A compact Silicon-Tungsten-Scintillator ECAL?

1. A compact Silicon-Tungsten-Scintillator ECAL? Graham W. Wilson Univ. of Kansas Cornell Workshop, July 13th 2003

2. Outline • Introduction • (Parametric Energy Flow Study -> Tuesday) • ECAL Design issues : Simulation studies of : i) Light collection uniformity in tile-fiber ii) EM energy resolution • Strawman ECAL design • Hardware activities

3. Big picture • Electromagnetic calorimetry critical for separating photon-hadron separation in p-flow jet energy measurement • ECAL cost likely to be a major issue • Need to justify various options in terms of performance and cost • Explore alternatives • If B R2/RM is the FoM, a compact design (small RM) which we can afford to place at large B R2 is interesting

4. ECAL requirements • Reasonable performance on single particle figures of merit : • Energy resolution :  10%/E  1% • Angular resolution :  1 mrad • Hermetic – measurement of missing energy very important. • Contribute to excellent jet energy resolution. Aim for 30%/ E. • Essential to separate photons from interacting charged hadrons • Higher B, higher R2, smaller Moliere radius (transverse size) • So large volume, compact and dense • Timing resolution. • Bunch crossings potentially every 1.4 ns. • Time resolution of 300 ps for photons helps • Affordable.

5. Tungsten-Silicon ECAL Sc-W-Sc-W-Si-W-Sc-W-Sc-W • Proposals exist for W-Si ECAL. • The TESLA design, R=1.7m, 40 layers of Silicon pads looks as if it can do the job (except maybe the timing), but is costed at 133 M$ (driven by 3$/cm2 Si cost) • Eres = 11%/E , Moliere radius = 16.5 mm • Our proposal centers on developing a cost optimized ECAL, with similar performance. This hybrid calorimeter would use Silicon sensors to do the fine pattern recognition and position measurement, plastic scintillators for fine sampling and timing. • TILE-HCAL aim 10$/tile  0.4 $/cm2 (25 cm2area) • 1.0 $/cm2 (10 cm2area)

6. Tile/fiber technique Use blue scintillator. Embed wavelength shifting fiber which absorbs blue scintillation light, re-emits in the green. (Y11 absorption length : 80 mm) Min. bending f : 100  fiber f Several fibers can be seen by one photo-detector (CMS photo)

7.  Recent developments • Excellent light-yield for tile-fiber reported by CALICE TILE-HCAL (V Korbel, Jeju 02) • 25 pe/mip with 5mm Scintillator, QE15% PMT • Higher QE devices (eg. APD) promise QE75% and ease of use in B-field • up to 20 pe/mip/mm looks possible • Elegant and cheap “on-tile” optical readout possibilities – Silicon PM’s => maybe 10$/tile cost Recent technique refinements suggest : I) can detect mips in each layer (if necessary) II) scintillator sampling is much more likely to be significantly cheaper than silicon sampling III) A more compact ECAL with scintillator sampling can be envisaged

8. Amsterdam workshop • Discussions with Stan Bentvelsen (NIKHEF) • Picked up a OPAL tile-endcap complete tile assembly • Stan helped get us started looking at GEANT4 simulation of tile response (he did similar GEANT3 studies for the OPAL design)

9. WIRED display of one photon rattling around in a perfect tile. GEANT4 study of tile-fiber response with Eric Benavidez Uses optical photon processes – exampleN06. Includes scintillation, Cerenkov radiation, Rayleigh scattering, absorption, Fresnel refraction, TI reflection. Data runs saved with AIDA, analyzed in JAS3, looking at light yield, uniformity, timing, spectral characteristics. Shopping list : Real data !, validate implementation (eg. Reflectivity,WLS properties). Anyone tried something similar? Study is making progress – but results not yet mature.

10. ECAL Design Studies • This spring – had looked at conceptual design issues of sampling frequency, sampling thickness and compactness in the framework of estimates based on parametrisations available in the literature expected to be good to around 10%. • Eg. Eres = 2.7% [tactive(mm)/fsamp] (Wigmans p190) (labelled WPAR in next plots) • Now using GEANT4 to repeat those studies and investigate actual hybrid geometries. GEANT4 results sensitive to range cut. • Despite “EM showers are understood mantra”, is there really good data in the literature which can be used to test/benchmark GEANT4 sampling ECAL results ?

11. ECAL Design Issues : Sampling Frequency WPAR WPAR Dense medium like Si has high sampling fraction • Issues : cost of many layers of active medium • Cost of thin sheets of absorber.

12. ECAL Design Study : Sampling Thickness WPAR WPAR • For thin scintillators, need enough photo-electron statistics. (Kawagoe et al, 3.2pe for 1mm scintillator) 1.5 pe/mip/mm looks like a sensible target for thin tiles

13. Compactness Upper curves, 1mm gap TESLA TDR Lower curves, no gap Need to minimise gaps, reduce space needed for fiber routing, by sharing fiber routing gaps among layers Assume 25% of scintillator thickness used for readout

14. New studies with Eric Benavidez (freshman) • All calorimeter designs have 30 X0 of W (105 mm) to ensure adequate longitudinal containment and a fair comparison. • GEANT4 studies are done primarily with 1 GeV photons (which are definitely longitudinally contained)

15. Scintillator-Tungsten Calorimeter GEANT4 study based on TestEm3 example. 3 curves correspond to range cuts of 100, 10, 1 mm (from top to bottom) 1.4mm W plates (75 layers) 1 GeV photon NB differs substantially from WPAR estimate Sampling contribution only

16. Comparison (75 layers) Focus just on the lowest line in both GEANT4 WPAR NB WPAR studies suggested E-res indep. of thickness. Not with G4.

17. Range cuts • 1 mm range cut is sufficient to estimate resolution with 2-3% accuracy. This corresponds to cutoffs of 1.6 and 8.9 keV for g and e in Tungsten • However, the EM response still requires basically no cut (0.1 mm setting) • (Sc/W 1.5mm/1.4mm), 1 GeV photon • 100 mm: 47.8 MeV, 11.6% • 10 mm: 50.5 MeV, 10.7% • 1 mm: 55.7 MeV, 10.4% • 0.1 mm: 63.5 MeV, 10.2%

18. E-resolution dependence on P.E. statistics GEANT4 (1mm range cut) 1.4mm W plates 1.4mm W plates Total: sampling  PE stats (2.5, 5, 10, 20 pe/mip/mm) PE stats contribution: 2.5, 5.0, 10.0, 20.0 pe/mip/mm Sampling contribution Lower curve : sampling only PE stats contribution: 2.5, 5.0, 10.0, 20.0 pe/mip/mm N.B. Suppressed zero For 5 pe/mip/mm, at t=1.5mm degrades from 10.5% to 10.9%

19. Three strawman hybrid designs Studied with GEANT4 (range cut = 1 mm) A super-layer (SL) Sc-W-Sc-W-Si-W-Sc-W-Sc-W HY75 (15 SL) HY135 (15 SL) HY42 (14 SL) In each case the Si layer is chosen as 400 mm Si + 2.0 mm G10 as in SDMar01 detector So far: study uniform sampling structures.

20. 1.4mm W plates 15 layers Si 60 layers of 1.5mm Scint. 4 layers ganged (30 pe/mip) ( if 5pe/mip/mm) 15 super-layers each with mip-detection 0.778mm W plates 15 layers Si 120 layers of 1mm Sc 8 layers ganged (40 pe/mip) 15 super-layers HY75 HY135 HY42 • 2.5 mm W plates • 14 layers Si • 28 layers of 2 mm Sc • 2 layers ganged (20 pe/mip) • 14 super-layers E res : 10.6%/E 7.9%/E 14.5%/ E Moliere radius : 19.3 mm 21.4 mm 16.5 mm 33% of the Silicon cost (SDMar01 30X0 W : 15.4%/E, 15.5 mm with 0.1 mm range cut)

21. Complementarity - Correlations HY135 : 7.9% E-resolution Correlation improves resolution slightly – compensates for PE stats Correlation coefficient = -0.19

22. Linearity check • Check HY75 with 50 GeV and 1 GeV • Study done with 100 mm range cut (need to recheck) • EM response consistent : Scint. Fraction : 3.78% (50), 3.77% (1) • Sc/Si response consistent: 1.861 (50), 1.865 (1) • E-res consistent : 12.1%/E (50), 12.0%/E (1) • Corr. Coeff. : -0.14 (50), -0.13 (1)

23. Possible photo-detectors Hamamatsu, multi-anode PMT, with 18mm18mm sensitive area. 16-channels has 164.5 4.5mm (16 1mm  fibers) 64-channels has 64  2.25  2.25 mm (4 1mm  fibers) Makes optical summing of several layers easy B-field sensitivity and modest Q.E. – probably not the real detector solution.

24. Hardware plans • Planning VME-based DAQ for cosmic and source tests, particularly for uniformity and light yield studies. • Basics ordered – including ADCs, discriminators and TDCs. • Start getting experience with tile-fiber technique and various photo-detectors.

25. Potential Collaborators • Kansas State: E. von Toerne, T. Bolton • Plan to collaborate • Colorado: Uriel et al • Lots in common – talking to each other • Padova, INFN : P. Checchia et al. • Discussed in Amsterdam – some commonalities • CALICE : JC Brient. Thin Sc/W/Sc elements conforming to alveolus structure can be accommodated in Si-W test beam • US Si-W groups: Oregon, SLAC • TILE-HCAL groups : DESY, Russia, NIU • Japanese ECAL : in contact with Kawagoe, Fujii, Takeshita

26. R&D issues • Calorimeter design optimization. • No. of layers, R, absorber thickness, detector thickness, sampling frequency vs depth, transverse granularity of Si/Sc, tile shapes, groove patterns, gap sizes. Should Si-layer be independent of scintillator layers ? Fiber routing. Timing resolution. • Studying with full shower simulation. • Using optics simulation in tile-fiber design and testing studies. • Demonstrating basic performance characteristics • Light yield for thin scintillating tiles • Response uniformity • Scintillator/WLS/Photo-detector for timing • Fiber routing for compact calorimeter • Sound mechanical design • Good quality thin absorber plates (sintering is cheap ..) • Photo-detector characteristics Getting started with lab test-stand with cosmic rays and sources. Good for involving students.