December 18, 2020 11:41 IJMPA S0217751X20440170 page 1

International Journal of Modern Physics A Vol. 35, Nos. 34 & 35 (2020) 2044017 (13 pages) © World Scientific Publishing Company DOI: 10.1142/S0217751X20440170

A high precision narrow-band beam: The ENUBET project

1,2, 3,4 1,16 1 5,6 1,2 M. Torti, ∗ F. Acerbi, A. Berra, M. Bonesini, A. Branca, C. Brizzolari, G. Brunetti,5 M. Calviani,18 S. Capelli,1,16 S. Carturan,5,7 M. G. Catanesi,8 N. Charitonidis,18 S. Cecchini,9 F. Cindolo,9 G. Collazuol,5,6 E. Conti,5 F. Dal Corso,5 C. Delogu,5,6 G. De Rosa,10,11 A. Falcone,1,2 A. Gola,3 C. Jollet,12 V. Kain,18 B. Kli´cek,13 Y. Kudenko,14 M. Laveder,5,6 A. Longhin,5,6 L. Ludovici,15 E. Lutsenko,1,16 L. Magaletti,8,17 G. Mandrioli,9 A. Margotti,9 V. Mascagna,1,16 N. Mauri,9 L. Meazza,1,2 A. Meregaglia,12 M. Mezzetto,5 M. Nessi,18 A. Paoloni,20 M. Pari,5,6 E. G. Parozzi,1,2 L. Pasqualini,9,19 G. Paternoster,3 L. Patrizii,9 M. Pozzato,9 M. Prest,1,16 F. Pupilli,5 E. Radicioni,8 C. Riccio,10,11 A. C. Ruggeri,10,11 C. Scian,5,6 G. Sirri,9 M. Stip´cevic,13 M. Tenti,9 F. Terranova,1,2 E. Vallazza,1 F. Velotti,18 M. Vesco5,6 and L. Votano20 1INFN Sezione di Milano-Bicocca, Piazza della Scienza 3, 20133 Milano, Italy 2Universit`adi Milano-Bicocca, Piazza della Scienza 3, 20133 Milano, Italy 3Fondazione Bruno Kessler (FBK ), Via Sommarive 18, 38123 Povo (TN ), Italy 4INFN-TIFPA, Universit`adi Trento, Via Sommarive 14, 38123 Povo (TN ), Italy 5INFN Sezione di Padova, via Marzolo 8, 35131 Padova, Italy 6Universit`adi Padova, via Marzolo 8, 35131 Padova, Italy 7INFN, Laboratori Nazionali di Legnaro, Viale dell’Universit`a2, 35020 Legnaro (PD), Italy 8INFN Sezione di Bari, Via Giovanni Amendola 173, 70126 Bari, Italy 9INFN Sezione di Bologna, viale Berti-Pichat 6/2, 40127 Bologna, Italy 10INFN, Sezione di Napoli, Strada Comunale Cinthia, 80126 Napoli, Italy 11Universit`a“Federico II” di Napoli, Corso Umberto I 40, 80138 Napoli, Italy 12CENBG, Universit`ede Bordeaux, CNRS/IN2P3, 33175 Gradignan, France 13Center of Excellence for Advanced Materials and Sensing Devices, Ruder Boskovic Institute, HR-10000 Zagreb, Croatia 14Institute of Nuclear Research of the Russian Academy of Science, Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com 142190 Moscow, Russia 15INFN Sezione di Roma 1, Piazzale A. Moro 2, 00185 Rome, Italy 16Universit`adegli Studi dell’Insubria, Via Valleggio 11, 22100 Como, Italy 17Universit`adegli Studi di Bari, Via Giovanni Amendola 173, 70126 Bari, Italy 18CERN, Esplanade des Particules, 1211 Gen`eve23, Switzerland 19Universit`adegli Studi di Bologna, viale Berti-Pichat 6/2, 40127 Bologna, Italy 20INFN, Laboratori Nazionali di Frascati, via Fermi 40, 00044 Frascati (Rome), Italy ∗[email protected]

Received 19 March 2020 Accepted 6 November 2020 Published 18 December 2020 by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles.

∗Corresponding author.

2044017-1 December 18, 2020 11:41 IJMPA S0217751X20440170 page 2

M. Torti et al.

The knowledge of the initial flux, energy and flavor of current neutrino beams is the main limitation for a precise measurement of neutrino cross-sections. The ENUBET ERC project is studying a facility based on a narrow-band neutrino beam capable of constraining the neutrino fluxes normalization through the monitoring of the associated charged in an instrumented decay tunnel. In ENUBET, the identification of large-angle from Ke3 decays at single particle level can potentially reduce the νe flux uncertainty at the level of 1%. This setup would allow for an unprecedented mea- surement of the νe cross-section at the GeV scale. This input would be highly beneficial to reduce the budget of systematic uncertainties in the next long baseline oscillation projects. Furthermore, in narrow-band beams, the transverse position of the neutrino interaction at the detector can be exploited to determine a priori with significant pre- cision the neutrino energy spectrum without relying on the final state reconstruction. This contribution will present the advances in the design and simulation of the hadronic beam line. Special emphasis will be given to a static focusing system of secondary that can be coupled to a slow extraction scheme. The consequent reduction of particle rates and pile-up effects makes the determination of the νµ flux through a direct monitoring of after the dump viable, and paves the way to a time-tagged neutrino beam. Time-coincidences among the at the source and the neutrino at the detector would enable an unprecedented purity and the possibility to reconstruct the neutrino kinematics at source on an event-by-event basis. We will also present the performance of tagger prototypes tested at CERN beamlines, a full simulation of the positron reconstruction chain and the expected physics reach of ENUBET.

Keywords: ; neutrino beam; calorimeter.

PACS numbers: 29.40. n, 29.90.+r −

1. Introduction Over the last 50 years, accelerator neutrino beams1 have been developed toward higher intensities, but uncertainties in the flux and in the flavor composition are still very large. Thanks to the progress in neutrino scattering experiments,2,3 the measurements of neutrino cross-sections are now limited by the knowledge of the initial fluxes, since the yield of neutrinos is not directly measured but it is extra- polated from hadro-production data and detailed simulations of the neutrino beam- line. This limitation bounds the precision that can be reached in the measurement Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com of the absolute cross-sections to (5–10%). Moreover, -based sources mainly O produce νµ while most of the next generation oscillation experiments will rely on the appearance of νe at the far detector. In fact, a direct measurement of the νe cross-sections is of great interest for the current (T2K, NOνA) and next (DUNE, HyperK) generation of oscillation experiments.4 These considerations motivated the development of “monitored” neutrino beams5 and the Enhanced NeUtrino BEams from Tagging (ENUBET) pro- posal:6 a neutrino beam facility where the dominant source of electron neutrino contamination is the three body semileptonic decay of the K+ π0e+ν → e (Ke3), in which the electron neutrino flux is monitored directly by the observation of large-angle positrons in the decay tunnel. 7

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. The ERC ENUBET (“Enhanced NeUtrino BEams from kaon Tagging”) project is aimed at building a detector capable of identifying positrons from Ke3 decays

2044017-2 December 18, 2020 11:41 IJMPA S0217751X20440170 page 3

ENUBET neutrino beam

while operating in the harsh environment of a conventional neutrino beam decay tunnel. Since March 2019, ENUBET is also a CERN Neutrino Platform project (NP06/ENUBET) developed in collaboration with CERN A&T and CERN-EN. The ENUBET neutrino beam is a conventional narrow-band beam with a short ( 20 m) transfer line followed by a 40 m long decay tunnel. Particles produced ∼ by the interaction of on the target are focused, momentum selected and transported at the entrance of the tunnel, as described in Sec. 2. The particles that reach the decay tunnel are hence , kaons and protons within the momentum bite of the transfer line (10% in ENUBET). Off-momentum particles are mostly low energy pions, electron, positron and photons from tertiary interactions in the collimators and other components of the beamline, and muons from pion decay that cross the collimators. The rates of these particles are several orders of magnitude smaller than beams currently in operation and particle creation in the decay tunnel can be monitored at single particle level by instrumenting a fraction of the decay tunnel. Three different prototypes were built and tested at CERN PS East Area in order to select the appropriate layout for the tagger demonstrator: two prototypes are shashlik-based calorimeters (see Sec. 3) while the third one was built with lateral scintillation light readout (see Sec. 4). Particles identification is performed by a Neural Network approach based on a Multivariate Data Analysis as described in Sec. 5. ENUBET will provide a precise measurement of νe and νµ fluxes and exploit correlation between the radial position of the neutrino interaction and its energy, as described in Sec. 6. Finally, ENUBET aims at to set the first milestone toward a “time-tagged neu- + trino beam,” where the νe at the detector is time-correlated with the produced e in the decay tunnel, as we will see in Sec. 7.

2. The ENUBET Beamline Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com The ENUBET beam is a conventional narrow-band beam with two differences with respect to the current beams: the decay tunnel is not located along the proton axis of the focusing system and the proton extraction length is slow ( (s)). Particles O produced by proton interactions in the target are focused, momentum selected and transported to the tunnel entrance. Noninteracting protons are stopped in a proton beam dump. Off-momentum particles reaching the decay tunnel are mostly low energy particles coming from interactions in the collimators and other beamline components together with muons that cross absorbers and collimators. The hadron beam considered has a reference momentum of 8.5 GeV/c with a momentum bite of 10% and we expect 50% of K+ to decay in a 40 m long tunnel. ∼ ENUBET goals can be achieved using conventional magnets by maximizing the + +

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. number of K and π at tunnel entrance, by minimizing the total length of the transfer line to reduce kaon decay losses and by keeping under control the level of

2044017-3 December 18, 2020 11:41 IJMPA S0217751X20440170 page 4

M. Torti et al.

Table 1. Expected π+ and K+ for proton on target (PoT) at the decay tunnel entrance for the two possible focusing schemes. The improvement factor in kaon transport with respect to Ref. 5 is shown in the last column.

Focusing π/PoT K/PoT Extraction π/cycle K/cycle Factor 3 3 10 10 [10− ] [10− ] length (10 ) (10 ) with respect to Ref. 5 Horn-based 97 7.9 2 ms 438 36 2 × Static 19 1.4 2s 85 6.3 4 ×

background transported. Momentum and charge-selected K+, π+ being injected in the instrumented decay tunnel, need to be collimated enough such that  any undecayed is capable of escaping the region without hitting the tagger inner surface: this limits the particles rate at the instrumentation. Primary proton interactions in the target are simulated with FLUKA,8 con- sidering various proton drivers (400 GeV, 120 GeV and 30 GeV protons) and target designs (Be or Graphite). The results reported in this document refer to 400 GeV protons and a Be target 110 cm long with a 3 mm diameter. The optic optimization is performed with TRANSPORT9 to match the ENUBET specifica- tions for momentum bite and beam envelope. The beam components and lattice are then implemented in G4Beamline10 that fully simulates particle transport and interactions. Two possible beamlines have been considered: the first one makes use of a focus- ing horn placed between the ENUBET target and the following transfer line (“horn- based transfer line”) while in the second one the transfer line quadrupoles are placed directly downstream the target (“static transfer line”). The performances of the static design, as shown in Table 1, turned out to be better than early estimates reported in the ENUBET proposal5 and it offers several advantages in terms of cost, simplification of technical implementation and performance of particle identi- fication.11 Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com 2.1. Static transfer line The static configuration allows to perform the focusing using DC operated devices (unlike pulsed magnetic horns) compatible with a traditional slow extraction of several seconds. The best configuration (see Fig. 1) consists in a quadrupole triplet followed by a dipole that provides a 7.4◦ bending angle and by another quadrupole triplet. The rate of background particles is several orders of magnitude smaller than present beams and the instrumentation located in the decay tunnel can monitor lepton production at single particle level. Figure 2 shows the momentum distribu- tion and the XY profile of K+ entering/exiting the decay tunnel.

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. The length of the decay tunnel is optimized in order to have Ke3 decays as the dominant νe source: electron neutrinos from decay in flight of kaons represent

2044017-4 March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

4 M. Torti et al.

Two possible beamlines have been considered: the first one makes use of a focusing horn placed between the ENUBET target and the following transferline (“horn- based transferline”) while in the second one the transferline quadrupoles are placed directly downstream the target (“static transferline”). The performances of the static design, as shown in Tab. 1, turned out to be better than early estimates re- ported in the ENUBET proposal5 and it offers several advantages in terms of cost, simplification of technical implementation and performance of particle identifica- tion.11

Table 1. Expected π+ and K+ for proton on target (PoT) at the decay tunnel entrance for the two possible focusing schemes. The improvement factor in kaon transport with respect to Ref.5 is shown in the last column. Focusing π/PoT K/PoT Extraction π/cycle K/cycle Factor 3 3 10 10 5 [10− ] [10− ] length (10 ) (10 ) w.r.t Horn-based 97 7.9 2 ms 438 36 2 × Static 19 1.4 2s 85 6.3 4 ×

2.1. Static transferline The static configuration allows to perform the focusing using DC operated devices December 18,(unlike 2020 11:41 pulsed magnetic horns) compatible IJMPA with a traditional slow extraction S0217751X20440170 of page 5 several seconds. The best configuration (see Fig. 1) consists in a quadrupole triplet followed by a dipole that provides a 7.4◦ bending angle and by another quadrupole triplet. The rate of background particles is several orders of magnitude smaller than present beams and the instrumentation located in the decay tunnel can monitor lepton production at single particle level. Fig. 2 shows the momentum distribution and the + XY profile of K entering/exiting the decay tunnel. ENUBET neutrino beam

March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

Fig. 1. Schematics of the ENUBET beam beam in in the the static static focusingENUBET focusing option. option. neutrino beam 5

The length of the decay tunnel is optimized in order to have Ke3 decays as the dominant ν source: electron neutrinos from decay in flight of kaons represent 97% e ∼ of the overall νe flux. Positrons from three body decays are emitted at large angles and hit the instrumented walls of the tunnel before exiting. 3 + The static beamline transports at the tunnel entrance 19 10− π /PoT and ×

Fig.Fig. 2. 2. Left: (Color momentum online) Left: distribution momentum of K distribution+ entering(black)/exiting(blue) of K+ entering (black)/exiting the decay tunnel. (blue) Mid- the dle:decay XY tunnel. profile Middle: of the KXY+ beamprofile at of tunnel the K entrance.+ beam at Right: tunnel XY entrance. profile of Right: the KXY+ beamprofile at of tunnel the exit.K+ beam at tunnel exit.

97% of3 the+ overall νe flux. Positrons from three body decays are emitted at large 1.4 10− K /PoT in [6.5 10.5 GeV/c] range, improving by 4 times the kaon yield ∼ × ÷ withangles respect and hit to the the instrumented first estimate walls reported of the in5 tunneland requiring before exiting. about 4.5 1019 PoT 3 + The static beamline transports at the tunnel entrance 19 10− π ×/PoT and at CERN SPS to carry out both νµ and νe cross section programs. An additional 3 + × 1.4 10− K /PoT in [6.5 10.5 GeV/c] range, improving by four times the kaon advantage× of the static solution÷ is the possibility to directly monitor the rate of muonsyield with from respectπ+ decays to afterthe first the hadronestimate dump: reported the reduced in Ref. yield 5 and with requiring respect about to the 4.5 1019 PoT at CERN SPS to carry out both ν and ν cross-section programs. horn-based× solution and the consequent lower particleµ ratee allows for a monitoring An additional advantage of the static solution is the possibility to directly monitor of the νµ flux at single particle level. the rate of muons from π+ decays after the hadron dump: the reduced yield with respect to the horn-based solution and the consequent lower particle rate allow for 3. Instrumentation of the decay tunnel a monitoring of the νµ flux at single particle level.

Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com The positron monitoring is achieved instrumenting the decay tunnel. The require- ments3. Instrumentation for the detector of are: the a Decayhigh e/π Tunnelseparation capability to remove the main source of background; radiation hardness, since it will operate in a harsh environ- The positron monitoring is achieved instrumenting the decay tunnel. The require- ment; cost effectiveness, since a large fraction of the decay tunnel will be instru- ments for the detector are: a high e/π separation capability to remove the main mented; fast recovery time to cope with the expected rate of 200 kHz/cm2. These source of background; radiation hardness, since it will operate in a harsh environ- requirements are fullfilled by a sampling calorimeter with a longitudinal segmenta- ment; cost effectiveness, since a large fraction of the decay tunnel will be instru- tion for positron tagging complemented by a photon veto, in order to discriminate mented; fast recovery time to cope with the expected rate of 200 kHz/cm2. These neutral pions and gammas. A shashlik calorimeter12, 13 with longitudinal segmenta- requirements are fullfilled by a sampling calorimeter with a longitudinal segmenta- tion was identified as a solution for the ENUBET requirements. tion for positron tagging complemented by a photon veto, in order to discriminate The final layout of the tagger will be composed by three calorimeter layers and neutral pions and gammas. A shashlik calorimeter12,13 with longitudinal segmen- equipped by the photon veto. tation was identified as a solution for the ENUBET requirements.

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. The final layout of the tagger will be composed by three calorimeter layers and 3.1.equippedShashlik by the prototype photon veto. The first shashlik prototype was composed by five, 15 mm thick, iron layers inter- 2044017-5 leaved by 5 mm thick plastic scintillator tiles14, 15 . The total length of the module (10 cm) and its transverse size (3 3 cm2) corresponds to 4.3 X and 1.7 Moliere × 0 radii, respectively. Nine wavelength shifting (WLS) fibers crossing the UCM are con- nected directly to 1 mm2 Silicon Photomultipliers (SiPMs) through a plastic holder. SiPMs are hosted on a PCB embedded in the calorimeter structure, as schematized in Fig. 3 right. In November 2016, a prototype consisting of a 7 4 2 array of UCMs was ex- × × posed, at the CERN PS East Area facility, to beam composed by electrons, muons, December 18, 2020 11:41 IJMPA S0217751X20440170 page 6

M. Torti et al.

3.1. Shashlik prototype The first shashlik prototype was composed by five, 15 mm thick, iron layers inter- leaved by 5 mm thick plastic scintillator tiles.14,15 The total length of the module (10 cm) and its transverse size (3 3 cm2) correspond to 4.3 X and 1.7 Moliere × 0 radii, respectively. Nine wavelength shifting (WLS) fibers crossing the UCM are connected directly to 1 mm2 Silicon Photomultipliers (SiPMs) through a plastic holder. SiPMs are hosted on a PCB embedded in the calorimeter structure, as schematized in Fig. 3 (right). In November 2016, a prototype consisting of a 7 4 2 array of UCMs was × × exposed, at the CERN PS East Area facility, to beam composed by electrons, muons, and pions with momentum in the range of interest for neutrino physics applications (1–5 GeV), in order to measure its performances.16 The scintillator tiles were machined and polished from EJ-200 and BC-412 sheets and painted with a diffusive TiO2 based coating (EJ-510) to increase the light collection efficiency. After painting, nine holes were drilled in each tile in order to accommodate the WLS fibers (Y11 and BCF92) read out by FBK 20 µm-pixel SiPMs. Ancillary detectors were used to identify particle type and isolate samples of electrons and muons. During the data taking the calorimeter was tilted at different angles (0, 50, 100, 200 mrad) with respect to the beam direction. The tilted geo- metry reproduces the operating condition of the calorimeter in the decay tunnel + of neutrino beams, where positrons from Ke3 reach the detector with an average angle of 100 mrad. ∼ The measured deposited energy is in good agreement with the results obtained from a Monte Carlo simulation that took into account all the different components of the beam. The electromagnetic energy resolution is well described by 17%/ E (GeV) ∼ showing that the dominant contribution to the resolution is due to the sampling p term and that the energy response is not affected by the light readout scheme em-

Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com ployed to achieve longitudinal segmentation. A good e/π separation, with a pion misidentification lower than 3%, is achieved exploiting the longitudinal segmenta- tion of the calorimeter and the different energy deposit patterns. Furthermore, the analysis of the tilted runs indicates that the performance of the calorimeter is similar to the 0 mrad run in the angular range of interest for ENUBET. Finally, the calorimeter shows linear response, within < 3%, in the whole range of interest in both standard (0 mrad) and tilted runs. The integration of the light readout into the calorimeter module is a very effec- tive solution but results in exposing the SiPMs to fast neutrons produced by hadronic showers. In order to evaluate the effect of the radiation on the perfor- mances of the SiPMs, three PCBs hosting 9 SiPMs and the single-SiPM PCB were irradiated from a minimum dose of 1.8 108 n/cm2 up to 1.7 1011 n/cm2 at the × × by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. irradiation facility of INFN-LNL (Laboratori Nazionali di Legnaro).

2044017-6 December 18, 2020 11:41 IJMPA S0217751X20440170 page 7 March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

ENUBETENUBET neutrino beam beam 7

Fig.Fig. 3. Left: 3. Left: scheme scheme of of the the baseline baseline shashlik shashlik UCM. Right: Right: pictures pictures of of the the polysiloxane-based polysiloxane-based calorimetercalorimeter prototype. prototype.

All the SiPMs show minor changes in the breakdown voltage, while the dark 3.2. The polysiloxane prototype current after breakdown increases by more than two orders of magnitude at a fluence Nonof conventional1011 n/cm2 options. Finally, based the sensitivity on polysiloxane to single scintillators photoelectron are is lost also at being fluences exam- ∼ ined.larger Polysiloxane than 3 10 is9 an/cm siliconic-based2. scintillator18 that offers several advantages × over plasticThe irradiated scintillators: PCBs better were radiation then installed tolerance, in two reduced calorimeter aging, prototypes no irreversible and deteriorationtested at CERN: caused theby mechanical electron and deformations, mip peak mean exposure value ratio to solvent is constant vapours after and irradiation and the integrated neutron fluence does not affect the dynamic range of high temperatures and, for shashlik calorimeters, no need to drill and insert the the photosensors.17 optical fibers. Silicone rubbers preserve their transparency even after a 10 kGy dose exposure3.2. The and polysiloxanetheir physical prototype properties are constant over a wide temperature range. PolysiloxaneNonconventional can be pouredoptions at based 60◦ C on in polysiloxane the shashlik scintillators module after are the also insertion being of the opticalexamined. fibers Polysiloxane in the absorber, is a siliconic-based simplifying scintillator the assembly.18 that On offers the several other advan- hand, the lighttages yield over is about plastic 30% scintillators: of the EJ-200 better yield. radiation tolerance, reduced aging, no ir- A polysiloxanereversible deterioration shashlik calorimeter caused by prototype mechanical (Fig. deformations, 3 right) composed exposureof to three solvent mod- ulesvapors (6 6 and high15 cm temperatures3 each, 13 X and,in for total) shashlik consisting calorimeters, of 12 no UCM need units to drill and and with × × 0 a 15insert mm scintillator the optical fibers. tile thickness Silicone rubbers was built preserve and exposed their transparency to particle even beams after at a the CERN10 kGy East dose Area exposure in October and their 2017 physical19 . properties are constant over a wide tem- perature range. Polysiloxane can be poured at 60◦C in the shashlik module after

Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com An electromagnetic energy resolution of 17%/ E(GeV) and good linearity (< the insertion of the optical fibers in the absorber,∼ simplifying the assembly. On the 3% in the 1-5 GeV range) were measured. other hand, the light yield is about 30% of the EJ-200p yield. A polysiloxane shashlik calorimeter prototype (Fig. 3 (right)) composed of three 4. Designmodules and(6 6 test15 cmfor3 theeach, lateral 13 X in scintillation total) consisting light of 12 readout UCM units and with × × 0 prototypea 15 mm scintillator tile thickness was built and exposed to particle beams at the CERN East Area in October 2017.19 The shashlikAn electromagnetic option has two energy disadvantages: resolution of SiPMs17%/ sufferE (GeV) large and radiation good linearity levels be- ∼ ing lying(< 3% within in the 1–5 the GeV calorimeter range) were bulk measured. and they are not accessible for maintenance during the data taking. For this reason we implementedp a different solution for the scintillation4. Design light and readout. Test for the Lateral Scintillation Light The currentReadout design Prototype consists in a scheme where light is collected from both sides The shashlik option has two disadvantages: SiPMs suffer large radiation levels lying by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. of each scintillator tile through WLS fibers installed in grooves machined on the plastic.within This the setup calorimeter replaces bulk the and shashlik-based they are not accessible UCM with for maintenance a Lateral readout during the Com- pact Module (LCM) in which fibers from the same LCM are bundled to a single 2044017-7 SiPM reading in groups of 10 at a distance of about 30 cm from the bulk of the calorimeter. In this layout SiPMs are not embedded in the calorimeter bulk, thus not exposed March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

8 M. Torti et al.

to the shower, and this setup allows for an easier access regarding maintenance or replacements. In fact, FLUKA simulations20 demonstrate that placing SiPM above 30 cm of bo- rated polyethylene, can reduce the neutron flux on the sensors by a factor 18. In 2018, a prototype made of 3 2 2 LCM was assembled as shown in Fig. 4. × × Each LCM was composed of 1.5 cm thick iron slabs interleaved with 0.5 cm plas- tic scintillators (EJ-204): the modules have thus the same sampling fraction as the shashlik UCM. The prototype was tested at CERN at the T9 beamline in May 2018. Later, in September, a larger prototype consisting of 84 LCMs in the 7 4 3 structure × × was exposed to CERN SPS beam. As for the previous prototypes, during the testbeam the calorimeter was tilted at dif- ferent angles (0, 50, 100, 200 mrad) with respect to the beam direction and exposed December 18,to 2020 electrons, 11:41 pions and muons. These IJMPA particles were identified thanks toS0217751X20440170 ancillary page 8 detectors present in the experimetal area. The testbeam results were compared with a Monte Carlo simulation. The calorimeter response to minimum ionizing particles, electrons and pions were evaluated and showed a good agreement with the predic- tion for the energy range 1-3 GeV. Preliminary results showed an electromagnetic energy resolution at 1 GeV of about 17% which is consistent with the prediction. In addition, data and prediction mutually agreed with a linear response in the range 20 1-3M. TortiGeV. et The al. analysis of testbeam data is still ongoing .

Fig.Fig. 4. 4. Pictures Pictures of the lateral scintillation light readout calorimeter prototype. prototype.

data taking. For this reason we implemented a different solution for the scintillation light readout. 4.1.TheThe current photon design veto consists in a scheme where light is collected from both sides + + 0 Photonsof each scintillator originating tile from throughK WLSπ fibersπ (Kπ installed2) decays in and grooves reaching machined the positron on the 5 → taggerplastic.must This setupbe discriminated replaces the from shashlik-based positrons by UCM a suitable with a photon Lateral veto readout (t0 layer). Com- Thispact detector Module (LCM)also provides in which coarse fibers information from the same on the LCM impact are point bundled of the to a particle single andSiPM timing reading information. in groups ofIn 10 the at baseline a distance option of about for the 30t cm0 layer from the the basic bulk unitof the is acalorimeter. doublet of plastic scintillator tiles with a 3 3 cm2 surface and a 0.5 cm In this layout SiPMs are not embedded in the∼ calorimeter× bulk, thus not exposed to the shower, and this setup allows for an easier access regarding maintenance or replacements. In fact, FLUKA simulations20 demonstrate that placing SiPM above 30 cm of borated polyethylene, can reduce the neutron flux on the sensors by a factor 18. In 2018, a prototype made of 3 2 2 LCM was assembled as shown in Fig. 4. × × Each LCM was composed of 1.5 cm thick iron slabs interleaved with 0.5 cm plastic scintillators (EJ-204): the modules thus have the same sampling fraction as the shashlik UCM. Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com The prototype was tested at CERN at the T9 beamline in May 2018. Later, in September, a larger prototype consisting of 84 LCMs in the 7 4 3 structure was × × exposed to CERN SPS beam. As for the previous prototypes, during the test beam the calorimeter was tilted at different angles (0, 50, 100, 200 mrad) with respect to the beam direction and exposed to electrons, pions and muons. These particles were identified thanks to ancillary detectors present in the experimental area. The test beam results were compared with a Monte Carlo simulation. The calorimeter response to minimum ionizing particles, electrons and pions was evaluated and showed a good agreement with the prediction for the energy range 1–3 GeV. Preliminary results showed an electromagnetic energy resolution at 1 GeV of about 17% which is consistent with

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. the prediction. In addition, data and prediction mutually agreed with a linear re- sponse in the range 1–3 GeV. The analysis of test beam data is still ongoing.20

2044017-8 March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

ENUBET neutrino beam 9

thickness ( 0.02 X ), readout by a WLS fiber optically linked to a SiPM. The ∼ 0 doublets are orthogonal to the tagger axis and mounted below the inner radius of the calorimeter. Adjacent tiles form a ring installed inside the positron tagger. Along the z-axis, t0 layer rings are separated by 7 cm to have on average about five hit doublets for Ke3 positrons. Photon conversions are discriminated from positrons

usingDecember the 18,information 2020 11:41 on the pulse height IJMPA in all hit scintillators. S0217751X20440170 Photon converted page 9 in the photon veto are rejected requiring a signal compatible with a single mip (1-2 mip separation). The t0 layer was tested in stand alone configuration and, in September 2018, coupled with the calorimeter (see Fig. 5) during the test beam at CERN. Data analysis is ongoing20 . ENUBET neutrino beam

Fig. 5. FourFig. doublets 5. Four of doublets the photon of the photon veto veto installed installed below below the the lateral lateral readout readout prototype during prototype the during the September 2018 test at CERN. The t0 layer is visible in the bottom right side of the picture. September 2018 test at CERN. The t0 layer is visible in the bottom right side of the picture. 4.1. The photon veto + + 0 Photons originating from K π π (Kπ2) decays and reaching the positron 5 → tagger must be discriminated from positrons by a suitable photon veto (t0 layer). 5. ParticleThis reconstruction detector also provides coarse information on the impact point of the particle and timing information. In the baseline option for the t0 layer the basic unit is Positron identificationa doublet of plastic has scintillator been simulated tiles with a starting3 3 cm2 surface from and particles a 0.5 cm transported ∼ × thickness ( 0.02 X ), readout by a WLS fiber optically linked to a SiPM. The by the ENUBET beamline∼ 0 at the entrance of the decay tunnel. The ENUBET 21–23doublets are orthogonal to the tagger axis and mounted below the inner radius GEANT4 of thesimulation calorimeter. Adjacent includes tiles particle form a ring tracking installed inside and thedetector positron response tagger. and the full particleAlong identification the z-axis, t0 (PID)layer rings chain are separated from by the 7 cm event to have builder on average to about the fivepositron iden- tification. PIDhit doublets is based for Ke3 onpositrons. a Multivariate Photon conversions Data are Analysis discriminated (MTVA) from positrons analysis that using the information on the pulse height in all hit scintillators. Photon converted employs variablesin the photon constructed veto are rejected from requiring the energy a signal deposit compatible ineach with a module. single mip The first step(1–2 needed mip separation). for PID is the definition of the event by the ENUBET Event The t0 layer was tested in stand alone configuration and, in September 2018, Builder (EB).coupled with the calorimeter (see Fig. 5) during the test beam at CERN. Data 20 24, 25 In theInt. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com standardanalysis positron is ongoing. identification analysis , all the modules of the first e.m. layer in which an energy deposition is found, are ordered in time and the first available UCM5. Particle with a Reconstruction visible energy larger than 20 MeV is selected as the seed of the event. UCMPositron and identificationt signals has close been in simulated time and starting space from to particles the seed transported one are clustered by the ENUBET0 beamline at the entrance of the decay tunnel. The ENUBET to build theGEANT4 event.21–23 Thesimulation procedure includes is particleiterated tracking over and all detector the recorded response and signals. the Charged particlesfull particle will identification deposit (PID) energy chain in from different the event UCMs builder to and the positron the geometrical iden- pat- tification. PID is based on a Multivariate Data Analysis (MTVA) that employs tern of thevariables energy constructed deposition from can the energy be exploited deposit in each to module. perform particle identification. Indeed positronsThe firstwill step produce needed for showers PID is the well definition contained of the event in the by the first ENUBET two calorimeter Event Builder (EB). layers and in few UCMs along the longitudinal coordinate,24,25 whereas pions will have

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. In the standard positron identification analysis, all the modules of the first a less compacte.m. layershower in which and an a energy larger deposition fraction is found, of energy are ordered deposited in time and in the the first third layer.

2044017-9 December 18, 2020 11:41 IJMPA S0217751X20440170 page 10

M. Torti et al.

available UCM with a visible energy larger than 20 MeV is selected as the seed of the event. UCM and t0 signals close in time and space to the seed one are clustered to build the event. The procedure is iterated over all the recorded signals. Charged particles will deposit energy in different UCMs and the geometrical pattern of the energy deposition can be exploited to perform particle identification. Indeed positrons will produce showers well contained in the first two calorimeter layers and in few UCMs along the longitudinal coordinate, whereas pions will have a less compact shower and a larger fraction of energy deposited in the third layer. To exploit the calorimeter information as much as possible, we used a Neural Network (NN) based on TMVA multivariate analysis. In particular we used vari- ables such as the fraction of energy in the first e.m. layer, the fraction in the event seed and several others containing information on the shower profile development. To further improve the background rejection, the longitudinal distribution of the events along the tunnel is also used, since the signal is more abundant in the second part of the tunnel due to its geometrical acceptance. The NN analysis allows to separate positrons from pions and muons but does not remove the contamination of photons from π0 decays. The e+/γ separation is performed using the signal of the t0 layer. Looking at the energy deposited in the two layers of the doublet we can discriminate between zero, one or two mip selecting positrons and rejecting photons converting in the first t0 layer. This analysis chain allows to select positrons with a signal-to-noise ratio of 0.5 ∼ and an efficiency of 20%, which is appropriate for the neutrino flux monitoring at ∼ the percent level. This level of reconstruction is achieved for the static focusing sys- tem described before. A recent tuning in the beamline design and the reconstruction algorithms improve these figures significantly.

6. The ENUBET Narrow-Band Beam ENUBET physics program can be enhanced by two additional features: provide a beam with a precisely measured flux and a measurement of the neutrino energy Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com that does not rely on the reconstruction of final state particles (narrow-band off-axis technique).26 In fact, the monitoring of positrons in the decay tunnel provides the νe flux, while the slow extraction scheme allows for a high precision νµ flux measuring the muons rate after the hadron dump. Neutrino fluxes were estimated simulating samples of positive pions and kaons with a momentum bite of 10% centered at 8.5 GeV/c, considering the static transfer line with 400 GeV protons and a liquid argon detector placed at 90 m from the entrance of the decay tunnel. Collecting data for about 1 year of SPS operations (4.5 1019 PoT), at the neutrino detector × about 1.13 106 ν CC and 1.4 104 ν interactions will be observed. × µ × e The narrow-band off-axis technique results from the narrow momentum band-

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. width of the beam and the finite transverse dimension of the neutrino detector. The energy measurement exploits the correlation between the energy of the neutrino

2044017-10 March 19, 2020 14:34 WSPC/INSTRUCTION FILE output March 19, 2020 14:34 WSPC/INSTRUCTION FILE output

MarchDecember 19, 18, 2020 2020 11:41 14:34 WSPC/INSTRUCTION IJMPA FILE output S0217751X20440170 page 11

ENUBET neutrino beam 11 ENUBET neutrino beam 11 ENUBET neutrino beam 11 ENUBET neutrino beam

Fig. 6. Left: radial and energy distribution of νµ CC interactions. The high energy neutrino Fig. 6. (Color online) Left: Radial and energy distribution of νµ CC interactions. The high componentFig.Fig. fromenergy 6. kaon neutrino Left: Left: decays radial radial component and is and shownenergy fromenergy distribution kaon with distribution decays red boxes, isof shownνµ ofCC whileν withµ interactions.CC red the interactions. boxes, low The while energy high the The energyone low high energy is neutrino reported energy neutrino with black boxes.componentcomponentone Right: is reportedν fromµ fromCC with kaon kaon spectra black decays decays boxes. normalized is shown is Right: shown withνµ withtoCC red unity spectra boxes, red boxes, obtained whilenormalized the while low selecting to theunity energy low obtained one interactions energy is reported selecting one is atwith reported different with blackinteractions boxes. Right: at differentν CC radial spectra distances normalized from the to unitybeam obtainedaxis. Region selecting of interest interactions for future at long different radial distancesblack boxes. from theRight: beamµνµ CC axis. spectra Region normalized of interest to unity for future obtained long selecting baseline interactions experiments at different are baseline experiments are enlightened. enlightened.radialradial distances distances from from the the beam beam axis. axis. Region Region of interest of interest for future for longfuture baseline long baseline experiments experiments are are enlightened. enlightened. interacting in the detector and the radial distance R of the interaction vertex from be performedbethe performed beam without axis without (Fig. relying 6 relying(left)). on Asthe on shown the reconstruction reconstruction in Fig. 6 (right), of of the itthe is final possible final state state to products collect products for for beνµ performedCC event samples without covering relying the onenergy the range reconstruction of interest for of future the long final baseline state products for the determinationthe determination of the of neutrino the neutrino energy. energy. Absolute Absolute ννµµcrosscross section section measurements measurements theoscillation determination experiments, of the selecting neutrino interaction energy. in radial Absolute windowsν ofcrossR 10 section cm. measurements can be performedcan be performed combining combining the thenarrow-band narrow-band off-axis off-axis technique techniqueµ with± with the measure- the measure- can beThe performed incoming neutrino combining energy the is determined narrow-band with a off-axis precision technique given by the with pion the measure- ment of thementpeakνµ of widthflux the ν of fromµ theflux spectrum pionfrom pion decay at decaya fixed during duringR as shown the the extraction. extraction. in Fig. 7. Only The The a flux rough flux measurement cut measurement on ment of the ν flux from pion decay during the extraction. The flux measurement can be exploitedcanthe be visible exploited using energyµ scintillatorsusing is needed scintillators to separate located located the afterν afterµ from the pion hadron hadron decay dump. from dump. the Studiesνµ Studiesorigi- on this on this possibilitycanpossibilitynating are be ongoing. exploited from are the ongoing. two-body using scintillators kaon decay. In located this way, after differential the hadron cross-section dump. mea- Studies on this possibilitysurements can are be ongoing. performed without relying on the reconstruction of the final state products for the determination of the neutrino energy. Absolute νµ cross-section Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com

Fig. 7. Beam energy spread (in black) and peak energy (in red) as a function of the distance R of the interaction vertex at the neutrino detector from the beam axis. by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. Fig. 7. (Color online) Beam energy spread (in black) and peak energy (in red) as a function of Fig. 7. Beamthe distanceenergyR spreadof the interaction (in black) vertex and at peakthe neutrino energy detector (in red) from the as beam a function axis. of the distance R Fig. 7. Beam energy spread (in black) and peak energy (in red) as a function of the distance R of the interaction vertex at the neutrino detector from the beam axis. of the interaction vertex at the neutrino2044017-11 detector from the beam axis. 7. From monitored to tagged neutrino beam A purely static focusing system opens up several opportunities beyond the original 7. From7.goals monitored From of ENUBET. monitored to Since tagged in to the tagged staticneutrino focusing neutrino beam system beam the proton extraction can last up to several seconds, the instantaneous rates of particles hitting the decay tunnel A purelyA static purely focusing static focusing system system opens opens up several up several opportunities opportunities beyond beyond the the original original walls is reduced by about two orders of magnitude compared with the horn option. goals of ENUBET.goals of ENUBET. Since in Since the static in the focusingstatic focusing system system the proton the proton extraction extraction can can last last up to severalup to seconds, several seconds, the instantaneous the instantaneous rates rates of particles of particles hitting hitting the the decay decay tunnel tunnel walls is reducedwalls is reduced by about by two about orders two orders of magnitude of magnitude compared compared with with the the horn horn option. option. December 18, 2020 11:41 IJMPA S0217751X20440170 page 12

M. Torti et al.

measurements can be performed by combining the narrow-band off-axis technique with the measurement of the νµ flux from pion decay during the extraction. The flux measurement can be exploited using scintillators located after the hadron dump. Studies on this possibility are ongoing.

7. From Monitored to Tagged Neutrino Beam A purely static focusing system opens up several opportunities beyond the original goals of ENUBET. Since in the static focusing system the proton extraction can last up to several seconds, the instantaneous rates of particles hitting the decay tunnel walls are reduced by about two orders of magnitude compared with the horn option. In the ENUBET static option the time between two Ke3 decays is 11.3 ns. A neutrino interaction in the detector can thus be time linked with the observation of its associated lepton in the decay tunnel, giving a “tagged neutrino beam.”28–32 In order to suppress accidental coincidences between the neutrino interaction and uncorrelated particles inside the beam pipe, the timing precision of the detectors that are used to instrument the decay tunnel must reach 100 ps. This precision is needed also to associate the positron with the other decay products of the kaons.33 The physics potential of tagged neutrino beams is enormous because they provide energy and flavor measurement on event-by-event basis and are the ideal tools to study cross-sections and nonstandard oscillation phenomena, including sterile neu- trinos. R&D on the photon tagger and the neutrino detector to bring the required time resolution have been started.

8. Conclusions During the first years of activities, ENUBET has achieved important milestones and extended remarkably its physics case. The R&D carried out includes the design of the transfer line and the tests of the prototypes at the CERN East Area. The project

Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com is aimed at delivering a complete design of the ENUBET beamline by 2021. Beyond νe cross-section measurements, the ENUBET physics potential is being investigated for the study of νµ interaction and the determination of differential cross-sections employing the narrow-band off-axis technique. The static focusing system opens up the possibility of transforming ENUBET in the world first tagged neutrino beam and the corresponding technical challenges are being addressed.

Acknowledgments This project has received funding from the European Union’s Horizon 2020 Research and Innovation program under Grant Agreement No. 681647 and by the Italian Ministry of Education and Research — MIUR (Bando “FARE,” progetto NuTech).

by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles. It is also supported by the Department of Physics “G. Occhialini,” University of Milano-Bicocca (project 2018-CONT-0128).

2044017-12 December 18, 2020 11:41 IJMPA S0217751X20440170 page 13

ENUBET neutrino beam

References 1. S. E. Kopp, Phys. Rep. 439, 101 (2007). 2. T. Katori and M. Martini, J. Phys. G 45, 013001 (2018). 3. L. Alvarez-Ruso et al., Prog. Part. Nucl. Phys. 100, 1 (2018). 4. A. M. Ankowski and C. Mariani, J. Phys. G 44, 054001 (2017). 5. A. Longhin, L. Ludovici and F. Terranova, Eur. Phys. J. C 75, 155 (2015). 6. A. Berra et al., CERN-SPSC-2016-036, SPSC-EOI-014, Geneva, 2016. 7. F. Acerbi et al., CERN-SPSC-2018-034, SPSC-I-248, Geneva, 2018. 8. A. Fass`o,A. Ferrari, J. Ranft and P. R. Sala, FLUKA: A multi-particle transport code, CERN-2005-10 (2005), INFN/TC-05/11, SLAC-R-773. 9. D. C. Carey, K. L. Brown and F. Rothacher, SLAC-R-95-462, Stanford, 1995. 10. T. J. Roberts, K. B. Beard, D. Huang, S. Ahmed, D. M. Kaplan and L. K. Spentzouris, Conf. Proc. C WEPP120, 0806233 (2008). 11. G. Brunetti et al., arXiv:1903.09044. 12. H. Fessler, P. Freund, J. Gebauer, K. M. Glas, K. Pretzl, P. Seyboth, J. Seyerlein and J. C. Thevenin, Nucl. Instrum. Methods A 228, 303 (1985). 13. G. S. Atoian et al., Nucl. Instrum. Methods A 320, 144 (1992). 14. A. Berra et al., Nucl. Instrum. Methods A 830, 345 (2016). 15. A. Berra et al., IEEE Trans. Nucl. Sci. 64, 1056 (2017). 16. G. Ballerini et al., J. Instrum. 13, P01028 (2018). 17. F. Acerbi et al., J. Instrum. 14, P02029 (2019). 18. A. Quaranta et al., J. Non-Cryst. Solids 357, 1921 (2011). 19. F. Acerbi et al., Nucl. Instrum. Methods A 956, 163379 (2020). 20. F. Acerbi et al., J. Instrum. 15, P08001 (2020). 21. S. Agostinelli et al., Nucl. Instrum. Methods A 506, 250 (2003). 22. J. Allison et al., IEEE Trans. Nucl. Sci. 53, 270 (2006). 23. J. Allison et al., Nucl. Instrum. Methods A 835, 186 (2016). 24. A. Meregaglia, Talk at 30th Rencontres de Blois, Blois, 3–8 June 2018. 25. F. Pupilli, PoS NEUTEL2017, 078 (2018). 26. F. Pupilli, PoS NOW2018, 030 (2018). 27. L. N. Hand, Proceedings of Second NAL Summer Study, Aspen, Colorado, 9 June–3 August 1969. 28. B. Pontecorvo, Lett. Nuovo Cimento 25, 257 (1979). 29. P. Denisov et al., preprint IHEP 81-98, Serpukhov, 1981.

Int. J. Mod. Phys. A 2020.35. Downloaded from www.worldscientific.com 30. R. H. Bernstein et al., -Proposal-0788, 1989. 31. L. Ludovici and P. Zucchelli, arXiv:hep-ex/9701007. 32. L. Ludovici and F. Terranova, Eur. Phys. J. C 69, 331 (2010). 33. F. Acerbi et al., Input document for the European Particle Physics Strategy, arXiv:1901.04768 [physics.ins-det]. by EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) on 01/13/21. Re-use and distribution is strictly not permitted, except for Open Access articles.

2044017-13