1. CBM2002@GSI 1 Resistive Plate Chambers for Time-of-Flight P. Fonte LIP/ISEC Coimbra, Portugal. Compressed Baryonic Matter May 13 – 16, 2002 GSI Darmstadt/Germany

2. CBM2002@GSI 2 Main physical mechanisms Description of several. interesting devices High frequency front-end electronics Small segmented counters Bidimensional position-sensitive single-gap counters Large area Practical difficulties to be addressed for future applications Crosstalk Ageing Tails Background Rate capability Plan of the Presentation

3. CBM2002@GSI 3 Timing RPCs – General features Several (~4) thin (~ 0.3 mm) gas gaps Atmospheric pressure operation Both charge and time readout Offline correction for time-charge correlation. Time-charge correlation 1-2 ns  100 ps Electronics rise time Detector-related (not yet fully understood) Experimental

4. CBM2002@GSI 4 Basic physical mechanisms 2 main questions: How can we reach 50 ps  resolution in a gas counter? How is it possible to reach efficiencies of 75% in a 0.3 mm gas gap? cathode anode i i Detection level independent from the cluster position  exact timing N= average number of visible primary clusters Av. cluster density >> 1.4/0.3 = 4.6

5. CBM2002@GSI 5 Time resolution-principles cathode i i anode [Mangiarotti and Gobbi, 2001] n 0eff =i 0 d/(ev)  Monte Carlo i0i0

6. CBM2002@GSI 6 Time resolution-principles [Mangiarotti and Gobbi, 2001] tt ()KN  v  =  vt  =first Townsend coefficient v = electron’s drift velocity

7. CBM2002@GSI 7 Signal properties 1-Signal shape cannot be changed by any linear system Th 1 2 TDC t 1 t 2 2-  v can be easily measured Experimental Linear system Z(s) 

8. CBM2002@GSI 8 Time resolution-theory vs. measurements Single 0.3mm gap 40 50 60 70 80 90 100 110 120 2.42.62.833.23.4 Applied Voltage (kV) Time resolution (ps  ) Time-charge corrected Uncorrected 0.75/(  v) Good agreement

9. CBM2002@GSI 9 The efficiency problem There should be at least one cluster in the efficient region of the gap d cathode anode i i d* Inefficient region (too small avalanches) Efficient region G min ~10 6 Experimental  Lmin=5/mm  Lmin=1.6/mm

10. CBM2002@GSI 10 F.Sauli CERN 77-09 C 4 H 10 C2F4H2C2F4H2 50607080 CH 4 [Fisher, NIMA238] L(cm -1 ) The efficiency problem-cluster density data [Finck, RPC2001] 0.1 mm gap in C 4 H 10 Experimental Lmin No contribution from cathode (would also provide an effect to explain the t-q correlation) Calculation (HEED) Use this data [Riegler, RPC2001]

11. CBM2002@GSI 11 The efficiency problem (maybe solved) Large values of G 0 must be assumed (much above the Raether limit of 10 8 ). Possible by an extremely strong avalanche saturation effect (space charge effect). d cathode anode i i d* G min ~10 6 G0G0

12. CBM2002@GSI 12 Some hardware

13. CBM2002@GSI 13 Timing RPCs – Small single counters (9 cm 2 ) 33 Resolution of the reference counter [ Fonte 2000]  = 99.5 % for MIPs (75%/gap) -HV 4x0.3 mm gaps AluminumGlass (optimum operating point  1% of discharges)

14. CBM2002@GSI 14 Timing RPCs – Array of 32 small counters  = 88 ± 9 ps  = 97 ± 0.5 % [ Spegel et al, 2000 ] Crosstalk < 1% (not tested for multiple hits)

15. CBM2002@GSI 15 Very high frequency front-end electronics based on commercial chips Amplifier-discriminator-delay module Pre-amplifier based on the INA-51063 chip (HP/Agilent) 2.5 GHz bandwidth 20 db power gain 3 dB noise figure 10 ps  resolution above 100 fC 0 10 20 30 40 50 60 70 101001000 Signal fast charge per channel (fC) Time resolution per channel (  2) (ps) RPC at Threshold=12.5 fC RPC at threshold=25 fC RPC at threshold=50 fC Pulser at threshold=25 fC Realistic tests using chamber-generated signals i=i 0 e st s =8.7 GHz Input signal (experimental) 10 ps  resolution above 100 fC [ Blanco 2000]

16. CBM2002@GSI 16 Double readout 4-gaps glass-metal counter [Finck, Gobbi et al, RPC2001] Edge effects

17. CBM2002@GSI 17 Timing RPCs – Large counter Active area = 10 cm  160 cm = 0.16 m 2 (400 cm 2 /electronic channel) 5 cm 4 timing channels 1,6 m Top viewCross section Ordinary 3 mm “window glass” Copper strips HV [Blanco 2001]

18. CBM2002@GSI 18 Timing RPCs – Large counter Charge distribution Time distribution 0246 0 100 200 300 02468 10 0 1 2 3 Fast charge (pC) Events / 20 fC 0.5 pC   98 % Time resolution essentially independent from electrode size -1000-500-30003005001000 10 0 1 2 3 Time difference (ps) Events/20 ps  = 63.4 ps “300 ps tails”  1.5  fit

19. CBM2002@GSI 19 Timing RPCs – Large counter 40 50 60 70 80 90 100 -80-70-60-50-40-30-20-1001020304050607080 Time resolution (ps  )  = 50 to 75 ps Center of the trigger region along the strips (cm ) 93% 94% 95% 96% 97% 98% 99% 100% -80-70-60-50-40-30-20-1001020304050607080 Time efficiency Strip A Strip B Strips A+B  = 95 to 98 % Center of the trigger region along the strips (cm ) No degradation when the area/channel was doubled to 800 cm 2 /channel [Blanco 2001]

20. CBM2002@GSI 20 Timing RPCs – Large counter [Blanco 2001] Position resolution along the strips -0.2 -0.1 0 0.1 0.2 0.3 -80-70-60-50-40-30-20-1001020304050607080 Center of the trigger region along the strips (cm) Fit residuals (ns) Very good linearity (  1.4 cm)  X = 1.2 cm (0.75% of detector length)

21. CBM2002@GSI 21 Timing RPCs – single gap 2D-position sensitive readout X leftX right Time signal HV XY readout plane Y-strips (on PCB) RC passive network X-strips (deposited on glass) out left out right 10 strips for each coordinate at 4 mm pitch 4 cm 2 mm thick black glass lapped to ~1  m flatness Well carved into the glass (avoid dark currents from the spacer) 300  m thick high  glass disk (corners) metal box (no crosstalk) Precise construction (for small and accurate TOF systems) [Blanco 2002]

22. CBM2002@GSI 22 Charge distributionTime distribution Time resolution of single-gap and 4-gap counters may be similar! -1000-500-30003005001000 10 0 1 2 3 Time difference (ps) Events/10ps  = 66.6 ps (54 ps) 3  Tails=1.9 % 300ps Tails=0.36 % Events=26186 33  1.5  fit Timing RPCs – single gap 02004006008001000 0 500 1000 1500 2000 2500 3000 Fast charge (a.u.) Events/bin 05001000 10 0 2 4  = 75 % ineficiency 2 - 2 =54ps = 2 - 2 =54ps  2 - 2 =54ps = 6767 2 - 40 2 =54ps 

23. CBM2002@GSI 23 Timing RPCs – Single gap Efficiency and time resolution  = 50 to 60 ps  = 75 to 80 % No influence from XY readout

24. CBM2002@GSI 24 Events/mm 2 Y(mm) X (mm) Timing RPCs – single gap Bidimensional position resolution edges  3 mm  resolution  3 mm FWHM (strips=4mm)

25. CBM2002@GSI 25 Timing RPCs – single gap Edge effects? Events/mm 2 Y (mm) X (mm) -1000-50005001000 10 0 1 2 Time difference (ps) Events/4ps Sigma=74.0 ps (62 ps) 3-Sigma Tails=3.0 % 300ps Tails=1.5 % Center Essentially no edge effects Sigma=76.9 ps (66 ps) 3-Sigma Tails=2.9 % 300ps Tails=1.8 % Edge & Corner -1000-50005001000 10 0 1 2 Time difference (ps) Events/4ps

26. CBM2002@GSI 26 Timing RPCs – multilayer 4 layers of single-gap chambers  = 50 ps @  = 99 % 3  Tails=1.5 % 300ps Tails=0.20% Single layer of 4-gap chambers Layer  = 75 %,  = 60 ps +  + 1% K @ 300 ps 3  Tails=1.4 % 300ps Tails=0.0% -1000-500-30003005001000 10 0 1 2 3 Time difference (ps) Events/10ps Events=51215 4  = 32 ps  = 94.9 % -3000300500 0 50 100 Time difference (ps) Events/10ps -3000300500 0 20 40 Events/10ps Time difference (ps) tail-dominated M.C. Experimental

27. CBM2002@GSI 27 Open problems

28. CBM2002@GSI 28 The subtle crosstalk problem 11.21.41.61.82 -10 -8 -6 -4 -2 0 2 4 6 8 10 Time (ns) Current (µA) Threshold Few 100 ps jitter Realistic (measured) current shape Baseline shift due to crosstalk

29. CBM2002@GSI 29 The crosstalk problem – main approaches 1 – Do nothing and hope the problem is small or correctable offline 2 – Accept crosstalk and use many more channels than strictly needed. 3 – Shield channels and try to totally avoid crosstalk None of these approaches has been proven so far (as of Nov 2001) + Robust performance. + Testable in the lab. - Segmented mechanics. + Simple and economic + Being massively implemented (ALICE, STAR) -Possible fragile performance + Good 2D position resolution - Expensive -Effective granularity must be defined by testing - Possible fragile performance.

30. CBM2002@GSI 30 Ageing in streamer mode glass RPCs (BELLE) [Kubo et al, RPC2001] Problem water+freon+streamers  Fluoridric acid  Glass corrosion  Dark current  Ineficiency Freonless gas Freon gas

31. CBM2002@GSI 31 Ageing in avalanche mode glass RPCs? 3 glass cathode and 3 aluminium cathode counters Gas: (85% C 2 H 2 F 4 +10% SF 6 +5%C 4 H 10 ) + 10% rel. humidity Glass cathode Aluminium cathode Dark current (nA) Days Temperature (ºC) Integrated charge (mC  10 8 avalanches) Test in progress

32. CBM2002@GSI 32 Tails -3000300500 0 50 100 Time difference (ps) Events/10ps -3000300500 0 20 40 Events/10ps Time difference (ps) tail-dominated Background Theoretical time distribution has a tail Almost completely compensated by t-q correction Safe cure: redundancy (in a real application counters will be imperfect) Highly ionising background  more streamers  lower rate capability  lower resolution Problem was not studied so far. Severity unknown.

33. CBM2002@GSI 33 Rate capability – Standard RPCs Streamer mode < 300 Hz/cm 2 Avalanche mode < 3000 Hz/cm 2 Typically max. rate corresponds to an efficiency drop of a few percent (survey of recently published results)

34. CBM2002@GSI 34 Rate capability – Special RPCs Drift gap Amplification gap Resistive anode on a metal base  =10 7  cm Wire meshes (50  m wires at 0.5 mm pitch) 15 mm 3.5 mm microRPC [Fonte 1997] [Carlson et al, NSS2001] Si plate  =10 4  cm -HV 0.1-0.5 mm PPAC microRPC PPAC 10 7 Hz/cm 2 Gain 5  10 4 Future GSI experiment

35. CBM2002@GSI 35 Main physical mechanisms – mostly understood except t-q correlation. Several interesting devices have been sucessfully tested with  <100 ps and very small tails or edge effects Small segmented counters. 2D position-sensitive single-gap counters. Large area counters. Practical difficulties to be addressed for future applications Crosstalk – several approaches proposed, none proven. Ageing – tests in progress, no problem so far. If problem: avoid water, freon or glass cathodes. Use chemistry to avoid formation of fluoridric acid. Tails – redundancy should solve any problem if really needed. Background – severity unknown. Rate capability – solutions should exist up to ~10 5 /cm 2. Conclusions