21 October 1999 Physics Letters B 465 Ž1999. 335–348 A new measurement of direct CP violation in two pion decays of the neutral kaon NA48 Collaboration V. Fanti, A. Lai, D. Marras, L. Musa 1 Dipartimento di Fisica dell’UniÕersita` e Sezione dell’INFN di Cagliari, I-09100 Cagliari, Italy A.J. Bevan, T.J. Gershon, B. Hay, R.W. Moore 2 , K.N. Moore 2 , D.J. Munday, M.D. Needham 3, M.A. Parker, S.F. Takach 4 , T.O. White, S.A. Wotton CaÕendish Laboratory, UniÕersity of Cambridge, Cambridge, CB3 0HE, UK 5 6 G. Barr, H. Blumer , G. Bocquet, J. Bremer, A. Ceccucci, J. Cogan 7, D. Cundy, ¨ N. Doble, W. Funk, L. Gatignon, A. Gianoli 8, A. Gonidec, G. Govi, P. Grafstrom, ¨ 9 G. Kesseler, W. Kubischta, A. Lacourt, S. Luitz , J.P. Matheys, A. Norton, S. Palestini, B. Panzer-Steindel, B. Peyaud 7, D. Schinzel, H. Taureg, M. Velasco, O. Vossnack, H. Wahl, G. Wirrer CERN, CH-1211 GeneÕa 23, Switzerland A. Gaponenko, V. Kekelidze, D. Madigojine, A. Mestvirishvili 10 , Yu. Potrebenikov, G. Tatishvili, A. Tkatchev, A. Zinchenko Joint Institute for Nuclear Research, Dubna, Russian Federation L. Bertolotto 1, O. Boyle, I.G. Knowles, V.J. Martin, H.L.C. Parsons, K.J. Peach, C. Talamonti 11 Department of Physics and Astronomy, UniÕersity of Edinburgh, JCMB King’s Buildings, Mayfield Road, Edinburgh, EH9 3JZ, UK 5 M. Contalbrigo, P. Dalpiaz, J. Duclos, A. Formica 7, P.L. Frabetti 12 , M. Martini, F. Petrucci, M. Savrié Dipartimento di Fisica dell’UniÕersita` e Sezione dell’INFN di Ferrara, I-44100 Ferrara, Italy 0370-2693r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 0 - 2 6 9 3 Ž 9 9 . 0 1 0 3 0 - 8 336 V. Fanti et al.r Physics Letters B 465 (1999) 335–348 A. Bizzeti 13, M. Calvetti, G. Collazuol, G. Graziani, E. Iacopini, M. Lenti, A. Michetti Dipartimento di Fisica dell’UniÕersita` e Sezione dell’INFN di Firenze, I-50125 Firenze, Italy H.G. Becker, P. Buchholz 14 , D. Coward 9, C. Ebersberger, H. Fox, A. Kalter, K. Kleinknecht, U. Koch, L. Kopke, B. Renk, J. Scheidt, J. Schmidt, V. Schonharting, ¨ ¨ Y. Schue, ´ R. Wilhelm, M. Wittgen Institut fur ¨ Physik, UniÕersitat ¨ Mainz, D-55099 Mainz, Germany 15 J.C. Chollet, S. Crepe, ´ ´ L. Fayard, L. Iconomidou-Fayard, J. Ocariz 16, G. Unal, D. Vattolo, I. Wingerter-Seez Laboratoire de l’Accelerateur Lineaire, IN2P3-CNRS, UniÕersite´ de Paris-Sud, F-91406 Orsay, France 17 ´´ ´ G. Anzivino, F. Bordacchini, P. Cenci, P. Lubrano, A. Nappi, M. Pepe, M. Punturo Dipartimento di Fisica dell’UniÕersita` e Sezione dell’INFN di Perugia, I-06100 Perugia, Italy L. Bertanza, A. Bigi, P. Calafiura 18, R. Carosi, R. Casali, C. Cerri, M. Cirilli, F. Costantini, R. Fantechi, S. Giudici, B. Gorini 1, I. Mannelli, V. Marzulli, G. Pierazzini, F. Raffaelli, M. Sozzi Dipartimento di Fisica dell’UniÕersita, ` Scuola Normale Superiore e Sezione dell’INFN di Pisa, I-56100 Pisa, Italy J.B. Cheze, M. De Beer, P. Debu, R. Granier de Cassagnac, P. Hristov 19, E. Mazzucato, S. Schanne, R. Turlay, B. Vallage DSM r DAPNIA - CEA Saclay, F-91191 Gif-sur-YÕette, France I. Augustin 1, M. Bender, M. Holder, M. Ziolkowski Fachbereich Physik, UniÕersitat ¨ Siegen, D-57068 Siegen, Germany 2 0 R. Arcidiacono, C. Biino, R. Cester, F. Marchetto, E. Menichetti, N. Pastrone Dipartimento di Fisica Sperimentale dell’UniÕersita` e Sezione dell’INFN di Torino, I-10125 Torino, Italy J. Nassalski, E. Rondio, M. Szleper, W. Wislicki, S. Wronka Soltan Institute for Nuclear Studies, Laboratory for High Energy Physics, PL-00-681 Warsaw, Poland 21 H. Dibon, G. Fischer, M. Jeitler, M. Markytan, I. Mikulec, G. Neuhofer, M. Pernicka, A. Taurok Osterreichische Akademie der Wissenschaften, Institut fur ¨ Hochenergiephysik, A-1050 Wien, Austria 2 2 Received 30 August 1999; accepted 31 August 1999 Editor: L. Montanet V. Fanti et al.r Physics Letters B 465 (1999) 335–348 337 Abstract The NA48 experiment at CERN has performed a new measurement of direct CP violation, based on data taken in 1997 by simultaneously collecting K L and K S decays into p 0p 0 and pq py. The result for the CP violating parameter Re Ž ´ Xr´ . is Ž18.5 " 4.5Žstat. " 5.8Žsyst.. = 10y4 . q 1999 Elsevier Science B.V. All rights reserved. 1 Present address: CERN, CH-1211 Geneva 23, Switzerland. Present address: Physics-Astronomy Building, Michigan State University, East Lansing, MI 48824 USA. 3 Present address: NIKHEF, PO Box 41882, 1009 DB Amsterdam, The Netherlands. 4 Present address: Department of Physics, Wayne State University, Detroit, MI 48201, USA. 5 Funded by the UK Particle Physics and Astronomy Research Council. 6 Present address: Universitat ¨ Karlsruhe ŽTH., Fakultat ¨ fur ¨ Physik, and Forschungszentrum Karlsruhe GmbH, Institut fur ¨ Kernphysik, D-76128 Karlsruhe. 7 Present address: DSMrDAPNIA - CEA Saclay, F-91191 Gif-sur-Yvette, France. 8 Present address: Dipartimento di Fisica dell’Universita` e Sezione dell’INFN di Ferrara, I-44100 Ferrara, Italy. 9 Present address: SLAC, Stanford, CA 94309, USA. 10 Present address: Dipartimento di Fisica dell’Universita` e Sezione dell’INFN di Perugia, I-06100 Perugia, Italy. 11 Present address: Dipartimento di Fisiopatologia Clinica dell’Universita` di Firenze, I-50134 Firenze, Italy. 12 Permanent address: Dipartimento di Fisica dell’Universita` e Sezione dell’INFN di Bologna, I-40126 Bologna, Italy. 13 Also at Dipartimento di Fisica dell’Universita` di Modena, I-41100 Modena, Italy. 14 Present address: Institut fur ¨ Physik, Universitat ¨ Dortmund. D-44221 Dortmund, Germany. 15 Funded by the German Federal Minister for Research and Technology ŽBMBF. under contract 7MZ18PŽ4.-TP2. 16 Permanent address: Departamento de Fısica, Universidad de ´ los Andes, Merida 5101-A, Venezuela. ´ 17 Funded by the Institut National de Physique des Particules et ŽIN2P3., France. de Physique Nucleaire ´ 18 Present address: E.O. Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA. 19 Present address: Joint Institute for Nuclear Research, Dubna, Russian Federation. 20 Funded by the German Federal Minister for Research and Technology ŽBMBF. under contract 056SI74. 21 Supported by the Committee for Scientific Research grant 2P03B07615 and using computing resources of the Interdisciplinary Center for Mathematical and Computational Modelling of the University of Warsaw. 22 Funded by the Austrian Ministery for Traffic and Research under the contract GZ 616.360r2-IV GZ 616.363r2-VIII, and by the Fonds fur ¨ Wissenschaft und Forschung FWF Nr. P08929-PHY. 2 1. Introduction CP violation has been observed so far only in the decays of neutral kaons w1x. The main component of the effect w2x occurs in the mixing between K 0 and K 0 eigenstates. The physical states K S and K L deviate from pure CP s "1 eigenstates, with the mixing described by the parameter ´ . On the other hand, direct CP violation can occur in the transition from the neutral kaon eigenstate with CP s y1 to a two-pion final state, which has CP s q1. This can be observed through the interference of the amplitudes of different final state isospin w3x and is described by the parameter ´ X . The measured quantity is the double ratio of decay widths: Rs G Ž K L ™ p 0p 0 . r G Ž K S ™ p 0p 0 . G Ž K L ™ pq py . r G Ž K S ™ pq py . , 1 y 6 = Re Ž ´ Xr´ . . Ž 1. The Standard Model relates CP violation to the existence of three generations of quarks and to the complex phase present in the CKM matrix w4x. Computations based on the Standard Model typically predict Re Ž ´ Xr´ . in the range 0 to 10 = 10y4 w5x. The first evidence for direct CP violation was published in 1988 w6x. Subsequent measurements Ž23.0 " 6.5. = 10y4 w7x and Ž7.4 " 5.9. = 10y4 w8x have been only marginally consistent with each other; the measurement in w7x shows evidence for direct CP violation while that in w8x is consistent with no effect. The KTeV collaboration has recently measured Ž28.0 " 4.1. = 10y4 w9x using a technique similar to the one used in w8x. In this letter, a new measurement of Re Ž ´ Xr´ . obtained by the NA48 collaboration at CERN is presented, based on the first data collected with the experiment in 1997. While the ultimate accuracy of the NA48 experiment will be significantly better than that reported in this study, this result is of 338 V. Fanti et al.r Physics Letters B 465 (1999) 335–348 sufficient precision to help clarify the current experimental situation. 2. Principle of measurement The measurement of Re Ž ´ Xr´ . proceeds by counting the number of events in each of the four decay modes and computing the double ratio R. The NA48 experiment is designed in such a way that many systematic uncertainties can be controlled with high accuracy by exploiting cancellations of effects which contribute symmetrically to different components of the double ratio. Data are collected simultaneously in the four decay channels, minimizing the sensitivity of the measurement to accidental activity and to variations in beam intensity and detection efficiency. K L and K S decays are provided by two nearly collinear beams, with similar momentum spectra, converging in the centre of the main detector. In order to reduce the difference in acceptance due to the large difference in average decay lengths, only K L decays occurring in the region also populated by K S decays are used for the measurement of the double ratio. Furthermore, a cancellation of the residual difference is obtained by weighting the K L events used in the double ratio with a function of the proper lifetime t , proportional to the expected ratio of K S and K L decays at time t . In this way, the systematic accuracy of the result does not rely on a detailed Monte Carlo simulation of the experiment. 3. Beam and detector 3.1. The kaon beams The neutral beams w10x are derived from 450 GeVrc protons extracted from the CERN SPS. Due to the different mean decay lengths Ž l L s 3.4 km, lS s 5.9 m at the average kaon momentum of 110 GeVrc ., two separate production targets are used, located 126 m and 6 m upstream of the beginning of the decay region. For each SPS pulse Ž2.4 s spill every 14.4 s., 1.1 = 10 12 protons hit the K L production target. A neutral secondary beam, with " 0.15 mrad divergence is generated at a production angle of 2.4 mrad. Three stages of collimation are used with the final element located just upstream of the beginning of the decay region. Part of the non-interacting primary protons impinge on a bent silicon mono-crystal w11x. A small fraction of these protons undergo channelling, and are bent to produce a collimated beam of 3 = 10 7 protons per pulse transported to the K S production target. The K S collimator selects secondary neutral particles at 4.2 mrad production angle, with a divergence of " 0.375 mrad. The different production angles for the K L and K S beams are chosen to minimize the difference in decay momentum spectra. The K S beam enters the fiducial decay volume 68 mm above the K L beam. The beams converge with an angle of 0.6 mrad and the axes of the two beams cross at the position of the electromagnetic calorimeter. The total flux of K L through the fiducial region is f 1.5 = 10 7 per spill, and the K S flux is f 2 = 10 2 per spill. The protons directed to the K S target are detected by an array of scintillation counters which comprise the tagging detector w12x. This device is used to tag K S decays. The time of any protons and the time of the event in the main detector are each reconstructed relative to a common free running clock. The presence Žabsence. of a proton reconstructed in coincidence with the event defines the event as K S Ž K L .. The signals are digitized by 960 MHz FADCs. The proton time resolution is 120 ps and the double-pulse separation is 4 ns. The neutral beam produced at the K S target traverses an anti-counter ŽAKS., formed by a set of three scintillation counters preceded by an aligned 3 mm thick iridium crystal which enhances photon conversions Ž1.8 effective radiation lengths. w13x. This detector is used to veto K S decays occurring upstream, thus providing an accurate definition of the beginning of the decay region. The decay region is contained in an evacuated Ž- 3 = 10y5 mbar. 90 m long tank with a 0.9 mm Ž0.003 radiation lengths. thick polyimide ŽKevlar. composite window at the end. The window extends from an inner diameter of 18.4 cm to an outer diameter of 2.3 m. The neutral beam continues in a 16 cm diameter evacuated tube to a Ž K L . beam V. Fanti et al.r Physics Letters B 465 (1999) 335–348 monitor and the beam dump, both downstream of all the detector elements. 3.2. Main detector The layout of the main detector is shown in Fig. 1. Kaons which decay into charged pions are measured by a magnetic spectrometer comprised of four drift chambers w14,15x, and a dipole magnet giving a horizontal transverse momentum kick of 265 MeVrc. The volume between the drift chambers is filled with helium at atmospheric pressure. The drift chambers are each comprised of eight planes of sense wires, two horizontal, two vertical and two along each of the 458 directions. Only horizontal and vertical planes are instrumented in DCH3. The resolution in each projection is 90 m m and the average efficiency is 99.5% per plane. The geometrical accuracy is better than 0.1 mmrm. The momentum resolution is sprp , 0.5% [ 0.009% = p Ž p in GeVrc ., where [ means that the contributions should be added in quadrature. Two plastic scintillator hodoscope planes are placed after the helium tank. They are segmented in horizontal and vertical strips, respectively, with widths varying between 6.5 cm and 10 cm, arranged in 4 quadrants. Fast logic provides a signal for the first level of the pqpy trigger. This trigger requires the presence of tracks in two opposite quadrants of the hodoscope. The hodoscope data are also used in 339 the analysis to determine the event time of pqpy events for the tagging procedure. The time resolution is , 200 ps per track. A quasi-homogeneous liquid krypton electromagnetic Že.m.. calorimeter with 127 cm Ž27 radiation lengths. long projective tower readout is used to measure the photons from p 0p 0 events. The electrode structure consists of copperrberyllium ribbons extending between the front and the back of the detector with a " 48 mrad accordion geometry. The 13212 readout cells each have a cross section of 2 = 2 cm2 at the back of the active region. The projective towers point to the average K S decay position Ž110 m upstream of the detector. so that the measurement of the photon positions is insensitive to the initial conversion depth. The transverse scale at the back plane of the calorimeter is measured to better than 0.2 mmrm. The initial current induced on the electrodes by the drift of the ionization is measured using pulse shapers with 80 ns FWHM and digitized with 40 MHz FADCs w16x. The energy resolution is s Ž E .rE , 0.125rE [ 0.032r 'E [ 0.005 Ž E in GeV.. The spatial resolution is better than 1.3 mm and the time resolution is better than 300 ps for 20 GeV photons. The e.m. calorimeter incorporates a detector made of a 4 mm thick plane of scintillating fibres located near the maximum of the shower development. This device provides an independent trigger for p 0p 0 decays, and an additional measurement of the event time. An iron-scintillator hadron calorimeter, 6.7 nuclear interaction lengths thick, is used to complement the krypton calorimeter to measure the energies of hadrons in the trigger. Three planes of scintillation counters, segmented into 25 cm wide counters, shielded by 80 cm thick iron walls, are used for muon identification to reduce the background from K L ™ pmn Ž Km 3 . events. Seven counter arrays Žindicated as ANTI 1 – ANTI 7 on Fig. 1., consisting of plastic scintillators and iron converters, surround the vacuum and helium tanks to detect photons which miss the calorimeter. In the present analysis, they are only used to monitor accidental activity. 3.3. Trigger and DAQ Fig. 1. Layout of the main detector components. The readout of the e.m. calorimeter sends the calibrated analogue sums of 2 = 8 cells to the p 0p 0 340 V. Fanti et al.r Physics Letters B 465 (1999) 335–348 trigger. The p 0p 0 trigger w17x digitizes the data and operates as a 40 MHz pipelined computation system with a latency of 3 ms. The calorimeter data are reduced to x and y projections, each consisting of 64 strips, which are used to reconstruct the total deposited energy E0 , the first and second moments of the energy distribution and to reconstruct and measure the time of energy peaks. This information is used to reconstruct the radial position of the centre of gravity C of the event and the proper decay time trt S in units of the mean K S lifetime t S . Events are selected if E0 ) 50 GeV, C - 15 cm, trt S - 5.5 and if there are no more than five peaks in both views in a 9 ns time window. The p 0p 0 trigger rate is about 4500 per pulse and the deadtime is negligible. A two level pqpy trigger is used. At the first level, the hodoscope trigger is placed in coincidence with a total energy condition ŽG 30 GeV., defined by adding the hadron calorimeter energy to E0 from the p 0p 0 trigger. The first level trigger is downscaled by a factor of two, the resulting trigger rate is about 100,000 per pulse. The second level pqpy trigger w18x uses information from the drift chambers. It consists of hardware coordinate builders for each view and a farm of asynchronous microprocessors which reconstructs tracks. Triggers are selected if the tracks converge to within 5 cm, their opening angle is less than 15 mrad, the reconstructed proper decay time is less than 4.5 t S and the reconstructed kaon mass is at least 95% of the nominal K 0 mass. The maximum decision latency is 100 ms. The second level trigger rate is about 2000 per pulse. Downscaled triggers, based on relaxed conditions are collected to compute the efficiency of the main triggers. ‘‘Random events’’ record any activity in the detectors 69 ms after a heavily downscaled beam counter signal. The 69 ms delay corresponds to the periodicity of the slow proton extraction Žthree revolution periods. of the SPS accelerator. These events contain signals which are representative of accidental activity in the detectors. The overall deadtime of the trigger system is below 1%, the largest contribution being due to the pqpy trigger. The deadtime condition is recorded and made completely symmetric offline. The detector readout systems provide information on all the activity in the detectors in a readout window of about 250 ns surrounding the trigger time. The read- out window for the drift chambers is larger, about 1.2 ms. 4. Event reconstruction and selection 4.1. p 0p 0 mode Photon showers are found in the calorimeter by looking for maxima in the digitized pulses from individual cells in both space and time and accumulating the energy within a radius of 11 cm. Photon energies are corrected for: energy outside the cluster boundary; energy lost in non-working cells Žabout 0.4% of the channels.; energy from the other photons in the event; small variations of the energy measurement depending on the impact point within the cell and small space charge effects from the accumulation of positive ions w19x because the calorimeter was operated at half the nominal 3 kV drift voltage in 1997. K ™ p 0p 0 decays are selected by requiring four clusters with energies between 3 GeV and 100 GeV which are in time within " 5 ns of the average of the four. Fiducial cuts are applied to ensure that photon energies are well measured. These cuts include the removal of photons within 2 cm of a non-working cell and photons within a " 4 cm wide column due to a failure in a HV connection during the 1997 run. The minimum distance between photon candidates is required to be greater than 10 cm. To reduce the background from K L ™ 3 p 0 decays, events with an additional cluster of energy above 1.5 GeV and within " 3 ns of the p 0p 0 candidate are rejected. The kaon energy is obtained with a resolution of ; 0.6% from the sum of the photon energies. The longitudinal position of the decay vertex relative to the front of the calorimeter D is reconstructed from the energies Ei and positions x i , yi of the four selected clusters, assuming that their invariant mass is the kaon mass Ž m K . as follows: Ds (Ý Ý i 2 Ei E j Ž x i y x j . q Ž y i y y j . 2 r mK j)i Ž 2. and is used to reconstruct the proper decay time t . The resolution of D is 50 to 70 cm depending on V. Fanti et al.r Physics Letters B 465 (1999) 335–348 energy. The invariant masses m1 and m 2 of the two photon pairs are computed using D and compared to the nominal p 0 mass Ž mp 0 .. For this, a x 2 variable is computed as follows: x 2s Ž m1 q m 2 . r2 y mp 0 sq 2 q Ž m1 y m 2 . r2 2 sy Ž 3. where s " are the resolutions of Ž m1 " m 2 .r2 observed in the data Žparameterised as a function of the lowest photon energy.. The three-fold choice of photon pairings is made by selecting the combination with the lowest x 2 . Due to the constraint on m K , Ž m1 " m 2 . are to good approximation uncorrelated. Typical values of sq and sy are 0.45 and 1.1 MeVrc 2 . To select good p 0p 0 candidates, the best combination is required to have x 2 - 13.5. This requirement removes about 7% of K S ™ p 0p 0 events in a pure K S beam where there is no background. About 70% of these are due to p 0 Dalitz decays and to photon conversions in the Kevlar window or in the spectrometer volume upstream of the magnet. The remaining part is due to the e.m. calorimeter response, and is dominated by occasional photon–nucleon interactions, which result in a reduction of the observed energy. Both effects are highly symmetric between K S and K L , and do not affect the measurement of the double ratio. The event time as used in the tagging procedure for p 0p 0 events is computed combining eight time estimators from the two most energetic cells of each cluster. 4.2. p qp y mode Tracks are reconstructed from the hits and drift times in the spectrometer. A vertex is defined at the point of closest approach of two oppositely charged tracks Žprovided they approach to within 3 cm.. The momenta of the tracks are derived using a detailed magnetic field map. The energy of the event is reconstructed using the angle between the tracks before the magnet and the ratio of the track momenta, assuming the two tracks originate from K ™ pqpy. This avoids systematic uncertainties from the knowledge of the magnetic field in the definition of the fiducial energy cut and lifetime measurement. 341 Small corrections on the tracks are made for the field Žintegral ; 2 = 10y3 T m. in the decay region before the spectrometer. The vertex resolution is typically 50 cm in the longitudinal direction z, and 2 mm in the transverse directions x, y. Since the beams are separated by about 6 cm in the decay region, a clean identification of K S ™ pqpy and K L ™ pqpy decays is possible Žthis is referred to as ‘‘ y-vertex tagging’’ in the following.. A variable A related to the decay orientation in the kaon rest frame is defined from the two track momenta p 1 and p 2 as A s < p 1 y p 2