1. 1 Calorimeter in G4MICE Berkeley 10 Feb 2005 Rikard Sandström Geneva University

2. 2 Outline Introduction Implementation –Geometry in G4MICE –Hit collection –Digitization Performance –Physics: energy loss, digitized events –e/mu separation Future plans

3. 3 Introduction New code has been written for EMCal in G4MICE by Rikard. The code consists of physical detector model, and digitization and unit test. The model uses latest design. –Dec -04, newer than TRD. Easy to use, ready for use. –But no reconstruction yet. All in G4MICE CVS repository.

4. 4 Geometry in G4MICE Calorimeter modeled on a fiber by fiber basis, grouped in cells. There are four sheets of cells along z. Cells: lead, 4x4x120 cm. 30 cells per layer. Fibers: scintillator, 120 cm long, 1 mm diameter. The default fiber spacing is 1.35 mm, center to center. Fibers are organized in hexagonal pattern. A cell is filled with as many fibers as dimensions allow. All cells parallel. z y 1.35 mm

5. 5 Snapshots

6. 6 Hit collection All fibers are made sensitive. Nothing else is. A G4Step with nonzero energy loss = a hit. All hits are stored and printed to standard G4MICE output file with a lot of information. The cells are numbered given an ID starting with 0 for most upstream layer, lowest cell. Above cell #0 is #1, etc. Next layer starts at #30, etc.

7. 7 Digitization (dE->ADC) All fiber hits in a cell are merged, then processed. Every cell is read out at both ends, producing pairs of digits. ADC amplitude depends on total energy loss in hits, attenuation length and position. –A test is made that verifies product of left-right amplitude is constant. The amplification was set by guessing, aiming for a reasonable value. (Tunable.) –Fluctuation sigma is sqrt. Digitization does not use time information, but everything necessary is there. –Only need to add an incomplete gamma function! –The real integration time is tunable (100 to 200 ns), and a goal of simulations could be to optimize it. The stored digits contain Monte Carlo truth information for comparisons. xADC[L]ADC[R]

8. 8 Reconstruction No code written yet, but ideas are: –Cell number gives y & z. –Comparing left-right pmt signal gives x. –Time (gate) gives track separation. –ADC comparison between layers gives PID.

9. 9 Performance – energy loss & ADC Reminder: –A hit is a step in a fiber with energy loss. (Any particle.) –A digit is the merged and digitized hits, one for each side of cell. Beams with e- and mu+ were shot onto EMCal. Simulations were using flat kinetic energy distribution from 100 to 250 MeV.

10. 10 Energy loss per hit, mu+ beam

11. 11 ADC, mu+ beam

12. 12 2 nd s to mu+ beam Energy loss, per hitADC

13. 13 Energy loss per hit, e- beam

14. 14 ADC, e- beam Primaries E range

15. 15 2 nd s e- beam

16. 16 Total energy loss, for each layer mu+ beam e- beam

17. 17 ADC per cell, mu+ beam

18. 18 ADC per cell, e- beam

19. 19 ADC per cell, e- beam (primaries)

20. 20 Energy loss, e- beam (primaries)

21. 21 Adding a 5 th layer, ADC(cell)

22. 22 Adding a 5 th layer The top of simulated energy spectrum is 250 MeV. Should lose all E after 5 layers of lead = 20 cm. But, fibers < dE/dx. Plot includes secondaries but is dominated by primary mu+. Real beam has lower energy -> lower population in layer 4 & 5.

23. 23 Performance – e/mu separation Previous results from Alessandra Tonazzo showed electron-muon separation capability. Trying to do the same. A few problems to be solved –Barycenter calculation requires event by event information. –So does dE 1 vs dE 1 /dE tot. –Information is there in G4MICE, but not easily accessible. –We already have a new interface design, which will simplify things.

24. 24 Future plans Simulation & detector modeling –Good shape. –Run with “real” mu+ beam. Decay? Digitization –Add final component to time response function. –Tune parameters & optimize design? –Make e/mu-separation plots when possible. Reconstruction –No code yet. Global reconstruction & PID –When EMCal recon ready, will provide global G4MICE recon with reconstructed data.