Evidence for
B
!
K
H.-C. Huang,25K. Abe,7K. Abe,40T. Abe,41I. Adachi,7H. Aihara,42M. Akatsu,21T. Aso,46V. Aulchenko,2T. Aushev,11 A. M. Bakich,37Y. Ban,32E. Banas,26A. Bay,17I. Bedny,2P. K. Behera,47I. Bizjak,12A. Bondar,2A. Bozek,26 M. Bracˇko,19,12T. E. Browder,6B. C. K. Casey,6P. Chang,25Y. Chao,25K.-F. Chen,25B. G. Cheon,36R. Chistov,11 Y. Choi,36Y. K. Choi,36M. Danilov,11L. Y. Dong,9A. Drutskoy,11S. Eidelman,2V. Eiges,11Y. Enari,21C. Fukunaga,44
N. Gabyshev,7A. Garmash,2,7T. Gershon,7B. Golob,18,12R. Guo,23J. Haba,7F. Handa,41T. Hara,30N. C. Hastings,7 H. Hayashii,22M. Hazumi,7L. Hinz,17T. Hokuue,21Y. Hoshi,40W.-S. Hou,25Y. B. Hsiung,25,* Y. Igarashi,7T. Iijima,21 K. Inami,21A. Ishikawa,21R. Itoh,7Y. Iwasaki,7H. K. Jang,35J. H. Kang,50J. S. Kang,14P. Kapusta,26N. Katayama,7
H. Kawai,3N. Kawamura,1T. Kawasaki,28H. Kichimi,7D.W. Kim,36H. J. Kim,50Hyunwoo Kim,14 J. H. Kim,36 K. Kinoshita,5P. Koppenburg,7S. Korpar,19,12 P. Krizˇan,18,12P. Krokovny,2R. Kulasiri,5S. Kumar,31Y.-J. Kwon,50
G. Leder,10S. H. Lee,35T. Lesiak,26J. Li,34A. Limosani,20S.-W. Lin,25D. Liventsev,11J. MacNaughton,10 G. Majumder,38F. Mandl,10D. Marlow,33H. Matsumoto,28T. Matsumoto,44W. Mitaroff,10H. Miyata,28 G. R. Moloney,20T. Mori,4T. Nagamine,41Y. Nagasaka,8T. Nakadaira,42E. Nakano,29M. Nakao,7H. Nakazawa,7 J.W. Nam,36Z. Natkaniec,26S. Nishida,15O. Nitoh,45T. Nozaki,7S. Ogawa,39T. Ohshima,21T. Okabe,21S. Okuno,13 S. L. Olsen,6W. Ostrowicz,26H. Ozaki,7H. Palka,26C.W. Park,14H. Park,16K. S. Park,36N. Parslow,37J.-P. Perroud,17
M. Peters,6L. E. Piilonen,48M. Rozanska,26H. Sagawa,7S. Saitoh,7Y. Sakai,7T. R. Sarangi,47A. Satpathy,7,5 O. Schneider,17J. Schu¨mann,25C. Schwanda,7,10A. J. Schwartz,5T. Seki,44S. Semenov,11M. E. Sevior,20T. Shibata,28 H. Shibuya,39V. Sidorov,2J. B. Singh,31S. Stanicˇ,7,†M. Staricˇ,12A. Sugi,21K. Sumisawa,7T. Sumiyoshi,44S. Suzuki,49
S. Y. Suzuki,7T. Takahashi,29F. Takasaki,7K. Tamai,7N. Tamura,28J. Tanaka,42M. Tanaka,7G. N. Taylor,20 Y. Teramoto,29T. Tomura,42S. N. Tovey,20K. Trabelsi,6T. Tsuboyama,7T. Tsukamoto,7S. Uehara,7S. Uno,7G. Varner,6
K. E. Varvell,37C. C. Wang,25C. H. Wang,24J. G. Wang,48M.-Z. Wang,25Y. Watanabe,43E. Won,14 B. D. Yabsley,48 Y. Yamada,7A. Yamaguchi,41Y. Yamashita,27M. Yamauchi,7H. Yanai,28Heyoung Yang,35Y. Yusa,41C. C. Zhang,9
Z. P. Zhang,34Y. Zheng,6V. Zhilich,2and D. Zˇ ontar18,12
(Belle Collaboration)
1Aomori University, Aomori
2Budker Institute of Nuclear Physics, Novosibirsk
3Chiba University, Chiba
4Chuo University, Tokyo
5University of Cincinnati, Cincinnati, Ohio 45221
6University of Hawaii, Honolulu, Hawaii 96822
7High Energy Accelerator Research Organization (KEK), Tsukuba
8Hiroshima Institute of Technology, Hiroshima
9Institute of High Energy Physics, Chinese Academy of Sciences, Beijing
10Institute of High Energy Physics, Vienna
11Institute for Theoretical and Experimental Physics, Moscow
12J. Stefan Institute, Ljubljana
13Kanagawa University, Yokohama
14Korea University, Seoul
15Kyoto University, Kyoto
16Kyungpook National University, Taegu
17Institut de Physique des Hautes E´ nergies, Universite´ de Lausanne, Lausanne
18University of Ljubljana, Ljubljana
19University of Maribor, Maribor
20University of Melbourne, Victoria
21Nagoya University, Nagoya
22Nara Women’s University, Nara
23National Kaohsiung Normal University, Kaohsiung
24National Lien-Ho Institute of Technology, Miao Li
25Department of Physics, National Taiwan University, Taipei
26H. Niewodniczanski Institute of Nuclear Physics, Krakow
27Nihon Dental College, Niigata
29Osaka City University, Osaka
30Osaka University, Osaka
31Panjab University, Chandigarh
32Peking University, Beijing
33Princeton University, Princeton, New Jersey 08545
34University of Science and Technology of China, Hefei
35Seoul National University, Seoul
36Sungkyunkwan University, Suwon
37University of Sydney, Sydney New South Wales
38Tata Institute of Fundamental Research, Bombay
39Toho University, Funabashi
40Tohoku Gakuin University, Tagajo
41Tohoku University, Sendai
42Department of Physics, University of Tokyo, Tokyo
43Tokyo Institute of Technology, Tokyo
44Tokyo Metropolitan University, Tokyo
45Tokyo University of Agriculture and Technology, Tokyo
46Toyama National College of Maritime Technology, Toyama
47Utkal University, Bhubaneswer
48Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
49Yokkaichi University, Yokkaichi
50Yonsei University, Seoul
(Received 22 May 2003; published 10 December 2003)
We report evidence for the decay modeB!Kbased on an analysis of78 fb1of data collected
with the Belle detector at KEKB. This is the first example of ab!sssss transition. The branching fraction for this decay is measured to beBB!K 2:61:1
0:90:3 10
6for ainvariant
mass below2:85 GeV=c2. Results for other related charmonium decay modes are also reported.
DOI: 10.1103/PhysRevLett.91.241802 PACS numbers: 13.25.Hw, 14.40.Nd
We report evidence for the decay modeB!K, the first example of a b!sssss transition. In the standard model (SM), this decay channel requires the creation of an additional final ss quark pair than in b!sss pro-cesses, which have been previously observed in modes such as B!K. In addition to improving our under-standing of charmless B decays, theK state may be sensitive to glueball production in B decays, where the glueball decays to [1]. Furthermore, with sufficient statistics, the decayB!Kcould be used to search for a possible non-SMCP-violating phase in theb!s tran-sition [2]. Direct CP violation could be enhanced to as high as the 40% level if there is sizable interference between transitions due to non-SM physics and decays via thecresonance.
We use a78 fb1data sample collected with the Belle detector at the KEKB asymmetric-energyee (3.5 on 8 GeV) collider [3] operating at the 4S resonance (ps10:58 GeV). The sample contains85:0106 pro-ducedBBpairs. The Belle detector is a large-solid-angle magnetic spectrometer consisting of a three-layer silicon vertex detector, a 50-layer central drift chamber (CDC), a system of aerogel threshold Cˇ erenkov counters (ACC), time-of-flight scintillation counters (TOF), and an array of CsI(Tl) crystals located inside a superconducting so-lenoid coil that provides a 1.5 T magnetic field. An iron flux-return located outside of the coil is instrumented to
identifyK0Land muons. The detector is described in detail elsewhere [4].
We select well measured charged tracks that have impact parameters with respect to the nominal inter-action point (IP) that are less than 0.2 cm in the radial direction and less than 2 cm along the beam direction (z). Each track is identified as a kaon or a pion according to a K= likelihood ratio,LK=LLK, whereLK are likelihoods derived from responses of the TOF and ACC systems anddE=dxmeasurements in the CDC. We select kaon candidates by requiring LK=LLK>0:6. This requirement has a kaon efficiency varying from 89:61:0%for momentum of 500 MeV to86:90:4% for momentum of 3 GeVand a misidentification rate from pions of 8:5%. Kaon candidates that are electronlike according to the information recorded in the CsI(Tl) calorimeter are rejected.
Candidate mesons are reconstructed via the ! KK decay mode; we require the KK invariant mass to be within 20 MeV=c2 (4:5 times the full
width) of the mass [5]. For the B0B0 !K0
S de-cay mode, we use K0
S!
candidates in the mass
window 482 MeV=c2< M<514 MeV=c2 (4),
0.22 cm. The difference in the angle between the pion-pair vertex direction from the IP and its reconstructed flight direction in thex-yplane is required to be less than 0.03 radians.
To isolate the signal, we form the beam-constrained
mass, Mbc
E2beam jPP~
reconj2
q
, and the energy differ-enceEEreconEbeam. HereEbeamis the beam energy,
and Erecon and PP~recon are the reconstructed energy and momentum of the signal candidate, in the4S center-of-mass frame. The signal region for E is 30 MeV which corresponds to 3:1, where is the resolu-tion determined from a Gaussian fit to the Monte Carlo (MC) simulation and verified using the decay of B ! D0 and D0 !K. The signal region for Mbc is 5:27 GeV=c2< Mbc<5:29 GeV=c2. The
beam-constrained mass resolution is 2:8 MeV=c2, which is mostly due to the beam energy spread of KEKB.
The major background for the B!K process is from continuum ee !qqq production, where q is a light quark (u,d,s, orc). Several event topology variables are used to discriminate the continuum background, which tends to be collimated along the original quark direction, from theBBevents, which are more isotropic than the former. Five modified Fox-Wolfram moments, theS?variable [6], and the cosine of the thrust angle are combined into a Fisher discriminant [7]. We form signal and background probability density functions (PDFs) for this Fisher discriminant and for the cosine of theBdecay angle with respect to thezaxis (cosB) for the signal MC and sideband (5:20 GeV=c2< Mbc<5:26 GeV=c2 and
0:1<jEj<0:2 GeV) data. The PDFs are multiplied together to form signal and background likelihoods,LS andLBG. The likelihood ratioLRLS=LSLBGis then required to be greater than 0.1. This requirement retains 97% of the signal while removing 55% of the continuum background.
Figure 1(a) shows theinvariant mass spectrum for events in theB!Ksignal region, where a clearc peak and some excess in the lower mass region are evident.
To extract signal yields, we apply an unbinned, ex-tended maximum likelihood (ML) fit to the events with jEj<0:2 GeV and Mbc >5:2 GeV=c2. The
ex-tended likelihood for a sample of N events is L eNSNBQN
i1NSPSi NBPBi, where P SB
i describes the probability for candidate event i to belong to the signal (background), based on its measuredMbc andE
values. The exponential factor in the likelihood accounts for Poisson fluctuations in the total number of observed events N. The signal yieldNS and the number of back-ground events NB are obtained by maximizingL. The statistical errors correspond to unit changes in the quan-tity!2 2 lnLaround its minimum value. The signifi-cance of the signal is defined as the square root of the change in !2 when constraining the number of signal
events to zero in the likelihood fit; it reflects the proba-bility for the background to fluctuate to the observed event yield.
The probability P for a given eventi is calculated as the product of independent PDFs for Mbc and E. The
signal PDFs are represented by a Gaussian forMbcand a double Gaussian forE. The background PDF forEis a linear function; for theMbc background we use a phase-space-like function with an empirical shape [8]. The parameters of the PDFs are determined from high-statistics MC samples for the signal and sideband data for the background.
For M<2:85 GeV=c2, the region below the
charm threshold, the ML fit gives an event yield of 7:332::25 with a significance of 5.1 standard deviations (). Projections of the E distribution (with 5:27 GeV=c2< Mbc<5:29 GeV=c2) and of theMbc
dis-tribution (with jEj<30 MeV) are shown in Figs. 2(a) and 2(b). As a consistency check, a ML fit to the pro-jected Edistribution [Fig. 2(b)] gives a signal yield of 7:53:3
2:7 with a 4:8 statistical significance. Figure 1(b)
shows a scatter plot of the twoKKinvariant masses for events in the B meson signal region with the mass requirements relaxed. Here there is a clear concentration in the overlap region of the twobands. To confirm that the observed signal is from B!K, we apply a tighter mass requirement (10 MeV=c2), which
re-duces the signal efficiency by 15%, and obtain a signal yield of 5.6 with 4:6 statistical significance. Using a signal efficiency of3:3%, obtained from a large-statistics MC that uses three-body phase space to model theB! K decays, we determine the branching fraction for charmlessB!KwithM<2:85 GeV=c2to be
BB!K 2:61:1
0:90:3 10
6;
where the first error is statistical and the second is systematic.
0 5 10
2 3 4 5
Mφφ (GeV/c2)
Events / (0.1 GeV/c
2 )
(a)
M(K+K-)1 (GeV/c2)
M(K
+ K - )2
(GeV/c
2 )
(b)
0.98 1 1.02 1.04 1.06
0.98 1 1.02 1.04 1.06
FIG. 1 (color online). (a)invariant mass spectrum. The open histogram corresponds to events from theB!K signal region and the shaded histogram corresponds to events from theEsidebands. (b)MKK of onemeson candidate versus MKK of the other for the events satisfyingM<
2:85 GeV=c2. Dots are forKand squares forK0
S. Each
Contributions to the systematic error include the un-certainties due to the tracking efficiency (5:4%), particle identification efficiency (5%), and the modeling of the likelihood ratio cut (2%). The error due to the modeling of the likelihood ratio cut is determined using B ! D0!K events in the same data sample; these events have the same number of final-state particles and an event topology that is similar to theB !K signal. The uncertainty due to the MC M modeling (4%) accounts for the M dependence of the detection efficiency. The systematic error in the signal yield (6%) is determined by varying the means andof the signal and the shape parameters of the background. We determine an upper limit of 5% on the possible contamination by non-resonantB!KK
NRKorB!2KKNRK
decays by redoing the fits with the mass requirement relaxed. The sources of systematic error are combined in quadrature to obtain the final systematic error of12%.
For the B0B0 !K0
S mode, there are only four signal candidates. We combine the B !K and B0B0 !K0
Smodes and perform a ML fit and obtain a signal event yield of 8:73:6
2:9 with 5:3 statistical
sig-nificance. Assuming isospin symmetry, we obtain
BB!K 2:300::980:3 10
6;
forM<2:85 GeV=c2.
No enhancement is observed in the M region cor-responding to the fJ2220 glueball candidate [5], also refered to as %. Assuming the mass and width of fJ2220to be2230 MeV=c2 and 20 MeV=c2, we define a signal region of 2:19 GeV=c2< M
<2:27 GeV=c2, 5:27GeV=c2< M
bc<5:29GeV=c2, and jEj<30 MeV.
One event is observed in this region with an expected background, estimated from the sideband, of 0.5. Using an extended Cousins-Highland method that uses the the Feldman-Cousins ordering scheme and takes systematic uncertainties into account [9], we obtain a 90% confi-dence level (C.L.) upper limit of 3.7 signal events, which corresponds to
BB!fJ2220KBfJ2220 !<1:2106:
We select B!cK, c ! candidates by re-quiring2:94 GeV=c2< M
<3:02 GeV=c2. This decay has been searched by previous experiments [10]. A clear signal is evident in Figs. 2(c) and 2(d), and the fitted yield of NS7:032::03 events has a significance of 8:8. The
corresponding product branching fraction is
BB!
cK Bc! 2:201::070:5 10
6:
In addition to the previously listed source of systematic errors, here the error also includes the possible contami-nation from charmless B!K decays, which is estimated to be less than 1.2 events. Using the measured branching fraction BB!
cK 1:250:42 103[11], we determine the
c!branching fraction to be
Bc ! 1:800::860:7 10
3;
which is smaller than the current world average value of 7:12:8 103 [5].
(a) φφK±
0 1 2 3 4 5
(b) φφK±
0 2 4 6
(c) ηc(φφ)K±
0 1 2 3 4
(h) J/ψ(4K) K±
0 2.5 5 7.5 10 (g) J/ψ(4K) K±
0 2 4 6
(e) ηc(4K) K±
0 2 4
(d) ηc(φφ) K±
0 2 4
(f) ηc(4K) K±
0 2 4 6 8
Mbc (GeV/c2) ∆E (GeV)
Ev
ents / ( 2 MeV/c
2)
Ev
ents / 16 MeV
5.2 5.3 -0.2 0 0.2
FIG. 2 (color online). Projections of Mbcand E overlaid
with the fitted curves for (a),(b) B!K with M<
2:85 GeV=c2, (c),(d)B!
cKandc!, (e),(f)B!
cK and c!2KK, and (g),(h) B!J= K and
J= !2KK.
0 5 10 15
2.8 2.9 3 3.1 3.2 M4K (GeV/c2)
Events / (20 MeV/c
)
(a)
MφK+K- (GeV/c2)
Events / (20 MeV/c
)
(b)
0 5
2.8 2.9 3 3.1 3.2
2 2
FIG. 3 (color online). (a)2KKand (b)KKinvariant mass spectra in thecandJ= regions. The open histograms
Since theJ= and c charmonium resonances decay to 2KK, we also measure branching fractions of the decays B!charmoniumK with charmonium! 2KK. To select B!2KKK candidates, we apply tighter particle identification and continuum sup-pression requirements than in the case of B!K in order to reduce the larger combinatorial background. Figure 3(a) shows the invariant mass distribution of any two pairs of KK, M4K, between 2:8 GeV=c2 and
3:2 GeV=c2 for the events in the B signal region.
Significant contributions from both c and J= inter-mediate states are seen.
To identify the signals from c and J= intermedi-ate stintermedi-ates, we require that the invariant mass of 2KK satisfy 2:94 GeV=c2< M
4K<3:02 GeV=c2 and 3:06 GeV=c2 < M
4K <3:14 GeV=c2, respectively. We use signal yields from ML fits to determine branch-ing fractions. Figures 2(e) – 2(h) show the Mbc and E
projection plots with the fitted curves superimposed. Table I summarizes the signal yields, efficiencies, statis-tical significances, and the branching-fraction products. By requiring the invariant mass of one of theKKpairs to correspond to ameson, we also measure the decays of B!cJ= K and
cJ= !KK. The re-sults are included in Table I.
Using the known branching fractions BB! J= K 1:010:05 103 [5] andBB !
cK, we obtain the secondary branching fractions forJ= and cdecays to2KKandKK listed in Table II.
Our measured branching fractions for c ! and c!2KK are smaller than those of previous ex-periments [5], while those forJ= decays are consistent. The decayc!2KKproceeds dominantly through c!KK with !KK. This is the first mea-surement ofc!KK. The decay ofc !with !KK makes up approximately1=3of the branch-ing fraction ofc!KK.
In summary, we have observed evidence for the charm-less three-body decay B!K, which is the first ex-ample of ab!sssss transition. The branching fraction
BB!K 2:61:1
0:90:3 10
6 for M
< 2:85 GeV=c2, is measured with a significance of 5:1.
No signal is observed for the decayB!fJ2220Kwith fJ2220 !. The corresponding upper limit at 90% C.L. is BB !fJ2220KBfJ2220 !< 1:2106. We have also observed significant signals for B!cK with c !, with c !KK, and with c!2KK, as well as a signal for B! J= K with J= !KK. We report the first mea-surement ofc!KK with a branching fraction of
Bc !KK 2:90:9
0:81:1 10
3. Our
mea-sured branching fractions for c! and 2KK are smaller than those of previous experiments.
We wish to thank the KEKB accelerator group for the excellent operation of the KEKB accelerator. We acknowl-edge support from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and the Japan Society for the Promotion of Science; the Australian Research Council and the Australian Department of Industry, Science and Resources; the National Science Foundation of China under Contract No. 10175071; the Department of Science and Technology of India; the BK21 program of the Ministry of Education of Korea and the CHEP SRC program of the Korea Science and Engineering Foundation; the Polish State Committee for Scientific Research under Contract No. 2P03B 01324; the TABLE I. Signal yields, efficiencies including secondary branching fractions, statistical significances, and branching fractions (or branching fraction products) ofB!Kand the related decays. The branching fractions for modes withKKpairs include contributions from!KK.
Mode Yield Efficiency (%) Significance () B(106)
B!K M<2:85 GeV=c2) 7:332::25 3:30:3 5.1 2:6
1:1 0:90:3
B!KM<2:85 GeV=c2 8:723::69 2:20:2 5.3 2:3
0:9 0:80:3
B!fJ2220K,fJ2220 ! <3:7 3:60:3 <1:2
B!cK,c! 7:032::03 3:70:3 8.8 2:2
1:0 0:70:5
B!cK,c!KK 14:143::47 4:60:4 7.7 3:6
1:1 0:90:8
B!cK,c!2KK 14:643::69 9:60:9 6.6 1:8 0:6 0:50:4
B!J= K,J= !KK 9:03:7
3:0 4:40:4 5.3 2:4
1:0 0:80:3
B!J= K,J= !2KK 11:04:3
3:5 9:20:9 4.8 1:4
0:6 0:40:2
TABLE II. Measured branching fractions of secondary char-monium decays and the world averages [5]. The branching fractions for modes with KK pairs include contributions from!KK.
Decay mode B(this work) B(PDG)
c! 1:800::860:7 10
3 7:12:8 103
c!KK 2:900::981:1 10 3
c!2KK 1:400::540:6 10
3 2:11:2%
J= !KK 2:41:0
0:80:3 10
3 7:41:1 104
J= !2KK 1:400::540:2 10
Ministry of Science and Technology of the Russian Federation; the Ministry of Education, Science and Sport of the Republic of Slovenia; the National Science Council and the Ministry of Education of Taiwan; and the U.S. Department of Energy.
*On leave from Fermi National Accelerator Laboratory, Batavia, IL 60510.
†On leave from Nova Gorica Polytechnic, Nova Gorica.
[1] C.-K. Chua, W.-S. Hou, and S.-Y. Tsai, Phys. Lett. B544, 139 (2002).
[2] M. Hazumi, hep-ph/0303089.
[3] S. Kurokawa and E. Kikutani, Nucl. Instrum. Methods Phys. Res., Sect. A499, 1 (2003).
[4] Belle Collaboration, A. Abashian et al., Nucl. Instrum. Methods Phys. Res., Sect. A479, 117 (2002).
[5] Particle Data Group, K. Hagiwaraet al., Phys. Rev. D66, 010001 (2002).
[6] CLEO Collaboration, R. Ammaret al., Phys. Rev. Lett. 71, 674 (1993).
[7] The Fox-Wolfram moment were introduced in G. C. Fox and S. Wolfram, Phys. Rev. Lett. 41, 1581 (1978). The Fisher discriminant used by Belle is described in K. Abe et al., Phys. Lett. B517, 309 (2001).
[8] ARGUS Collaboration, H. Albrechtet al., Phys. Lett. B 241, 278 (1990).
[9] J. Conradet al., Phys. Rev. D67, 012002 (2003); R. D. Cousins and V. L. Highland, Nucl. Instrum. Methods Phys. Res., Sect. A 320, 331 (1992); G. J. Feldman and R. D. Cousins, Phys. Rev. D57, 3873 (1998).
[10] CLEO Collaboration, K.W. Edwards et al., Phys. Rev. Lett.86, 30 (2001).