TPC GAIN STUDIES

Low Energy Reconstruction Efficiency

Sergio Manthey Corchado

1

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Goal

Methodology

Study effect of TPC gain decrease in low-energy event detection.

• Use standard low energy production fcls for HD.
• Use Wirecell and change gain setting and noise file (provided by TPC electronics group) in configuration files. Evaluate waveforms.
• Use SolarNuAna analyzer to evaluate detection efficiency:
• Use recob:hit (gaushit).
• Cluster hits according to time & channel number (max 25 tick & 3 ch).
• Match col + induction (at least 1) planes for y-z cluster reconstruction.
• Group clusters into “primary” and “adjacent” (R adj cl. < 1m & charge adj. < charge primary).
• Compute efficiency: probability of finding at least one cluster per event.

2

DUNE - ELECTRONICS MEETING - 22 OCT 2024

FHICL Settings: dune10kt_1x2x6

• All modification in custom branch (dunereco/tree/gain_dunereco)
• Found gain setting in params.jsonnet file (see figure).
• Gets called from wcls-sim-drift-simchannel-nf-sp.jsonnet.
• Finally configuring standard_detsim_dune10kt_1x2x6.fcl

• Generated new set of *_gain*.jsonnet/fcl files for new workflow (see branch).

3

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Productions

Using dune10kt_1x2x6, vary gain setting 14 (default) and 7.8 [mV/fC].

• Production: Flat elastic scattering neutrino interactions 0-5 MeV homogeneous across detector producing electron tracks (see backup slide).

→ Use alternative detsim config (wcls-sim-drift-simchannel.jsonnet) to export raw waveforms.

→ Follow standard workflow. Reco efficiency from events with reconstructed clusters (see slide)

4

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Raw Waveform Comparison

• Waveform (~5 MeV electron) for gain 14.0 [mV/fC].
• No zoom, y-scale set to waveform amp.

5

signal

Selected wvf with maximum signal in full sample

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Raw Waveform Comparison

• Waveform (~5 MeV electron) for gain 7.8 [mV/fC].
• No zoom, y-scale set to waveform amp.

• Gain changes with 14/7.8 = 1.8 & as expected: Signal amp. 18.5/10.5 = 1.8 and STD 3.4/1.9 = 1.8.
• Because of this, no changes are expected in overall results.

6

signal

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Raw Waveform Comparison

• Waveform (~5 MeV electron) gain 14.0 [mV/fC].
• Zoomed, y-scale set to 40 ADC.

7

signal

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Raw Waveform Comparison

• Waveform (~5 MeV electron) gain 7.8 [mV/fC].
• Zoomed, y-scale set to 40 ADC.

• Results are effectively a noise and signal scaling with the same factor 1.8.

8

signal

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Waveform Max. Amplitude Distributions

• Showing noise wvf. max amp distributions.
• Waveform digitization checks to 1 ADC increments.

9

gain 7.8 [mV/fC]

gain 14 [mV/fC]

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Waveform Max. Amplitude Per Channel

• Noise vs channel reveals a certain pattern. Ref. line showing max value.
• What is the reason for this channel grouping?

10

gain 7.8 [mV/fC]

gain 14 [mV/fC]

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Waveform STD Distributions

• STD distributions replicates noise amplitude results in terms of scaling.
• For 1 MeV electron → amp. of *3.7 - 2.1 & S/N of ~1.4 or ~1.1.

11

gain 7.8 [mV/fC]

gain 14 [mV/fC]

1

2

1

2

*Scaled down from wvfs. in slides 5, 6.

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Waveform STD Distributions

• Clear channel grouping conserved in updated noise model. Same distribution in both productions but different scaling.

12

gain 7.8 [mV/fC]

gain 14 [mV/fC]

Ref. line showing max value.

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Detection Efficiency

• Wirecell correctly adapts to the changes in waveform.
• Both gain configurations present identical detection efficiency.

13

Detection threshold

0.5 MeV

Error bars show counting statistics

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Gain Comparison

• Waveforms look identically scaled (by factor 1.8) in terms of signal amplitude and noise.
• New gain configuration leads to expected changes in signal and reconstruction adapts accordingly.
• Nevertheless, at low energies 1 MeV electron S/N is quite low: 1.4 - 1.1.
• Is this scaling realistic for full scale detector?
• Do we understand noise distributions per channel?

14

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Detection Efficiency

• Showing change in detection eff. with different fiduzializations in drift coordinate.
• No effect on threshold.
• No effect on max. efficiency.
• Only small effect on “slope”.
• Possibility to compare against other LArTPC experiment.

15

Drift length shows no effect on detection thld. of 0.5 Mev

*Electrons are distributed homogeneously across detector

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Detection Efficiency: MicrobooNE Comparison

16

MicrobooNE + Blip

• With same reco. algo. MicrobooNE simulation delivers better efficiency.
• With tuned wirecell Low-threshold config even better results!

MicroBooNE Ambient Radon

• Presented in LowE meeting Blip Reconstruction in DUNE.

DUNE + Blip

3D (2-3 planes) identical to my results in prev. slide

DUNE - ELECTRONICS MEETING - 22 OCT 2024

MicrobooNE Comparison

• Showing nice low energy result from MicrobooNE. Is this possible in DUNE?

17

MicroBooNE Ambient Radon

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Cluster Statistics

• Average number of hits below 3 for e⁻ < 5 MeV.
• Charge behaves linearly with slope ~140 ADC x tick / MeV and 60 ADC x tick thld.

18

Error bars show STD

Error bars show STD

Charge thld. ~60 ADC x tick

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Questions

• Limiting factor for low energy reco.? S/N not enough below 1 MeV?
• How does wirecell calculate the hit threshold?
• Can we review/improve wirecell reco./deconvolution for low energy studies?
• Ready to implement gain change into standard workflow?

19

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Backup

20

DUNE - ELECTRONICS MEETING - 22 OCT 2024

True Electron Spectrum

• Due to the process of elastic scattering, the flat neutrino spectrum leads to an exponential electron sample in energy.
• Distribution favours threshold statistics.
• All values are calculated per bin avoiding spectral bias.
• Errors will be statistical or STD depending on convenience.

21

Sample

Region of interest for threshold evaluation

DUNE - ELECTRONICS MEETING - 22 OCT 2024

True Electron Spectrum

• Due to the process of elastic scattering, the flat neutrino spectrum leads to an exponential electron sample in energy.
• Distribution favours threshold statistics.
• All values are calculated per bin avoiding spectral bias.
• Errors will be statistical or STD depending on convenience.

22

Sample

DUNE - ELECTRONICS MEETING - 22 OCT 2024

True Electron Spectrum

• Due to the process of elastic scattering, the flat neutrino spectrum leads to an exponential electron sample in energy.
• Distribution favours threshold statistics.
• All values are calculated per bin avoiding spectral bias.
• Errors will be statistical or STD depending on convenience.

23

Sample

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Energy Reconstruction: Microboone Comparison

24

See Will Foreman’s Talk in last LowE meeting

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Electron Threshold: Experiment Comparison

Zoom

Blip: MicrobooNE vs DUNE

DUNE: Blip vs SolarNuAna

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Energy Threshold: MicrobooNE Comparison

26

DUNE + SolarNuAna

DUNE + Blip

MicrobooNE + Blip

• Blip: MicrobooNE low energy reconstruction ported to DUNE.
• MicrobooNE simulation delivers better efficiency.

MicroBooNE Ambient Radon

DUNE - ELECTRONICS MEETING - 22 OCT 2024

Energy Resolution: MicrobooNE Comparison

27

DUNE + SolarNuAna

DUNE + Blip

Microboone + Blip

• MicrobooNE implemented lowe changes in wirecell allowing better energy resolution and threshold.

MicroBooNE Ambient Radon

Marley CC prod.

DUNE - ELECTRONICS MEETING - 22 OCT 2024