Simulation of the spark rate in a Micromegas detector with Geant4 Sébastien Procureur CEA-Saclay

Motivation Spark rate simulationsCT review, 12/02/2009 S.Procureur “We suggest the following further simulation studies: […] Study the rate of production of highly-ionizing particles (e.g. low energy protons or alpha particles) which might cause sparking; for example, from elastic e-p events or from scattering from nuclear targets. A study of the rate of highly ionizing events and their energy deposition profile in the detector at CLAS12 design luminosity could provide an estimate of dead-time losses due to sparking and could lead to a better overall detector design.” → This requires to first estimate, with a simulation, the measured spark rates From JLab review (May 2009):

Outline Spark rate simulationsCT review, 12/02/2009 S.Procureur Introduction – existing data on sparks Geant4 simulation of a MM in hadron beams (Gemc) - Highly Ionizing Particles (HIPs) production - Deposited energy Spark rate estimation - Normalized and real gains - Systematics (production threshold, integration size) - Comparison with data Predictions and conclusion - Bulk vs non bulk Micromegas - Effect of the drift electrode material - Applicability for CLAS12? (+ 1 slide on the tracking)

Sparks in Micromegas Spark rate simulationsCT review, 12/02/2009 S.Procureur - a spark/discharge is a rapid breakdown appearing in the amplification gap (see Brahim’s talk), usually with hadron beams (see Brahim’s talk), usually with hadron beams - pioneering work: D. Thers et al., NIM A469 (2001), 133 → Spark rate is a power law of the gain → Strong dependence with Z (gas) « A likely explanation is that a discharge is triggered when an energy of a few hundred keV is released in the detector conversion gap resulting in a large number of primary ions » (+ additional studies with α sources… similar conclusions) This behaviour was confirmed by several experiments

Geant4 simulation of a MM (Gemc) Spark rate simulationsCT review, 12/02/2009 S.Procureur Simulation of the detector tested by Thers et al. 400 µm Epoxy 2.45 mm conversion gap (Ar+11%iC 4 H 10 ) 7 µm Copper strips 100 µm amplif. gap 4 µm Nickel grid 15 GeV π → Physics list: QGSC_BERT → Integration of Edep in 300x300 µm² → Production threshold: 300 µm Requires large simulation (rare processes, ~10 -6 )

Secondary particles production Spark rate simulationsCT review, 12/02/2009 S.Procureur Many different types of secondary particles can be produced: → small probability to produce HIPs P [MeV]

Origin of the HIPs Spark rate simulationsCT review, 12/02/2009 S.Procureur Origin of particles leaving at least 0.2 MeV in 300x300 µm² : → 42% from the drift electrode → 22% from the gas → 10% from the micro-mesh → 23% from the strips driftstripsmesh

Deposited energy distribution Spark rate simulationsCT review, 12/02/2009 S.Procureur → Energy above 1 MeV can be deposited! (~2000 times be deposited! (~2000 times more than a MIP) more than a MIP) We will assume that : → the number of primary electrons is proportional to E dep (N p =E dep /w i ) → if a deposited energy of 1 MeV produces a spark at a gain equals to unity (normalization), then a 500 keV deposit will produce a spark with a gain of 2 (normalization), then a 500 keV deposit will produce a spark with a gain of 2

Spark rate estimate Spark rate simulationsCT review, 12/02/2009 S.Procureur → Power law dependence is clearly observed, as in the data observed, as in the data How to relate the « normalized » gain (from simulation) to the real one? 1) Make use of the Raether’s limit (spark if N p xG ~ 2x10 7 ) 2) Relate the spark rate with a α source and deposited energy

Comparison with data Spark rate simulationsCT review, 12/02/2009 S.Procureur 1)Correct order of magnitude 2)Similar slope

Some systematics Spark rate simulationsCT review, 12/02/2009 S.Procureur → Dependence only on the integration size (should be close to transverse diffusion size) diffusion size) Production threshold Integration size

Spark rate predictions Spark rate simulationsCT review, 12/02/2009 S.Procureur → Expect similar spark rate for the 2 types of detectors 1) Bulk (woven, thick mesh) vs non bulk (electroformed, thin grid):

Spark rate predictions Spark rate simulationsCT review, 12/02/2009 S.Procureur → Spark rate ~2 times larger with the Inox drift 2) Aluminized mylar vs woven Inox drift electrode:

CERN tests (150 GeV) Spark rate simulationsCT review, 12/02/2009 S.Procureur → Bulk~non bulk; Inox drift>~ mylar drift (but large dispersion)

1st comparison with simulation Spark rate simulationsCT review, 12/02/2009 S.Procureur → Preliminary comparison (using Raether’s limit) shows nice agreement between measurements and simulation between measurements and simulation

Towards a SR estimate for CLAS12 Spark rate simulationsCT review, 12/02/2009 S.Procureur → Looks suspiciously stable down to 300 MeV down to 300 MeV It would be really useful to experimentally investigate the SR at energies below 2 GeV (JLab? CERN? …)

Conclusion  Better understanding of sparks (HIPs production)  Still doubts on the reliability of the simulation at CLAS12 energies  Simulation can give a correct estimate of the spark rate  Can use it to optimize the detector design (material, mesh segmentation)  The simulation will never replace an experimental measurement! Spark rate simulationsCT review, 12/02/2009 S.Procureur

Central Tracking Effect of the number of Silicons in the tracking  More multiple scattering, but limited effect Spark rate simulationsCT review, 12/02/2009 S.Procureur

Backup Micromegas simulationsJLab review, 05/07/2009 S.Procureur

Reconstruction (Socrat) - Forward Micromegas simulationsJLab review, 05/07/2009 S.Procureur  Comparison between FST and FMT performance (with DC) -Very similar performance  large improvement of vertex determination wrt DC

Reconstruction (Socrat) - Forward Micromegas simulationsJLab review, 05/07/2009 S.Procureur  Comparison between FST and FMT performance (with DC)