Experimental setup

Neutron production and transport are typically accompanied by gamma rays resulting from the radiation capture effect [21, 46]. Neutrons and gamma rays typically coexist in the form of complex radiation fields. In contrast, isotope radiation sources or heavy-ion accelerators can provide relatively pure charged particles with well-defined energy distributions. We examined the waveform discrimination capabilities of GZO single crystals with respect to various charged particles by analyzing the excitation luminescence response time generated when different types of charged particles irradiate a GZO single crystal.

The dimensions of the GZO single crystal were 30 mm×30 mm×2 ×, fabricated using a hydrothermal technique with a carrier concentration of 1.07×10¹⁹ cm^-3 [56]. The luminescence response time was evaluated using the time-correlation single-photon counting (TCSPC) technique [76, 77]. The experimental setup is illustrated in Fig. 1 [78]. The particle emission source was housed in a dark box with the GZO single crystal and a two-way photomultiplier tube (PMT) to prevent the interference of external photons during the single-photon counting method. PMT1#, as shown in Fig. 1, was strategically positioned near the crystal to maximize collection photons as a zero-time signal. Subsequently, PMT2# was moved away from the sample to capture a random single-photon signal. The temporal correlation between the two PMTs signals was then recorded using an electronic acquisition system located behind the PMT. When the energy of the irradiated particle was insufficient, the zero-time signal generated by PMT1# exhibited instability, which could compromise the accuracy of the zero-time signal measurement. Consequently, the TCSPC system was reconfigured into a dual-channel single-photon counting (DCSPC) system to measure the lower-energy charged particles by moving the PMT away from the sample. In this system, an autocorrelation function for the fluorescence time spectrum is represented by the agreement between the signals of the two PMTs, which are both far from the sample. The decay time constant of the sample can be fitted from one branch of the autocorrelation function as a simple exponential function [75]. In the case of the DCSPC, the temporal resolution of the system exceeded 80 ps following calibration using a picosecond light source (EPL-375, Edinburgh; typical pulse width 60 ps).

Fig. 1Full-screen View

(Color online) Schematic representation of the experimental layout for the measurement of charged-particle excitation luminescence decay time

Five charged-particle sources were used in this experiment: α, β, H⁺,Li⁺, and O⁸⁺. Monoenergetic α particles with energies of 5.48 MeV were generated using an α ²⁴¹Am radioactive source, whereas β particles with energies ranging from 0 MeV to 1.173 MeV were emitted by a ¹³⁷Cs radioactive source. Protons (H⁺) with energies of 15 MeV, lithium nuclei (Li⁺) with energies of 42 MeV, and oxygen nuclei (O⁸⁺) with energies of 100 MeV were obtained using the HI-13 tandem accelerator at Chinese Institute of Atomic Energy (CIAE). The ions were guided from the vacuum tube through a 10 μM-thick titanium window in the accelerator tube and impinged on the GZO sample to generate excited luminescence after traversing 5 mm of air. The optical signals were acquired using two microchannel plate PMTs (MCP-PMT, Hamamatsu R3809U).

Pulse shape characteristic result

The time-resolved radioluminescence spectra of GZO excited with high-energy H⁺, Li⁺ and O⁸⁺ ions, as measured using the TCSPC method, are presented in Fig. 2(a). The corresponding radioluminescence spectra for low-energy α and β particles, obtained using the DCSPC method, are shown in Fig. 2(b). A statistical approach to the two channels resulted in the temporal distribution shown in Fig. 2(b), which was closer to a Gaussian distribution. The time constants were determined by fitting the data to a single exponential decay function, and the resulting radioluminescence response time constants for the five charged particles are given in Table 1.

Fig. 2Full-screen View

(Color online) (a) Excitation time spectrum for H⁺, Li⁺, and O⁸⁺. (b) Excitation time spectrum for α and β, where the insert curve is the α particle excited temporal profile obtained using the TCSPC method

Analysis of results

As shown in Fig. 2(a), the three types of charged particles generated by the HI-13 tandem accelerator exhibited distinct differences at the trailing edge of the excitation time spectrum, as measured using the TCSPC method. The excitation time spectra of the charged particles generated by the two isotope sources, obtained using the DCSPC method, exhibited an approximately Gaussian distribution. The β-particle time spectrum was characterized by a narrower (or sharper) peak, which significantly contrasted with the broader distribution observed for α particles. The data in Table 1 demonstrated significant differences in the GZO scintillation crystal decay time constants due to the excitation of the five charged particles. The results suggest that the response time of the GZO charged particles to irradiation is contingent on the dE/dx values of the irradiated charged particles. The variation in dE/dx indicates the energy dissipation of charged particles within the unit interval, which corresponds to the the ionization potentials of the particles. The β particles exhibited the lowest dE/dx values, whereas the O⁸⁺ ions exhibited the highest dE/dx values. Although the mass of Li⁺ is greater than that of α particles, the energy of Li⁺ ions was significantly elevated compared with that of α particles in this experiment, resulting in a lower dE/dx for Li⁺ ions relative to α particles. Therefore, the smaller the dE/dx value, the lower the ionization power of charged particles and more rapid the response of the excited GZO scintillation crystal. These findings suggest that GZO can discriminate between different charged particles, and this experimental approach offers a promising method for discriminating between charged particles.

Experimental Setup

The 600 kV nanosecond-pulse neutron generator (CPNG-6) at the CIAE can operate in the pulsed mode by implementing deuterium ion source cutting [79]. When the deuterium ion beam is accelerated and used to bombard a deuterium/tritium target, neutrons are generated via D-D or D-T nuclear fusion reactions, with a simultaneous neutron/gamma mixed-pulse radiation field. The ion beam cutting device has demonstrated excellent stability in emitting a half-width 5 μs deuterium ion beam at a working frequency of 1 Hz and generating a 5 μs wide pulsed neutron beam according to the principle of nuclear reactions. The radiation field can produce a composite neutron/gamma signal featuring a distinct neutron multiplication time spectrum. Initially, a sequence of mixed neutron/gamma signals was obtained using a scintillator detector, followed by waveform discrimination. Subsequently, based on the post-discrimination neutron signal, the neutron time spectrum of the pulsed neutron radiation field was reconstituted to validate the efficacy of the waveform discrimination.

A schematic of the CPNG-6 experiment based on a pulse mode is presented in Fig. 3. Upon the activation of the ion source chopper, a square wave signal was transmitted to the oscilloscope as an external trigger signal, thus synchronizing with the time base of the oscilloscope. The oscilloscope captured the voltage signals generated in channels 1 and 2 within the -1 to 9 μs time window. The cylindrical scintillation crystal was directly coupled to the PMT and housed within a stainless-steel shielding shell to protect against light and electromagnetic interference. To acquire precise signal details, a HDO8108A oscilloscope (Teledyne LeCroy) with a maximum sampling rate of 10 GHz and a 12-bit bandwidth was employed. The neutron detector and beam channel were situated on the same horizontal plane. To prevent high-flux neutrons in the straight channel from saturating the detector and forfeiting the single-particle waveform signal, the detector was offset from the central axis and target by specific distances, which were strictly recorded. In this experiment, a deuterium ion beam was used to impact a tritium target, generating 14.9 MeV monoenergetic neutrons. A second set of scintillation detectors was constructed on the opposite side of the detector relative to the beam, using a PMT (model ETL-9815) and commercial stilbene crystal (Φ50 mm×5 mm). A comparative experiment was conducted, and the results of neutron/gamma discrimination and the neutron pulse–time spectra were obtained. To optimize the measurement efficiency, a larger section of the GZO scintillator from the same batch with a diameter of Φ40 mm and thickness of 0.3 mm was employed.

Fig. 3Full-screen View

(Color online) Schematic representation of the layout of the neutron/gamma mixed-field experiment based on the pulsed mode CPNG-6

The experimental process and data treatment employed in this study were based on the current-mode counting method [79]. Each neutron/gamma pulse generated a current or voltage signal that was fully captured by a digital oscilloscope. Numerous single-particle signal waveforms were extracted from continuous signals using digital processing methods. The signal analysis extracted critical information, such as the amplitude, the charge value generated by the PMT, type of pulse, and the signal arrival time. The waveforms of typical pulses recorded by the oscilloscope during the experiment after exposing the two types of scintillators to a pulsed radiation field are shown in Fig. 4. The ordinate system was calibrated based on the impedance of the oscilloscope (50 Ω), and the voltage values were converted into current values, facilitating the subsequent calculation of the charge values corresponding to the individual signal waveforms. The initial signals generated by the commercial stilbene scintillator (Φ50 mm×5 mm) and GZO scintillator (Φ40 mm×0.3 mm) are shown in Fig. 4(a) and 4(b), respectively. The dashed black line represents the ion source open signal as the external trigger signal recorded by the pulsed mode oscilloscope. The neutron/gamma signal obtained by the GZO scintillator was faster, but smaller in amplitude. This was owing to the faster response time characteristics and lower irradiated light yield of GZO and the MCP tube.

Fig. 4Full-screen View

(Color online) Original waveform generated in (a) the stilbene detector, and (b) the GZO crystal

Pulse shape discrimination results for a single particle

Based on the composite neutron/gamma radiation field generated under the stated experimental conditions, 1800 pulse arrays, each with a duration of 10 μs, were collected. From these, 961 single-particle waveform signals for GZO were extracted, and the lower threshold was set to 5 mV. The most prominent and widely employed time-domain PSD algorithm in the field of neutron/gamma waveform discrimination, referred to as the charge integration method [69, 80], was initially employed to discriminate the pulses associated with the GZO single crystal. In this method, two distinct time intervals were selected to integrate the single-particle signal. Q_total was derived by integrating from a fixed time before the peak (denoted by t_re, where ‘re’ signifies the ‘rise edge’) to a specific time following the peak (t_end). Q_slow was obtained by integrating from t_re to a fixed time near the peak (denoted as t_fe, where ‘fe’ signifies the ‘falling edge’). Because the neutron pulse drop time was slower than that of gamma, and Q_slow constituted a larger portion of Q_total, Q_total/Q_slow was used for neutron/gamma discrimination. The integration intervals of the GZO waveforms in this study were established as -1 to 1 ns (Q_slow) and -1 to 12 ns (Q_total), respectively, with the peak position designated as the 0 time point. In the case of the commercial stilbene crystal, the integration intervals varied from -10 to 60 ns (Q_total), 10 to 60 ns (the difference between (Q_total), and 10 to 60 ns and Q_slow). The PSD results are shown in Fig. 5.

Fig. 5Full-screen View

(Color online) Neutron/gamma discrimination obtained using the charge integration method. (a) GZO (Φ40 mm×0.3 mm). (b) Commercial stilbene (Φ50 mm×5 mm)

GZO scintillation demonstrated the ability to distinguish between neutron/gamma single-particle waveforms, which are discernible to the naked eye. Compared with the commercial organic scintillator, the response time for the GZO scintillation single crystals was notably faster, and almost no pulse stacking was observed. However, the detection efficiency was relatively low, resulting in fewer neutron/gamma single-particle signals. After discrimination, the single neutron particle pulses captured by GZO accounted for 77.93% of the total single particles. For the organic scintillator, the neutron contribution was 59.85%, excluding the stacking pulse waveform. The disparate ratios of the neutrons can be attributed to two factors: (i) the disparity in the neutron/gamma detection efficiency of the two types of scintillators, and (ii) the varying crystal thickness, where that of the GZO scintillator was thinner than that of stilbene. Consequently, the detection efficiency of gamma rays with a superior penetrating ability was comparatively lower for GZO.

We also applied an established frequency-domain discrimination algorithm [81, 82], the fast Fourier transform (FFT) [83], for pulse discrimination. Typical results derived from the GZO experimental data are shown in Fig. 6. Following normalization, application of FFT, and ordinal summation of the low-frequency components associated with the single-particle pulse, the discrimination coefficient was derived corresponding to the ordinate in Fig. 6(a) and the abscissa in Fig. 6(b). The data presented in Fig. 6(a) represent 961 individual particle signal pulses, individually plotted as the coordinate coefficients, where neutron/gamma discrimination was executed through the use of the discrimination coefficient size. In Fig. 6(b), the the discrimination coefficients of the 961 single-particle signals were partitioned into intervals using the operating principle of the multichannel analyzer [46, 84, 85], with the interval counts provided; two near-Gaussian peaks were observed. Utilizing FFT to discriminate the single-particle pulse acquired by GZO enabled an effective neutron/gamma discrimination outcome.

Fig. 6Full-screen View

(Color online) Neutron/gamma discrimination results for GZO obtained via FFT. (a) FFT discriminating coefficients for neutron/gamma distribution. (b) Statistical plot of the neutron/gamma FFT resolution coefficient

However, the PSD results for the GZO scintillator revealed certain drawbacks compared to the commercial organic scintillator. The edges of the two types of peaks shown in Fig. 6(b) lacked smoothness, the peaks were asymmetrical, and the relative distance between the peak positions was insufficient. The single-particle neutron waveform utilizing the frequency-domain discrimination algorithm based on FFT accounted for 82.21% of all single particles. This result corresponds to less than a 5% deviation from the classical time-domain algorithm, indicating that the processing outcomes of both algorithms were reliable. These findings suggest that the classical frequency-domain methodology exhibits a degree of applicability in discriminating neutrons/gamma for the GZO scintillation crystal. However, owing to the constraints imposed by the performance of the scintillation crystals, achieving the same level of discrimination as commercial organic scintillators remains a challenge.

Reconstruction of pulsed neutron radiation-field information based on PSD

C.1. Radiation-field information reconstruction method

The theoretical basis underlying information reconstruction using pulsed radiation field measurement outcomes has been discussed in detail in a previous publication [79]. The radiation-field reconstruction approach predicated on pulsed-radiation-field measurement results is termed the current-mode counting method. Following neutron/gamma pulse PSD, the digital neutron single-particle waveform was carefully selected from the mixed field. The extraction of multi-dimensional radiation information was performed on each single-particle waveform. Finally, the radiation-field information was reconstructed based on comprehensive multi-particle data. The energy spectrum [86-88] and neutron multiplication time spectrum are the most crucial factors within the radiation-field information associated with pulsed neutrons [89-91].

For radiation fields with long neutron flight distances, the neutron time-of-flight method [92-94] offers an accurate treatment of the energy spectrum. Once a correlation is established between the detector output signal amplitude (or deposited charge) and the neutron energy, the spectrum can be reconstructed using either the neutron pulse height spectrum (PHS) [40, 47] or charge height spectrum (QHS) [79]. Based on the neutron/gamma discrimination results, the amplitude data of the selected single-particle pulse were systematically enumerated from the minimum to maximum values in accordance with the operational principle of the multichannel analyzer [84, 85]. The number of single particles was determined according to a pre-defined interval of change, enabling the derivation of the neutron PHS. Similarly, the pulse of a single particle (inset of Fig. 4), when integrated along the abscissa, can be used to determine the charge value deposited on the particle detector. A subsequent application of the multichannel analysis method enabled the derivation of the neutron QHS.

The neutron multiplication time spectrum, also known as the neutron time spectrum, describes the neutron multiplication process as a function of time and can be employed to characterize the nuclear reaction process and total reaction quantity in fusion and fission research [95, 96]. The continuous signal derived from each reaction pulse, as illustrated in Fig. 4, can be used to determine the appearance time of a neutron signal after neutron/gamma discrimination. This information facilitates the extraction of time-related details related to the emergence of a sequence of neutrons during the time-spectrum analysis. The window-by-window statistical algorithm (WWS) or time statistics algorithm based on the wavelet packet transform (WPT) [79] can be employed to identify the time at which the neutron signal appears. The neutron time spectrum can then be reconstructed.

C.2. Neutron amplitude spectrum reconstruction results

Employing the discrimination outcomes of the charge integration method illustrated in Fig. 5, 740 neutron signals were extracted from the 961 single-particle pulses. The amplitude of the pulse was directly extracted from the digitally recorded signal using the oscilloscope, followed by an amplitude statistical analysis of the 740 neutron signals. Based on the voltage amplitude of the neutron pulse signal, the PHS was segregated into 50 lanes ranging from low to high, the result of which is shown in Fig. 7(a). The solid red squares represent the number of neutrons corresponding to a specific pulse height, as measured by the GZO crystal, after normalization for each reaction pulse. The error bars indicate the uncertainty (68% confidence level) associated with the neutron count in the channel address at this height. The height spectrum resulting from the conversion of the normalized count to logarithmic coordinates is shown in the inset of Fig. 7(a). The neutron PHS shown in Fig. 7(b) was derived from a total of 6453 neutron signals using the commercial stilbene crystal, where the blue data points represent the number of neutrons corresponding to the specific pulse-height channel address normalized for each reaction pulse. The neutron PHS obtained using stilbene was analogously partitioned into 50 channels; the error bars indicate the measurement uncertainty (68% confidence level).

Fig. 7Full-screen View

(Color online) Neutron PHS obtained using GZO and commercial stilbene crystals. (a) PHS obtained with GZO. (b) PHS obtained with stilbene

The total deposited charge in the detector corresponding to each neutron single-particle signal was determined by integrating the neutron single-particle pulse signals obtained using GZO and commercial stilbene crystals with respect to time. This process was repeated for 50 channels, generating a neutron QHS for both GZO and commercial stilbene crystals, as illustrated in Fig. 8.

Fig. 8Full-screen View

(Color online) Neutron QHS obtained using GZO and commercial stilbene crystals. (a) QHS obtained using GZO. (b) QHS obtained using stilbene

The commercially available stilbene crystal exhibited superior neutron detection efficiency in both the voltage height spectrum and QHS. The neutron detection efficiency of commercial organic scintillators is approximately 8.44E-2 n/pulse, whereas that of the GZO crystal is one order of magnitude lower (8.06×10^-3 n/pulse). Absolute or relative counts obtained from the statistical analysis reflected the probability distribution of various energy signals generated by the reaction between neutrons and scintillators. The amplitude spectra presented in Figs. 7(b) and 8(b) revealed a rectangular distribution for the commercial stilbene system, which became particularly evident as the voltage or charge varied. In addition, a distinct shoulder peak was observed. The experiment employed monoenergetic neutrons with energies of 14.9 MeV generated by D-T fusion, and the amplitude spectrum distribution was in line with the characteristics of the elastic scattering between monoenergetic neutrons and H atoms and the rectangular distribution of the recoil proton energy [96]. The neutron PHS and QHS measured using the GZO crystal exhibited an asymmetric Gaussian distribution characterized by a peak followed by a decline in the e-index without evidence of a flat interval. Transient peaks were observed at 38 mV and 53 mV in the PHS for GZO (Fig. 7(a)) and at 1.9 pC and 2.2 pC in the QHS (Fig. 8(a)). We propose that the collisions between 14.9 MeV fast neutrons and ¹⁶O nuclei in GZO can result in various responses, including elastic scattering, (n,α), (n,p), (n,d), (n,nα), and (n,np) [95, 97]. The two peaks observed can be attributed to the reactions of (n,α) and (n, nα), whereas the predominant peak in the low-energy region may result from a combination of several factors, including the contribution of low-energy neutrons interacting with material atoms following elastic scattering, inelastic scattering, and reactions involving 14.9 MeV neutrons, as well as the influence of system noise.

The results show that, following neutron/gamma discrimination, the neutron QHS or PHS can be obtained using GZO. The interaction cross section between fast neutrons and ¹⁶O is significantly intricate. If the objective is to achieve a comprehensive resolution of the charge amplitude spectrum, or even a reconstruction of the neutron energy spectrum, considerable additional effort will be required. The application of the amplitude spectrum is not further discussed in this paper because the neutron energy spectrum could not be directly obtained from Fig. 7(b) or Fig. 8(b), even for the commercial stilbene crystals.

C.3. Pulsed neutron time spectrum reconstruction results

In this experiment, the distance between the front end of the detector and the target head of the accelerator was L=50 cm (projected length in the north–south direction) and d=10 cm (projected length in the east–west direction). Owing to the short flight distance of neutrons, any broadening of the neutrons in the detector time spectrum caused by the flight time could be ignored and the arrival time of the neutron signal after neutron/gamma discrimination was directly used for neutron time spectrum reconstruction. The WPT [79] was employed to extract the neutron time structure with the CPNG-6 device in pulsed mode, which was recorded using both GZO and the commercial stilbene crystal. For comparison, the results obtained using GZO and the commercial stilbene crystal are shown in Fig. 9. The abscissa represents absolute time. The red time spectrum refers to GZO (with the associated ordinate in red), and the commercial stilbene spectrum is indicated by the dashed blue line (with the associated ordinate in blue). The critical spectral parameters are listed in Table 1. The relative discrepancy between the obtained results and the theoretical value (5 μs) is indicated in parentheses. Moreover, the comparative errors between the measured outcomes for the two crystals can be assessed using the values in Table 2.

Fig. 9Full-screen View

(Color online) Neutron time spectra obtained using GZO and commercial stilbene crystals

According to the operational principle of the CPNG-6 pulsed neutron source, the predicted neutron burst time spectrum profile for the pulsed radiation field should exhibit a square waveform with a width of 5 μs. The results obtained for GZO and commercial stilbene crystals were essentially in line with the theoretical values. The stilbene measurement revealed a 5.40% deviation from the theoretical value, with a smaller base width value relative to the GZO measurement. The full width at half maximum (FWHM) of the time spectrum for GZO was reduced by 0.4% when compared with the theoretical value, representing an improvement over the stilbene measurement. Considering the neutron burst time point, the relative error for the two crystal results was 2.05%. The detection efficiency of GZO at the neutron burst peak was ca. 0.074 n/μs, which was 11 times smaller than that of the commercial stilbene crystal (0.847 n/μs). The detection efficiency of the commercial stilbene crystals was higher during the interval in which the neutron burst-peak plateau occurred, leading to a smaller statistical uncertainty in the measurement and a smoother plateau curve. Based on the detection efficiency, the number of neutrons collected by GZO was smaller, resulting in a more pronounced waveform oscillation in the plateau curve. However, pulsed neutron radiation fields with a relatively straightforward time structure were detected effectively using a system based on GZO that obtained a pulsed neutron time spectrum by measuring the bottom width or half maximum width of the time spectrum.