X-ray-induced Scintillation Governed by Energy Transfer Process in Glasses

The efficiency of X-ray-induced scintillation in glasses roughly depends on both the effective atomic number Zeff and the photoluminescence quantum efficiency Qeff of glass, which are useful tools for searching high-performance phosphors. Here, we demonstrate that the energy transfer from host to activators is also an important factor for attaining high scintillation efficiency in Ce-doped oxide glasses. The scintillation intensity of glasses with coexisting fractions of Ce3+ and Ce4+ species is found to be higher than that of a pure-Ce3+-containing glass with a lower Zeff value. Values of total attenuation of each sample indicate that there is a non-linear correlation between the scintillation intensity and the product of total attenuation and Qeff. The obtained results illustrate the difficulty in understanding the luminescence induced by ionizing radiation, including the energy absorption and subsequent energy transfer. Our findings may provide a new approach for synthesizing novel scintillators by tailoring the local structure.

SiO 2 is equivalent to a change in the value of Z eff . Therefore, it is worthwhile to examine the PL and scintillation properties of Ce 3+ in this glass system.
The aim of this study is to investigate the relationship between the valence state of activators in the lithium borosilicate glasses possessing different Z eff values and the PL and scintillation efficiency. In order to discuss the valence state of cerium, the Ce 3+ ratio in the glasses is introduced and defined as the ratio of the Ce 3+ concentration to the sum of the concentrations for Ce 3+ and Ce 4+ . Based on several analytical data, we have found that there is an anomalous relationship between the scintillation properties and the chemical composition of glass.

Results
The chemical composition of the present glass system is xCe 3+ -40Li 2 O-yB 2 O 3 -(60-y)SiO 2 (in molar ratio), where an excess amount of Ce is added. Herein, the general glass system is abbreviated as xCe:LBSy. First, we examined several Ce-doped Li 2 O-B 2 O 3 -SiO 2 glasses in order to change the Z eff value. An increase in the amount of SiO 2 increases the value of Z eff , which determines the effective absorption of X-ray energy. The chemical composition and the nominal Z eff values of these glasses are listed in Table S1. Figure 1(a) shows the optical absorption spectra of 0.5Ce:LBSy glasses at room temperature (RT) for different values of y. Comparison of the absorption spectra for 0.5Ce:LBSy glasses with those of non-doped LBSy glasses ( Fig. 1(b)) demonstrates that most of the absorption is due to the addition of Ce. Furthermore, the shape of the spectra in Fig. 1(a) changes considerably with the value of y (i.e. the B 2 O 3 -SiO 2 ratio). On the other hand, Fig. 1(c) shows that the shape of the spectra slightly changes with the value of x (i.e. the Ce concentration) 27 . As shown in the inset of Fig. 1(c), when the chemical composition of LBSy is fixed, the optical absorption edge is slightly red-shifted with increasing amounts of Ce 3+ due to be a broadening of the tail, i.e. a local coordination change (see Fig. S1). However, as shown in the inset of Fig. 1(a), the absorption coefficient at the tail region is largely red-shifted with increasing SiO 2 fractions. Therefore, it is expected that the absorption shape depends on both parameters x and y. Since the absorption tail of Ce 4+ is observed at low energy regions 28,29 , it is assumed that the red-shift of the absorption tail is correlated with the generation of Ce 4+ species. Clear absorption bands are observed for LBS30 and LBS40 glasses. After a peak deconvolution using six absorption peaks with a half-width at half-maximum of approximately 2250 cm −1 , we found that the photon energy of each excitation peak is almost the same. The results suggest that the Ce 3+ coordination is almost the same for both glasses and that the activators are dispersed homogenously in the glass matrix. In order to examine the valence state, we measured Ce L III -edge XANES spectra of 0.5Ce:LBSy glasses, as shown in Fig. 2(a). These white lines change with the B 2 O 3 -SiO 2 ratio, especially near y = 10. The shape of the spectrum for the 0.5Ce:LBS40 glass is very similar to that of Ce(OCOCH 3 ) 3 ·H 2 O, as shown in Fig. S2, and noticeable differences for varying Ce concentrations are not observed ( Fig. 2(b)). We can, therefore, conclude that the valence state of almost all (>95%) Ce centres in these LBS40 glasses are Ce 3+ states, which is independent of the Ce concentration. Although precise fitting is difficult, the Ce 3+ ratio of these glasses can be evaluated by spectra deconvolution using the spectra of Ce(OCOCH 3 ) 3 ·H 2 O and CeO 2 . Using these two reference materials, the Ce 3+ ratios can be calculated as shown in Fig. 3. In the case of Ce:LBS30 and LBS40 glasses, the valence state of Ce is mostly the trivalent state. However, when the SiO 2 fraction increases, the Ce 3+ ratio decreases. It is notable that the XANES spectrum of the 0.5Ce:LBS10 glass is very similar to that of the 0.5Ce:LBS40 glass prepared in air (Fig S3), and that the Ce 3+ ratio is less than 40 %, although the preparation of 0.5Ce:LBS10 was performed in an Ar atmosphere. Figure 4 shows PL and PL excitation (PLE) spectra of 0.5Ce:LBSy glasses at RT. The wavenumbers of both the excitation and emission peaks of Ce 3+ for the present glass are lower than those in phosphate glasses 19,20 while higher than those in silicate glasses 20 . As the B 2 O 3 fraction decreases, both peaks are slightly red-shifted, i.e. a smaller excitation energy induces a smaller emission energy. This might be correlated with the behaviour of the optical absorption spectra shown in Fig. 1(a), in which the absorption tail red-shifts with decreasing B 2 O 3 fraction. Figure 5 shows contour plots of the PL-PLE spectra of 0.5Ce:LBSy glasses, where the PL intensity was normalized in order to understand the shapes of the spectra. The vertical and horizontal axes show the photon wavenumbers of excitation and emission, respectively. The fact that the excitation band is broad suggests that it is associated with the continuous excitation band, which is characteristic of Ce 3+ states. However, as shown in Figs 4 and 5, the spectrum shape of the LBS10 glass is quite different from the shapes of the spectra for other B 2 O 3 fractions. Irregularities associated with the LBS10 glass are also evident in the PL decay curves of xCe:LBSy glasses shown in Fig Table S2. The internal quantum efficiencies Q eff of xCe:LBSy glasses are shown in Table S3 and Fig. 7. The values of Q eff roughly depend on the Ce concentration and variations of Q eff are probably due to differences in the local coordination state. Figure 8(a) shows X-ray induced scintillation spectra of 0.5Ce:LBSy glasses obtained by and irradiation dose of 10 Gy. The scintillation intensities are normalized using the volume of the sample. We have confirmed that the scintillation spectra were unchanged during irradiation and that there is a linear relationship between the irradiation dose and the scintillation intensity ( Fig. S5(a) and (b)). Figure 8(a) also shows that emission peak wavenumbers of Ce 3+ red-shift with decreasing B 2 O 3 fraction, as was observed in the PL spectra. It is noteworthy that the emission peak area of the 0.5Ce:LBS10 glass is much larger than that of the 0.5Ce:LBS40 glass, although we have confirmed that many Ce species are oxidized into Ce 4+ during melting. In order to discuss the Ce 3+ ratio quantitatively, the values of Q eff , and the scintillation peak area (normalized to the peak area of the 0.5Ce:LBS40 glass) are plotted in Fig. 8(b) as a function of Z eff (bottom axis) and the B 2 O 3 fraction (upper axis). It is evident that the scintillation intensity is proportional to Z eff and inversely proportional to Q eff and the Ce 3+ ratio.

Discussion
We have found that the chemical composition of glass affects the valence state of the activator in glasses. The results clearly suggest that the average Ce 3+ ratio is affected by the chemical composition of glass, i.e. the macroscopic basicity of glass. In order to explain the results, we use the concept of the 'optical basicity' defined by Duffy 42,43 . Optical basicity, i.e. the average basicity of oxides in the glass, is a concept based on the polarization of electrons. The idea of basicity of glasses is sometimes useful for evaluation of the physical properties of bulk glasses. The optical basicity of Li 2 O, B 2 O 3 , and SiO 2 are reported to be 1, 0.42, and 0.48, respectively 43 . Therefore, when the optical basicity of glass increases by substitution of SiO 2 for B 2 O 3 , it is expected that an oxidation reaction of Ce 3+ into Ce 4+ occurs even in an Ar atmosphere. Since the starting materials of glass can affect the valence state of Ce cations 29 , it is not possible to reach a direct conclusion from the observed phenomena. However, an increase of the optical absorption in SiO 2 -rich glasses is expected to be brought about by a redox reaction transforming Ce 3+ into Ce 4+ . To the best of our knowledge, the physics of ionizing radiation is still unclear because of the complexity of the process. Therefore, research on scintillators is often conducted by focusing on specific parameters. Although Q eff is generally a useful parameter to develop scintillators, Z eff has been found to play a more dominant role for X-ray-induced scintillators 34 .
As mentioned above, an increase in the SiO 2 fraction causes an increase of Z eff , which in turn increases the effective absorption of X-rays. Figure 9 shows the X-ray-induced scintillation peak area of xCe:LBSy glasses as a function of the product ρ·Z eff 4 . Since the dopant concentration is less than 2 mol%, the density of glass, ρ, shown in Table S4 44 can be used for the discussion. With the exception of Ce:LBS10 glasses, in which a decrease in scintillation intensity is observed due to the strong self-absorption in the visible region, the scintillation peak areas are roughly dependent on the Ce concentration. Although the value of Q eff for the 0.5Ce:LBS10 glass is much lower than that of the 0.5Ce:LBS40 glass because of the generation of Ce 4+ species, the scintillation peak area of the 0.5Ce:LBS10 glass is higher than the peak areas of most 0.5Ce:LBSy glasses in Fig. S9. Figure 10(a) shows the total attenuation with coherent scattering of 0.5Ce:LBSy glasses, which was calculated using a previously published fomula 45 that takes into account the influence of Z eff . The energy spectrum of the X-rays used in the present study [46][47][48] is also shown in the figure with a scale given on the right axis. Here, the X-ray source is a conventional X-ray tube with a W target and a Be window. In the energy region of irradiated X-rays, the total attenuation of the 0.5Ce:LBS10 glass is the highest among the present samples. Moreover, the attenuation values without coherent scattering exhibit a similar tendency. Here, we determined the total absorption energy using the following expression: where ζ is the absorbed energy in the sample along the irradiation axis per unit area, E is the incident radiation energy, N 0 is the number of incident photons per unit area, μ T (E) is the total attenuation coefficient of the sample, μ EA is the energy absorption coefficient of sample, and t is the thickness of the sample. Figure 10(b) shows the total absorption energy, ζ relative , relative to that of the 0.5Ce:LBS10 glass. In the present X-ray energy region, the value of ζ for the 0.5Ce:LBS10 glass is approximately 1.2 times larger than that of the 0.5Ce:LBS40 glass. As mentioned above, the scintillation intensity I scinti is a product of the total absorption energy ζ and the scintillation efficiency η = β e-h ·S trans ·Q eff and is given by . .
Since we have no quantitative information about the values of β e-h and S trans , their product, (β e-h . S trans ), is treated as a coefficient that can be evaluated using I scinti , ζ relative , and Q eff , and represents the efficiency for generating electron-hole pairs followed by energy transfer to luminescent centres in each glass. Using the values depicted in Figs 3(b) and 4(b), we have found that the value of (β e-h . S trans ) for the 0.5Ce:LBS10 glass is more than 14 times larger than that of the 0.5Ce:LBS40 glass (see right axis of Fig. 4(b)). In other words, the absorbed X-ray energy is not converted into scintillation photons effectively in 0.5Ce:LBS40 glasses.
Plausible reasons for the low conversion efficiency are the physical parameters of non-doped LBS glasses shown in Table S4 44 . Since the molar volume of the LBS10 glass is smaller than that of the LBS40 glass, the network of the LBS10 glass is spatially denser than that of the LBS40 glass, i.e. there is larger free volume in the LBS40 glass. If there is no large difference in the phonon vibration energies of LBSy glasses, the free volume in the glasses may work as an attenuator and inhibit the effective energy transfer to activators. On the other hand, another reason for the low conversion efficiency is the storage mechanism of irradiated energy proposed by Yanagida 34 . It was reported that a B 2 O 3 -containing glass exhibits storage luminescence by X-ray irradiation 35 . Because the irradiated energy is converted into scintillation, storage luminescence, or thermal vibration (non-radiative relaxation), high storage luminescence means low scintillation. Considering that the origin of storage luminescence is defects in glasses, we speculate that there are many defects that affect the energy transfer process to activators in B 2 O 3 -rich glass. As shown in Fig. S6 and Table S5, there are only small differences in the band gaps for LBSy glasses and these differences cannot provide a plausible explanation for changes in the conversion efficiencies.
Recent studies have suggested that the fraction of Ce 4+ in scintillators has an effect on scintillation properties [49][50][51][52] and several of them claimed that coexistence of Ce 3+ and Ce 4+ is important for high scintillation efficiency [49][50][51] . However, if the coexistence of Ce 3+ and Ce 4+ was a critical factor for determining the intensity, the correlation between chemical composition and scintillation intensity, as shown in Fig. 3(a), would be quite different; i.e. Ce:LBSy glasses would exhibit similar intensities with the exception of the Ce:LBS10 glass. Therefore, the present results do not support the hypothesis that coexistence of Ce 3+ and Ce 4+ is important for high scintillation efficiency, at least in the present glass system. In turn, this work shows that the energy transfer process of the generated charged secondary particles to activators is important for attaining high scintillation efficiency. Therefore, tailoring the energy transfer process is expected to enable fabrication of high-performance scintillators.

Conclusion
We have examined PL and X-ray-induced scintillation properties of several Ce-doped lithium borosilicate glasses. It was confirmed that only Ce 3+ valence states exist in Ce:LBS40 glasses and that the Ce 3+ ratio decreases with increasing SiO 2 fraction in the glasses. The oxidation reaction in the glass melt in an inert atmosphere can be explained by the optical basicity of the glass, and the amount of Ce 4+ generated is the origin of the absorption tail in the visible region of the absorption spectra. Although the value of Q eff for the Ce:LBS10 glass is the smallest among all Q eff values for the present LBS glasses, the scintillation intensity of the Ce:LBS10 glass is the highest because it has the highest attenuation values. In terms of the emission mechanism of scintillators, the effective energy conversion after absorbing the ionizing radiation is prevented in the B 2 O 3 -rich glasses. Such energy transfer path will be important for further materials design of radiation detectors.

Methods
Preparation of Ce-doped lithium borosilicate glass. The xCe 3+ -40Li 2 O-xB 2 O 3 -(60-y)SiO 2 (xCe:LBSy) glasses were prepared according to a conventional melt-quenching method by employing a platinum crucible 24 . A  The glass melt was quenched on a stainless plate at 200°C and then annealed at a temperature T g , which was measured by differential thermal analysis (DTA) for 1 h. The bulk glasses were cut into several glass pieces (10 mm × 10 mm) using a cutting machine, and then, samples were mechanically polished (thickness ~ 1 mm) to obtain mirror surfaces. The temperature T g was determined by a DTA system operating at a heating rate of 10 °C/min using a TG8120 instrument (Rigaku, Japan). The density of the samples was measured using the Archimedes method with pure water as an immersion liquid.
Luminescence properties. The PL and PLE spectra were recorded at 1 nm intervals at RT using an F7000 fluorescence spectrophotometer (Hitachi High-Tech. Japan). Band pass filters of 2.5 nm for the PL measurement were used for both excitation and emission. The absorption spectra at RT were recorded at 1 nm intervals using a U3500 UV-vis-NIR spectrometer (Hitachi High-Tech. Japan). The absolute quantum efficiencies, also known as quantum yields (QYs), of the glasses were measured using an integrating sphere Quantaurus-QY (Hamamatsu  Photonics, Japan). The error bars were ±2. The emission decay at RT was measured using a Quantaurus-Tau system (Hamamatsu Photonics, Japan) with a 340 nm LED. The accumulated counts for evaluation were 50,000. Scintillation (radioluminescence) spectra were measured by using a CCD-based spectrometer (Andor DU920P CCD and SR163 monochromator) under X-ray exposure 23 . The supplied bias voltage and tube current were 40 kV and 0.52 ~ 5.2 mA, respectively. XANES measurement. The Ce L III -edge XANES spectra were measured at the BL01B1 and BL14B2 beamlines of SPring-8 (Hyogo, Japan). The storage ring energy was operated at 8 GeV with a typical current of 100 mA. The measurements were performed using a Si (111) double-crystal monochromator in the transmission mode (Quick Scan method), or in the fluorescence mode using 19-SSD detector at RT. The XANES spectra were recorded from 5.52 to 6.18 keV. Pellet samples for the measurements were prepared by mixing the granular sample with boron nitride. As references, XANES data for Ce(OCOCH 3 ) 3 ·2H 2 O and CeO 2 were collected using the same conditions. The corresponding analyses were performed by using Athena software 53 .