Search for Charged Current Coherent Pion Production on Carbon in a Few-GeV Neutrino Beam K. Hiraide,10 J. L. Alcaraz-Aunion,1 S. J. Brice,4 L. Bugel,13 J. Catala-Perez,18 G. Cheng,3 J. M. Conrad,13 Z. Djurcic,3 U. Dore,15 D. A. Finley,4 A. J. Franke,3 C. Giganti∗,15 J. J. Gomez-Cadenas,18 P. Guzowski,6 A. Hanson,7 Y. Hayato,8 G. Jover-Manas,1 G. Karagiorgi,13 T. Katori,7 Y. K. Kobayashi,17 T. Kobilarcik,4 H. Kubo,10 Y. Kurimoto,10 W. C. Louis,11 P. F. Loverre,15 L. Ludovici,15 K. B. M. Mahn,3 C. Mariani†,15 S. Masuike,17 K. Matsuoka,10 W. Metcalf,12 G. Mills,11 G. Mitsuka,9 Y. Miyachi,17 S. Mizugashira,17 C. D. Moore,4 Y. Nakajima,10 T. Nakaya,10 R. Napora,14 P. Nienaber,16 V. Nguyen,13 D. Orme,10 M. Otani,10 A. D. Russell,4 F. Sanchez,1 M. H. Shaevitz,3 T.-A. Shibata,17 M. Sorel,18 R. J. Stefanski,4 H. Takei,17 H.-K. Tanaka,3 M. Tanaka,5 R. Tayloe,7 I. J. Taylor,6 R. J. Tesarek,4 Y. Uchida,6 R. Van de Water,11 J. J. Walding,6 M. O. Wascko,6 H. White,4 M. J. Wilking,2 M. Yokoyama,10 G. P. Zeller,11 and E. D. Zimmerman2 9 (The SciBooNE Collaboration) 0 1Institut de Fisica d’Altes Energies, Universitat Autonoma de Barcelona, E-08193 Bellaterra (Barcelona), Spain 0 2Department of Physics, University of Colorado, Boulder, Colorado 80309, USA 2 3Department of Physics, Columbia University, New York, NY 10027, USA n 4Fermi National Accelerator Laboratory; Batavia, IL 60150, USA a 5High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305-0801, Japan J 6Department of Physics, Imperial College London, London SW7 2AZ, UK 5 7Department of Physics, Indiana University, Bloomington, IN 47405, USA 8Kamioka Observatory, Institute for Cosmic Ray Research, University of Tokyo, Gifu 506-1205, Japan ] 9Research Center for Cosmic Neutrinos, Institute for Cosmic Ray x Research, University of Tokyo, Kashiwa, Chiba 277-8582, Japan e - 10Department of Physics, Kyoto University, Kyoto 606-8502, Japan p 11Los Alamos National Laboratory; Los Alamos, NM 87545, USA e 12Department of Physics and Astronomy, Louisiana State University, Baton Rouge, LA 70803, USA h 13Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA [ 14Department of Chemistry and Physics, Purdue University Calumet, Hammond, IN 46323, USA 2 15Universita di Roma La Sapienza, Dipartmento di Fisica and INFN, 1-000185 Rome, Italy v 16Physics Department, Saint Mary’s University of Minnesota, Winona, MN 55987, USA 9 17Department of Physics, Tokyo Institute of Technology, Tokyo 152-8551, Japan 6 18Instituto de Fisica Corpuscular, Universidad de Valencia and CSIC, E-46071 Valencia, Spain 3 (Dated: January 5, 2009) 0 TheSciBooNECollaborationhasperformedasearchforchargedcurrentcoherentpionproduction . 1 from muon neutrinosscattering on carbon, ν 12C→µ−12Cπ+,with twodistinct datasamples. No µ 1 evidencefor coherent pionproduction isobserved. Weset 90% confidencelevelupperlimits on the 8 cross section ratio of charged current coherent pion production to the total charged current cross 0 section at 0.67×10−2 at mean neutrino energy 1.1 GeV and 1.36×10−2 at mean neutrino energy : v 2.2 GeV. i X PACSnumbers: 13.15.+g,13.60.Le,25.30.Pt,95.55.Vj r a I. INTRODUCTION [1, 2]. Several interaction channels contribute to the to- tal neutrino-nucleus cross section in the neutrino energy range of a few GeV. Interactions producing single pions Althoughtheyhavebeenstudiedfordecades,neutrino- (chargedorneutral)accountforalargecrosssectionfrac- nucleus cross sections between 100 MeV and 10 GeV tion, which must be understood because they form sig- energy are still known with very poor accuracy. The nificant backgrounds for neutrino oscillation searches. demand for precise cross section measurements in this Ithasbeenknownforyearsthatneutrinoscanproduce energy regime is driven by the needs of the next genera- pionsbyinteractingcoherentlywiththenucleonsforming tionofneutrinooscillationexperimentsintheirpursuitof the target nucleus. The cross section for this process is sub-leading flavor oscillation and charge-parityviolation expected to be smaller than incoherent pion production, thelatterbeingdominatedbyneutrino-inducedbaryonic resonance excitation off a single nucleon bound in a nu- cleus. Moreover, coherent pion production is compara- ∗Present address: DSM/Irfu/SPP, CEA Saclay, F-91191 Gif-sur- tively poorly understood, although it is characterizedby Yvette, France †Present address: Department of Physics, Columbia University, a distinct signature consisting of a nucleus left in the NewYork,NY10027, USA ground state (no nuclear breakup occurs) and a forward 2 scattered pion. Both charged current and neutral cur- in the SciBooNE experiment [22]. This is a particularly rent coherent modes are possible, ν A → µ−Aπ+ and interesting test of the K2K null result, probing a similar µ ν A→ν Aπ0, where A is a nucleus. neutrino energy range and same target material. Also, µ µ Several theoretical models describing coherent pion compared to K2K, SciBooNE’s result presented here is production have been proposed, using different for- basedon a higher-statistics data sample and uses an im- malisms to describe the relevant physics. A first class proved analysis, as will be described below. of models is built on the basis of Adler’s PCAC the- Thispaperisorganizedasfollows. SectionIIdescribes orem [3], relating the neutrino-nucleus cross section to the neutrino beam-line andthe neutrino flux simulation. that of a pion interacting with a nucleus at Q2 = 0; the The simulation of neutrino interactions with nuclei are extrapolation to Q2 6= 0 is performed via a propagator described in Section III. The detector configuration and term [4, 5, 6, 7, 8]. A second commonly-used formal- simulation are described in Section IV. A summary of ism is based on the description of the coherent produc- the data set and experimental performance is given in tion of ∆ resonances on nuclei by using a modified ∆- Section V. The data analysis, including the event selec- propagator and a distorted wave-function for the pion tionandMonteCarlo(MC)tuning,isdescribedindetail [9, 10, 11, 12]. While the relationship between neutral in Section VI. The results of the analysis and discussion current and charged current modes, and that between are presented along with a summary of systematic un- neutrinoandantineutrinocoherentpionproductioncross certainties in Section VII, and the final conclusions are sections, are relatively well known, order-of-magnitude given in Section VIII. variationsonabsolutecoherentpionproductioncrosssec- tions are expected within these models. In addition, the cross section dependence on neutrino energy and on tar- II. NEUTRINO BEAM get material is also uncertain. It is therefore imperative that more experimental input on coherent pion produc- The SciBooNE detector has been exposed to the tion in neutrino-nucleus interactions is gathered in the BoosterNeutrinoBeam(BNB)locatedatFermilab. The near future. BNB is a high-intensity, conventional neutrino beam Coherent pion production in neutrino-nucleus interac- whichhasbeenservingtheMiniBooNEexperimentsince tions has already been the subject of several experimen- 2002. tal campaigns. The neutrino energy range between 1 and 100 GeV has been investigated, including both the charged current and neutral current modes, and using A. Beam-line Description bothneutrinoandantineutrinoprobes. Aresultthathas drawnmuchattentionintheneutrinophysicscommunity has been the recent non-observation of charged current The primary beam uses protons accelerated to 8 GeV coherent pion production by the K2K experiment with kineticenergybytheFermilabBooster. Selectedbatches a 1.3 GeV wide-band neutrino beam [13]. This is moti- containingapproximately4-5×1012protonsareextracted vatedbythefactthattheK2KCollaborationhasquoted andbenttowardtheBNBtargethallviadipolemagnets. an upper limit for the ratio of the charged current co- Eachspillis composedof81bunches of protons,approx- herent pion production cross section to the charged cur- imately 6 ns wide each and 19 ns apart, for a total spill rent inclusive cross section that is well below the predic- duration of 1.6 µs. tion of the original Rein-Sehgal model [5] that has been Beam proton trajectories and positions are monitored adoptedinthepasttodescribecoherentpionproduction on a pulse-by-pulse basis. The typical beam alignment processes. In addition, even within more recently pro- anddivergence measuredby the beam positionmonitors posed models, it is often difficult to reconcile this new locatednearthetargetarewithin1mmand1mradofthe and accurate null result at low energies with previous nominaltargetcenterandaxisdirection,respectively;the measurements. On the one hand, evidence for neutral typicalbeamfocusingontargetmeasuredbybeamprofile current coherent pion production in a neutrino energy monitors is of the order of 1-2 mm (RMS) in both the range that is similar to K2K has been unambiguously horizontal and vertical directions. These parameters are reported from the Aachen-Padova [14] and Gargamelle wellwithintheexperimentrequirements. Thenumberof experimental data [15] first, and more recently also by protonsdeliveredtotheBNBtargetismeasuredforeach the MiniBooNE Collaboration [16]. On the other hand, proton batch using two toroids located near the target while no measurements of chargedcurrentcoherent pion along the beam-line. The toroid calibration, performed productionotherthantheK2Koneexistinasimilarneu- onapulse-by-pulsebasis,providesameasurementofthe trino energy range, there exist charged current coherent number of protons to BNB with a 2% accuracy. pion production positive results at higher energies (7- Primary protons from the 8 GeV beamline strike a 100GeVneutrinoenergy)fromtheSKAT[17],CHARM thick beryllium target located in the BNB target hall. [18], BEBC [19, 20], and FNAL E632 [21] experiments. Hadronicinteractionsofthe protonswith the targetma- In this paper, we discuss the first measurement of terial produce a beam of secondary mesons (pions and charged current coherent pion production by neutrinos kaons). The target is made of seven cylindrical slugs for 3 a total target length of 71.1 cm, or about 1.7 inelastic from meson and muon decays, and for obtaining the fi- interaction lengths. nal neutrino fluxes extrapolated to the SciBooNE detec- The beryllium target is surrounded by a magnetic fo- tor with negligible beam Monte Carlo statistical errors. cusing horn, bending and sign-selecting the secondary Currentbestknowledgeofneutrino-producingmesonand particles that emerge from the interactions in the tar- muon decay branching fractions, and decay form factors get along the direction pointing to the SciBooNE detec- in three-body semi-leptonic decays, are used. Polariza- tor. The focusing is produced by the toroidal magnetic tion effects in muon decays are also accounted for. field present in the air volume between the horn’s two Once produced by the simulation, neutrinos are ex- coaxial conductors made of aluminum alloy. The horn trapolated along straightlines towardthe SciBooNE de- current pulse is approximately a half-sinusoid of ampli- tector. All neutrinos whose ray traces cross any part of tude 174 kA, 143 µs long, synchronized to each beam the detector volume are considered for SciBooNE flux spill. The polarity of the horn current flow can be (and predictions. Basedonaccuratesurveydata,the distance has been) switched, in order to focus negatively-charged between the center of the beryllium target and the cen- mesons,andthereforeproduceanantineutrinoinsteadof terofthe SciBardetectoris takento be99.9m,withthe a neutrino beam. SciBooNE detector located on beam axis within a toler- Thebeamoffocused,secondarymesonsemergingfrom ance of a few cm. Each simulated neutrino interaction the target/horn region is further collimated via passive is linked to its detailed beam information and history, shielding, and allowedto decayinto neutrinos in a cylin- which includes neutrino flavor, energy, parent type and dricaldecayregionfilledwithairatatmosphericpressure, kinematics, and ray trace entry and exit points within 50mlongand90cminradius. Abeamabsorberlocated the detectorvolume; the raytraceinformationisusedto at the end of the decay region stops the hadronic and determine the incoming neutrino’s direction and inter- muonic component of the beam, and only a pure neu- action location. Proper weights for each beam neutrino trinobeampointingtowardthedetectorremains,mostly event are computed, using this beam neutrino informa- from π+ →µ+ν decays. tion, as well as information from the interaction and de- µ tector simulation: neutrino interaction probability, and detailedSciBooNEdetectorgeometryandspecifications. B. Neutrino Flux Prediction Neutrino flux predictions at the SciBooNE detector location are obtained via a GEANT4-based [23] beam T)10-9 O Monte Carlo simulation. The same simulation code de- P veloped by the MiniBooNE Collaboration is used [24]. eV/ n m all M In the simulation code, a realistic description of the 510-10 geometry and materials present in the BNB target hall 2/2m n m and decay region is used. Primary protons are gener- x (/c autpesdtreaacmcorodfinthgettoartgheet.eTxpheecitnetderbaecatimonospotficpsripmraorpyerptrieos- Flu10-11 n e tons with the beryllium target are simulated according tostate-of-the-arthadroninteractiondata. Ofparticular importance for this analysis is π+ production in proton- 10-12 beryllium interactions, which uses experimental input n e from the HARP [25] and BNL E910 [26] experiments. Production of secondary protons, neutrons, charged pi- 10-13 ons,andchargedandneutralkaonsistakenintoaccount, and elastic and quasi-elastic scattering of protons in the target are also simulated. Particles emanating from the 10-14 0 1 2 3 4 5 primary proton-beryllium interaction in the target are En (GeV) then propagated within the GEANT4 framework,which accounts for all relevant physics processes. Hadronic re- FIG.1: NeutrinofluxpredictionattheSciBooNEdetectoras interactions of pions and nucleons with beryllium and a function of neutrino energy E , normalized per unit area, ν aluminum materials are particularly important and are proton on target (POT) andneutrinoenergy bin width. The described by custom models, while other hadronic pro- spectrum is averaged within 2.12 m from the beam center. cesses and all electromagnetic processes (energy loss, Thetotalfluxandcontributionsfromindividualneutrinofla- multiple scattering, effect of horn magnetic field, etc.) vors are shown. aredescribedaccordingtodefaultGEANT4physicslists. A second, FORTRAN-based Monte Carlo code uses the outputofthe GEANT4programasinput, andisrespon- TheneutrinofluxpredictionattheSciBooNEdetector sibleforgeneratingtheneutrinokinematicsdistributions location and as a function of neutrino energy is shown 4 in Fig. 1. A total neutrino flux per proton on target s)100 of 2.2×10−8 cm−2 is expected at the SciBooNE detec- eu 90 cl torlocationandinneutrinorunningmode(positivehorn nu 80 polarity), with a mean neutrino energy of 0.7 GeV. The bon 70 fluxisdominatedbymuonneutrinos(93%oftotal),with ar C 60 smallcontributionsfrommuonantineutrinos(6.4%),and 2/m electronneutrinosandantineutrinos(0.6%intotal). For -40c0 4500 the neutrino flux predictions used in this analysis, no 1 C( information from BNB (SciBooNE or MiniBooNE) neu- C 30 s trino data is used as experimental input. 20 10 0 0 0.5 1 1.5 2 2.5 3 III. NEUTRINO INTERACTION SIMULATION En(GeV) Theneutrinointeractionswithnucleartargetsaresim- FIG.2: Cross section for νµ12C→µ−π+12Cinteraction. The solid line represents the Rein and Sehgal model with lepton ulated with the NEUT program library [27, 28] which mass effects [8](the default model of the signal for this anal- is used in the Kamiokande, Super-Kamiokande, K2K, ysis), the dashed line represents the Rein and Sehgal model and T2K experiments. NEUT handles protons, oxy- withoutleptonmasseffects[5],thedottedlinerepresentsthe gen, carbon, and iron as nuclear targets in the energy modelofKartavtsevetal.[7],andthedashed-dottedlinerep- range from 100 MeV to 100 TeV. In NEUT, the follow- resents the model of Alvarez-Ruso et al. [11]. The model of ing neutrino interactions in both neutral and charged Singh et al. [9] gives a cross section similar to the model of currents are simulated: quasi-elastic scattering (νN → Alvarez-Rusoet al. ℓN′), single meson production (νN → ℓN′m), single gamma production (νN → ℓN′γ), coherent π produc- tion(ν12C(or56Fe)→ℓπ12C(or56Fe)),anddeepinelastic mented using the model of Llewellyn-Smith [29]. For scattering(νN →ℓN′hadrons),whereN andN′ arethe scattering off nucleons in the nucleus, we use the rela- nucleons (proton or neutron), ℓ is the lepton, and m is tivistic Fermi gas model of Smith and Moniz [30]. The the meson. Following the primary neutrino interactions nucleonsaretreatedasquasi-freeparticlesandtheFermi innuclei,re-interactionsofthe mesonsandhadronswith motion of nucleons along with the Pauli exclusion prin- the nuclear medium are also simulated. ciple is taken into account. The momentum distribution of the target nucleon is assumed to be flat up to a fixed Fermi surface momentum of 217 MeV/c for carbon and A. Coherent π production 250 MeV/c for iron. The same Fermi momentum distri- butionisalsousedforalloftheothernuclearinteractions. The nuclear potential is set to 27 MeV for carbon and The signal for this analysis,coherentpion production, 32 MeV for iron. Both vector and axial-vector form fac- is a neutrino interaction with a nucleus which remains tor are assumed to be dipole. The vector mass in quasi- intact, releasing one pion with the same charge as the elastic scattering is set to be 0.84 GeV/c2. The axial incoming weak current. Because of the small momen- vector mass, M , is set to be 1.21 GeV/c2 as suggested tum transfer to the target nucleus the outgoing lepton A by recent results [31, 32]. and pion tend to go in the forward direction (in the lab frame). The formalismdevelopedbyRein andSehgal[5] is used to simulate the interactions, including the recent correction of lepton mass effects [8]. The axial vector C. Single meson production via baryon resonances mass, M , is set to 1.0 GeV/c2. The nuclear radius pa- A rameter R0 is set to 1.0 fm. For the total and inelastic The second most probable interaction in SciBooNE is pion-nucleon cross sections used in the formalism, the the resonantsingle meson production of π, K, and η de- fitted results given in Rein and Sehgal’s paper are em- scribedbythemodelofReinandSehgal[33]. Themodel ployed. Thetotalcrosssectionon12CisshowninFig.2, assumes an intermediate baryon resonance, N∗, in the with comparisonsof other models discussedin the intro- reaction of νN → ℓN∗,N∗ → N′m. The differential duction. The Rein and Sehgal model predicts charged cross section of single meson production depends on the current coherent pion production to be approximately amplitude for the production of a given resonance and 1% of the total neutrino interactions in SciBooNE. the probabilityofthe baryonresonancedecaytothe me- son. All intermediate baryon resonances with mass less than 2 GeV/c2 are included. Those baryon resonances B. Quasi-elastic scattering with mass greater than 2 GeV/c2 are simulated as deep inelastic scattering. Lepton mass effects from the non- The dominant interaction in the SciBooNE neutrino conservation of lepton current and the pion-pole term energy range is quasi-elastic scattering, which is imple- in the hadronic axial vector current are included in the 5 simulation [34, 35]. eventtype. Nucleon-nucleoninteractionsmodifytheout- To determine the angular distribution of a pion in the going nucleon’s momentum and direction. Both elastic final state, Rein’s method [36] is used for the P (1232) scattering and pion production are considered. In or- 33 resonance. For other resonances, the directional distri- der to simulate these interactions, a cascade model is bution of the generated pion is set to be isotropic in the againusedandthe generatedparticlesinthe nucleusare resonance rest frame. The angular distribution of π+ tracked using the same code as for the mesons. has been measured for ν p → µ−pπ+ [37] and the re- No de-excitation gamma-ray from the carbon nucleus µ sults agree well with NEUT’s prediction. Pauli blocking is simulated when nuclear breakup occurs. is accountedfor in the decay of the baryonresonance by requiringthemomentumofthenucleontobelargerthan the Fermi surface momentum. Pion-less ∆ decay is also IV. NEUTRINO DETECTOR takenintoaccount,where20%oftheeventsdonothavea pion and only the lepton and nucleonare generated[38]. The SciBooNE detector is located 100 m downstream The axial vector mass, M , is set to be 1.21 GeV/c2. A from the beryllium target on the axis of the beam. The detectorcomprisesthreesub-detectors: afullyactiveand finely segmentedscintillatortracker(SciBar),anelectro- D. Deep inelastic scattering magnetic calorimeter (EC), and a muon range detector (MRD). The cross section for deep inelastic scattering (DIS) is calculated using the GRV98 parton distribution func- tions[39]. Additionally,wehaveincludedthecorrections A. Detector Description inthesmallQ2regiondevelopedbyBodekandYang[40]. In the calculation, the hadronic invariant mass, W, is Fig. 3 shows an event display of a typical muon neu- required to be larger than 1.3 GeV/c2. Also, the mul- trino charged current single charged pion event candi- tiplicity of pions is restricted to be larger than or equal date. Detector coordinates are shown in the figure. Sci- to two for 1.3 < W < 2.0 GeV/c2, because single pion BooNEusesa right-handedCartesiancoordinatesystem production is already included in the simulation, as de- in which the z axis is the beam direction and the y axis scribed above. The multi-hadron final states are sim- is the vertical upward direction. The origin is located ulated with two models: a custom-made program [41] on the most upstream surface of SciBar in the z dimen- for the event with W between 1.3 and 2.0 GeV/c2 and PYTHIA/JETSET[42]fortheeventswithW largerthan 2 GeV/c2. m)150 c Y ( E. Intra-nuclear interactions 100 The intra-nuclear interactions of mesons and nucleons 50 produced in neutrino interactions in the nuclei are sim- ulated. These interactions are treated using a cascade 0 model,andeachofthe particlesis traceduntilitescapes from the nucleus. Amongalltheinteractionsofmesonsandnucleons,the -50 interactionsofpionsaremostimportanttothis analysis. Theinelasticscattering,chargeexchangeandabsorption -100 ofpionsinthenucleiaresimulated. Theinteractioncross sections of pions in the nuclei are calculated using the model by Salcedo et al. [43], which agrees well with past -150 experimental data [44]. If inelastic scattering or charge 0 50 100 150 200 250 300 350 exchange occurs, the direction and momentum of pions Z (cm) are determined by using results from a phase shift anal- FIG.3: Eventdisplayofatypicalmuonneutrinochargedcur- ysis of pion-nucleus scattering experiments [45]. When rent single charged pion event candidate in SciBooNE data. calculatingthepionscatteringamplitude,Pauliblocking The neutrino beam runs from left to right in this figure, en- istakenintoaccountbyrequiringthenucleonmomentum countering SciBar, the EC and MRD, in that order. The after the interaction to be larger than the Fermi surface circlesonSciBarandtheECindicateADChitsforwhichthe momentum at the interaction point. area of the circle is proportional to the energy deposition in Re-interactionsoftherecoilprotonsandneutronspro- that channel. Filled boxes in the MRD show ADC hits in duced in neutrino interactions are also important, be- time with thebeam window. cause the proton tracks are used to classify the neutrino 6 sion, and at the center of the SciBar scintillator plane ameter scintillating fibers embedded in lead foil. The in the x and y dimensions. Since each sub-detector is calorimeteris made of modules of dimensions 262 × 8 × read out both vertically and horizontally, two views are 4 cm3. Each module is read out by two green-extended defined; the top view (z-x projection) and the side view 1inchHamamatsuPMTsperside,256PMTstotal. The (z-y projection). modules were originally built for the CHORUS neutrino The SciBardetector [46]is positionedupstreamof the experiment at CERN [48] and later used in HARP and other sub-detectors. The primary role of SciBar is to then K2K. The modules construct one vertical and one reconstruct the neutrino-nucleus interaction vertex and horizontalplane,andeachplanehas32modules. TheEC detect charged particles produced by neutrino interac- hasathicknessof11radiationlengthsalongthebeamdi- tions. Moreover, SciBar is capable of particle identifica- rection. The planes coveranactive areaof2.7 ×2.6 m2. tionbasedondepositedenergy. SciBarwasdesignedand The charge information from each PMT is recorded. built as a near detector for the K2K experiment. After A minimum ionizing particle with a minimal path K2K’s completion, SciBar was relocatedto the Fermilab length deposits approximately 91 MeV in the EC. The BNB for SciBooNE. energy resolution for electrons was measured to be SciBar consists of 14,336 extruded plastic scintillator 14%/ E (GeV) using a test beam [48]. strips which serve as the target for the neutrino beam The MRD is installed downstream of the EC and is p as well as the active detection medium. Originally pro- designed to measure the momentum of muons produced duced by Fermilab, each strip has a dimension of 1.3 × bycharged-currentneutrino interactions. The MRDwas 2.5 × 300 cm3. The scintillators are arranged vertically constructed for SciBooNE at Fermilab, primarily out of and horizontally to construct a 3 × 3 × 1.7 m3 volume parts recycled from past experiments. It has 12 iron with a total mass of 15 tons. Eachstrip is readout by a plates with thickness of 5 cm which are sandwiched be- wavelengthshifting(WLS)fiberattachedtoa64-channel tween planes of 6 mm thick scintillation counters, 13 al- multi-anode PMT. Charge and timing information from ternating horizontal and vertical planes, which are read each PMT is recorded by front-end electronics boards out via 2 inch PMTs from a variety of past experiments; (FEB) attached directly to the PMT and a back-end thereare362PMTstotal. Eachironplatecoversanarea VME module [47]. The FEB uses VA/TA ASICs; the of 274 × 305 cm2. The total mass of absorber material VAhandleschargeinformationfromthePMTwitha32- is approximately 48 tons. The MRD measures the mo- channelpreamplifier chipwith ashaper andmultiplexer, mentum of muons up to 1.2 GeV/c using the observed while the TA provides timing information by taking the muon range. Charge and timing information from each logical“OR”of32channels. Thechargeandtiminginfor- PMT are recorded. Hit finding efficiency was continu- mationaredigitizedbyADCandmulti-hitTDCmodules ously monitored using cosmic ray data taken between on back-end electronics, and read out through the VME beam spills; the average hit finding efficiency is 99%. bus. SciBooNE has two global triggers, the beam trigger The gains of all PMT channels were measured prior andtheoff-beamtrigger. Twotypesofdataarecollected to installation in K2K. SciBar is equipped with a gain in one beam cycle, neutrino data with the beam trigger calibration system comprised of LEDs to monitor and andcalibrationdatawiththeoff-beamtrigger. Onecycle correctgaindriftduring the datataking;the gainstabil- is about 2 sec which is defined by the accelerator timing ity is monitored with precision better than 1%. Cosmic sequence. The BNB receives one train of proton beam ray data are also employed to calibrate the PMT gains pulses per cycle, with a maximum of 10 pulses in a row and scintillator light yield, including attenuation of the at 15 Hz. WLS-fibers. These calibration data, LED and cosmic, A fast timing signal sent by the extraction magnet on are taken between beam spills and continuously moni- BNB pulses establishes a beam-trigger. Once the beam tored. Calibration data verify the light yield was sta- triggerconditionisset,allsub-detectorsystemsreadout ble within 1% during operation. The timing resolution allchannelsirrespectiveofhitoccupancy(i.e. whetheror for minimum-ionizing particles was evaluated with cos- not a neutrino interaction occurred), ensuring unbiased mic ray data to be 1.6 ns. The average light yield for neutrino data. minimum-ionizing particles is approximately 20 photo- After the beam trigger turns off, the off-beam trig- electronsper1.3cmpathlength,andthetypicalpedestal ger condition is automatically set and each sub-detector width is below 0.3 photoelectron. The hit finding ef- takes calibration data. There are three types of calibra- ficiency evaluated with cosmic ray data is more than tion data: pedestal, LED (only for SciBar) and cosmic 99.8%. The minimum length of a reconstructable track ray data. The pedestal and LED data are collected once is approximately 8 cm (three layers hit in each view). per cycle. For cosmic ray data, there are two indepen- The track finding efficiency for single tracks of 10 cm or dent trigger blocks: SciBar/EC and MRD. SciBar and longer is more than 99%. the EC use a common cosmic ray trigger which is gen- The EC is located just downstream of SciBar, and is erated using fast signals from the TA. The MRD has its designedtomeasuretheelectronneutrinocontamination own cosmic ray trigger which is also self-generated by in the beam and tag photons from π0 decay. The EC is discriminator outputs. Both SciBar/EC and the MRD a “spaghetti” type calorimeter comprised of 1 mm di- collect 20 cosmic ray triggers in a cycle. 7 B. Detector simulation 0) 2 Delivered E 1 For analysis latTiohne.GTEhAeNBTer4tifnraimcaeswcaodrkeimsuosdeedl fwoirththinedGeEteActNoTr4sim[4u9-] get (x 2 is used to simulate the interactions of hadronic parti- ar n t cles with detector materials. The detector simulation o s includes a detailed geometric model of the detector, in- n o cludingthedetectorframeandexperimentalhallandsoil, ot 1 whichis basedonsurveymeasurementstakenduringde- Pr tector construction. In the detector simulation of SciBar, low level data 0 parametersareusedasinputtothe simulationwhenever Jun Jul Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug '07 '08 Date possible. Theenergylossofachargedparticleinasingle strip is simulated by GEANT, and this energy scale is tuned using cosmic ray data. Scintillator quenching is T) 30 O simulated using Birk’s law [50] with a value of Birk’s 6 P constant, measured for K2K, of 0.0208 cm/MeV [51]. E1 4 The energy deposited by a charged particle is converted e (/ 20 to photoelectrons using conversion factors measured for at each channel with cosmic muons. The measured light nt r e attenuation length of each fiber (approximately 350 cm v E on average)is usedin the simulation. Crosstalkbetween 10 nearbyMA-PMTchannelsissimulatedusingvaluesmea- sured in a test stand prior to installation. The number of photoelectrons is smeared by Poisson statistics, and the single photoelectron resolution of the MA-PMT is 0 Jun Jul OctNov Dec Jan Feb Mar Apr May Jun Jul Aug simulated. To simulate the digitization of the PMT sig- '07 '08 Date nal, the number of photoelectrons is converted to ADC counts, and then electronics noise and threshold effects FIG.4: Experimentalperformance. Inthetoppanel,thesolid of the TA are simulated. line shows thehistory of the accumulated numberof protons ontargetandthedashedlineshowsthenumberofprotonson TDC hit simulation includes light propagation delays targetpassingalldataqualitycuts. Thebottompanelshows intheWLSfibers. AlogicalORof32MA-PMTchannels the number of charged current candidate events in SciBar is made for each TDC channel, and the time of each hit normalized to the number of protons on target. The event is converted to TDC counts. Multiple TDC hits in each ratedifference between neutrinoand antineutrinomodes can channel are simulated. beseen clearly. In the EC detector simulation, true energy deposition in scintillating fibers in the detector is converted to the V. DATA SUMMARY numberofphotoelectronsusingaconversionfactorwhich ismeasuredforeachchannelwithcosmic-raymuons. The attenuation of light in the fiber is simulated using the The SciBooNE experiment took data from June 2007 measured attenuation length value. The number of pho- until August 2008. The data-takingis divided into three toelectrons is smeared by Poisson statistics and by the periods depending on the polarity of the horn, as sum- PMTresolution,andthenconvertedtoADCcounts. The marized in Table I. Fig. 4(top) shows a history of the time-dependent ADC gain due to the overshoot of the accumulatednumberofprotonsontarget;thetwocurves PMT signal is simulated based on a measurement with show the total protons on target for all events and the cosmic muons. Electronics noise is also simulated. protonsontargetforeventspassingalldataqualitycuts, described below. For the detector simulation of the MRD, true energy Intotal,2.64×1020protonsontargetweredeliveredto depositionineachscintillatorisconvertedtoADCcounts theberylliumtargetduringtheSciBooNEdatarun. The usingtheconversionfactormeasuredwithcosmicmuons. beam datastream (measuring, for example, magnet cur- Theattenuationoflightinthescintillatoraswellaselec- rent settings, measured beam intensity, measured peak tronics noise are simulated. Gaps between scintillator horncurrent)issynchronizedandmergedwiththecorre- counters in each plane, which cause inefficiency, are in- sponding SciBooNE detector datastream, provided that cluded in the simulation. The time of energy deposition the spill time as measured by the beam instrumentation is digitized and converted into TDC counts. and by the detector match within 10 ms of each other. 8 layer penetration, and therefore the minimum length of TABLE I: Summary of SciBooNE data-taking. The table a reconstructedtrack is 8 cm in the beam direction. Ac- showsthenumberofprotonsontarget (POT)collected after cording to the MC simulation, 96% of charged current application of data quality cuts, as described in thetext. interactions in SciBar are reconstructed to have at least one track. Run Period POT To identify charged current events, we look for events Run1 (Antineutrino) Jun. 2007 - Aug. 2007 0.52×1020 in which at least one reconstructed track in SciBar is Run2 (Neutrino) Oct. 2007 - Apr. 2008 0.99×1020 Run3 (Antineutrino) Apr. 2008 - Aug. 2008 1.01×1020 matched with a track or hits in the MRD. Such a track is defined as a SciBar-MRD matched track. The most energeticSciBar-MRDmatchedtrackinanyeventiscon- sideredasamuoncandidate. FormatchingaMRDtrack to a SciBar track, the upstream edge of the MRD track Only spills that satisfy certain beam quality cuts are is required to be on either one of the first two layers used for analysis. We require that the beam intensity is at least 0.1×1012 protons per spill, that the agree- of the MRD. The transverse distance between the two tracks at the first layer of the MRD must be less than ment between the two toroid readouts along the beam- 30 cm. The requirement on the difference between track line is within 10%, that the absolute peak horn current angles with respect to the beam direction is given by is greater than 170 kA, and that the targeting efficiency |θ − θ | < θ , where θ is a function of is greater than 95%. Overall, beam quality cuts reject MRD SB max max thelengthoftheMRDtrack,varyingbetween0.4radian less than 1% of the total number of protons on target and 1.1 radians. For track reconstruction in the MRD, accumulated during the run. A somewhat larger frac- at least two hit layers in each view are needed, and thus tion of protons on target is rejected because of detector this matching method is used for tracks which penetrate dead time, yielding about a 95% efficiency to satisfy all at least three steel plates. If no MRD track is found, we (beam plus detector) data quality cuts. After all beam and detector quality cuts, 2.52×1020 protons on target extrapolate the SciBar track to the MRD and search for nearby contiguous hits in the MRD identifying a short are usable for physics analyses. muon track. For matching MRD hits to a SciBar track, In this analysis, the full neutrino data sample is used, correspondingto 0.99×1020 protonsontargetsatisfying theMRDhitisrequiredtobewithinaconewithanaper- ture of ±0.5 radian and a transverse offset within 10 cm alldataqualitycuts,collectedbetweenOctober2007and of the extrapolated SciBar track at the upstream edge April 2008. During that time, all detector channels were of the MRD. The timing difference between the SciBar operationalonbeamtriggers. Theexperimentalstability track and the track or hits in the MRD is required to be isdemonstratedinFig.4(bottom),whichshowsthenum- within 100 nsec. The matching criteria impose a muon ber of charged current event candidates per protons on momentum threshold of 350 MeV/c. target. Inthisfigure,theeventreconstructionisasimple χ2 track finder which is used only for operations related studies, and not for the analysis described in this paper. B. Particle Identification The figure illustrates the event rate difference between neutrino mode and antineutrino mode running. The antineutrino data sample collected before and af- The SciBar detector has the capability to distinguish ter the neutrino data-taking period is not considered in protonsfrommuonsandpionsusingdE/dx. Theparticle this analysis. identification variable, Muon Confidence Level (MuCL) is calculated as follows. The confidence level at each plane is first defined as the fraction of events in the ex- VI. EVENT RECONSTRUCTION AND pected dE/dx distribution of muons above the observed ANALYSIS value, (dE/dx) . The expected dE/dx distribution of obs muons is obtained by using cosmic-ray muons. Each A. Track Reconstruction plane’s confidence level is combined to form a total con- fidence level, assuming the confidence level at each layer is independent. The MuCL is calculated as The first step of the event reconstruction is to search for two-dimensional tracks in each view of SciBar using n−1(−lnP)i a cellular automaton algorithm [52]. For tracking, the MuCL=P × (1) hit threshold is set to two photoelectrons, correspond- i! i=0 ing to approximately0.2MeV.Threedimensionaltracks X are reconstructed by matching the timing and z-edges wherenisthenumberofplanespenetratedbythetrack, of the two dimensional tracks. The timing difference P = n CL , CL is the confidence level at the i-th i=1 i i between two two dimensional tracks is required to be plane. Q less than 50 nsec, and the z-edge difference must be less Fig. 5 shows the dE/dx distributions of muon and than 6.6 cm for upstream and downstream edges. Re- proton enriched samples. The predicted distributions constructed tracks are required to have at least three- of true muon and proton tracks are shown as hatched 9 es1500 es80 date on the most upstream layer of SciBar to eliminate Entri mDATA Entri DpATA incomingparticlesduetoneutrinointeractionsintheup- 60 streamwallor soil. The hit thresholdfor this veto cut is other other 1000 set to two photoelectrons. The neutrino interaction ver- 40 tex is reconstructed as the upstream edge of the muon track. The vertex resolution is approximately 0.5 cm 500 20 in each dimension, estimated with the MC simulation. We select events whose vertices are in the SciBar fidu- 0 0 cial volume, defined to be ±130 cm in both the x and 0 2 4 6 8 10 0 2 4 6 8 10 dE/dx (MeV/cm) dE/dx (MeV/cm) y dimensions, and 2.62 cm< z <157.2 cm, a total mass of10.6tons. The backgroundcontaminationdue toneu- trinoeventswhichoccurintheECandMRDis2.0%and FIG.5: (Color online) dE/dxof muonenrichedsample (left) 0.5%, respectively. Finally, the time of the muon candi- and proton enriched sample (right). date is required to be within a 2 µsec window around the beam pulse. The cosmic-raybackground contamina- s104 s103 tion in the beam timing window is only 0.5%, estimated e e ntri mDATA ntri DpATA using a beam-off timing window. According to the MC E E simulation, the selection efficiency and purity of true ν other 102 other µ 103 chargedcurrenteventsare27.9%and92.8%,respectively. Impurity comes from ν neutral current events (3.0%), µ 10 ν charged current events (1.6%), and neutrino events 102 µ which occur in the EC/MRD (2.5%). The average neu- 1 trino beam energy for true chargedcurrent events in the 10 sample is 1.2 GeV. This SciBar-MRD matched sample 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 MuCL MuCL is our standard charged current data set and defines the MCnormalization,i.e. theMCdistributionsarenormal- ized to the number of SciBar-MRD matched events in FIG. 6: (Color online) MuCL of muon enriched (left) and data. proton enriched (right) samples. Two sub-samples of the SciBar-MRD matched sample are further defined; the MRD stopped sample and the MRD penetrated sample. Events with the muon stop- histograms. To select muon candidates for this study, pingintheMRDareclassifiedasMRDstoppedevents,in we first select SciBar-MRD matched tracks. According whichwecanmeasurethemuonmomentum. Eventswith to the MC simulation, the sample is 94.7% pure muons the muon exiting from the downstream end of the MRD with a small contamination of protons and charged pi- are defined as the MRD penetrated sample, in which we ons. Forprotoncandidates,weselectthesecondtrackin can measure only a part of the muon momentum. The a charged current quasi-elastic (CC-QE) scattering en- average neutrino beam energy for true charged current riched sample made by cutting on a kinematic variable events in the MRD stopped and MRD penetrated sam- described later. The fraction of protons in the sample ples are 1.0 GeV and 2.0 GeV, respectively, enabling a is 92.1%, estimated with the MC simulation. The con- measurement of charged current coherent pion produc- tamination ofchargedpions and muons are estimated to tion at two different mean neutrino energies. be 5.5% and 1.6%, respectively. Proton candidates are Theslopesofthemuonangleswithrespecttothebeam clearly separated from muon candidates. in the two SciBar views are used to calculate the three The MuCL distributions for the muon enriched sam- dimensional muon angle with respect to the beam (θ ). µ ple and the proton enriched sample are shown in Fig. 6. Thekineticenergyofthemuoniscalculatedbytherange Tracks with MuCL greater than 0.05 are considered and expected energy deposition per unit length (dE/dx) muon-like (or pion-like) and the others are classified as in SciBar, the EC and the MRD, proton-like. The probability of misidentification is esti- mated to be 1.1% for muons and 12% for protons, aver- E = ESB+EEC+EMRD kin aged over track length in the muon and proton enriched dE ∆EEC samples. = L + 0 +EMRD(L ) (2) SB MRD dx cosθ (cid:12)SB µ (cid:12) (cid:12) where ESB, EEC,(cid:12)and EMRD are the energy deposition C. Charged Current Event Selection in each detector. L and L are the track length of SB MRD the muon in SciBar and the range in the MRD, respec- Events with at least one SciBar-MRD matched track tively. We set dE/dx| to 2.04 MeV/cm, and ∆EEC, SB 0 are selected as charged current event candidates. We whichistheenergydepositedintheECbyahorizontally reject events with hits associated with the muon candi- traversing minimum ionizing particle, is set to 91 MeV, 10 estimated with the GEANT4 simulation. EMRD is cal- culated from a range to energy lookup table based on s e the MC simulation. For muons stopping in the MRD, ntri DATA the average muon momentum and muon angular reso- E 1155000000 CC coherent p lutions are 50 MeV/c and 0.9 degree, respectively. For muonsexitingthe MRD,onlyalowerlimitonmuonmo- CC resonant p mentumisobtained,while themuonangleisdetermined Other withthe sameresolutionasthatofstoppingmuons. The 1100000000 CC QE systematic uncertainty in the muon momentum scale is estimatedto be 2%whichis dominatedby the difference amongvariouscalculationsoftherangetoenergylookup 55000000 table. 00 00 11 22 33 44 55 D. Event Classification Number of tracks The MRD stopped and MRD penetrated samples are FIG. 7: (Color online) Number of vertex-matched tracks for further divided into sub-samples with the same selection the MRD stopped sample. The MC distribution shown here criteria. Once the muon candidate and the neutrino in- is before tuning. teraction vertex are reconstructed, we search for other tracks originating from the vertex. For this purpose, the track edge distance is defined as the 3D distance between the vertex and the closer edge of another re- constructed track. Tracks whose edge distance is within p-like p -like 10cmarecalledvertex-matchedtracks. Fig.7showsthe es distributionofthe numberoftracksatthe vertexforthe ntri DATA E p MRD stopped sample. For the MC simulation, the con- 110033 tributions from charged current coherent pion, charged p m currentresonantpion, chargedcurrentquasi-elastic,and other interactions are shownseparately. Most events are e reconstructed as either one track or two track events. 110022 The two track sample is further divided based on parti- cle identification. We first require that the MuCL of the SciBar-MRDmatchedtrackisgreaterthan0.05toreject events with a proton penetrating into the MRD. Then the second track in the event is classified as a pion-like 1100 or a proton-like track with the same MuCL threshold. 00 00..22 00..44 00..66 00..88 11 MuCL Fig. 8 shows the contributions to the second track from trueproton,pion,muon,andelectrontracksaspredicted by the MC simulation. FIG. 8: (Color online) MuCL for the second track in the In a charged current resonant pion event, νp → two-track sample. The MC distribution shown here is before µ−pπ+, the proton is often not reconstructed due to tuning. its low energy, and such an event is therefore identified as a two track µ+ π event. To separate charged cur- rent coherent pion events from chargedcurrent resonant pion events, additional protons with momentum below thetrackingthresholdareinsteaddetectedbytheirlarge scribed in Sec. IVB). De-excitation gamma-rays from energydepositionaroundthe vertex,so-calledvertexac- the carbon nucleus do not affect the distribution since tivity. We search for the maximum deposited energy in most of the gamma-rays first interact outside the ver- a strip around the vertex, an area of 12.5 cm×12.5 cm tex region. Events with energy deposition greater than in both views. Fig. 9 shows the maximum energy for 10 MeV are considered to have activity at the vertex. µ+πeventsintheMRDstoppedsample. Apeakaround Charged current coherent pion candidates are extracted 6 MeV corresponds to the energy deposited in the strip from the sample of µ+π events without vertex activ- containing the vertex by two minimum ionizing parti- ity. Foursub-samples,theonetrackevents,µ+pevents, cles, and a high energy tail is mainly due to the low µ+πeventswithvertexactivity,andµ+πeventswithout energy proton. To simulate such protons, we consider vertex activity in the MRD stopped sample are used for re-interactions of nucleons in the nucleus as described constraining systematic uncertainties in the simulation, in Sec. IIIE as well as ones outside the nucleus (de- described next.