1. 17JUL13 1 ATLAS Forward Protons: Fast Time-of-Flight Detectors ATLAS Forward Protons: Fast Time-of-Flight Detectors Michael Rijssenbeek – Stony Brook University for the ATLAS Forward Proton group Quartic - the ATLAS Forward Proton Fast ToF Detector see also next talk: Diamond detectors by Gabriele Chiodini (Universita del Salento)

2. AFP Time-of-Flight ATLAS Forward Timing Detectors Collaborating Institutions: Canada: U Alberta (CFD, HPTDC), U Toronto (Detector mounting); France: Saclay (SAMPIC RO chip) Germany: U Giessen (Radiators) Italy: Lecce, Roma2 (Diamond R&D) Portugal: Lisbon U (Trigger) USA: U Texas at Arlington (Detectors, MCP-PMT), Oklahoma State U (RO – Optoboards), U New Mexico (Irradiation), SLAC (Readout), Stony Brook U (Electronics, Readout, RPs) AFP Time-of-Flight Project Leader: Andrew Brandt (UTA) Many Thanks to all my colleagues for fruitful collaboration and help! 17JUL13 2

3. AFP Time-of-Flight AFP – ATLAS Forward Protons AFP measurements: Tag and measure momentum of intact protons from interactions seen in the central ATLAS detector Soft QCD (Diffraction) in special low/medium-luminosity runs – avoid backgrounds from additional interactions in the same BX  μ≃1 cross sections are rather high: many pb’s need clean interactions in ATLAS, i.e. low pile-up – need ~3 weeks-equivalent of data taking at μ≃1 (or ~1 week at μ≃3 ?) at μ>1, require proton time-of-flight measurement to correlate forward protons with interaction vertex measured in central ATLAS detector  σ t =30 ps ⇔ σ z =7 mm Hard Central Diffraction in standard running (μ~50) – huge background from pile-up: 1 proton per side in each BX from soft QCD (Single Diffraction, etc.) pile-up suppression requires precise proton time-of-flight measurement. any increase spatial and temporal granularity improves efficiency and rejection 17JUL13 3 AFP206AFP214 AFP206

4. AFP Time-of-Flight Fast Time-of-Flight Main CEP background: overlap of SD protons with non-diffractive events = ‘pile-up’ background Reduce by: – central mass matching: M central = M AFP = (s ξ Left ξ Right ) ½ – ToF: z vtx = c(t Left – t Right )/2 E.g.: σ t = 10 ps  σ zvtx = 2.1 mm – not a new idea; FP420: 17JUL13 4 σ t =20 ps: 0.1 σ t =10 ps: 0.5 M X >800: 0.05

5. AFP Time-of-Flight Diffractive Protons in AFP Number of protons per 100 fb –1 (~1 LHC yr) per Si pixel (50 μm × 250 μm): – Proton energy loss ξ is related to x : – Central Mass M is related to both protons’ energy losses ξ 1, ξ 2 : 17JUL13 5 ----- detector area (20 mm × 20 mm)

6. AFP Time-of-Flight Hamburg Beam Pipe ATLAS design: Be floor and windows in Al structure Tilted windows (11  ) minimize beam coupling and losses Beryllium windows and floor, and Al structure minimize interactions and multiple scattering Ample space for tracking and timing devices Results of detailed RF simulations: Impedance Z long is at the level of 0.5% /station at 1 mm from the beam Similar for Z trans Power loss (heating) is manageable ~30 W, mostly in conical sections Bellows are not yet included, but we are confident we can minimize their effect 17JUL13 6 ALUMINUM - AUSTENITIC STEEL FLANGEs ALUMINUM BERYLLIUM 450 mm thin

7. AFP Time-of-Flight AFP Roman Pot & Station AFP Pot adaptation from TOTEM design – shown with a possible timing detector … Copy RP Station design of ALFA & TOTEM: – Ample operational experience – Known cost and construction & installation procedures 17JUL13 7 AFP Pot beam AFP timing TOTEM horizontal RP station (beam view)

8. AFP Time-of-Flight Major Development Challenges MCP-PMT Rate and Lifetime: – Have tube capable of 5 MHz and 5 C/cm 2 (equivalent to 50 fb –1 !) – expect further 2-3× improvement HPTDC board capable of 15 MHz 5 ps resolution CFD Clock Distribution Circuit <5 ps All achieved ! 17JUL13 8

9. AFP Time-of-Flight AFP Fast Time-of-Flight QUARTIC concept: Mike Albrow for FP420 (joint ATLAS/ CMS effort) (2004) based on Nagoya Detector. – Initial design (~2006): 4 trains of 8 Q bars: 6mm × 6mm ×100mm – mounted at Cherenkov angle θ Č ≃ 48° – Isochronous – Cherenkov light reaches tube at ~same time for each bar in a train – arrival time of proton is multiply measured: bar + readout resolution less stringent! e.g 30 ps / bar  11 ps for train of 8 bars 2011 DOE Advanced Detector Research award for electronics development: 17JUL13 9 proton Č photons MCP-PMT trains 1 2 3 4 θČθČ SMA pigtails PA-b Programmable Gain AmpCFD Daughter Board HPTDC Board 8-Channel Preamplifier (PA-a) Detector & PMT R&D: U Texas at Arlington (A. Brandt et al.); Electronics R&D: Stony Brook (M.R. et al)

10. AFP Time-of-Flight Electronics Layout Phase 0 Baseline layout (8×8 channels/side):  if the CFD is sufficiently radiation-hard, it can be located at 214 m  if the HPTDC is sufficiently radiation-tolerant, it can be located at 214 m 17JUL13 10 Quartic Feed through (32 ch REDEL-HV) 8× HV 64× Signal 8× Temp 2× Pressure (SMA) (μD) AFP2 crate Signals Att DCS Trigger RR13 ? crate USA15 crates Trigger (LMR600) TDC ToT Opto Board DCS BOC-ROD or RCE CTF DCS iseg HV LV +6V 5A (50 Ω) CFD ToT LE Data LV +6V 20A DCS PA-a PA-b 214 m 240 m 0 m

11. AFP Time-of-Flight Beam Test – FNAL 2012 (A.Brandt, UTA) 30 mm long Quartz bar // beam read by SiPM: σ t ≃ 10 ps for a SiPM (CFD only!) – excellent resolution! – not very radiation hard  2 mm wide × 6 mm deep (in beam direction) Quartz bar positioned at 48° with beam (Cherenkov angle), read by 10 μm pore MCP-MAPMT – single bar: σ t ≃ 20 ps (CFD only!) 4 bars at 48° (~32 mm): expect ~10 ps – single bar with HPTDC: σ t ≃ 26 ps 6 bar train measurement (Test Beam): ~11 ps – rad hard tube (no degradation seen yet up to 5 C) Multiple measurements  ‘tunable’ resolution, size, and interaction length … 17JUL13 11 SiPM 1 – SiPM 3 SiPM 3 – Qbar 3

12. AFP Time-of-Flight MCP-PMT Life Time (A.Brandt, UTA) Historically MCP-PMT’s have not been extremely robust, their performance (QE) degrades from positive ion feedback UTA Formed a collaboration with Arradiance and Photonis for coating … 17JUL13 12 Hamamatsu ion barrier SL10 Arradiance 10 (25)  m pore Planacon 20 ps single bar resolution at 5MHz proton rate (10 pe per proton) at 5E4 gain; x3-5 better with 10  m pore tube Lehman et al. (Panda): As of 5/13 no loss in QE with  Q>5 C/cm 2 ! >10× improvement over typical tube 1C~10 fb -1 expect 3× more with next version

13. AFP Time-of-Flight T958 DAQ FNAL 2012 17JUL13 13 2 3 4 5 6 7 Avg Using a 20-ch, 20 GHz, 40 GS/s (25ps/point) 500k$ LeCroy 9Zi scope! Thanks for the loan LeCroy ! Time difference between SiPM and average of 6 Q-bars: σ t = 20 ps (SiPM: σ t = 14-15 ps) (A. Brandt, UTA)

14. AFP Time-of-Flight Timing System Resolution Reached and extrapolated timing resolution: – Currently at 11-12 ps (Fall 2012 Test beam) with 6 bars; – ultimate performance of this system is probably about 8 ps 17JUL13 14 Component σ t (ps) Current σ t (ps) Projected Action Radiator/MCP-PMT (~10 pe’s with 10 µ pore MCP) 1917Optimize radiator CFD55Larger dynamic range HPTDC18<9New HPTDC chip Reference Clock 33- Total/bar2720 Total/ detector (6 ch)118-

15. AFP Time-of-Flight Time of Flight in Roman Pot Bend the Quartic bars by 90° – disadvantage: loose light in the bend – advantage: extra degree of freedom in projecting the bar onto the MA-PMT ! Note: the Quartic concept is modular; – make ‘trains’ of ‘arbitrary’ length (limited by λ int )  choose σ t – choose granularity: make trains of arbitrary width beam test: 2 mm wide bar has same resolution as a 6 mm wide bar – optimize ToF detector’s size vs. σ t vs. λ int ! Possibility: make the light guide part of the bar into a mirrored air-guide! – reduce amount of material exposed to particles – reduce dispersion compared to quartz 17JUL13 15

16. AFP Time-of-Flight Quartz vs. Air Light Guide … Simulations by Libor Nozka (Prague): run 0: straight Q-bar 150 mm long run 5: Bent Q-Bar 30/120 mm with mirror on elbow (R=90%) run 6: Q-bar 30 mm + bent Air guide 120 mm with mirror on elbow (R=90%) (R = reflectivity of air guide) 17JUL13 16 straight Q-bar 15 cm bent Q-bar 3+12 cm Q-bar 3 cm + bent airguide12 cm ns this design gives σ t ≃ 20 ps in beam test

17. AFP Time-of-Flight Diffractive Protons in AFP Number of protons per 100 fb –1 (~1 LHC yr) per Si pixel (50 μm × 250 μm): – Proton energy loss ξ is related to x : – Central Mass M is related to both protons’ energy losses ξ 1, ξ 2 : 17JUL13 17 detector area (20 mm × 20 mm)

18. AFP Time-of-Flight Efficiency & Backgrounds Royon, Sampert confirm pixellation of ~10 rows is adequate: – inefficiency per train: 17JUL13 18 7 trains: 2, 6×3.25 mm 10 trains of 2 mm width 20 trains of 1 mm width

19. AFP Time-of-Flight New Nuclear Interaction Studies Concerns: – Scattering in first (upstream) station this destroys proton which will neither be tracked nor timed  global inefficiency In case where this proton was kinematically disallowed it might create another proton inside the 2 nd station’s acceptance – Scattering in the thin ‘floor’, spraying ‘sideways’ into the detector inside – At 14 TeV find about λ Int ≃2% per Q-bar (~8 mm of quartz) 15% of events have an interaction by bar 8 These interactions have a high multiplicity – Too many particles in quartz bar (shower) would saturate amps dynamic range about 8-10 O(10s) particles, which would saturate amps and cause that bar (and following) timing to be mismeasured – Time over Threshold functionality allows some recovery … All above influence the timing detector optimization 17JUL13 19 Tom Sykora et al.

20. AFP Time-of-Flight Severe Backgrounds at the LHC Sources: 1.IP: single diffraction pile-up 2.secondary interactions in upstream beam elements 3.Beam Halo Low-μ (special) runs: backgrounds are OK – see: ALFA runs at β* = 90 m, 1 km – OK for the soft diffraction program of AFP High-μ (standard) runs: backgrounds are very high – see: TOTEM standard-optics runs (Joachim Baechler’s talk) evidence that the source is primarily IP and secondary interactions in collimators (1 & 2) – we are analyzing recently recovered ALFA run at β*=0.55 m (15’ run, 2 Mevts) – we are simulating the high-μ environment with β*=0.55 m optics … 17JUL13 20 dominant !

21. Horizontal RP Rate at 14  56-F45-N45-F Rate for 1368 b with beam separation 2 MHz1 MHz3 MHz (incl. showers from N) separation lumi factor 1 / 15.71 / 18.61 / 22.6 Rate for 1368 b without separation 31 MHz19 MHz68 MHz (22.6* 3 MHz) Rate for 1 b without separation 23 kHz14 kHz50 kHz Hits per bx w/o separation 2.01.24.4 (50 kHz/11.2kHz) Beam conditions (fill # 3288): 1.6 x 10 11 p/b E = 4 TeV  * = 0.6 m  n = 2.8  m rad  = 31 (without separation) L = 6.7 x 10 33 expected SD rate per arm within acceptance: ~ 0.4 / bx (event rate / bunch crossing) Revolution frequency: 11.2 kHz average crossing rate : 11.2 * 1368 = 15.3 MHz average interaction rate (without separation) : 15.3 * 31= 47.4 MHz Expected rates after LS1 are different (L, bunch scheme ) insertion at low β* beam heating – LHC vacuum – RP optimization- rates from: Joachim Baechler’s talk of yesterday 17JUL1321

22. AFP Time-of-Flight Summary AFP has a baseline fast timing detector – 10 ps or better resolution for 8 Q-bars – Long-lifetime MCP-PMT – Electronics Optimization in progress: – Needs for μ≃1 physics – Backgrounds and efficiency … – Housing in a Roman Pot ? – Triggering 17JUL13 22

23. AFP Time-of-Flight Backup Slides 17JUL13 23

24. AFP Time-of-Flight AFP – HBP plus Tracker … 17JUL13 24 thin floorsensors evaporative cooling readout flex ATLAS AFP206AFP214 AFP206 AFP

25. AFP Time-of-Flight AFP POT Modifications AFP needs changes in the POT design: – the TOTEM design has a different thin window size, not optimally matched to our acceptance – the TOTEM floor is a groove in the pot bottom: requires a bump-out of the tracking sensor, making it difficult to insert a Quartic detector close to the beam … We have more time than TOTEM  use to investigate improvements – We should make the AFP pot a bit larger (to ~144 mm) by reducing the 2.5 mm gap between the bellows and the pot itself to ~1 mm. Making it even larger than that requires a different ‘Tee’ design and RF calculations will have to be repeated to validate a larger cylinder. Unless absolutely necessary, I would prefer to keep the pot to 144 mm ID. – We should investigate alternative pot and window materials, coatings, etc. e.g., a Be window of 200-400 μm thickness welded to an Al pot (cfr. Daniela) would be a huge improvement over the current TOTEM pot in terms of conductivity, radiation length (MS), and interaction length. Need our own feedthrough plate with the services as we require them, adapting the plate as designed for TOTEM to AFP needs. Possibly the plate for the ‘timing’ will be different from the plate for the ‘tracking’ 17JUL13 25