Channel-Plate Photomultipliers

Transcription

1 The New Micro-Channel Channel-Plate Photomultipliers A revolution in lifetime spectroscopy? * ) F. Bečvář Charles University, Prague * ) and also New Ultra-Fast Digitizers

2 OUTLINE The state-of-the-art and persisting problems of positron lifetime spectroscopy Upgrading based on employment of the ultra-fast digitizers About the method Results of testing measurements Further perspectives: the use of the MCP Photomultipliers Simulations Discussion of the results Conclusions

3 Positron-lifetime measurements with 22 Na source 2.60 yr Na 3.7 ps γ 1274 kev E.C. (9.5.%) β kev (90.4 %) BaF 2 detectors Ne 0 β + (0.1 %) 511 kev 1274 kev 22 Na + Sample Delayed-coincidence spectra 180 o Positron lifetimes 511 kev Detection time difference

4 Lifetime measurements with a pulsed positron source BaF 2 γ 511 kev γ 511 kev e + Source of pulsed positron beam BaF 2 Σ Detection-time Machine s time zero

5 Developments during last two decades 2.60 yr Na What E.C. (9.5.%) remains and will (probably) remain unchanged 3.7 ps β kev (90.4 %) Ne BaF 2 detectors - the basic principles of the method γ 1274 kev - the use of β + (0.1 BaF %) 2 scintillators (?) 0 Promising progress in 1274 kev 511 kev - electronics - photonics Delayed-coincidence up to now no convincing impact on spectra the quality of lifetime measurements is seen 22 Na + Sample 180 o Positron lifetimes 511 kev Detection time difference

6 Positron-lifetime spectrometers at Charles University PMTs Helmholtz coils Assembled in 1992 and still going strong

7 Detail I.

8 Detail II. 22 Na sandwitched between two parts of a sample BaF 2 crystals

9 An example of positron-lifetime data Number of Annihilation Events per Channel Residual/σ Pile-up effect No significant residuals Positron annihilation in reactor steel 6.25 ps/channel Resolution power 147 ps (FWHM) Radiation-induced defects Channel Number χ 2 /ν = Free positrons sometimes an extremely short group Annihilation in the source To dig out a weak component with τ = 25 ps -- a difficult task even for the best positron-lifetime spectrometers

10 Problems and limitations

11 Systems using BaF 2 and Philips XP2020/Q PMTs Conditions for the best timing properties: fast dynode output (M. Moszynski, 80 s) increased voltage between the photocathode and 1 st dynode over the factory-permissible limit compensation of the external magnetic field the use of customized CF discriminators At expense of drawbacks: parasite oscillation of the output signal (space-charge artifacts) pronounced degradation of timing properties with time the most serious! Signal from D10 (mv) kev The anode-like prelude carries the best timing information Output dynode signal 511 kev XP2020/Q ser. No H.V. Divider IV Pile-up effects expected! Time (ns)

12 Ageing of XP 2020/Q PMTs 1.2 Output from D9 (rel. units) I XP2020/Q ser. No I XP2020/Q ser. No II II III III 170 III FWHM (ps) H.V. off I II Time (days)

13 Ageing in an alternative regime of operation Output from D9 (rel. units) XP2020/Q ser. No Divider I Divider IV Time (days)

14 Long-term behaviour of timing resolution Weak 22 Na, "fast-slow" Regular 22 Na, "fast-slow" Regular 22 Na, strobing with E γ sum FWHM (ps) Helmholtz coils installed Two medium-gain preamplifiers installed 140 One low-gain preamplifier installed & H.V. increased 130 Detectors interchanged H.V. off Age (years)

15 White-noise-based absolute calibration of time scale 3.3 Channel width (ps) Calibration of time scale Valid starts: x 10 8 Stop frequency: 10 MHz Width precision 6144 of % Channel number for each individual channel

16 Persisting problems of positron lifetime spectroscopy Ageing of PMTs Limited timing resolution power Its instability and degradation with time Not-well controlled role of pile-up effects Difficulties with calibration of time scale Difficulties with tuning (many parameters to be optimized) Limited throughput (usually less than 100 coincidences/s)

17 Upgrading based on the use of the ultra-fast digitizers

18 How ultra-fast (4 GS/s) digitizers work 1GS/s digitizer No. Right now 8-bit code ps FIFO buffer 3 4 U(t) t The main (segmented) memory

19 Block scheme of the PLT spectrometer using the DC-241 s* ) Na22+αFe Dynode Out Det. 1 Det. 2 Dynode Out Diff. CFD Diff. CFD Blk. Out Blk. Out Passive mixer Trigger Level In Master DC-241 Slave DC-241 In C C C To PC * ) Digitizers developed and produced by Acqiris, S.A., Geneva

20 Shapes of digitized waveforms (full scale) 128 A set of BaF 2 detector signals as seen by an externally triggered DC Digitized voltage Time in units of 0.25 ns

21 As obtained waveform pairs from real measurements Digitized signal 0-50 Legend: Ch1 -start Ch3 -stop Time in units of 0.25 ns Each waveform formed by 240 sampling points 2000 pairs of waveforms per one transfer to a PC Totally accumulated 40 GByte of sampling points

22 Determination of start-stop time difference from digital filtering Parabolic interpolations R.m.s. Feet Maxima 60 CF1 40 CF2 20 Parameter δ = Signals -40 Minimima

23 As obtained pairs of waveforms Digitized signal 0-50 Legend: Ch1 Ch Time in units of 0.25 ns

24 and after digital filtering

25 Behaviour of normalized waveforms stop waveforms C.F. = 0.5 Delta = Normalized signal start waveforms C.F. = 0.4 Delta = Time in units of 250 ps

26 Behaviour of normalized waveforms near the C.F. point 0.6 Normalized signal start waveforms C.F. = 0.4 Delta = ps C.F.point Residual digital noise of only 20 ps (FWHM) Time in units of 250 ps

27 DC-241 positron-lifetime spectrum I Residuals Counts per ps channel α-fe sample Wide energy windows x 10 6 events coincidences s -1 χ 2 /ν = orders of magnitude! ps Normalized response function (ns -1 ) Time (ns)

28 DC-241 positron-lifetime spectrum II Residuals Counts per ps channel α-fe sample Energy windows of medium widths x 10 6 events 59.1 coincidences s -1 χ 2 /ν = ps Normalized response function (ns -1 ) Time (ns)

29 DC-241 positron-lifetime spectrum III Residuals Counts per ps channel α-fe sample Narrow energy windows x 10 6 events 33.8 coincidences s -1 χ 2 /ν = ps Normalized response function (ns -1 ) Time (ns)

30 Positron-lifetime spectrum obtained from the fast-fast configuration 4 Residuals Counts per ps channel α-fe sample Wide energy windows x 10 6 events 125 coincidences s -1 χ 2 /ν = Less than 3 orders 173 ps Normalized response function (ns -1 ) Time (ns)

31 Timing resolution: influence of age and various improvements Weak 22 Na, "fast-slow" Regular 22 Na, "fast-slow" Regular 22 Na, strobing with E γ sum Regular 22 Na "fast-fast" 170 FWHM (ps) ? World record! 140 Regular 22 Na DC-241 Digitizers 130 H.V. off Age (years) July 2003

32 Comparison with other similar attempts 1. R. Aarvikk et al., 13th International Conference On Positron Annihilation, Kyoto, Japan, 2003 Configuration with ACQIRIS DP-210: FWHM = 240 ps, throughput = 300 coi/s Fast-fast configuration: 220 ps Plastic scintillators 2. H. Saito et al., Nucl. Instruments and Methods A 487, 612 (2002) Configuration with Ocsilloscope LeCroy pro 960: FWHM = 140 ps, throughput very low, 20 coi/s BaF 2 scintillators 3. H. Saito and T. Hyodo, Physical Review Letters 90, (2003) Configuration with Ocsilloscope LeCroy pro 960, equivalent for two-detector setup: FWHM = ps throughput very low, 20 coi/s BaF 2 scintillators

33 Advantages of the digitization 1. Simplicity of the system 2. Flexibility 3. Ex post optimization 4. Low digital noise 5. Unprecedented timing resolution achieved 6. Reasonably high throughput 7. Significant suppression of parasite pile-up effects 8. Clean response function 9. Immunity to ageing of PMTs and BaF 2 crystals 10. No detectable influence of non-uniformity of time sale and differential non-linearity of output codes

34 Perspectives of the use of state-of-the-art MCP PMTs

35 Why MCP PMTs? Compared to standard PMTs substantially smaller fluctuations of the electron avalanche transport time A very short leading edge of the single-photoelectron output signal Uniformity in collection of photoelectrons from the photocathode Immunity of the output signal to the external magnetic field A very large area, possibly cm 2, from which the last generation of secondary electrons are emitted The efficiency for trapping the photoelectrons by microchannels is expected to be even as high as 70 %

36 Photek MCP PMT125

37 Photek MCP PMT125 MCP - pitch 6µm/7µm wall thickness of 1 µm, gain 10 3 UV window Dia 25 mm 50 Ohm vaweguide Accelerating grid Anode Drawing by courtesy of Photek, Ltd.

38 Simulation of the MCP PMT signals basic assumptions Emission of λ = 220 nm photons from BaF 2 follows the exponential law with a relaxation constant τ = 730 ps and so also the emission of electrons from the photocathode The number of useful photoelectron in each event of γ detection obeys the Poisson distribution The single-photoelectron signals at the PMT anode are of Gaussian shape Pulse heights of these signals follow a χ 2 distribution with a specified parameter ν A time difference between the emission of a photoelectron and the occurrence of the corresponding contribution to the anode signal follows a normal distribution (jitter time) The resulting signal at the anode is subject to RC shaping The γ-detection time is deduced using the method of true Constant Fraction (C.F.)

39 Simulation of the MCP PMT output signals the algorithm Proba Emission of a single photoelectron Exponential distribution, τ = 0.67 ns (BaF 2 ) BaF 2 +PMT125: 320 photoelectrons/mev 0 0 Time Proba Creation of of one of many e-h pairs Absorption of γ-ray by BaF 2 Voltage 0 0 Time Random emission of p.e. Voltage 0 The signal formed by many contributions 0 0 Time 0 Distribution of the electron avalanche transit time Transit time jitter: FWHM = 150 ps Normal distribution assumed Time Single-photoelectron responses at the anode Signal: FWHM = 140 ps Pulse-height distribution: χ 2 with ν 2 A contribution to the resulting signal The anode signal For 511 kev γ rays 160 photoelectrons are contributing A random process! Role of RC not illustrated

40 A set of 40 randomly simulated anode signals -1.4 Output voltage (arbitrary units) Photek PMT125 Trial No Time (ps)

41 Behaviour of C.F. points Output voltage (arbitrary units) Photek PMT125 Trial No. 11 C.F. = 0.15 FWHM = 65.7 ps Time (ps)

42 Behaviour of C.F. points Output voltage (arbitrary units) FWHM = 66.0 ps Photek PMT125 Trial No. 12 C.F. = Time (ps)

43 Timing response function Probability density (ps -1 ) Photek PMT125 Trial 12 Eγ = 511 kev C.F. = 0.15 FWHM = 66.0 ps Time (ps)

44 Behaviour of area C.F. points Output voltage (arbitrary units) FWHM = 62.9 ps Photek PMT125 Trial No. 13 "Area" C.F. = Time (ps)

45 Timing response function Deduced from analysis of waveforms Probability density (ps -1 ) Photek PMT125 Trial 13 Eγ = 511 kev "Area" C.F. = 0.15 FWHM = 62.9 ps Time (ps)

46 A crucial role of jitter time Output voltage (arbitrary units) Time (ps) Photek PMT125 Trial No. 13 Jitter time 150 ps (FWHM) "Area" C.F. = 0.15 FWHM = 62.9 ps Output voltage (arbitrary units) Photek PMT125 Trial No. 13a Jitter time 5 ps (FWHM) "Area" C.F. = 0.15 FWHM = 36.1 ps Time (ps) FWHM = 36 ps -- an ultimate limit set by BaF 2 scintillators and a finite Q.E.

47 Summary of the results for signals from 511 kev γ-rays Trial No. SPE signal risetime (ps) FWHM of SPE transit jitter time (ps) Number of SPEs per 511 kev Shaping RC (ps) ν eff for SPE PH distributrion FWHM of timing CF ratio FWHM of PH distribution (%) E γ = 1274 kev BaF 2 + XP2020/Q a) b) e) c) d) f) -- a) Quantum efficiency QE = 25 % and efficiency of photoelectron collection EPC = 70% at 220 nm, as declared by Photek b) QE = 25 %, EPC = 70% and γ-ray energy of 1274 kev c) The case of PMTs Philips XP2020/Q with QE = 25 % at 220 nm d) PMTs Philips XP2020/Q with QE = 20 % at 220 nm e) QE = 25 % and EPC = 44% f) Area C.F.

48 Are the estimates of the 220 nm light output realistic? Pulse-height spectrum for BaF 2 + XP2020/Q Cs + 60 Co 662 kev ( 137 Cs) Counts per channel v v FWHM = 6.8 % Output from D9 (channels) 1173 kev ( 60 Co) 1332 kev ( 60 Co) γ-ray energy (kev) FWHM ot the 661 kev γ line: 6.8 % Q.E. for 220 nm: 25 % Rel. var. of the single ph.el. signal PH distribution: 50 % Philips PMTs N ph.el. (220 nm)/511 kev = 256 even more optimistic!

49 Summary of the results for signals from 511 kev γ-rays Trial No. SPE signal risetime (ps) FWHM of SPE transit jitter time (ps) Number of SPEs per 511 kev Shaping RC (ps) ν eff for SPE PH distributrion FWHM of timing CF ratio FWHM of PH distribution (%) E γ = 1274 kev More realistic BaF 2 + XP2020/Q a) b) e) c) d) f) -- a) Quantum efficiency QE = 25 % and efficiency of photoelectron collection EPC = 70% at 220 nm, as declared by Photek b) QE = 25 %, EPC = 70% and γ-ray energy of 1274 kev c) The case of PMTs Philips XP2020/Q with QE = 25 % at 220 nm d) PMTs Philips XP2020/Q with QE = 20 % at 220 nm e) QE = 25 % and EPC = 44% f) Area C.F.

50 Results of simulations for BaF 2 +XP2020/Q 2000 Simulace výstupního signálu z BaF 2 + XP2020Q Výstupní Minus output napì voltage tí (rel. (rel. jednotky) units) FWHM = ps 22 Na γ-cascade 130 ps As obtained signal FWHM = 135 ps Digital const. fract. Emulation of hardware const. fract. Input parameters τ = 670 ps E γ = 511 kev <N f.el. > = 225 t jitter = 750 ps (FWHM) t ph.el.signal = 700 ps (FWHM) Attenuated original signal minus delayed inverted signal Èas Time (ps) (ps) The FWHMs digital deduced constant fraction from 10 5 is simulated definitely waveforms better!

51 Summary of simulation trial No. 11 Assumptions Lifetime of the fast BaF2 scintillation component 670 ps The number of fast, 220 nm photons per MeV 1800 Quantum efficiency at 220 nm 25 % Efficiency for collection of photoelectrons 70 % The number of useful photoelectrons per 511 kev 167 Single-photoelectron (SPE) output signal shape Gaussian SPE signal rise time 120 ps Shape of the PH distribution of the SPE output signal exponential Uncertainty of the electron avalanche transport time (FWMH) 150 ps RC shaping time ( C = 15 pf, R = 50 Ω ) 750 ps Photek: these values are conservative! Results Uncertainty of timing for 511 kev γ rays (FWHM) at CF ratio of ps Timing resolution power (FWHM) of a two-detector system installed at a pulsed positron beam 46.5 ps Timing resolution power (FWHM) for a standard two-detector configuration with the 22 Na positron source 77.5 ps* ) EPOS * ) To be compared with the best value 131 ps achieved with standard PMTs

52 The problem of the slow BaF 2 component Expected number of γ-ray detections per bunch: Probability of two detections in two consecutive bunches: 0.33 % Approximate next-neighbor spacing of the single-photoelectron signals from the slow component immediately after the fast signal : 1.2 ns FWHM of the single-photoelectron signals 0.15 ns Presence of pollution from the slow component can be easily detected by checking the size of the r.m.s. fluctuations of the waveform baseline

53 Characteristics of output signals of PMT125s Current at the peak of the anode output signal Peak output voltage at 50 Ohm anode resistor Effective width of output pulses 25 µ A 1.3 mv 1 ns Count rate < s -1 Average current Total charge collected at the anode over the period of 10 years < 12 na < < 5 C verified that these values are well below limits set by the manufacturer

54 SUMMARY The ultra-fast digitizers significantly improve the quality of positronlifetime measuremets the resolution power the makeup of the response function (suppressed pile-up effects) In the start-stop/stop-start regime they will double the coincidence count rate With the state-of-the-art MCP PMTs in combination with the ultra-fast digitizers a timing resolution power of ps (FWHM) is expected for the planned EPOS γ-detection system to be tested in May 2004 Software filtering of events polluted with photons from the slow scintillation component of BaF 2 seems possible The need for digitizers with the speed higher than 4 GS/s? simulations may provide an answer

55