Deeply Virtual Compton Scattering on the Neutron with CLAS12 at 11 GeV

Transcription

1 New Research Prpsal t Jeffersn Lab PAC 7 Deeply Virtual Cmptn Scattering n the Neutrn with CLAS1 at 11 GeV A. Fradi, B. Guegan, M. Guidal, S. Nicclai 1,, S. Pisan, D. Skhan Institut de Physique Nucléaire d Orsay, 9146 Orsay, France A. Avakian, V. Baturin, V.D. Burkert, L. Eluadrhiri, T. Kageya, V. Kubarvsky 1 Jeffersn Lab, Newprt News, VA 66, USA A. El Alaui 1 Argnne Natinal Labratry, Argnne, IL 649, USA M. Aghasyan, S. Anefals Pereira, E. De Sanctis, D. Hasch, V. Lucherini, M. Mirazita 1, P. Rssi INFN, Labratri Nazinali di Frascati, 44 Frascati, Italy M. Battaglieri, R. De Vita, M. Osipenk, G. Ricc, M. Ripani, M. Taiuti INFN, Sezine di Genva, Genva, Italy C. Maiern, Y. Perrin, E. Vutier LPSC Grenble, 8 Grenble, France J. Ball, M. Garçn, P. Knczykwski, B. Mren, H. Mutarde, S. Prcureur, F. Sabatié SPhN-CEA Saclay, Gif-sur-Yvette, France A. D Angel, C. Schaerf, V. Vegna Università di Rma - Tr Vergata, 17 Rma, Italy J. Annand, M. Hek, D. Ireland, R. Kaiser, K. Livingstn, D. MacGregr, G. Rsner, B. Seitz, G. Smith University f Glasgw, Glasgw G1 8QQ, United Kingdm A. Kubarvsky, N. Saylr, P. Stler Renesslaer Plytechnic Institute, Try, NY 118-9, USA A. Biselli Fairfield University, Fairfield Cnnecticut 684 M. Ungar University f Cnnecticut, Strrs, Cnnecticut c-spkespersn cntact persn, silvia@jlab.rg 1

2 L. Barin, M. Cntalbrig, G. Ciull, P. Lenisa, L. Pappalard INFN Sezine di Ferrara, 44 Ferrara, Italy F. Meddi, G.M. Urcili INFN Sezine di Rma, 18 Rma, Italy D.M. Castellucci, E. Cisbani, F. Garibaldi, S. Frullani INFN Rma - Sanità, 161 Rma, Italy M. Capgni INFN Rma and ENEA Casaccia, 1 S. Maria di Galeria - Rma, Italy V. Bellini, A. Giusa, F. Mammliti, G. Russ, L. Sperdut, C. Sutera INFN Sezine di Catania, 91 Catania, Italy R. De Le Università di Bari, 711 Bari, Italy R. Perrin INFN Sezine di Lecce, 7 Lecce, Italy A CLAS Cllabratin prpsal

3 Abstract Measuring Deeply Virtual Cmptn Scattering n a neutrn target is ne f the necessary steps t cmplete ur understanding f the structure f the nuclen in terms f Generalized Partn Distributins (GPDs). DVCS n a neutrn target allws t perate a flavr decmpsitin f the GPDs and plays a cmplementary rle t DVCS n a transversely plarized prtn target in the determinatin f the GPD E, the least knwn and least cnstrained GPD that enters Ji s angular mmentum sum rule. T start the experimental prgram f DVCS n the neutrn, we prpse t measure beamspin asymmetries fr n-dvcs (ed e nγ(p)) with the upgraded 11-GeV CEBAF plarized-electrn beam and the CLAS1 detectr. Fr the detectin f the recil neutrn, necessary t ensure the exclusivity f the reactin after having detected the scattered electrn and the DVCS phtn, we will cnstruct a scintillatr-barrel detectr t be placed in the Central Detectr, between the CTOF and the slenid magnet. This Central Neutrn Detectr (CND) will be made f three layers f scintillatr paddles (48 paddles per layer), cupled tw-by-tw at the frnt with semi-circular light guides and read at the back by phtmultipliers placed utside f the high magneticfield regin and cnnected t the bars via 1-meter-lng bent light guides. Simulatins and measurements n a prttype, cvering ne radial layer and tw azimuthal bins, have prven the feasibility f this prject. In rder t prvide an accurate mapping f the n-dvcs beam-spin asymmetry ver the available 4-dimensinal (Q, x B, t, ) phase space, we request 9 days f running n a deuterium target with the maximum available beam energy, 11 GeV. 1

4 Cntents 1 Intrductin 8 Physics mtivatin: neutrn GPDs 9 First n-dvcs experiment: JLab Hall A 11 4 Central Neutrn Detectr: mtivatin and requirements 1 CND: R & D studies and final detectr design.1 Summary f early R&D studies Final design and perfrmances Csts and financial supprt fr the CND Simulatin and recnstructin Efficiency Angular and mmentum reslutins Particle Identificatin Backgrunds n the CND Physics backgrund Electrmagnetic backgrund enπ (p) backgrund 4 9 Cunt-rate estimate 47 Systematic uncertainties Summary f experimental setup and trigger cnfiguratin 1 1 Beam-time request 1 Cnclusins A Details n simulatin and recnstructin 4 A.1 Digitisatin f signals frm CND paddles in GEMC A. Hit recnstructin B Prjected results 8

5 List f Tables 1 Expected 4-fld differential crss sectins, 4-dimensinal acceptance times neutrn detectin efficiency and number f events fr n-dvcs/bh with CLAS1 and the CND, as a functin f. < t >=. GeV, < Q >=.7 GeV, < x B >=.7, =, Q = 1. GeV, x B =.1, t =. GeV. The calculatin was dne fr a luminsity L = cm s 1 per nuclen and fr 8 days f running time Expected systematic uncertainties n the prpsed measurement Beam-time request List f Figures 1 The handbag diagram fr the DVCS prcess n a nuclen en e N γ. Here x + ξ and x ξ are the lngitudinal mmentum fractins f the initial and final quark, respectively, and t = (p p ) is the squared mmentum transfer between the initial and final prtns (r equivalently between the tw phtns). There is als a crssed diagram which is nt shwn here Beam-spin asymmetry fr DVCS n a neutrn target, pltted as a functin f (frm left t right), t, x B, and Q, as predicted by the VGG mdel. The kinematics are: E e =11 GeV, x B =.17, Q = GeV, t=.4 GeV and =6. All distributins have been calculated at these kinematics, except fr the variable against which each distributin is pltted. The curves are btained fr different cmbinatins f values f J u and J d : (J u =.,J d =.1) - slid curve, (J u =.8,J d =.1) - thin dashed curve, (J u =.,J d =.1) - thin dash-dtted curve, (J u =.,J d =.8) - thick dashed curve, (J u =.,J d =.) - thick dash-dtted curve Beam-spin asymmetry fr DVCS n a prtn target, as predicted by the VGG mdel, pltted as a functin f (frm left t right), t, x B and Q. E e =11 GeV, x B =., Q = GeV, t=. GeV and =6. Otherwise, same cnditins and cnventins as fr Fig VGG-mdel predictin fr the unplarized crss sectin fr DVCS n a prtn target (slid curves) and n a neutrn target (dashed curve), pltted as a functin f (frm left t right), t and x B. Same kinematics as fr Fig. (fr the prtn) and Fig. (fr the neutrn)

6 n-dvcs analysis results frm the Hall A experiment []. Tp: helicity signal (defined as S h = π (N+ N )d Φ π π (N+ N )d Φ), fr D(e,e,γ) and H(e,e,γ) events; H data are flded with a mmentum distributin f the prtn in deuterium and scaled t the D data luminsity; the simulatin curve is fr the Fermi-bradened H(e,e,γ)p reactin. Bttm: residual helicity signal after H subtractin; the arrws indicate the MX average psitin f n-dvcs and d-dvcs events fr < t >=. GeV ; the simulatin curves are integrated ver the acceptance and btained fr the arbitrary values Im[Cn] I exp = Im[Cd I]exp = 1, where Cn I and Cd I depend n the interference f the BH amplitude with the twist- Cmptn frm factrs n-dvcs results frm the Hall A experiment []: t dependence f the extracted sin( γγ ) mments fr cherent d-dvcs (tp) and incherent n-dvcs (bttm). Errr bars shw statistical uncertainties; systematical uncertainties are indicated by the shaded bands Distributins f kinematic variables fr n-dvcs events. Frward- CLAS1 acceptance cuts and physics cuts are included. Tp: Q as a functin f x B. Middle: t as a functin f x B. Bttm: t as a functin f Q Electrn energy as a functin f electrn plar angle, fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included Phtn energy as a functin f phtn plar angle, fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. The tw znes crrespnd t the IC/Frward tagger (frm. t 4. ) and t the FEC (frm nwards) Neutrn mmentum as a functin f neutrn plar angle, fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included Neutrn mmentum (tp) and plar angle (bttm), fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. The drp at p n 1. is due t the t < 1. GeV/c cut, applied t ensure the applicability f the GPD frmalism Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr, assuming abslute precisin n the phtn and electrn kinematic variables. Tp: fixed neutrn θ and reslutins, mmentum reslutin varying between.1% and %. Middle: fixed neutrn mmentum and reslutins, θ reslutin varying between.1 and.bttm: fixed neutrn mmentum and θ reslutins, reslutin varying between.1 and Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr, assuming the nminal CLAS1 reslutins n the phtn and electrn kinematic variables, fixing the neutrn θ and reslutins and varying the mmentum reslutin, between.1% and %. Tp plt: phtn reslutins fr the Frward- Tagger ptin. Bttm plt: phtn reslutins fr the IC ptin... 4

7 14 Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr. The different clrs crrespnd t different cmbinatins f chices f particles being detected with abslute precisin r with realistic reslutins. Tp plt: phtn reslutins fr the Frward-Tagger ptin. Bttm plt: phtn reslutins fr the IC ptin Drawing f the Central Detectr: the red area represents the free space between the magnet (shaded area) and the CTOF (represented by the bar and its bent light guides) Gemetry f the scintillatr barrel fr the Central Neutrn Detectr. The current design cnsists f radial layers each made f 48 trapezidal scintillatr paddles Magnetic field map fr the Central Detectr (radial cmpnent n the tp, axial cmpnent n the bttm). The black semi-circle in the tp plt shws the psitin f the PMTs f the Central Neutrn Detectr Design f the Central Neutrn Detectr Drawing (side cut view) f the CND, placed int the slenid magnet. 9 The ne-layer prttype f the CND during the timing reslutin measurements with csmic rays carried ut at Orsay Drawing f the three-layer prttype f the CND, under cnstructin. Results f the csmic rays measurements n the ne-layer prttype. Tp: charge cllected by the tw PMTs as a functin f the hit psitin. Bttm: time reslutin f each PMT as a functin f the hit psitin Efficiency fr the detectin f neutrns having.4 GeV/c f mmentum, as a functin f the threshld n the depsited energy. The efficiency is shwn fr different values f θ n, between and Efficiency fr the detectin f neutrns emitted at 6, as a functin f mmentum, fr 7 different values f the threshld n the depsited energy, frm 1 t MeV Efficiency fr the detectin f neutrns, as a functin f neutrn mmentum, fr a -MeV threshld n the depsited energy. The efficiency is shwn fr three different values f θ n, between and 7. 6 Angular reslutin σ θ as a functin f θ fr neutrns f mmentum.4 GeV/c, fr a -MeV threshld n the depsited energy. The three clrs f the pints crrespnd t the three radial layers f the CND Mmentum reslutin σ p /p as a functin f p fr neutrns having θ = 6, fr a -MeV threshld n the depsited energy. The three clrs f the pints crrespnd t the three radial layers f the CND Efficiency fr the detectin f phtns, as a functin f phtn mmentum, fr a -MeV threshld n the depsited energy. The efficiency is shwn fr θ γ = 6. Belw E γ =.1 GeV, the phtn efficiency drps t zer

8 9 β distributins fr neutrns with p n =. GeV/c (green), p n =.4 GeV/c (purple), p n =.7 GeV/c (blue), p n = 1 GeV/c (red), and phtns with E = 1 GeV. Each bx shws the results fr ne f the three radial layers that cmpse the CND (the innermst is n the tp, the utermst is n the bttm). The threshld n the depsited energy is MeV. The plts shw all hits, integrated ver. Equal neutrn and phtn yields have been assumed here β versus mmentum fr neutrns (red) and phtns (blue) with mmenta between. and 1 GeV. The errr bars are defined as σ, where σ is the fitted width f each β peak. The threshld n the depsited energy is MeV θ versus energy fr the phtns emitted at backward angles assciated t an electrn e and an energetic phtn γ detected in the frward part f CLAS1. The cut MM(eγ) < 1 GeV/c is applied View (frm the beam s perspective) f the Central Detectr, fr ne simulated backgrund event. Red tracks crrespnd t negatively charged particles, green tracks crrespnd t neutrals. This picture has been btained fr a luminsity L = cm s 1, crrespnding t 1/ f the nminal luminsity, fr practical reasns related t the graphical interface Distributin f the energy depsit in the CND, integrated ver all the azimuthal and radial bins, f the hits cming frm the generated electrmagnetic backgrund. The majrity f the events crrespnds t an energy depsitin belw 1 MeV Event distributin fr the electrmagnetic backgrund in the CND, as a functin f the x and y crdinates in the lab frame (z being the beam directin), withut any cut n the depsited energy. It can be seen that the majrity f the events are cncentrated in the innermst layer f the CND Missing mass f the enγ system, fr the n-dvcs/bh channel (in red), and the ed e nπ (p) channel (in black), bth simulated with ur event generatr. CLAS1 and CND reslutins are applied. Tp plt: phtn reslutins fr the Frward-Tagger ptin. Bttm: phtn reslutins fr the IC ptin Tp: crss sectin fr n-dvcs/bh. Middle: acceptance fr the eγ(p) final state, including nly the frward part f CLAS1, cmputed with ur event generatr and FASTMC. Bttm: expected cunt rate fr 8 days f beam time. All three plts are prduced fr the kinematic bin f Table Beam-spin asymmetry fr n-dvcs/bh as predicted by the VGG mdel (fr J u =. and J d =.1), pltted as a functin f fr the kinematic bin t =. GeV, Q =.7 GeV, x B =.7. The errr bars reflect the expected uncertainties fr ur experiment, crrespnding t 8 hurs f beam time at a luminsity f cm s 1 per nuclen.. 6

9 8 Schematic drawings f the cases the recnstructin deals with when signals are cllected at bth D and N PMTs Prjected cunt rates fr n-dvcs/bh, as a functin f, fr each Q, x B bin and fr. < t <. GeV (tp) and. < t <. GeV (bttm) Prjected cunt rates fr n-dvcs/bh, as a functin f, fr each Q, x B bin and fr. < t <.8 GeV (tp) and.8 < t < 1. GeV (bttm) Prjected BSAs fr n-dvcs/bh, as a functin f, fr each Q, x B bin and fr. < t <. GeV (tp) and. < t <. GeV (bttm) Prjected BSAs fr n-dvcs/bh, as a functin f, fr each Q, x B bin and fr. < t <.8 GeV (tp) and.8 < t < 1. GeV (bttm)

10 1 Intrductin Generalized Partn Distributins are nwadays the bject f an intense effrt f research, in the perspective f understanding nuclen structure. They describe the crrelatins between the lngitudinal mmentum and transverse spatial psitin f the partns inside the nuclen, they give access t the cntributin f the rbital mmentum f the quarks t the nuclen spin, they are sensitive t the crrelated q q cmpnents, etc. The riginal articles and general reviews n GPDs and details n the frmalism can be fund in Refs. [1,,, 4,, 6, 7]. The nuclen GPDs are the structure functins which are accessed in the measurement f the exclusive leptprductin f a phtn (DVCS, which stands fr Deeply Virtual Cmptn Scattering) r f a mesn n the nuclen, at sufficiently large Q, where Q is the virtuality f the phtn emitted by the initial leptn. Figure 1 illustrates the leading prcess fr DVCS. Cnsidering nly helicity-cnserving quantities and the quark sectr, there are fur GPDs, H, H,E,Ẽ, which depend, in leading-rder and leading-twist QCD, upn three variables: x, ξ and t. x ξ and x + ξ are the lngitudinal mmentum fractins f the quarks, respectively, cming ut and ging back int the nuclen and t is the squared fur-mmentum transfer between the final and initial nuclen. Figure 1: The handbag diagram fr the DVCS prcess n a nuclen en e N γ. Here x + ξ and x ξ are the lngitudinal mmentum fractins f the initial and final quark, respectively, and t = (p p ) is the squared mmentum transfer between the initial and final prtns (r equivalently between the tw phtns). There is als a crssed diagram which is nt shwn here. Amng the three variables, x, ξ and t, nly tw, ξ and t, are accessible experimen- 8

11 tally. In the Bjrken limit, ξ = x B/ 1 x B /, where x B is the standard Bjrken variable. Frmally, the DVCS amplitude is prprtinal t: +1 1 dx H( x,ξ,t) x ± ξ iǫ +... (1) where the ellipsis stands fr similar terms fr E, H and Ẽ. Decmpsing this expressin int its real and imaginary parts, it is fund that the maximum infrmatin that can be extracted frm the experimental data at a given (ξ, t) pint is H(±ξ, ξ, t), when measuring an bservable sensitive t the imaginary part f the DVCS amplitude, and +1 1 dxh( x,ξ,t) x±ξ, when measuring an bservable sensitive t the real part f the DVCS amplitude. Knwing the GPDs at sme particular pint (±ξ, ξ, t) and their weighted integral ver x des nt, f curse, uniquely define them. A mdel input will be required, t make the interplatin ver the variable x. The DVCS prcess is accmpanied by the Bethe-Heitler (BH) prcess, in which the final-state phtn is radiated by the incming r scattered electrn and nt by the nuclen itself. The BH prcess, which is nt sensitive t GPDs, is indistinguishable frm the DVCS and interferes with it, cmplicating the matter. Hwever, cnsidering that the nuclen frm factrs are well knwn at small t, the BH prcess is precisely calculable theretically. It is clearly a highly nn-trivial task t actually measure the GPDs. It calls fr a lng-term experimental prgram cmprising the measurement f different bservables: crss sectins, beam-, lngitudinal and transverse target- single plarizatin bservables, duble plarizatin bservables and als pssibly beam-charge asymmetries, timelike Cmptn scattering, etc. Refs. [8, 9] shw the infrmatin brught by the varius bservables. Such dedicated experimental prgram, cncentrating n a prtn target, has started wrldwide in these past few years. JLab has prvided the first measurement, in the valence regin, f beam-plarized and unplarized DVCS crss sectins, in a limited phase-space dmain, with the Hall A [], and several beam-spin and target-spin asymmetries (BSA, TSA), ver a large kinematic range, btained with the CLAS detectr [11], [1]. Beam-charge asymmetries, BSAs, lngitudinally and transverselyplarized target-spin asymmetries, as well as duble-spin asymmetries, have als been measured by the HERMES cllabratin [1]. These first data will sn be cmpleted with a series f new experiments nging and planned at JLab and aimed t measure accurately lngitudinally [14] and transversely [1] plarized target-spin asymmetries and crss sectins (alng with duble-plarizatin bservables) and new precise unplarized and beam-plarized crss sectins at new kinematics [16, 17]. Measurements f DVCS crss sectins, BSA and lngitudinal TSA with JLab at 1 GeV have als been apprved [18, 19]. Physics mtivatin: neutrn GPDs The aim f this prpsal is t start a similar experimental prgram with a neutrn target. The imprtance f neutrn targets in the DVCS phenmenlgy was clearly es- 9

12 tablished in the pineering Hall A experiment, where the plarized-beam crss sectin difference ff a neutrn, frm a deuterium target, was measured [] (see Sectin ). Measuring neutrn GPDs is highly cmplementary t measuring prtn GPDs. Neutrn and prtn GPDs are independent quantities, like neutrn and prtn frm factrs. Measuring bth GPDs allws t carry ut a flavr separatin. Fr instance, H p (ξ,ξ,t) = 4 9 Hu (ξ,ξ,t) Hd (ξ,ξ,t) () and H n (ξ,ξ,t) = 1 9 Hu (ξ,ξ,t) Hd (ξ,ξ,t) () (and similarly fr E, H and Ẽ), frm which ne can btain and H u (ξ,ξ,t) = 9 1 (4Hp (ξ,ξ,t) H n (ξ,ξ,t)) (4) H d (ξ,ξ,t) = 9 1 (4Hn (ξ,ξ,t) H p (ξ,ξ,t)). () Cncerning the BSA, which is the main gal f this prpsal, it can be shwn that, in the case f DVCS n the neutrn, its amplitude is mainly gverned by the GPD E, the least knwn f the GPDs. In particular, E is ne f the tw GPDs entering Ji s sum rule: J q = dxx [H q (x,ξ,t = ) + E q (x,ξ,t = )], (6) which links the ttal angular mmentum (J q ) carried by each quark q t the sum f the secnd mments ver x f the GPDs H and E. It is therefre crucial t btain experimental cnstraints n E in rder t make sme first steps twards the estimatin f the cntributin f the rbital mmentum f the quarks t the nuclen spin. In rder t make a quark-flavr separatin, bth E n and E p are needed: this prpsal mainly aims at determining E n. E p can be accessed thrugh transverse-target plarizatin r duble (beam-target) plarizatin bservables n the prtn [8], which are, as previusly mentined, the gals f experiments already planned at JLab. Hereafter, the VGG mdel [1, ], which parametrizes GPDs and calculates the assciated DVCS bservables, has been adpted, in rder t quantify (albeit in a mdel-dependent way) the sensitivity f the neutrn-dvcs BSA t the GPD E. An interesting feature f the VGG mdel is that the parametrizatin f the GPD E is dependent n the tw parameters J u and J d, i.e. the ttal spin (rbital mmentum+intrinsic spin) cntributins f the u and d quarks respectively. The idea is that a given shape in x fr the GPD E q is assumed, and then the verall nrmalizatin is prprtinal t J q (see ref. [] fr mre details). Figure shws the BSA fr n- DVCS as a functin f the fur independent variables describing the DVCS prcess,

13 , t, x B and Q, fr different values f J u and J d, as predicted by the VGG mdel. The kinematics fr Fig. are E e =11 GeV, x B =.17, Q = GeV, t=.4 GeV and =6. Althugh sme f the J u, J d values are unlikely (fr instance J d =.8), nevertheless this shws the strng sensitivity f this BSA t E and, in the framewrk f the VGG mdel, t J q. One sees that these BSAs can extend frm up t %, with spectacular changes f sign depending n the relative signs f J u and J d, and therefre they can be as large, in magnitude, as the prtn-dvcs beam-spin asymmetries that have been recently measured [11]. Hwever, it is imprtant t ntice that these large neutrn-dvcs asymmetries are btained nly in a specific phase space regin, i.e. nly arund x B =.1 r.1. T reach such lw values f x B, at sufficiently large Q, an 11-GeV electrn beam is needed. The current 6-GeV beam allws t explre mainly the x B. regin where the BSA far frm its maximum. This is cnfirmed by the explratry measurement f the JLab Hall A cllabratin [] where neutrn-dvcs BSAs essentially cnsistent with zer were btained and fr which the sensitivity t J q was therefre minimal. Figure shws the crrespnding BSAs, at apprximatively the same kinematics, fr the prtn case. It is clear that the sensitivity t E r, alternatively t J u and J d, is much less. This is mainly due t the fact that the prtn-dvcs BSA is mstly sensitive t the H GPD, the weight f the E GPD being suppressed by kinematical factrs. Finally, Fig. 4 shws the cmparisn f the unplarized crss sectins fr DVCS n the prtn and n the neutrn at apprximately the same kinematics, as a functin f, t and x B, accrding t the VGG mdel with J u =. and J d =.1. One sees that the neutrn-dvcs crss sectins are, depending n the kinematics, a factr t belw the prtn-dvcs crss sectins. First n-dvcs experiment: JLab Hall A The neutrn DVCS channel was explred fr the first time in the E-6 experiment [] perfrmed in the Hall A f Jeffersn Lab. The plarized-beam crss sectin difference was measured n deuterium and hydrgen targets, and the neutrn DVCS and deutern DVCS signals were extracted frm the cmparisn f experimental yields within the impulse apprximatin (Fig. ). On the ne hand, this pineering wrk did experimentally establish the imprtance f the measurement f the n-dvcs reactin fr the investigatin r quark angular mmentum []. On the ther hand, these data are limited t ne specific regin f the physics phase space and suffer frm significant statistic and systematic errrs riginating frm the crrelatin f the neutrn and deutern mments, the relative calibratin f the phtn calrimeter between hydrgen and deuterium targets, and the neutral pin cntaminatin. The experiment prpsed here aims at investigating the n-dvcs reactin in a wide phase space, prviding a systematic study f the beam-spin asymmetry. The detectin f the struck neutrn will insure the full exclusivity f the reactin and well-established techniques, cmmn t all DVCS measurements perfrmed with CLAS and prpsed fr CLAS1, will allw fr a precise subtractin f the neutral pin backgrund (see Sectin 8). 11

14 Φ (deg.). 1 -t (GeV ).. x B 4 Q (GeV ) Figure : Beam-spin asymmetry fr DVCS n a neutrn target, pltted as a functin f (frm left t right), t, x B, and Q, as predicted by the VGG mdel. The kinematics are: E e =11 GeV, x B =.17, Q = GeV, t=.4 GeV and =6. All distributins have been calculated at these kinematics, except fr the variable against which each distributin is pltted. The curves are btained fr different cmbinatins f values f J u and J d : (J u =.,J d =.1) - slid curve, (J u =.8,J d =.1) - thin dashed curve, (J u =.,J d =.1) - thin dash-dtted curve, (J u =.,J d =.8) - thick dashed curve, (J u =.,J d =.) - thick dash-dtted curve. 1

15 Φ (deg.). 1 -t (GeV )..4. x B Q (GeV ) Figure : Beam-spin asymmetry fr DVCS n a prtn target, as predicted by the VGG mdel, pltted as a functin f (frm left t right), t, x B and Q. E e =11 GeV, x B =., Q = GeV, t=. GeV and =6. Otherwise, same cnditins and cnventins as fr Fig.. 1

16 dσ/dq dx B dtdφ (nb/gev 4 ) Φ (deg.). 1 -t (GeV )..4 x B Figure 4: VGG-mdel predictin fr the unplarized crss sectin fr DVCS n a prtn target (slid curves) and n a neutrn target (dashed curve), pltted as a functin f (frm left t right), t and x B. Same kinematics as fr Fig. (fr the prtn) and Fig. (fr the neutrn). 14

17 S h data 1 M X cut D data H p-dvcs simulatin S h n-dvcs d-dvcs simulatin M X (GeV ) d-dvcs n-dvcs simulatin - - n-dvcs + d-dvcs M (GeV ) Figure : n-dvcs analysis results frm the Hall A experiment []. Tp: helicity signal (defined as S h = π (N+ N )d Φ π π (N+ N )d Φ), fr D(e, e, γ) and H(e, e, γ) events; H data are flded with a mmentum distributin f the prtn in deuterium and scaled t the D data luminsity; the simulatin curve is fr the Fermi-bradened H(e, e, γ)p reactin. Bttm: residual helicity signal after H subtractin; the arrws indicate the MX average psitin f n-dvcs and d-dvcs events fr < t >=. GeV ; the simulatin curves are integrated ver the acceptance and btained fr the arbitrary values Im[Cn I]exp = Im[Cd I]exp = 1, where Cn I and CI d depend n the interference f the BH amplitude with the twist- Cmptn frm factrs.. While the neutrn detectin was als implemented in the Hall A n-dvcs experiment, it was never successfully used in the data analysis, presumably because f the effects f a large neutral lw energy backgrund at frward angles [4]. Because f the actual lcatin f ur neutrn detectr at large angles (mtivated in Sectin 4) and the additinal bst frm the increased beam energy, these effects are expected t be highly suppressed in the prpsed experiment (see Sectin 7.). 4 Central Neutrn Detectr: mtivatin and requirements An event generatr fr DVCS/BH and exclusive π electrprductin n the neutrn inside a deuterium target has been develped []. The DVCS amplitude is calculated accrding t the BKM frmalism [9], while the GPDs have been taken frm the standard CLAS DVCS generatr [6]. The Fermi-mtin distributin is calculated with the Paris ptential [7]. The utput f the event generatr was fed thrugh CLAS1 FASTMC, t simu- 1

18 exp ) d Im(C I I 6 4 This experiment Can & Pire calculatin [4] - exp ) n Im(C This experiment t (GeV ) J u =-.4 1 J d =-.6 J u J =. d= AHLT calculatin [6] VGG calculatin [7] J u =.6 J d = t (GeV ) Figure 6: n-dvcs results frm the Hall A experiment []: t dependence f the extracted sin( γγ ) mments fr cherent d-dvcs (tp) and incherent n-dvcs (bttm). Errr bars shw statistical uncertainties; systematical uncertainties are indicated by the shaded bands. late the acceptance and reslutins f electrns and phtns in the Frward Detectr. Fr the detectin f phtns with plar angles between. and 4., tw ptins have been studied: the current Inner Calrimeter (IC), which shuld be used in the apprved CLAS1 experiment fr DVCS n the prtn [19], and its pssible upgrade, currently being prpsed, the Frward Tagger (FT) [8]. Kinematic cuts t ensure the applicability f the GPD frmalism (Q > 1 GeV /c, t > 1. GeV /c, W > GeV/c ) have been applied. Figure 7 shws the cverage in Q, x B and t that is btained frm the event generatr fr the n-dvcs/bh reactin, with an electrn-beam energy f 11 GeV. Figures 8, 9, and shw θ as a functin f mmentum in the lab frame fr, respectively, the electrn, the phtn and the neutrn. The tw panels f Fig. 11 are ne-dimensinal plts, shwing, respectively, the mmentum and the plar angle f the recil neutrn. As expected, the electrn and the phtn are mstly emitted at frward angles, while the recil neutrn is ging at backwards angles. In the hypthesis f absence f Final State Interactins (FSI), the minimal requirement t ensure the exclusivity f the n-dvcs reactin frm a deuterium target and t determine the final and initial state is t fully detect (PID, angles and mmentum) the scattered electrn, the phtn, and the neutrn. In fact, using fur-vectrs, the energy-mmentum cnservatin fr the n-dvcs reactin can be written as: p µ e + p µ n + p µ p = p µ e + p µ n + p µ p + p µ γ. (7) The absence f FSI implies that the kinematics f the initial and final spectatr 16

19 ) (GeV Q Xbj - ) t(gev. t Vs x Bj Xbj ) t(gev..4 t Vs Q Q (GeV ) Figure 7: Distributins f kinematic variables fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. Tp: Q as a functin f x B. Middle: t as a functin f x B. Bttm: t as a functin f Q. 17

20 P e (GeV) 9 8 P e Vs θ e θ e (deg) Figure 8: Electrn energy as a functin f electrn plar angle, fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. Figure 9: Phtn energy as a functin f phtn plar angle, fr n-dvcs events. Frward- CLAS1 acceptance cuts and physics cuts are included. The tw znes crrespnd t the IC/Frward tagger (frm. t 4. ) and t the FEC (frm nwards). 18

21 P N Vs θ N P N (GeV) θ N (deg) Figure : Neutrn mmentum as a functin f neutrn plar angle, fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. prtn are equal: Substituting Eq. 8 in Eq. 7 ne btains: p µ p = p µ p. (8) p µ e + pµ n = pµ e + p µ n + p µ γ. (9) Knwing the beam energy, if ne identifies the final electrn, phtn and neutrn and measures their angles and mmenta, fur unknwns (the cmpnents f the initialneutrn fur vectr) and fur equatins remain. The spectatr prtn kinematics can then be retrieved using the fact that, since the deutern target is at rest: p µ n + p µ p =,E n + E p = m d. () As shwn in the previus sectin, the electrn and the DVCS phtn will be emitted at small angles, and thus will be detected in the frward part f CLAS1 (with the phtn either in the EC r in the IC/FT), while the neutrn will be emitted predminantly (fr 8% f the events) at θ > 4 in the labratry frame, with average mmentum arund.4 GeV/c. This pints t the necessity t add a neutrn detectr (hereafter named Central Neutrn Detectr, r CND) t the Central Detectr f CLAS1, that in the present design has very limited detectin efficiency fr neutrns they can be detected in the CTOF, with abut -% f efficiency. 19

22 Neutrn mmentum P N (GeV) θ N θ N (deg) Figure 11: Neutrn mmentum (tp) and plar angle (bttm), fr n-dvcs events. Frward-CLAS1 acceptance cuts and physics cuts are included. The drp at p n 1. is due t the t < 1. GeV/c cut, applied t ensure the applicability f the GPD frmalism.

23 With the aid f the CLAS1 FASTMC tl, the requirements in terms f angular and mmentum reslutins n the detected neutrns were determined. The kinematical variables f the scattered electrn (e) and f the DVCS phtn (γ), cmputed by the n-dvcs generatr described in the previus sectin, were smeared using the values f reslutins prduced by FASTMC. As mentined ealier n, fr the phtn detectin at lw angles (. 4. ), tw ptins were studied: the standard Inner Calrimeter (IC) and the upgraded prpsed ne, r Frward Tagger (FT). The energy and angular reslutins were parametrized, respectively, as: fr the IC case: σ E /E = (.4/E) + (.1/ E) +.19 [9] and σ θ = 186 E. (taking the spatial angular reslutin f the current IC σ x =.cm E and assuming a target-ic distance f 186 cm) [19] fr the FT case: σ E /E = (./E) +.1 and the same angular reslutin as fr the IC case. The CND requirements were determined by lking at the missing mass f the enγ system, which is the nly quantity ne can cut n, in this detectin tplgy, t ensure exclusivity fr the n-dvcs channel by minimizing the enπ cntaminatin. First f all, withut applying any reslutins n the electrn and phtn kinematical variables, and varying instead the smearing n the neutrn kinematical variables, it was shwn that the reslutin n the neutrn mmentum plays the majr rle in determining the width f MM(enγ), while the effect f the angular reslutins is less imprtant. This can be seen cmparing the three panels f Fig. 1, where the missing mass is cmputed varying, respectively, the neutrn mmentum, the plar and the azimuthal angle, while keeping the ther tw variables cnstant. Varying either σ θ r σ by a factr (frm.1 t ) increases the width f MM(enγ) by nly 6-8 MeV in abslute (crrespnding t abut % mre), while the same increase by a factr f (frm.1% t %) n the neutrn mmentum reslutin σ P /P wrsens the reslutin f the missing mass by a factr f 4 (its width passes frm MeV t 86 MeV). Intrducing the realistic reslutins n the electrn and phtn calculated by FASTMC, it appears (Figs. 1 and 14) that if the neutrn mmentum reslutin is kept up t % its effect is negligeable with respect t the ther particles. In particular (tw panels f Fig. 14, green curve), the phtn reslutins is respnsible f 94% f the width f the missing mass fr the FT case (tp figure) and 97% fr the IC case (bttm figure). Therefre, cnsidering that the detectin capabilities f CLAS1 fr electrns and high-energy phtns are fixed, the requirements f the CND will be: gd neutrn identificatin capabilities fr the kinematic range f interest (. < p n < 1. GeV/c, 4 < θ n < 8 ) and neutrn mmentum reslutin σ P /P within %. Fr these figures, the fixed values f the reslutins (σ P /P = %, σ θ =., σ =.7 ) are an average f the realistic nes, btained frm the GEANT4 simulatin f the CND see Sectin 6. Hwever, as it will be explained in the fllwing, the cnclusins f the study d nt depend n these particular values. 1

24 θ N =., =.7 N N reslutin n e and γ Mm (ed >en γ[x]) Entries P P =% 1777 χ / ndf.18e+4 / 149 Cnstant 499 ±.1 Mean.8749 ±. Sigma.8641 ±.1 Entries P P =% 1777 χ / ndf 1.77e+4 / 141 Cnstant 19 ± 48.8 Mean.87 ±. Sigma.478 ±.4 Entries P P =% 1777 χ / ndf 1.6e+4 / 7 Cnstant 691 ± 74. Mean.874 ±.1 Sigma. ±. Entries P P =1% 1777 χ / ndf 148 / 16 Cnstant 1.6e+4 ± 117 Mean.8749 ±. Sigma.8 ±.8 Entries P P =.1% 1777 χ / ndf 6.61 / Cnstant.e+4 ± 141 Mean.87 ±. Sigma.897 ± GeV Mm (ed >en γ[x]) Mm (ed >en γ[x]) Entries = 1777 χ / ndf 4819 / 118 Cnstant 1717 ±. Mean.8641 ±.4 4 Sigma.787 ±.1 P = θ N Entries 1777 N =., =% P N χ / ndf Cnstant / ± 9.1 Entries 1777 N reslutin n e and γ Mean Sigma.8718 ±..1 ±.4 = χ / ndf 69 / 7 Cnstant 199 ± 4.8 Mean.8747 ±. Sigma.4 ±.4 Entries = χ / ndf 48 / 7 1 Cnstant 1997 ± 4.9 Mean.874 ±. Sigma. ±. =.1 Entries 1777 χ / ndf 194 / 7 Cnstant 97 ± 4. Mean.878 ±. Sigma.196 ± GeV Figure 1: Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr, assuming abslute precisin n the phtn and electrn kinematic variables. Tp: fixed neutrn θ and reslutins, mmentum reslutin varying between.1% and %. Middle: fixed neutrn mmentum and reslutins, θ reslutin varying between.1 and.bttm: fixed neutrn mmentum and θ reslutins, reslutin varying between.1 and.

25 θ N =., =.7 N Mm (ed->en γ[x]) Entries 1196 P χ / ndf / 197 Cnstant 18.4 ±. Mean.86 ±. Sigma.41 ±.19 Entries 1196 P χ / ndf 8.96 / 196 Cnstant 1.7 ±.6 Mean.8618 ±. Sigma.1964 ±.16 Entries P 1196 χ / ndf 86. / 194 Cnstant ±.6 Mean.8646 ±. Sigma.187 ±.16 Entries P 1196 χ / ndf.8 / 19 Cnstant ±.7 Mean.867 ±.19 Sigma.1814 ±.16 P =.1% Entries 1196 P χ / ndf.1 / 194 Cnstant. ±.8 Mean.866 ±.19 Sigma.181 ± GeV θ N =., =.7 N Mm (ed >en γ[x]) P =% P =% P =% P =1% Entries 1777 χ / ndf 7.8 / 197 Cnstant 4.7 ±. Mean.849 ±.6 Sigma. ±.1 Entries χ / ndf / 197 Cnstant. ±. Mean Sigma.861 ±.4.94 ±. Entries χ / ndf / 197 Cnstant ±. Mean Sigma.8678 ±.4.86 ±.1 Entries 1777 χ / ndf 64.1 / 197 Cnstant 19 ±. Mean.869 ±.4 Sigma.8 ±.1 P P =.1% Entries 1777 χ / ndf 77.6 / 197 Cnstant. ±. Mean.869 ±.4 Sigma.8 ± GeV P P =% P P =% P P =% P P =1% Figure 1: Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr, assuming the nminal CLAS1 reslutins n the phtn and electrn kinematic variables, fixing the neutrn θ and reslutins and varying the mmentum reslutin, between.1% and %. Tp plt: phtn reslutins fr the Frward- Tagger ptin. Bttm plt: phtn reslutins fr the IC ptin.

26 P θ =.7, N N =., =% N Mm (ed >en γ[x]) P N Entries Smear: e, γ,n 1196 χ / ndf 89.7 / 194 Cnstant 1.9 ±.7 Entries 1196 Mean.8646 ±.19 Mean.876 Sigma.187 ±.16 Entries Smear: e, γ 1196 χ / ndf / 19 Cnstant ±.8 Mean.867 ±. Sigma.181 ±.1 Entries Smear: γ 1196 χ / ndf 116. / 19 Cnstant ±.8 Mean.8667 ±.19 Sigma.174 ±.16 Entries Smear: e 1196 χ / ndf 9.9 / 9 Cnstant ± 6.6 Mean.8764 ±.9 Sigma.86 ±.1 Entries Smear: N 1196 χ / ndf 16. / 77 Cnstant ± 9.8 Mean Sigma.874 ±.8.6 ± GeV 4 P θ N N =., =.7, =% N P N 4 Mm (ed >en γ[x]) Entries Smear:e, γ,n 1777 χ Entries / ndf / 197 Cnstant 199. ±. Mean.876 Mean.8678 ±.4 Sigma.86 ±.1 Entries Smear:e, γ 1777 χ / ndf 9.6 / 197 Cnstant ±. Mean.874 ±.4 Sigma.8 ±. Entries Smear:γ 1777 χ / ndf.8 / 197 Cnstant 19.9 ±. Mean.8699 ±.4 Sigma.799 ±.1 Entries Smear:e 1777 χ / ndf 96.4 / 9 Cnstant 19. ± 7. Mean.876 ±.8 Sigma.48 ±.9 Entries Smear:N 1777 χ / ndf 181 / 7 Cnstant 118. ±. Mean.871 ±.9 Sigma.196 ± GeV Figure 14: Missing mass squared f the enγ system, fr the n-dvcs channel, simulated with ur event generatr. The different clrs crrespnd t different cmbinatins f chices f particles being detected with abslute precisin r with realistic reslutins. Tp plt: phtn reslutins fr the Frward-Tagger ptin. Bttm plt: phtn reslutins fr the IC ptin. 4

27 Figure 1: Drawing f the Central Detectr: the red area represents the free space between the magnet (shaded area) and the CTOF (represented by the bar and its bent light guides). CND: R & D studies and final detectr design The available space in the CLAS1 Central Detectr is limited by the presence f the CTOF and f the magnet, which leave abut cm free (Fig. 1). Hwever, the CTOF can als be used t detect neutrns, adding a cuple f percent f efficiency. The central tracker will be used as a vet fr charged particles. Finally, it is imprtant t remind that there will be a surrunding magnetic field f T, which cmplicates the issue f light cllectin. Mre than ne year f simulatins and R&D studies have been devted t studying the varius ptins fr the CND and its pssible phtdetectrs. After cnsidering and then rejecting the ptin f a spaghetti calrimeter made f lead and scintillating fibers - it has a t high efficiency fr phtns with respect t neutrns - the retained design fr the detectr is a barrel f standard plastic scintillatr bars f trapezidal shape, all with their lng sides parallel t the beam directin (Fig. 16). This gemetry is similar t the ne f the CTOF. As stated in the previus sectin, ne f the tw requirements f the CND is gd neutrn identificatin capabilities. If the charged particles are veted by the central tracker, the nly particles left that can be mistaken fr neutrns are the phtns. Using plastic scintillatrs, the mst straightfrward way t distinguish neutrns frm phtns is by measuring their time f flight (TOF) and cmpare their β s. β is defined as β = l TOF c, (11) where c is the speed f light and l is flight path f the particle frm the target t the

28 Figure 16: Gemetry f the scintillatr barrel fr the Central Neutrn Detectr. The current design cnsists f radial layers each made f 48 trapezidal scintillatr paddles. scintillatr bar, that can be btained, in ur gemetry, as l = z + h, (1) where z and h are the hit psitin alng the z axis (riented, in ur gemetry, with the beam directin) and in the radial directin. Measuring the time f the hit at bth sides f the scintillatr bar gives access t z z = 1 v eff (t left t right ) (1) where v eff is the effective velcity f light prpagatin in the scintillatr material, and t knw h it is necessary t have radial segmentatin (thus h will be given by the distance between the target and the middle f the hit paddle). Early simulatin studies [] had shwn that t ensure a gd phtn/neutrn separatin fr the neutrn mmentum range f the n-dvcs reactin the CND had t be equipped with phtdetectrs ensuring a time reslutin f abut 1 ps..1 Summary f early R&D studies The first part f ur R&D studies had been fcused n studying the timing perfrmances f varius magnetic-field resistant phtdetectrs, t be placed at the tw ends f the scintillatr bars, in the high-magnetic-field regin f the Central Detectr. Measurements f time reslutin with csmic rays have been carried ut using silicn 6

29 phtmultipliers (SiPMs), avalanche pht-dides (APDs), and micr-channel-plate phtmultipliers (MCP-PMTs). Nne f these devices has been retained. A 1x1mm SiPM was tested, and it was rejected because, due its small active surface, it had a t small number f phtelectrns ( 1) and hence yielded a t big time reslutin ( 1 ns). Matrices f SiPM (matrices 4x4 f xmm chips) have als been tested, giving prmising results in terms f time perfrmances [1], but they wuld require a very cmplex custmized read-ut electrnics, calling fr a cuple f years f dedicated R&D. The APD gave t high time reslutin (σ t 1.4 ns), due t its big rise time. Gd timing perfrmances (σ t 1 ps) were btained fr the MCP-PMTs in the measurement withut magnetic field, but when tested in a T magnetic field they displayed a t strng lss f gain []. Anther reasn t abandn the micr-channel- PMTs ptin was lifetime: we cmputed the expected flux f ptical phtns n the CND phtdetectrs due t electrmagnetic backgrund prduced ver the duratin f ur experiment, and it turned ut t be mre than a factr f magnitude higher than the limit quted in the literature [] after which the quantum efficiency f the MCP-PMT drps [4].. Final design and perfrmances As the magnetic-field-resistant phtdetectrs prved t be nt suited fr the requirements f the CND, anther slutin was fund: reading the light nly at the backward end f each scintillatr bar, with an rdinary PMT placed in the lw-field regin (Fig. 17) and cnnected t the bar by a 1-m-lng bent light guide, while the frnt end f the bar is cnnected via a u-turn light guide t the neighbring paddle. The light emitted at the frnt end f ne scintillatr is therefre fed thrugh its neighbring paddle and read by the PMT cnnected t its end (Fig. 18). The current plan fr the detectr segmentatin is t have 48 azimuthal segments and layers in the radial directin, fr a ttal f 144 scintillatr bars, cupled tw-bytw (Fig. 18). This chice has been made t ptimize the light cllectin by matching the surfaces f the scintillatr and f the phtcathde f the PMT. This cnfiguratin has been tested with measurements f time reslutin with csmic rays. A ne-layer prttype has been built fr this gal at the IPN Orsay (Fig. ). It cnsists f tw scintillatr bars (BC48), each 66 cm lng, cm thick and. cm wide, jined at ne end by a u-turn light guide and each cnnected at the ther end t a 1-m-lng bent light guide cupled t tw rdinary PMTs (Hamamatsu R8, at this stage 4 ) (Fig. 1). A semi-circular shape fr the u-turn light guide has been chsen, as it gave a lwer lss f light than the ther slutin tested (triangular shape). Fr the wrapping, aluminum fil has been preferred t Mylar r VM fr its better timing perfrmances and fr its higher pacity, which minimizes crss talk between adjacent paddles []. In the middle f ne f the tw bars, abve and belw it, are placed tw smaller scintillatrs (1 cm thick, having x cm f surface), each read by a fast PMT (Hamamatsu R8), which are used t trigger the data acquisitin and t ensure that the psitin f the hit is knwn. 4 The PMT Hamamatsu R7997, csting abut a third f R8, will be tested in the next weeks. 7

30 Figure 17: Magnetic field map fr the Central Detectr (radial cmpnent n the tp, axial cmpnent n the bttm). The black semi-circle in the tp plt shws the psitin f the PMTs f the Central Neutrn Detectr. 8

31 Figure 18: Design f the Central Neutrn Detectr. Figure 19: Drawing (side cut view) f the CND, placed int the slenid magnet. 9

32 Figure : The ne-layer prttype f the CND during the timing reslutin measurements with csmic rays carried ut at Orsay. Figure 1: Drawing f the three-layer prttype f the CND, under cnstructin.

33 The utput signals f the tp (T) and bttm (B) trigger PMTs, as well as the nes frm the direct (D, the PMT f the bar where the hit takes place) and neighbr (N, the PMT f the adjacent bar) ne, are fed t ADCs and t TDCs (after discriminatin). B and T are als averaged with a mean-timer (MT), which gives the start t the DAQ. The timing reslutin fr the D r N PMTs is given by []: σ D(N) = σ D(N) TRG σ TRG (14) where σd(n) TRG is the reslutin f the time between the trigger and the Direct (Neighbr) signal, and σ TRG = t MT t T = t MT t B (1) is the time reslutin f the trigger. The results f the timing measurements with csmic rays, perfrmed varying the psitin f the trigger PMTs thrughut the length f the scintillatr bar, are shwn in Fig.. Psitin crrespnds t the center f the bar, is clse t the PMT, and - is near the u-turn. Frm these figures, ne can infer that: the u-turn brings abut a factr f f lss f cllected charge the average timing reslutins fr the tw PMTs are: σ D 1 ps and σ N ps. These experimental results have been used in the simulatin and the recnstructin fr the CND see Sectin 6 and Appendix A... Csts and financial supprt fr the CND A significant part f the R&D n the phtdetectrs f the Central Neutrn Detectr and the tests n the prttype, leading t the psitive results presented in this prpsal, have been supprted by the EU Framewrk Prgram 7 thrugh the Integrating Activity HadrnPhysics and the Jint Research Activity "Hardex. The financing f the full CND, estimated t amunt t rughly keurs (withut salaries), is anticipated t stem mstly frm the funding agencies f the Eurpean grups signing this prpsal upn apprval f this prpsal. 6 Simulatin and recnstructin In rder t study the perfrmances f this detectr, its gemetry has been added t the CLAS1 GEANT4-based simulatin package, GEMC [6]. As respect t earlier studies [] nw the Birks effect, fr which the amunt f ptical phtns prduced after a certain energy depsitin in the scintillatr depends n the particle lsing that energy, and the hit digitizatin fr the CND have been intrduced in GEMC [7]. The timing reslutin and the energy lss due t the u-turn gemetry have been included in the simulatin using the values measured in the csmic-rays tests. Details 1

34 Charge One layer prttype, csmic rays 14 1 Neighbr PMT Direct PMT σ t Psitin One layer prttype, csmic rays Trigger Neighbr PMT Direct PMT Psitin Figure : Results f the csmic rays measurements n the ne-layer prttype. Tp: charge cllected by the tw PMTs as a functin f the hit psitin. Bttm: time reslutin f each PMT as a functin f the hit psitin.

35 n the digitizatin and n the hit and event recnstructin are explained in the Appendix (Sectins A.1 and A.). The main pints f the neutrn recnstructin are the fllwing: if n hits are recrded in the Central Tracker, a hit in the CND is cnsidered a neutral (neutrn r phtn); amng all recnstructed neutral hits in the CND, nly thse passing a certain threshld n the energy depsitin are kept; amng these surviving hits, nly thse fr which the hit psitin recnstructed frm the Direct and Neighbr PMTs timings is within the length f the paddle are kept; β is calculated fr each selected hit, t exclude phtns; the angles θ and and the mmentum are cmputed (see Sectin 6. and Appendix A.) fr the identified neutrns. Simulatins, which included all the cmpnents f the Central Detectr, have been run t evaluate the efficiency f the CND fr neutrns, its ability t discriminate between neutrns and phtns, and its angular and mmentum reslutins. Neutrns and phtns f mmenta varying between.1 and 1 GeV/c and having plar angles θ varying between and 7 have been generated at fixed azimuthal angle ( = ), pinting t the center f ne f the scintillatr bars. The results btained with these simulatins are described in Sectins 6.1, 6. and Efficiency The detectin efficiency is defined here as the rati between the number f events fr which a gd hit (see Appendix A.) was recnstructed in the crrect azimuthal bin f the CND and the ttal number f neutrns generated. Several values f energy threshlds, between 1 and MeV, have been tested. Figure shws the efficiency as a functin f the threshld, fr neutrns with mmentum f.4 GeV/c. The different clrs crrespnd t different values f the neutrn plar angle, θ n. The efficiency, which decreases with increasing threshld, ranges between 1% at the lwest threshlds and 7% at the highest nes. This can als be seen in Fig. 4, where the efficiency fr neutrns emitted at 6 is pltted as a functin f mmentum, fr varius values f the threshld n the energy depsitin. Figure shws instead the efficiency as a functin f the mmentum f the neutrn, at a fixed energy threshld f MeV, and fr different values f θ n. All f these plts have been dne with a cut rejecting hits with time f flight larger than 8 ns. This cut has been applied t suppress the events in which the neutrns interact in the magnet (withut depsiting energy in the CND) and rescatter r prduce secndary particles hitting the CND at a later time, cmprmising the PID and the determinatin f the angles. This cut, alng with a chice f threshld n the recnstructed depsited energy f a few MeV ( is the value chsen at the present stage), is effective in remving these secndary hits (mre details n this aspect can be fund at [7, 8]).

36 Efficiency.14.1 θ angles: Depsited energy threshld (MeV) Figure : Efficiency fr the detectin f neutrns having.4 GeV/c f mmentum, as a functin f the threshld n the depsited energy. The efficiency is shwn fr different values f θ n, between and 9. Efficiency Threshlds: 1 MeV 1. MeV MeV. MeV MeV 4 MeV MeV Mmentum (GeV/c) Figure 4: Efficiency fr the detectin f neutrns emitted at 6, as a functin f mmentum, fr 7 different values f the threshld n the depsited energy, frm 1 t MeV. 4

37 Efficiency θ angles: Mmentum (GeV/c) 6 7 Figure : Efficiency fr the detectin f neutrns, as a functin f neutrn mmentum, fr a -MeV threshld n the depsited energy. The efficiency is shwn fr three different values f θ n, between and Angular and mmentum reslutins The reslutins n the plar angle θ f the neutrn that can be btained with the CND are strngly linked t its TOF reslutin. The angle θ is in fact given by θ = (18/π) arccs( z ave ) (16) l where l and z ave, defined in Appendix A., bth depend n the time measurement. Using the value A =.4 ns MeV 1/, deduced frm the measurements n the prpttype, fr the gaussian smearing n the timing (see Appendix A.), the θ reslutin was studied with GEMC, as a functin f neutrn mmentum and θ itself. The results are shwn in Fig. 6, where the angular reslutin σ θ, btained via gaussian fits f the simulated θ distributins, is pltted as a functin f θ, fr a particular value f neutrn mmentum (.4 GeV/c). σ θ increases slightly with the angle and als is fairly cnstant as a functin f neutrn mmentum, and its value is between 1. and.. The reslutin n the azimuthal angle is directly cnnected t the ttal number f scintillatr bars alng. In fact, the bin size is given by = 6 N = 7. (17) where N paddle is the ID number f the paddle where the hit tk place, and N is the ttal number f paddles in (48 fr the current design f the CND). σ can be taken as half f, therefre.7.

38 ( ) σ θ Layer 1 Layer Layer θ ( ) Figure 6: Angular reslutin σ θ as a functin f θ fr neutrns f mmentum.4 GeV/c, fr a -MeV threshld n the depsited energy. The three clrs f the pints crrespnd t the three radial layers f the CND. The reslutin n the neutrn mmentum, which is btained knwing β and having perfrmed the particle identificatin, accrding t the frmula p = β m n 1 β, (18) is als strictly cnnected t the TOF reslutin. Figure 7 shws the mmentum reslutin σ p /p as a functin f mmentum fr neutrns emitted with θ = 6 : it increases with increasing mmentum, and ranges between 4% and 11%. N appreciable variatins f mmentum reslutin are bserved by varying the neutrn plar angle. 6. Particle Identificatin Since the charged particles passing thrugh the CND will be veted by the Central Tracker, the nly particles that culd be mistaken fr neutrns in the CND are the phtns. The efficiency f the CND fr phtns has been estimated by simulatins, and it is cmparable t the ne fr neutrns (f the rder f %, see Fig. 8, fr phtn energies dwn t. GeV, while it drps t zer fr lwer energies). Neutrns can be discriminated frm phtns by means f their β. Therefre, the β distributins that can be btained with the CND fr neutrns and phtns have been studied with the help f the GEMC simulatin. After chsing a gd hit as described in Appendix A., β is cmputed as β = l TOF true c, (19) 6

39 / p σ p Layer 1 Layer Layer p (GeV/c) Figure 7: Mmentum reslutin σ p /p as a functin f p fr neutrns having θ = 6, fr a -MeV threshld n the depsited energy. The three clrs f the pints crrespnd t the three radial layers f the CND. Efficiency p (GeV/c) γ Figure 8: Efficiency fr the detectin f phtns, as a functin f phtn mmentum, fr a -MeV threshld n the depsited energy. The efficiency is shwn fr θ γ = 6. Belw E γ =.1 GeV, the phtn efficiency drps t zer. 7

40 where l = h + z ave, () h is the distance frm the vertex t the middle f the layer where the hit tk place, TOF true is the recnstructed time f flight and z ave is the recnstructed psitin f the hit (see Appendix A. fr mre details n hw the latter tw quantities are btained fr the u-turn design f the CND). Figure 9 shws the cmparisn, fr each f the radial layers and integrating ver the azimuthal angle, between the β distributins btained fr neutrns f varius mmenta (.,.4,.7 and 1 GeV/c) and fr 1-GeV phtns (in black). All particles in this plt are emitted at θ = 6. Neutrns f mmentum f.9-1 GeV/c can be taken as phtns, as their β distributins begin t verlap, while the n/γ separatin is clear fr lwer mmenta which crrespnd t mst f the range f interest fr n-dvcs, as nly abut 8% f the events are expected t have p n >.9 GeV/c. This is evident als frm Fig., where the errr bars crrespnd actually t σ, where σ is the gaussian width f each β distributin. Equal neutrns and phtn yields have been assumed fr this study. This assumptin is addressed and justified in Sectins 7.1 and Backgrunds n the CND As described earlier, phtns are the main surce f backgrund fr the CND, as they can be mistaken fr neutrns. Charged particles, instead, will be veted by the Central Tracker. Tw kinds f phtns can cntribute t this backgrund: physical events, fr instance π prductin where ne f the tw decay phtns is emitted at backwards angles, and phtns prduced by electrmagnetic reactins f the electrn beam in the target. 7.1 Physics backgrund An estimate f the hadrnic backgrund has been deduced, using the clasdis event generatr (based upn PYTHIA). The backgrund events that culd mimic a n-dvcs event are thse having: ne energetic phtn (E γ > 1 GeV) in the frward directin, and ne phtn in the central detectr. Fr these kinds f events, the estimated rate at full luminsity ( cm s 1 per nuclen) in the DIS kinematics is.4 KHz. If ne als requires the missing mass fr the eγ system (calculated n a neutrn target) t be belw 1 GeV/c, the rate drps t Hz. Assuming a 6% acceptance fr the electrns and fr the phtns detected in the frward detectr, the rate ges dwn t abut 4 Hz. Figure 1 shws the θ distributin as a functin f the energy fr the remaining phtns in the CND. They are mstly emitted at energies belw MeV. Finally, keeping int accunt % f efficiency f the CND fr phtns and this is an upper limit, cnsidering that the 8

41 Events 4 Layer 1 Neutrns. GeV/c Neutrns.4 GeV/c Neutrns.7 GeV/c Neutrns 1 GeV/c Phtns 1 GeV/c β Events 4 Layer Neutrns. GeV/c Neutrns.4 GeV/c Neutrns.7 GeV/c Neutrns 1 GeV/c Phtns 1 GeV/c β Events Layer Neutrns. GeV/c Neutrns.4 GeV/c 4 Neutrns.7 GeV/c Neutrns 1 GeV/c Phtns 1 GeV/c β Figure 9: β distributins fr neutrns with p n =. GeV/c (green), p n =.4 GeV/c (purple), p n =.7 GeV/c (blue), p n = 1 GeV/c (red), and phtns with E = 1 GeV. Each bx shws the results fr ne f the three radial layers that cmpse the CND (the innermst is n the tp, the utermst is n the bttm). The threshld n the depsited energy is MeV. The plts shw all hits, integrated ver. Equal neutrn and phtn yields have been assumed here. 9

42 β Layer Phtns Neutrns Mmentum (GeV) β Layer Phtns Neutrns Mmentum (GeV) β Layer Phtns Neutrns Mmentum (GeV) Figure : β versus mmentum fr neutrns (red) and phtns (blue) with mmenta between. and 1 GeV. The errr bars are defined as σ, where σ is the fitted width f each β peak. The threshld n the depsited energy is MeV. 4

43 θ γ (deg) E γ (GeV) Figure 1: θ versus energy fr the phtns emitted at backward angles assciated t an electrn e and an energetic phtn γ detected in the frward part f CLAS1. The cut MM(eγ) < 1 GeV/c is applied. CND efficiency fr phtns drps t zer fr E γ <.1 GeV, the resulting rate is abut.6 hz. This shuld be cmpared t the ne f neutrns frm n-dvcs, which, as is reprted in Sectin 9, is abut 4 Hz. The assumptin f equal rates, made when studying the phtn/neutrn separatin capabilities f the CND (see Sectin 6.), is therefre a very cnservative ne. Under this assumptin, it was shwn that in the CND phtns can be distinguished frm neutrns prvided that the latter have mmenta belw 1 GeV. This crrespnds t the majrity f the n-dvcs events, fr which neutrns are mstly emitted with mmentum arund.4 GeV. 7. Electrmagnetic backgrund In rder t evaluate the effects f the electrmagnetic backgrund n the Central Neutrn Detectr, in particular t estimate the actual rates seen by the CND due t the backgrund and the energy and timing distributins f the backgrund hits, GEMC simulatins have been run in the fllwing cnditins [9]: the primary electrn has been generated ging frward (t simulate the real hadrnic event), plus rughly 8 ther electrns have been thrwn, distributed in a 14 ns windw in bunches ns apart, riginating cm upstream the target. 8 is apprximately the number f beam electrns that wuld pass thrugh the target in a 14 ns time windw at the nminal CLAS1 luminsity. 14 ns is the typical time windw f the DAQ expected fr CLAS1, which crrespnds t ne event in CLAS1. These electrns then interact 41

44 with the target itself, prducing an electrmagnetic backgrund hitting the neutrn detectr. Figure, prduced with the interactive versin f GEMC, shws ne typical backgrund event in the Central Detectr: the red tracks crrespnd t negatively charged particles (electrns) while the green nes are neutrals (mstly phtns). The hits in the CND, in green/blue-ish, are mainly due t phtns. The utput f these simulatins has been analyzed using the event-recnstructin algrithm adpted t recnstruct neutrns in the CND. In general, the energy depsited in the CND by the electrmagnetic-backgrund phtns is quite small, as it can be seen in Fig., where the energy depsitin in the whle CND is pltted, befre any recnstructin cuts are applied. These phtns tend t release their energy mainly in the first radial layers f the CND, as shwn in Figure 4. If n threshld n the depsited energy r timing cuts are applied, the ttal rate n the CND due t the electrmagnetic backgrund is abut GHz. Cutting n the depsited energy at MeV and n the time at 9 ns, values which has been chsen t ptimize the PID and angular reslutin (Sectin 6.1), the rate drps t abut KHz. These hits can mimic a fake n-dvcs event by accidental cincidence with hadrnic events where an electrn and a phtn are detected in the frward part f CLAS1. Assuming a rate fr such events f the rder f 1 KHz, the accidental cincidence rate is r the rder f. Hz, which is almst tw rders f magnitude less than the rate f n-dvcs neutrns (see Sectin 9). Als fr this type f backgrund, the assumptin f equal rates between neutrns and phtns, made in Sectin 6., is quite cnservative. 8 enπ (p) backgrund Once the events cntaining ne electrn, ne neutrn and ne phtn are selected, the n-dvcs/bh final state can be islated by cutting n the enγ missing mass. Hwever, due t the finite reslutins n the varius kinematic variables measured, the final event sample will still be cntaminated by enγ events cming frm the enπ (p) channel, where ne phtn frm the π decay is detected in the frward part f CLAS1 while the ther escapes detectin. This cntaminatin will be evaluated and subtracted as dne in previus DVCS CLAS analyses [11], by extracting exclusive enπ (p) events detecting bth decay phtns frm the data, and using Mnte Carl simulatins t evaluate the rati f acceptances f π events with 1 and phtns detected. The final number f n-dvcs/bh events, in each 4-dimensinal bin, will be btained as: N DV CS (Q,x B, t,) = N enγx (Q,x B, t,) N π 1γ(Q,x B, t,) (1) where N π 1γ(Q,x B, t,) = N data π (Q,x B, t,) NMC π 1γ (Q,x B, t,) N MC π γ (Q,x B, t,) With the aid f ur event generatrs, we have estimated the expected level f π cntaminatin fr the prpsed experiment. Bth the angular and mmentum reslu- 4 ()

45 Figure : View (frm the beam s perspective) f the Central Detectr, fr ne simulated backgrund event. Red tracks crrespnd t negatively charged particles, green tracks crrespnd t neutrals. This picture has been btained fr a luminsity L = cm s 1, crrespnding t 1/ f the nminal luminsity, fr practical reasns related t the graphical interface. 4

46 Cunts Energy (MeV) Figure : Distributin f the energy depsit in the CND, integrated ver all the azimuthal and radial bins, f the hits cming frm the generated electrmagnetic backgrund. The majrity f the events crrespnds t an energy depsitin belw 1 MeV. tins fr neutrns btained with the simulatin f the CND, as well as the reslutins n electrns and phtns cming frm CLAS1 FASTMC, have been implemented in the n-dvcs/bh and enπ (p) event generatrs. Fiducial cuts have been applied n all the three final-state particles. Fr the enπ (p) crss sectin, we have used the mdel fr exclusive π electrprductin n the nuclen develped by J.M. Laget [4]. This mdel is based n Regge thery with the inclusin f rescattering prcesses. It prvides estimatins f crss sectins which are in agreement with the nes measured by the JLab cllabratins f Hall A [41] and Hall B [4] n the prtn. In this mdel, in the kinematical dmain explred in this prpsal, the exclusive π crss sectin n the prtn is basically equal t the ne n the neutrn. The tw plts f Fig. shw, superimpsed, the enγ missing mass squared fr n-dvcs/bh (red) and enπ (p) (black) events, prduced by ur event generatrs, integrated ver the full kinematic range f interest. The tp plt shws the expected distributin if the lw-angle phtns are detected in the Frward Tagger. Applying the cut MM < 1. GeV the π cntaminatin is arund 1%. The bttm plt is dne assuming that the DVCS phtn is detected with the current IC. In this case, given the larger width f the prtn-mass peak, in rder t keep rughly the same amunt f DVCS events, the cut MM < 1. GeV is applied. Belw this cut, abut 19% f the events cme frm the enπ (p) reactin. 44

47 y (mm) x (mm) 1 Figure 4: Event distributin fr the electrmagnetic backgrund in the CND, as a functin f the x and y crdinates in the lab frame (z being the beam directin), withut any cut n the depsited energy. It can be seen that the majrity f the events are cncentrated in the innermst layer f the CND. 4

48 1 8 DVCS Signal=91% Pi cntaminatin=1% Mm (GeV ) DVCS Signal=89% Pi cntaminatin=19% Mm (GeV ) Figure : Missing mass f the enγ system, fr the n-dvcs/bh channel (in red), and the ed e nπ (p) channel (in black), bth simulated with ur event generatr. CLAS1 and CND reslutins are applied. Tp plt: phtn reslutins fr the Frward-Tagger ptin. Bttm: phtn reslutins fr the IC ptin. 46

49 9 Cunt-rate estimate The n-dvcs/bh final state will be recnstructed by detecting the scattered electrn and the DVCS/BH phtn in the frward part f CLAS1 and the recil neutrn mstly in the CND, as very few neutrns are emitted in the frward directin with enugh mmentum t be detected in EC with appreciable efficiency. The expected number f recnstructed events fr n-dvcs/bh has been calculated, as a functin f the kinematics, with the event generatr described in Sectin 4. The frward-clas1 fiducial cuts have been included, and an verall % neutrn-detectin efficiency (keeping int accunt the few percents f efficiency that can be btained with the CTOF) fr neutrns with θ > 4 has been assumed. The electrn and phtn-detectin efficiencies fr the Frward Detectr have been assumed t be %, within the fiducial cuts. The calculatin has been dne fr a luminsity L = cm s 1 per nuclen and fr 8 days f running time. The fllwing 4-dimensinal grid f bins has been adpted here: 4 bins in Q [1,,.,, GeV /c ] 4 bins in t [,.,.,.8,1. GeV /c ] 4 bins in x B [.,.1,.,.4,.7] 1 bins in, each wide. The number f events, fr each 4-dimensinal bin (Q, x B, t and ), has been cmputed as: N = dσ dq dx B dtd t Q x B L T Acc Eff, () dσ where dq dx B dtd is the 4-fld differential crss sectin, T is the running time, L the luminsity, Acc is the bin-by-bin acceptance and Ef f is the neutrn-detectin efficiency. In Table 1 the expected 4-fld differential crss sectins, the 4-dimensinal acceptance (times the neutrn detectin efficiency) and the crrespnding number f events are listed fr ne particular kinematic bin (< t >=. GeV, < Q >=.7 GeV, < x B >=.7) as a functin f. These yields have statistical errrs between.% (fr the lwest and highest bins, where Bethe-Heitler dminates) and %. The quantities listed in Table 1 are als shwn in Fig. 6. The statistical errrs n the beam-spin asymmetries will then depend n the values f the BSA itself (A) and f the beam plarizatin (P ), thrugh the frmula: σ A = 1 P (1 P A) N. (4) Figure 7 shws the expected accuracy n the n-dvcs/bh beam-spin asymmetry, cmputed using the VGG mdel and assuming J u =. and J d =.1, fr the kinematic bin f Table 1. A beam plarizatin f 8% has been assumed. The errr bars are in average f the rder f % in relative. The prjectins fr the cunt rates and fr the 47

50 dσ/dq dtdx B d (nb/gev 4 ) 1 1 (degrees) Acceptance (degrees) 4 1 (degrees) Figure 6: Tp: crss sectin fr n-dvcs/bh. Middle: acceptance fr the eγ(p) final state, including nly the frward part f CLAS1, cmputed with ur event generatr and FASTMC. Bttm: expected cunt rate fr 8 days f beam time. All three plts are prduced fr the kinematic bin f Table 1. 48

51 Table 1: Expected 4-fld differential crss sectins, 4-dimensinal acceptance times neutrn detectin efficiency and number f events fr n-dvcs/bh with CLAS1 and the CND, as a functin f. < t >=. GeV, < Q >=.7 GeV, < x B >=.7, =, Q = 1. GeV, x B =.1, t =. GeV. The calculatin was dne fr a luminsity L = cm s 1 per nuclen and fr 8 days f running time. ( dσ ) dq dx B dtd (nb/gev4 ) Acc Eff Nb events % % % % % % % % % % % % 1864 BSAs ver the whle grid f bins are shwn in Appendix B. Hwever, as the statistical errr depends n the value f the asymmetry, we will be able t ptimize the bin size f the 4-dimensinal grid nly after having extracted experimentally the BSA. By summing n all the cunt rates btained fr the full grid f bins, we can have an estimate f the ttal expected cunt rate. Overall, rughly millin f n-dvcs/bh events are expected t be cllected ver the full kinematic range f interest, crrespnding t an integrated rate f 4 Hz fr the 8 days f running time. Systematic uncertainties The gal f this experiment is t extract beam-spin asymmetries, which are ratis f plarized crss sectins. In the rati, helicity-independent terms, such as acceptances, efficiencies, radiative crrectins and luminsity, cancel ut, in a first apprximatin. One f the main surces f systematic uncertainty fr the prpsed experiment will be the π backgrund estimatin, which due t the finite size f ur bins will depend n the accuracy f the descriptin f the detectr acceptance and efficiency and n the mdel used in the Mnte-Carl simulatin t describe the enπ (p) reactin (see Eq. ). We estimate this surce t cntribute with % t the verall systematic uncertainties. A similar cntributin will cme frm neutrn/phtn misidentificatin. Due t its strng variatins as a functin and t the size f ur bins, the acceptance will bring an additinal % systematic errr. The measurement f the beam plarizatin will intrduce % f systematic uncertainties. A summary f the uncertainties induced 49

52 Φ (deg.) Figure 7: Beam-spin asymmetry fr n-dvcs/bh as predicted by the VGG mdel (fr J u =. and J d =.1), pltted as a functin f fr the kinematic bin t =. GeV, Q =.7 GeV, x B =.7. The errr bars reflect the expected uncertainties fr ur experiment, crrespnding t 8 hurs f beam time at a luminsity f cm s 1 per nuclen.

53 Surce f errr BSA Beam plarizatin % π cntaminatin % Acceptance % Radiative crrectins 1% n-γ misidentificatin % Ttal 8% Table : Expected systematic uncertainties n the prpsed measurement. by these varius surces can be fund in Table. The ttal systematic uncertainty will be therefre f the rder f 8%, averaged ver all the kinematics (the π -backgrund uncertainty will actually vary depending n the bin). 11 Summary f experimental setup and trigger cnfiguratin We plan t measure beam-spin asymmetries fr the DVCS/BH reactin n the neutrn using a liquid deuterium target and an 11-GeV highly plarized electrn beam. T detect the scattered electrn and phtn we will use the CLAS1 detectr in its baseline cnfiguratin plus, at small angles, a frward electrmagnetic calrimeter either the IC r the Frward Tagger (the secnd ptin being preferable fr its better reslutin perfrmances). Fr the detectin f the recil neutrn we will add ur neutrn detectr t the CLAS1 Central Detectr. T define the trigger fr the data acquisitin, we plan t make use f the experience with CLAS at 6 GeV. The current CLAS electrn trigger is based n the cincidence between Lw Threshld Cherenkv Cunter (LTCC) and EC calrimeter with an energy threshld in the regin f -6 MeV. The cincidence scheme wrks at the sectr level: Trigger = 6 sectr=1 (LTCC sectr EC sectr ) The trigger rate at L = 4 s 1 cm was arund 4 khz in the e1-dvcs experiment at 6 GeV. Due t the lw threshlds in LTCC (N phtelectrns >.) and in the EC calrimeter, nly % f the events were identified as real electrns in the ff-line analysis. Detailed studies f the trigger events shwed that increasing the threshld in the Cherenkv Cunter up t phtelectrns makes the electrn trigger much mre selective. Hwever this methd was nt implemented in the CLAS trigger due t the limited average number f phtelectrns in the LTCC. The CLAS1 trigger system will be significantly imprved. It will be flexible enugh t include different types f detectrs and even spatial crrelatins f the 1

54 Testing and cmmissining Prductin data taking at L = cm s 1 /nuclen Meller plarimeter runs 7 days 8 days days Table : Beam-time request. trigger elements. First f all the High Threshld Cherenkv Cunter (HTCC), with expected extrardinary perfrmances, will participate in the electrn trigger. It was shwn by Mnte-Carl simulatins that even at threshld N phtelectrns > the electrns will be identified with almst % efficiency and the pin-rejectin factr will be much better than fr the present CLAS. The Preshwer Calrimeter can be included int the trigger scheme in case the trigger rate will be unacceptably high. It will imprve the pin rejectin factr keeping the electrn efficiency at high level even with lw energy threshld. The level- CLAS1 trigger will give us the pssibility t gemetrically match the electrn candidate track with the signals frm different detectrs: HTCC, LTCC, Preshwer Calrimeter and EC calrimeter. The CLAS experience tells us that level- trigger is a very pwerful tl t suppress randm cincidences between CC and EC. In summary we can say that the selectivity f the electrn trigger will be much better in cmparisn t the current CLAS. Taking int accunt the imprved perfrmance f the CLAS1 DAQ we hpe that the trigger rate will be at the acceptable level at design luminsity L = s 1 cm per nuclen. Hwever if the backgrund level will be higher than expected we can add additinal detectrs fr the trigger lgic. The trigger fr exclusive reactins may be enfrced by including the energy depsitin in the IC calrimeter (E γ > 1 GeV) that will significantly reduce the trigger rate as we learned frm the CLAS/e1-dvcs experiments. As we plan t use the Central Tracker as vet fr neutrns, we will pssibly intrduce this detectr in the trigger lgic as well. 1 Beam-time request In rder t cllect gd statistics n the BSAs ver the wide phase space cvered by CLAS1 fr n-dvcs (see Appendix B), we request 9 days f beam time, 7 f which will be spent testing the apparatus and in the cmmissining f the experiment, while 8 will be devted t prductin data taking at a luminsity f cm s 1 per nuclen. Meller runs t measure and mnitr the beam plarizatin will take additinal days. Table summarizes ur request. 1 Cnclusins The strng sensitivity t the GPD E f the beam-spin asymmetry fr DVCS n a neutrn target makes the measurement f this bservable very imprtant fr the experimental GPD prgram f Jeffersn Lab. This sensitivity is maximal fr values f

55 x B which are attainable nly with a 11-GeV beam. Mdel predictins shw that fr kinematics that will be available with CLAS1 this asymmetry can be cmparable in size t the ne btained fr prtn DVCS. In rder t measure this reactin ensuring its exclusivity, the detectin f the recil neutrn, which will be mstly emitted at backwards angles, is necessary. We plan t cnstruct a neutrn detectr that will fit in the CLAS1 Central Detectr in the free space between the CTOF and the slenid, cnsisting f a barrel f three layers f scintillatrs cupled at their frnt ends with u-turn light guides and read ut at their back sides by rdinary PMTs cnnected t the bars via 1-m-lng bent light guides and placed in the lw-field regin f the CND. Our GEANT4-based simulatins, calibrated with measurements carried ut n a prttype, shw that the efficiencies btainable with this detectr and its phtn-rejectin capabilities will be sufficient t cllect gd statistics n the BSAs fr the n-dvcs reactin ver a wide phase space, using a ttal f 9 days f beam time. Althugh this is ut f the scpe f this prpsal, this detectr culd als be used in ther experiments requiring the detectin f the recil neutrn (N prgram, fr instance, r all the deeply-virtual mesn prductin reactins n a neutrn), and it can als be useful fr the PID f charged particles via measurement f de/dx and time f flight.

56 A Details n simulatin and recnstructin A.1 Digitisatin f signals frm CND paddles in GEMC GEMC currently accumulates all energy-lss steps within a 4 ns time-windw int ne hit. The digitisatin f the signals frm the CND paddles fllws the fllwing prcedure. Fr the energy digitisatin (ADCs), the calculatin f the depsited energy which will be cnverted int light, E b, fr EACH STEP s in the hit is dne as fllws: E b = E dep (1 + B E dep /step l ) () where E dep is the depsited energy, B is Birk s cnstant (which depends n the material), and step l is the length f the step calculated frm the difference f the vectrs f the current and previus step psitins (except fr the first step, where the first and secnd step psitins are taken). The attenuated energies arriving at the PMTs attached t the ends f the scintillatr paddles, E l, E r (fr left and right ends f paddle), E d (where d stands fr direct ), E n ( neighbr, fr the u-turn cnfiguratin) are summed, fr each hit, ver the depsit frm each step making up the hit: The calculatin fr E l, E r and E d fllws the same frmula, e.g., fr E r we have: E r = s E b / e ( dr/latt) L cll (6) where s is the index f the particular step, d r is the distance frm the step psitin t the end f the scintillatr, l att is the attenuatin length in the material (l att = m) and L cll is the light cllectin efficiency (fractin f phtns which make it int the PMT). L cll is calculated as the crss-sectinal area f the PMT divided by the crss-sectinal area f the paddle, and therefre varies fr each radial layer. Fr the case f E n : E n = s E b / e ( (dr+l) /l att ) L bend L cll (7) where l is the whle length f the paddle and L bend is the energy fractin lst at the u-turn bend cnnecting the neighburing paddles. Its values,., is deduced frm the measurements n the ne-layer prttype carried ut in Orsay (see Fig., tp plt). The cnversin f the energy int ADC values fllws the same methd fr all fur ADC readings (ADCl, ADCr, ADCd, ADCn): ADC = P(E yield Q eff ) gain C ADC + ped (8) where E is ne f E r, E l, E d r E n, yield is the light yield (number f phtns prduced in the scintillatr per unit f depsited energy, namely /MeV), Q eff (=.) is the quantum efficiency f the PMT (fractin f phtelectrns prduced per 4

57 phtn), gain is the PMT gain (=.8 pc/phtelectrn), C ADC (=/pc) is the cnversin factr frm charge t ADC channels, ped is the ADC pedestal (currently set t ) and P(m) is the Pissn distributin functin with mean m, that smears the number f phtelectrns prduced by each hit. The riginal treatment f time f the hit in GEMC was t take an average f the signal time frm all the steps. Since a TDC will trigger when the analgue signal reaches a particular level, the mean ver all times verestimates the actual time. Withut knwing the shape f the ttal signal frm all the steps in a hit as it arrives at the TDC we cannt deduce the time precisely, but we have chsen t set it t the time f the first step in the hit passing a given energy depsitin threshld, as this may be clser t the true time the TDC will trigger at than the mean. The time f arrival f the signal at the PMT is calculated in a similar way fr l, r and d, taking the time f the first step in the hit abve a -kev threshld. Fr r, fr example: t r = T + d r /v eff (9) where T is the time the first energy depsit (first step) in the hit happened (with respect t the event start time), d r is as befre the distance frm the first step energy depsit t the right end f the paddle and v eff is the effective velcity f light in the scintillatr material. Fr t n, the calculatins is: t n = T + (d r + l)/v eff + d u /v eff () where d u is the effective distance thrugh the u-turn light guide. The time is then digitised (in the same way fr all fur TDC branches, TDCl, TDCr, TDCd, TDCn). Fr TDCr fr example: TDCr = t r + G(,A/ E r ) C TDC (1) where G(mean,sigma) is a Gaussian distributin functin, A is a cnstant which determines the smearing in the timing. The value chsen fr A (A =.4 ns MeV 1/ ) is deduced frm the time reslutin csmic-rays measurements n the CND ne-layer prttype described in Sectin []. C TDC (=/ns) is the cnversin factr frm time t TDC channels. A. Hit recnstructin The algrithm fr the chice f the gd hit is here described. Let us assume paddle 1 is ptically cupled t paddle. We get N1 signals in TDC f paddle 1 and N signals in TDC f paddle. The pssible cmbinatins ne can have are: Case A: a signal frm a "direct" hit in paddle 1 is recnstructed alng with a signal frm a "direct" hit in paddle (Fig. 8, tp image). This is wrng.

58 Figure 8: Schematic drawings f the cases the recnstructin deals with when signals are cllected at bth D and N PMTs. Case B: a signal frm a "direct" hit in paddle 1 is recnstructed alng with a signal frm anther "direct" hit in the same paddle, prpagated t paddle. (Fig. 8, secnd image frm the tp). This is als wrng. Case C: a signal frm a "direct" hit in paddle 1 is recnstructed alng with a signal frm the same hit prpagated t paddle (Fig. 8, bttm). This is right. Case D: a signal registered in paddle 1, but it actually prpagated frm "direct" hit in paddle, which was then recnstructed with a signal registered in paddle which had prpagated frm "direct" hit in paddle 1 (Fig. 8, third image frm the tp). This is bviusly als wrng. If nly ne hit was registered in paddle 1, and n hits in its neighbur paddle, then cases A, B and D are nt an issue. Hwever, it is pssible that the recnstructin will fail anyway if the signal frm the "direct" hit was registered but its prpagatin t neighbur wasn t (because f energy attenuatin perhaps it just didn t make it past the threshld, fr example). Thse hits are lst. Fr each decent hit (having energy depsit abve zer and the time in its "direct" paddle s TDC reading smething physical, nt zer r an unstpped TDC value), ne needs t determine whether its cupled neighbur is t its left r right. Then an array is filled with all the physical hits in the neighbur and in the same paddle (discarding thse that have energy depsit r unphysical TDC times (zer r unstpped TDC)). These hits are called "partners". Next, fr each hit, ne iterates thrugh the array f partners. If a partner is a neighbur, ne has t take its "direct" TDC time tp (as recrded in its paddle). If a partner is frm the same paddle as the hit in questin, ne must take its prpagated time tp t neighbur. We are nw wrking with tw time values the ne just selected frm the partner, 6