Understanding Ageing of Gaseous Detectors: Science or Magic?

1. MPGD ageing Fred Hartjes NIKHEF Magic or science? 2nd RD51 collaboration meeting Paris, October 13 - 15, 2008

2. What is ageing of a gaseous detector? • Loss of avalanche gain • Rapid or slower • Broadening amplitude spectrum • => More variation in gas gain • Increased sparking tendency • => Damage on electrodes • Secondary emission from cathode • Increased dark current • After pulsing • Self-sustained discharge J. Va'vra, NIM A387(1997)183 F.Hartjes, MSGC damage

3. Loss of gain in gaseous detectors • Figure of merit: accumulated charge on the anode surface • Kadyk (I. Juricic, J.A. Kadyk, Proceedings Workshop on Radiation Damage to Wire Chambers, Berkeley 1986, p. 141) where • Q is the accumulated charge per cm anode (wires or strips) or cm2 (MPGD, PPC, ….) • G is the gas gain • D is the dose (particles per cm resp cm2 • ne is the primary ionisation per hit • Define the ageing rate R (%/C/cm) or (%/C/cm2) as where • A is the average magnitude of the charge signal • Example for wire chambers • R = 1 %/ C/cm => excellent • R = 200 %/C/cm => bad

4. Competition: ageing of silicon sensors • Figure of merit: neq dose • Damage from applied radiation converted into damage from radiation caused by 1 MeV neutrons • => easy evaluation of radiation hardness • Often using neutrons from nuclear plant for ageing characterisation • Nature of silicon sensor damage • Increase of depletion voltage • Presently “solved” using oxygenated silicon (RD50 activities) • Decrease resistivity • Rock hard phenomenon • => go to low temperature (-30 C) • Decrease of the mean free path • Rock hard phenomenon • => practical limit for regular oxygenated silicon sensors 3*1015 neq / cm2 (Not speaking about 3D silicon, diamond, …..)

5. Do gaseous detectors have better ageing prospects than silicon? • Not possible to give an absolute comparison • Too many different sensor properties involved • Response to uncharged particles (neutrons) • Gas gain • geometry • Required ageing performance of MPGDs to be competitive with silicon • Gas gain by Micromegas or GEM • Assume thin drift space • MIP irradiation ne = 30/hit • Gas gain = 2000 • 3*1015 neq/cm2 ≈ 6*1015MIPs/cm2 • => accumulated charge 29 C/cm2(GEM, Micromegas) • => 0.2 C/cm (wire chambers, integrated anode surface) Conclusion:Gaseous pixel detectors can be more radhard than silicon, but we have to be careful

6. Obtained so far for GEM and Micromegas GEM M. Alfonsi et al, Nucl. Instr. and Meth. A518(2004)106 Micromegas (Nikhef measurement)

7. Micromegas ageing Normalized unit • Mesh current Ar/CF4/Iso (95:3:2) 16,1 C / cm² ~ 20 LHC years David Attié, MPGD workshop CERN Sept. 2007 Time (d)

8. Loss of gain: rapid 0.3 C/cm2 • Rapid ageing is generally caused by the formation of a polymer on the anode surface • Catalysed by pollutants • mC/cm range for wires/strips • May be removed by etchants • CF4, O2, H2O • Possible polymer reaction • C2H4 → 2CH2: • CH2: extremely reactive radical, can easily build polymer chains • Studied by plasma physicists Nikhef measurement The most reactive fragment is assumed to be CH2:

9. Polymerization in a plasma • Extensively studied by plasma physicists • But I find no explanation for the rapid ageing catalyzed by minor traces of pollutants H. Yasuda, Desy Ageing workshop 2001

10. Example of rapid ageing: MSGC • Gas: DME/CO2 60/40 • Low dose applied (0.5 mC/cm) • Result: anode strip covered by a thick transparent wax-like layer • (Scratch made on purpose for better visibility) • => big decrease in gas gain Irradiated (0.5 mC/cm) Non-irradiated 100 µm

11. Preventing/ curing rapid ageing • Very clean gas system • Do not use outgassing materials in the chamber • Use CF4 • (controversial) • Use DME • (controversial) • Add some water (who knows) • Few percent O2 • May help in certain occasions • Is this all sufficient? NO

12. Ar/CF4/CH4 Ar/CF4/CO2 CF4, a controversial gas • CF4 + CH4 induces polymerisation; while • CF4 + CO2 removes polymerisation • In general CF4 + oxygen containing mols helps • Polymerisation also observed in low pressure plasma discharge of pure CF4 Winfred and Eva Stoffels, TU Eindhoven, Netherlands (http://www.phys.tue.nl/EPG/epghome/papers/1999/lunter1.ppt) • High primary ionisation (51 clusters/cm, only DME and isobutane are better) • Non-flammable, Fast drifting, Low diffusion but • Aggressive with water • Formation of HF • Used in silicon photolithography as plasma etchant • => May damage chamber materials, electrodes • May also clean electrodes (with water) • Highly electronegative at fields (>4 kV/cm) (unpublished private experience) • => “eats” part of the primary ionisation in the transition between drift area and amplification area Mar Capeans, MPGD workshop CERN Sept. 2007 A. Romaniuk et al, Nucl. Instr. and Meth. A515(2003)166

13. CF4 + water • Water (0.14%) is needed to avoid ageing • Hermes wire chambers S. Belostotski et al, Nucl. Instr. and Meth. A591 (2008) 353

14. DME, another controversial gas • Best quencher of all commonly used HC • Good absorber for 6.5 eV photons •  good cure for rapid after-pulsing • Excellent ageing properties • R = 0.7 +/- 0.3 %/C/cm V. Blinov, Desy ageing workshop 2001 • Very low diffusion • High primary ionization but • Highly flammable • Chemically aggressive • Attaches most plastics • Even Kapton • Not delivered in high purity grade Va’vra, Desy ageing workshop 2001 M. Capeans, Desy ageing workshop 2001

15. Water • Mostly not too good • With CF4 is may cure ageing • But often makes ageing worse • May stabilize HV behaviour by adding conductivity to insulating surfaces • Often hard to bring down to the ppM level • high diffusion through most plastics Discharge probability • May also enhance discharges • Much more discharges with 100 ppM than with 10 ppM M. Hildebrandt, Desy Ageing workshop 2001

16. Inorganic deposits • Inorganic deposits often consist of silicon or carbon whiskers • Normally in the higher dose range (100 mC/cm or more) • Induce drop in gain • Broaden charge signal spectrum • Often induce sparking or dark current • Often originate from very low concentration pollutants • ppb level • Bubbler with silicon oil downstream • A few silicon containing gases exist: SiH4, CSiH6, C2SiH8 • The mechanism of creation is mostly unknown • Not easy or impossible to cure M. Binkley et al, Nucl. Instr. and Meth.A515(2003)53

17. Dark current • Secondary emission from cathode • Malter effect L.Malter, Phys. Rev. 50 (1936) • Electrons drawn from a metal surface by a high field from ions attached to an insulating layer on the surface • Oily residues • Polymer from ageing process • Skin of conductive glue • Dark currents may also be induced by • Sharp points on cathode or anode • Small cathode surface • Basically adds up to anode ageing • Avoiding Malter by applying large equally charged cathode surfaces • circular drift tube  good • single cathode wire in wire chambers  bad Malter effect Insulating film + - + - - + - + cathode + - - + + - + -

18. Material selection • Choose low-out gassing materials from • Database from CERN gasgroup/RD10 (piping + elastomers) http://detector-gas-systems.web.cern.ch/detector-gas-systems/HomePage/homePage.htm • Nasa database (very extensive) http://outgassing.nasa.gov/ • Why difference between Araldite 103 and 106?? Low Outgassing room-T epoxies Outgassing room-T epoxies

19. ANODE STRIP CATHODE STRIPS Field geometry of most common gaseous detectors MSGC: dipole amplification field Very high field at cathode edge Wire chamber: 1/R amplification field GEM: amplification field across ~ 25 µm (high at the edges of the hole) Anode NOT close to avalanche A.Oed, Nucl. Instr. and Meth. A263(1988)351 Micromegas: homogeneous amplification field across 50 µm Y. Giomataris et al, Nucl. Instr. and Meth. A376(1996)239 50 µm F. Sauli, Nucl. Instr. and Methods A386(1997)531

20. Dependence on detector technology • Polymerisation will be mainly at the end of the avalanche where the electron density is highest • A few µm away from the anode • Exception: GEM • Key issue • Whta is the field at the anode surface? • High field => high avalanche temperature • => more dissociation organic molecules • => more sensitive to ageing • How big is the anode surface near avalanche? • MSGC: very small (edge of anode strip) • Wire chamber: quite small • Micromegas: large • GEM: avalanche not in vicinity of anode • => GEM and Micromegas less vulnerable for ageing MSGC ageing: In the µC/cm range

21. Interesting example of wire chamber ageing:Production of LHCB straw tracker • Tracker from boxes filled with straws (Ar/CO2 70/30) • Uniformity of response automatically scanned with a 90Sr source across the full surface • Radhard test during production • Scan • Single point irradiation with a 2 mCi 90Sr source (20 h) • accumulated charge 2.8 mC/cm (peak value) • Verification scan Irradiation profile across the straws Ref:, Ageing in the LHCb Outer Tracker Niels Tuning (Nikhef) IEEE NSS (N48-3) Nov. 1, 2007

22. Result 2nd scan /1st scan • At accumulated charge 2.8 mC/cm (peak value) • Strong unexpected ageing effect • No ageing downstream • At prototype tests no ageing observed • Until 3 C/cm Gas flow Accumulated 3 C/cm in 120 days No effects seen

23. Intensity dependence of ageing Deposit on the surface? • Strongest ageing at moderate intensity • Not much ageing at the highest intensity • => not proportional to accumulated charge • Highest ageing at ~ 0.2 mC/cm • Deposit on wire surface visible

24. Cause of the LHCB ageing phenomenon • Analysis of wire deposit • Carbon, no silicon • Culprit: fresh glue • Araldite AW103 + HY991 • On the green CERN material list Analysis of wire deposit Carbon

25.  Relative Gain  Irradiation Time (hours) 1st HV training 2nd HV training 3rd HV training Possible curing method: HV training • HV training • 24h @ 10 µA/wire • After training plateau occurs of ~ 30 mC/cm • Comparable effect seen by Blinov in DME • Plateau ~ 300 mC/cm • => plateau caused by plasma cleaning of anode surface? V. Blinov, I. N. Popkov, A. N. Yushkov,Aging measurements in wire chambers, Nucl. Instr. and Meth. A515(2003)95

26. Possible curing method: outgassing • Outgassing • Heating up for 2 weeks @ 45ºC or • Flushing for ~6 months • Preventing ageing by O3 • 2.5% O2 added • => O3 formed in the avalanche • Less ageing with low flow • Gas cleaned by avalanche  Gain loss (%)  Flushing Time (weeks)

27. Comparing MPGD ageing to silicon sensor ageing • Gaseous detectors might be much more radhard but • Silicon sensor ageing • Can be well characterised in lab conditions • Results relatively easy to scale up to a big tracker setup • Testing samples of the final production batch at various particles • Lab measurements give a good prediction • Gaseous detector ageing • Characterising gaseous detector ageing by the ageing rate parameter R (%/C/cm) alone is a too simple approach • Lab characterisations at different sites often conflict •  Too many parameters involved • Particle type • Gas purity • Cleanliness of the gas system • Irradiation rate (we cannot cope with every rate) • Irradiated surface • ............ • there is always a risk when running a big tracker system with gaseous detectors • Gaseous detector ageing might suddenly occur and proceed very fast • Ageing tests with Micromegas and GEM have not reached the silicon limit yet •  many ageing studies to be done

28. General recommendations • For a big experiment it is hard to exactly reproduce the laboratory conditions that were used for the ageing tests • Be very alert on unexpected ageing phenomena • Experience on ageing from other groups is helpful, but don’t take it too absolute • Everybody has its own ageing experience, they don’t reproduce well for other sites and with other detectors • Take notice of the experience of plasma physicists • But I haven’t seen an explanation for ageing caused by extremely low pollutant levels in gaseous detectors

29. Design recommendations • Reduce field on cathode surface as much as possible • Use the cleanest materials you can afford (NASA and CERN database) • But there’s no need getting bankrupt • Add filter at the in coming gas close to the detector (molecular sieve 5A) • Consider adding special ageing chamber for advance cleaning (see LHCB experience) • But don’t expect this to be absolutely safe to prevent ageing • Do as much ageing prototype tests as you can on as much different conditions to get an impression of the robustness of your detector • Different particles • Different irradiation rates • Different sites

30. Operational recommendations • While running, monitor the chamber performance on a daily basis and take immediate action when observing ageing phenomena • Don’t change from gas supplier while running an experiment • Be prepared to apply additives • CF4 + oxygen containing molecule like CO2 or alcohols • Water (active moisture control and monitoring), don’t let it pass the 1% limit • => But be aware that these measures might worsen the situation in your specific case

31. A last word of encouragement • After a lot of testing you will get to know the characteristics of your own system quite well • You will learn how to operate it safely and to avoid unexpected ageing

32. Studying MPGD ageing? Magic in a scientific environment

33. Spare

34. Possible ionisation sources for ageing studies • PS beam at CERN  Best radiation test  Mip ionisation profile (24 GeV/c p) Average rate 3 - 9 GHz/cm2  but current during spill order of magnitude higher (> 50 GHz/cm2) • Can Gossip handle this??  Running a testbeam experiment is time and money consuming  Tight time schedules • Powerful (5 GBq) 90Sr source at Nikhef Up to ~ 1.5 GHz/cm2 continuously  Rate still OK for Gossip  Mip ionisation profile (1 – 2 MeV/c e-) • simulating b-layer environment  Always available  Not ultimate radiation test • UV light source at Nikhef (Harry van der Graaf)  Continuous operation  Easy in use (no personal danger)  Different ionisation profile • individual photoelectrons mainly liberated from metal surfaces • Reference: pixel b-layer at SLHC • rate up to 0.4 GHz/cm2 • dose up to 1016 cm-2 • mostly hadrons: p, p in GeV range (mips) • g (MeV - GeV range) • n (MeV range)

35. Irradiation facility with 5 GBq source • Remote controlled irradiation stage • Sample can be moved in and out by pneumatic piston • => separating induced signal from background signal source Ageing sample

36. 90Sr source at Nikhef • 5GBq 90Sr • 1 – 2 Mev e- simulating mips • Rate calibrated with ionisation chamber • Distance sample to source 4.7 mm => • 1.33 GHz/cm2 (sphere) • 1.25 GHz/cm2 (parallel surface) • => 1.15*1014 mips/cm2/day in bulk material • SLHC aim: 1016 mips/cm2 for the ATLAS b-layer

37. Dummy detector for GOSSIP ageing studies • Dummy glass ROC • Circular pads on 60 µm pitch • Drift volume 13 x 13 mm, 1.24 +/- 0.01 mm high • Centre signal electrode: 2 x 2 mm structure of 1089 pads (red) • => Final detection volume is a block of 2 x 2 x 1.2 mm • Gas gain by Micromegas • Closed gas volume of 210 µl • Gas flow ~ 0.5 l/h • => more than 2000 volume exchanges/hour Iguard electrode Icentre electrode

38. 5 GBq 90Sr source 1.0 Ø5.0 4.7 drift cathode Guard electrode Micromegas 1.2 Dummy ROC centre electrode 2.0 Source geometry during irradiation • Distance centre source to centre drift volume 4.7 mm • Homogeneous irradiation of centre electrode Dimensions uniformly scaled

39. Gas: Ar/iC4H10 70/30 Rate 1.33 GHz/cm2 Rate dependence investigated by gain curve No sign of saturation until gain = 900 Rate dependence

40. Ageing method • Two gas mixtures tested so far • Gossip21: He/iC4H10 78/22 • Gossip23: Ar/iC4H10 70/30 • Data taking every 2s • Background current periodically measured for reference • 20 s per 4 hour Centre current Guard current

41.  No sign of decay of gas gain Instabilities partly caused by variations in temperature and pressure?  But measurement had to be terminated because of increased sensitivity for HV trips Tripping induced by radiation Gossip21: He/iC4H10 78/22 • Vgrid = -418 V • Vcathode = -627 V

42. Gossip23: Ar/iC4H10 78/22 • More fluctuation but no significant indication for decay of gas gain • Trip (Iguard > 2 µA) at about once a week • Again measurement had to be terminated after 22 days because of increased sensitivity for tripping • Vgrid = -635 V • Vcathode = -889 V

43. Comparison to earlier measurement • 8.04 keV X-rays at Panalytical • Here 40% reduction in gain but no tripping problems • Using X rays instead of MIPs? • Anode: solid aluminium plate instead of small pads glass ROC? • Gas: Ar/CH4 90/10 vs Ar or He / iC4H10 mixtures? • At Panalytical ~ 5x higher charge rate? • Much lower gas refreshment rate • Other reasons???

44. Examining Gossip21 after irradiation • Field foil removed • Coloured spot on top of Micromegas near one of the gas pipes • Probably inlet • When removing Micromegas no other pollutions/damages found • Dummy ROC and Micromegas were still clean • => no visual cause for HV tripping traced

45. Ageing test chamber with ionisation by UV light Quartz window • Initiated by Harry van der Graaf • Long, thin chambers from clean materials • SS • glass • ceramics • No plastics, epoxies • Closed loop gas system • Gas gain by InGrid structure • Inserting suspicious material in test container • Possible purification of the gas by avalanches • Downstream chambers have less ageing (LHC-b experience) • Still in development Test container

46. Conclusions • For the Gossip prototypes we do not see the common ageing behaviour of gaseous detectors • No significant decrease of gas gain even at a high dose • But for iC4H10 mixtures deterioration of HV stability • Not in agreement with earlier X-ray CH4 test for unknown reasons • Will continue tests with other quenchers • CO2 • DME • CH4 • Verification required with other kinds of irradiation (hadrons, g, n) • SiProt ageing looks promising • No significant effect observed until 4 *1015 cm-2 • Protection and ROC operation remain intact • To be repeated with neutrons • UV light ageing • Convenient experimenting • Easy way tracing ageing compounds

47. spare