DHCAL Response Simulation with Geant 4 Burak Bilki
DHCAL Response Simulation with Geant 4 Burak Bilki University of Iowa Argonne National Laboratory
The Digital Hadron Calorimeter (DHCAL) Active element Thin Resistive Plate Chambers (RPCs) Glass as resistive plates Single 1. 15 mm thick gas gap Readout 1 x 1 cm 2 pads 1 -bit per pad/channel → digital readout 100 -ns level time-stamping Calorimeter 54 active layers 1 x 1 m 2 planes with each 9, 216 readout channels 3 RPCs (32 x 96 cm 2) per plane Absorber Either Steel or Tungsten (or no absorber) 2
Resistive Plate Chambers (RPCs) Pick-up pads Resistive layer Signal HV Gas Resistive plates How it works 1. 2. 3. 4. Charged particle ionizes gas (Freon+Iso-butane+SF 6) The free e- initiates an avalanche in the gas gap in the electric field Charge induced on pads Charge pulse flows to ASICs for processing (amplifier, threshold, etc. ) and data acquisition 3
1 Data analysis Noise measurement Muons Alignment and Calibration Fe-DHCAL Positrons Pions W-DHCAL Positrons Pions Monte Carlo Simulation Density 3 x 3: Number of neighbors in 3 x 3 pads surrounding the hit 4 Event: Trigger time stamp Čerenkov/muon tagger bits Hit: x, y, z, hit time stamp … … … Nearest neighbor clustering: Combine hits with a common edge in each layer Cluster: x, y, z 4
Monte Carlo Simulation Strategy Experimental set-up Beam (E, particle, x, y, x’, y’) GEANT 4 Measured signal Q distribution Points (E depositions in gas gap: x, y, z) RPC response simulation Hits DATA Hits Comparison With muons – tune the charge spread functions, T, (dcut), and Q 0 With positrons – tune dcut Pions – no additional tuning (absolute prediction of pion response) Parameters Distance cut dcut (within which only 1 avalanche) Charge adjustment Q 0 (if needed) Threshold T (of discriminator) The charge spread function - RPCSim (next slide)
Tuning of dcut Values Use Positron distributions at 8 Ge. V Mean of hit distribution Sigma of hit distribution Density distribution (0÷ 8) Longitudinal profile Measure difference to measured distributions Define a χ2 Tuning Identify smallest χ2 Consider points in pairs (in time order: first point sooner, second point later) Calculate the probability to remove the second point if it is within r of the first point Remove the second point 6
The 4 RPC_sim Versions RPC_sim_ Spread functions Comments 3 R e-ar + (1 -R) e-br To help the tail 4 e-ar Measurement from STAR 5 R e-(r/σ1)^2+ (1 -R) e-(r/σ2)^2 Commonly used 6 1/(a + r 2)3/2 Recently came across RPC_sim_ Slope a Slope b 3 0. 0678 0. 671 4 0. 0843 5 6 Sigma 1 0. 120 0. 0761 Sigma 2 0. 983 R Q 0 dcut T 0. 345 0. 201 0. 262 0. 3645 0. 199 0. 092 0. 286 0. 114 0. 092 0. 250 0. 384 0. 092 0. 3405 0. 241 7
Muon Response Simulation is very successful. Fine details are reproduced with no specific adjustments. x-distribution Well reproduced y-distribution Inter-RPC gaps well reproduced Fishing lines well reproduced Edges well reproduced Possible reasons Top layer gets gas last, is warmer RPCSim 3
Positron Response Simulation 8 Ge. V looks OK at first sight, but has some issues in the details. π+ No tu se d fo r tu ni ng e+ Number of Hits e+ Average Number of Hits / Layer Number of Hits μ+ e+ Density of Hits (3 x 3) Number of Hits / Layer
Positron Response Simulation looks OK at first sight, but has some issues in the details. π+ EM showers not reproduced very well No tu se d fo r tu ni ng e+ 8 Ge. V Number of Hits e+ Average Number of Hits / Layer Number of Hits μ+ e+ Density of Hits (3 x 3) Number of Hits / Layer
Positron Response Simulation 8 Ge. V Problem was reduced to the contribution from photon interactions in the gas gap. No tu se d fo r tu n in g π+ e+ Number of Hits e+ e+ Average Number of Hits / Layer Density of Hits (3 x 3) Added all photons up to 100 ke. V to the list of avalanche starters. Number of Hits / Layer
Positron Response Simulation 8 Ge. V Problem was reduced to the contribution from photon interactions in the gas gap. No tu se d fo r tu n in g π+ e+ Number of Hits e+ e+ Average Number of Hits / Layer Density of Hits (3 x 3) Added all photons up to 100 ke. V to the list of avalanche starters, if they interacted or not. Number of Hits / Layer
RPC Gas in Geant 4 Freon (94. 5%) + Iso-butane (5%) + SF 6 (0. 5%) Default EM package EM_opt 0 EM_opt 1 EM_opt 2 EM_opt 3 EM_opt 4 EM_livermore EM_penelope PAI_photon Photoionization cross-section for Ar (for cross-check) Ar Energy range of useful photons: 1 e. V – 1 ke. V Modified cuts to enable production of photons in the entire range → this helps! G 4 Production. Cuts. Table: : Get. Production. Cuts. Table()->Set. Energy. Range(10. *e. V, 1. *Ge. V);
RPCSim 5 Optimization (Best so far) e+ No tu se d fo r tu n in g π+ 10 Ge. V Number of Hits e+ e+ Average Number of Hits / Layer μ+ Density of Hits (3 x 3) RPCSim 5: R e-(r/σ1)^2+ (1 -R) e-(r/σ2)^2 σ1=0. 07, σ2=0. 7, R=0. 18 T=0. 15, dcut=0. 14, Q 0=0. 8 Number of Hits / Layer
Conclusions and Outlook The muon response is simulated successfully EM shower response still has some issues in different configurations (e. g. November 2011 minimal absorber setup) Try an independent optimization with a completely different dataset - November 2011 (Fermilab no absorbers) and November 2012 (CERN with W absorbers) – and investigate the differences 15
- Slides: 15