1. Peter Križan, Ljubljana May 10, 2007 SuperB V Peter Križan University of Ljubljana and J. Stefan Institute PID in the endcap region: proximity focusing RICH with TOF capabilities SuperB V, Paris, May 10, 2007

2. Peter Križan, Ljubljana May 10, 2007 SuperB V Contents Introduction Proof of principle: beam tests Multiple layer radiator, focusing Photon detector, beam and bench tests Read-out electronics Mechanics Cost and material budget estimates Summary

3. Peter Križan, Ljubljana May 10, 2007 SuperB V PID upgrade in the forward region Possibility for the PID in the endcap: proximity focusing RICH as studied for SuperBelle

4. Peter Križan, Ljubljana May 10, 2007 SuperB V Proximity focusing RICH K/  separation at 4 GeV/c  c (  ) ~ 308 mrad ( n = 1.05 )  c (  )–  c (K) ~ 23 mrad  c (meas.) =   ~ 14 mrad, typical value for a 20mm thick radiator and 6mm PMT pad size Separation: [  c (  )–  c (K)]/  track  5  separation with N pe ~10

5. Peter Križan, Ljubljana May 10, 2007 SuperB V Beam test: Cherenkov angle resolution and number of photons Beam test results with 2cm thick aerogel tiles: >4  K/  separation  Number of photons has to be increased.  0 ~ 15mrad N pe ~6 NIM A521(2004)367; NIM A553(2005)58

6. Peter Križan, Ljubljana May 10, 2007 SuperB V PID capability on test beam data From typical values (single photon resolution 15mrad and 6 detected photons) we can estimate the Cherenkov resolution per track: 5.3mrad;  ~4   /K separation at 4GeV/c. Illustration of PID performance: Cherenkov angle distribution for pions at 4GeV/c and 'kaons' (pions at 1.1GeV/c with the same Cherenkov angle as kaons at 4GeV/c).

7. Peter Križan, Ljubljana May 10, 2007 SuperB V PID capability on test beam data 2: identification of low momentum leptons Illustration of PID performance: Cherenkov angle distribution for single photons (peaks would be three times narrower for 10 photons) for a 0.55 GeV/c beam with e, muons and pions.  identification of low momentum leptons 0.55 GeV/c with e veto

8. Peter Križan, Ljubljana May 10, 2007 SuperB V How to increase the number of photons? What is the optimal radiator thickness? Use beam test data on  0 and N pe Minimize the error per track: Optimum is close to 2 cm 00 N pe

9. Peter Križan, Ljubljana May 10, 2007 SuperB V How to increase the number of photons without degrading the resolution? normal Radiator with multiple refractive indices NIM A548 (2005) 383  stack two tiles with different refractive indices: “focusing” configuration

10. Peter Križan, Ljubljana May 10, 2007 SuperB V Such a configuration is only possible with aerogel (a form of Si x O y ) – material with a tunable refractive index between 1.01 and 1.07. Radiator with multiple refractive indices 2

11. Peter Križan, Ljubljana May 10, 2007 SuperB V 4cm aerogel single index 2+2cm aerogel Focusing configuration – data

12. Peter Križan, Ljubljana May 10, 2007 SuperB V ● upstream aerogel: d=11mm, n=1.045 ● downstream layer: vary refractive index ● measured resolution in good agreement with prediction ● a wide minimum allows for some tolerance in aerogel production Single photon resolution refractive index difference Curve: optimisation study NIM A565 (2006) 457 Focusing configuration – vary n 2 -n 1 1.045varied Data points: S. Korpar et al, Pisa meeting 2006.

13. Peter Križan, Ljubljana May 10, 2007 SuperB V Resolution per track Single photon resolution Number of detected photons Multi-layer extensions: data  N pe Multiple layer radiators combined from 5mm and 10mm tiles Cherenkov angle resolution per track: around 4.3 mrad  p/ K separation at 4 GeV: >5 s

14. Peter Križan, Ljubljana May 10, 2007 SuperB V PID capability - MC study MC simulation of the device and reconstruction of data: Assume detector performance as seen in the beam tests Including background sources: Rayleigh scattered photons + backgrounds observed in the beam tests + off-time rates from random triggers at Belle Maximum likelihood analysis of the images on the photon detector

15. Peter Križan, Ljubljana May 10, 2007 SuperB V ● simulation of the test setup with the level of background expected for Super Belle ● 3cm single radiator compared to 2x1.5cm focusing radiator (n~1.05) ● focusing radiator improves PID for momenta above ~3GeV PID capability - MC results

16. Peter Križan, Ljubljana May 10, 2007 SuperB V Focusing vs. defocusing radiator K ID efficiency Dip at 2GeV/c: pion ring from low n coincides with K ring from high n

17. Peter Križan, Ljubljana May 10, 2007 SuperB V Photon detector candidates BURLE 85011 MCP-PMT: ● multi-anode PMT with 2 MCPs ● 25  m pores ● bialkali photocathode ● gain ~ 0.6 x 10 6 ● collection efficiency ~ 60% ● box dimensions ~ 71mm square ● 64(8x8) anode pads ● pitch ~ 6.45mm, gap ~ 0.5mm ● active area fraction ~ 52% Photon detector has to work at 1.5 T!

18. Peter Križan, Ljubljana May 10, 2007 SuperB V ● accumulated rings on MCP-PMT and R5900-M16 PMTs Beam tests: Cherenkov rings Cherenkov photons emitted in the PMT window NIM A567(2006)124

19. Peter Križan, Ljubljana May 10, 2007 SuperB V ●  ~ 13 mrad (single cluster) ● number of clusters per track N ~ 4.5 ●  ~ 6 mrad (per track)  ~ 4  / K separation at 4 GeV/c Increase the number of photons, poss ible improvements : ● bare tubes (52%->63%)  OK ● increase active area fraction (bare tube 63%->85%)  RD ● increase the photo-electron collection efficiency (from 60% at present to 70%)  RD Extrapolation from the present data 4.5 clusters per ring  8.5  per track: 6 mrad  4.5 mrad  >5  /K sep. at 4 GeV  looks OK on paper, see what Burle manages Resolution per track

20. Peter Križan, Ljubljana May 10, 2007 SuperB V uniformity of response, cross talk, timing properties, operation in magnetic field, ageing Bench tests  see my slides from November meeting

21. Peter Križan, Ljubljana May 10, 2007 SuperB V Surface scan with single photons using a narrow time (~10ns) window. Photon detector candidate: SiPM ● immune to magnetic field ● high photon detection efficiency up to 70% ● good timing properties (~300ps FWHM) ● no high voltage ● low material budget ● high noise rate ~ 1MHz/mm2 ● radiation damage - increase of dark count rate 1mm

22. Peter Križan, Ljubljana May 10, 2007 SuperB V Photon detector candidate: SiPM Increase signal to noise ratio by using narrow time (<10ns) window and light guides. MC studies: kaon ID efficiency for different background rates. 1MHz with 10ns time window  1% occupancy

23. Peter Križan, Ljubljana May 10, 2007 SuperB V With a fast photon detector, a proximity focusing RICH counter can be used also as a time-of-flight counter. Time difference between  and K  Cherenkov photons from PMT window Cherenkov photons from aerogel MCP-PMT aerogel track 2GeV/c p /K: D t ~ 180ps 4GeV/c p /K: D t ~ 45ps TOF capability Cherenkov photons from two sources can be used: ● photons emitted in the aerogel radiator ● photons emitted in the PMT window

24. Peter Križan, Ljubljana May 10, 2007 SuperB V Time resolution for Cherenkov photons from the aerogel radiator: 50ps  agrees well with the value from the bench tests Resolution for full ring (~10 photons) would be around 20ps TOF capability: photons from the ring Beam tests: study timing properties of such a counter.

25. Peter Križan, Ljubljana May 10, 2007 SuperB V TOF test with pions and protons at 2 GeV/c. Distance between start counter and MCP-PMT is 65cm Expected number of detected Cherenkov photons emitted in the PMT window (2mm) is ~15 Expected resolution ~35 ps  TOF capability: window photons

26. Peter Križan, Ljubljana May 10, 2007 SuperB V Time-of-flight with photons from the PMT window Cherenkov angle aerogel, n=1.05 Aerogel: kaons (protons) have no signal below 1.6 GeV (3.1 GeV): identification in the veto mode (works OK in Belle!). Benefits: Čerenkov threshold in glass (or quartz) is much lower than in aerogel. Window: threshold for kaons (protons) is at ~0.5 GeV (~0.9 GeV):  positive identification possible. Threshold in the window:  K p

27. Peter Križan, Ljubljana May 10, 2007 SuperB V Read-out electronics Default: binary output. Prefered: excellent time and (rough) amplitude determination (for time walk correction): necessary for TOF, good for background and cross talk handling. Possible mixed solution: binary for each anode, time for the or-ed signal in each tube (or for a part of it). Options (some, in most case RD needed): VA64TAP ASIC developed at TMU/KEK wave sampler based read-out (G. Varner, Hawaii)  More RD needed....

28. Peter Križan, Ljubljana May 10, 2007 SuperB V Tiling of the radiator –Cut into hexagonal shape from a square block –Machining device: use “water-jet”, possible with hydrophobic tiles Minimize photon yield losses at the aerogel tile boundary: hexagonal tiling scheme

29. Peter Križan, Ljubljana May 10, 2007 SuperB V Photon detector tiling 92% of the surface covered by PMTs minimal distance between modules: 0.5~mm max. distance (few mm) allows for feeding in the HV supply cable (has to come to the front side of the HPD) Can also make a more symmetric tiling, e.g. 6 identical sectors.

30. Peter Križan, Ljubljana May 10, 2007 SuperB V Mechanics Donut structure: front wall: support for aerogel (e.g. 0.5-1mm Al), rear wall: photon detector+electronics (~5mm Al) aerogel photon detector + electronics ~110cm ~40cm

31. Peter Križan, Ljubljana May 10, 2007 SuperB V Summary ● A proximity focusing RICH with ~20 cm radiator to photon detector distance and ~6x6mm 2 pads is a very promissing option for the endcap region of a Super B factory. ● Single refractive index radiator has an optimal radiator thickness of ~2cm; increasing the thickness results in a degradation of Cherenkov angle resolution per track. ● A multi layer radiator with varying refractive index in the focusing configuration allows to improve the performance, a ~5 sigma p/K separation up to 4 GeV/c is expected. ● Such a counter can also be used for TOF measurement → extend PID capabilities into low momentum region ● RD issues: read-out electronics, photon detector

32. Peter Križan, Ljubljana May 10, 2007 SuperB V Summary of properties 1 Aim: measure Cherenkov angle and – optionally - also the time of flight. PID coverage: 4-5  /K at 4GeV/c; pion threshold at ~0.6GeV/c, kaon threshold ~1.6GeV/c, proton ~3 GeV/c. For the time-of-flight measurement photons from the PMT window can be used, kaon (proton) threshold at ~0.5 GeV/c (1 GeV/c). Building blocks: radiator, expansion gap, photon detector, read-out electronics, mechanical structure. Radiator: multilayer aerogel in the focusing configuration, total thickness ~4cm (depends on the available space), tiles as large as possible (to minimize the losses at the tile boundaries), preferably hexagonal. Expansion gap: as large as possible, ~20cm

33. Peter Križan, Ljubljana May 10, 2007 SuperB V Summary of properties 2 Photon detector: working candidate Burle MCP PMT with 10micron pores, photocathode to MCP distance=<1mm, bare tube version with reduced margins; RD: aging. Read-out electronics: default binary output. Prefered excellent time and (rough) amplitude determination (for time walk correction). Possible mixed solution: binary for each anode, time for the or-ed signal in each tube (or for a part of it). Mechanical structure: donut shaped, one wall with the radiator plane, one to support the photon detector and electronics. Material budget – rough estimate: 0.35 X 0 Cost – rough estimate: 3MUSD

34. Peter Križan, Ljubljana May 10, 2007 SuperB V Material budget estimate Front support wall (0.5-1mm Al): 0.005-0.01 Aerogel radiator (~4cm): 0.03 Photon detector + electronics: ~200g/36cm 2 ~ 0.25 Rear support wall (~5mm Al): ~0.05 Total: ~ 0.30-0.35 X 0 Reduce material: use carbon structures

35. Peter Križan, Ljubljana May 10, 2007 SuperB V Cost estimate, very rough Aerogel radiator: ~140k USD Photon detector: ~600 tubes x 3k USD = 1.8M USD Electronics: 40k channels x ~20 USD ~ 800k USD Mechanical support structure: 130k USD Total ~3M USD N.B. Estimates for SuperKEKB, has to be reevaluated for the smaller area coverage in SuperB.

36. Peter Križan, Ljubljana May 10, 2007 SuperB V Back-up slides

37. Peter Križan, Ljubljana May 10, 2007 SuperB V READOUT ELECTRONICS: ● signals from anodes are amplified and discriminated by ASD8 boards ● digital signals are converted to ECL levels and fed to VME modules LIGHT SOURCE: ● blue LED (470nm) focused by microscope to ~ 30  m ● 2D position of the light source is computer controlled in steps of 12.5  m ASD8 BOARDS: ● used in the HERA-B RICH ● 16 channels (2 x ASD8 chips) ASD8 = 8 channel amplifier, shaper and discriminator: ● ENC ~ 900 + 70/pF ● shaping time ~ 10ns ● sensitivity ~ 2.5mV/fC Bench test set-up

38. Peter Križan, Ljubljana May 10, 2007 SuperB V - Amplifier: FTA 820 Ortec, rise time <1ns - ADC: dual input range C.A.E.N. V965 - Laser PiLas diode laser, 404 and 636nm optical heads - 6mm bifocal lens, 0.3% filter Bench test set-up 2

39. Peter Križan, Ljubljana May 10, 2007 SuperB V Operation in magnetic field Burle MCP PMT 25 micron pore tube works well up to 0.8T Akatsu et al.NIM A528(2994)763 10 micron tube: good performance at 1.5T (Jerry Va’vra, submitted to NIMA)

40. Peter Križan, Ljubljana May 10, 2007 SuperB V count rates - all channels: ● charge sharing at pad boundaries 2100 V Single photon counting vs position on the tube Slice of the counting rate distribution (single channels - colored, sum of all channels - black) NIM A567(2006)124

41. Peter Križan, Ljubljana May 10, 2007 SuperB V MCP PMT: Cross-talk mechanisms L=6mm

42. Peter Križan, Ljubljana May 10, 2007 SuperB V Single channel hit Coincidence of channels Charge sharing 10micron pore tube, 2x2 channels

43. Peter Križan, Ljubljana May 10, 2007 SuperB V 1 Largest signal on ch. 1, all signals; fraction of hits Largest signal on ch. 1, prompt signals; fraction of hits Largest signal on ch. 1, delayed signals; fraction of hits Signal area extends far beyond the pad boundary due to the long range of the backscattered photoelectrons (delayed signals) Cross talk studies: channel 1 TDC

44. Peter Križan, Ljubljana May 10, 2007 SuperB V Charge sharing and backscattering Ch. 1 hits; larger signal on the neighbouring pad Ch1 hit, all signals; fraction of hits Ch1 hit, prompt signal; fraction of hits Ch1 hit delayed signals; fraction of hits 1

45. Peter Križan, Ljubljana May 10, 2007 SuperB V Single channel response for photon incidence angles of 0 o and 45 o (reflections) logarithmic scale! Yet another cross-talk source: refected photon NIM A567(2006)124

46. Peter Križan, Ljubljana May 10, 2007 SuperB V ● relative intensities of the main peak, first and second reflections are 0.92, 0.07 and 0.01, respectively ● displacement of secondary image consistent with reflection from MCP surface ● impact on spatial resolution (+10% @18 o ) ● impact on timing resolution:  t ~ 40ps 45 o 0.92 0.07 0.01  Internal reflections  x vs.   photon

47. Peter Križan, Ljubljana May 10, 2007 SuperB V MCP PMT timing properties Response: Main component: narrow Gaussian + Tail  Excellent  =65ps t (ps)

48. Peter Križan, Ljubljana May 10, 2007 SuperB V ● pion beam 0.5 GeV/c - 4 GeV/c ● two MWPCs for tracking ● same front end electronics (ASD8) as bench tests ● digital signals read out by VME TDCs ● different aerogel samples used Beam test set-up

49. Peter Križan, Ljubljana May 10, 2007 SuperB V Impact on -> resolution -> # of rec. photons Charge sharing -> expect clusters instead of single hits Number of hits per cluster Hit clustering

50. Peter Križan, Ljubljana May 10, 2007 SuperB V * Poisson # of hits Number of hits # of clusters

51. Peter Križan, Ljubljana May 10, 2007 SuperB V clusters ● N = ~ 1.2 (@ 13%) ● full ring ~ 9 ● full coverage ~ 4.5 (@ 52%) photons (from Poisson zero hit probability P(0)) ● N = ~ 1.8 (@ 13%) ● full ring ~ 14 ● full coverage ~ 7 (@ 52%) Number of hits 2

52. Peter Križan, Ljubljana May 10, 2007 SuperB V MCP-PMT photons (from Poisson) ● N = ~ 1.8 (@ 13%) ● full ring ~ 14 ● full coverage ~ 7 (@ 52%) R5900-M16-PMT photons ● N = ~ 1.95 (@ 11%) ● full ring ~ 17.5 ● full coverage ~ 6.5 (@ 36%) Photons per ring MCP - PMT: 14 R5900- M16: 17.5 consistent with the ratio of collection eff. 60% vs 75% Number of hits 3 Cross-check: comparison with the reference detector

53. Peter Križan, Ljubljana May 10, 2007 SuperB V ● charge sharing at the edges of pads -> resolution improves if using center of gravity of the cluster s J : 17.8 mrad -> 13.1 mrad Resolution

54. Peter Križan, Ljubljana May 10, 2007 SuperB V Photon detector summary I Burle MCP PMT 64 channel tube with: 10  m pores (for 1.5T operation) photocathode to MCP distance=<1mm (reduce cross-talk) bare tube version with reduced margins (to be proven) sufficient lifetime Detector tested in the test beam and on the bench. Cross talk sources identified, found not critical. Possible upgrade: 1024 channel version with a finer granularity Open issue: ageing of the MCP-PMTs (due to ion feedback, can be reduced by a protective foil - ~1/2 photons lost; better option adding a third MCP layer or a better vaccuum)

55. Peter Križan, Ljubljana May 10, 2007 SuperB V Yield losses at tile boundaries Scan with the beam across the tile boundary. As expected, the yield is affected over a few mm in the vicinity of the boundary. A simple model (all photons hitting the boundary get lost) accounts for most of the dependence How to design radiator tiles: check losses at the tile boundary. Reduce the fraction of tracks close to tile boundaries and corners. NIM A553(2005)58

56. Peter Križan, Ljubljana May 10, 2007 SuperB V Effects at the tile boundary Npe Angular resolution 30mm 150mm 120° 150mm reduced to ½ at the boundary. 13.21±0.63mrad

57. Peter Križan, Ljubljana May 10, 2007 SuperB V ● 2+2cm aerogel ● angle 30 o ● 2+2cm aerogel ● angle 20 o Works as well! Focusing configuration - inclined tracks

58. Peter Križan, Ljubljana May 10, 2007 SuperB V Good overlapping down to 0.6 GeV/c ● Matching of indices: done for high momentum tracks (4GeV/c) ● How is the overlapping of rings at lower momenta? Focusing configuration – low momentum

59. Peter Križan, Ljubljana May 10, 2007 SuperB V ● number of detected hits: dual radiator has a clear advantage ● single photon resolution: dual radiator ~same as single (of half the thickness) for the full momentum range Focusing configuration – momentum scan Overlapp optimized at 4GeV/c  OK at low momenta as well

60. Peter Križan, Ljubljana May 10, 2007 SuperB V Optimisation of the two layer configuration Fix: total length L (20-25 cm) attenuation length: 3-4 cm first refractive index: ~1.05 pad size 6 mm,  rest = 7-8 mrad Vary: Thicknesses and the second refractive index Mimimize: NIM A565 (2006) 457

61. Peter Križan, Ljubljana May 10, 2007 SuperB V Optimisation of the two layer configuration Mimimize:  track nn k Only a very weak dependence on k, relative rad. thickness (k=0.5 - same thickness) Vary difference in refractive index  n NIM A565 (2006) 457

62. Peter Križan, Ljubljana May 10, 2007 SuperB V Extension to many layers – expectations 1  2 layers: a big jump in resolution per track 2  3, 4: small steps But: more radiator layers  shallower minimum  of particular importance in the vicinity of threshold, where the number of photons (  thicker radiator) is more important Curves: simple model Points: full optimisation NIM A565 (2006) 457

63. Peter Križan, Ljubljana May 10, 2007 SuperB V Optimisation of the two layer configuration – 2D Mimimize:  track nn k No correllation between the refractive index difference dn and k, relative rad. thickness (k=0.5 - same thickness) NIM A565 (2006) 457

64. Peter Križan, Ljubljana May 10, 2007 SuperB V Extension to many layers – expectations 1  2 layers: a big jump in resolution per track 2  3, 4,.. : small steps number of radiator layers NIM A565 (2006) 457

65. Peter Križan, Ljubljana May 10, 2007 SuperB V Photon detectors for the aerogel RICH Needs: Operation in high magnetic field (1.5T) High efficiency at >350nm Pad size ~5-6mm Candidates: large area HPD of the proximity focusing type MCP PMT (Burle 85011)

66. Peter Križan, Ljubljana May 10, 2007 SuperB V BURLE 85011 MCP-PMT: ● multi-anode PMT with two MCP steps ● 25  m pores ● bialkali photocathode ● gain ~ 0.6 x 10 6 ● collection efficiency ~ 60% ● box dimensions ~ 71mm square ● 64(8x8) anode pads ● pitch ~ 6.45mm, gap ~ 0.5mm ● active area fraction ~ 52% ●  ~13 mrad (single cluster) ● number of clusters per track N ~ 4.5 ●  ~ 6 mrad (per track) ● -> ~ 4  /K separation at 4 GeV/c MCP-PMT multi-anode PMTs ● Tested in combination with multi-anode PMTs ● 10  m pores required for 1.5T ● collection eff. and active area fraction should be improved ● aging study should be carried out Photon detector candidate: MCP-PMT

67. Peter Križan, Ljubljana May 10, 2007 SuperB V Read-out electronics: ASIC under development Basic parameters for the ASIC (Rohm CMOS 0.35μm) –Gain ： 5 [V/pC] –Shaping time ： 0.15 [μs] –VGA ： 1-16 –S/N ： 8 (@2000[e]) –Readout ： pipeline with shift register –Package : 18 channels/chip –Control : LVDS –Power consumption : 5 m W/channel Detailed evaluation is under way. □ 4.93[mm] PreampShaper VGA ComparatorShift register

68. Peter Križan, Ljubljana May 10, 2007 SuperB V VA64TAP is a low-power, low-noise ASIC with 64 channels, each with: ● preamplifier (ENC ~ 500 @ 10 pF) ● amplifier (can be switched off) ● CR-RC shaper (75 ns) ● discriminator with 4-bit trim-DAC ● threshold uniformity: +-200e- ● threshold nominal value: 3000e- ● power: 2.3 mW/ch. ● parallel output ● die size: 5.5mm x 5.4mm Auxiliary chip LS64: logic level adapter, converts current logic (from VA64TAP) into CMOS logic (0V, 2.5V - 5V). Same lateral dimensions, direct channel to channel bonding to VA64TAP VA64TAP: backup read-out electronics