UPDATE ON THE ARGON GAS TPC NEAR DETECTOR

Justo Martín-Albo (Oxford U.)

DUNE ND WG MEETING – 20th APRIL 2016

▸ United Kingdom: Imperial, Lancaster, Liverpool, Oxford, QMUL, RAL, RHUL, Warwick.

▸ Spain: IFAE Barcelona, IFIC Valencia.

▸ France: Saclay.

▸ Switzerland: Geneva.

▸ Poland: WrocĿaw, NCBJ.

▸ Germany: Aachen.

▸ USA: Colorado, Fermilab.

COUNTRIES & INSTITUTIONS 2

Growing community interested in gas TPCs for neutrino physics (DUNE, T2K, HyperK and neutrino cross sections).

CONCEPTUAL DETECTOR DESIGN

ARGON GAS TPC

ECAL

/ MR

D

ECAL / MRD

ECAL / MRDEC

AL /

MRD

NEUTRINO BEAM B,E

3

Monolithic, fully active Ar target. Gas TPC: low detection thresholds, intrinsic particle ID (dE/dx) and 3D reconstruction capabilities. Magnetized detector for charge/momentum measurement. Electromagnetic calorimeter surrounding TPC for detection of neutral particles.

ARGON GAS TPC: MOTIVATIONS 4

▸ Same nuclear target as that of far detector.

▸ Low detection thresholds: sensitivity to <100 MeV/c protons and <25 MeV/c muons and pions.

▸ Measurement of nuclear effects (2p2h, spectral functions, FSI…).

▸ Model testing and generator tuning.

▸ Target argon mass of the order of 1 tonne.

▸ Pressurized gas (10–20 bar) to reduce TPC dimensions.

▸ Gas density: O(16–32 kg/m3).

▸ TPC dimensions: 3–5 m long, 3–5 m diameter.

▸ Finding optimal dimensions requires (some) engineering input.

TPC DESIGN: DIMENSIONS 5

TPC PERFORMANCE 6

Rep.Prog.Phys.73

(2010)116201H

JH

ilke

Table 3. Characteristics and performance of some TPCs.

Parameter/Experiment PEP4 TRIUMF TOPAZ AlEPH DELPHI STAR ALICEa

Operation 1982/1984 1982/1983 1987 1989 1989 2000 2009Inner/Outer radius (m) 0.2/1.0 ∼ 0.15/0.50 0.38/1.1 0.35/1.8 0.35/1.4 0.5/2.0 0.85/2.5Max. driftlength (L/2) (m) 1 0.34 1.1 2.2 1.34 2.1 2.5Magnetic field (T) 0.4/1.325 0.9 1 1.5 1.23 0.25/0.5 0.5Gas : Ar/CH4 Ar/CH4 Ar/CH4 Ar/CH4 Ar/CH4 Ar/CH4 Ne /CO2/ N2

Mixture 80/20 80/20 90/10 91/9 80/20 90/10 90/ 10/ 5Pressure (atm) 8.5 1 3.5 1 1 1 1

Drift field (kV cm−1 atm−1) 0.088 0.25 0.1 0.11 0.15 0.14 0.4Electron drift velocity (cm µs−1) 5 7 5.3 5 6.69 5.45 2.7ωτ (see section 2.2.1.3) 0.2/0.7 2 1.5 7 5 1.15/2.3 <1Pads: Size w × L (mm × mm) 7.5 × 7.5 (5.3–6.4) × 19 (9–11) × 12 6.2 × 30 ∼7 × 7 2.85 × 11.5 4 × 7.5

6.2 × 19.5 6 × 10/15Max. no. 3D points 15—straight 12 10—linear 9 + 12—circular 16—circular 13 + 32—straight 63 + 64 + 32dE/dx: Max. no. samples/track 183 12 175 148 + 196 192 13 + 32 63 + 64 + 32Sample size (mm atm); w or p 4 × 8.5; wires 6.35; wires 4 × 3.5; wires 4; wires 4; wires 11.5 + 19.5; pads 7.5 + 10 + 15; padsGas amplification 1000 50 000 3000–5000 5000 3000/1100 20 000Gap a–p; a–c; c–gateb 4; 4; 8 6 4; 4; 8 4; 4; 6 4; 4; 6 2; 2; 6/4; 4 ; 6 2; 2; 3/3; 3; 3Pitch a–a; cathode; gate 4; 1; 1 4; 1; 1 4; 1; 2 4; 1; 1 4; 1; 1/ 4; 1; 1 2.5; 2.5; 1.5Pulse sampling (MHz/no. samples) 10/455, CCD only 1 digitiz., ADC 10/ 455, CCD 11/ 512, FADC 14/300, FADC 9.6/400 5–10/500–1000, ADCGatingc !1984 o.on tr. !1983 o.on tr. o. on tr. synchr. cl.wo.tr static o.on tr. o.on tr.Pads, total number 15 000 7800 8200 41 000 20 000 137 000 560 000

Performance#xT (µm)-best/typ. 130–200 200/ 185/230 170/200–450 180/190–280 300–600 spec:800–1100#xL (µm)-best/typ. 160–260 3000 335/900 500–1700 900 500–1200 spec:1100–1250Two-track separation (mm), T/L 20 25 15 15 8 - 13/30∂p/p2 (GeV/c) −1 : TPC alone; high p 0.0065 0.015 0.0012 0.005 0.006 spec:0.005dE/dx (%) Single tracks/ in jets 2.7/4.0 4.4 / 4.4 / 5.7/7.4 7.4/7.6 spec:4.9/6.8Comments a in single PCs chevron pads circular pad rows circular pad rows No field wires No field wires

strong E × B effect >3000 tracks "20 000 tracks

a Expected performance.b a = anode, p = pads, c = cathode grid.c o. on tr.: gate opens on trigger; cl.wo.tr. : opens before collision and closes without trigger; static : closed for ions only (see text).

16

Hilke, Rep. Prog. Phys. 73 (2010) 116201

TPC PERFORMANCE 7

Rep. Prog. Phys. 73 (2010) 116201 H J Hilke

Figure 26. PID from ionization measurements (dE/dx) in the PEP4TPC. (Reprinted with permission from [15]. Copyright 2008, TheRegents of the University of California.)

calculated straggling functions for different sample lengths xin argon. As these distributions are normalized to x, the meanloss ⟨!/x⟩ is the same, but the most probable loss changes.This means that straggling functions cannot be scaled witha single parameter (e.g. Imp). It turns out, however, that a2-parameter scaling is adequate [77].

The quality of PID is determined by the separation powerD with

D = (IA − IB)/[σ (IA) + σ (IB)]/2, (3.5)

i.e. the separation of particle types A and B expressed in num-ber of standard deviations. For the PEP4 TPC, the best K- πseparation obtained was about 3σ (Iπ ) with σ (Iπ )/Iπ = 3.0%for minimum ionizing pions [64]. Figure 30 shows for anumber of gas mixtures the measured separation power at15 GeV/c, obtained from the lowest 40% pulse heights from64 samples of 4 cm length [78]. The relativistic rise at 1 atm ishighest for noble gases, around 1.6–1.7 with respect to the min-imum dE/dx. The low-Z molecular gases show better resolu-tion σ (I ) but have a lower relativistic rise. There is no ‘magicgas’ with outstanding PID properties in all respects, althoughthere are significant differences. For the choice of gas in a par-ticular experiment, other characteristics such as diffusion, driftvelocity and radiation length will enter, in addition to PID.

For rough estimates of the relative truncated meanresolution σ (I )/I as a function of pressure p, sample size dand number N of samples in mixtures of Ar with 10–20% CH4

the following relation has been derived from the PAI model forpure argon [75]:

σ (I )/I = 0.41(pd)−0.32N−0.46 = 0.41(pL)−0.32N−0.14,

(3.6)

where L is the total length over which I is measured. To allowcomparison of different gases, an extension was proposed:

σ (I )/I = 0.345(Apd)−0.32N−0.46,

with A = 6.83νpD/(β2I ), (3.7)

where ν is the mean number of electrons per molecule. ForHe, Ne, Ar, Kr and Xe, A = 0.32, 0.50, 0.62, 0.65 and 0.70,respectively. Values obtained from these theoretical relationsbased on a likelihood treatment were intended as a guide onlyto estimate best possible resolutions. Based on experimentalresults, it has been argued that below a sample size of about5 cm atm in argon the gain in resolution is negligible [79].More recent studies, however, indicate that sampling downto 0.5 cm atm still improves the accuracy [80].

Table 4 gives an overview of the PID performance of someTPCs. Included are also the jet chamber of OPAL at LEP andtwo big detectors, the EPI and ISIS2, which were dedicated todE/dx measurements with only restricted tracking capability.The best measured values for σ (I ) are compared with the idealresolution obtained from the relations cited above, assumingthe maximum number Nmax of samples can be used. The lastrow shows the calculated resolution for the case that only 70%of Nmax samples are useful, which is closer to reality in mostcases. The measured values are very close to these theoreticalnumbers. For PEP4 and EPI, the best values are obtained forvery clean tracks, where the assumption of Nmax contributingsamples is reasonable.

4. Particularities of some TPCs

4.1. Global aspects

Many TPCs have been constructed. They may be groupedcrudely according to the type of experiments they were or arepart of.

(a) Electron/positron storage rings: PEP4, TOPAZ, ALEPHand DELPHI.These experiments demanded large TPCs but had tohandle only low rates of events with multiplicities of2–30 charged tracks. Their designs are similar. PEP4and TOPAZ were pressurized to maximize PID. ALEPH,the largest of these four TPCs, with the longest driftlength (2.2 m) of all TPCs up to now and DELPHI putemphasis on momentum resolution, pattern recognitionand reduced material in front of other components andchose operation at 1 atm. The size of the DELPHI TPC hadto be reduced because of the addition of a RICH for morepowerful PID. Both ALEPH and DELPHI used circularpad rows to improve spatial resolution. ALEPH choselonger pads for better momentum resolution at the highestenergies, DELPHI preferred shorter pads for better two-track separation of lower momentum particles.A reconstructed event from the PEP4 TPC is shown infigure 31.

(b) Heavy ions: fixed-target and collider experiments.Medium-sized and large TPCs have been used in a numberof fixed-target experiments: EOS/HISS at the BEVALEC,the BNL 810 TPC at Brookhaven and NA35, NA36 andNA49 at CERN. A very large TPC is operating for STARat Brookhaven and an even larger one, ALICE, is ready fordata taking at the LHC at CERN. All these TPCs have tohandle high particle multiplicities up to several thousandper event in STAR and even more in ALICE. They use(d)

25

[ .;. _ ._*, ..;.:. ,__. & ..*. L...x‘ r.,.pf.. _ . . _. ._I ., . . .., . . . . . . :: * .I .

RUN 2-26 ~11/16/811

0’ I I I I I I -30 -20 -10 0 10 20 a0

w- wo. n- 757

ALPHA XY Resolution et 6.6 Atm

Fig. 7. Dependence of rmsxy on the angle a for 8.6 atm data.

w4

u = 260 u. XY

the mean of this distribution and add it in quadrature to its standard deviation. Note that the distribution of Fig. 8 is not gaussian and that the peak value falls well below the quoted resolution. At the different pressures we obtain the values shown in Table III.

We use a similar method to measure the z res- olution. We see little dependence of rmsz on L, the drift distance, therefore we average over all tracks. The distribution of rmsz has a similar behavior as Fig. 8, and the average is calculated in the same way as for uxy. The results are shown in Table III.

Table III. Resolution in xy and x at different pressures.

P(atm) UxyoJ > o&J 1

8.6 260 654 4.0 299 810 1.5 439 857

The xy resolution of 8.6 atm is worse than the 150 u we expected to achieve. There are many ~ factors that can deteriorate the resolution.

(a) Electronic noise. The noise level at this time is about 2.4 counts, i.e., 1% of the signal for a minimum ionizing track. This is expected to contribute 6411 to the resolution.

(b) The electronic cal ibration is not optimized. We are working on improving the stability of the system.

(c) We expect distortions due to E x B effects, that is effects due to radial components of B. We have not included these corrections in the analysis yet. Corrections are expected to be as large as 120 n.

(d) Electrostatic distortions. The electric field is not uni form across the drift volume. We have observed some distortions in regions very close to the inner and outer radius of the TPC. Corrections for these distortions are not included.

The z resolution is also worse than expected, but again electronic noise and unrel iable cal ibra- tion affect this resolution.

We expect to improve the resolution when these effects are better understood and corrected.

4.0 Momentum Resolution

This can be calculated by comparing the curvature of the cosmic rays in the two track sections in opposi te sectors. The distribution of the difference of curvatures for one of the data runs at 8.6 atm is shown in Fig. 9. The rms

PEP4 RUN 2-26

CURVATURE DIFFERENCE (Cm-‘ ) MOMENTUM RESOLUTION AT 6.8 ATM

Fig. 9. Distribution of curvature difference for the same cosmic ray in two track portions. The standard deviation of this distribution yields dp/p2 = O.OBl(GeV/c)-l.

deviation of this distribution, after normalization, gives dp/p2 = 0.081 (GeV/c)-l . The resolution we expect to achieve for a 4 KG field is Wp2 = 0.016 (GeV/c)-1.

- 49 -

dE/dx and transverse spatial resolution of the PEP4 TPC.

CHARGE AVALANCHE READOUT: MPGD 8

CHARGE AVALANCHE READOUT 9

▸ Modern MPGDs solve traditional problems of TPCs (ExB homogeneity, ion feedback, operational stability…) allowing even pure noble gas operation.

▸ Used in existing detectors (T2K-ND280) and future ones (LHC upgrades, ILC TPC…).

▸ High-pressure operation requires quenchers to achieve high enough gains. (Most quenchers suppress scintillation light, but probably not needed.)

▸ Reaching pressures above 10 bar will require dedicated R&D.

Amplification field (kV/cm)30 40 50 60 70 80 90 100

gain

1

10

210

3102 bar 3 bar

4 bar

Figure 5: Gain curves with both alpha (circles) and gamma (crosses) peaks, for Ar+5% isobutane, at 2,3 and 4 bar.

of the mesh. The curve is compatible with usual curves taken with the 55Fe source atatmospheric pressure and shows that no significant attachment is present in the system.The relevant data were taken in conditions corresponding to the full electron transmissionthrough the mesh (on the plateau of the curve 6).

The FWHM of the alpha peak shows a characteristic dependency on the drift field,shown in Fig. 7, clearly correlated with the electron transmission plot of Fig. 6. Bestenergy resolutions are obtained for larger drift fields inside the electron transmissionplateau, although resolutions around the level of 2% and 2.3% (respectively for 2% and5% isobutane mixtures) are obtained for a relatively wide range of values. The bestobtained values were 1.8% and 2% respectively.

In the best cases, the peak showed a clearly asymmetric shape as it is shown inFig. 8. As this shape is not expected from just fundamental fluctuations in the primaryor secondary charge, it is indicative of (small) external effects related, for example, toincomplete charge collection for some events (for example for alphas coming from thebulk of the Americium and leaving some energy in it before exiting, or cases where thealpha tracks are in small angle with the cathode plane and part of its ionization is lostback to the cathode). Let us add that this asymmetry cannot be attributed to the effectof attachment, as the voltage configurations for which it is observed lie always well withinthe electron transmission plateau. In any case the amount of charge loss would veryplausibly follow a Landau distribution. In fact, the peak shape is very well parameterizedby a negative Landau function convoluted by a gaussian. If we attribute the mentionedexternal effects to the Landau function we can see the remaining gaussian as the intrinsicenergy resolution of the Micromegas. This deconvolution analysis, illustrated in the fitof Fig. 8, points to an energy resolution of 0.7% FWHM (0.9% for the 5% isobutane

7

Dafni et al., arXiv:0906.0534

OPTICAL READOUT 10

Electroluminescence detected with array of photosensors. (Photo: NEXT Collaboration.)

OPTICAL READOUT 11

▸ Popular in rare-event searches (dark matter, double beta decay).

▸ High-pressure operation (>10 bar) with pure gas feasible (cf. NEXT TPCs).

▸ VUV light detection requires wavelength shifters.

▸ Potentially, better energy resolution but worse spatial resolution.

▸ Possible weakness: large stretched meshes –> development of MPGD-like structures for generation of EL light.

UK PROTOTYPE 12

▸ Imperial, RHUL, Warwick, and Lancaster were awarded a grant from the Projects R&D Scheme in the UK.

▸ Physics goal: build something to put in a proton/pion test beam at CERN.

▸ Physics goal: do the ν generator and simulation development to evaluate impact on future experiments

▸ Detector design: 5 bar, 0.5 m drift, 1 m2 readout.

1.2 m

0.6 m

UK PROTOTYPE: OPTICAL READOUT 13

▸ RHUL runs the DMTPC project, which uses CCD cameras to read out a gas TPC for dark matter

▸ Pressures & energies scale correctly so that HPTPC can build on this work.

Optical Readout๏ Big problem is how to

instrument a large gas volume

๏ RHUL runs the DMTPC project, which uses CCD cameras to read out a gas TPC for dark matter

๏ Pressures & energies scale correctly so that HPTPC can build on this work—if we can make it fast enough

12

x pixel200 300 400

y pi

xel

750

800

850

900

950

-40-20020406080100120140

10 MeV simulated proton at 2bar

100 keV data nuclear recoil at 120 torr

Optical Readout๏ Big problem is how to

instrument a large gas volume

๏ RHUL runs the DMTPC project, which uses CCD cameras to read out a gas TPC for dark matter

๏ Pressures & energies scale correctly so that HPTPC can build on this work—if we can make it fast enough

12

x pixel200 300 400

y pi

xel

750

800

850

900

950

-40-20020406080100120140

10 MeV simulated proton at 2bar

100 keV data nuclear recoil at 120 torr

SIMULATION AND RECONSTRUCTION 14

▸ Provide realistic simulated and reconstructed data for optimisation studies and task force detector comparison.

▸ Experiment-agnostic software framework.

▸ Baseline design:

▸ 5 m long, 5 m diameter vessel

▸ Argon + CF4 + C4H10 (95:3:2)

▸ 10 bar –> 1.3 tonnes

▸ ND280 ECAL design

▸ Dipole magnet, 0.4 T

SIMULATION AND RECONSTRUCTION 15

EVENT GENERATION (GENIE)

PARTICLE TRACKING (GEANT4)

PRIMARY SIGNAL GENERATION

DIGITIZATION & ELECTRONICS

SIMULATION RECONSTRUCTION ANALYSIS

RECO TPC 3D HITS

CLUSTERING / PATTERN RECOGN.

TRACK FITTING

DETECTOR MATCHING

EVENT SELECTION

SIMULATION AND RECONSTRUCTION: STATUS 16

▸ Basic development of G4 simulation [Liv, Oxf] –> DONE

▸ Simulation of drift and readout [RHUL, Oxf] –> DONE

▸ Clustering / pattern recognition using TREx [Warw]

▸ Work in progress; first release for June.

▸ Track fitting using RecPack [Oxf]

▸ Work in progress; first release for DUNE CM.

▸ First detector performance results (momentum and dE/dx resolutions) [Oxf]

▸ Work in progress; aiming at DUNE CM.

▸ Geometry optimisation studies (TPC and ECAL) [ColB, Oxf, Liv]

▸ In progress.

TREX 17

▸ TREx is a reconstruction package developed for finding and reconstructing interactions inside the volume of the gas TPC in T2K-ND280.

▸ Well validated and now being used as main ND280 TPC reconstruction, as well as for specific gas interaction analyses.

▸ Explicitly designed to find primary vertices inside gas, and to handle tracks with large curvature. Builds events containing paths and junctions (vertices).

TREX 18

Eddy Larkin TPC Gas Interactions 6/31

Pathfinding

Basis of pattern recognition● Join two points up via a series of 'connected' hits● With a little creativity, can use this to find vertices

*There are a whole bunch of algorithms for doing this – we use the A* algorithm. See backups!

Eddy Larkin TPC Gas Interactions 8/31

Edge detection

Find hits on the edge of the pattern● Just look for groups of cells at maximum or minimum in x, y, z

Eddy Larkin TPC Gas Interactions 9/31

Vertex finding

Done via path finding● Connect edges via paths● Look for points where pairs of paths start to diverge

Eddy Larkin TPC Gas Interactions 10/31

Path finding

Either● Connect vertices to path ends, or● Connect track ends to each other (if there are only two)

Eddy Larkin TPC Gas Interactions 12/31

Likelihood fitting

Run on returned paths, work out kinematics● See tpcRecon

Eddy Larkin TPC Gas Interactions 13/31

PID

Work out what stuff is from dE/dx● See tpcRecon

muonproton

protonmaybe?

TREX: RECONSTRUCTED EVENT FROM ND280 19

SUMMARY 20

▸ Simulation work proceeding well in the context of the DUNE task force:

▸ MC productions (run-throughs) in December 2015 and April 2016.

▸ First reconstruction results expected soon.

▸ Full reconstruction and event selection ready for 3rd run-through (September 2016).

▸ A couple of TPC prototypes under development by different groups and with different readout technologies.