System and Method to Count Neutrons

Abstract

A radiation detection system and a method for a parallel detection of gamma-rays and neutrons are provided, comprising a gamma-ray detector comprising a scintillator crystal comprising .sup.127I, a digitizer to generate digitized time series and an analyzer, characterized in that the analyzer is adapted to identify a primary signal component, a first delayed signal component and a second delayed signal component in the digitized time series. The first and second delayed signal components, respectively, correspond to an energy deposition of about 30 keV and about 138 keV, and follow the primary and first delayed signal components in time. The analyzer is further adapted to count the number of digitized time series comprising at least the first and the second delayed signal components as neutron events, thereby providing a measure for a neutron flux the scintillator crystal is exposed to.

Claims

1. A radiation detection system for a parallel detection of gamma rays and neutrons, comprising: a gamma-ray detector comprising a scintillator crystal comprising .sup.127I and a photodetector with an amplifier, wherein said scintillator crystal is adapted to convert energy deposited by gamma rays or conversion electrons to optical photons, the photodetector is adapted to convert the optical photons to an electrical signal and the amplifier is adapted to amplify the electrical signal, wherein the electrical signal is in a known relationship with the energy deposited by the detected gamma rays or conversion electrons in said scintillator crystal, a digitizer comprising sampling Analog to Digital Converters, ADC, wherein said digitizer is adapted to sample the electrical signal of said gamma-ray detector with a predetermined frequency of at least 20 mega samples per second to generate digitized time series of the electrical signal, and an analyzer which is coupled operatively to said digitizer, wherein said digitizer is adapted to transmit the digitized time series to the analyzer, and wherein the analyzer is adapted to analyze the digitized time series in order to identify signal components in the digitized time series with consecutive time delays between each other of at least 20 ns and at most 10 ?s, wherein each signal component is due to an energy deposition in the scintillator, wherein the analyzer is further adapted to: identify a primary signal component in the digitized time series corresponding to an energy deposition E.sub.0 in a predefined range, identify a first delayed signal component in the digitized time series, the first delayed signal component corresponding to an energy deposition E.sub.1 of about 30 keV, preferably between 15 keV and 50 keV, and following the primary signal component in time, identify a second delayed signal component in the digitized time series, the second delayed signal component corresponding to an energy deposition E.sub.2 of about 138 keV, preferably between 100 keV and 200 keV, and following the first delayed signal component in time, and count the number of digitized time series comprising at least the first delayed signal component and the second delayed signal component as neutron events, thereby providing a measure for a neutron flux the scintillator crystal is exposed to.

2. The radiation detection system of claim 1, wherein the analyzer is further adapted to quantify at least one event parameter of a group of event parameters for each of said digitized time series, the group of event parameters comprising the time difference between said primary signal component and said first delayed signal component the time difference between said primary component and said second delayed signal component the time difference between said first delayed signal component and said second delayed signal component, the energy deposition in the scintillator crystal corresponding to the primary signal component, the energy deposition in the scintillator crystal corresponding to the first delayed signal component, the energy deposition in the scintillator crystal corresponding to the second delayed signal components, and combinations thereof, evaluate for each of said digitized time series whether at least one event parameter of the group of event parameters fulfills a predefined criteria to classify said digitized time series as neutron event, and count the number of digitized time series classified as neutron event, thereby providing a measure for the neutron flux the scintillator crystal is exposed to.

3. The radiation detection system of claim 1, wherein the photodetector of the gamma-ray detector is a photomultiplier tube, a silicon photomultiplier (SiPM), or an avalanche photodiode.

4. The radiation detection system of claim 2, wherein the photomultiplier tube comprises a super-bialkali photokathode or an ultra-bialkali photokathode.

5. The radiation detection system of claim 1, wherein said scintillator crystal is a NaI crystal, preferably with Tl doping, or a CsI crystal, preferably with Na or Tl doping, or a NaI crystal, preferably with Tl doping, comprising Li or B as co-dopant.

6. The radiation detection system of claim 1. Wherein the gamma-ray detector, the digitizer and the analyzer are small enough to be arranged in a handheld device.

7. The radiation detection system of claim 1, wherein the gamma-ray detector, the digitizer and the analyzer are small enough to be arranged in a backpack.

8. A method to detect neutrons and gamma rays, utilizing the radiation detection system of claim 1, wherein the method comprises the following steps: neutrons interacting with .sup.127I of the scintillator crystal generating .sup.128I upon neutron capture, wherein .sup.128I de-excites both under a prompt emission of gamma radiation and at least partially via at least one long-lived excited state feeding another long-lived excited state, which at least sometimes leads to two delayed energy depositions in the scintillator crystal corresponding to the two delayed de-excitation steps producing either conversion electrons or gamma radiation interacting with the scintillator crystal, following a primary signal component which is due to the prompt emission, said gamma-ray detector generating an electrical signal in succession of an interaction between gamma rays or conversion electrons and said scintillator crystal, wherein said electrical signal is in a known relationship with the energy deposited by the detected gamma rays or conversion electrons in said scintillator crystal, said digitizer sampling the electrical signal of said gamma-ray detector with a predetermined frequency of at least 20 mega samples per second to generate a digitized time series of the detected gamma rays, and transmitting the digitized time series to the analyzer, said analyzer identifying a primary signal component in the digitized time series corresponding to an energy deposition E.sub.0 in a predefined range, said analyzer searching for a first delayed signal component in the digitized time series, the first delayed signal component corresponding to an energy deposition of about 30 keV, preferably between 15 kV and 50 keV, and following the primary signal in time, said analyzer searching for a second delayed signal component, the second delayed signal component corresponding to an energy deposition of about 138 keV, preferably between 100 keV and 200 keV, and following the first delayed signal component in time, said analyzer counting the number of digitized time series comprising at least the first delayed signal component and the second delayed signal component as neutron events, thereby providing a measure for the neutron flux the scintillator crystal is exposed to.

9. The method of claim 8, further comprising the steps: said analyzer quantifying at least one event parameter of a group of event parameters for each of said digitized time series, the group of event parameters comprising the time difference between said primary signal component and said first delayed signal component the time difference between said primary component and said second delayed signal component the time difference between said first delayed signal component and said second delayed signal component, the energy deposition in the scintillator crystal corresponding to the primary signal component, the energy deposition in the scintillator crystal corresponding to the first delayed signal component, the energy deposition in the scintillator crystal corresponding to the second delayed signal components, and combinations thereof, said analyzer evaluating for each of said digitized time series whether at least one event parameter of the group of event parameters fulfills a predefined criteria to classify said digitized time series as neutron event, and said analyzer counting the number of digitized time series classified as neutron event, thereby providing a measure for the neutron flux the scintillator crystal is exposed to.

10. The method of claim 9, wherein said analyzer finds and identifies at least one delayed signal component in said digitized time series by means of pulse pile-up reconstruction techniques, utilizing a method to decompose the digitized time series comprising piled-up signal components into these components, thereby quantifying at least one of the group of event parameters.

11. The method of claim 8, wherein said analyzer further: identifies another primary signal component in the digitized time series corresponding to an energy deposition in a predefined range, searches for a delayed signal component in the digitized time series, the delayed signal component corresponding to an energy deposition of around 138 keV preferably between 100 keV and 200 keV, and following said primary signal component in time, quantifies at least one event parameter of the group of event parameters for each of said digitized time series, the group further comprising the time difference between said primary component and said delayed signal component, the energy deposition in the scintillator crystal corresponding to said primary signal component, and the energy deposition in the scintillator crystal corresponding to said delayed signal components, said analyzer evaluating for each of said digitized time series whether at least one event parameter of the group of event parameters fulfills a predefined criteria to classify said digitized time series as thermal neutron event, and said analyzer counting the number of digitized time series classified as neutron event, thereby providing a measure for the neutron flux the scintillator crystal is exposed to.

Description

BRIEF DESCRIPTION OF THE DRAWING

[0076] Some embodiments of the invention are explained in the following with reference to the figures. The described examples are provided for illustrative purposes and are not intended to limit the scope of the invention.

[0077] FIG. 1 is a schematic set-up of an embodiment of the radiation detection system for a parallel detection of gamma rays and neutrons as provided by the present invention;

[0078] FIG. 2 is an exemplary digitized time series comprising delayed triple coincidences indicating neutron captures in .sup.127I;

[0079] FIG. 3 shows distributions of time differences t32 between the second delayed signal component and the first delayed signal component of delayed triple coincidences, measured at different detector loads;

[0080] FIG. 4 shows distributions of time differences t21 between a delayed signal component and the primary signal component of delayed double coincidences, measured at different detector loads and analyzed using pulse pile-up reconstruction, the analysis excluding all detected delayed triple-coincidence events;

[0081] FIG. 5 shows distributions of time differences t21 between a delayed signal component corresponding to an energy deposition of 138 keV and the primary signal component of delayed double coincidences, measured at different detector loads with a trigger holdoff of 1.8 ?s, thereby corresponding to prior art. This analysis also includes delayed triple coincidences, however, without being detected as such, as the first delayed signal component is hidden due to the trigger holdoff.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0082] FIG. 1 shows a first embodiment of the radiation detection system for a parallel detection of gamma rays and neutrons provided by the present invention.

[0083] Shown is a radiation detection system 1 comprising a gamma-ray detector 10, a digitizer 40 and an analyzer 50. The gamma-ray detector 10 comprises a scintillator crystal 20 comprising 127 J and a photodetector 30 with an amplifier, wherein said scintillator crystal 20 is adapted to convert energy deposited by gamma rays or conversion electrons to optical photons. The photodetector 30 is adapted to convert the optical photons to an electrical signal and the amplifier is adapted to amplify the electrical signal, wherein the electrical signal is in a known relationship with the energy deposited by the detected gamma rays or conversion electrons in said scintillator crystal 20.

[0084] Preferably, the photodetector 30 is a photomultiplier tube with intrinsic signal amplification. The photomultiplier may as well be an avalanche photodiode or a silicon photomultiplier (SiPM).

[0085] Most preferably, the photomultiplier tube comprises a super-bialkali photokathode or an ultra-bialkali photokathode.

[0086] The scintillator crystal 20 comprises .sup.127I intrinsically, acting as neutron-gamma converter. A neutron-gamma converter is adapted to emit gamma radiation upon capturing neutrons.

[0087] More specifically, said scintillator crystal 20 is a NaI crystal, preferably with Tl doping, or a CsI crystal, preferably with Na or Tl doping, or a NaI crystal, preferably with Tl doping, comprising Li or B as co-dopant.

[0088] Needless to say, scintillator crystal 20 and the photodetector 30 may be wrapped in light reflecting materials and arranged in a housing to protect them from external light and humidity (not shown).

[0089] The embodiment of the radiation detection system of FIG. 1 further comprises a digitizer 40 comprising sampling Analog to Digital Converters, ADC. The digitizer 40 is adapted to sample the electrical signal of said gamma-ray detector with a predetermined frequency of at least 20 mega samples per second, preferably at least 100 mega samples per second, to generate digitized time series of the electrical signal.

[0090] The embodiment of the radiation detection system of FIG. 1 also comprises an analyzer 50 which is coupled operatively to said digitizer 40. The digitizer 40 is adapted to transmit the digitized time series to the analyzer 50, and the analyzer 50 is adapted to analyze the digitized time series in order to determine a measure of the neutron flux, particularly the slow neutron flux, the detector 10 is exposed to. For this purpose, the analyzer 50 is adapted to identify signal components in the digitized time series with consecutive time delays between each other of at least 20 ns and at most 10 ?s, wherein each signal component is due to an energy deposition in the scintillator crystal 20.

[0091] Preferably, the analyzer 50 is adapted to analyze the digitized time series and to determine a measure for the neutron flux, particularly the slow neutron flux, the detector 10 is exposed to in real-time.

[0092] The analyzer 50 is further adapted to identify a primary signal component in the digitized time series corresponding to an energy deposition E.sub.0 in a predefined range. The analyzer 50 is adapted to search for a first delayed signal component in the digitized time series, the first delayed signal component corresponding to an energy deposition E.sub.1 of about 30 keV and following the primary signal component in time, and to search for a second delayed signal component in the digitized time series, the second delayed signal component corresponding to an energy deposition E.sub.2 of about 138 keV and following the first delayed signal component in time. The analyzer 50 is further adapted to count the number of digitized time series comprising at least the first delayed signal component and the second delayed signal component as neutron events, thereby providing a measure for the neutron flux the scintillator crystal is exposed to.

[0093] In practice, it may be sufficient to determine the energy depositions of the delayed signal components coarsely within time and energy windows, the energy windows e.g. ranging from 15 keV to 50 keV for the first delayed signal component and from 100 keV to 200 keV for the second delayed signal component. For single neutron counts, in particular in environments with very intense ambient gamma radiation (count rates far beyond 1000 counts per second), it has been proven useful to further suppress the background of random triple pulse events by quantifying at least one event parameter of a group of event parameters more precisely.

[0094] Preferably, the group of event parameters comprises the time difference between the primary signal component and the first delayed signal component, the time difference between the primary signal component and the second delayed signal component, the time difference between the first delayed signal component and the second delayed signal component, and the corresponding energy depositions of each signal component in the scintillator crystal 20, and combinations thereof.

[0095] The inventors were the first who realized that digitized time series representing triple coincidence events, wherein each pulse of the triple coincidence events fulfills relatively coarse grained time and energy windows, may be attributed with high accuracy to neutron events if at least one of the event parameters fulfills a predefined fine-grained criteria. Examples for such a predefined criteria are a time difference between the first delayed signal component and the second delayed signal component of less than 4 ?s, an energy deposition corresponding to the primary signal component of less than 300 keV, or the energy depositions corresponding to the first delayed signal component and second delayed signal component of 30 keV and 138 keV, respectively, within the resolution of the scintillator crystal 20.

[0096] More specifically, the analyzer 50 is preferably adapted to carry out pulse pile-up reconstruction techniques, wherein the pulse pile-up reconstruction techniques comprise a method to decompose the digitized time series into their signal components. Each signal component may have a pulse shape comprising an onset, the onset being characterized by a sharp rise in finite time, and an exponential decay, the exponential decay being essentially characterized by the light decay time of the scintillator crystal 20. Based on the reconstructed signal components, the analyzer 50 may quantify at least one of the group of event parameters. Time differences may be computed between the onsets of two consecutive signal components and energy depositions in the scintillator crystal 20 of a corresponding signal component may be determined by integrating the corresponding signal component.

[0097] The analyzer 50 may preferably carry out the pulse pile-up reconstruction techniques and subsequent computation of event parameters in real-time to enable an online detection of neutron events, e.g. for homeland security applications with moving sources.

[0098] Preferably, the gamma-ray detector 10 and corresponding electronics 40 and 50 are small enough to be arranged in a handheld device. Equally preferably, the gamma-ray detector 10 and corresponding electronics 40 and 50 are small enough to be arranged in a backpack. These form factors of the radiation detection system according to embodiments of the invention have been proven particularly useful for homeland security applications.

[0099] FIG. 2 shows exemplary digitized time series comprising delayed triple coincidences indicating neutron capture in .sup.127I followed by gamma-ray cascades involving the two long-lived excited states in .sup.128I, at about 167 keV (half-life of 175 ns) and at about 138 keV (half-life at 845 ns), respectively.

[0100] The digitized time series shown in FIG. 2 were measured as directly digitized anode currents in an embodiment of the radiation detection system comprising a 2?2 NaI (Tl) crystal 20 coupled to a photomultiplier tube (PMT), the PMT coupled with a digital spectrometer 40 operated at a sampling rate of 250 MHz. The gamma-ray detector 10 was exposed to a moderated .sup.252Cf neutron source. The digitized time series representing the signal components were stored and analyzed offline for delayed double and triple coincidences.

[0101] Digitized time series representing delayed triple coincidences comprise three signal components, namely a primary signal component recorded at a time t1, a first delayed signal component recorded at a time t2, and a second delayed signal component recorded at a time t3. In FIG. 2 these signal components are denoted as Pulse 1, Pulse 2, and Pulse 3, respectively. Triple coincidences as determined by the method according to embodiments of the present invention are characterized by a first delayed signal component, Pulse 2, corresponding to an energy deposition of around 30 keV in the scintillator crystal 20 and a second delayed signal component, Pulse 3, corresponding to an energy deposition of around 138 keV in the scintillator crystal 20. Note that the sub-pulse energies expressed in FIG. 2 are slightly overestimated, since the linear energy calibration used for this stage of the signal analysis does not consider the non-linearity of the NaI(Tl) light output.

[0102] Digitized time series representing delayed double coincidences comprise only two signal components, namely a primary signal component recorded at a time t1 and a delayed signal component recorded at a time t2. Delayed double coincidences as mentioned by Yakushev et al. are characterized by a delayed signal component corresponding to an energy deposition of around 138 keV in the scintillator crystal 20.

[0103] FIG. 2 shows only representations of delayed triple coincidence events which were analyzed using pulse pile-up reconstruction techniques. In case of a measurement involving a trigger holdoff, these delayed triple coincidence events are mostly mistaken as a delayed double coincidences since in most cases the first delayed signal component cannot be separated from the primary signal component because of the trigger holdoff such that a delayed trip coincidence event actually appears as a double pulse structure comprising a primary signal component and a delayed signal component corresponding to an energy deposition of about 138 keV. Real delayed double-coincidence events, however, only comprise one delayed signal component corresponding to an energy deposition of about 138 keV since the de-excitation cascades do not involve the excited state at around 167 keV (half-life of 175 ns) of 128 I in these cases. Without using pulse pile-up reconstruction techniques, real delayed double-coincidences cannot be distinguished from delayed triple coincidences since the latter are also detected as double pulses, e.g. due to a trigger holdoff.

[0104] The three panels of FIG. 2 highlight three different scenarios of delayed triple coincidences found by embodiments of the method according to the present invention. The upper panel, FIG. 2a, shows well isolated signal components of a delayed triple coincidence which were rather rare in the recorded dataset. In particular, the first delayed signal component, Pulse 2, corresponding to an energy deposition of around 30 keV in the scintillator crystal 20, mostly piles up with the primary signal component, Pulse 1, as shown in the middle (FIG. 2b) and the lower panels (FIG. 2c) of FIG. 2. Even the second delayed signal component, Pulse 3, corresponding to an energy deposition of around 138 keV in the scintillator crystal 20, often piles up with the prompt signal component or with the first delayed signal component, as shown in FIG. 2c. Hence, pulse pile-up reconstruction techniques are essential to decompose the digitized time series in order to identify the relevant signal components of delayed triple coincidences indicating neutron captures in .sup.127I.

[0105] Based on the sampling rate of the digital spectrometer 40 used for this demonstration, 1.8 ?s correspond to 450 samples in the example of FIG. 2. The exemplary digitized time series of FIG. 2 therefore clearly demonstrate, that Yakushev et al., being limited by a trigger holdoff of the spectrometer of 1.8 ?s, would have not only missed the first delayed signal component, Pulse 2, corresponding to an energy of around 30 keV in the scintillator crystal 20, but even also the second delayed signal component, Pulse 3, corresponding to an energy deposition of around 138 keV in the scintillator crystal 20 in all of the shown exemplary digitized time series.

[0106] The examples of delayed triple coincidences shown in FIG. 2 result from the method to detect neutrons and gamma rays, utilizing the radiation detection system according.

[0107] In a first step of the method, neutrons interact with 127I of the scintillator crystal generating .sup.128I up on neutron capture. .sup.128I de-excites both under a prompt emission of gamma radiation and at least partially via at least one long-lived excited state feeding another long-lived excited state, which at least sometimes leads to two delayed energy depositions in the scintillator crystal 20 corresponding to the two delayed de-excitation steps producing either conversion electrons or gamma radiation interacting with the scintillator crystal 20, following a primary signal component which is due to the prompt emission.

[0108] In a next step of the method, the gamma-ray detector 10 generates an electrical signal in succession of an interaction between gamma rays or conversion electrons and said scintillator crystal 20. The electrical signal is in a known relationship with the energy deposited by the detected gamma rays or conversion electrons in said scintillator crystal 20.

[0109] Next, the digitizer samples the electrical signal of said gamma-ray detector with a predetermined frequency of at least 20 mega samples per second, preferably at least 100 mega samples per second, to generate digitized time series of the detected gamma rays, and transmits the digitized time series to the analyzer 50.

[0110] In a further step of the method the analyzer 50 identifies a primary signal component in a digitized time series, denoted as Pulse 1 in FIG. 2, corresponding to an energy deposition E.sub.0 in a predefined range. Further, the analyzer 50 searches for a first delayed signal component in the digitized time series, denoted as Pulse 2 in FIG. 2. The first delayed signal component corresponds to an energy deposition of about 30 keV and follows the primary signal with a time delay of at most 2 ?s. Further, the analyzer 50 searches for a second delayed signal component, denoted as Pulse 3 in FIG. 2. The second delayed signal component corresponds to an energy deposition of about 138 keV and follows the first delayed signal component with a time delay of at most 10 ?s.

[0111] It may be sufficient to simply count the number of digitizes time series in which a primary signal component, a first delayed signal component and a second delayed signal component are identified to determine a measure the neutron flux the scintillator crystal is exposed to. Preferably, however, to further suppress random background events, the analyzer 50 quantifies at least one of a group of event parameters for each of the digitized time series, and evaluates for each of said digitized time series whether at least one event parameter of the group of event parameters fulfills a predefined criteria to classify said digitized time series as neutron event. With counting the number of digitized time series classified as neutron event the analyzer 50 provides a measure for a thermal neutron flux the scintillator crystal 20 is exposed to.

[0112] As demonstrated by the examples shown in FIG. 2 the analyzer 50 preferably is adapted to find and to identify the delayed triple coincidences in the digitized time series by means of pulse pile-up reconstruction techniques. For pulse pile-up reconstruction, the analyzer uses a method to decompose the digitized time series to their components. Thereby, the analyzer 50 quantifies at least one of the group of event parameters for each of the digitized time series.

[0113] The event parameters quantified by the analyzer 50 of the delayed triple coincidences shown in FIG. 2 comprise the time differences between the primary signal component (Pulse 1) and the first delayed signal component (Pulse 2), denoted by t21, and between the first delayed signal component (Pulse 2) and the second delayed signal component (Pulse 3), denoted by t32. The latter time difference, t32, will be discussed further with respect to FIG. 3.

[0114] The analyzer 50 is preferably adapted to compute a measure for the neutron flux, particularly the slow neutron flux, the detector 10 is exposed to, from the event parameters, as explained with respect to FIG. 3 and FIG. 4.

[0115] FIG. 3 shows distributions of time differences t32 between the second delayed signal component, corresponding to an energy deposition of around 138 keV in the scintillator crystal 20, and the first delayed signal component, corresponding to an energy deposition of around 30 keV in the scintillator crystal 20, of delayed triple coincidences as shown in FIG. 2, measured at different detector loads as indicated in the figure.

[0116] The data was measured with the detector 10, as described with respect to FIG. 2, being exposed to a moderated .sup.252Cf neutron source. The signal analysis was performed with the same algorithm as used for the analysis of signals shown in FIG. 2.

[0117] Additional detector load was generated by complementary Th ore and .sup.137Cs sources arranged at various distances from the scintillator crystal 20. The distributions of time differences measured at different detector loads are denoted by different symbols.

[0118] The t32 distributions were fitted with the fit function f(t)=a.Math.e.sup.??.Math.t+b, with ?=0.693/0.845 ?s representing the decay constant corresponding to a half-life of 845 ns, and then normalized with respect to the fit parameter a. The fits curves are also shown in the figure.

[0119] The fit parameter a measures the strength of the second delayed signal component (half-life of 845 ns) in the t32 distribution characterizing the second delayed transition, and thus the number of neutrons detected in a corresponding measurement. This means that all distributions shown in FIG. 3 are normalized to the same number of neutrons. The fit parameter b measures the background of random delayed triple coincidences which are not due to neutron captures in 127I. Also given in FIG. 3 are effect-to-background ratios E/B which are derived from the fit parameters. The effect-to-background ratios are computed from the number of neutrons

[00001]$N_n={?}_{t1}^{t2}a.Math.e^{{}^{}-?.Math.t}\mathrm{dt}=\frac{a}{?}\left[e^{{}^{}-?.Math.t_1}-e^{{}^{}-?.Math.t_2}\right]$

[0120] divided by the background contribution in the same time interval

N.sub.bck=?.sub.t1.sup.t2b dt=b.Math.(t.sub.2t.sub.1)

[0121] for an integration time ranging from t.sub.1=0.16 ?s, which is the minimum detectable time difference between piled-up signal components with the pulse pile-up reconstruction algorithm used for analyzing the data, to t.sub.2=4.0 ?s, which is a reasonable limit covering most neutron signals but not too much background. Note that the choice of the integration time range affects the E/B ratio. While t.sub.1 should always be set to the minimum detectable pulse delay to maximize the number of neutron counts considered, the optimum t.sub.2 value is affected by the background contribution, which varies with the actual detector load. The E/B ratios given here thus represent no general result, but exemplary values illustrating the trend: Time distributions as well as the corresponding E/B ratios reflect the rising background contribution towards higher detector load. However, the neutron contribution, distinguished by an exponential decay with a decay constant corresponding to 845 ns half-life, is still visible and can be quantified even at a detector load around 15 kcps in terms of the fit parameter a.

[0122] Preferably, the analyzer 50 computes a measure for the thermal neutron flux the scintillator crystal 20 is exposed to by determining a distribution of time differences t32 between the second delayed signal component and the first delayed signal component, fitting said distribution using a function comprising an exponential decay component corresponding to the half-life of the second delayed signal component and a scaling factor, wherein the scaling factor represents the number of (thermal) neutrons detected. Similarly, the number of neutrons may be obtained from the distribution of time differences between the first delayed signal component and the primary signal component, and even from the distribution between the second delayed signal component and the primary signal component, as long as the functions are adapted to comprise the corresponding exponential decay components.

[0123] Also preferably, the analyzer 50 computes another event parameter, the time difference t21 between the first delayed signal component and the primary signal. By selecting only such delayed triple-coincidence events as potential neutron counts being considered for further analysis, e.g. for accumulating the t32 distribution, where t21 does not exceed a predefined limit, preferably 2 ?s, more preferably 1.5 ?s, the effect-to-background ratio is increased at the expense of the effect count rate.

[0124] Also preferably, the analyzer 50 computes another event parameter, the energy deposition in the scintillator crystal 20 corresponding to the primary signal, E.sub.0. By selecting only such delayed triple-coincidence events as potential neutron counts being considered for further analysis, e.g. for accumulating the t32 distribution, where E_0 corresponds, within the energy resolution of the gamma detector, to the energy of a strong transition leading to the 167.3 keV state in .sup.128I, preferably 142 keV or 67 keV, or to the sum energy of such transitions, the effect-to-background ratio is increased at the expense of the effect count rate.

[0125] Similarly preferably, the analyzer 50 computes other event parameters, the energy depositions in the scintillator crystal 20 corresponding to the first and/or the second delayed signal components, E.sub.1 and/or E.sub.2, respectively. By selecting only such delayed triple-coincidence events as potential neutron counts being considered for further analysis, e.g. for accumulating the t32 distribution, where E.sub.1 corresponds, within the energy resolution of the gamma detector, to the 30 keV transition feeding the 137.8 keV state in .sup.128I, and/or E.sub.2 corresponds, within the energy resolution of the gamma detector, to the 138 keV energy released during de-excitation of the 137.8 keV state in .sup.128I, the effect-to-background ratio is increased at the expense of the effect count rate.

[0126] In general, said predefined criteria applied to one or more event parameters, eventually combined to a set of criteria, can be used to further increase the effect-to-background ration on the expense of the effect (i.e., neutron) count rate. This allows dynamically adapting the sensitivity of the radiation detection system 1 with respect to neutron detection to the ambient gamma-ray background by using sets of more or less restrictive criteria, depending on said ambient background. In case of low ambient gamma background, weak or even no criteria applied to the event parameters maximize the neutron count rate at still acceptable background in the neutron counts. In case of a strong ambient gamma background, leading to an increased number of random delayed triple coincidences, more restrictive event selection criteria reduce the neutron sensitivity but may still allow quantifying the neutron flux the scintillator crystal 20 is exposed to.

[0127] FIG. 4 shows distributions of time differences t21 between a delayed signal component and the primary signal component of delayed double coincidences comprised in the same data sets as referenced above with respect to FIGS. 2 and 3.

[0128] The event parameter t21, representing the time difference (delay) between the delayed signal component corresponding to an energy deposition of around 138 keV in the scintillator crystal 20 and the primary signal component, was quantified by the analyzer 50. The time distributions shown in FIG. 4 exclusively comprise delayed double coincidence events according to which a delayed signal component corresponding to an energy depositions of about 138 keV but no preceding delayed signal component corresponding to an energy deposition of about 30 keV was detected. FIG. 4 thus exhibits data that may be used for neutron detection in addition to the delayed triple coincidence events discussed with respect to FIGS. 2 and 3.

[0129] The t21 distributions were fitted with the function f(t)=a.Math.e.sup.??.Math.t+b, with ?=0.693/0.845 ?s, but (for better precision) normalized with the neutron contributions obtained from the delayed triple coincidence data shown in FIG. 3. The effect-to-background ratios E/B were determined from fits applying the formulae and integration time ranges as used for FIG. 3 to the t21 distributions shown in FIG. 4. The numbers N.sub.D/N.sub.T given in FIG. 4 denote the ratios of neutron counts derived from the t21 distribution of delayed double coincidences (N.sub.D) as given in the FIG. 4 to the neutron counts derived from the t32 distribution of delayed triple coincidences (N.sub.T) in the same measurement as referenced above with respect to FIG. 3. These numbers quantify, in other words, the ratio of the effect count rates exploiting delayed double coincidences only, or delayed triple coincidences only, wherein the effect count rates were both obtained from the same data set and using the same pulse pile-up reconstruction algorithm.

[0130] It is obvious that the delayed triple coincidences provide a much betterin these exemplary cases by about an order of magnitude bettereffect-to-background ratio for neutron counting than the delayed double coincidences, while the neutron contribution measured with delayed double coincidences is only about twice the neutron contribution measured with delayed triple coincidences. Nevertheless the neutron contribution measured with delayed double coincidences provides an additional and valuable measure for the neutron flux the scintillator crystal 20 is exposed to.

[0131] In a preferred embodiment of the invention the analyzer 50 measures the neutron flux the scintillator crystal 20 is exposed to from the neutron contribution derived from delayed double coincidences and from the neutron contribution derived from delayed triple coincidences, both contributions obtained from a pile-up re-construction algorithm that allows decomposing piled-up pulses and a subsequent event parameter analysis, thus increasing the accuracy of the neutron flux measurement.

[0132] FIG. 5 shows distributions of time differences t21 between a delayed signal component corresponding to an energy deposition of 138 keV and the primary signal component of delayed double coincidences, measured at different detector loads with a simulated trigger holdoff of 1.8 ?s, the analysis also including delayed triple coincidences, however, without being detected as such, as the first delayed signal component is hidden due to the trigger holdoff. thereby simulating prior art. All detected delayed coincidences are comprised in the same data as references above with respect to FIGS. 2 to 4.

[0133] In contrast to FIG. 4, the time distributions of FIG. 5 comprise all events exhibiting a delayed 138 keV signal, even if a preceding 30 keV signal was originally detected by the pulse pile-up reconstruction algorithm but ignored because of the simulated trigger holdoff. Also in contrast to FIGS. 3-4, the integration time range here starts at t.sub.1=1.8?, which corresponds to the trigger holdoff applied in Yakushev's measurement. The data shown in FIG. 5 thus represent t21 distributions and effect-to-background ratios E/B that would be measured under equivalent conditions with a setup according to prior art as described by Yakushev et al. The numbers N.sub.DY/N.sub.T denote the ratios of the neutron counts, achievable with a delayed double-coincidence analysis according to prior art, to the neutron counts obtained by delayed triple-coincidence analysis with the method under equivalent conditions as shown in FIG. 3.

[0134] The analysis of delayed triple coincidence events obviously provides a much better, by more than an order of magnitude better, effect-to-background ratio than prior art, while the effect rates are rather comparable. Moreover, if delayed double-coincidence are not excluded but also considered separately as shown in FIG. 4, the effect rate with the method is about twice that of the rate achievable with prior art, while the effective effect-to-background ratio is much higher.

[0135] Hence, FIG. 3 and FIG. 5 demonstrate clearly the advantage of the method and radiation detection system according to embodiments of the present invention based on delayed triple coincidences compared with the method described by Yakushev et al. based on delayed double coincidences by a direct comparison of effect-to-background ratios and effect rates using exemplary, consistent experimental data obtained under realistic conditions.

REFERENCE NUMERALS

[0136] 1 Radiation detection system
10 Gamma-ray detector
20 Scintillator crystal
30 Photodetector with an amplifier

40 Digitizer

50 Analyzer

[0137] E/B Effect-to-background ratio
N.sub.D Number of detectible neutron counts based on delayed double coincidences
N.sub.T Number of detectible neutron counts based on delayed triple coincidences
N.sub.DY Number of detectible neutron counts using prior art [0138] t1 Detection time of primary signal component [0139] t2 Detection time of first delayed signal component [0140] t3 Detection time of second delayed signal component [0141] t21 Time difference between a delayed signal component (138 keV) and the primary signal component [0142] t32 Time difference between the second delayed signal component (138 keV) and the first delayed signal component (30 keV)