Feasibility study on signal separation for spontaneous alpha decay in LaBr3:Ce scintillator by signal peak-to-charge discrimination

A novel analysis method named peak-to-charge ratio (V p / Q total ) discrimination, aiming at background rejection especially for alpha decay self-activity in LaBr 3 :Ce scintillators has been developed. This method is based on a waveform analysis using the peak-to-charge ratio in the output waveform of a photomultiplier tube. The discrimination of alpha-induced events was achieved by using a threshold function based on the error propagation of the V p / Q total value. The accidental rejection ratio of gamma-induced events was evaluated to be 0.17%. Furthermore, a total absorption peak spectrum processed with the V p / Q total discrimination method for 68 Ga 1.883 MeV gamma rays, where the energy was overlapped with background alpha events, reproduced exactly the same result as that of the background subtraction method. The di ff erence in measured peak counts of both methods was 0.716%, and the statistical error in the V p / Q total discrimination method and background subtraction was 4.81% and 8.70%, respectively. Thus a higher-accuracy measurement could be achieved using the V p / Q total discrimination method. The present study demonstrates that the V p / Q total discrimination method is a promising method for background rejection of the spontaneous alpha decay in LaBr 3 :Ce scintillators. C 2015 AIP Publishing LLC. [http: // dx.doi.org / 10.1063 / 1.4928115]


I. INTRODUCTION
Recent developments of inorganic scintillators have been remarkable in the field of radiation measurement. In particular, the LaBr 3 :Ce scintillator yields excellent energy resolution (<3% at 662 keV) with intrinsic time resolution of about 107 ps. 1,2 These characteristics are useful for gamma ray spectrometry, 3 development of gamma cameras, 4 and time of flight measurements. 5 According to previous studies, LaBr 3 :Ce scintillators have also been used for lifetime measurement of unstable nuclei 6,7 and observation of the response to fusion neutrons. 8 Furthermore, LaBr 3 :Ce scintillators have large stopping power in spite of their relatively lower effective atomic number, since it is possible to make a large crystal with high density (5.3 g/cm 3 ). [9][10][11] They are also suitable for high energy gamma ray measurements (e.g., cosmic ray, characteristic gamma ray due to nuclear reactions 12 and environmental radiation 13 ). Additionally, fluorescence characteristics of LaBr 3 :Ce scintillators have only a small dependency on thermal changes 14 and have strong radiation-resistance. 15,16 In recent years, CeBr 3 scintillators with high energy resolution (4.0% at 662 keV) and fast timing resolution of 119 ps with a self-activity much lower than the LaBr 3 :Ce scintillator have also been developed. [17][18][19] However, the characteristics of CeBr 3 scintillators are slightly inferior to LaBr 3 :Ce scintillators at present. Yet, self-activity of LaBr 3 :Ce crystals is considered to be a serious defect in LaBr 3 :Ce scintillators. 20 The major components of this self-activity are 1436 keV gamma rays from electron capture disintegration (66.4%) of 138 La nuclei with 32 keV K-X ray of Ba atom, and 789 keV a) Electronic mail: masayori@med.hokudai.ac.jp gamma rays with 255 keV electron from beta decay (33.6%) of 138 La nuclei. Moreover, the self-activity background due to alpha decay from Ac-series nuclei is distributed at a range of 1.5-3 MeV in the energy spectrum and a characteristics gamma ray induced from the alpha decay is also emitted almost at the same time. Much research has been carried out on particle identification of alpha-decay nuclei using coincidence measurement of alpha-decay events and its induced characteristic gamma ray. 1,20 This inevitable background can be negligible if the count rate of true events is sufficiently larger than self-activity background; however, it is difficult to observe the true event when a low count rate is taken into account. In general, large volume LaBr 3 :Ce crystals are used for high energy gamma ray measurements, but the event rate of self-activity increases with crystal volume. Therefore, research on the self-activity suppression carries a significant importance for the development of large volume LaBr 3 :Ce crystals.
Research by Hoel et al. 21 and Crespi et al. 22 suggested that there exists a difference between alpha and gamma induced events in the vicinity of the peak voltage in the output waveform of a photomultiplier tube (PMT) equipped with a LaBr 3 :Ce scintillator. However, to the best of our knowledge, almost no research has been done in the field of pulse shape discrimination (PSD) for rejected alpha-decay events. This is because the waveform difference obtained is almost imperceptible. In research by Crespi et al., 22 self-activity rejection of alpha-decay was achieved by comparing the total integrated charge and the partial integrated charge in a suitable time domain in the vicinity of the peak voltage. In the present work, the domain of integration and threshold constant in the whole energy range have been determined discretionally, and a post hoc analysis is necessary for the optimization of these parameters. Furthermore, there is a possibility that the rejection accuracy of alpha-induced events depends on the domain size of the charge integration, as the threshold for the discrimination of alpha and gamma induced events changes with this size. Therefore, the signal peak-to-charge ratio V p /Q total determined uniquely in all waveforms was noted in this work. Generally, Q total is proportional to V p and for gamma ray events at adequate gamma ray energy region, 23,24 while the V p /Q total value is expected to be a constant. However, we observed different values between alpha and gamma induced events. By using this difference, we developed an alpha-event discriminating method dedicated for LaBr 3 :Ce scintillators using the V p /Q total value. In this paper, we will report the V p /Q total discrimination method and demonstrate the effectiveness of an LaBr 3 :Ce scintillation detector in gamma ray spectrometry.

A. Data acquisition system and noise reduction
As shown in Figure 1, the electric signal waveforms from a PMT (R6231-100, Hamamatsu Photonics) equipped with a LaBr 3 :Ce scintillator (38S382/B380, BrilLanCe™ Saint-Gobain) were measured using an oscilloscope (Wave Runner 64xi, LeCroy, 5 GS/s) without electronic devices such as a shaping amplifier. The raw waveforms were stored to an external storage device. The values of V p and Q total were obtained from the analysis of the measured waveforms using in-house software which has several functions such as offset correction, a moving average for waveform smoothing, a frequency analysis using Fast Fourier Transform (FFT), a digital low pass filter using FFT with inverse FFT (IFFT), and spectrum generation for all measured data. In particular, the function related with noise suppression is important for the V p /Q total value, since the V p is sensitive to high-frequency noise. Therefore, the high-frequency noise was suppressed using a moving average with a time width of 2 ns, and a digital low pass filter using FFT and IFFT suppressed the noise component of >50 MHz in the entire waveform.

B. Alpha-event discrimination using V p / Q total
Although the background suppression method for selfactivity has been discussed in previous research, 22 it is still difficult to assess the rejection accuracy quantitatively. This is because a threshold for discriminating alpha and gamma events could not be determined theoretically as described in Section I. In the present work, the peak-to-charge ratio V p /Q total is expected to be constant for gamma events, and a different constant value is expected for alpha events. However, the V p /Q total value fluctuated according to the uncertainties of V p and Q total values, and a function of σ V p /Q total (Equation (1)) deduced from an error propagation of V p /Q total value was therefore used for discriminating alpha and gamma events. Since the σ V p /Q total function varies according to the supplied voltage to the PMT, the parameters for the function should be obtained as a calibration. This process requires a gamma ray source. However, LaBr 3 :Ce self-activity generates a gamma ray emission statistically sufficient for the calibration, (1)

C. Evaluation of alpha event rejection using V p / Q total discrimination method
In order to confirm the efficacy of alpha-event discrimination, a quantitative assessment was performed using an external gamma ray source. Since the energy range of background alpha-decay is distributed from 1.5 to 3 MeV (gamma equivalent), a Ge-68/Ga-68 source (CS-6-14, 12.4 kBq, Siemens Medical Solutions) which emits 1.883 MeV gamma rays from 68 Ga was used as the external gamma-ray source. The rejection accuracy of alpha-induced events was assessed by comparing the energy spectrum between the V p /Q total discrimination method and conventional method (background subtraction). The 68 Ga source was placed 0.6 m away from the LaBr 3 :Ce scintillation detector. Waveforms of 60 000 events were used for the comparison.

A. Noise reduction to improve measurement accuracy
The raw waveform data at a detection energy of 511 keV and the waveform of high-frequency noise suppressed by a moving average with a time width of 2 ns and 50 MHz digital low pass filter using FFT with IFFT are shown in Figures 2(a.1)-2(a.3). Since components of instantaneous noise affect the raw waveform, the values of V p obtained from raw data was about 4.7% larger than noise-suppressed waveform of 511 keV gamma ray, and in the case of V p obtained from 1.883 MeV gamma ray which is emitted form 68 Ga was 0.71% larger. The signal to noise ratio in the V p value improves with increasing the injected gamma ray energy. The value of V p obtained from the moving average and digital low pass filters were almost identical since high-frequency noise was adequately suppressed. On the other hand, no significant difference in the total integrated charge Q total was observed among the raw data and the two smoothing methods. Figure 2(b) shows the energy spectra consisting of about 10 5 Q total data with and without noise suppression. An almost identical energy spectrum was obtained. This supports the fact that the Q total value is not affected by the noise reduction technique using a low pass filter.
B. Demonstration of the V p / Q total discrimination method using LaBr 3 :Ce self-activity Figure 3(a.1) shows the energy (Q total equivalent) dependence of the V p /Q total value for self-activity of the LaBr 3 :Ce scintillator. Two components of the V p /Q total value at the energy range of 1.5-MeV were observed. According to Crespi et al., the waveform of alpha-induced events contains a higher value of peak intensity than the gamma-induced events in the same detection energy. 22 Hence, the alpha-induced events should be distributed at a higher value than the gammainduced events. For the lower component of V p /Q total , it is considered to be a gamma-induced event from environmental radiation such as 208 Tl (2.615 MeV). On the other hand, the V p /Q total value in Fig. 3(a.1) decreases with increase in energy, although V p /Q total was expected to be a constant. The reason for the decline seems to be due to saturation of V p as shown in Figure 3(b.1). The linearity between V p and Q total was only kept in the low energy region. In order to correct the saturated V p , the saturation curve of V p was assumed to be in the form of Equation (2) which consists of a linear term αQ total (α = const.) and a saturation term 1 + βQ total ( β = const.). Equation (3) was deduced by assuming that the corrected peak voltage V pCorr is proportional to Q total such as V pCorr = αQ total , Note that these parameters α and β which are necessary for the saturation correction were obtained by a fitting analysis with Equation (2) in the experimental value without alphainduced events. The coefficient of correlation R 2 between the measurement value and Equation (2) was found to be 0.9976, which shows a strong correlation between them. Figs. 3(a.2) and 3(b.2) show that the saturation was corrected properly using Equation (3). The optimized threshold function on the V p /Q total value was configured theoretically using its standard deviation of σ V p /Q total . As mentioned for Equation (1), σ V p /Q total is a function of Q total in approximate inverse proportion. Threshold functions of 1 to 3 σ V p /Q total were obtained by the calibration using a fitting analysis with σ V p /Q total = kQ −1 total + l (k = const., l = const.). The fitting curves adequately expressed their measured values as shown in Figure 4(a). In this analysis, since the component of alpha-induced events ranged from 1.5 to 3 MeV, only the data lower than 1.4 MeV were used. Figure 4 V p /Q total scatter plot in Figure 3(a.2), alpha-induced events seem to be adequately discriminated using the threshold function. Figures 5(a.1)-5(a.3) shows the energy spectrum with and without alpha event discrimination in LaBr 3 :Ce self-activity using the present method. It is shown that the alpha-decay event can be discriminated effectively. On the other hand, the accidental rejection of gamma-induced events occurred by the threshold function as shown in Figures 5 . Accidental gamma-rejection depends on the threshold function; the probability of gamma-rejection for 1, 2, 3 σ V p /Q total was evaluated as 4.661%, 0.809%, and 0.172%, respectively, in the energy range of <1.5 MeV where alpha-induced events are not contained.

C. Assessment of alpha-discrimination accuracy
In the present experiment, the rejection accuracy of alphainduced events was assessed using an external radiation source ( 68 Ga 1.883 MeV). Figures 6(a) and 6(b) show the energy spectrum for the background subtraction and V p /Q total discrimination methods, respectively. The energy spectrum without radiation source was normalized by the integrated event counts in the energy range of >2 MeV for background subtraction. As shown in Figure 6(b), only the 68 Ga 1.883 MeV gamma ray peak remained clear by using the 3 σ V p /Q total threshold function. Similarly, Figure 6(c) shows the enlarged energy spectra dedicated for 68 Ga 1.883 MeV gamma rays for detailed comparison. The V p /Q total discrimination method accurately reproduced the result as that using the background subtraction method. In both methods, the difference in measured peak counts was 0.716%. And the difference in parameters obtained by the fitting analysis with a gauss function in the spectrum such as peak counts, mean energy, full width at half maximum (FWHM) was 1.694%, 0.021%, and 0.165%, respectively, as shown in Table I. Additionally, statistical errors in measured total peak counts for the V p /Q total discrimination method and subtraction method were 4.81% and 8.70%, respectively, at a range of ±3σ. A reduction in statistical error was thus observed using the former method.

A. Calibration of threshold function and correction for V p saturation
The V p /Q total value increases along with the supplied voltage of the PMT because the pulse width in the PMT declines with the rise in supplied voltage. Accordingly, the threshold function must be calibrated for each measurement condition especially for the supplied voltage. The actual supplied voltage must be taken into account for a successful calibration, but the V p /Q total difference between alpha and gamma events could be observed in the range of typical voltage (500-1000 V) for gamma-ray spectrometry using a LaBr 3 :Ce scintillator. Once the calibration is done for a specific supplied voltage, the stability and reproducibility of the threshold function is confirmed with the fixed supplied voltage. Moreover, the advantage of this calibration process FIG . 5. (a) Energy spectrums with and without alpha event rejection using each threshold functions in LaBr 3 :Ce self-activity. For the clarity of the present method effect, event counts increase by a factor of ten in the energy range of alpha self-activity. (b) The accidental gamma rejection rate in the V p /Q total discrimination method in the energy range of 1.5 MeV, where alpha events are not existent. The energy range without a data point means an accidental gamma rejection of 0%. This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: does not require an external radiation source because the LaBr 3 :Ce scintillator itself has self-activity. This makes the calibration process much easier and can be performed just before a measurement. In Figures 3(a.1) and 3(b.1), the saturated V p was seen as a function of Q total . As the V p decreases with increase in the pulse width of PMT output due to the space charge effect (increase in the electron transit time spread in the PMT), the saturation becomes more obvious in higher supplied voltage. The analytical solution of this saturation curve is extremely complex; however, the saturated V p fitted using Equation (2) could reproduce exactly the experimental values.

B. System dead time of the data acquisition
In an energy range of about 2.6 MeV, the contribution of alpha events from 215 Po in the LaBr 3 :Ce crystal should be observed. However, the contribution was not seen in the energy spectrum in Figure 2(b). The alpha decay of the 215 Po with shorter lifetime of 1.785 ms occurs immediately after the disintegration of the parent nuclide such as 219 Rn. The reason for the unrecorded alpha event from 215 Po is due to long dead time of the present data acquisition system. The acquisition system requires at least 50 ms for storing measured waveform, and this latency rises to about 200 ms with increasing a number of stored waveform to the external hard drive. The dead time for this data acquisition system is difficult to measure precisely due to low reproducibility. However, event losses for alpha and gamma are not essential in discriminating alpha event because the event loss occurred not only for alpha event from 215 Po nuclei but also following events. For that reason, it was difficult to estimate the real measured time. We therefore used event counts for normalization to compare the subtraction method and V p /Q total discrimination methods in Section III C.

C. Detection of low-rate environmental gamma ray
As shown in Figure 7, the contribution of 208 Tl (2.614 MeV) gamma rays was clearly detected. This means that Thorium series nuclei exist in the measurement environment. In the raw data, 214 Bi (1.764 MeV) of Uranium series could not be identified; however, they were clearly identified using the V p /Q total discrimination method with the 3 σ V p /Q total threshold function. On the other hand, 212 Bi (1.620 MeV) of Thorium series nuclei could be observed in both cases of raw data and alpha rejection as shown in Fig. 7. For the subtraction method, the actual background spectrum also depends on the surrounding materials and it would therefore be quite difficult to eliminate from it the ubiquitous presence of 212 Bi. With a background spectrum containing 212 Bi peak, it will thus be difficult to identify 212 Bi in the actual measurement. Therefore, the V p /Q total discrimination method can be considered a more practical approach for surveying environmental radiation.

D. The V p / Q total discrimination method extensibility
Although the V p /Q total discrimination seems to be categorized in the PSD, the use of this method is not restricted only for alpha discrimination in LaBr3 self-activity. The PSD method is applied in discriminating neutron and gamma event using a plastic scintillator. Rise time to fall time ratio is analyzed by a partial integration window in the traditional method. Recently, oscilloscope and flash analog-to-digital FIG. 7. The energy spectrum of environmental radiation obtained by using the LaBr 3 :Ce scintillator with the V p /Q total analysis (3 σ Vp/Q total threshold function).
converter (ADC) driven in GHz are available. Time resolution in few picosecond can thus be obtained. The V p /Q total discrimination method is expected to be simpler and more accurate than the conventional PSD methods. For our future work, we are planning to make a hardwarebased V p /Q total discrimination system using a flash ADC (ADC12J4000, Texas Instruments). The flash ADC can acquire 4 Giga sampling per second, implies a time resolution of 250 ps, and intrinsic latency of about 200 ns. However, the current V p /Q total discrimination method applies FFT/IFFT for noise reduction, which might take a few millisecond for calculation. Making the sampling rate lower is a possible solution for the realistic implementation of V p /Q total discrimination method, but discrimination accuracy may be worse depending on the time resolution because the lower sampling rate makes the peak voltage lower.

V. CONCLUSION
In the present research, alpha-decay events of the LaBr 3 : Ce scintillator self-activity were discriminated by analyzing the peak-to-charge ratio V p /Q total of the PMT-output waveform. The V p /Q total discrimination method has demonstrated only small accidental gamma rejection, and the energy spectrum after alpha-rejection could be reproduced exactly as that of the background subtraction method with higher-accuracy. The presented method uses measurable parameters (V p and Q total ) using flash ADC, and the entire process can be implemented with an electric circuit. The authors strongly believe that the V p /Q total discrimination method will be used as one of the standard techniques for gamma ray energy spectrometry using LaBr 3 :Ce scintillators especially for environmental gamma ray surveys.