Update on the HPgTPC System Alan Bross Near Detector General - - PowerPoint PPT Presentation

Alan Bross Near Detector General Meeting September 5, 2018

Update on the HPgTPC System

• 4 Components

–HPgTPC itself –ECAL

• Also neutron detector

–Magnet system (coils + steel) –µ tag (using steel above)

High-Pressure gas TPC (HPgTPC) System

2

• Sept. 5, 2018

Near Detector MTG

• The HPgTPC provides two functions

– Aids in event containment for ArgonCube – Very-high resolution stand-alone experiment (1t fiducial target mass)

• ~ 4-5M nµ CC events in 3 yrs n running
• Build copy of ALICE TPC

– Re-purpose ALICE readout chambers – Represents significant savings in both hardware and engineering

• Tremendous synergy with ArgonCube

– Target nucleus

• 100% Ar in LAr and 97% Ar in gas.

– Raw 3-D data – Similar FE electronics architecture

• Front-end chip design (LArPix GArPix)

– Same DAQ – Common data structure

HPgTPC: Review

3

• Sept. 5, 2018

Near Detector MTG

HPgTPC: Review II

4

• Sept. 5, 2018

~5.3m ~5.2m

• Build copy of ALICE

TPC re-purposing their readout chambers (available in 2019)

– But run at 10 Atm as

• pposed to 1
• Active volume ~ 97 m3

– Central hole instrumented

• ~550,000 pads in

existing chambers 1 Atm Ne-CO2

Near Detector MTG

High Pessure TPC test stand: GOAT

5

• Sept. 5, 2018

Now Complete

• Reproduce ALICE
• peration at 1 Atm

with Ar-CO2

– Measure gain, wire stability, noise, etc.

• Operate at 10 Atm

and study various gas mixtures

• Electronics test

stand

– GArPix

GOAT: GAr Operation of the ALICE TPC Guillermo Fernandez-Moroni, Jen Raaf, Tanaz Moyahai &AB

Near Detector MTG

GOAT: ALICE TPC part

6

• Sept. 5, 2018
• Test vehicle is an ALICE inner readout chamber (IROC)

– ~5500 pads

• Field cage with 8 cm drift

Near Detector MTG

GOAT: Cosmic Data

7

• Sept. 5, 2018
• Looking for signal coincidence between two sets of pads that have been

ganged together.

• The distribution should go to zero for time diference larger than the drift

length (5µs).

Near Detector MTG

• Is performed using an 55Fe radiation source which has a X-

ray peak at 5.9keV

• The active volume illuminated by the source is recorded

using high-bandwidth electronics.

• An off-line pass band digital filtering scheme is used to

extract the X-ray signal.

GOAT: Gain Calibration

8

• Sept. 5, 2018

Test pads Reference pads

Near Detector MTG

Raw signal from the DAQ (anode 1375 V)

Time (125ns) Raw signal (0.48 mV) X-rays events Each negative jump corresponds to an event in the IROC

• Sept. 5, 2018

9 Near Detector MTG

Signal pulse height vs. Anode voltage (after Digital Filter)

Noise peak 5.9 keV peak double event peak

• Sept. 5, 2018

10 Near Detector MTG

Gas Gain for Ar-CO2 (90-10), 16 psi abs. press.

Preliminary ALICE BIT LOW?

• Sept. 5, 2018

11 Near Detector MTG

• The TPC is an inherently slow device

– Drift over 2.5m ~ 50 µs (for vdrift = 5cm/µs)

• 10 µs spill

– Yields ~ 50 cm uncertainty w/r to where the vertex is.

• Not an issue for any event with a charge track leaving the TPC and

entering the ECAL.

– Having a time stamp would help with the rest

• Matching to ECAL hits then becomes straightforward
• This is a bit tricky, since many of the conventional gas

mixtures (P10) quench all fluorescence.

• Promising ternary systems:

– Ar + Xe + CH4, Ar + Xe + CF4, Ar + N2 + CF4 – Fluorescence light in the visible?

Gas studies: Fluorescence tag

12

AB

• Sept. 5, 2018

Near Detector MTG

• There has been recent work in LAr to measure and use NIR

fluorescence for light detection in order to vastly reduce Rayleigh scattering.

– http://iopscience.iop.org/article/10.1209/0295-5075/106/32001 – Carlo Escobar’s work at PAB

• From IOP article

Another approach

13

• Sept. 5, 2018

Neumeier et al.

Near Detector MTG

10 ppm XE

Very strong emission in IR: ~ 127nm for pure LAr

14

• Sept. 5, 2018

Near Detector MTG

• Those data were in LAr. Are there data for gas?
• Well some, Neumeier, EPL, 106 (2014) 32001

What about in Gas?

15

• Sept. 5, 2018
• Ar-Xe mixtures
• ~1.5 Atm @ 100K
• Electron beam excitation
• Setting up to do

spectroscopic measurments at 10 Atm Ar-Xe mixtures @ 300K

1000ppm Xe Near Detector MTG

• The ~900nm emission is matched well to a new type of SiPM
• Hamamatsu & FBK SiPMs for LIDAR applications

– TOF imager for automobiles: self-driving cars

• Potentially big market

Why is this so interesting?

16

• Sept. 5, 2018
• Good match to the fluorescence in Lindbloom

paper

– FBK even higher PDE

• Fast
• Low noise: 1.5 kHz dark count rate

– This is VERY low for SiPM!

• Cheap?
• Fluorescence timing/tagging within the HPgTPC

active volume, while allowing for stable MWPC

• peration at the gains we need 2-5k, looks

possible.

• The NIR approach has many details to work out,

but would be a game changer, if workable.

Hamamatsu

Near Detector MTG

• Produce a modified design of LArPix (LBNL)
• Modifications:

– Polarity: induced signal in HPgTPC vs. charge in ArgonCube – Higher clock rate to accommodate larger vdrift

• Promising in the following sense

– LArPix exhibits similar noise for room temperature operation. – HPgTPC chamber pad capacitance similar to LAr pads capacitance. – Noise requirement < 1000ENC less stringent than in LAr – Much larger (easily X10) can be tolerated for the HPgTPC

HPgTPC electronics R&D

17

• Sept. 5, 2018

Dan Dwyer, Guillermo Fernandez-Moroni, Jen Raaf, Tanaz Moyahai

Near Detector MTG

LArPix ➛ GArPix

18

• Sept. 5, 2018

All plots from Dan Dwyer

Near Detector MTG

Software Choices: GArSoft

• ALICE software is very thorough and battle-tested.
• But it depends on ALICE geometry and is tuned to their physics

requirements (heavy ions at the LHC)

• The sector geometry of the pads is baked in the lowest levels of

simulation, hit finding, and reconstruction.

• Tracking efficiency for tracks within 45 degrees of the E field is poor.
• We need our own software to work with the DUNE ND complex, and

the rest of the experiment.

• So – use GArSoft, which was stared by Brian Rebel about a year and

a half ago.

• GArSoft is based on the art framework and NuTools, like LArSoft
• Sept. 5, 2018 Near Detector MTG

19

Tom Junk

Simplified Geometry to Get Started

• Just to get started, I made a cylinder of 10 bars of 90% argon and 10%

methane with the dimensions of the ALICE TPC

• No inner field cage since

there's no beam pipe.

• Added a 30-cm thick ECAL

made out of 10 g/cm3 Pb-C-H mixture

• ECAL stops particles

(otherwise they loop)

• Very little dependence on

features of this geometry, so upgrading to new GDML should be easy

• Sept. 5, 2018 Near Detector MTG

20

A Better Geometry

• From Mike Kordosky. Eldwan Brianne has been working on

ECAL simulation in GArSoft using this GDML:

• GArSoft now can handle non-(0,0,0) locations of the center of

the gas TPC

• Sept. 5, 2018 Near Detector MTG

21

The Channel Grid

(ALICE chamber approximation)

• Sept. 5, 2018 Near Detector MTG

22

292140 x 2 = 584280 channels (c.f. 557568 in ALICE, or about 4.6 extra per pad row) 18 sectors 63 pad rows on the inner chambers, 64 pad rows in the inner outer chambers 32 pad rows in the

• uter outer

chambers

Filling the Hole (new chambers)

• Sept. 5, 2018 Near Detector MTG

23

• Rectangular array of pixels

in a disk

• Pixel size: 6mm x 6mm

c.f. 4 mm x 7.5mm for inner pad rows.

• Central disk has 62060 pads

per detector side.

• Total channels per side is

now 354200. About 18%

• f channels are in the disk.

Software Status

• Simulation is with GEANT4.
• GENIE and single-particle interface
• Data products produced:
• MC Particles (and MC Truth for the neutrino interaction)
• Simulated energy deposits (just newly introduced into LArSoft, in GArSoft from the

beginning)

• Raw Digits (waveforms on each pad) – zero suppressed by default (as ALICE does)
• Hits (in LArSoft terminology) – reconstructed pulses
• Tracks
• Vertices
• Other data products, like flux and trigger simulations have initial placeholders but no

work yet on them.

• 3D Event display shows: MC Particle trajectories, raw digits (thresholded),

hits, hits grouped on tracks.

• Sept. 5, 2018 Near Detector MTG

24

Raw Digits in Green on the 3D event display

• Sept. 5, 2018 Near Detector MTG

25

Uses the channel geomety, raw::RawDigit simulation, and a drift model. I had to cheat the event time however. A 10 !s beam spill window width causes visible displacements in raw digits (not shown here). Visible when overlaying MC Truth, but also visible for cathode-crossing tracks

Hit Finder

• Simple initial hit finder -- Just use the Zero-Suppressed blocks
• f Raw Digits to make hits.
• Threshold applied when making zero-suppressed blocks. 5

ADC counts in simulation. No noise, so no fake hits.

• 5 extra ticks before and 5 after the above-threshold region to get

sub-threshold tails.

• Nearby hits are now clustered as ALICE does to improve

resolution and simplify pattern recognition and fitting.

• Clustering resolution needs to be tested.
• Sept. 5, 2018 Near Detector MTG

26

Track Finder and Fitter

• A "greedy" clustering algorithm:
• Hits are sorted in x, and grouped in clusters in y and z as x is

scanned from either the +x side or the –x side

• Tracks are required to be clustered in both directions, otherwise

make new tracks.

• Initial track parameters estimated from three hits, either on one

end of the track or the other.

• Kalman fit (running) or simple Helix fit (written, needs testing)

available.

• Two-pass fit: Find initial tracks, drop hits that don't fit.

Reassign dropped hits to other tracks and re-fit.

• Sept. 5, 2018 Near Detector MTG

27

Vertex Finding

• We don't know a priori which end of the track is the beginning

and which is the end. (ALICE does, at least for tracks they are interested in)

• Linearly extrapolate tracks whose endpoints are close in space

and find the best-fit point

• To do: Iterate this with a helix instead of just a line. Fitting for

intersection of lines is easy, so linearly extrapolating around the lines tangent to the helix at the vertex candidate should work.

• Track extrapolation methods written. Need error matrix

extrapolation (not written).

• Track charge depends on which end is the beginning. So have

track curvature and slope instead.

• Sept. 5, 2018 Near Detector MTG

28

Example Event

• Sept. 5, 2018 Near Detector MTG

29

neCC GENIE event in the center of the TPC MC truth: Electron: red Protons: purple Neutrons: gray Hits – different colors depending on which track they're assigned to. Blue hits on short proton there, just hard to see. MC Trajectories obscure many of the hits.

Build and Release

• Jenkins build script written so GArSoft builds on SL6, SL7,

macOS 10.12 and macOS 10.13.

• Debug and optimized versions
• gcc and Clang, where available.
• v01_01_01 available on CVMFS.
• Sept. 5, 2018 Near Detector MTG

30

To Do

• Currently working on Analysis Tree module: enable performance

metrics and plot them (efficiency, completeness, resolutions)

• Work on tracking: test helix fit, hit reassignment. Third pass for

very short tracks near vertex will be needed.

• Include ECAL, new geometry (Eldwan, Mike)
• Improve event display
• Understand what to do to Integrate with LBL physics analysis
• If at all at this time?
• Sept. 5, 2018 Near Detector MTG

31

ECAL: Status of geometry

32

• Sept. 5, 2018
• Detector Geometry

– Tool is available (Thanks to Mike K!) in dunendggd to generate the ND geometry (HPgTPC + ECAL)

• Any geometry can be generated
• Baseline with

– 2 mm Cu / 5 mm Scint / 1 mm FR4 – 30 inner ECAL layers – 50 outer ECAL layers – 2 cm PV thickness – 10x10 mm2 cells

• Try to allow for flexible geometry

for ECAL optimisation efforts

• Inconvenient: huge gdml file (144

Mb)

Eldwan Brianne (DESY)

Near Detector MTG

ECAL: Status of simulation

33

• Sept. 5, 2018
• HPgTPC + ECAL is now fully

integrated into GArSoft

– Simulation – Readout + Digitisation – Reconstruction

• Few things still ongoing

– Event Display -> TEve ROOT based – Ganging of hits to allow for different granularity – Segmentation (for now not needed) – Cross-check of the full chain (sim -> reco)

Near Detector MTG

• Not much since Frank Simon’s student defended his master

thesis

• University of Mainz is looking into the effect of using different

absorber thicknesses in the ECAL on the energy resolution and angular resolution

• Once integration into GArSoft is done

– Refocus on ECAL optimisation (reduction of the number of channels, number of layers…) based on Frank’s student results

• Projective (Shashlik) geometry in back section?

– Study influence on energy resolution, angular resolution… and GENIE events – Fix the ECAL geometry

ECAL: Status of detector optimization

34

• Sept. 5, 2018

Near Detector MTG

LBL Physics Analysis

35

• Sept. 5, 2018

Current ND geometry Courtesy of Mike Kordosky Justo Martin-Albo

Near Detector MTG

• The new geometry includes a magnet and ECAL of cylindric

shape, better suited for integration with the HPTPC pressure vessel.

• It includes as well a wider (7 m) LArTPC to improve

containment of muons emitted at very open angles.

• We’ve started a new MC production with this geometry and

we’ll repeat some of the studies done for the previous phase (muon acceptance, HPgTPC event rate…) as a test and cross-check.

LBL Analysis: New MC Production

36

• Sept. 5, 2018

Near Detector MTG

• Work on several areas has already started:

– Optimization of ECAL geometry, including number and thickness of layers, tile (or bar) configuration, number of resulting channels, etc. – Development of task-force-style pseudo-reconstruction for TDR physics studies (similar work already done by Chris

• M. for LArTPC).

– Study of most significant HPgTPC samples for LBL physics. – Investigate neutron detection sensitivity and reconstruction (En from TOF) for full system.

• First results in 4-6 weeks

LBL Analysis: Plans

37

• Sept. 5, 2018

Near Detector MTG

• Thomas Strauss, George Velev and AB met with the

Spokesperson's to discuss a new conceptual design study for the magnet system needed for the DUNE ND.

• Given the large infrastructure costs (upwards of $5M) and

large operating costs ($1.5M/yr), it was agreed that a superconducting option should be explored

• It was also agreed that in parallel, work on conductor design

for a conventional iron-dominated normal-conducting magnet should proceed in order to address some of the shortcomings

• f the existing design and to reach a better understanding of

total power, cooling and infrastructure costs

• 3-4 month time scale

Magnet design

38

• Sept. 5, 2018

Near Detector MTG

• Vladimir Kashikhin had looked into this option about a year

ago and has recently (last 2 weeks) taken another look at the possibilities.

• He first looked at a 3-coil system design.

Magnet design: Superconducting Helmholtz Coil

39

• Sept. 5, 2018

Near Detector MTG

Magnet design: Superconducting Helmholtz Coil

40

• Sept. 5, 2018

The stray field > 100g in all of the Hall (DUNE Prism, not included) Area with the field below 0.1 T (white), center field is 0.5 T

Near Detector MTG

• Vladimir then proposed a design with

bucking coils

–Now universally used in MRI systems –Minimizes stray field while maintaining excellent central field

• In MRI’s central field uniformity

requirement is at the 1 part in 105 level

Magnet design: Helmholtz + bucking coils

41

• Sept. 5, 2018

Near Detector MTG

Magnet design: Helmholtz + bucking coils II

42

• Sept. 5, 2018

Field area below 0.1 T with the active shielding (NO STEEL) Now field reaches 50g at ~ 6m from center

Near Detector MTG

Magnet design: Helmholtz + bucking coils III

43

• Sept. 5, 2018

Add 25 cm thick ½ cylinder of steel

Near Detector MTG

Magnet design: Helmholtz + bucking coils IV

44

• Sept. 5, 2018

Field area below 0.1 T with the active shielding Only small reduction in stray field volume

Near Detector MTG

• These slides represent only a first quick pass at the concept
• Initial simulations indicate that the Helmholtz design with bucking

coils can provide the required central field (or greater) with an acceptable stray field in the hall without the need for any steel.

– Steel requirement would then be driven by muon tag detector requirements.

• Total stored energy in this system is ~ 115MJ
• Cost based on stored energy (Strauss model) is ~ $16M

– Not more than conventional magnet costs, when infrastructure and

• perating costs included

– TJNAF Hall B Toroid system (recently completed) similar in scope. Looking into costs for that.

• Main issue at this moment – large size

– Requires bigger access shaft – Study done for KLOE magnet excellent start (this requirement somewhat bigger, however)

• Just at the very beginning of this design effort

Magnet design: Comments

45

• Sept. 5, 2018

Near Detector MTG

• Excellent progress in many areas on the HPgTPC System
• ver the last couple of months

– Test stand now fully operational – Program for gas studies to study the possibilities for a fluorescence fast-timing tag has been defined – Exploring the extension of the LArPix design to the gas (GArPix) – Significant progress on GArSoft with integration of the baseline ECAL design – Work has begun on integrating the HPgTPC system into the LBL physics analysis – Design study for magnet system has been launched and a SC

• ption looks promising
• Still much to do in order to reach a solid baseline for the CDR

and beyond.

Conclusions

46

• Sept. 5, 2018

Near Detector MTG

THANKS

BACK UPS

48

• Sept. 5, 2018

Near Detector MTG

ALICE TPC Cross-sectional view

49

• Sept. 5, 2018

Near Detector MTG