Report on the 3D-projection Scintillator Tracker (3DST) as part of - - PDF document

Report on the 3D-projection Scintillator Tracker (3DST) as part

• f the DUNE Near Detector Concept Study
• S. Kettell and E. Worcester

Brookhaven National Laboratory

• K. Siyeon and C.H. Jang

Chung Ang University

• S. Bordoni, A. De Roeck, F. Pietropaolo, M. Nessi, and D. Sgalaberna

CERN, European Organization for Nuclear Research

• Y. Kudenko

INR, Institute for Nuclear Research

• J. Maneira

University of Lisbon

• T. Kutter

Lousiana State University

• R. Gran

University of Minnesota, Duluth

• C. Mauger

University Pennsylvania

• H. Su, D. Naples, and V. Paolone

University of Pittsburgh

• T. Cai, R. Flight, S. Manly, K. McFarland, and A. Olivier

University of Rochester C.K. Jung, C. McGrew, J. Palomino, K. Wood, and G. Yang Stony Brook University 1

• M. Kordosky and E. Valencia

College of William and Mary

(Dated: March 14, 2018)

Abstract

This note summarizes ongoing efforts to explore the case for including a 3D-projection scintillator tracker (3DST) as part of the DUNE near detector. The 3DST is a fully active detector with 3 dimensional, fast readout. It is dense enough to provide a large statistics sample with reasonable containment of hadrons and photons from neutrino interactions, and sensitivity to neutrons. The high statistics and granularity of the 3DST will allow the study/use of many different interaction morphologies and differential studies. The physics mission of the 3DST depends to some extent on the other components of the near detector and how they are configured, as well as whether or not elements of the detector are moved off-axis during running. In the DUNE era, data from the 3DST will be easily comparable to the large catalog of past neutrino results taken on plastic scintillator, and (potentially) with data recorded by the functionally identical SuperFGD in the lower energy, narrow band T2K beam, which is peaked at (∼0.6 GeV) near the DUNE 2nd oscillation maximum. The comparisons will allow us to fine-tune the neutrino interaction models which will be critical in reducing systematic uncertainties. These characteristics combine to form an addition to a DUNE near detector complex that is relatively insensitive to backgrounds and capable of providing useful handles for understanding the beam, near detector environment, and constraints for the analysis in the DUNE far detector.

2

CONTENTS

• I. Introduction

4

• II. Statistics

7

• III. Detector simulation

7

• IV. Tracking efficiency

9

• V. Answers to questions posed by concept study group

14

• A. What is the angular and energy resolution of the 3DST

14

• 1. Angular resolution

14

• 2. muon energy resolution

17

• 3. electron energy resolution

18

• B. How well can the 3DST do neutrino-electron scattering?

18

• C. How large does the 3DST need to be to do reasonably well on pizero and

neutron topologies? 21

• 1. pizero photon containment

21

• 2. neutron containment

22

• D. Can the 3DST do something with neutron counting/angles?

25

• 1. Neutrons in scintillator and the 3DST

28

• E. Does the 3DST need to be in a magnetic field?

31

• 1. Charge separation efficiency

32

• 2. wrong-sign backgrounds

32

• 3. electron neutrino constraints on hadroproduction

33

• F. What is the complementary physics for the 3DST relative to the other

trackers? How would it improve CP sensitivity? 33

• 1. Synergy with the Liquid argon TPC

36

• 2. Synergy with the straw tube tracker

38

• 3. Synergy with the high pressure gaseous argon TPC

38

• G. Selected physics processes

39

• 1. neutrino-electron scattering

39

• 2. coherent charged and neutral pion production

39 3

• 3. low-ν

40

• 4. inclusive CC

42

• 5. NC/CC neutral pion production

42

• VI. Potential U.S.-Japan Cooperation funding

43

• VII. Other geometries

43

• A. 3DST inside HPGArTPC

43

• VIII. R&D

45

• IX. References

50 Acknowledgments 50 References 50

I. INTRODUCTION

The 3DST detector will be made of many optically independent cubes of extruded scin-

• tillator. The design is very similar to that being proposed by the T2K collaboration as

part of the T2K ND280 detector upgrade [12, 14]. The cubes in this detector are read out along three orthogonal directions by wavelength shifting (WLS) fibers. A schematic of the detector is shown in Fig. 1.

Scintillator cube WLS fibers

• FIG. 1: Schematic of the 3DST/Super-FGD structure. This figure is from reference [14].

4

The 3DST is proposed as a component of the DUNE near detector for a number of

• reasons. The complementarity of the 3DST to the rest of the detector is contingent to some

extent on the other components of the near detector and the location and size of the 3DST. In general, the 3DST provides:

• a large statistics sample of neutrino interactions with true 4π coverage in a magnetic

field;

• a fine-grained spatial resolution sufficient to identify and measure most neutrino in-

teraction morphologies;

• a transparent connection to the vast catalog of cross section results coming out of

plastic scintillator experiments, i.e., MINERvA and T2K;

• a detector that is functionally identical to the T2K Super-FGD which, by the time

DUNE sees a neutrino beam, will have taken much data in a narrow-band, lower energy beam that happens to peak in the energy region of the DUNE second oscillation maximum;

• a fast detector, with the insensitivity to backgrounds that implies;
• a detector with substantial sensitivity to neutrons in the final state of neutrino inter-

actions;

• a relatively high density detector that will convert a substantial fraction of photons

from π0 decays. Achieving a full 4π acceptance at the near detector is fundamental to avoid any bias in the extrapolation of the cross-section model to the far detector. We know the neutrino cross section is poorly understood and it is not totally clear this issue will be solved by when the DUNE experiment will start collecting data. There are many reasons why we should be careful about a detector without a full solid angle acceptance, in particular considering the fact that the LAr near detector would not be magnetized. The wrong-sign background may largely affect the precision on δCP and we know that the difference between neutrino and antineutrino cross sections is poorly modeled, resulting in large systematic uncertainties. 5

A clear example is given by the so-called “2p2h” process, where the (anti)neutrino inter- acts with a neutron (proton) but, because of short range correlations, two nucleons (e.g. a proton and a neutron) or more can be ejected instead of one, providing a bias in the neutrino energy reconstruction. This process occurs for neutrino energies of about 1 GeV/c and has a strong effect on the neutrino energy reconstruction, in particular for the DUNE neutrino flux. We do not expect this process to be the same for neutrinos and antineutrinos and this difference is correlated with the direction of the outgoing lepton angle and the neutrino

• energy. Another issue is given by the fact that the theoretical models (for example the

so-called “Martini” and “Valencia” models) show larger discrepancies for muons produced at high angles than along the neutrino direction [2]. The constraint on the cross-section model systematic uncertainties would certainly profit by the detection of all the pions, protons and, possibly, neutrons produced at any angle, also to provide a precise measurement of the transverse missing momentum. Recently the GiBUU theoretical group presented an updated model that expects the 2p2h processes to mainly populate that part of the phase space where a muon ejected with angle large with respect to the neutrino direction. Consequently a large effect in the forward region would be due to final state interactions (FSI) [4]. This is in contrast with other models (e.g. “Martini” and “Valencia”) that expect to find 2p2h neutrino events mainly in the forward region. The neutrino physics community is aiming to a unified neutrino-interaction model that could describe all the data from all the experiments at different energies and nuclei. This implies that the model must suit well the neutrino-carbon (oxygen), neutrino-argon and neutrino-hydrogen interaction data. However, these nuclei are very different (C12, Ar40) and the solution may not be trivial. For example the FSI and secondary interactions (SI) are expected to rapidly change with the size of the nucleus and, consequently, to be larger in argon than carbon. The possibility to measure neutrino interactions not only in argon but also in plastic, would allow us to cross check our data with many other experiments both in the energy range of the first (Minerva) and second (T2K) oscillation probability maximum. Furthermore it would help to solve the degeneracy between the effects due to the neutrino- nucleus interaction and FSI and SI. The situation is still quite unclear and we may find issues in the extrapolation from carbon- to argon- based models. A high-precision 4π plastic 6

scintillator detector could protect us by future unexpected problems in the understanding

• f the neutrino cross-section models. All these issues tell us about the importance of a fast

near detector capable to measure low momentum threshold particles in the full solid angle.

II. STATISTICS

A 3DST detector could be configured in different ways as part of the DUNE near detector. The arguments for placing the 3DST inside the magnetic volume are discussed in Section V E. The possibility of placing it inside a high pressure gaseous Ar TPC as a target is discussed in Section VII A. Size options and corresponding statistical implications for this latter configuration are discussed in that section. Unless stated otherwise, the default configuration of the 3DST used for studies presented here is inside the magnetic volume upstream of a low density tracker. An ECAL is placed immediately downstream of the 3DST for studies that require it. The default size of the 3DST used in these studies is 2.4 m x 2.4 m in transverse dimensions by 2.0 m in depth along the neutrino beam direction. This gives a total target mass for the 3DST of 12.2 t. Implementing a conservative veto region around each side of the detector

• f 25 cm, gives a fiducial mass of 5.7 t. The 25 cm fiducial cut is similar to that used by the

MINERvA collaboration for many analyses done with a scintillator detector in the NuMI low energy beam, which is similar in flux energy profile to the DUNE beam. Table I gives the number of events expected per year in the fiducial volume of a 3DST of the size described above. The numbers given are assuming the 80 GeV, 3 horn, optimized LBNF beam flux and 1.46×1021 POT/year.

III. DETECTOR SIMULATION

The chain of the detector simulation is as:

• Use DUNENDGGD [5] package to generate whatever geometry we desire.
• Shoot particle guns OR Use GENIE event generator to generate interactions in the

geometry with DUNE flux. 7

Channel FHC RHC νµ CC inclusive 8.4×106 3.1×106 CCQE 1.8×106 1.0×106 CC π◦ inclusive 2.4×106 0.6×106 NC total 3.0×106 1.3×106 νµ-e− scattering 960 660 νe CC inclusive 1.6×105 0.35×105

TABLE I: This table summarizes the projected event rates per year for a 2.4 x 2.4 x 2.0 m3 3DST detector, assuming a 25 cm veto region at each side and the 80 GeV, three horn, optimized LBNF beam.

• Use edep-sim [9] to obtain energy deposits in the detector
• Go through the 3DST hit simulation, which accounts for the cube geometry effect and

convert energy deposit to number of photo-electrons (PE) with fiber attenuation.

• Do analyses.

The steps are briefly described here. DUNENDGGD gives, for example, a 3DST + ECAL geometry, which is shown in fig. 2. The 3DST consists of 1 cubic centimeter small cubes and the size is 2.4m x 2.4m x 2m and the ECAL is made of lead sandwiched by scintillator module planes. Each scintillator module plane consists of a large number of scintillator bars along X direction or along Y direction alternatively. For details of the ECAL geometry, refer to the CDR design [6]. Note that for most of following studies, we only care about the physics inside 3DST, therefore, ECAL information is not always included in the simulation processes. The geometry is input into GENIE and we run edep-sim to get the energy deposits. An example muon neutrino CC multi-pi production event is shown in fig. 3. Then the energy deposit information is input into the hit simulation in order to obtain the PE information in cube geometry. An example muon neutrino CC multi-pion production event after the hit simulation only in 3DST is shown in fig. 4. 8

• FIG. 2: Geometry of a 3DST plus an ECAL.

IV. TRACKING EFFICIENCY

3DST has a isotropic geometry so we expect it to have a 4π coverage of the charged

• particles. A study has been performed to confirm this. We define the reconstruct-able track

as:

• In one of the three 2D projections, if 3 cubes are fired, we call that 2D projection is
• good. We need at least two good 2D projections.
• The track can be separated from the other longest track.

9

• FIG. 3: Energy deposit in 3DST + ECAL for an example CC event.
• FIG. 4: An example event in 3DST with hit simulation.

We only care about the contained muon tracks. For the tracks that go outside, we assume they can be perfectly reconstructed with low-mass tracker. For contained ones with above definition, the tracking efficiency as function of particle angle and be seen in the left panel

• f fig. 5. In the right panel, the efficiency as function of muon energy is shown. We do see

flat distribution for muon angle and the threshold for muon energy is very low. Since we really care about the contained events, We would like to look into the event containment in 3DST. Events are classified to three categories: contained, end-escaping and side-escaping. Figs. 6- 8 show the particle energy-angle distributions of muons, pions and protons for contained events, end-escaping events and side-escaping events, respectively.

• Figs. 9- 11 show the same things but in the neutrino energy-particle angle space. The end-

10

• FIG. 5: Tracking efficiency as functions of muon angle and muon energy.
• FIG. 6: Distribution in deposited energy vs. particle angle for contained events.

escaping events are likely to be measured in the low-mass tracker and the side-escaping events are likely to be measured in the range side detectors or gaseous argon TPC, as proposed. In addition, we look into the tracking efficiencies for the three cases. Fig. 12- 14 show

• FIG. 7: Distribution in particle energy vs. particle angle for end-escaping events.

11

• FIG. 8: Distribution in particle energy vs. particle angle for side-escaping events.
• FIG. 9: Distribution in neutrino energy vs. particle angle for contained events.

the tracking efficiency as functions of energy and angle for each particle type. We care more about the efficiency in neutrino energy since this is applied to obtain the wrong-sign components. For all cases, the tracking efficiencies are very close to 1. So the performance inside the 3DST is good and we expect even better tracking performance in the low-mass tracker.

• FIG. 10: Distribution in neutrino energy vs. particle angle for end-escaping events.

12

• FIG. 11: Distribution in neutrino energy vs. particle angle for side-escaping events.
• FIG. 12: Tracking efficiency for the contained events.
• FIG. 13: Tracking efficiency for the end-escaping events.

13

• FIG. 14: Tracking efficiency for the side-escaping events.
• FIG. 15: A 4 x 4 x 4 cube3 zoom of the staggered geometry used in this study. Cube
• utlines of two successive planes for each projection demonstrate the increase of effective

granularity in this particular staggering scheme.

V. ANSWERS TO QUESTIONS POSED BY CONCEPT STUDY GROUP A. What is the angular and energy resolution of the 3DST 1. Angular resolution

A 240 × 240 × 200 array of 1 × 1 × 1 cm3 scintillator cubes was encoded into a gdml file via the software package Dune-ND-GGD [5]. The default configuration corresponds to all cubes being stacked in such a way that all cubes in a given row are in line with one

• another. In an effort to improve the angular resolution, a geometry in which alternating

rows (checkerboard pattern) in the zy plane are staggered by half a cube length, 5 mm, in x was also generated. See Figure 15. Monoenergetic particles were simulated to traverse 14

this geometry with angles uniformly distributed within a cone extending 30◦ around the z-axis with edep-sim [9], a Geant4-based energy deposition simulation package. Several samples of 10K muons and electrons with energies ranging from 0.5 to 3.0 GeV were utilized in this study. These samples were put through a rather crude detector response simulation, called 3DSTSim [10], adapted from T2K’s ND280 upgrade software. The electronics were not simulated. The electron angular resolution is of particular interest for measuring the ν-e elastic scat- tering rate, which has been shown to be a powerful constraint on neutrino flux uncertainties. We find a similar resolution as MINERνA’s scintillator detector, which has demonstrated success in this channel. [11] While the information made available by the design of the 3DST consists of three 2D projections (xy, zy, and zx) of the activity in the detector with sub-nanosecond timing resolution, the reconstruction of the particle’s angle only uses one list for each dimension (x, y, and z) of the photoelectron (PE) yield in 1 cm bins. No timing information is used. We recognize the potential of a more sophisticated fitting algorithm that utilizes all of the information a 3DST detector would offer. However, due to time constraints we were not able to study an optimization of the reconstruction deeply. The 3D fitter used in this study finds the average position weighted by PE yield for each dimension and constructs a covariance matrix of PE-weighted hit positions. The eigenvector

• f this matrix with the largest eigenvalue gives the reconstructed direction of the track, ˆ

r. For each track, the angle between the reconstructed and true directions is computed: θreco−true = cos−1 ˆ r · p0 |p0|

• ∈ [0, π],

(1) where p0 is the true initial 3-momentum of the particle saved by edep-sim. θreco−true is computed for all 10K tracks in a given sample. The angular resolution is defined as the upper bound to which one needs to integrate from θreco−true = 0 rad. in order to capture 68% of the resulting distribution. See Figure 16. This procedure was applied to each sample several times using various track lengths to find the optimal resolution. The resolution suffers when the track length is too short due to a geometrical effect and when the track length is too long due to multiple scattering. Somewhere inbetween exists the optimal fit length that minimizes the angular resolution. Furthermore, this optimal length is expected to be energy dependent as is the impact of multiple scattering. 15

• FIG. 16: A sample θreco−true distribution with the upper bound that contains 68% of the

entries defining the angular resolution.

Fit Length [mm] 100 200 300 400 500 600 Resolution [rad] 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

500 MeV 1500 MeV 3000 MeV

Electron Angular Resolution

(a)

Fit Length [mm] 100 200 300 400 500 600 Resolution [rad] 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

500 MeV 1500 MeV 3000 MeV

Muon Angular Resolution

(b)

• FIG. 17: Angular resolution as a function of the length of track used in the 3D fit for 0.5,

1.5 and 3.0 GeV (a) electrons and (b) muons. The angular resolution as a function of fit length for electrons and muons at various energies is shown in Figure 17 (a) and (b) respectively. These plots correspond to the default configuration (i.e. no cube staggering). We find an optimal 3D angular resolution

• f ∼27 mrad for electrons and ∼12 mrad for muons.

To study the effect the described staggering scheme has on the resolution, we compare 16

Fit Length [mm] 100 200 300 400 500 600 Resolution [rad] 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

1.5 GeV Muons Without Staggering 1.5 GeV Muons With Staggering 1.5 GeV Electrons Without Staggering 1.5 GeV Electrons With Staggering

3DST Angular Resolution

• FIG. 18: Angular resolution as a function of the length of track used for the 3D fit for 1.5

GeV muons and electrons in both a staggered and non-staggered 3DST geometry. resolution curves for 1.5 GeV electrons and muons in both the default and staggered geome- tries in Figure 18. We note an improvement of the muon angular resolution in the staggered geometry, but no improvement for electrons.

2. muon energy resolution

We care mostly about the contained events. So we can only look into the muons with energy less than 1 GeV givin the fact that higher energy muons would escape the scintillator tracker regardless of moving directions. We shoot particle guns with a 30 degree direction to Z in various kinimatic energies and look into the energy deposit resolution at each energy. All decay electron contributions have been removed, which means we assume that the decay electrons can be separated perfectly from the parent muons. Figure 19 shows the muon energy resolution as a function of kinetic

• energy. The energy resolution can reach a few percent. This resolution is actually only
• btained with the dedx information. We are trying to fit the magnetic field bending curve,

in which way better resolutions for muons with a wide range of energy are expected. 17

• FIG. 19: Muon energy resolution as a function of kinetic energy.

3. electron energy resolution

Similar to the muon energy resolution. We shoot particl gun with Z direction in various kinimatic energies and look into the energy deposit resolution at each energy. The deposited energy is roughly the electron energy if the electron is below 600 MeV in a 0.4T magnetic

• field. Fig. 20 shows the electron energy resolution as a function of electron kinetic energy.

Again, the energy resolution is juat few percent for electrons. So far, this result only takes the 3DST contained energy deposits into account. A better way is to include the ECAL. The result of electron energy resolution with a 3DST+ECAL model is on-going.

B. How well can the 3DST do neutrino-electron scattering?

Neutrino-electron scattering (NES) cross section is well measured, thus this channel pos- sibly can provide us good flux constraint. NES tends to give a forward-going electron so measuring the electron angle is the most important thing in this study. As shown in previous sections, 3DST can provide a comparable electron angular resolution to Minerva [11]. So similar to Minerva, we expect to be able to use this channel to constrain the neutrino flux. 18

• FIG. 20: Muon energy resolution as a function of kinetic energy.
• Ref. [7] shows a detailed study for potential ND designs. Some of those materials can be

directly borrowed here to examine our situation. For a simulated liquid argon detector, 3 mm position resolution with 1 mm sampling step size has been used to be fitted event-by-event. The angular resolution this way gives can be interpreted as that two Gaussian functions are used to determine the angular resolution of electron. One is from the geometry effect (position and sampling resolutions) and the other one is from multiple scattering (MCS), which has a long tail.

• Fig. 21 shows the angular resolution as a function of electron energy for the case that

fitting with the peak only and the other case that fitting with the MCS tail. The electron traveling in the first 15 cm has been used for the fit. 3DST two dimensional projected resolution with 15 cm travel length is also shown in the plot. Note that we are comparing the liquid argon detector with 3DST here. We can conclude that 3DST’s angular resolution is between the two extreme cases of liquid argon. Then we can project this piece of information to obtain a flux constraint [23].

• Ref. [7] also shows a direct flux constraint with the NES channel.

They use fitting 19

• FIG. 21: Angular resolution as functions of electron energy. See text for details.

templates that in different neutrino energies and fit the combination of those templates with the systematics to obtain the sensitivities. More details can be checked in the talk. A conclusion plot is shown in fig. 22. Those plots show the post-fit and pre-fit ratios. That gives us a sense of potential flux constraints. The systematic uncertainties of beam divergence, beam related νe and γ backgrounds, as well the detector smearing from above study have been included. In the case of CH, which has similar material to 3DST (thus similar MCS strength), the assumption for angular resolution is 3 mm position resolution and 2 cm sampling step size [8]. These numbers are expected to be similar to 3DST given the 1 cubic cm cube size. The energy resolution is assumed to be flat 5%, but this does not really matter as similar results have been obtained with different energy resolution assumptions [8]. So we could expect to have a similar flux constraint to the CH case in the reference study. 20

• FIG. 22: Neutrino flux constraints for various detector configurations. See text for details

C. How large does the 3DST need to be to do reasonably well on pizero and neutron topologies? 1. pizero photon containment

To establish the size of a 3DST detector that can contain gammas decaying from neu- tral pions a special geometry was generated. This geometry includes the 3DST detector surrounded by an extra liquid argon detector large enough to be able to fully contain the

• gammas. The 3DST detector size is 2.4m x 2.4m x 2m and the external volume adds an

extra 20 radiation lengths based on previous studies done by the MINERvA collaboration. This detector was located inside the DUNE ND hall using the geometry packaged called DUNENDGGD[5]. The total exposure processed for this study was 1.46x1021 POT in the neutrino mode (80 GeV protons), which correspond to one year of exposure. The FERMI- LAB grid resources were used for this sample generation. Because the purpose of the study is to find the depth necessary for containment, all gammas from the neutral pion decays in the inclusive neutrino interaction sample. There was no requirement that both gammas from a pizero in an event pass a set of selection cuts. The event selection applied for this 21

study is described as follows:

• The distance from the neutrino interaction vertex to the nearest 3DST detector wall

must be larger than 25 cm.

• Every gamma must leave visible energy inside the 3DST detector.
• The energy deposit per gamma must be at least 95% inside the 3DST detector. ***?***

Figure 23 shows the distribution for the angle of each gamma with respect to the neutrino beam direction versus the traversed distance calculated from the point where the gamma starts showering. A significant fraction of of the gammas from pizeros make an angle with respect to the beam of less than 40 degrees. This is an important feature to consider for the detector design. In to order quantify the depth along the beam direction necessary to study neutral pions in the 3DST detector, the efficiency of gamma containment with respect to the traversed distance was calculated for events with an angle of less than 40 degrees. This is shown in Figure 24. For the ∼five radiation length depth in the design considered here, the efficiency for gammas from pizero decays having ≥95% energy containment is over 60%. If a downstream ECAL is included in the design, the containment can be easily improved. At larger angles, the 2.4 m transverse size is sufficient to contain the bulk of the gammas.

2. neutron containment

The MINERvA collaboration recently demonstrated the ability to tag neutrons with a high efficiency and fast timing in a plastic scintillator detector [16]. The work presented by MINERvA was optimized to see low energy depositions in a low recoil analysis. Recently MINERvA has also shown studies looking at neutron interactions with enough energy to penetrate multiple planes in the detector and allow for 3 dimensional reconstruction. Figure 25 shows results of a neutron particle gun study in a simulated MINERvA detector. These plots show the distance traveled by neutrons before interacting on hydrogen or carbon as a function of the neutron KE. These plots do not require the interaction leave visible

• energy. However, as shown Section V D, at low neutron kinetic energies the probability for

the neutron to interact with hydrogen in the detector is high and it often leaves observable energy (though typically not enough for the MINERvA 3 dimensional reconstruction). 22

2 4 6 8 10 12 14 16 18 20

rad len N

20 40 60 80 100 120 140 160 180

γ

and

ν

Angle btw.

1 2 3 4 5 6 7 8 9 10

3

10 ×

s with Containment > 0.95

γ

All -

• FIG. 23: Angle between gamma and the neutrino beam versus the traversed distance

calculated from the point where the gamma starts showering. Figure 26 shows the result of a neutron particle gun study where the distance from the neutron origin to the start of the 3 dimensional cluster of energy in MINERvA that contains the largest hit energy (aiming to capture the Bragg peak of a proton). These clusters necessarily have sufficient energy to spread over at least three layers in MINERvA in order to satisfy the conditions for 3 dimensional reconstruction. From the figures in this section, it can be seen that the distribution of neutron energy deposits seen by MINERvA (with high efficiency as will be shown in Section V D) is contained within a distance of ∼1 m. The actual distance in the 3DST should be smaller because it lacks space between layers that is part of the MINERvA design. This implies that a 3DST with dimensions of 2 m on a side is sufficient to contain neutron energy depositions that can provide multiplicity and angular information for a large fraction of events in the fiducial volume. 23

2 4 6 8 10 12 14 16 18 20

rad len N

0.2 0.4 0.6 0.8 1 1.2

Efficiency

s with Containment > 0.95

γ

• FIG. 24: Efficiency to contain at least 95% of the gamma energy, as a function of the

depth along the beam in terms of radiation lengths. For our current 3DST dimension the efficiency reach values higher than 60%.

• FIG. 25: Path length for neutrons before first interaction as a function of target nucleus

and neutron kinetic energy. From a study done with the MINERvA simulation as part of the DUNE near detector concept study. 24

• FIG. 26: Distance from neutrino interaction vertex to tagged energy deposition position

for neutrons depositing enough energy for 3 dimensional reconstruction in MINERvA.

D. Can the 3DST do something with neutron counting/angles?

The reconstruction of neutrons produced in neutrino interactions is problematic. Ener- getic inelastic interactions happen rarely and cannot be used reliably to tag the presence

• f neutrons. Such interactions, when they happen, do not necessarily allow for the recon-

struction of the neutron energy. Neutron capture can be done with high efficiency but the inherent time delay and loss of angular information during thermalization in the detector limit the usefulness of this technique in a high intensity environment like DUNE. The MINERvA collaboration recently demonstrated the ability to tag neutrons with a high efficiency and fast timing in a plastic scintillator detector [16]. The simulation used in the MINERvA results, based on GEANT, was found to describe the data fairly well for neutron candidate multiplicity, deposited energy, timing, and longitudinal position. Fig. 27 shows distributions of data-validated simulations from MINERvA that are relevant for illus- trating the potential of the 3DST due to the similarities of the two detectors. On the left is a plot of the deposited energy for neutron candidates in the MINERvA analysis. Note that low energy deposits dominate and that the deposited energy is relatively unchanged for a wide range of neutron kinetic energies. On the right is plotted the kinetic energy distribu- tion for neutrons in neutrino interactions in a low recoil sample in the MINERvA low energy beam along with the same distribution for neutrons that form candidates in the MINERvA

• analysis. The MINERvA analysis tags neutrons in these events with an efficiency that rises

25

• FIG. 27: Distributions for neutrons in MINERvA according to simulation. These plots are

approved by MINERvA as part of the analysis released in November 2017 [? ]. with neutron kinetic energy up to a level of 60% at a kinetic energy of 300 MeV. This is an exciting development. The fast tagging of neutrons in the DUNE near detector with a high efficiency is desirable.

• The CCQE process is the golden channel for use in oscillation experiments because the

neutrino energy can be reconstructed from the lepton information alone. One problem with this is the CCQE-like samples used for this reconstruction contain contributions from non-CCQE processes. The neutrino energy reconstructed under the CCQE as- sumption in events that are not true CCQE can have worse resolution and/or a bias. See for example, reference [17] where events arising from a 2p2h interaction are shown to give a large low energy tail when reconstructed as true-CCQE. Multinucleon inter- actions might be expected to produce more nucleons in the final state. In particular, for antineutrino interactions, the neutrino interaction of a correlated pn pair should produce 2 neutrons in the final state (disregarding final state interactions). Tagging neutrons on an event-by-event basis can provide an additional handle for identifying events where the neutrino energy reconstruction might be suspect due to the interac- tion not being true CCQE. By rejecting such events or by adjusting the reconstruction

• n tagged events, the neutrino energy resolution should be improved.
• The far detector flux constraint will be built from the observation and model tuning
• n many different event morphologies. The use of exclusive event classes containing

tagged neutrons will add information to the overall constraining model which may 26

improve the constraint at the far detector using the VALOR or a VALOR-like infras- tructure.

• Recent results from T2K and MINERvA have made use of transverse momentum

balance variables to study final state interaction effects in neutrino interactions [18][19]. These variables are useful only when strict reconstruction requirements are placed on the final state particles. In the results presented to date, information on neutrons when present in the final state was ignored except as an “unattached” energy veto. It may be that with the inclusion of angular information, events containing neutrons might be used as a signal sample category in a transverse variable analysis.

• The NOMAD collaboration demonstrated the usefulness of transverse momentum bal-

ance for separating NC from CC interactions [20]. At an early stage of the DUNE near detector concept study, C. Marshall showed that the expected performance of such a separation at DUNE energies is severely degraded in the limit of using a detector blind to neutrons [21]. Neutron tagging might be used as a way of flagging events where the transverse momentum separation technique is problematic, allowing for a cleaner separation of non-tagged events. The reasons for excitement at fast and efficient neutron tagging which are listed above suggest several potentially interesting studies. To what extent is the neutrino energy reso- lution improved for CCQE-like samples when events with tagged neutrons are used as an additional handle? To what extent is the far detector flux constraint improved with the in- clusion of exclusive final states containing neutrons in the near detector constraint? Can the transverse momentum variable resolution be improved through the use of neutron multiplic- ity and angle information? Can NC/CC separation be improved using neutron multiplicity and angular information? Given the time constraints of the near detector concept study, these studies were not done in time for this report. The emphasis for the work included here was on understanding better the MINERvA signal and demonstrating, to some extent, the expected 3DST performance with regard to neutrons. 27

• FIG. 28: Interaction probability distributions for neutrons in MINERvA on hydrogen and
• carbon. From a study done with the MINERvA simulation as part of the DUNE near

detector concept study.

1. Neutrons in scintillator and the 3DST

Further work has been done with the MINERvA low energy flux simulation as part of the DUNE near detector concept study. One part of the study aimed to determine the target nucleus for neutron interactions yielding observable candidates in MINERvA. The probability distributions for neutrons to interact on hydrogen and carbon are seen in Fig. 28. The probability to interact on hydrogen is dominant at low neutron kinetic energies. For higher kinetic energy neutrons the interaction is usually on carbon and the energy deposition comes from particles produced when the carbon nucleus is fragmented in the interaction. The probability for a neutron interaction to produce an observable energy deposit, defined as an energy deposition of more than 1 MeV, is shown in Figure 29. The path length traversed by the neutron before interacting is shown as a function of the kinetic energy of the neutron for interactions on hydrogen and carbon in Fig. 25. It should be noted that the path length in a 3DST detector will be less than MINERvA due to the air gaps between layers in MINERvA that would not be present in the 3DST. A study of neutron interactions in the 3DST was done with a simulation chain beginning with 5x105 neutrino events processed with edep-sim (described in Section III). Simulated hits are formed for a 3DST with 1 cm subdetectors by voxelizing the detector in regions near Geant energy deposits. Voxels are assigned an energy proportional to the distance each track depositing energy travels through them, and voxels are considered visible if they have 28

• FIG. 29: The probability of a neutron in MINERvA interacting on hydrogen and carbon

to deposit more than 1 MeV of energy in the scintillator. From a study done with the MINERvA simulation as part of the DUNE near detector concept study. at least 1.5 MeV of energy. The energy in each voxel is grouped by whether it was produced by a final state neutron. To require that neutron hits are visible, a voxel must have at least 3 times as much energy from neutron descendants as from all other particles to be considered a neutron hit. Neutron hits are thrown out if there are any visible voxels within a 5 cm cube that are not neutron hits. Neutron candidates are formed as clusters of neutron hits. All neutron hits that are in within a specified number of rows of cubes are formed into clusters. If a hit could be combined into multiple clusters, those clusters are merged into one cluster with the new hit. A cluster’s center is the energy-weighted centroid of the hits it contains. Hits and clusters are assigned to the final state particles that created the neutron energy deposits that contributed to them. This study is still work in progress. The number of candidate energy depositions are a function of the energy clustering algorithm. That algorithm has not been carefully studied

• r tuned to provide an optimal correlation between the actual final state neutron multiplicity

and the measured candidate multiplicity. This problem is exacerbated currently by the fact that the simulation lacks the saturation effect described by Birk’s law. This leads to an unrealistic excess of low energy deposits. This document will be modified as improvements in this simulation and analysis allow. Figure 30 shows the candidate deposited energy in the detector versus the neutron kinetic

• energy. Figure 31 gives the probability that a final state neutron creates an observable energy

29

• FIG. 30: Deposited energy in clustered hits identified as energy deposits associated with

final state neutron activity versus the neutron kinetic energy.

• FIG. 31: The probability that a final state neutron leaves an observable energy deposit in

the detector as a function of neutron kinetic energy. deposit in the detector as a function of the neutron kinetic energy. Finally, Figure 32 shows the opening angle between the vector from the interaction vertex and the neutron candidate position versus the true initial direction of the final state neutron responsible for the energy

• deposit. In instances where a single final state neutron produced multiple observable energy

deposits, the candidate closest to the neutrino interaction was selected for the angle plot. 30

• FIG. 32: Opening angle between the initial direction of a final state neutron and the

direction to the closest (to the neutrino interaction) observable candidate energy deposit.

E. Does the 3DST need to be in a magnetic field?

*** low p muon sign separation need and effect on cp (perhaps most at second max). don’t get sign sep from lar and don’t get stats (unless stt), asked Seb Jones for plot using cafana The capability of a detector to separate between positive and negative tracks is very important also to reject the NC background, in particular those events at lower energy that populate the second oscillation maximum energy region. Though the proton - muon PID is very good in plastic scintillators, there can still be some NC elastic events with high-momentum protons (almost MIP), misidentified as muons. NC elastic or NCπ0 (where the π0 is not detected) events can mimic a CCQE event: the neutrino scatters with the proton that is recoiled with enough large momentum to produce a relatively long track and misidentified as a muon. If running in neutrino-beam mode, a good charge identification capability of the detector can measure the positive charge of the proton and reject the ND elastic event (opposite charge sign if antineutrino-beam mode). At slightly higher energy, in a similar way, also NC1π±, not negligible at the DUNE neutrino energy, can produce background to the CC0π sample. Another possible source of background is from NCπ0 events that can affect the electron- neutrino samples. In fact a photon produced by a π0 → γγ decay can convert in the 3DST and sometimes only the positron or the electron can be detected. As shown in [14], a 3DST- like detector has good capability of identifying gamma conversions by detecting two tracks 31

with opposite charge sign produced at the same vertex, so consistent γ → γγ. It can also reject some of the γ → γγ events by precisely measuring the pulse shape, double when e+ and e− are in the same cube In addition about half of those background events that pass the selection could be still rejected by looking at the charge of the single track, for example a positron-like track with neutrino-running beam mode.

1. Charge separation efficiency

The charge separation for 3DST is expected to be good, especially for muons. A particle gun study has been done in order to have a sense of the charge separation efficiencies for muons and electrons. The algorithm we use is very simple:

• Use first 20 cubes to do a straight line fit
• Count all hit points in the full track above and below the fitted line
• compare the above and below numbers to get the sign

Electrons and muons with various energy particle guns have been used, including 300 MeV, 1, 2, 3 GeV. In order to obtain the efficeincy, we look into the fractions that we can identify electrons or muons as electrons or muons. Another important term is the purity. In order to obtain this quantitatively. We look into the fractions that we mis-identify the positrons

• r anti-muons as electrons or muons. Fig. 33 shows the efficiencies and purities for electron

and muon charge separation. We see a almost perfect charge separation for muons with the energy range we mostly care about. The electron efficiency goes to 80% and impurity goes to 20%. However, with more sophisticated algorithms, we expect to have better electron charge separation results.

2. wrong-sign backgrounds

We use a similar-to-DUNE beam, NuMI low-energy beam, to obtain the wrong sign fluxes for FHC(nu mode) and RHC (nubar mode). Fig. 36 shows the fluxes for muon neutrinos and anti-neutrinos in the nu and nubar modes. First column is the signals and the second column is the backgrounds. The wrong sign fraction is defined as 32

• FIG. 33: Charge separation efficiences and purities for electrons and muons.

ǫb = F(b) × containment(b) × ǫreco.(b) × ǫq(b) F(s) × containment(s) × ǫreco.(s) × ǫq(s), (2) where F, ǫreco. and ǫq are the flux, reconstruction efficiency and charge sign separation effi- ciency, while s and b are signals and backgrounds. The last factor has been obtained from

• above. An average number over the whole energy range has been used since the charge

separation efficiencies are flat. The second and third factors have been calculated with the detector simulation in the previous section. If we put all of them together, the wrong-sign background fractions in FHC and RHC can be obtained. Figs. ?? and ?? show the results.

3. electron neutrino constraints on hadroproduction F. What is the complementary physics for the 3DST relative to the other trackers? How would it improve CP sensitivity?

The 3DST detector placed on the inside, upstream part of the magnetic volume provides physics program complementarity to the other components of the near detector. The nature 33

• FIG. 34: Fluxes of muon neutrino and anti-neutrino in the FHC (nu mode) and RHC

(nubar mode) as functions of neutrino energy and lepton angle.

• f that complementarity is described below.

A unique feature of the 3DST detector is that it is made of scintillating plastic. This means, as discussed in Section ??, the scintillating plastic material in the 3DST will provide a transparent connection between the 3DST dataset and the vast catalog of cross section data from MINERvA and T2K’s ND280 detector. The similarity between the 3DST data and these other data sets should facilitate the understanding of systematic errors and com- parisons used in flux constraints. Another unique feature of the 3DST is that it is very similar (perhaps identical except in form factor) to the SuperFGD target under serious consideration for use in the T2K ND280

• upgrade. Assuming the SuperFGD becomes part of ND280, T2K will provide valuable cross

section results with very similar systematics in a narrow band of energy nicely overlapping the DUNE second oscillation maximum, as shown in Fig. 37. DUNE has events in this 34

• FIG. 35: Wrong-sign fraction in FHC in the neutrino energy-lepton angle space.

region but is vulnerable from confusion from feed down from the substantial flux at higher

• energy. In a very real sense the superFGD in the T2K beam acts as a test beam calibration

for DUNE 3DST events in an important region of phase space. The T2K experience should contribute toward a deeper understanding of the systematic errors in the DUNE 3DST. The 3DST detector will collect relatively large, sign-selected, charged current νe and νe samples. These samples are potentially helpful for tuning the kaon component of the hadroproduction model for the flux. The production of νe and νe above a neutrino energy

• f *** are dominated by the secondary kaons produced in the beam. **** still needs work,

emailed mike and laura *** 35

• FIG. 36: Wrong-sign fraction in RHC in the neutrino energy-lepton angle space.

1. Synergy with the Liquid argon TPC

The 3DST, like the liquid argon detector upstream, will generate high statistics samples. In the event that the liquid argon detector has problems, the high statistics sample from the 3DST would be an essential substitute for the core mission of the near detector of constraining the muon neutrino flux if the low density tracker is a gas TPC. In a scenario where a DUNE-prism system is implemented and the liquid argon detector moves off-axis ∼50% of the time, a 3DST on-axis provides continual monitoring of the on-axis flux with high statistics. The liquid argon detector is not in a magnetic field. Muons produced in the interactions in liquid argon will be analyzed by range up to ∼1.5 GeV/c. Higher momentum tracks in

• nly a relatively small angular region penetrate the light density, magnetized tracker where

they are momentum analyzed with precision. There is dip in the acceptance for muons from 36

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Energy (GeV)

µ

ν 0.2 0.4 0.6 0.8 1 Flux (A.U.) DUNE FD Osc. T2K ND

• Osc. regions
• FIG. 37: The DUNE (blue) and T2K [3] (red) neutrino beam fluxes as a function of the

neutrino energy are shown. The gray bands show the energy regions where the expected DUNE oscillation maxima occur. the liquid argon due to tracks stopping in passive material (the magnet coil) before reaching the magnet analyzing volume. A 3DST detector in the magnetic volume can perform charge separation and magnetic tracking over large acceptance. High momentum muons will be precisely momentum analyzed by the precision tracker over a larger solid angle than those coming from the liquid argon target. The threshold for sign separation for muons can be pushed down to 150 MeV. *** Cafana study by Seb Jones showing effect of this on CP sensitvity *** The 3DST has very fast timing. It should be more robust to issues arising from high rate

• r backgrounds than the liquid argon detector.

A low-ν analysis done in the 3DST offers a flux shape extracted from a statistically independent sample from the liquid argon detector. This flux can be used for CCQE-like liquid argon measurements that can be correlated with a low-ν flux determined in the liquid argon detector. This correlation can be mitigated at the cost of statistical power. Such a sacrifice is not necessary using the low-ν flux from the 3DST. 37

2. Synergy with the straw tube tracker

The straw tube tracker has excellent tracking resolution, significant mass and fast timing, as well as a substantial data set coming from a hydrocarbon target. As such, the 3DST plays less of a complementary role to the straw tube tracker than it does to the HPGasArTPC and liquid argon TPC. As mentioned above, the 3DST data should be more transparently comparable to the trove of results from MINERvA and ND280 (both before and after the current upgrade). Also, the similarity between the superFGD and 3DST may be helpful in understanding systematic issues for interactions in the second oscillation maximum region of the flux, as discussed above. The 3DST has true three dimensional readout and a smooth tracking efficiency as a function of 4π. The STT has a planar structure that may, in principle, lead to some accep- tance/acceptance modeling/reconstruction issues at 90 degrees. That has not been studied as part of this work.

3. Synergy with the high pressure gaseous argon TPC

In contrast to the situation with the straw tube tracker, the 3DST is nicely complemen- tary to the HPGArTPC. Whereas the HPGArTPC has exquisite resolution, low detection thresholds, and an argon target, the 3DST is very fast, has mass for high statistics, good containment for photons and sensitivity to neutrons (all of which are discussed in other places in this document). In a scenario that the 3DST and HPGArPTC share a magnetic volume, the photon containment and fast readout speed of the 3DST will help insure the backgrounds in the volume are understood. This, in turn, will help validate the analysis/understanding of interactions producing pizeros in the HPGArTPC. The 3DST can provide high statistics in the event that the liquid argon detector has problems or is moved off-axis. Outside of tuning the a priori flux model with measured interactions in the near de- tector, there are two analyses that can provide useful additional constraints on the flux: neutrino-electron scattering and the low-ν technique. The scattering measurement requires high statistics that the 3DST has in abundance relative to the HPGArTPC. Statistics is 38

also likely to be more of a limitation for the HPGArTPC in a low-ν analysis. In addi- tion, the HPGArTPC may have more uncertainty in the recoil energy reconstruction from backgrounds in the ECAL. This has not been studied carefully as part of this work. Finally, as mentioned above, the 3DST data will be transparently comparable to the catalog of cross section results from MINERvA and ND280. In particular, data taken with the very similar superFGD in the T3K beam may be very useful in facilitating an understanding of the DUNE flux near the second oscillation maximum via the 3DST data set.

G. Selected physics processes

Below are five physics processes identified by the near detector concept study group as important measures of near detector capability. Given the compressed time frame for the 3DST studies, detailed and formal analyses of these processes were not practical. Still, some work was done and other things can be inferred from experience with scintillator detectors in other experiments.

1. neutrino-electron scattering

This process is discussed above in Section V B.

2. coherent charged and neutral pion production

The state-of-the-art in charged-current coherent pion production measurements are the low energy NuMI beam results published by MINERvA recently [? ]. These results are a useful point of comparison for the 3DST in DUNE since the low energy tune for NuMI is similar to the expected DUNE flux and the detector is similar in many ways. In terms of statistics, the MINERvA coherent analysis, based on the low energy run, was based on a fiducial volume of 5.47 metric tons. This is slightly smaller than the fiducial volume of the 2.4 x 2.4 x2.0 m3 3DST in a beam that is less intense. Important aspects of the coherent charged pion analysis to consider are sensitivity to vertex activity and full reconstruction of the pion. The vertex activity sensitivity is necessary 39

to eliminate events where the nucleus is excited or broken up, and therefore not part of a coherent process by definition. The pion reconstruction is necessary for the reconstruction

• f the kinematic quantity, t, which is the energy transfer to the nucleus ***. In this variable,

coherent events can be tagged in a model independent fashion. Relative to MINERvA’s strip-plane readout, the 3DST has a full 3 dimensional fine- grained readout. The 3DST cell width is less than the MINERvA strip thickness of 1.5 cm. This leads to the expectation that the 3DST should have somewhat better sensitivity to vertex activity. The 3DST has the advantage of a magnetic field relative to MINERvA. On the other hand, the current concept of the 3DST does not contain a hadron calorimeter downstream. Charged pions used in a coherent analysis will need to meet requirements of tracking mo- mentum reconstruction quality or containment in the 3DST plus ECAL. For a coherent pizero analysis, the final state is a lone pizero. The neutrino energy and t cannot be determined. The pizero energy,angle and the lack of observable vertex activity are the handles to tag coherence. Due to the missing information, model-dependent assumptions about the Q2 dependence must be made in the course of efficiency corrections. In addition, the neutrino energy cannot be determined and the results must be averaged

• ver the neutrino energy. These experimental limitations enhance the value of the charged

pion coherent production, which can be related to the neutral current reaction theoretically. In spite of the experimental limitations of the measurement, this process is something that should not be very challenging for the 3DST. The 3DST has high statistics, relatively good vertex sensitivity and good photon containment (shown in Section ??). The ability to reconstruct and point photon showers to a nominal pizero production point will be useful in ruling out vertex activity.

3. low-ν

The low-ν technique to measure the flux shape was proposed by Mishra[? ] and used by several experiments, including a recent analysis by MINERvA in the NuMI low energy beam[? ]. In the limit of low ν the neutrino and anti-neutrino cross sections are approxi- mately constant with (anti-)neutrino energy. This means a measurement of the low-ν inter- action rate is a measure of the shape of the flux. This flux shape is then normalized at a high 40

neutrino energy where the flux is well-measured to provide an absolute flux measurement. With a similar beam energy profile and similar detector material/technology, the recent MINERvA analysis provides a reasonable point of comparison for the expected 3DST low-ν

• performance. Significant issues to consider in this comparison are the momentum range for

muon acceptance, the recoil energy resolution, the low-ν cutoff threshold, and the interplay between different data sets and the need to maintain statistical independence. An important part of a low-ν analysis is the position of the cutoff in ν, ν◦, below which is defined as the low-ν sample. The lower the cutoff, the less is the energy dependence of the low-ν cross section which reduces the flux model dependence. On the other hand, a lower limit of ν◦≥300 MeV was used by MINERvA to avoid modeling uncertainties at lower energy transfer arising from disagreements between the data and the GENIE simulation in CCQE scattering and ∆ resonance production [? ]. Given the fast-moving progress in the modeling of the CCQE process and pion production (e.g., addition of 2p2h and RPA and studies of low recoil samples and transverse variables), it is reasonable to assume that a DUNE-era low-ν analysis will face a situation where ν◦ can be pushed lower. This will allow the low-ν flux to be pushed to lower (anti-)neutrino

• energy. This, in turn, will benefit from good acceptance for muons at low momentum.

MINERvA uses muons with momenta greater than 1.5 GeV/c in order to insure they are analyzed in the magnetized MINOS near detector downstream. (This is fairly similar to the situation faced by the DUNE near detector for a low-ν analysis of neutrino interactions in the liquid argon detector.) For a low-ν analysis of interactions in the 3DST, the muon momentum cut could be as low as 200 MeV ***?**. MINERvA shows a fractional recoil system energy resolution of σ ν = 0.132(0.163) 0.329(0.283) √ν (3) where the numbers in parentheses are for anti-neutrinos [? ]. The 3DST should plausibly achieve similar to somewhat better performance. The higher granularity could lead to better

• performance. Also, a careful program of test beam studies might lead to a reduced systematic

error on the recoil energy resolution relative to MINERvA. One of the issues that concerns MINERvA in using the low-ν flux is potential sample

• verlap and correlation with samples used for CCQE cross section analyses. Care must be

taken to avoid overlaps between samples used for these two purposes and this can cause issues 41

with statistics. In the case of the DUNE near detector. the 3DST offers the opportunity to make a low-ν flux measurement that is statistically independent of the data sample collected in the liquid argon detector (and vice versa). This will allow for the use of the low-ν (CCQE-like) data in the liquid argon detector for the determination of cross sections using the full statistical power without concerns for overlap. Similarly, cross sections can be determined in the 3DST using the low-ν flux as evaluated in the liquid argon detector. The independent determinations of the low-ν flux in the liquid argon and 3DST will also afford a nice systematic cross check of many aspects of the analyses and the flux. *** can 3DST plausibly do low-nu for electron neutrinos to get at the electron neutrino flux shape, Laura Fields suggestion ***

4. inclusive CC 5. NC/CC neutral pion production

Using a similar beam and detector technology to the 3DST, the MINERvA experiment has made very nice measurements of charged-current neutral pion production by neutrinos [? ] and anti-neutrinos [? ]. One essential ingredient to this work is that a substantial fraction of the decay photons are contained by the detector. Section ?? shows that the 2.4 x 2.4 x 2.0 m3 3DST detector has good containment for these photons. In addition, the true 3 dimensional readout and high granularity of the 3DST should offer substantial advantages to the 3DST in separating and reconstructing the photons from pizeros relative to MINERvA. With good photon containment and fast timing, it is plausible that the 3DST can recon- struct neutral current pizero events well. (As a point of comparison, MINERvA has not yet published a paper on this process. The reason for that is not technical but rather one of topic prioritization with limited manpower.) The significant difference from the charged-current version of this process is that there is no muon emanating from the interaction vertex. There may be vertex activity that can be used to help anchor the pizero decay position. Without that anchor point the pointing resolution of the reconstructed photon showers becomes more

• important. The true 3 dimensional readout and high granularity of the 3DST should allow

for good performance on photon shower performance. A detailed study on photon shower 42

• FIG. 38: Schematic of the 3DST inside the HPGArTPC. On the left, the position of the

3DST is shown. On the right, the active region of the TPC that is removed by presence of the 3DST is highlighted. pointing resolution in the 3DST has not yet been done.

VI. POTENTIAL U.S.-JAPAN COOPERATION FUNDING VII. OTHER GEOMETRIES A. 3DST inside HPGArTPC

As mentioned in Section V F, the 3DST is nicely complementary to the HPGArTPC. The 3DST data would: connect well with the vast catalog of neutrino-plastic cross section data; provide high statistics and substantial containment for photons; provide sensitivity to neutrons; and create a transparent link to the SuperFGD dataset anticipated in T2K. Also, the 3DST has very fast readout. The HPGArTPC, on the other hand, has a lower momentum threshold for tracking and exquisite charged particle tracking resolution. In the limit where the magnet volume may need to be minimized to reduce cost, ten- sion arises between these two potential components of the near detector. A sizable 3DST component inside the magnet volume will reduce the potential size of the TPC and its corresponding data set. Given the complementarity of these two detectors, it is interesting to explore the potential reduction in the spatial tension between them by placing the 3DST on the upstream interior 43

3DST dimensions (m) 3DST mass (tons) 3DST HPGArTPC Height Width Depth Total Fiducial νµ CC events (FV) fraction active νµ CC events (FV) 1.5 2.0 2.0 6.4 4.7 6.7×106 67% 1.1×106 1.5 4.0 1.5 9.5 7.3 1.1×107 79% 1.3×106 1.5 4.0 2.0 12.7 9.9 1.5×107 67% 1.1×106 2.0 2.0 2.0 8.5 6.5 9.5×106 65% 1.07×106 2.0 4.0 2.0 17.0 13.8 2.0×107 65% 1.07×106 2.0 2.0 2.5 10.6 8.2 1.2×107 53% 8.7×105 2.0 4.0 2.5 21.2 17.4 2.5×107 53% 8.7×105

TABLE II: This table summarizes the event rates for varying 3DST sizes in the scheme where the 3DST is placed up against the interior, upstream edge of the HPGArTPC.

• f the HPGArTPC volume, as shown in Fig. 38. In such an arrangement the 3DST has a

smaller impact on the active volume of the TPC. Table ?? summarizes the event rates for various 3DST sizes in the scheme where the 3DST is placed up against the interior, upstream edge of the HPGArTPC. The outer 10 cm of the 3DST is removed on the upstream edge and sides to define the detector fiducial

• volume. The rates are given in νµ CC events per year per metric ton, assuming ***beam***.

‘Fraction active’ is the fraction of the HPGArTPC active volume that remains relative to the size without the 3DST. The width of the 3DST could be as large as 5 m in this scheme, though smaller values were chosen for illustration. In the event that the width of the 3DST is significantly less than 5 m, regions to the sides of the 3DST might be made active for the

• TPC. For these calculations it was assumed that this was not the case.

This scheme has the additional advantage of making the 3DST/HPGArTPC a true hybrid detector in the sense that, with high acceptance and minimal intervening material, 3DST tracks can be measured precisely in the TPC. The ECAL of the TPC (interior and exterior) can be shared by both the 3DST and the TPC. The 3DST acts as a partial Ecal on the downstream side of the TPC. A GEANT simulation of this concept has not yet been created. Such a simulation will be necessary/useful for quantifying the charged particle acceptance in the TPC for tracks ema- nating from 3DST interactions and studying the reconstruction of electromagnetic showers 44

• FIG. 39: Picture of a small Super-FGD prototype. Several cubes of extruded plastic

scintillator with three fibers inserted in the three holes are assembled. The size of each cube is 1 × 1 × 1 cm3. From [12, 14] that are not contained in the 3DST.

VIII. R&D

Many core aspects of the 3DST design are the same as the Super-FGD detector being discussed by the T2K collaboration as part of the ND280 upgrade [14]. As mentioned in Section VI a number of groups on both DUNE and T2K are collaborating on the development

• f both detectors. Though the detailed designs of the 3DST and Super-FGD will differ in

the end, the basic structure and readout are expected to be the same and much of the work done in evaluating the Super-FGD design and developing construction techniques can be considered as R&D for the 3DST as well. Some of that R&D is described in the ND280 Upgrade Proposal and summarized here [14]. The candidate scintillator is a composition of a polystyrene doped with 1.5% of parater- phenyl (PTP) and 0.01% of POPOP. The cubes for Super-FGD prototypes to date were produced by Uniplast in Vladimir, Russia. The cubes are covered by a ∼ 50µm thick chem- ical reflector, obtained by etching the scintillator surface with a chemical agent that results in the formation of a white micropore deposit over the polystyrene [13]. Each cube has three orthogonal cylindrical holes of 1.5 mm diameter drilled along X, Y and Z axes. Three 1.0 mm diameter WLS fibers are inserted through the holes. The current R&D is based on the design of 1×1 cm2 cube. If this design turns out to be difficult to achieve (e.g. due to large number of channels), a possible alternative option is 45

10 300 mkm 75.4 microns

• FIG. 40: Measured dimensions of 166 scintillator cubes. Three sides are measured and

plotted together. The distance between the two vertical lines corresponds to 300 µm. From [12, 14]

.

to enlarge the size of the cube to 1.5×1.5 cm2 or 2×2 cm2, which could drastically reduce the number of cubes and readout channels at the risk of granularity. A picture of a small prototype is shown in Fig. 39. Properties of the scintillator cubes have been measured in INR Moscow. The measured dimensions of the first 166 cubes are shown in Figure 40. The size was measured for three directions and all are plotted together. The mean and sigma are measured to be 1004.2 µm and 75 µm, respectively, for this set of cubes. A production of 10,000 cubes is in progress to study the mechanical issues. R&D is ongoing to improve the cube size manufacturing precision. As part of the Super-FGD R&D, the light yield and timing resolution were measured with cosmic rays. Figure 41 shows a schematic view of the measurement setup as reported in [14]. Scintillator cubes were read out via wavelength shifting fibers and Multi Pixel Photon Counters (MPPCs). The length of fiber was 1.3 m, with a distance between scintillator and MPPC of 1 m. The measured light yield for this prototype test is shown in Fig. 42. On average, the light yield was 50-60 photoelectrons at 1 m from MPPC. The timing resolution was found to be 0.91 ns RMS for 1 fiber readout and 0.63 ns RMS in the case where the signal from two fibers was used. The Super-FGD group carried out a test beam experiment at CERN in October 2017 to evaluate the performance in more detail. A small detector with 5×5×5 cubes was assembled 46

• large trigger counter

large trigger counter small trigger counter small trigger counter small trigger counters: 8x8 mm2 Distance from MPPC to cube – 100 cm MPPC MPPC

• FIG. 41: Setup of light yield measurement with cosmic rays. From [12, 14].

Entries 267 Mean 55.33 RMS 19.9

Light Yield [# of photoelectrons]

50 100 150 200

# of events

5 10 15 20 25 Entries 267 Mean 55.33 RMS 19.9

Light Yield

Entries 1024 Mean 4.777 RMS 0.9109

Time [ns]

2 4 6 8 10

# of events

2 4 6 8 10 12 14 16 18 20 22

Entries 1024 Mean 4.777 RMS 0.9109

Time resolution with 1 WLS fiber

• FIG. 42: Measured light yield (left) and timing resolution (right) of 1×1×1 cm3 scintillator
• cubes. From [12, 14]

as shown in Fig. 44. The prototype was equipped with 75 readout channels, 1 × 1 mm2 MPPCs and 1.3 m long wavelength shifting fibers. A wave-form digitizer with a sampling rate of 5 GHz was used for the signal readout in order to measure the intrinsic time resolution

• f the detector. The beam tests were performed in the T10 beam test area at CERN with

6 GeV/c beams of pions and muons. A trigger system composed by two 3×3 mm2 counters, used to select particles crossing cubes along a defined direction, and a 9 mm diameter

• veto. First the light yield from a single WLS fiber perpendicular to the beam direction was
• measured. On average the light yield was 41 photoelectrons (p.e.) per fiber. Lower light

yield compared to the previous cosmic data run is due to the shadowing between fibers, confirmed by dedicated studies. However this effect has not any major impact on the total light collected from a cube. Indeed the light yield collected by two fibers is about 79.8 p.e., 47

• FIG. 43: Small detector with 5×5×5 cubes for the 2017 beam test at CERN (left) and T10

beam test area (right). consistent with two times the light measured with a single fiber. The cross talk, defined as the amount of light collected in a cube adjacent to the one crossed by the beam, was measured to be about 3.7% and is mainly due to light penetration through the white reflector. This is an upper limit, since corrections for the dark rate are not included. The time resolution was measured for different cases. If the light from a single fiber is considered, the time resolution is about 0.92 ns. While if the light from two fibers in the same cube is collected, it improves down 0.68 ns. If a track crosses two cubes and the light collected by four fibers is taken into account, the time resolution becomes 0.53 ns. Additional tests were done with the same prototype using cosmic data. For tracks crossing at least 5 cubes, a time resolution better than 0.3 ns was obtained. While the response of the detector has been understood and shown to have unprecedented performances, the assembly of the detector is one of the main challenges for the construction

• f this detector. R&D was done at INR in order to address this item and it was demon-

strated the feasibility to assemble 5,000 cubes in a single module. “Fishing lines” of 1.3 mm diameter, thicker than the WLS fibers, were used to align the cubes along the three orthog-

• nal directions. First, planes of 600 cubes were aligned, finally the “fishing lines” along the

upward direction were inserted. The final step consists of removing the fishing lines and insert the WLS fibers. The procedure was smooth and relatively fast. We believe that this procedure can be scaled up to a module of the dimension of 60 × 60 × 60 cm3. In addition, 48

• FIG. 44: Prototype of 5,000 cubes during the assembling procedure (left) and final

prototype (right). R&D to improve the tolerance in ongoing. The goal is to achieve a tolerance as low as 20 µm

• n the cube edge, similar to the one of the current plastic scintillator bars of the Fine-Grain

Detector (FGD) currently installed in ND280. The Super-FGD group along with DUNE collaborators interested in furthering our un- derstanding of the 3DST detector have proposed building a 25x25x25 cube prototype in 2018, as discussed in Section VI. The hope is to take cosmic ray and beam test data with this prototype. Both the SuperFGD and the 3DST detectors will be installed inside similar near detector complexes composed by several sub-detectors immersed in a magnetic field. The design must carefully minimize the material budget between sub-detector in order to avoid particles stopping outside the fiducial volume regions and worsen the event reconstruction. The sub- detectors must be very close to each other also to improve the acceptance and the efficiency for matching shared tracks. An important part of the R&D is focusing on the design of a very compact signal read-

• ut system, that includes new generation MPPCs, the Front-End Electronics (FEE) and a

cooling system to deal the heating due to the several thousands readout channels in a reduce amount of space. We are developing a preliminary design, to be tested on the big prototype after the test beam. 49

IX. REFERENCES ACKNOWLEDGMENTS

This work was

[1] Y. Fukuda, et al. Phys. Rev. Lett. 81:1562 (1998). [2] K. Abe, et al., Phys.Rev. D96 (2017) no.9, 092006 [3] K. Abe, et al., Phys.Rev. D87 (2013) no.1, 012001 [4] U. Mosel, K. Gallmeister, arXiv:1712.07134 [5] https://github.com/gyang9/dunendggd [6] R. Acciarri et al., “LBNF and DUNE Conceptual Design Report”, arXiv. 1512.06148 [7] C. Marshall and C. Wilkinson, DUNE ND 2017 November workshop talk [8] Private conversation with Chris Marshall [9] https://github.com/ClarkMcGrew/edep-sim [10] https://github.com/gyang9/3DSTSim [11] J. Park et al. (MINERνA Collaboration)

• Phys. Rev. D 93, 112007 - Published 10 June 2016

[12] A. Blondel et al 2018 JINST 13 P02006 [13] Yu. G. Kudenko, L.S. Littenberg, V.A. Mayatsky, O.V. Mineev and N.V. Ershov, Extruded plastic counters with WLS fiber readout, Nucl. Instrum. Meth. A 469 (2001) 340. [14] CERN-SPSC-2018-001 (SPSC-P-357), https://cds.cern.ch/record/2299599/ [15] P.-A.Amaudruz, et al., “The T2K Fine-Grained Detectors”, Nucl. Instrum. Meth. A 696, 1 (2012) [16] put in minerva neutron Wc reference [17] martini 2p2h recon as ccqe reference [18] t2k transverse variables paper [19] minerva xianguo wc talk [20] nomad cc-nc sep result [21] CM ND presentation on CC-NC sep in limit of no neutron sensitivity

50

[22] J. Park, Neutrino-Electron Scattering in MINERA for Constraining the NuMI Neutrino Flux, thesis, 2013. [23] Flux constraint results of liquid argon case in the following study is obtained by using the fitting without the MCS tail, i.e. peak case.

51