Pulse-shape discrimination potential of new scintillator material: Ce:La-GPS

Ce:(Gd,La)$_2$Si$_2$O$_7$ (Ce:La-GPS) is a new scintillator material with high light output, high energy resolution, and fast decay time. Moreover, the scintillator has a high temperature ($\sim$150$^\circ$C) tolerance and is non-hygroscopic in nature; thus, it is especially suitable for underground resource exploration. Particle identification greatly expands the possible applications of scintillator. For resource exploration, the particle identification should be completed in a single pulse only. The pulse-shape discrimination of the scintillator was confirmed. We compared two methods; a different gate method and a digital filter method. Using digital filter method (shape indicator), the F-measure between $\alpha$ and $\gamma$ particles was obtained to 0.92 at 0.66 MeVee.


Introduction
Recently, Ce:(Gd,La) 2 Si 2 O 7 (Ce:La-GPS) with good scintillation property was developed and reported [1][2][3][4]. This scintillator crystal has become attractive due to its characteristic properties such as high light yield of over 40,000 photons/MeV and good energy resolution of approximately 4-5 % FWHM at 662 keV γ-ray. The emission spectrum has a peak at 390 nm, which is a very good match to the sensitive wavelength of typical photomultiplier tubes (PMT). In addition, the decay time constant, which is important for high rate measurement, was less than 100 ns, which is relatively short compared with other inorganic scintillator crystals. The scintillation properties of La-GPS are summarized in Table 1 [5].
Since Ce:La-GPS also possesses non-deliquescent properties and a huge light output up to 150 • C, it is expected to be applied in underground resource exploration. Uranium deposit and oil fields are rich in natural radioactive elements of uranium chain and thorium chain. Ce:La-GPS is considered as an alternative to other scintillators, such as NaI(Tl) and Ce:Gd 2 SiO 5 due to its good temperature resistance and non-deliquescent properties.
We measured the pulse-shape discrimination (PSD) ability of (Ce 0.005 ,La 0.245 ,Gd 0.75 ) 2 Si 2 O 7 for γ-ray and α-ray using a single crystal grown by the Czochralski (Cz) process. PSD makes it possible to distinguish γ-ray and α-ray and individual α-ray measurement from uranium or radon, which may provide additional information with regard to resource exploration. Moreover, PSD can also enable the reduction of background contamination, and there is a possibility to find applications of Ce:La-GPS in environmental measurement or in astroparticle physics experiment (e.g. double beta decay of 160 Gd).

Experiment
The Ce:La-GPS scintillator used was a 1 cm 3 cube. Pulses were observed for the α-ray source ( 241 Am, 5.49 MeV and 5.44 MeV) and γ-ray source ( 137 Cs, 0.662 MeV) by a photomultiplier (R9869   Figure 1. The α source was placed on the scintillator because the α-ray stops at a short range. Since the internal α particle could not be realized, thus we only considered the external α-ray events. The γ-ray source was placed several centimeters away to avoid a pile-up signal. Both sources were strong; hence, we ignored the effects of cosmic muons and the environmental γ-ray. The peaks arising from these effects were not observed.

Energy calibration and reference pulses
A total of 11,009 and 10,291 events were recorded for γ and α sources, respectively. The obtained spectra are shown in Figure 2. The energy was calibrated using the 0.66 MeV peak of 137 Cs γ-ray.
The alpha peak of 241 Am (5.49 MeV) was observed below its quenching factor of ∼ 20%. The energy electron equivalent calibrated by the γ source was used as the energy (e.g., MeVee) in the analysis described below. Using the 2,560 and 1,628 events whose corresponding energies were between 0.5 MeVee and 0.8 MeVee, reference pulses were calculated for α and γ, respectively. The reference pulses were constructed from the average waveform normalized by each integration and are shown in Figure 3.
There is a clear difference between the two reference pulses. First, we used a different gate method.

Different gate method
The different gates method is commonly used for PSD [6]. This method uses two different gates; one is a usual gate, which includes the total pulse, and the other is a short gate for the fast component of the pulse. The ratio of the two charge integrations (Fast/Total) clearly shows the difference in the waveform components. This method requires only two charge integrations in different gate widths. The waveforms were recorded, and thus, we tried to use the method in a pseudo manner. We defined the short gate time width as t s .  The results are shown in Figure 4 when t s is 100 ns. In the high energy region (∼ 0.7 MeVee), α and γ events were clearly distinguished. However, in the low energy region, these events were contaminated. Generally, the PSD is difficult for the low energy pulse.
This method requires the optimization of the gate width (t s ). To select the best cut criteria in each energy region and evaluate the efficiency of the PSD, we introduced an evaluation function, F-measure [7]. The events were defined as the follows; Then, F-measure was obtained as An example of the three evaluation functions is shown in Figure 5. Since we cannot use only the precision or the recall alone to evaluate the best cut criteria and the efficiency, the F-measure, i.e., the harmonic mean of the precision and the recall was used. The closer the F-measure value is to 1, the better the separation between α and γ events.
In each minute energy regions, we obtained the best cut criteria from the maximum value of F-measure and also considered the best gate, t s . Figure 6 shows the F-measure of the different gate  (t s ) method for each energy. The t s of ∼80 ns shows the best F-measure regardless of the energy region for Ce:La-GPS in this experiment. In this t s , F-measure > 0.8 was realized at 0.66 MeVee. This method requires only two different gates; thus, it is easy to analyze the PSD even online. However, Figure 6 shows that the gate should be determined for a clearer PSD.

Digital filter method
A digital filter method by a shape indicator (SI) is also used for the PSD [8]. The SI is defined as follows; . f (t i ) is the ith data point of an observed waveform. P(t i ) is defined by f α (t i ) and f γ (t i ), which are the referenced pulses of α and γ, respectively, and is given by In Figure 7, the calculated SIs are shown for each energy. Two clear clusters were observed. This method seems to be better than the different gate method. Figure 8 shows the event fraction in the SI for all events between 0.6 MeVee to 0.9 MeVee. Using SI = 0.0057, the best cut line for the events; F-measure was 0.96. This F-measure value depends on only this experiments (the number of events of α and γ particles). The F-measure should be evaluated for each energy region. The digital filter method is difficult in terms of discriminating between α and γ in the low energy region. We used the best cut criteria in each energy region. Then, the obtained F-measure was dependent on the energy. Figure 9 shows the F-measure of the two methods in each energy region. A great separation (F-measure = 0.92) in the digital filter method was observed at 0.66 MeVee for the photoelectric region of 137 Cs γ-ray. An underground resource exploration requires discrimination between α-ray from elements of uranium and thorium chain and environmental γray. Therefore, the result shows the excellent potential of the new scintillator material: Ce:La-GPS. In the low energy region, a fewer number of photons are generated. Thus, the PSD is statistically limited. In addition, it was difficult to distinguish them because of background contamination. The discrimination at the lower energy region can possibly be further improved.
For all energy region, the digital filter method using the SI realized better separation than the different gates method. However, the optimized gates method could discriminate α and γ events similar to the digital filter method.

Summary
We studied the PSD of new scintillator material, Ce:La-GPS, using the different gates method and the digital filter method (SI). The potential of the PSD was observed in both the methods. The digital filter method was found to be better than the different gates method. However, the optimized gate method also produced good results.
This study is a simple analysis by F-measure to evaluate the efficiency of the discrimination. For various purposes that require good precision or recall analysis, the optimized the PSD will be able to produce expected result, as we obtained an F-measure > 0.9 for 0.66 MeVee pulse.