PhysicsLetters B 305 (1993) 177-181 North-Holland PHYSICS LETTERS B Measurement of relative branching fractions for D ÷ K-K+K + and Ds+ - - , n - n +n ÷ decays WA82 Collaboration M Adamovich e, y Alexandrov e, F Antinori b, D Barberis c, W Beusch b, A Buys d, V Casanova c, M Dameri c, M Davenport b, j.p Dufey b, A Forino ", B.R French b, R Gessaroli ", F Grard d, K Harrison b, R Hurst ¢, A Jacholkowski b, A Kirk b, E Lamanna b, J.C Lassalle b, p Legros d, p Mazzanti ", F Muller b.~, p Netchaeva *, B Osculati c, A Quareni a, N Redaelli b, C Roda b, L Rossi ~, G Tomasini c, M Weymann b and M Zavertyaev ~ a Dipartimento di Fisica and INFN, Bologna, Italy b CERN, European OrganizationforNuclearResearch, Geneva, Switzerland Dipartimento di Fisica and INFN, Genova, Italy d Universitd de Mons-Hainaut and IISN, Mons, Belgium Lebedev Physical Institute, Moscow, Russian Federation Received 27 February 1993 The decaymodesD,+ -,lt-n +n + and D + - - , K - K +K + havebeen studied in the hadroproductionexperimentWA82.The following ratios of branchingfractionshave been measured: B(D{ - ~ n - n + n + ) / B ( D { *On+ ) =0.33+_0.10_+0.04, B ( D + ~ K - K + K + ) / B ( D + - , K - n + n + ) =0.057_+0,020+0.007, B(D+-.K-K+K+)/B(D+ ~On+)=0.49+0.23_+O.06 Much theoretical work has been carried out in trying to understand the so-called charm-decay puzzle [ ] and, in particular, the difference between the lifetime of the D + meson and the lifetimes of the D° and D + mesons Effects of possible relevance include non-leptonic decays occurring via processes not described by spectator diagrams, final-state interactions and the destructive interference that may arise in the Cabibbo-favoured decays of the D ÷ but not in the decays of the D ° It is noted that doubly Cabibbosuppressed D ÷ decays are not subject to destructive interference, therefore their relative branching fractions might be enhanced [ ] In a previous paper [ ] we reported on the WA82 t Deceased measurements of the relative branching fractions of D O Cabibbo-suppressed decays Here we present results based on an analysis of the three-body decays D~+ - * n - n +n + and D + ~ K - K + K + (the charge conjugate states will be implicitly included throughout this paper) These decays cannot be described by simple spectator diagrams, but must involve annihilation subprocesses or final-state rescattering Furthermore, the D + - - , K - K + K + mode is doubly Cabibbo-suppressed The E691 Collaboration has observed the decay Ds+ - ~ n - n + n + [4] in addition to the decay D +-~ n - n + n + previously reported by the Mark lII Collaboration [5] The decay D + ~ K - K + K + has been studied by the E691 Collaboration [ ] in the c-resonant mode only In the WA82 experiment the CERN Omega Spec- 0370-2693/93/$ 06.00 © 1993 Elsevier SciencePublishers B.V All rights reserved 177 Volume 305, number 1,2 PHYSICS LETTERS B trometer, supplemented with a silicon microstrip detector, was used to study charmed hadrons produced by interactions of 340 G e V / c n - mesons on a m m thick target On-line enrichment for charmed particles was achieved by triggering on events in which at least one track was found to miss the primary interaction vertex by more than 100 gm Some × 107 events were recorded and are used in the present analysis For 60% o f the data, particle identification was provided by a Ring Imaging Cherenkov detector ( R I C H ) The experimental set-up, the trigger requirements and the data processing have been fully described in previous publications [ ] Identification o f charmed particles was made possible by the high precision in secondary vertex reconstruction (30 /am transversally, 500 g m longitudinally) provided by the silicon microstrip telescope The present analysis o f three-body decays is similar to our previous analysis o f decays of the D O [ ] Invariant mass distributions were obtained for events where a secondary vertex was found outside the thin target and where the longitudinal separation between the secondary vertex and the primary interaction vertex was greater than standard deviations When extrapolated backwards the reconstructed momentum vector o f a decay candidate was required to pass within 30 g m o f the primary vertex Particle identification with the R I C H was performed using the following procedure: first chisquared probabilities, derived from expected and observed numbers of photo-electrons and ring radii, were obtained for the different mass hypotheses; next conditional probabilities were calculated using a priori probabilities determined from the data through an iterative method [8]; then the results were expressed in terms o f ratio between the conditional and the a priori probability for each mass hypothesis ( x e denotes this ratio for particle hypothesis P = n, K, p in the following discussion); finally, particles were identified by cutting on xe The R I C H identification efficiency for kaons has been determined experimentally as a function of m o m e n t u m by considering the signals for D Oand D + Cabibbo-favoured decays Fig shows our signals for 0-resonant D~+ , K - K + n + and D + - - , K - K + n + , without and with R I C H identification, the latter obtained by requiring that xK, the ratio o f the kaon probability to the kaon 178 May 1993 (a) 3o 09 25 2o > LLJ 17 175 18 185 19 195 205 m ( K K*7~+) 21 215 Oev"~ (b) (D 60 10 c > hA J 175 18 185 19 195 rn ( K-K *7~+ ) 205 21 215 0eV/" Fig h (a) K - K + ~ + invariant mass distribution for full data sample, m ( K - K +) within 10 MeV/c of the ¢ mass, no RICH identification (b) K - K * g ÷ invariant mass distribution for the 60°/0 of the data for which RICH information is available, m ( K - K + ) within 10 MeV/c of the ¢ mass, x~>0.5 for both kaon candidates a priori probability, be greater than 0.5 for both kaon candidates Fig 2a shows the D + - - - , K - g + n ÷ signal without R I C H identification The signal over noise ratio can be further improved by cutting on the proper lifetime of the decay candidate, as shown in fig 2b Fig shows the g - g + g + invariant mass distribution for the full data sample, where the large peak appearing to the left o f the D + and D~+ signals is due to the reflection of D + - - , K - g +g ÷ in which the kaon is wrongly assumed to be a pion The R I C H identification is not used for this mode since the background under the signals is mostly made ofpions Instead we have asked that the calculated proper lifetime, r, of a decay candidate should be greater than some specified value Results are shown in fig 3a for r > 0.5 ps and in fig 3b for r > ps The cut z > ps eliminates a substantial part of the D~+ signal, while the Volume 305, n u m b e r 1,2 PHYSICS LETTERS B ~(j May 1993 45 (a) (o) ~ " 200 @ ~75 50 ~ 125 O9 ~ 25 ~ 100 2ksJ to T>O.5ps 55 ~5o 20 15 25 75 I 85 95 t D > 2.5 ~ 18 21 85 19 OeV/c m(K-Tz+7~ ÷) (1) 205 t/b/ 195 2.05 21 215 rYl ( 7s-7s+ 7s + 22 OeV/c ~25 b 2o Q~) 17 O3 125 1.5 @ 75 O 0.5 25 o12 ' 'oi, ' '0,6' ' '0,8' lJ~u~ ' ' I ' ' I l l12 T~,o(ps) Fig (a) K - n +zt + invariant mass distribution for the full data sample, no RICH identification (b) Relative values of signalto-background ratio (S/B) and signal (S) as a function of the cut on proper lifetime ~> zmi, Values are norrnalised to for no explicit proper lifetime cut (i.e when there are only the requirements of trigger and filter and the requested vertex separation) D + signal is little affected, as is to be expected from the D ÷ and D~+ lifetimes For the calculation of relative branching fractions we obtain a reduced background, with no significant loss of statistics, by requiring r > 0.8 ps when considering D + and ~> 0.5 ps when considering D + Fig 4a shows the K - K + K ÷ invariant mass distribution for the data sample for which RICH information was available In order to optimise background reduction, while keeping evidence of a signal, we request x r > 0.5 for both K + candidates In this analysis we not consider secondary vertices that are compatible within 15 MeV/c with D+ K - T t + n +, with D+ / D s+ .(bzt +, with A~+ ,p K - z t + or within 10 MeV/c with K ° - - , n - n + to which a third track was associated The RICH cut approximately halves, to 13.1 + 4.5, the number of events in '18' ' '165 ]9 195 205 m (7r-7~+7~ + ) 21 2}5 22 OeV/c: Fig ~ - n + ~ + invariant mass distribution for full data sample, no RICH identification, cuts on proper lifetimes as indicated the peak at the D ÷ mass This is consistent with the RICH efficiency for the mode considered (0.40+0.03) By contrast, the background is reduced by more than a factor of To further reduce the background and check the compatibility of the observed peak with a D + signal, we have also applied cuts on proper lifetime, obtaining the invariant mass distribution shown in fig 4b for z> 1.3 ps There is then a peak of 5.7+2.8 events at the D + mass, in agreement with the expected number of 4.1 + 1.5 according to fig 2b The number of events in the various decay modes was determined by maximum likelihood fits to the invariant mass distributions, as in ref [3] Signal peaks are fitted with gaussians Linear functions are used to describe the structureless backgrounds except in the case of the K - K + K ÷ mode, where a quadratic function is adopted The tail of the reflection peak in the n-z~+n + distribution is modelled as a Breit179 Volume 305, number 1,2 > "~ 35 PHYSICS LETTERS B x~ > O × > (o) F: ;l " ii 50 rl :; [k, r, ~'-ir, 25 M a y 1993 Table Number of events in the three-body D ÷ and D decay modes for the different selection criteria Decay mode r[ ,,L, m r(ps) xx D+~K-rc+zc + 939 425 >0.5 Ld 15 D + ,¢~r + 16 17 ~8 lg 21 m (k-I~ +t< ÷ ) % > I0 [ ~ T 32.7_+ 9.0 19.3_+ 6.0 >0.5 45.8_+ 8.3 19.4_+ 6.0 g > LI D+-,n-n+n+ D + *rr-n +~r + >Ops T > ]5 22 Oev/c: x~ > ps D+~,K-K+K + >0.5 >0.5 >0.5 19.5_+ 7.1 >0.8 19.7_+ 7.0 >0.5 >0.8 28.6_+ 8.4 11.7_+ 5.1 > 1.3 13.1_+ 4.5 5.7_+ 2.8 Table Comparison of WA82 results with P D G values 16 17 18 19 m(K-K+K +) 21 22 CeV/c F(D+~n-n+n +) F(D + ~K-n+n + ) WA82 PDG 0.032+0.011 +0.003 0.035_+0.007 0.075_+0.007 Fig ( a ) K - K + K + invariant mass distribution for the 60% of data for which RICH information is available, with no R I C H identification (dotted line) and with x x > 0.5 for both K ÷ candidates (solid line ) ( b ) Upper curve as solid line of ( a ) ; shaded area with additional requirement z > 1.3 ps F(D+ '~rt+) F ( D + *K - n + n + ) 0.062+0.017+0.006 F(D+~n-n+n+) F(D +~On +) F(D~n-n+~r +) 0.52 _+0.23 + Wigner curve, as given by simulated data for F ( D+ ~ K - K + K + ) 0.057-+0.020_+0.007 F(D+-,K-rt+rt + ) F(D + ,K-K+K +) 0.49 + _+0.06 F ( D + , (bn + ) D + - - , K - n +n + When considering signals other than D+~K-n+n +, the D + and D~+ masses are fixed to the P D G values [ ] and peak widths are those determined from simulated events: a = M e V / c for D + - - , n - n + n +, 7.1 M e V / c for D + - - , n - n + n + and 5.3 MeV/c for D + - , K - K + K + The simulated events, generated using PYTHIA [ 10] and passed through a detector simulation program, have been subjected to the same selection criteria and analysis procedures as the experimental data The numbers of events obtained in the different modes are given in table Relative detection efficiencies for the decay modes considered have been determined from the simulated data [ 11 ] The effect on the rr-rt+n+ and K - K + K + invariant mass distributions o f the proper lifetime cuts used in our analysis is compatible with 180 _+45 _+34 >0.5 D + ,¢zr + 15 Number of events F(D~+ ~On + ) 0.33 + _+0.04 0.44 + +0.04 the expectations both from simulation and from the observed reduction of the D + ~ K - n + n ÷ signal Our results are summarized in table and are compared with the current P D G data [ ] For each of our measurements the first error quoted is statistical and the second is systematic, taking into account uncertainties in the calculated detection efficiencies and different parameterisations o f the background Our results for previously established modes are in agreement with the P D G values Of particular interest, we find B(D~+ ~rc-z~ +~r÷ ) = 3 + _ + B ( D + ,On + ) Volume 305, number 1,2 PHYSICS LETTERS B F o r D + - - K - K + K +, w i t h o u t cut o n p r o p e r lifetime, we o b t a i n B(D+-*K-K+K+ ) =0.057+0.020+0.007, B(D+-,K-rc+~r +) B ( D + o K - K + K + ) = 0.49 + 0,23 + 0.06 B(D+ ~¢zt + ) W i t h the cut z > 1.3 ps, we m e a s u r e B ( D + ~ K - K + K + ) = 0.069 + 0.034 + 0 , B(D+~K-rc+rc +) c o n s i s t e n t with the result given above We f i n d n o e v i d e n c e for D + ~ K - K + K + or for a significant D + ~q~K + r e s o n a n t c o n t r i b u t i o n i n D + - - , K - K + K + This latter o b s e r v a t i o n is n o t i n c o m p a t i b l e with the E691 result [6] o f B ( D + ~ ¢ ) K + ) / B ( D + - - , ¢ ) n + ) = (~J o -Q+3.2 -2 -+ - 0.7) × 10-% w h i c h in o u r data w o u l d give a signal o f 1.3 events (with n o cut o n proper lifetime) In c o n c l u s i o n , we have m e a s u r e d relative b r a n c h ing fractions for some D + a n d D + decays This analysis i n c l u d e s the study o f a n e w d o u b l y C a b i b b o - s u p pressed decay D + ~ K - K + K + in the n o n - r e s o n a n t May 1993 m o d e , the i d e n t i f i c a t i o n of which was possible using RICH information References [ ] I.I Bigi, in: Heavy quark physics, eds P.S Drell and D.L Rubin, AlP Conf Proc No 196 (American Institute of Physics, New York, 1989), and references therein [2] I.I Bigi, Proc 16th SLAC Summer Institute on Particle physics, ed E.C Brennan (Stanford University, Stanford, CA, 1988) [ ] M Adamovich et al., Phys Lett B 280 (1992) 163 [4] J.C Anjos et al., Phys Rev Lett 62 (1989) 125 [ ] R.M Baltrusaitis et al., Phys Rev Lett 56 (1986) 2140 [6] J.C Anjos et al., Rev Len 69 (1992) 2892 [7] M Adamovich et al., IEEE Trans Nucl Sci 37 (1990) 236; Nucl Instrum Methods A 309 ( 1991 ) 401 [8 ] A Buys, Dissertation de doctorat, Universit~ de MonsHainaut (1992) [9] Particle Data Group, K Hikasa et al., Review of particle properties, Phys Rev D 45 (1992) I.l [ 10] Lund Monte Carlo Programs Manual (CERN, November 1989) [11 ] A Buys, Internal report PNPE-312, Universit6 de MonsHainaut (1993) 181 ... Fig 2a shows the D + - - - , K - g + n ÷ signal without R I C H identification The signal over noise ratio can be further improved by cutting on the proper lifetime of the decay candidate, as shown... shows the g - g + g + invariant mass distribution for the full data sample, where the large peak appearing to the left o f the D + and D~+ signals is due to the reflection of D + - - , K - g +g... simulated data for F ( D+ ~ K - K + K + ) 0.05 7-+ 0.020_+0.007 F(D +-, K-rt+rt + ) F(D + ,K-K+K +) 0.49 + _+0.06 F ( D + , (bn + ) D + - - , K - n +n + When considering signals other than D+~K-n+n