Digital techniques for neutron detection and pulse shape discrimination in liquid scintillators

1. Digital techniques for neutron detection and pulse shape discrimination in liquid scintillators • P.J. Sellin, S. Jastaniah, G. Jaffar • Department of Physics • University of Surrey • Guildford, UK • p.sellin@surrey.ac.uk • www.ph.surrey.ac.uk/cnrp

2. Contents • Motivation for this work • Pulse shape discrimination (PSD) in organic scintillators: • traditional PSD in liquid scintillators • direct detection of neutron scatter events • digital PSD algorithms • Results from the Surrey digital setup: • Digital PSD from integrated and current pulses • PSD Figure of Merit (FOM) • 10B-loaded scintillator for fast neutron detection: • review of capture-gated neutron detection in BC454 • the use of BC523/BC523A boron-loaded liquid scintillators • current status and limitations of a portable capture-gated neutron detector • New material developments • Conclusions

3. Emphasis on fast computationally-simple digital algorithms suitable for field instruments Efficient n/g discrimination is essential - the extraction of a weak fast neutron flux against a strong gamma ray background Full-energy fast neutron spectrometry has particular advantages for dosimetry detectors: Motivation for this work: Development of digital neutron monitors for neutron field measurements, homeland security, and neutron dosimetry Portable instruments can take advantage of compact digital pulse processing technology Introduction See also: A. Rasolonjatovo et al, NIM A492 2002 423-433

4. Pulse shape discrimination • Pulse shape discrimination (PSD) in organic scintillators has been known for many years - particularly liquid scintillators (NE213 / BC501A) • PSD is due to long-lived decay of scintillator light caused by high de/dx particles - neutron scatter interactions events causing proton recoils: mean decay time t

5. Integrated vs current pulses • Extraction of scintillation decay lifetime t depends on the RC time constant of the external circuit: • Large time constant RC>>t: • integrated pulse - event energy extracted from pulse amplitude • t extracted from pulse risetime: • Short time constant RC<<t: • current pulse - event energy extracted from pulse integral • t extracted from pulse decay time:

6. Pulse risetime algorithms (1) • Integrated pulses - using a PMT preamplifier • Improved signal-noise ratio • Risetime limited by preamp (~10ns) • 1. 10-90% risetime algorithm • Current pulses - anode connected directly to 50W • Simple circuitry, fastest response • Two PSD algorithms have been investigated: • 2. ‘time over threshold’ algorithm • Other techniques use a full least-squares fit to the pulse shape, eg. N.V. Kornilov et al, NIM A497 (2003) 467-478. • S. Marrone et al, NIM A490 (2002) 299-307

7. Pulse risetime algorithms (2) • 3. ‘Q-Ratio’ algorithm • A digital implementation of the common charge integration PSD algorithm - the current pulse is integrated within a ‘short’ and a ‘long’ time window • eg. D. Wolski et al, NIM A360 (1995) 584-593 • Advantage of this technique compared to ‘Time over Threshold’ is that all the data in the pulse is sampled •  better S/R ratio The Q-Ratio ‘signal amplitude’ A is: PSD parameter is:

8. Digital PSD on inorganic scintillators • Digital implementations of PSD algorithms have been already applied to commercial systems, suitable for slower inorganic scintillators • Eg. The XIA digital data acquisition system, sampling at 40 MHz, time interval 25 ns. See: W. Skulski and M. Momayezi, NIM A458 (2001) 759-771 photon interaction in silicon photodiode scintillation interactions in CsI(Tl)

9. XIA performance Simple rise time inspection gives reasonable a, g separation More sophisticated algorithms allow good discrimination of p, a, g:

10. Other PSD techniques • Other techniques use a full least-squares fit to the pulse shape: • eg. by de-convolution of the scintillator light pulse from the detector response function: • N.V. Kornilov et al, NIM A497 (2003) 467-478. where s(t) is the measured pulse signal, r(t,t’) is the detector response function, and f(t’) is the scintillator light pulse PMT response function s(t) expt data and fit This technique is computationally intensive and not suitable for portable instruments

11. Least square fitting of scintillator pulses • Fast digital sampling of liquid scintillators has been combined with full linear-regression curve fitting: • S. Marrone et al, NIM A490 (2002) 299-307 • Convolution of the detector response function with a single exponential decay term does not fit the observed pulse shapes • a two-component exponential function is required: • a complex iterative fitting procedure is required to optimise all 6 free parameters  very computationally intensive

12. Direct discrimination of fast neutrons • In principal, direct discrimination of fast neutrons can be attempted by observing the time delays between fast neutron scatters. • This has been reported by Reeder et al, NIM A422 (1999) 84-88. • 1 MeV neutron travels at 5% of c, with a 90% chance of interaction in 10cm of plastic scintillator • Time delay between 1st and 2nd neutron scatter is ~3 ns • 1 MeV gamma has mean free path of ~13 cm, with a flight time of 0.45 ns • The fast neutron pulse in plastic • scintillator should be broader than • from gammas • Technique need as fast digitiser • with nanosecond timing. Graph shows calculated average time between hydrogen recoils vs neutron energy

13. Requirements for the direct technique • Reeder’s method used a digital oscilloscope to capture pulse shapes - direct record of fast neutron scatters prior to significant moderation. • Better efficiency that ‘capture gated’ methods since only 2-3 scatters are required - the neutron can then escape from the scintillator. • Requires timing resolution ~1 ns or better • Single neutron scatter events cannot be distinguished from gammas • 252Cf time-of-flight system used to provide tagged 1 MeV neutrons

14. Results of direct discrimination • Results: • average width of 100 gamma pulses: 3.3 ns • average width of 100 neutron pulses: 3.5 ns • Why are the gamma pulses so broad (not expected by MCNP studies)? • Fast light pulses directly into PMT gives width ~1.4ns • single photon fluorescence confirmed plastic decay time • scintillator shows asymmetric pulse shape which washes out the expected time differences

15. Single channel specification: 8 bit resolution 1 GS/s, 500 MHz 2 Mpoints waveform memory 80 MB/s sustained data transfer rate to PC (12 bit cards, up to 400 MS/s also available) Custom LabView software for real-time pulse analysis and histogramming High speed waveform digitisers now provide 1ns sampling times (1 GS/s), 8 bit resolution, high speed data transfer to PC: We use the Cougar system from Acqiris - www.acqiris.com 4 channel compactPCI crate-based system, expandable up to 80 channels The Surrey waveform digitiser system

16. PSD measurements were initially made with small-volume (100 ml) commercial cells, containing BC501A (no boron) and BC523A (5% 10B enriched) A similar size cell of BC454 plastic was also studied (5% natural boron, ~1% 10B) A larger 700 ml cell was the constructed to investigate capture-gated neutron detection. This cell included an embedded 30mm diameter BGO scintillator Detector Cells When filling the cells, the scintillator was bubbled with N2 gas to purge the oxygen. A fume cupboard is required, and careful adhesion (Torrseal) of the glass window to the metal canister is necessary to prevent evaporation/leakage

17. 10B capture peak • Typical pulse height spectrum from a BC523A cell, acquired with the digital data acquisition system: The 10B capture peak is observed at 60 keV electron-equivalent energy.

18. Energy Calibration 44 keV Tb X-ray 8-bit digital DAQ • Liquid scintillator operated at 2 gain settings, with separate energy calibrations: • High Gain: • photopeak for X/g-rays < 60 keV: Ba, Tb K X-rays 241Am g-ray • Low Gain: • Compton edge for high energy g-rays: 57Co 137Cs 60Co 44 keV Tb X-ray 12-bit analogue DAQ

19. Digital DAQ calibration low energy photopeak calibration typical photopeak spectra - 8 bit digital system high energy Compton edge calibration

20. PSD at low gain • Risetime versus pulse height plot at low gain setting showing n/g PSD from (a) BC501A, and (b) from BC523A.

21. No PSD in plastic BC454 • We also tested PSD in plastic scintillator BC454 - no discrimination was seen for neutron scatter events all events

22. PSD at high gain • At high gain, the 10B capture peak is visible due to simultaneous detection of 7Li and a no significant PSD is observed Lack of PSD is due to quenching of slow component from heavy ions - limited PSD has been seen in ‘special’ 10B-loaded scintillator S. Normand et al, NIM A484 2002 342-350

23. PSD Figure of Merit • Quality of PSD is described using a Figure of Merit (FOM): • Vertical ‘slices’ from the 2D spectra give risetime histograms: Sng = separation of two peaks Fn,g = n,g peak centroid position high energy FOM = 1.5 low energy FOM = 1.4 Method is similar to conventional analogue PSD techniques FOM is extracted digitally in software FOM>1 required for ‘good’ PSD g n

24. PSD from current pulses (1) • ‘Time over Threshold’ current pulse algorithm - the 2D plot has a different shape • FOM is slightly worse than for integrated pulses with poorer valley separation, particularly at low signal amplitude

25. PSD from current pulses (2) • ‘Q-Ratio’ current pulse algorithm - the 2D plot has well separated locii across the full energy range • PSD performance at low signal amplitude is considerably better than ‘time over threshold’ algorithm

26. FOM plots from Q-Ratio algorithm FOM values are 1.1 for both energy ranges - the Q-ratio algorithm gives better overall PSD performance for current pulses

27. We have investigated liquid scintillator enriched with 10B - BC523A Often used for thermal neutron detection, 10B-loaded scintillator can also be used for ‘capture-gated’ neutron spectroscopy: Fast neutron spectroscopy routinely measures the energy of proton recoil events: 10B loaded liquid scintillator • where ERMAX is the maximum recoil energy of nucleus with atomic mass A • For protons, A=1 and ERMAX=EN

28. The method of ‘capture-gated’ neutron spectroscopy uses the technique of ‘moderate + capture’. If moderation occurs within the active detector, the full energy of the neutron EN can be uniquely measured Characteristic double-pulse sequence of moderation + capture provides clean fast neutron signature. Capture pulse has fixed amplitude (10B+n Q value) Amplitude of moderation pulse gives incident neutron kinetic energy  true ‘full energy’ neutron spectrometer Capture gated timing signals Neutron capture: n + 10B  7Li* + a + 478 keV g (Q = 2.31 MeV, 92%) n + 10B  7Li + a(Q = 2.79 MeV, 6%)

29. First capture-gated experiments Capture-gated neutron measurements were first reported in 1986 - 1991, initially with BC454 - plastic loaded with 5% natural boron WC Feldman et al (NIM A306 (1991) 350-365 and NIM A422 (1999) 562-566) developed a BC454 + BGO detector for the NASA Lunar Prospector The neutron capture lifetime was measured as 2.2 ms The BGO provides an additional signature for the coincident 478 keV gamma ray from deexcitation of 7Li* -> 7Li

30. Large-volume experiments Large-volume capture-gated experiments, again with BC454, were carried out by Miller. An array of 10 BC-454 detectors, each optically coupled to BGO and a photomultiplier. The 10B capture peak (Q ~ 2.3 MeV) was observed at an electron equivalent energy of 93 keV:

31. Multi-detector system The array of 10 detectors was arranged in a ring, to accommodate a central sample chamber. Designed at Los Alamos for neutron assay measurements MC Miller et al, Appl Rad Isotopes 47 (1997) 1549-1555 and NIM A422 (1999) 89-94 In both the Los Alamos and NASA systems, no PSD was available from the plastic scintillator, and only analogue readout electronics was used.

32. First measurements with liquid BC523 Boron-loaded liquid scintillator was developed to combine fast neutron detection properties with PSD for gamma rejection. T Aoyama et al, NIM A333 (1993) 492-501 measure a neutron capture lifetime of 2.2 ms in BC523 - 5% natural Boron The capture-gated spectroscopic performance of BC523 to monoenergetic neutrons was measured:  non-linear light yield vs recoil energy produces poor resolution spectra  a major limitation to the spectroscopic performance of this technique

33. Neutron capture lifetimes • After moderation in the scintillator, the neutron capture lifetime is dependent only on the 10B concentration (s 1/v): • and the thermal neutron probability distribution is given by: • The calculated capture lifetimes for the various commercially-available boron loaded scintillators are:

34. The Surrey BC523A detector head • The 700ml volume BC523A cell was fabricated from aluminium, with an embedded BGO detector to measure coincident 478 keV gamma rays from 10B reaction

35. Capture-gated neutron detection • Capture-gated neutron detection gives very clean fast neutron signature • Trigger event rate is low: requires full moderation of neutron within the scintillator  volume dependant • Full energy spectrometer - fast neutron energy obtained from amplitude of recoil pulse • PSD can be used to further reject false TAC start pulses Neutron capture: n + 10B  7Li + a Q = 2.31 MeV (92%) Q= 2.79 MeV (6%) neutron capture lifetime

36. Capture-gated TAC spectrum

37. Fast neutron capture lifetime • Neutron capture lifetime t has an exponential distribution: • where t depends only on 10B • concentration, since s1/v: Scintillator 10B (%)  (s) BC523A ~ 5 0.49 BC523 ~ 1 2.25 BC454 ~ 1 2.13 Short neutron capture times allow high event rates for the capture-gated detection mode Event rate with our 10GBq AmBe neutron source: ~20Hz for 700ml BC523A cell

38. New materials • New loaded scintillator materials offer much potential for future development of neutron detection methods. Some promising candidates include: • 1. Boron loaded plastics showing n/g PSD • Norman et al (NIM A484 (2002) 432-350) have shown limited fast neutron - gamma PSD from boron-loaded plastic, not previously observed in BC454: • limited PSD was seen from • scintillator grown at CEA, not • from BC454 • no alpha/lithium - gamma PSD • observed in either material • Boron loaded pastics quench the • long-lived triplet state that is normally • filled mainly by heavy charged particles

39. New materials (2) • 2. Lithium gadolinium borate • J Bart Czirr et al (NIM A476 (2002) 309-312) have produced a new loaded plastic scintillator, lithium gadolinium borate, which contains a mixture of high cross-section materials: • This material is still under test - obtaining large-volume samples • is still difficult

40. Conclusions • Digital PSD techniques in organic scintillators are being developed that rival traditional analogue methods - the performance of high speed waveform digitisers is key to these developments • Good n/g PSD performance of 1 ns sampling time, 8-bit resolution, digitisers has been successfully demonstrated, using computationally-simple algorithms suitable for field-portable instruments • The application of digital techniques to capture-gated fast neutron detection is under development, and offers a useful technique for fast neutron monitors • Issues for the future: • Fast waveform digitisers are still expensive and non-portable • True neutron spectroscopy from capture-gated 10B-loaded scintillator is currently limited by the non-linear light output of these materials • New loaded scintillators need to be developed offering good PSD of the neutron capture reaction (eg. 7Li+a from 10B).

41. References: • SD Jastaniah and PJ Sellin, “Digital pulse-shape algorithms for scintillation-based neutron detectors”, IEEE Trans Nucl Sci 49/4 (2002) 1824-1828. • SD Jastaniah and PJ Sellin, “Digital techniques for n/g pulse shape discrimination and capture-gated neutron spectroscopy using liquid scintillators”, in press NIM A.