Charmless B Decays at CDF

0.53 MB·

by  Diego Tonelli

Charmless B Decays at CDF
0
1
0
2
n Diego Tonelli∗†
a
Fermilab,P.O.Box500,Batavia,IL,60510-5011,USA
J
E-mail: [email protected]
0
1

] TheCDFexperimentattheFermilabTevatronpp¯ colliderpioneeredtheexplorationofcharmless
x
e b–hadrondecaysinhadroncollisions.Thisprovidedaunique,richandhighlysuccessfulprogram
-
p of measurements that is currently reaching its full maturity. I report here results on two-body
e decaysandthenewanalysisofB0→φφ decaysinaneventsamplecorrespondingto2.9fb−1.
h s
[

2
v
8
6
8
0
.
1
0
0
1
:
v
i
X
r
a
12thInternationalConferenceonB-PhysicsatHadronMachines
September7-11,2009
Heidelberg,Germany.
∗Speaker.
†OnbehalfoftheCDFCollaboration.
(cid:13)c Copyrightownedbytheauthor(s)underthetermsoftheCreativeCommonsAttribution-NonCommercial-ShareAlikeLicence. http://pos.sissa.it/
CharmlessBDecaysatCDF DiegoTonelli
1. Introduction
Non-leptoniccharmlessdecaysofb–hadronsareamongthemostwidelystudiedprocessesin
quark-flavorphysics. TheColliderDetectoratFermilab(CDF)isthefirstexperimentthatexplores
these decays in hadron collisions, and has so far unique access to large samples of charmless B0
s
(and Λ0) decays, necessary supplements to the wealth of precision measurements in the B0 and
b
B+ sectors available from the B-factories [1]. This is made possible by identifying long-lived de-
cays in the trigger, where dedicated custom electronics [2] reconstructs accurately trajectories of
chargedparticles(tracks)inthemicro-vertexsilicondetector. Tracksintheplanetransversetothe
beam are reconstructed at the first trigger stage (synchronous with the 2.52 MHz bunch-crossing)
inthemulti-wiredriftchamber(thatcovers|η|<1)andmatchedatthesecondtriggerstage(asyn-
chronous with 25 µs average latency) with tracks found in the silicon detector. The innermost
silicon layer, at 2.5 cm radius from the beam, allow measurement of impact parameter (distance
of closest approach to the beam) with offline-like, 48 µm resolution (including the beam spread).
This is small with respect to 150 µm values typical of relativistic b–hadron decays, allowing dis-
criminationoflong-livedheavyflavordecaysfromtheoverwhelminglight-quarkbackground. The
resultingsamplesofhadronicBdecaysprovidedarichandfruitfulprogramthatiscurrentlyreach-
ing its maturity, and whose latest results are shown in what follows. All conjugate modes are
implied,branchingfractionsareCP-averages,andK∗ isshorthandforK∗(892)0.
2. UpdatedanalysisofB0 →φφ decays
s
TheB0 →φφ decayismediatedbyapenguin-dominatedb→ss¯stransitionandwasfirstde-
s
tected in year 2005 by CDF, with 8 events in a data sample corresponding to a time-integrated
luminosity of 180 pb−1 [3]. The decay-rate is predicted with large uncertainty and the phe-
nomenologyisenrichedbythepresenceofthreedifferentpolarizationamplitudes,whichenhance
sensitivity to non–standard-model (non-SM) physics. We improved the measurement of branch-
ing fraction using 2.9 fb−1 of data. The B0 →φ(→K+K−)φ(→K+K−) rate is measured using
s
B0 →J/ψ(→µ+µ−)φ(→K+K−) decays, collected by the same trigger, as a reference mode as
s
follows:
B(B0→φφ) N B(J/ψ →µ+µ−)ε
s = φφ J/ψφε , (2.1)
B(B0→J/ψφ) N B(φ →K+K−) ε µ
s J/ψφ φφ
where N (N ) is the number of signal (reference) decays determined in data; the branching
φφ J/ψφ
ratiosofthevectormesonsareknown[4];thetriggerandselectionefficiencytoreconstructasignal
decayrelativetoareferenceone,ε /ε ,isdeterminedfromsimulation;andtheefficiencyfor
φφ J/ψφ
requiringmuonidentificationononeoftheJ/ψ prongsisaccountedforseparatelybytheε term,
µ
determined from inclusive J/ψ in data. We chose B0 →J/ψφ as a reference because it involves
s
a φ decay, and has same number of vertices and final state particles as the signal. The decay
B0→φ(→K+K−)K∗(→K+π−)isevenmoresimilartothesignal,butitsuseasreferencecarries
thepenaltyofthelargeuncertaintyontheratioofproductionfractions f /f .
s d
The trigger requires two oppositely-charged particles originated in a point displaced trans-
verselybytheprimarypp¯ collisionvertex. Triggertracksarerequiredtohaveanimpactparameter
of0.012<d <0.1cm,tooriginateatleast200µmfromtheprimaryvertexintheplanetransverse
0
2
CharmlessBDecaysatCDF DiegoTonelli
tothebeam,andtobeazimuthallyseparatedby2◦<∆φ <90◦. Differentthresholdsontransverse
momenta of particles and their scalar sum (p > 2 or 2.5 GeV/c, ∑p > 5.5 or 6.5 GeV/c) are
T T
applied as a function of luminosity to control trigger rates. In the offline analysis, a kinematic fit
toacommonspace-pointisappliedtoeachtrackpair. Pairsconsistentwithbeingkaonsoriginated
fromφ mesondecaysarethenfittogethertoa4-tracksvertex.
The mass distribution of the four kaons shows a large, featureless flat background. Samples
of simulated data (to characterize signal, S) and B0 mass sidebands (for background, B) are used
s
in an unbiased optimization to extract a visible signal. We maximized, over a large set of con-
√
figurations of kinematic and vertex quality requirements, the quantity S/ S+B. This is close to
√
optimalforabranchingfractionmeasurement,sinceS/ S+B∝1/σB,whereσB2 isthevariance
ofarateinacountingexperiment. Themostdiscriminatingobservablesarethetransversedecay-
length of the candidate, L (B0)>330 µm, and the transverse momentum of the lower–p kaon,
T s T
p (K)>0.7GeV/c. Otherusedquantitiesincludeimpactparameterofthehighest–p vectorme-
T T
son,d (φ)>85µm;impactparameterofthecandidate,d (B0)<65µm;andthefit-qualityofthe
0 0 s
reconstructedsecondaryvertex,χ2<17.
Figure1: DistributionofK+K−K+K−mass(left)andJ/ψK+K−mass(right)withfitfunctionoverlaid.
Thefinalsamplecontains295±20signaleventsoveramoderatebackground(fig.1,left)dom-
inatedbyaccidentalcombinationsofrandomtracksthatmeettheselectionrequirements(combina-
torial). Simulationshowsthata2.5%contaminationofB0→φK∗ isalsopresentandcontributions
ofthe(asyetunobserved)B0→K∗K∗decayarenegligibleforbranchingfractionsO(10−5),aspre-
s
dictedbytheSM.Theeventsclusteringatmassesof5.25–5.30GeV/c2suggestapossibleB0→φφ
contribution. AfitwherearesonanceattheB0massisallowedtofloatyieldsresultsconsistentwith
abackgroundfluctuationoflessthan2σ,withacorrespondingupperlimitonB(B0→φφ)consis-
tentwithcurrentlyknownvalues. ThishasbeenincludedasasystematiceffectintheB(B0→φφ)
s
measurement. Thefactor 40gainin signalyieldwith respecttothe previousresult[3], in spiteof
just a factor 16 increase in sample size quantifies the effect of the optimized trigger and offline
selection. Sevenlayersofsiliconsensorsextendingradiallyupto22cm,andthedriftchamberthat
provides96measurementsbetween30and140cm,allimmersedinthe1.4Taxialmagneticfield,
provideamassresolutionofapproximately17MeV/c2 ontheB0→φφ peak.
s
Offline reconstruction of B0 → J/ψφ decays is similar to the signal reconstruction, except
s
that the 4-track vertex is fit using the known J/ψ mass as a constraint and at least one track
3
CharmlessBDecaysatCDF DiegoTonelli
from the candidate J/ψ is required to have a matching extrapolation outward to track-segments
reconstructedinthemuondetectors(planardriftchambersat|η|<1). Thisrequirementselectsa
sufficientlyabundantandpuresamplerejectinganyJ/ψ →e+e−contamination. Adedicatedopti-
mizationyields1766±48B0→J/ψφ eventsoveralowbackgrounddominatedbycombinatorics
s
(fig. 1, right) and a residual 4% contribution from mis-reconstructed B0 →J/ψK∗ decays. Note
that B0 →J/ψφ decays reconstructed from the displaced-track trigger provide an additional 25%
s
signalfortheanalysisoftheB0-mixingphase,whichsuffersfromlow-statisticsissues[5].
s
The relative trigger and selection efficiency is extracted from simulation. Simulated decays
were reweighed to match the transverse momentum distribution of data because the Monte Carlo
doesnotreproduceaccuratelytheobservedadmixtureoftriggerswithdifferent p -thresholds. The
T
resultisε /ε =0.939±0.009±0.030. Thefirstuncertaintyisduetofinitestatisticscollected
J/ψφ φφ
byeachtrigger,thesecondonearisesfromthestatisticaluncertaintyinthereweighing.
The efficiency for identifying a muon is extracted from data using a few thousands inclusive
J/ψ → µ+µ− decays as a function of muon p and in two ranges of pseudorapidity, |η|<0.6
T
and 0.6<|η|<1.0, corresponding to distinct muon detectors. The J/ψ decays are collected by
the same trigger that collects signal decays, which is unbiased for muons, and selected offline as
follows: L (J/ψ)>200 µm, χ2 <5, p (µ)>1.5GeV/c,andatleastoneJ/ψ-trackthatextrap-
T T
olates outward to a track segment in the muon chambers. The resulting p –averaged efficiency is
T
ε =N /(N +N )=0.8695±0.0044(stat.), where N (N ) are J/ψ candidates that have
µ 2µ 1µ 2µ 1µ 2µ
exactlyone(two)muon(s)identifiedoffline,andtheefficiencyforidentifyingdi-muonsisassumed
factorizableintotheproductofsinglemuonefficiencies.
The dominant systematic uncertainty arises from the unknown polarizations of signal and
referencedecays(6-7%), whichimpactacceptance. OthercontributionsincludeaK/µ difference
intriggerefficiencyduetodifferentionizationinthetrackingchamber(3-4%),uncertaintyinsignal
massmodel(3%),unmodeledbackgrounds(3%), p -reweighingandbackgroundsubtraction(1%).
T
From the observed number of events, N =295±20 (stat.)±12 (syst.) and N =1766±
φφ J/ψφ
48 (stat.)±41 (syst.),wedetermine
B(B0→φφ)
s =[1.78±0.14(stat.)±0.20(syst.)]×10−2 (2.2)
B(B0→J/ψφ)
s
which, using B(B0 →J/ψφ)=(1.45±0.46)×10−3 (the known value updated with the current
s
value of f /f [4]), yields B(B0 →φφ)=[2.40±0.21(stat.)±0.27(syst.)±0.82(B)]×10−5.
s d s
The last uncertainty derives from the uncertainty in B(B0 →J/ψφ). The result agrees with the
s
previous determination, [1.40+0.6 (stat.)±0.6(syst.)]×10−5 [3], and with predictions based on
−0.5
QCDfactorization,[2.18±0.1+3.04]×10−5orperturbativeQCD,[3.53+0.83+1.67]×10−5[6],allof
−1.78 −0.69−1.02
whichhavelargeruncertainties.
This sample of 300 events is already sufficient to explore the rich polarization structure of
the B0 →φφ final state. Three states of polarization correspond to the allowed values of orbital
s
angularmomentum((cid:96)=0,1,or2)inthedecayofapseudo-scalarparticleintotwovectorparticles
(B→VV). A measurement of their relative contributions provides useful information to the puz-
zlingscenarioofB→VV polarizations. Theorypredictsthetransversepolarizationfractiontobe
small, f =O(m2/m2)≈4%, where m (m ) is the mass of the vector (B) meson, but measure-
T V B V B
mentsof f ≈50%inB0(+)→φK∗(+) decaysdisfavorthishierarchy[7]. Severaladhocsolutions
T
4
CharmlessBDecaysatCDF DiegoTonelli
may accommodate the discrepancy without invoking non-SM physics. These include enhanced
annihilation contributions, presence of transversely-polarized gluons, effects of electro-magnetic
or charming penguins, long-distance re-scattering and others [8]. However, all these explanations
are either model-dependent or non-conclusive, and the new physics option, e. g., 1+γ5 terms in
theamplitudefromscalarinteractionsorsuper-symmetricparticles,remainsvalid. Polarizationof
the B0 →φφ is considered crucial to discriminate among models. Ref.[9], for instance, proposes
s
to combine it with SU(3) symmetry to constrain the role of penguin annihilation amplitudes. The
analysisofpolarizationsisinprogressatCDFandO(5%)resolutionsonamplitudesareexpected.
3. AnalysisofB0 andΛ0 decaysintopairsofcharmlesshadrons
(s) b
Two-body decays mediated by the b→u quark-level transition have amplitudes sensitive to
γ (or φ ), the least known angle of the CKM matrix; significant contributions from ‘penguin’
3
transitions probe the presence of non–standard-model physics in loops; in addition, the variety of
openchannelswithsimilarfinalstatesallowcancellationofmanycommonsystematiceffectsand
provides information to improve effective models of low-energy QCD. Analysis of these decays
haveakeyroleintheCDFflavorprogram,providinguniqueaccesstocharmlessB0andΛ0decays,
s b
andmeasurementsintheB0 sectorcompetitivewiththeB–factories[10][11]. Webrieflyoverview
here the results based on 1 fb−1 of data collected by a dedicated displaced-track trigger similar to
theonedescribedintheprevioussection.
Anoptimizedofflineselectionisolatesaclearsignalofafewthousandseventsoverasmooth
backgrounddominatedbycombinatorialcontributions(fig.2,left). Absenceofevent-specificpar-
22er 20 MeV/er 20 MeV/cc 111111111110000000000033333333333 dtoattaal BBBB(cid:82)(cid:82)B0s00s0b0b00 (cid:65)(cid:65)(cid:65)(cid:65)(cid:65)(cid:65)(cid:65) pKK(cid:47)KpKK+(cid:47)++-+(cid:47)(cid:47)(cid:47)K-K--+-- - (cid:103)2 231 BB00 (cid:65)(cid:65) (cid:47)K++K(cid:47)--
es pes p Bcos0 m(cid:65)b (cid:47)in+(cid:47)a-torial backg. 0
atat 111111111110000000000022222222222 multi-body B decays
dd
didi -1
nn
aa
CC -2
-3
55555555555 55555555555...........11111111111 55555555555...........22222222222 55555555555...........33333333333 55555555555...........44444444444 55555555555...........55555555555 55555555555...........66666666666 55555555555...........77777777777 55555555555...........88888888888 -3 -2 -1 0 1 2 3
IInnvvaarriiaanntt (cid:47)(cid:47)(cid:47)(cid:47)--mmaassss [[GGeeVV//cc22]] (cid:103)1
Figure2:Distributionofππ–masswithfitprojectionsoverlaid(left).SeparationbetweenK+π−andK−π+
statesinthespaceofthedE/dxofthetwoparticles(right). Thequantityκ istheobserveddE/dxaroundthe
averagepionresponsedividedbythedifferencebetweenaveragekaonandaveragepionresponses.
ticle identification and limited mass resolution (approximately 22 MeV/c2 in these decays) cause
several channels to appear overlapping in a single peak. A five-dimensional likelihood fit rely-
ing on kinematics differences between decays (correlation between the two-body mass with ar-
bitrary mass assignment to the final state particles and their momentum imbalance), and particle
identification from the measurement of specific ionization in the drift chamber (1.5σ kaon-pion
separation from dE/dx, fig. 2, right) allow statistical determination of the individual contribu-
tions. Large D0 → K−π+ samples are used both to model accurately mass line-shapes and for
5
CharmlessBDecaysatCDF DiegoTonelli
Decaymode Relativebranchingfraction(B) AbsoluteB(10−6)
B0→π+π− B(B0→π+π−) = 0.259±0.017±0.016 5.02±0.33±0.35
B(B0→K+π−)
B0→K+K− fs ×B(B0s→K+K−) = 0.347±0.020±0.021 24.4±1.4±3.5
s fd B(B0→K+π−)
B0→K−π+ fs ×B(B0s→K−π+) = 0.071±0.010±0.007 5.0±0.7±0.8
s fd B(B0→K+π−)
B0→π+π− fs × B(B0s→π+π−) = 0.007±0.004±0.005 <1.2at90%C.L.
s fd B(B0→K+π−)
B0→K+K− B(B0→K+K−) = 0.020±0.008±0.006 <0.7at90%C.L.
B(B0→K+π−)
Λ0→pK− fL × B(Λ0b→pK−) = 0.066±0.009±0.008 5.6±0.8±1.5
b fd B(B0→K+π−)
Λ0→pπ− fL × B(Λ0b→pπ−) = 0.042±0.007±0.006 3.5±0.6±0.9
b fd B(B0→K+π−)
Table1: Measuredbranchingfractions. Absoluteresultswerederivedbynormalizingtothecurrentknown
value B(B0 →K+π−)=(19.4±0.6)×10−6, and assuming f /f =0.276±0.034 and f /f =0.230±
s d L d
0.052fortheproductionfractions[4]. Thefirstquoteduncertaintyisstatistical,thesecondissystematic.
dE/dx calibration. Event fractions for each channel are corrected for trigger and selection effi-
ciencies (from simulation and data) to extract the decay-rates (tab. 1). We report the first obser-
vation of the decays B0 →K−π+, Λ0 →pK−, and Λ0 →pπ−, and world-leading measurements
s b b
of (upper limits on) the B0 →K+K− (B0 →π+π−) branching fractions [11]. The measured ra-
s s
tioB(Λ0 →pπ−)/B(Λ0 →pK−)=0.66±0.14(stat.)±0.08(syst.)showsthatsignificantpen-
b b
guin contributions compensate the Cabibbo (and kinematic) suppression expected at tree level.
We also report first measurements of the direct CP-violating asymmetries A (B0 → K−π+) =
CP s
[39±15(stat.)±8(syst.)]%, A (Λ0 → pπ−) = [3±17(stat.)±5(syst.)]%, and A (Λ0 →
CP b CP b
pK−)=[37±17(stat.)±3(syst.)]%. DirectCPviolationintheB0→K+π−modeismeasuredas
A (B0 →K+π−)=[−8.6±2.3(stat.)±0.9(syst.)]%, whichisconsistentandclosetobecom-
CP
petitivewiththefinalB–factoriesresults. Theanalysisofthe5fb−1 sampleisnowinprogress. We
expecttoreconstructmorethan15,000B0 →h+h(cid:48)− decays,wherenewmodescouldbeobserved
(s)
andCP-violatingasymmetriesmeasuredwithdoubledprecision.
4. Concludingremarksandoutlook
CDFkeepsharvestingfromitsrichanduniqueprogramoncharmlessBdecaysinhadroncol-
lisions, pioneered since year 2002. The update of the B0 →φφ analysis to 2.9 fb−1 is the latest
s
addition. The 37% uncertainty on the branching fraction is almost half of the previous determi-
nation, and is dominated by the uncertainty on B(B0 →J/ψφ), which may soon reduced by the
s
Belle experiment. The polarization analysis is in progress with expected uncertainties of 5%. In
theupdateto1fb−1ofthetwo-bodyanalysis,thedecaysB0→K−π+,Λ0→pπ−,andΛ0→pK−
s b b
were newly observed and their branching fractions and CP-violating asymmetries measured. The
observedhierarchyB(Λ0→pK−)>B(Λ0→pπ−)isinconsistentwithrecenttheorypredictions
b b
[12],suggestingsignificantcontributionsfrompenguinamplitudesinhadronicΛ0 decays.
b
This fruitful program is reaching its maturity, but it’s far from being exhausted. By October
2010 CDF will have more than 8 fb−1 of physics quality data on permanent storage, which may
reach 10 fb−1 after one year, if Tevatron Run II will be further extended. These 2.5–10 factors in
6
CharmlessBDecaysatCDF DiegoTonelli
increase of sample-size over analyses shown here set the scale of what’s coming next. Not only
current analyses, mostly limited by statistical uncertainties and by systematic effects that scale
with statistics, will be improved. Avenues for new measurements are opening-up: an accurate
determination of the polarization structure of B0 →φφ decays will contribute useful information
s
onthepuzzlingscenarioinB→VV decays;searchforalargemixingphaseinpenguin-dominated
B0 →φφ decays will supplement the analogous effort in tree-dominated B0 →J/ψφ modes [5];
s s
comparison of B0 and B0 decay rates in common K+π− final states will provide stringent model-
s
independentconstraintsonnon-SMphysicscontributions[13],explorationoftime-dependentCP-
violatingasymmetriesinB0 →h+h(cid:48)− decayswillincreaseourknowledgeoftheangleγ (φ ).
(s) 3
While the first, exciting whimpers of LHC are emerging these days, CDF sits on a goldmine
ofdataandafewexcitingyearsofcompetitionwithLHCbarecoming.
Acknowledgments
ThankstotheCDFandTevatronpeopleforproducingtheseresultsandtotheconference
organizersforhavingpatientlywaitedmylongoverduecontribution.
References
[1] M.Antonellietal.,arXiv:0907.5386 (hep-ph),submittedtoPhys.Rept.
[2] W.Ashmanskasetal.,Nucl.Instrum.MethodsA518532(2004)[physics/0306169].
[3] D.Acostaetal.(CDFCollaboration),Phys.Rev.Lett.95031801(2005)[hep-ex/0502044].
[4] C.Amsleretal.(ParticleDataGroup),Phys.Lett.B6671(2008).
[5] T.Aaltonenetal.(CDFCollaboration),Phys.Rev.Lett.100161802(2008)
[0712.2397(hep-ex)];D.Tonelli(fortheCDFCollaboration),0810.3229(hep-ex).
[6] see,forinstance,M.Beneke,J.Rohrer,andD.Yang,Nucl.Phys.B77464(2007)
[hep-ph/0612290]orA.Alietal.,Phys.Rev.D76074018(2007)[hep-ph/0703162].
[7] see,forinstance,B.Aubertetal.(BaBarCollaboration),Phys.Rev.Lett.98051801(2007)
[hep-ex/0610073].
[8] examplesincludeA.L.Kagan,Phys.Lett.B601151(2004)[hep-ph/0405134];W.-S.Houand
M.Nagashima,hep-ph/0408007;M.Beneke,J.Rohrer,andD.Yang,Phys.Rev.Lett.96141801
(2006)[hep-ph/0512258];C.W.Baueretal.,Phys.Rev.D70054015(2004)
[hep-ph/0401188];H.-Y.Cheng,C.-K.Chua,andA.Soni,Phys.Rev.D71014030(2005)
[hep-ph/0409317].
[9] A.Dattaetal.,Eur.Phys.J. C60279(2009)[0802.0897(hep-ph)].
[10] A.Abulenciaetal.(CDFCollaboration),Phys.Rev.Lett.97211802(2006)[hep-ex/0607021].
[11] T.Aaltonenetal.(CDFCollaboration),Phys.Rev.Lett.103031801(2009)
[0812.4271(hep-ex)].
[12] C.-D..Luetal.,Phys.Rev.D80034011(2009),[0906.1479(hep-ph)]orZ.-T.Wei,H.-W.Ke,
andX.-Q.Li,ibidem094016,[0909.0100(hep-ph)].
[13] H.J.Lipkin,Phys.Lett.B621126(2005),[hep-ph/0503022].
7