QCD an not be treated asa perturbative theory.As a result of (1.4) the oupling
α
s
(µ)
varies logarithmi ally withµ
, so thatatveryshortdistan es,gluonex hangebe omesweaker. This property,knownasasymptoti freedom,isresponsibleforthequasi-freebehavior
exhib-ited by quarks inhadronsprobedat very short distan es by deeply inelasti
s attering.
At large distan e, that means at momentum s ales smaller than
Λ ≃
200M eV
the onnement termisdominating. It an bewrittenintheform:V (r) ∼ kr,
"#$%&' ()* !"#"!$%&'()'*+!",')-"../'+"-+&$.
+,'-.&#/%01'&20.&'3'0 '3#61,'7/88/9#6$*
( !"#$%&'"()1,'0:#6;/&5#11'&20:8#101,'01.1'09#1,1,'0.2'/&5#1.8.6$%;
8.&2/2'61%23':'63#6$/61,' !!'<:' 1.1#/6-.8%'=>6'01&% 1%&'?* " ! @= !? ) #" " #$ #" ! #$ ! %=A& $?
9,'&' "! .63 "" .&' 1,' 0 .8.& .63 1,' -' 1/& /2:/6'610 /7 1,' 6/6;
&'8.1#-#01# :/1'61#.8"=$?(
Figure1.2: PlotoftheQCDpotential(1.1),forquark-gluon oupling
α
s
= 0.20
and
k = 1 Gev/f m
in nature is explained exa tly by the onnement term, be ause it implies
that the energy of a
By putting together these two behaviors, one an write the Cornell
po-tential, shown inFig. 1.2 [3℄:
V
(r) ∼ −
4
3
α
s
(r)
r
+ kr,
(1.5)Withthis potential, the harmonium wave fun tion an beexpressed as:
Ψ(r, θ, φ) = R
nl
(r)Y
l
m
(θ, φ).
(1.6)This des ription, however, is not enough to reprodu e the mass
dier-en e for harmonium states in the same orbital angular momentum or spin
multiplets.
H
′
inthe equation (1.2 ) in ludesthe spin (S
) and orbital(L
) dependent partof thestrongintera tion, explainingthe harmoniumneand hypernestru ture [4℄:
H
′
= V
LS
+ V
SS
+ V
tens
.
(1.7) The various terms of intera tion are des ribedin the following: spin-orbit (
V
LS
): spin-orbit for es between quarks are present for both ve tor and s alar intera tions, but in dierent form. We nd forquarks ofequal mass
m
c
:V
LS
= (L · S)(3
dV
V
dr
−
dV
S
dr
)/(2m
2
c
r)
(1.8)where
V
S
andV
V
are the s alar and ve tor omponents of the non-relativisti potential V(r). This term splits the states with the sameorbitalangularmomentum dependingonthe
(L · S)
expe tation value(ne stru ture);
spin-spin (
V
SS
): the hyperne ele tromagneti intera tion between a proton and an ele tron leads to a 1420 MHz level splitting betweensinglet and triplet states of atomi hydrogen. In light-quark systems,
a similar spin-spin for e due to single-gluon-ex hange between quarks
generates the splittings between the masses of the pion and the
ρ
res-onan e, the nu leon and the
∆
resonan e, theΣ
and theΛ
hyperons,and so on. The spin-spin intera tion is of the form:
V
SS
=
2(S
1
· S
2
)
3m
2
c
∇
2
V
V
(r)
(1.9)and the expe tation valuefor
S
1
· S
2
is +1/4forS
=1and -3/4 forS
=0; tensor (
V
T
): the tensor potential, in analogy with ele trodynami s, ontains the tensor ee ts of the ve tor potential:V
T
=
S
12
12m
2
c
(
1
3
dV
V
dr
−
d
2
V
V
dr
2
),
S
12
= 2 [3(S · ˆr)(S · ˆr) − S
2
].
where
S
12
has nonzero matrix elementsonly forL 6= 0
.Evenifthe QCDtheory allowstodes ribeonthebasis thefoundamental
theoryofthe intera tions,asexplainedinthenextse tion,othersuggestions
for the fun tional form of the binding potential V(r) exist, but they are
essentially oin identwith thevaluesfrom(1.1)inthe regionfrom0.1to1.0
fm, the dimension s ale of the
cc
system, and lead to similar results.Another possibility topredi tthe harmoniummass spe trumis to
on a dis rete Eu lidean spa e-time grid. Indeed, QCD has been very
su - essful inpredi ting phenomena involving largemomentum transfer. In this
regime the oupling onstant is small and perturbation theory be omes a
reliabletool.
On the other hand, at the s ale of the hadroni world,
µ
≤ 1GeV
, theoupling onstant is of order unity and perturbative methods fail. In this
domain latti e QCD provides a non-perturbative tool for al ulating the
hadroni spe trum and the matrix elements of any operator within these
hadroni states from rst prin iples. Sin e no new parameters oreld
vari-ables are introdu ed in this dis retization, LQCD retains the fundamental
hara ter of QCD.
The eld theory fundamental prin iples and the path integral an be
used to al ulate on a omputer the properties of the strong intera tion,
with Monte Carlo integration of the Eu lidean path integral. The value of
thelatti espa ing,usually denotedwith
a
, an bede ideddependingonthespe i problem that has to be solved.
The only tunable input parameters in these simulations are the strong
oupling onstantandthe baremassesofthequarks. Our beliefisthatthese
parametersare pres ribedby someyetmorefundamental underlying theory,
however,within the ontext of the standard model they have to be xed in
terms of anequal numberof experimental quantities.
1.2 Quantum Chromodynami s
Quantum Chromodynami sisa quantumeld theory obtained fromthe full
StandardModel(SM)bysettingthe weakandele tromagneti oupling
on-stantstozeroandfreezingthes alardoublettoitsva uumexpe tationvalue.
What remains is a Yang-Mills (YM) theory with lo al gauge group SU(3)
( olour) ve torially oupled to six Dira elds (quarks) of dierent masses
(avours). The ve tor elds in the YM Lagrangian (gluons) live in the
ad-joint representation and transform like onne tions under the lo al gauge
group whereas the quark elds live in the fundamental representation and
transform ovariantly. The QCD Lagrangian is
L
QCD
= −
1
4
F
a
µν
F
a µν
+
X
{q}
¯
q(iγ
µ
D
µ
− m
q
)q
(1.10)where
{q} = u, d, s, c, b, t,
F
µν
a
= ∂
µ
A
a
ν
− ∂
ν
A
a
µ
+ gf
abc
A
b
µ
A
c
ν
,
D
µ
= ∂
µ
− iT
a
A
a
µ
and
f
abc
aretheSU(3)stru ture onstantsandT
a
formabasisofthe
founda-mental representation of the SU(3) algebra. When oupled to
ele tromag-netism, gluons behave as neutral parti les whereas
u
,c
andt
quarks haveharges +2/3 and
d
,s
andb
quarks have harges -1/3.The mainproperties ofQCD,whi hhavebeenpartiallyillustrated inthe
previous se tions,are the following:
It is Poin aré, parity, time reversal and hen e harge onjugation
in-variant. ItisinadditioninvariantunderU(1)
6
whi himpliesindividual
avour onservation.
Being a non-Abelian gauge theory, the physi al spe trum onsists of
olour singlet states only. The simplest of these states have the
quan-tum numbers of quark-antiquark pairs (mesons) or of three quarks
(baryons) although other possibilities are not ex luded.
TheQCDee tive oupling onstant
α
s
(q)
de reasesasthemomentum transfer s aleq
in reases (asymptoti freedom) [8 , 9℄, as also alreadyexplainedinse tion1.1. Thisallowstomakeperturbative al ulations
in
α
s
athigh energies. At low energies it develops an intrinsi s ale (mass gap), usually
re-ferred as
Λ
QCD
, whi h provides themain ontributiontothe masses of most light hadrons. At s alesq ∼ Λ
QCD
, α
s
(q) ∼ 1
and perturbation theory annot be used. Investigations must be arried out usingnon-perturbative te hniques, the bestestablished of whi h islatti e QCD.
Quarks are onventionally divided intolight
m
q
l
+
l
−
Λ
QCD
(q = u, d, s)
and heavy
m
Q
≫ Λ
QCD
(Q = c, b, t)
: 1m
u
= 1.5 − 3.3 MeV, m
u
= 3.5 − 6.0 MeV, m
s
= 70 − 130 MeV,
m
c
= 1.27
+0.07
−0.11
M eV, m
b
= 4.20
+0.17
−0.07
GeV, m
t
= 171.2 ± 2.1 GeV
Iflightquarkmassesarenegle ted,theU(1)
3
avour onservation
sym-metryoftheQCDLagrangianinthisse torisenlargedtoaU(3)
N
U(3)
1
group. The axial U(1) subgroup is expli itly broken by quantum
ef-fe ts(axial anomaly). Theve tor U(1)subgroupprovideslightavour
onservation. The remainingSU(3)
N
SU(3) subgroup,knownas hiral
symmetry group, turns out to be spontaneously broken down to the
diagonal SU(3) (avour symmetry). This produ es eight Goldstone
bosons, whi h, upon taking into a ount the expli it breaking of the
symmetry due to the non-zero quark masses, a quire masses that are
mu hsmaller than
Λ
QCD
. Hadrons ontainingheavyquarkshavemassesoftheorderof
m
Q
rather than of the orderΛ
QCD
. They enjoy parti ular kinemati al features that allow for spe i theoreti al treatments.1.3 Ee tive Field Theories
From the point of view of QCD the des ription of hadrons ontaining two
heavy quarks is a rather hallenging problem, whi h adds to the
ompli a-tionsof thebound stateineld theorythose omingfromanonperturbative
low-energy dynami s. A proper relativisti treatment of the bound state
basedon the Bethe-Salpeter equation [10 ℄ has proveddi ult. Perturbative
al ulationshaveturnedout unpra ti alathigherorderandthe methodhas
been abandonedinre entQCD al ulations. Moreover, theentanglementof
allenergymodes inafullyrelativisti treatmentismoreanobsta lethanan
advantageforthefa torizationofphysi alquantitiesintohigh-energy
pertur-bativeandlowenergynonperturbative ontributions. Partialsemirelativisti
redu tions and models have been often adopted to over ome these
di ul-ties at the pri e to introdu e un ontrolled approximations and lose onta t
withQCD.Thefully relativisti dynami s an, inprin iple,betreated
with-out approximations inlatti e gaugetheories. This isin perspe tive the best
founded and most promising approa h,as already said inse tion 1.1 .
A nonrelativisti treatment of the heavyquarkoniumdynami s, whi h is
suggested by the large mass of the heavyquarks, has lear advantages. The
velo ityofthequarksintheboundstateprovidesasmallparameterinwhi h
thedynami als alesmaybehierar hi allyorderedandthe QCDamplitudes
systemati ally expanded. Fa torization formulas be ome easier to a hieve.
Aprioriwedonotknowifanonrelativisti des riptionwillworkwellenough
for all heavy quarkonium systems in nature: for instan e, the harm quark
may not be heavyenough. The fa t that most of the theoreti al predi tions
are based on su h a nonrelativisti assumption and the su ess of most of
We may, however, also take advantage of the existen e of a hierar hy
of s ales by substituting QCD with simpler but equivalent Ee tive Field
Theories (EFTs). EFTshavebe omein reasinglypopularinparti lephysi s
during the last de ades.
They provide a realization of Wilson renormalization group ideas [11℄
and fullyexploit the properties oflo alquantum eld theories. AnEFT isa
quantum eld theory with the following properties:
a) it ontainsthe relevantdegreesoffreedom todes ribephenomenathat
o ur in ertainlimited rangeof energies and momenta;
b) it ontainsanintrinsi energys ale
Λ
thatsetsthelimitofappli abilityof the EFT.
TheLagrangianofanEFTisorganizedinoperatorsofin reasingdimension,
hen e, an EFT is in general non-renormalizable in the usual sense. In spite
of this, it an be made nite to any nite order in
1/Λ
by renormalizing(mat hing) the onstants (mat hing oe ients)infrontof the operatorsin
the Lagrangian untilthat order. This meansthat one needs more
renormal-ization onditions when the order in
1/Λ
is in reased. However, even if theonly way of xing the onstants would be by means of experimental data,
this would redu ebut not spoilthe predi tive powerof the EFT.If the data
are abundant, the onstants an be t on e for ever and used later on to
make predi tions onnew experiments.
The prototype of EFT for heavy quarks is the Heavy Quark Ee tive
Theory (HQET), whi h is the EFT of QCD suitable to des ribe systems
with onlyone heavyquark [12 , 13 ℄. These systems are hara terizedby two
energy s ales:
m
andΛ
QCD
. HQET is obtained by integrating out the s alem
and built as asystemati expansionin powersofΛ
QCD
/m
.As dis ussedabove, bound states made of two heavy quarks are
hara -terized by more s ales. Integrating out only the s ale
m
, whi h for heavyquarks an be done perturbatively, leads to an EFT, Nonrelativisti QCD
(NRQCD) [14, 15℄, that still ontains the lower s ales as dynami al degrees
of freedom. Disentangling the remaining s ales is relevant both te hni ally,
sin e it enables perturbative al ulations otherwise quite ompli ate, and
more fundamentally, sin e it allows to fa torize nonperturbative
ontribu-tions intothe expe tation values ormatrixelements offewoperators. These
may be eventually evaluated on the latti e, extra ted from the data or
al- ulatedin QCDva uummodels.
In the next se tion we will give a briefgeneral introdu tion to NRQCD,
1.4 Nonrelativisti QCD
Aparti ularlyelegantapproa hforseparatingrelativisti fromnonrelativisti
s alesistore asttheanalysisintermsofnonrelativisti quantum
hromody-nami s(NRQCD)[15℄,anee tiveeldtheorydesignedpre iselytoseparate
therelativisti physi sofannihilation(whi hinvolvesmomenta
p ∼ M
)fromthenonrelativisti physi sofquarkoniumstru ture(whi hinvolves
p ∼ Mv
).NRQCD onsists of a nonrelativisti S hroedinger eld theory for the
heavyquarkand antiquarkthat is oupled tothe usualrelativisti eld
the-ory for light quarks and gluons. The theory is made pre isely equivalent to
full QCD through the addition of lo al intera tions that systemati ally
in- orporaterelativisti orre tionsthrough anygivenorderintheheavy-quark
velo ity
v
. It is an ee tive eld theory, with a nite ultraviolet uto oforder
M
thatex ludesrelativisti states (statesthatare poorlydes ribedbynonrelativisti dynami s). A heavy quark in the meson an u tuate into
a relativisti state, but these u tuations are ne essarily short-lived. This
meansthat the ee tsof the ex ludedrelativisti states anbemimi kedby
lo al intera tions and an, therefore,be in orporated into NRQCD through
renormalizations of its innitely many oupling onstants. Thus,
nonrela-tivisti physi s is orre tly des ribed by the nonperturbative dynami s of
NRQCD, while all relativisti ee ts are absorbed into oupling onstants
that an be omputed asperturbationseries in
α
s
(M).The main advantage oered by NRQCD overordinary QCD inthis
on-text is that it is easier to separate ontributions of dierent orders in
v
inNRQCD. Thus, we are able not only to organize al ulations to all orders
in
α
s
, but also to elaborate systemati ally the relativisti orre tions to the onventional formulas.1.4.1 The NRQCD lagrangian
Themost importantenergy s alesforthe stru ture andspe trumofaheavy
quarkonium system are
M v
andM v
2
, where
M
is the mass of the heavyquark
Q
andvl
+
l
−
1
isitsaveragevelo ityinthemesonrestframe. Momenta
of order
M
play only a minor role in the omplex binding dynami s of thesystem. We an take advantage of this fa t in our analysis of heavyquark
mesonsby modifyingQCD intwosteps.
We start with full QCD, in whi h the heavy quarks are des ribed by
4- omponent Dira spinor elds. In the rst step, we introdu e an ultraviolet
momentum uto that is of order
M
. This uto expli itly ex ludesrela-tivisti heavyquarksfromthetheory,aswellasgluonsandlightquarkswith
sin e the importantnonperturbative physi sinvolvesmomenta of order
M v
orless. Of ourse,therelativisti stateswearedis ardingdohavesomeee t
on the low energy physi s of the theory. However, any intera tion involving
relativisti intermediate states is approximately lo al, sin e the
intermedi-ate states are ne essarily highly virtual and so annot propagate over long
distan es. Thus, generalizing standard renormalization pro edures, we
sys-temati ally ompensate for the removal of relativisti states by adding new
lo al intera tionstothelagrangian. Toleadingorderin
1/Λ
or, equivalently,1/M
, these new intera tions are identi al in form to intera tions already present in the theory, and so the net ee t is simply to shift bare massesand harges. Beyond leading order in
1/M
, one mustextend the lagrangianto in lude nonrenormalizable intera tions that orre t the low energy
dy-nami sorder-by-order in
1/M
. In this uto formulationof QCD,allee tsthat arise from relativisti states, and only these ee ts, are in orporated
into renormalizations of the oupling onstants of the extended lagrangian.
Thus, in the uto theory, relativisti and nonrelativisti ontributions are
automati ally separated. This separation is the basis for an analysis of the
annihilationde ays of heavyquarkonia.
The lagrangian for NRQCDis:
L
N RQCD
= L
light
+ L
heavy
+ δL
(1.11)The gluons and the
n
f
avors of light quarks are des ribed by the fully relativisti lagrangianL
light
= −
1
2
trG
µν
G
µν
+
X
¯
qiDq
(1.12)where
G
µ
is the gluon eld-strength tensor expressed in the form of an SU(3) matrix, andq
is the Dira spinor eld for a light quark. Thegauge- ovariant derivative is
D
µ
+ igA
µ
, where
A
µ
= (φ, A)
is the SU(3)
matrix-valued gaugeeldand
g
is theQCD oupling onstant. Thesum in(1.12)isoverthe
n
f
avorsof light quarks.The heavy quarksand antiquarks are des ribed by the term
L
heavy
= ψ
†
iD
t
+
D
2
2M
ψ + χ
†
iD
t
−
D
2
2M
χ,
(1.13)where
ψ
is the Pauli spinor eld that annihilates a heavy quark,χ
isthe Pauli spinor eld that reates a heavy antiquark, and
D
t
andD
are the time and spa e omponentsof the gauge- ovariant derivativeD
µ
. Color
L
light
+L
heavy
des ribesordinaryQCD oupledtoaS hroedingereldtheory for the heavyquarks and antiquarks.Therelativisti ee tsoffullQCDarereprodu edthroughthe orre tion
term
δL
in the lagrangianL
N RQCD
[14℄.In parti ular the orre tionterms most important forheavyquarkonium
are bilinearin the quark eld orantiquark eld:
L
bilinear
=
c
1
8M
3
ψ
†
(D
2
)
2
ψ − χ
†
(D
2
)
2
χ
+
c
2
8M
2
ψ
†
(D · gE − gE · D)ψ − χ
†
(D · gE − gE · D)χ
+
c
3
8M
2
ψ
†
(iD × gE − gE × iD)ψ − χ
†
(iD × gE − gE × iD)χ
+
c
4
2M
ψ
†
(gB · σ)ψ − χ
†
(gB · σ)χ
,
(1.14) whereE
i
= G
0i
andB
i
=
1
2
ǫ
ijk
G
jk
arethe ele tri andmagneti omponents
of the gluoneld strength tensor
G
µν
. By harge onjugation symmetry, for
every termin(1.14)involving
ψ
,there isa orrespondingterm involving theantiquarkeld
χ
,withthe same oe ientc
i
,up toasign. Theoperatorsin (1.14 )mustberegularized,andtheythereforedependontheultraviolet utoorrenormalizations ale
Λ
of NRQCD.The oe ientsc
i
(Λ)
alsodepend onΛ
insu h away as to an el theΛ
-dependen e of the operators.Noti e that
L
bilinear
doesn't ontain mixed two-fermionoperators involv-ingχ
†
and
ψ
(orψ
†
and
χ
), orrespondingtotheannihilation(orthe reation)of a
Q ¯
Q
pair. Indeed su h terms are ex luded from the lagrangian as partof the denition of NRQCD: if su h an operator annihilates a
Q ¯
Q
pair, itwould, by energy onservation, have to reate gluons (or light quarks) with
energiesof order
M
. The amplitude forannihilation of aQ ¯
Q
pair into su hhighenergygluons annotbedes ribeda uratelyinanonrelativisti theory
su has NRQCD.
The oe ients
c
i
must be tuned as fun tions of the oupling onstantα
s
, the heavy-quark mass parameter infull QCD, and the ultraviolet utoΛ
of NRQCD, sothat physi al observables are the same asin fullQCD. Inprin iple,innitelymanytermsarerequiredintheNRQCDlagrangianinordertoreprodu e fullQCD,but inpra ti e onlyanite numberofthese
isneededforpre ision toanygivenorder inthetypi alheavy-quark velo ity
1.5 Experimental study of harmonium
1.5.1 Quarkonium produ tion
Quarkonia an be produ ed in several ways, whi h rea h dierent states
within the spe trum. The rst three listed here are mere reversals of
Q ¯
Q
de ay pro esses and are sket hed inFig. 1.3 a), b), and ).
In ele tron-positron olliders, the rea tion results in
e
+
e
−
→ γ
∗
→ Q ¯
Q
states that an ouple to a virtual photon, namely
n
3
S
1
su h asJ/ψ
andΥ
with a tiny admixture ofn
3
D
1
. Dire t resonan e formation oers the advantageoflargeprodu tionrates,givinga esstobran hingfra tionsevenas small as
10
−5
, as well as higher a ura y in the measurements of masses
and widths..
Two-photon ollisions allow dire t reation of
J = 0, 2
states, e.g.η
[c,b]
,χ
[c,b][0,2]
. While theyare readilyavailableate
+
e
−
ma hines,theysuerfrom
small produ tionrates. Still theyprovidean important ontribution in that
they an be used fordis overy purposes.
Hadron ma hines, being able to form any quarkonia state in prin iple
by annihilation of the
p¯
p
pair into gluons, ontinue to ontribute mostly tothe study of produ tion of harmonia. This environment suers from large
ba kground;thereby one has to fo us onex lusive de ays.
Twomore s enarios: downward transitions within the system providean
importantroute tootherwise not rea hable states. AnyB-fa tory has a ess
to
cc
states through weak de ays of theb
quark, with the two pro essessket hed inFig. 1.3 d) and e).
An important ba kground for the rea tion
e
+
e
−
→ Q ¯
Q → X
, or more
expli itly,
e
+
e
−
→ γ
∗
→ Q ¯
Q → X
, is the ase in whi h no intermediate
Q ¯
Q
resonan e isformed. The presen e ofthis hanneladdsto themeasured ross-se tion both dire tly and by interferen e, whi h an bea sizeableon-tribution. Inmostmeasurements,this ontribution isnottakenintoa ount
orsubtra ted. This ba kgroundneeds tobe either measured,by running o
the relevant resonan e, or al ulated. In measurements of the ross-se tion
as fun tionof energy (s ans),the non-resonant produ tion an be expli itly
taken intoa ount when tting the line shape.
In addition tothat, the produ tion of double harmonium in
e
+
e
−
anni-hilations has re ently been observed at the B fa tories. The produ tion of
double harmoniumin
e
+
e
−
annihilationwasdis overedbytheBelle
ollabo-rationfromdata olle tedatthe
Υ (4S)
resonan e ata enter-of-massenergys =
p(10.6)GeV
by studying the re oil momentum spe trum of theJ/ψ
ine
+
e
−
→ J/ψ
+ X [16 ℄. The measured ross se tion for double harmonium
predi tion of NRQCD in the non-relativisti limit. This large dis repan y
was rather puzzling. This way to produ e harmonium is the main topi of
thisthesis,andthenext hapterwillbededi atedtothedouble harmonium
produ tionphysi s.
1.5.2 Charmonium spe trum
The spe trum of harmonium states is shown in Fig. 1.4 . The potential
model, des ribed in se tion 1.1 , an explain with the spin-spin intera tion
term (
V
SS
) the splitting among spin singlet and triplet states likeJ/ψ
andη
c
, and with the spin-orbit intera tion (V
LS
) the splitting among states likeχ
c 0,1,2
.The harmoniumspe trum onsistsofeightnarrowstatesbelowtheopen
harm threshold (3.73 GeV) and several tens of states abovethe threshold.
All eight states below
D ¯
D
threshold are wellestablished, but whereasthetriplet statesare measuredwith verygooda ura y,the same annotbe
said for the singlet states.
The
η
c
wasdis overedalmostthirtyyearsagoandmanymeasurementsof itsmass andtotal widthexist. Despite the largevariety ofavailabledata onit,the pre isedetermination of its mass and width is still an openproblem.
TheParti leDataGroup(PDG)[5℄valueofthemassis2980.3
±
1.2MeV/c
2
:
the erroronthe
η
c
mass isstill aslarge as1.2 MeV/c
2
, tobe omparedwith
few tens of KeV/
c
2
for the
J/ψ
andψ
′
and few hundreds of KeV/
c
2
for the
χ
c 0,1,2
. The situation is even worse for the total width: the PDG averageF
F
4
4
4
4
H
H
H
H
H
H
4
4
4
4
H
H
( )
4
4
S
S
4
4
S
S
E
T
( )
4
4
V
T
:
( )
a)
b)
c)
d)
e)
–
Figure1.3: Heavyquarkoniaprodu tiondiagrams. Produ tion(left)andtheir
orrespondingde ay (right)pro esses: a)
e
+
e
−
→ γ
∗
→ Q ¯
Q
; b)
γγ → Q ¯
Q
; )p¯
p →
gluons→ Q ¯
Q
; d) Quarkoniumde-ex itation byemission oftwo pions; e) reating harmoniumfroma Bmeson.Figure 1.4: The harmoniumspe trum.
is 26.7
±
3.0 MeV. The most re ent measurements have shown that theη
c
width is larger than was previously believed, with values whi h are di ultto a omodate in quark models. This situation points to the need for new
high-pre ision measurements of the
η
c
parameters.Therstexperimentaleviden eofthe
η
c
(2S)
wasreportedbythe Crystal Ball Collaboration [17℄, but this nding was not onrmed in subsequentsear hes in
p¯
p
ore
+
e
−
experiments. The
η
c
(2S) was nally dis overed by the Belle ollaboration [18 ℄ in the hadroni de ay of the B mesonB →
K + η
c
(2s) → K + (K
s
K
−
π
+
)
with amass whi hwasin ompatiblewiththeCrystal Ball andidate. The Bellendingwasthen onrmed by CLEOand
BaBar [19, 20℄, whi h observed this state in twophoton fusion. The PDG
value of the mass is 3637
±
4 MeV/c
2
, and the width isonly measured with
ana ura y of 50%: 14
±
7 MeV/c
2
.
The
1
P
1
stateof harmonium(h
c
)isofparti ularimportan einthe deter-mination of the spin-dependent omponentof theThe
h
c
has been observed by CLEO [21℄ in the rea tionψ(2S) → π
0
h
c
→
(γγ)(γη
c
)
with amass of3524.4±
0.6±
0.4 MeV/c
2
than 5
σ
.Parti le
n
2S+1
L
J
J
P C
Mass(MeV) Width(MeV)η
c
1
1
S
0
0
−+
2980.3±
1.2 26.7±
3.0J/ψ
1
3
S
1
1
−−
3096.916±
0.011 (93.2±
2.1)×
10−3
χ
c0
1
3
P
0
0
++
3414.75±
0.31 10.2±
0.7χ
c1
1
3
P
1
1
++
3510.66±
0.07 0.89±
0.05χ
c2
1
1
P
2
2
++
3556.20±
0.09 2.03±
0.12η
c
(2S)
2
1
S
0
0
−+
3637±
4 14±
7ψ(2S)
2
3
S
1
1
−−
3686.09±
0.04 0.317±
0.009Table1.1: Quantumnumbers,massesandwidthofthe harmoniumstateswith
massbelowtheopen harmprodu tionthresholdfrom PDG.[5 ℄.
The region above
D ¯
D
threshold is ri h in interesting new physi s. Inthisregion, losetothe
D ¯
D
threshold, oneexpe tstondthefour1Dstates.Oftheseonlythe
1
3
D
1
,identiedwiththeψ
(3770)resonan eanddis overed bythe MarkI ollaboration[22℄,has beenestablished. Itisawideresonan e(
Γ(ψ(3770) = 27.3 ± 1.0MeV/c
2
), whi h de ays predominantly to
D ¯
D
. TheJ
= 2 states (1
1
D
2
and1
3
D
2
) are predi ted to be narrow, be ause parity onservation forbids their de ay toD ¯
D
. In addition to theD
states, theradial ex itations of the
S
andP
states are predi ted to o ur above theopen harm threshold. None of these stateshave been positivelyidentied.
Some of the features of harmonium states are summarized in table 1.1.
In the next se tion, there will be more details on the experimental on the
energyregionabove
D ¯
D
threshold,inparti ularforthosenewstatesre ently,mainly dis overed at the B-fa tories.
1.5.3 New harmonia
A lot of new states have re ently been dis overed (X, Y, Z mesons),mainly
inthe hadroni de aysof the B meson: thesenew states are asso iatedwith
harmoniumbe ausetheyde aypredominantlyinto harmoniumstatessu h
as the
J/ψ
or theψ
′
, but their interpretation is far from obvious. In this
se tion,abriefsummary ofthe experimental dataand the possible
interpre-tationis presented.
Curiously,three harmonium-likestateswereobservedwithsimilarmasses
near 3.94 GeV/
c
2
, but in quite dierent pro esses, as summarized in table
1.2 [26 ℄.
The harmoniumlike state X(3940) has been observed by Belle in the
double harmonium produ tioninthe pro ess
e
+
e
−
→ J/ψ D ¯
D
∗
inthe mass
State
J
P C
Mass(MeV/c
2
)Γ
(MeV/c
2
) De ay Produ tion Collaboration
X(3940)
?
?+
3942+7
−6
±
8 37+26
−28
±
8D ¯
D
∗
e
+
e
−
→ J/ψ X(3940)
Belle X(4160)?
?+
4156+25
−20
±
15 139+11
−61
±
21D
∗
D
¯
∗
e
+
e
−
→ J/ψ X(4160)
Belle Y(3940)?
?+
3943±
11±
13 87±
22±
26ωJ/ψ
B → KY (3940)
Belle Y(3940)?
?+
3914.6+3.8
−3.4
±
2.0 34+12
−8
±
5ωJ/ψ
B → KY (3940)
BaBar Z(3930)2
++
3929±
5±
2 29±
10±
2γγ → Z(3940)
Belle Table1.2: MeasuredparametersoftheXYZ(3940)states.for the BABAR ollaboration. X(3940) state is tentatively identied with
η
c
(3S)
. In addition Belle found a new harmoniumlike state, X(4160), in thepro essese
+
e
−
→ J/ψ X(4160)
de ayinginto
D
∗
D
¯
∗
withasigni an eof
5.1
σ
[23 ℄. BoththeX(3940)andtheX(4160)de aytoopen harmnalstatesandtherefore ouldbeattributedto
3
1
S
0
and4
1
S
0
onventional harmonium states. However, the problem with this assignment is that potential modelspredi tmassesfortheselevelstobesigni antlyhigherthanthosemeasured
for the X(3940) and X(4160).
TheY(3940)statewasobservedbyBelleasanear-thresholdenhan ement
in the
ωJ/ψ
invariant mass distribution for ex lusiveB → KωJ/ψ
de ayswith astatisti al signi an e of8.1
σ
[24 ℄. AlsoBABAR found anωJ/ψ
massenhan ement at
∼
3.915 GeV/c
2
in the de ays
B → K
0,+
→ ωJ/ψ
[25 ℄ and
onrmed the Belle result. The Y(3940) mass is two standard deviations
lower than the Z(3930) mass, and three standard deviations lower than for
the X(3940); the width agrees with the Z(3930) and X(3940) values.
The Z(3930) state was found by Belle in two-photon ollisions
γγ →
D ¯
D
with a mass∼
3.930 GeV/ 2. The produ tion rate and the angular distributionintheγγ
enter-of-massframefavortheinterpretationofZ(3930)as the
χ
c2
(2P) harmoniumstate.For all other new states (X(3872), Y(4260), Y(4320) and so on) the
in-terpretationisnotatall lear,withspe ulations rangingfromthe missing
cc
Double harmonium produ tion
2.1 Introdu tion
The ex lusive produ tion of a pair of double heavy mesons with -quarks
in
e
+
e
−
annihilation has attra ted onsiderable attention in the last years.
In fa t, at the beginning of these studies, the ross se tion of the pro ess
e
+
e
−
→ J/ψ η
c
,whi hwasmeasured inthe experimentsonBABARand Belle dete tors at the energy√
s
= 10.6 GeV, resultedto beσ(e
+
e
−
→ J/ψ + η
c
) × B(η
c
→≥ 2 charged) =
(
25.6 ± 2.8 ± 3.4 [
27]
17.6 ± 2.8
+1.5
−2.1
[
28]
(2.1)and led to a dis repan y with the theoreti al al ulation in the framework
of nonrelativisti QCD (NRQCD) by an order of magnitude. This
on lu-sion is basedon al ulationsin whi h the relativemomentaof heavyquarks
and bound state ee ts inthe produ tionamplitude were not takeninto
a - ount. A set of al ulations was performed to improve the nonrelativisti
approximation for the pro ess.
Inparti ular,relativisti orre tionstothe rossse tion
σ(e
+
e
−
→ J/ψ +
η
c
)
were onsideredina olorsingletmodelinreferen e[29 ℄usingthemethods of NRQCD[14℄. A synthesis of this method will be done inse tion 2.2.1 .Anotherattempttotakeintoa ounttherelativisti orre tionswasdone
intheframeworkofthelight- oneformalism [30,31℄,des ribedhereinse tion
2.2.4 . With this formalism the dis repan y between experiment and theory
an be eliminated ompletely by onsidering the intrinsi motion of heavy
quarks formingthe doubly heavymesons.
In addition, perturbative orre tions of order
α
s
to the produ tion am-plitudewere al ulated inreferen e [32℄, where Zhang, Gao and Chao ouldOn a ount of dierent values of relativisti orre tions obtained in
ref-eren es [29 , 30 , 31 ℄ and the importan eof a relativisti onsideration of the
pro ess
e
+
e
−
→ J/ψ + η
c
in solving the doubly heavy meson produ tion problem, Ebert and Martynenko [33 ℄ have performed a new investigation ofrelativisti and bound state ee ts. This investigation is based on the
rela-tivisti quarkmodelwhi hprovidesthesolutioninmanytasksofheavyquark
physi s. In [34, 35℄ theyhave demonstrated how the originalamplitude,
de-s ribing the physi al pro ess, must be transformed in order to preserve the
relativisti plusboundstate orre tions onne tedwiththeone-parti lewave
fun tionsand the wave fun tionof a two-parti le bound state.
In parti ular, in paper [33 ℄ they extend the method to the ase of the
produ tion of a pair (
P + V
) of double heavy mesons ontaining quarks ofdierentavours
b
andc
. They onsidertheinternalmotionofheavyquarksin both produ ed pseudos alar
P
and ve torV
mesons, and the results ofthe ross-se tion will be presented in se tion2.2.3 .
Two more se tions are in this hapter: se tion 2.4 where a synthesis
of the re ent results on the analysis
e
+
e
−
→ γ
∗
→ J/ψ
+ X is done and
the possible interpretations of the state X(3940), whi h is expe ted to be
seen in the double harmonium produ tion pro ess via one virtual photon,
are illustrated. This last se tion is parti ularly interesting for this analysis,
whi hhasinitsaimsalsoto onrmthisstate,also seeninBelleinthe re oil
spe trum. Finallyinse tion2.5wewillbrieydes ribethetheory on erning
thedouble harmoniumprodu tionwithtwovirtual photonsinvolvedinthe
pro ess (
e
+
e
−
→ γ
∗
γ
∗
→ cccc
).
2.2 Cross se tion
If harmoniumisthe onlyhadronintheinitialornalstate,the olor-singlet
model should be a urate up to orre tions that are higher order in
v
. Thesimplestexamplesofsu hpro esses areele tromagneti annihilationde ays,
su has
J/ψ → e
+
e
−
and
η
c
→ γγ
,and ex lusiveele tromagneti produ tion pro esses,su h asγγ → η
c
.Another pro ess for whi h the olor-singlet model should be a urate is
e
+
e
−
annihilation intoexa tly two harmonia. There are nohadrons in the
initial state, and the absen e of additional hadrons in the nal state an
beguaranteed experimentallyby the monoenergeti natureofa 2-bodynal
state. For many harmonia,the NRQCDmatrix element an be determined
from the ele tromagneti annihilation de ay rate of either the harmonium
state itself or of another state related to it by spin symmetry. Cross
suppressed by powersof
v
2
without any unknown phenomenologi al fa tors.
One problem with
e
+
e
−
annihilation into ex lusive double harmonium
isthat the rossse tions arevery smallatenergieslarge enoughtotrustthe
predi tions of perturbative QCD. A naive estimate of the ross se tion for
J/ψ + η
c
inunits of the ross se tion forµ
+
µ
−
is:R[J/ψ + η
c
] ∼ α
2
s
m
c
v
E
beam
6
.
(2.2)The2powersof
α
s
arethefewestrequiredtoprodu eacc + cc
nalstate. There is a fa tor of (m
c
v
3
) asso iated with the wavefun tion at the origin
for ea h harmonium. These fa tors in the numerator are ompensated by
fa tors of the beam energy
E
beam
in the denominatorto get a dimensionless ratio.As an example, onsider
e
+
e
−
annihilation with enter-of-mass energy
2
E
beam
= 10.6 GeV. If we setv
2
≈
0.3,
α
s
≈
0.2, andm
c
≈
1.4 GeV, we get the naive estimateR[J/ψ + η
c
] ≈
4×
10−7
. This should be ompared
to the total ratio
R[hadrons] ≈
3.6 for all hadroni nal states [36 ℄. Thede ay of the
J/ψ
into the easily dete tablee
+
e
−
or
µ
+
µ
−
modes suppresses
the observable ross se tionby anotherorder of magnitude.
Fortunately,theeraofhigh-luminosityBfa torieshasmadethe
measure-mentofsu hsmall rossse tionsfeasible. BraatenandLee[29 ℄ al ulatedthe
ross se tions for ex lusive double- harmoniumprodu tion via
e
+
e
−
annihi-lation intoa virtual photon. This pro ess produ es only harmonium states
with opposite harge onjugation. The ross se tionsfor harmonium states
with thesame harge onjugation, whi h pro eed through
e
+
e
−
annihilation
into two virtual photons[37, 38℄ will be illustrated inse tion 2.5 .
2.2.1 Color-singlet model al ulation
In this se tion, the ross se tions for
e
+
e
−
annihilation through a virtual
photonintoadouble- harmoniumnalstate
H
1
+ H
2
are al ulatedbyusing the olor-singletmodel. The olor-singletmodel(CSM) anbeobtainedfromthe NRQCD fa torization formula by dropping all of the olour-o tet terms
and all but one of the olour-singletterms. The term that is retained isthe
one in whi h the quantum numbers of the
Q ¯
Q
pair are the same as those ofthe quarkonium.
Charge onjugation symmetryrequiresone ofthe harmoniatobea
C =
−
state and the other to be aC = +
state. TheC = −
states with narrow widths are theJ
P C
= 1
−−
states
J/ψ
andψ(2S)
, the1
+−
state
h
c
,and the yet-to-be-dis overed2
−−
The
C = +
states with narrowwidths are the0
−+
states
η
c
andη
c
(2S)
, theJ
++
states
χ
cJ
(1P),J = 0, 1, 2
, and the yet-to-be-dis overed2
−+
state
η
c2
(1D).The resultswill be express intermsof the ratioR[H
1
+ H
2
]
dened byR[H
1
+ H
2
] =
σ[e
+
e
−
→ H
1
+ H
2
]
σ[e
+
e
−
→ µ
+
µ
−
]
(2.3) In the text, only the results forR
summed over heli ity states will begiven. Theseresults mayfa ilitatethe use ofpartialwaveanalysis toresolve
theexperimental double- harmoniumsignalinto ontributions fromthe
var-ious harmonium states.
When the
e
+
e
−
beam energy
E
beam
ismu h largerthan the harm quark massm
c
, the relative sizes of the various double- harmonium ross se tions aregovernedlargelybythenumberofkinemati suppressionfa torsr
2
,where
the variable
r
is denedbyr
2
=
4m
2
c
E
2
beam
.
(2.4)If we set
m
c
= 1.4 GeV andE
beam
= 5.3 GeV, the value of this small parameter isr
2
= 0.28. The asymptoti behavior of the ratio
R[H
1
+ H
2
]
asr →
0 an be determined from the heli ity sele tion rules for ex lusivepro essesinperturbativeQCD.Forea hofthe
cc
pairsinthenalstate,thereisasuppressionfa torof
r
2
duetothe largemomentumtransfer requiredfor
the
c
andc
¯
to emerge with small relative momentum. Thus, at any orderin
α
s
, the ratioR[H
1
+ H
2
]
must de rease at least as fast asr
4
as
r → 0
.Howeveritmayde rease morerapidly dependingonthe heli itystatesofthe
two hadrons. There is of ourse a onstraint on the possible heli ities from
angularmomentum onservation:
|λ
1
− λ
2
|
=0 or1.The asymptoti behavior of the ratio
R[H
1
(λ
1
) + H
2
(λ
2
)]
depends on the heli itiesλ
1
andλ
2
. The heli ity sele tion rules imply that the slowest asymptoti de reaseR ∼ r
4
an o ur only if the sum of the heli ities of
the hadrons is onserved. Sin e there are no hadrons in the initial state,
hadron heli ity onservation requires
λ
1
+ λ
2
= 0. The only heli ity state that satises both this onstraint and the onstraint of angular momentumonservationis(
λ
1
, λ
2
)=(0, 0). Foreveryunitof heli ity by whi h thisrule isviolated, there is afurther suppression fa tor ofr
2
.
So, the resulting estimatefor the ratio R atleading order in
α
s
isR
QCD
[H
1
(λ
1
) + H
2
(λ
2
)] ∼ α
2
s
(v
2
)
3+L
1
+L
2
(r
2
)
2+|λ
1
+λ
2
|
.
(2.5)The fa tor of
v
3+2L
for a harmoniumstate with orbitalangular
ourse be further suppression fa tors of
r
2
that arise from the simple
stru -ture of the leading-order diagrams for
e
+
e
−
→ cc + cc
inFig. 2.1, but these
suppression fa tors are unlikely to persist tohigher orders in
α
s
. TheQEDdiagrams fore
+
e
−
→ cc(
3
S
1
) + cc
inFig. 2.2 give ontributionsto
R[J/ψ + H
2
]
that s ale in a dierent way withr
. This ase is a tuallyinteresting for the analysis do umented in this thesis. As
r →
0, there is aontribution to the ross se tion from these diagrams into the ross se tion
for
γ + H
2
and the fragmentation fun tionforγ → J/ψ
. Thisfragmentationpro essprodu es
J/ψ
inaλ
J/ψ
=±
1heli ity state. Thehard-s attering part of the pro ess produ es only onecc
pair with small relative momentum, sothere is one fewer fa tor of
r
2
relative to equation2.5 . The ross se tion for
γ + H
1
is still subje t tothe heli ity sele tion rules of perturbative QCD,sothe pure QED ontribution tothe ratio Rhas the behavior
R
QED
[J/ψ (±1) + H
2
(λ
2
)] ∼ α
2
(v
2
)
3+L
2
(r
2
)
1+|λ
2
|
.
(2.6)There may also be interferen e terms between the QCD and QED
on-tributions whoses aling behavioris intermediatebetween equations2.5 and
2.6 .
2.2.2 Cal ulation of the ross se tions
In this se tion, the ross se tions for ex lusive double- harmonium
produ -tion in
e
+
e
−
annihilationattheB fa tories ispresented, and partially
al u-lated.
The resultsinse tion2.2.1were expressedintermsof theratioRdened
inequation 2.3. The orresponding ross se tions are:
σ[H
1
+ H
2
] =
4πα
2
3s
R[H
1
+ H
2
]
(2.7)The ratios R depend on a number of inputs: the oupling onstants
α
s
andα
, the harmquark massm
c
, and the NRQCD matrix elementshO
1
i
.The value of the QCD oupling onstant
α
s
depends onthe hoi e ofthe s aleµ
. In the QCDdiagrams of Fig. 2.1 , the invariant mass ofthe gluon isps/2
. Wetherefore hoosethe s aletobeµ
=5.3GeV.Theresulting value of the QCD oupling onstant isα
s
(µ)
= 0.21.The numeri al value forthe pole mass
m
c
ofthe harm quarkisunstable under perturbative orre tions, so it must be treated with are. Sin e theexpressionsfor the ele tromagneti annihilation de ay rates in ludethe
per-turbative orre tion of order
α
s
the appropriate hoi e for the harm quark massm
c
in these expressions is the pole mass with orre tions of orderα
s
in luded. It an be expressed asP
1
P
2
(a)
(b)
(c)
(d)
Figure 2.1: QCD diagrams that an ontribute to the olor-singlet pro ess
γ
∗
→ cc + cc
→
P
2
P
1
(b)
(a)
Figure 2.2: QED diagrams that an ontribute to the olor-singlet pro ess
γ
∗
→ cc(
3
S
1
) + cc
m
c
= ¯
m
c
( ¯
m
c
)
1 +
4
3
α
s
π
.
(2.8)Taking the running mass of the harm quark to be
m
¯
c
( ¯
m
c
)
= 1.2±
0.2 GeV, the NLO pole mass ism
c
= 1.4±
0.2 GeV.The Braaten-Lee predi tions for the double harmonium ross se tions
without relativisti orre tions are givenin table 2.1 . 1
TheBraaten-Leepredi tionsforthedouble harmonium rossse tionsfor
the S-wave states (
η
c
, η
c
(2S), J/ψ , ψ(2S)
) in luding the leading relativisti orre tion are obtained by multiplyingthe values intable 2.1 by the fa tor:1
Onlyvalues interestingfor this analysis have been reported. For all al ulations see