AIDA Advanced European Infrastructures for Detectors at Accelerators. Deliverable Report

Transcription

1 AIDA-D8.11 AIDA Advanced European Infrastructures for Detectors at Accelerators Deliverable Report Infrastructure performance and utilization: TASD and MIND are constructed and tested for their performance Noah, E. (University of Geneva) et al 29 January 2015 The research leading to these results has received funding from the European Commission under the FP7 Research Infrastructures project AIDA, grant agreement no This work is part of AIDA Work Package 8: Improvement and equipment of irradiation and test beam lines. The electronic version of this AIDA Publication is available via the AIDA web site < or on the CERN Document Server at the following URL: < AIDA-D8.11

2 Grant Agreement No: AIDA Advanced European Infrastructures for Detectors at Accelerators Seventh Framework Programme, Capacities Specific Programme, Research Infrastructures, Combination of Collaborative Project and Coordination and Support Action DELIVERABLE REPORT UTILIZATION: TASD AND MIND ARE CONSTRUCTED AND TESTED FOR THEIR PERFORMANCE DELIVERABLE: D8.11 Document identifier: Due date of deliverable: End of Month 48 (January 2015) Report release date: 29/01/2015 Work package: Lead beneficiary: WP8: Improvement and equipment of irradiation and test beam lines STFC Document status: Final Abstract: Two prototype neutrino detectors based on plastic scintillators are presented in this deliverable report D8.11 for the AIDA project. The Totally Active Scintillator Detector (TASD) is fully operational on the MICE beam line at RAL where it measures muon sample purity since October A new magnetization scheme is proposed for the Magnetized Iron Neutrino Detector (MIND) improving charge identification efficiencies for muons with momenta < 1 GeV/c. Extensive tests of MIND detector components are presented. Copyright AIDA Consortium, 2015 Grant Agreement PUBLIC 1 / 24

3 Copyright notice: Copyright AIDA Consortium, 2015 For more information on AIDA, its partners and contributors please see The Advanced European Infrastructures for Detectors at Accelerators (AIDA) is a project co-funded by the European Commission under FP7 Research Infrastructures, grant agreement no AIDA began in February 2011 and will run for 4 years. The information herein only reflects the views of its authors and not those of the European Commission and no warranty expressed or implied is made with regard to such information or its use. Delivery Slip Name Partner Date Authored by A. Blondel, E. Noah P. Soler UNIGE UNIGLA 20/01/2015 Edited by A. Blondel, E. Noah P. Soler UNIGE UNIGLA 27/01/2015 G. Mazitelli [WP coordinator] INFN Reviewed by M Moll [WP coordinator] CERN 27/01/2015 L. Serin [Scientific coordinator] CNRS Approved by Steering Committee 29/01/2015 Grant Agreement PUBLIC 2 / 24

4 TABLE OF CONTENTS 1. INTRODUCTION TASD TASD DESIGN ASSEMBLY AND TESTING INSTALLATION ON BEAM LINE DETECTOR PERFORMANCE MIND SIMULATIONS MAGNET REDESIGN ELECTRONICS COMPONENT PERFORMANCE FUTURE PLANS CONCLUSION REFERENCES ANNEX: GLOSSARY Grant Agreement PUBLIC 3 / 24

5 Executive summary At its inception, the AIDA project foresaw a task dedicated to prototyping neutrino detectors based on plastic scintillator bars: a Totally Active Scintillator Detector (TASD) and a Magnetized Iron Neutrino Detector (MIND). The TASD was designed, constructed and commissioned with cosmic rays at the University of Geneva. It was then transported to the MICE beam line at RAL in the UK where it measured its first charged particle beams in October Its function is to ensure good muon purity by rejecting electrons originating from the decay of muons in the MICE cooling channel. The requirement was for 1% contamination, which it fulfils by achieving a muon sample purity of 99.85%. No dead channels have been recorded to date, and a software routine corrects for the 2 mismatched optical channels out of a total of 5664 channels corresponding to the double-sided readout of 2832 triangular plastic scintillator bars. It is due to instrument the MICE beam line until The Daya Bay results on measurements of 13 published in March 2012 signalled a shift in prospects for neutrino facilities worldwide. Plans to build a standard MIND were reformulated to design a MIND in line with requirements for future experiments. The resulting redesign of the magnetization scheme enables better muon charge identification efficiencies at lower momenta. The design of AIDA MIND detector modules was based on experience gained with the TASD. Specific differences include the adoption of silicon photomultipliers rather than photomultiplier tubes due to the magnetic field, and a new electronics readout system. Extensive tests of the rectangular plastic scintillator bars and several silicon photomultiplier variants are presented, showing excellent light yields of >50 photo-electrons per photo sensor, with optical crosstalk at the level of 5%. 1. INTRODUCTION This document reports on activities carried out within the context of Task of the ECfunded AIDA project. The goal of the task was to design, construct and assess the performance of two prototype detectors: a Totally Active Scintillator Detector (TASD) prototype a Magnetised Iron Neutrino Detector (MIND) prototype When the AIDA project was first proposed in 2010, these prototypes were foreseen to be tested at CERN. However, test beams were not available at CERN during 2013 and 2014, due to a shutdown of the accelerator complex for upgrades and maintenance work. Following completion of the design and construction of the TASD detector at the University of Geneva, it was installed at the MICE facility at the Rutherford Appleton Laboratory for testing. The publication of Daya Bay measurements of 13 in March 2012 prompted a re-evaluation of the relevance of the MIND prototype for future neutrino studies. The AIDA management and steering group chose to go through a redesign phase for the MIND, affecting the magnet and electronics readout chain. MIND prototyping was redefined towards a new design of the magnet [1], and selection, procurement and testing of detector elements. A Milestone MS33 report submitted February 2014 outlined first results of the TASD detector on the MICE beam line [2]. We report here on the main design aspects and detector performance. In a separate section we report on the MIND redesign, and characterization of MIND detector components. Grant Agreement PUBLIC 4 / 24

6 2. TASD The TASD activities covered the following major items: 1) Detector design 2) Development of numerous procedures for construction 3) Implementation of quality assurance plans 4) Development of electronics readout scheme 5) Development of DAQ software for control and readout of data. 6) Assembly of the detector 7) Detector simulation software 8) Full functionality tests with cosmic rays 9) Transportation of the detector from Geneva to the UK 10) Installation on the MICE beam line of the Rutherford Appleton Laboratory 11) Beam tests at MICE 12) Data analysis A detailed account of the items above is well beyond the scope of this report. The reader is referred to a Ph.D. thesis available in the public domain, which covers extensively over 250 pages all aspects [3]. In this document, we choose to highlight arbitrarily some features of the design and summarise the performance assessment. Notable subjects absent from this account are the design of the electronics system, and the complete redesign of the optical readout system following test results that showed the initial design led to 10% faulty channels. They both represent a significant body of work but are omitted here due to space limitations TASD DESIGN The design is based on 24 detector modules, each module consisting of two planes covering an active area of 1.27 m 2 as shown in Figure 1. A plane houses 59 triangular scintillator bars made of extruded plastic scintillator supplied by Fermilab. A total of 2832 bars instrument the detector. A wavelength shifting fiber is embedded inside each plastic scintillator bar. It's function is to collect the light produced when a charged particle impinges on the detector, shift the wavelength, and transport the light to both ends of the bar, where custom optical connectors allow coupling to light-collecting clear fibers. The process of gluing the WLS fiber inside the plastic scintillator bars was carried out at a rate of 60 bars/day with 3150 bars processed in total to ensure enough bars were available for assembly, following testing of every individual bar according to a quality assurance protocol. There are two light readout systems, one for each end of every bar: On one side, every individual bar is connected to clear fibers that are coupled to a 64- channel multi-anode photomultiplier tube (PMT), which converts the light to 64 equivalent electrical signals, keeping the granularity of the detector. On the other side, all bars are connected to a clear fiber bundle coupled to a singleanode PMT which sums the light output from all channels in a plane, and converts it to an equivalent electrical signal for the whole plane. Grant Agreement PUBLIC 5 / 24

7 Figure 1: TASD detector module showing two planes of scintillator bars, the top plane aligned perpendicular to the bottom plane. The mechanics for the detector were constructed to support the modules, the associated electronics and designed for straightforward installation at Geneva and RAL and the stresses of transportation between those two sites, Figure 2. Figure 2: CAD drawing of the mechanics of the TASD detector. The mass of the active part of the detector is approximately 1 t. Grant Agreement PUBLIC 6 / 24

8 2.2. ASSEMBLY AND TESTING The effectiveness of the quality assurance procedures was demonstrated by the absence of dead channels following assembly, Figure 3. Cosmic ray tests determined that two channels were mismatched, i.e. there was an incorrect clear fiber channel match to the corresponding scintillator bar in one of the fiber bundles for a total of 5664 clear fibers. This represents an excellent achievement considering the fiber bundles were manufactured in-house by technicians with no prior experience of handling optical fibers. Characterization of optical crosstalk was also performed with cosmic rays. Some uncertainties in the alignment of the clear fiber bundle to the 64-ch PMT could lead to light from a given scintillator bar illuminating the wrong PMT cell. The crosstalk was seen to increase as a function of signal strength. It is around 0.1% for typical MIP particles and does not require specific attention. Misalignment of the fiber bundle connector was estimated from a weighted average of signals extracted from 8 neighbouring channels of the calibration channel, Figure 4. This misalignment ranges from -0.8 to 0.6 mm in the X-direction and from -0.2 to 1 mm in the Y- direction. Figure 3: Assembly of the TASD detector planes at the University of Geneva. Left: March 2013, 2 planes; Center: May 2013, 24 planes; Right: August 2013, 48 planes. Figure 4: Measurement of fiber misalignment, Left: Fiber bundle connector used for the 64-channel PMT, showing the calibration channel illuminated with a light source; Right: misalignment of the fiber connector with respect to the 64-channel PMT. Each point is a weighted average of the 8 channels surrounding the calibration channel. Grant Agreement PUBLIC 7 / 24

9 2.3. INSTALLATION ON BEAM LINE After confirmation of the functionality of the detector with cosmic rays at the University of Geneva, the detector was transported to the UK, Figure 5, and installed at the end of the MICE beam line at RAL in September 2013, Figures 6, 7. It is referred to as the Electron Muon Ranger (EMR) and carries out the function of measuring the range of muons. It performs a rejection of electrons from the decay of muons inside the MICE cooling channel. The particle composition and beam momentum can be selected by varying currents on the beam line magnets. Six types of beam settings are possible, positive or negative electron, muon and pion beams. The settings optimize particle rates for a given particle species, although all species are present in all types of beam. For the beam tests, it was decided to scan momenta over the range 280 to 540 MeV/c. Figure 5: Left: Loading of the TASD and associated electronics rack onto a truck on 25 th September 2013 for transportation from Geneva to RAL in the UK. Right: photo taken 27 th September 2013 at the MICE facility at RAL. Figure 6: MICE beam line at the Rutherford Appleton Laboratory in the UK. The TASD, referred to as the Electron Muon Ranger (EMR), was installed at the end of the beam line. Grant Agreement PUBLIC 8 / 24

10 Figure 7: EMR installation in the MICE beam line. Left: patch panels, Center: control rack (not in final position), Right: cabling DETECTOR PERFORMANCE Results of the detailed characterisation offline provided confidence the detector would perform well online. The data flow for analysis is shown in Figure 8. As can be seen, data analysis can result from very different initial data sets such as simulations with Geant 4, cosmic ray runs, beam tests or calibration tests with an LED system. Figure 9 illustrates an event display for a simulated 200 MeV negative muon impinging on the detector. Such simulations with Geant 4 provided detailed energy loss rates and event topologies, Figure 10. Figure 8: Data flow for the TASD installed at the MICE beam line. Grant Agreement PUBLIC 9 / 24

11 Figure 9: Event simulation in the TASD with Geant 4. A 200 MeV negative muon stops in plane 21. After 2.2 microseconds corresponding to the muon lifetime, it decays into an electron and electron anti-neutrino and muon neutrino. Figure 10: Simulated shower shapes of 250 MeV/c electrons (top), muons (middle), pions (bottom). Both negative (left) and positive (right) particles are simulated. The particle rates per spill over this range, and the detector occupancy for a 200 MeV/c electron beam are shown in Figure 11. The range of muons in the TASD as a function of momentum was determined to be R = (0.5 pz 68) [cm], Figure 12. In order to assess detector efficiency, several parameters were identified as potentially useable for electron discrimination such as plane density, average bar multiplicity, bar multiplicity in the first plane, track depth, transverse energy deposition in the two detector projections. The selection and detailed analysis of all these parameters is the Grant Agreement PUBLIC 10 / 24

12 subject of a second Ph.D. thesis to be submitted by the University of Geneva. In summary, the plane density and the average bar multiplicity are the most obvious differences between electrons and muons. The plane density is the ratio of number of hit planes over total number of planes along a particle track and is approximately 60% for electrons and 100% for muons. Using this parameter alone to discriminate the 11.7% electrons present in the beam, the purity of the muon sample reaches 99.85%. This is well within the requirement of 1% beam contamination downstream of the cooling channel set for MICE. Multivariate analysis is underway to determine limits for this technology concerning electron discrimination. Figure 11: Left: Particle rate per spill for all particle types (e,, ) as a function of beam momentum. Right: Detector occupancy for 200 MeV/c electron beam. Figure 12: Muon range in the detector. Left: track depth in units of number of planes as a function of momentum. Right: data used in the range fit. Grant Agreement PUBLIC 11 / 24

13 3. MIND MIND-type detectors have been successfully designed, built and operated over several decades. They are particularly well suited for charged current (CC) neutrino interactions with an outgoing muon in the final state, see Figure 13. A MIND is the baseline detector for a Neutrino Factory, where a muon of opposite charge to that expected from the neutrino content of the beam, is detected through CC interactions of the oscillated muon neutrino, hence the requirement for magnetization for charge separation. Figure 13: Sketch of basic event topologies in a MIND. The dashed lines with the W/Z bosons are for illustration purposes only and do not represent tracks. Despite the considerable amount of experience gained on this detector technology, there are several gaps in our knowledge which merit further detailed studies. This was the original motivation for a MIND prototype proposal within AIDA, designed to address specifically: Muon charge identification, for wrong sign muon signature of a neutrino oscillation event: golden channel at a neutrino factory: requires correct sign background rejection of 1 in 10 4, 0.8 to 5 GeV/c. Hadronic shower reconstruction for identification of charged current neutrino interactions and rejection of neutral current neutrino interactions. Test beam: protons/pions 0.5 to 9 GeV/c. The case for a Neutrino Factory has weakened due to the large value of 13, published by the Daya Bay collaboration in March 2012, which opens up possibilities for measurements of mass hierarchy and CP violation in the leptonic sector with other neutrino sources and detectors. During the course of the AIDA project, the importance of water cherenkov (SK, HK, ESS), liquid scintillator and liquid argon detectors (LBNE, LBNO) for future neutrino facilities was re-affirmed. These large-scale detectors are not magnetized and charge identification of outgoing leptons from the interaction vertices of neutrinos contributes to systematic errors. Recently, several studies have been launched to explore how a MIND downstream of a water cherekov (TITUS-HK) or LAr (LBNO) could reduce systematic errors by providing muon charge ID. These changes prompted a re-evaluation of the pertinence of the initial goals of the project as stated in the AIDA report entitled "MIND redesign justification and status for Deliverable D8.11" [1]. In agreement with AIDA management, the community decided to redesign the magnet in order to improve charge identification efficiencies below 1.0 GeV/c, to potentially as low as 0.3 GeV/c where Multiple scattering Grant Agreement PUBLIC 12 / 24

14 significantly affects angular resolution. We report on several aspects of the MIND prototype namely simulations, magnet design, electronics and components performance SIMULATIONS Simulations of the initial design of the MIND prototype with standard transmission lines for magnetization were carried out with the simulation framework developed for the Neutrino Factory and nustorm studies [4]. Octagonal plates 2 m 1 m and two transmission lines, with a 2.8 m detector depth (along the beam axis) are simulated [5]. Four different scenarios were tested to validate the steel thickness and for an indication of whether the choice of scintillator geometry is acceptable (0.7 cm high rectangular bars vs 1.7 cm high triangular bars), scintillator pitch = 1.0 cm in all scenarios: 3 cm steel plate, 1.5 cm scintillator module (i.e cm X plane cm Y plane), 2 cm steel plate, 1.5 cm scintillator module, 3 cm steel plate, 3.5 cm scintillator module, 2 cm steel plate, 3.5 cm scintillator module. In assessing the various efficiencies from simulations for a small MIND prototype exposed to a charged particle beam, the following were taken into consideration: a) The total number of tracks (or simulated particles in the detector); b) The number of tracks reconstructed (using the Kalman filter); c) The number of successful tracks (where successful means that the correct charge is identified. The efficiencies are then defined as: 1) The reconstruction efficiency is b/a; 2) The charge identification efficiency is c/a. Muon reconstruction efficiencies are shown in Figure 14. All four combinations of steel and scintillator thicknesses show good efficiencies at low momenta < 2 GeV/c. Above 1 GeV/c, the combination showing the best performance over the widest momentum range is 3.0 cm of steel and 1.5 cm of scintillator, with efficiencies close to 100 % up to 6 GeV/c, staying above 99 % up to 10 GeV/c. These efficiencies remain good for the other scenarios, dropping to 97 % at high momenta. A small fraction of events are not successfully reconstructed for the MIND prototype. Track reconstruction using the Kalman filter from RecPack is based on a number of criteria, the most important being the number of hits along the track and the apparent curvature of the track (i.e. is the curvature well enough defined as to determine the momentum). The Kalman filter algorithm allows for the propagation of track parameters back through successive detector planes using a helix model that considers multiple scattering and energy loss. In a neutrino detector where a charged current interaction leads to hadronic activity at the interaction vertex in addition to an outgoing muon, the filter works well when the muon travels much further than particles related to the hadronic interactions. By ranking hits as a function of distance from the interaction vertex, it is possible to distinguish hits furthest away from this vertex as due only to a muon. Those hits then act as a seed for the Kalman filter. A typical reconstruction analysis will consider a number of planes (e.g. five) furthest downstream where hits are due to muons only. At high momenta, the technique works well. It has however limitations at high Q 2 or low neutrino energy, when the muon range is Grant Agreement PUBLIC 13 / 24

15 comparable to the range of the hadronic activity and identification of the muon becomes more challenging. When the Kalman filter fails, a cellular automaton method can be applied. It forms possible trajectories from a ranking of hits using a neighbourhood function. In a charged particle beam scenario with a reduced detector depth along the charged particle beam axis, the Kalman filter can fail to reconstruct a track at low momentum if the number of hits is insufficient, or at high momentum if the curvature of the track is insufficient. Analyses for the much larger Neutrino Factory MIND or nustorm SuperBIND with true neutrino interactions yield efficiencies which are much closer to 100%, with a few events << 1 % failing reconstruction due to significant scattering affecting track curvature. For these detectors, pattern recognition and event reconstruction are more complex, and merit some comparison with test beam data. Charge identification efficiencies are similar for all scenarios, close to 100 % for all momenta above 1 GeV/c, Figure 15. Detailed simulations were not carried out below 1 GeV/c where the thickness of steel and scintillator play a significant role because of multiple scattering. A hint of the fall in charge identification efficiencies can be seen in Figure 15 with a couple of data points below 1 GeV/c where the thinnest steel and plastic scintillator combination has marginally better efficiency. Figure 14: Reconstruction efficiencies for m+ for different steel and scintillator thickness combinations: from left to right: a) 3 cm steel, 1.5 cm scintillator; b) 2 cm steel, 1.5 cm scintillator; c) 3 cm steel, 3.5 cm scintillator; d) 2 cm steel, 3.5 cm scintillator. Figure 15: Reconstruction efficiencies for m+ for different steel and scintillator thickness combinations: from left to right: a) 3 cm steel, 1.5 cm scintillator; b) 2 cm steel, 1.5 cm scintillator; c) 3 cm steel, 3.5 cm scintillator; d) 2 cm steel, 3.5 cm scintillator. Grant Agreement PUBLIC 14 / 24

16 3.2. MAGNET REDESIGN The traditional approach for a MIND magnet design is to have large transmission lines carrying the magnetizing current in loops around the whole stack of iron plates, Figure 16. Figure 16: Left) Traditional approach to MIND magnet design, with one or more transmission lines around all iron plates. Right) New design proposed for the AIDA MIND where every individual iron plate has its own coil windings, leading to increased modularity and flexibility in choosing optimal separation gap between plates. Given the requirement to improve charge identification efficiencies for muons with momenta below 1 GeV/c, we reviewed the magnet design to address the major challenge: multiple scattering in the iron. As was outlined in the redesign document we propose [1]: a) Larger gaps between the first 3 iron plates: We have adopted a design where the gap between the first and second iron plates is 30 cm, i.e. much larger than typical. The gap between the second and third iron plates is 10 cm, also larger than the typical 3-5 cm. With these larger gaps between the first 3 iron plates, the transverse displacement on a given detector plane is greater, leading to increased angular resolution and therefore better charge ID efficiencies, Figure 16. b) Modular approach to the magnetization scheme: A magnet design where every individual iron plate has its own coil windings was studied in collaboration with a team at CERN with extensive experience in magnets, Figure 17. Because the power supply requirements are relatively modest, air-cooling of the coil is sufficient. The absence of a water cooling channel leads to much smaller coil cross-sections, thicknesses of 1-5 mm, so multiple scattering due to the coil material is negligible. The slit in the iron plate for the winding of the two coils is relatively narrow, of order 10 mm, with no significant loss of acceptance for the detector. Figure 17: Conductor coil winding on each individual iron plate of the MIND (left). Zoom on the slit engineered into the iron plate (right). Grant Agreement PUBLIC 15 / 24

17 Simulations of the magnetic field in the iron demonstrate the excellent attainable field maps, Figure 18. Figure 18: Contour map of magnetic field lines in units of T for one magnetized iron plate, with a total current of 2.5 ka, and a slit width of 2 m. We derived an expression for the power required, in order to identify the most critical parameters. For a given iron/steel grade, conductor material (aluminium) and solenoid length, it is the coil thickness that drives the power dissipation, with decreased resistance in the coil for an increase in thickness as is seen in the expression below: P = I 2 R coil = ( H material ) 2 r where: t coil l coil " ( l solenoid ) P is the power dissipated by the coil through ohmic losses, I is the current flowing through the coil, Rcoil is the resistance of the coil, Hmaterial is the material-dependent magnetizing force in Amp.turns/m, is the resistivity of the coil material, tcoil is the thickness of the coil, l " coil is the current path length for one turn of the coil, lsolenoid is the length of the solenoid, i.e. the width of iron plate over which the coil is wound. First estimations show that a coil thickness of between 1 and 4 mm is required. This range is driven: at the lower end by the requirement to keep power dissipation low, and at the higher end by the requirement to minimize multiple scattering in the conductor coil. The dependence of power on aluminium coil thickness is given in Figure 19. For a coil thickness of 1, 2, 4 mm, the power required to magnetize 32 Armco iron plates to 1.5 T is 4.5 kw, 2.2 kw and 1.1 kw respectively. Grant Agreement PUBLIC 16 / 24

18 Figure 19: Power dissipated for magnetization of one iron plate (module), and 32 iron plates, as a function of aluminium coil thickness, for two different steels: Armco and low carbon S1010 steel ELECTRONICS A new readout scheme was proposed for the MIND based on the EASIROC/CITIROC 1 family of chips, a clear shift away from using readout electronics based on the Trip-t readout chip as was initially planned. The layout for each AIDA front-end board includes three EASIROC/CITIROC chips, with a total of 96 channels possible, Figure 20. The two types of chips are interchangeable. The triggering and timing scheme of the FPGA on the board is shown in Figure 21. Figure 20: AIDA Front-end board layout with the three readout chips (EASIROC/CITIROC) mounted along the lower edge, close to the SiPM co-axial inputs. 1 These chips are based on a larger family of chips designed and developed in the AIDA WP9 for the read-out of calorimeters. Grant Agreement PUBLIC 17 / 24

19 Figure 21: Triggering and timing scheme for the FPGA of the AIDA Front-end board. Extensive testing of both the EASIROC and CITIROC chips with prototype readout boards was carried out in order to assess functionality and compatibility with photo-detectors, Figure 22. Figure 22: EASIROC/CITIROC test setup with prototype readout board. A typical charge spectrum obtained with this setup for a Hamamatsu silicon photomultiplier is shown, top right. Grant Agreement PUBLIC 18 / 24

20 3.4. COMPONENT PERFORMANCE The R&D work for the MIND was in part defined from experience with the TASD, notably affecting the detector module and electronics design, and integrating thoughts about detector construction, assembly and quality assurance in the design process in order to ensure better efficiency for processes involved. A detailed account of the work carried out is reported in [5]. Although the basic technology for detector modules is similar to the TASD there are some very significant differences: plastic scintillator bar manufacturing process and geometry readout based on silicon photomultipliers (SiPM) simplified optical readout scheme including custom optical connector electronics readout scheme based on the EASIROC/CITIROC family of asics. The plastic scintillator bars are supplied by the Institute for Nuclear Research (INR) of the Russian Academy of Sciences. The nominal parameters for the geometry are bars of 90 cm long, 0.7 cm in height and 1.0 cm in width, examples are shown in Figure 23. These extruded scintillator slabs are polystyrene-based with 1.5 % of paraterphenyl (PTP) and 0.01 % of POPOP, similar to the plastics used for the T2K SMRD detector counters. The surface is etched with a chemical agent (Uniplast) to create a m layer acting as a diffusive reflector. A small batch of prototypes was manufactured by Uniplast based in Vladimir (Russia) consisting of three different sizes ( mm 3, mm 3, mm 3 ) with 2 mm-deep grooves of different widths (1.1 mm, 1.3 mm or 1.7 mm) to embed optical fibres of different diameters. Figure 23: AIDA MIND plastic scintillator bars with wavelength shifting fiber and optical connectors at both ends. Tests were carried out at INR to determine basic light yield and timing properties. A wavelength shifting fiber (WLS) from Kuraray (200 ppm, S-type) of d = 1.0 mm was embedded into the 1.1 mm wide groove with a silicon grease (TSF451-50M) to improve optical contact between the scintillator groove surface and the fiber. Hamamatsu MPPC photosensors ( mm^2, 667 pixels, m 2, gain = C) were connected to both ends of the ~1 m long WLS fibers. A cosmic telescope was set up with two trigger counters. Measurements were made at the center of the scintillator slabs. The temperature during testing was C. Results are summarised in Table 1. Typical response to a minimum ionising particle is shown in Figure 24. Results show good light yield for all bar thicknesses, the highest light yield was obtained with the narrowest 10 mm width with 83 p.e. Comparisons with/without chemical reflector show an increase of light yield of a factor 2.5 when the chemical reflector is present. Grant Agreement PUBLIC 19 / 24

21 The effect of the silicon grease is close to 60 %. For the final assembly, the silicon grease is replaced by glue, which has roughly the same effect. An additional Tyvek reflector provides a 20 % increase in light yield, though this reflector is not implemented in the detector modules. Timing properties were studied for the two-sided readout, combining both ends with the result: (t1 t2)/2) = 0.5 ns, Figure 24. The timing is mostly determined by the fiber decay constant, fiber ~ 12 ns. Table 1: Light yield from plastic scintillator slabs of different dimensions, with different reflector configurations. Bar width [mm] Bar with no chemical reflector Light yields [photo-electrons (p.e.)] MPPC 1 MPPC 2 Sum Tyvek reflector Bar with chemical reflector (1) w/o grease (1) (1) + Tyvek reflector (2) Figure 24: Typical response of the plastic scintillator bar and silicon photomultiplier to a minimum ionizing particle: left) light yield, right) timing properties of the AIDA MIND bars. A total of 8400 plastic scintillator bars were manufactured for the AIDA MIND prototype. They were all tested for light yield, a fraction of those light yield results is reported in Figure 25. The light yield was measured to be ~56 p.e./mip on average from each photosensor, in the case of the previous generation Hamamatsu MPPC mm 2. Results of the latest generation process from Hamamatsu show a similar average of 58.5 p.e./mip despite the smaller sensitive area of mm 2 and smaller fill factor due to the 25 m cell size. Grant Agreement PUBLIC 20 / 24

22 Figure 25: Light yield from both ends of 3849 AIDA MIND plastic scintillators. The choice of photo-sensor is detailed in [5, 6]. Table 2 summarises the comparative test results for different manufacturers. These tests were carried out relatively early on in the project, and these manufacturers have all since developed new processes. We decided to choose Hamamatsu on the basis of those results, and on the basis of the extensive experience with MPPCs at the T2K experiment where 50'000 MPPCs equip near detector modules. Since cost scales roughly with surface area, we decided to purchase the cheaper mm 2 devices. These are new generation low after pulse devices, but not the low cross talk (LCT4) variant. Table 2: Comparative tests carried out on silicon photomultipliers from 4 different manufacturers under conditions representative of the AIDA MIND detector modules. Parameter Unit MPPC-T2K ASD-40 KETEK SensL Manufacturer reported specifications Pixel size m Number of pixels Sensitive area mm dia Gain Dark rate MHz 1 ~ Bias voltage V ~ Performance Overvoltage V ~ Dark rate khz Crosstalk % Pulse shape good good long tails Good Peak separation good good bad Bad PDE % The chemical reflector coating on the plastic scintillator is particularly susceptible to the addition of different materials such as structural glues. Tests were carried out with different glues, confirming a drop of light yield of order 35 %. The use of structural glues was limited to the strict minimum during assembly of the module shown in Figure 26. Double-sided adhesive tape was used instead where possible. Grant Agreement PUBLIC 21 / 24

23 Figure 26: Assembly of the AIDA MIND detector modules. Clockwise from top left: positioning of a scintillator bar onto the aluminium support frame with guide pins to ensure accurate pitch, placing the carbon fiber envelope, the complete module, installation of the heavy top plate of the assembly jig which sets the final height of the module by clamping the module for 24hr whilst the structural epoxy glue holding the aluminium support frame and carbon fiber envelope sets. A study of optical crosstalk between adjacent bars was undertaken to assess its significance. Some results are shown in Figure 27. The crosstalk was measured to be below 5 %. With > 50 photo-electrons in the main channel, this does not represent a significant issue. It can be addressed by increasing the discriminator threshold on the readout chip, or in analysis software. Figure 27: Crosstalk studies for the AIDA MIND detector modules: Left) Test setup schematic showing the position of trigger, veto, crosstalk and main channel bars; Middle) Typical charge spectrum in the crosstalk channel for cosmic rays; Right) Typical charge spectrum in the crosstalk channel with an LED source. Grant Agreement PUBLIC 22 / 24

24 4. FUTURE PLANS The TASD detector is due to continue fulfilling its role measuring the range of electrons and muons on the MICE beam line at RAL in the UK until It has already undergone some light upgrades ahead of planned beam tests in 2015, notably with replacements of the single anode PMTs that read out the total charge in each plane. Analysis of the rejection rate for electrons shows the muon sample purity reaches an excellent level of 99.85% with one variable. Further studies with multivariate analysis (MVA) are expected to lead to even greater purity and are the subject of a Ph.D. thesis to be submitted by By redesigning the MIND prototype, specifically targeting better muon charge identification at much lower momenta, 0.3 < p < 1.0 GeV/c, it is now possible to combine the MIND with other detector types located upstream of particle beams. Such plans are being drawn for the WA105 Liquid Argon prototype to be assembled at CERN by 2017, and for the WAGASCI experiment that will be operational at J-PARC end Modularity in the design of the MIND prototype could potentially lead to a host of applications. Magnetization schemes will be further explored under the newly approved EC-funded AIDA-2020 project. The MIND prototype is integrated in the CERN Neutrino Platform approved in 2014, which foresees some space and resources allocated to neutrino detector prototypes in a dedicated zone in the North Area building EHN1. 5. CONCLUSION The scope of the work carried out under the AIDA project within WP8.5.2 was defined from motivations to test neutrino detector prototypes based on plastic scintillator modules. A Totally Active Scintillator Detector was designed, constructed with stringent quality assurance principles and operated at the MICE beam line in the UK where it measured its first charged particle beams in October With an effective rejection of the 11.7% electron contamination in the beam and a muon sample purity of 99.85%, it performs well above the requirement of 99% purity. No dead channels were observed, and only 2 out of 5664 channels were mismatched mechanically, an achievement when taking into consideration the complexity of the fiber bundles and testament to the dedication of the technical teams that were relatively inexperienced in the handling of optical fibers at the beginning of the project. With the announcement of the Daya Bay results of March 2012 signalling a shift in prospects for neutrino facilities worldwide, the initial plans to build a standard MIND prototype were reformulated within AIDA to design a MIND in line with requirements for future experiments. The result is a redesign of the magnetization of the iron plates that enables better muon charge identification efficiencies at lower momenta. Detector modules were designed specifically for the AIDA MIND, with custom optical connectors and electronics based on the EASIROC/CITIROC family of chips. Extensive testing of the electronics with prototype evaluation boards, the 8400 plastic scintillator bars, and several types of silicon photomultipliers validates the design. Grant Agreement PUBLIC 23 / 24

25 6. REFERENCES [1] Noah, E. et al. (2015) AIDA MIND redesign status, AIDA Collaboration Report, AIDA- REP [2] Noah, E. et al. (2014) Installation of TASD and MIND, AIDA Milestone Report, AIDA- MS33. [3] Asfandiyarov, R. (2014) Totally Active Scintillator Tracker-Calorimeter for the Muon Ionization Cooling Experiment, Ph.D. Thesis N 4701, University of Geneva, CERN- THESIS [4] Bayes, R. et al. (2014) Light sterile neutrino sensitivity at the nustorm facility, Physical Review D 89, (R). [5] Noah, E. et al. (2014) Proposal for SPS beam time for the baby MIND and TASD neutrino detector prototypes, arxiv: v1. [6] Bron, S. et al. (2014) EASIROC and CITIROC chip studies for neutrino detector prototypes, AIDA Collaboration Note, AIDA-NOTE ANNEX: GLOSSARY Acronym AIDA EMR FPGA MIND MPPC MVA PMT SiPM TASD Definition Advanced European Infrastructures for Detectors at Accelerators Electron Muon Ranger Field-programmable gate array Magnetized Iron Neutrino Detector Multi-pixel photon counter Multivariate analysis Photo-multiplier tube Silicon photo-multiplier Totally Active Scintillator Detector Grant Agreement PUBLIC 24 / 24