Oxidation suppression of Cu in alkaline aluminophosphate glass and the effects for radiation-induced luminescence characteristics

A glass phosphor is an attractive material for applications in radiation detections because of its high workability and availability with a wide range of chemical compositions. Recently, X-ray-induced luminescence of glasses containing various luminescent activators are actively investigated worldwide. In applications as phosphor, tailoring valence state of activators, which can take multiple valence states in glass, is very important. In this research, we studied effects of glass melting atmosphere on the valence state of copper-activator ion in alkaline aluminophosphate glasses and the radiation-induced luminescence characteristics. Optical absorption and X-ray absorption near edge structure spectra of Cu-doped glasses showed that the glass fused in Ar atmosphere contains higher concentration of Cu+ than those prepared in air. In addition, the presence of Cu+ enhances the photoluminescence (PL) quantum yield and PL kinetic constant. Furthermore, the increase of Cu+ concentration resulted an improvement of the X-ray-induced scintillation and thermally-stimulated luminescence intensity.


Results
The density and thickness of all the specimens are shown in Figure S1. From the XRD measurement results, only a halo peak which is a characteristic of glass was observed, and no crystallization was confirmed ( Figure S1). Thermal properties of all the specimens examined by differential thermal analysis (DTA) are shown in Table 1 and Figure S2. Each specimen exhibited an endotherm corresponding to T g at around 450 °C, followed by another endotherm corresponding to the re-melting of glass matrix. Thermal annealing was performed for all the glass specimens at the corresponding T g temperature to remove the internal stress. The T g of the Cu-Ar specimens were higher than those of the Cu-air, suggesting that their thermal properties were different by the melting atmosphere. Figure 1 shows normalized Cu K-edge XANES spectra of the Cu-doped specimens. CuO and Cu 2 O are reference raw compounds of divalent and monovalent Cu, respectively, and we estimated the valence state of Cu in the specimens compared with these references. The Cu-Ar specimen and Cu 2 O had a distinct pre-edge around 8981 eV which was assigned to the 1s-4p electronic transition of Cu 21,22 . In contrast, we observed only a shoulder Table 1. Glass transition temperature T g of all the specimens from DTA measurement. www.nature.com/scientificreports/ structure in the spectra of the Cu-air specimen and CuO. It has been recognized that the pre-edge feature corresponds to the absorption edge of Cu + , but Cu 2+23- 25 . These results indicate that the Cu-Ar specimen includes most of the Cu in the monovalent state while most Cu ions in the Cu-air specimen was present in the divalent state. We also estimated the valence state of each Cu-doped specimen quantitatively by the linear combination fitting method using Athena program. In the linear combination fitting method, the X-ray absorption spectrum is approximated by a least-square fitting with a linear combination of spectra of known species. The linear combination fitting has been carried out for the normalized spectrum in the range from 8970 to 9020 eV in this analysis. The fitted spectra and calculated results are shown in the Figure S3 and Table 2. Here, we assumed that all the Cu ions take a state of either CuO (Cu 2+ ) or Cu 2 O (Cu + ), and Cu total is defined as a total number of Cu ions (i.e. sum the numbers of Cu 2+ and Cu + ions). Comparing between the two different specimens prepared in Ar and air (both are doped with 0.01% Cu), the former includes approximately 90% of Cu ion in the monovalent state while only 20% was found in the latter specimen. Figure 2 shows the optical absorption spectra in the UV and near infrared (NIR) regions for the specimens prepared under (a,b) Ar and (c,d) air. In the undoped specimens, an absorption band feature was observed in the UV region at around 220 nm (Fig. 2a,c). The absorption is originated from unavoidable trace Fe impurities included in the raw starting materials of the glasses 26,27 . For the Cu-doped specimens, an absorption band appeared near 250 nm. The latter absorption band is assigned to 3d 10 -3d 9 4s 1 transition of Cu +28 . The absorption bands shift to longer wavelength with increasing the Cu concentration. Moreover, the absorption band of the Cu-air specimen appeared on the longer wavelength side than that of the Cu-Ar specimen. This behavior is not attributed to the presence of the 3d 10 -3d 9 4s 1 transition of Cu + but a charge transfer of O-Cu-O complex 29,30 , which implies that the generation of O-Cu-O is suppressed when the glass is molten in Ar atmosphere. Figure 2b,d are the absorption spectra in the NIR range. The Cu-air specimens showed more remarkable absorption band near 800 nm than those of the Cu-Ar specimens, and any absorption band was not confirmed in the undoped specimens. This absorption band is considered to be due to 2 E g -2 T g transition of Cu 2+ placed in an distorted octahedral field [30][31][32] . This result suggests that Cu 2+ can be present in a large proportion when the glass was molten in air atmosphere compared with those molten in Ar atmosphere. In other words, the melting under the Ar atmosphere can increase the concentration of Cu + . These observations in the absorption spectra as a function of Cu concentration are consistent with those by the linear combination fittings of the XANES spectra. On the other hand, these results also suggest that oxidation does somewhat occurred even when the material was molten in Ar atmosphere. We assume that the oxidation reaction occurred during the quenching process since it was performed in air, although we ensured the glass melt to be exposed to air as short time as possible during the quenching process. An increase in absorbance observed in the region longer than 2500 nm is due to absorption by OH stretching 17 . The absorption attributed to OH stretching was more remarkable in those molten in air than Ar. Figure 3a shows photoluminescence (PL) excitation/emission spectra of the 0.10% Cu-doped specimens at room temperature (R.T.), and PL excitation/emission contour maps and spectra of all the Cu-doped specimens are also shown in Figure S4 and S5. Each specimen had a broad excitation band in the UV region and showed a broad emission band peaking at around 440 nm. This emission is due to the 3d 9 4s 1 -3d 10 transition of Cu +31 . Here, we observed a spectral peak shift in the excitation spectra of the Cu-air specimens more effectively than those of the Cu-Ar specimens. The shift in excitation spectra is because the Cu-air specimens have an absorption band due to the O-Cu-O complexes in addition to the 3d 10 -3d 9 4s 1 band, which overlaps with a part of the excitation band. In contrast, we did not observe a spectral peak shift of an emission band in all the Cu-doped specimens (see Figure S5). In the emission spectra, all the specimens indicated a weak shoulder feature on the longer wavelength side (around 450-500 nm), which suggests that the emission has two different origins. Figure 3b,c show peak deconvolution results by gaussian fitting of 0.10% Cu-Ar and Cu-air. Each spectrum was well-reproduced by a convolution of two gaussians peaking at 424 and 467 nm. Figure 3d is a ratio of the integrated gaussians (424/467 nm) as a function of Cu concentration. The PL band peaking at 424 nm was dominant than the other for all the specimens. In addition, an anticorrelation of the emission ratio was observed between the specimens molten under Ar and air atmospheres. This fact suggests that glass melting atmosphere controls the coordination environment around Cu + . The emission origins of these bands will be discussed in detail below together with the results of PL decay constants.
The maximum PL quantum yields (QY) for emission by Cu + are summarized in Table 2. The QY of each specimen showed a maximum value when excited at 250 nm. The QY was significantly higher in the Cu-Ar specimens than the Cu-air specimens. These results fully reflected the effect of melting in Ar atmosphere. Furthermore, the QY of Cu-Ar tended to increase with increasing Cu concentration while it tended to decrease for Cu-air Table 2. PL QYs, valence ratio, and relative Cu + concentration in the all Cu-doped specimens.

Glass specimens
QYs (%) Valence ratio (Cu + /Cu total ) (%) Cu + concentration (mol%)  Figure 4 is a correlation between QY, valence ratio (Cu + /Cu total ), and Cu + concentration as a function of doping concertation of Cu. The valence ratio calculated from the linear combination fitting analysis of XANES spectra tends to decrease with increasing the Cu concentration (except for Cu(0.10%)-Ar), while the actual Cu + concentration increases with Cu concentration. Focusing on the relationship between Cu + and QY, we can see that there is a positive correlation in the Cu-Ar specimens and a negative correlation in Cu-air.  Table 3. The excitation and monitored wavelengths were 260 and 440 nm, respectively. PL lifetimes were deduced by approximating the experimental data with a sum of two exponential decay functions. The calculated decay constants were on an order of several tens of microseconds, and the experimental results agreed well with the previously reported values of luminescence by Cu + activator 33 . According to a previous study, in case of amorphous materials such as glass, the triplet state ( 3 E g ) of the 3d 9 4s 1 configuration, which is an excited state of Cu + , is split into two levels (T 2g and T 1g ) because the Cu coordinates at the distorted tetrahedral sites 33 . The first and second components of the PL lifetime are assigned to the emission transitions from the triplet states T 2g and T 1g to the 1 A 1g ground state, respectively. The two PL peaks at 424 and 467 nm correspond to the emissions by transitions from T 2g and T 1g , respectively. Figure 5c,d are PL decay constants, and kinetic constants as a function of the Cu concentration of Cu-Ar and Cu-air specimens. Here, the kinetic constant is a probability of radiative transition per unit time and is denoted as k r in this paper. Furthermore, if the probability of the non-radiative transition is defined as k nr , the decay constant can be expressed as τ = 1/(k r + k nr ). www.nature.com/scientificreports/ In addition, a PL QY is expressed using these kinetic constants as QY = k r /(k r + k nr ). From these relationships, the kinetic constant can be calculated by the following equation.
There was a positive correlation between Cu concentration and PL decay constants or kinetic constants of each Cu-Ar specimen. In contrast, we confirmed a negative correlation in Cu-air. This result suggests that the Cu + concentration increases with increasing Cu concentration for Cu-Ar whereas it decreases with increasing Cu concentration in Cu-air. Figure 6 shows X-ray-induced scintillation spectra of (a) Cu-Ar and (b) Cu-air specimens, and (c) a relationship between PL QY, the Cu + concentration, and the scintillation peak area as a function of effective atomic number (Z eff : explained in the discussion section) and Cu concentration. In each undoped specimen, we observed a weak broad emission from 300 to 500 nm with a peak around 320 nm. The latter broad emission is attributed to intrinsic luminescence center called L-center in glass [34][35][36] . In the case of our specimens, this L-center is interpreted as P-O-K linkages 34,37 . In contrast, Cu-doped glass specimens melted both in Ar and air had different emission peak features from the undoped ones. Following the spectral shapes of the PL from Cu-doped specimen, we  www.nature.com/scientificreports/ could attribute the origin of scintillation emission to the 3d 9 4s 1 -3d 10 transition of Cu + . Focusing on the peak position, we can see that these scintillation peak positions are longer than those of PL. Furthermore, the peak wavelength of the Cu-air is longer than that of the Cu-Ar specimens (see Figure S6). Based on a previous report, the reason is probably because an emission peak due to Cu + -Cu + dimer appeared in scintillation 38 . The trend of increasing or decreasing scintillation emission area is positively correlated with QY among the Cu-Ar specimens. In contrast, it was negatively correlated with QY among the Cu-air specimens. Thermally stimulated luminescence (TSL) glow curve represents luminescence intensity as a function of temperature while heating at a constant rate. Prior to the measurement, the specimens were irradiated with X-rays of 10 Gy. The obtained glow curves are illustrated in Fig. 7a-c. We could observe two glow peaks from the undoped specimens. Following the results from scintillation, the luminescence center in TSL of the undoped specimens seems to be the L-center. On the other hand, the Cu-doped specimens showed only a single glow peak, and its intensity significantly increased compared with the undoped specimens. Both Cu-Ar and Cu-air types had a glow peak nearby 110 °C, and no significant differences in the glow curve shape was observed regardless of the difference of the melting atmosphere. It is assumed that the luminescence center in TSL of the Cu-doped specimens is the 3d 9 4s 1 -3d 10 transition of Cu + . Figure 7d shows the PL QY and normalized TSL intensity as a function of Cu concentration. In the Cu-Ar specimens, the fluctuation trend of the normalized TSL intensity coincides with QY. In the Cu-air specimens, both QY and normalized TSL intensity showed a decreasing trend from 0.01 to 0.05% while, from 0.05 to 0.10%, QY decreased and normalized TSL intensity increased.

Discussion
First, we discuss the physical properties of the glass specimens. In the TG-DTA measurements, T g of Cu-doped specimens exhibit different tendency depending on the preparation atmosphere. Although T g of Cu-doped glass decreases with increasing Cu concentration, the specimens prepared in air atmosphere possess steeper change compared with those prepared in Ar atmosphere. Considering the present results, we assume that Cu 2+ species are more effective than Cu + to decrease the T g . We could not confirm metallic Cu related spectrum from the optical absorption and Cu K-edge XANES spectra. Therefore, our glass specimens have not included metallic Cu.
It should be noted that the results of linear combination fitting of the XANES spectra are for reference only because the host material of the fitted spectra is different from that of the reference samples. As illustrated in Fig. 5c,d, PL QYs, decay constants and kinetic constants are correlated with Cu concentration. Therefore, actually, the concentration of Cu + in Cu-air is expected to decrease even more strongly with Cu concentration, and the Cu + concentration is expected to show the same decreasing trend as in QY. Based on the above observations, we will discuss the results of X-ray-induced luminescence properties.
When X or γ rays pass through a material with atomic number Z, the incident radiation decreases exponentially with the penetration depth whose degree is well-defined with a linear attenuation coefficient. The linear attenuation coefficient is proportional to the fourth power of Z, since the photoelectric effect is dominant in the case of X-rays with energy of several tens of keV, which is also the case of this research. If the substance is a compound, it is expressed as a sum of the linear attenuation coefficient weighted in weight fractions of each constituent element. In other words, the fourth power of the compound atomic number Z eff corresponds to the total sum of products of the weight fraction of each element and the fourth power of respective constituent elements, we can calculate Z eff by taking these biquadratic roots. In the present glass composition, an increase in the  www.nature.com/scientificreports/ Cu concentration causes an increase of Z eff , which in turn increases the effective absorption of X-rays. Therefore, as in Fig. 6c, the scintillation intensity of Cu-air showed an upward trend despite the decrease in the QY. The TSL intensity of the Cu(0.10%)-Ar specimen was greater than that of the Cu(0.10%)-air specimen. Here, the luminescence efficiency η i in TSL is defined by the following equation 39 : Here n eh , η tr , η TSL , hν , and η esc are defined as the number of e-h pairs per unit mass and per unit absorbed dose, the fraction that is captured in traps which can be thermally-stimulated, TSL efficiency, the average energy of the emitted TSL photons, and the fraction of the produced photons that will escape without being absorbed, respectively. In addition, η TSL is also defined as a product of a release probability of the charges from the traps: p, a transport efficiency of released charges to a luminescent center: S, and a recombination efficiency on the luminescent center (i.e. PL QY). Now, since eacspecimen has the same base material, n eh , p, S, and hν , are consistent whereas η i depends only on η tr , QY, and η esc . Therefore, the TSL intensity of the Cu(0.10%)-Ar specimen with high QY and low self-absorption by Cu 2+ was higher than that of the Cu(0.10%)-air specimens. Figure 7d shows that the TSL intensity increased with increasing Cu concentration in the Cu-Ar specimen while the TLS (2) η i = n eh η tr η TSL hνη esc . www.nature.com/scientificreports/ intensity decreased from 0.01% to 0.05% in the Cu-air specimen and again increased from 0.05% to 0.10%. Considering the formula for η i , it is reasonable to think that the increase of TSL intensity depends on η tr . because the integrated TSL intensity is not being correlated with QY. In the Cu-air specimens, the Cu + concentration did not increase with increasing Cu concentration of the specimens, however, the concentration of Cu 2+ increased. This suggests that Cu 2+ is likely to function as an electron trap center. As a result, the TSL intensity is predominantly contributed by the QY at and below 0.05% while the number of electron trapping centers (i.e. Cu 2+ ) dominantly contributes at and above 0.05% of Cu concentration.

Conclusion
The absorption spectra of the glass molten under the air atmosphere showed a pronounced optical absorption attributed to the 2 E g -2 T 2g transition of Cu 2+ whereas we could scarcely observe the absorption band in the glass fabricated by melting in Ar atmosphere. Furthermore, as a result of analyzing XANES at the copper K-edge, we found that most of the copper in the glass have a low valence state (Cu + ) when they are molten in the Ar atmosphere. Fabrication of a glass under a reducing atmosphere improved the PL QY of Cu + luminescence. Along with an increase of luminescence efficiency, TSL intensity of the Cu-doped specimen improved. These results showed that the control of the melting atmosphere is effective for improving radiation-induced luminescence characteristics due to luminescent center ion such as copper which can assume several valence states.   (2N) was added to the host material to include 0, 0.01, 0.05, and 0.10 mol.% of Cu ion while the host material is dealt as 100%. In this paper, 'Cu concentration' indicates the nominal value. These powders were weighed in a total amount of 10.0 g and mixed homogeneously using a mortar. The mixed powder was put into a Pt crucible and then melted at 1100 °C for 20 min in Ar or air atmosphere. The glass preparation under an Ar gas atmosphere was carried out by the following procedure. The mixed powder in a Pt crucible was loaded into an electric tube furnace in which a SiO 2 tammann tube is equipped. The tammann tube was vacuumed to less than − 0.1 MPa, and then a high purity Ar gas (5N) was purged. The above vacuum and purge processes were repeated three times to ensure all the air was removed. During the glass melting process, a flow of Ar gas was controlled at a rate of 0.5 L/min. After the melting for 20 min, the electric furnace was opened in air, and the Pt crucible was quickly taken out from the furnace using a tongue to quench the glass melt on a stainless-steel plate preheated to 200 °C. After cooling to R.T., the glasses were annealed at the glass transition temperature T g for 1 h in air. Following the annealing procedure, the glass specimens were cut into the dimension of 9.0 × 9.0 × ~ 1.4 mm 3 and then the surfaces were mechanically polished. In this paper, we denote that the obtained glasses molten in Ar are Cu-Ar specimens while those molten in air are Cu-air specimens. The T g of the glass specimen was determined using a TG-DTA system (Rigaku) operating at a heating rate of 10 °C/min. The densities of the glass samples were determined by the Archimedes method using analytical balances (GR-120, A&D Company, Limited). XRD patterns were measured by MiniFlex 600 (Rigaku).
Cu K-edge XANES measurement. In order to evaluate valence states of Cu in the glass specimens prepared under Ar and air atmosphere, we have measured XANES at the BL01B1 beamline of the synchrotron radiation facility, SPring-8. The Cu K-edge XANES measurements were carried out using a Si (111) double-crystal monochromator in the fluorescence mode using 19-SSD at R.T. Reference samples of CuO and Cu 2 O, which were prepared by mixing the granular sample with boron nitride, were also measured in the transmittance mode.
Optical/luminescence properties. Transmittance spectra of the glass specimens were measured using a spectrophotometer (V670, JASCO) across a spectral range from 190 to 2700 nm with 1 nm intervals. The PL excitation/emission spectra were measured by a spectrofluorometer (FP-8600, JASCO) which equipped a Xelamp. The range of monitored wavelength was from 300 to 600 nm while the excitation wavelength was 260 nm. Excitation spectra was measured from 200 to 350 nm when the monitored wavelength was 440 nm. PL QYs were measured using Quantaurus-QY (C11347, Hamamatsu Photonics) at R.T. QY includes the measurement errors in ± 2%. The monitored excitation/emission wavelength ranges were 250-350 and 350-750 nm with 10 nm intervals, respectively. To determine an origin of luminescence, PL lifetime measurements were performed using Quantaurus-Tau (C11367, Hamamatsu Photonics).
TSL measurement. TSL glow curves were measured using TL-2000 (Nanogray Inc.), and we set up temperature range and heating rate to 50-400 °C and 1 °C/s, respectively 41 .
Received: 26 October 2020; Accepted: 23 November 2020 Scientific Reports | (2020) 10:21403 | https://doi.org/10.1038/s41598-020-78510-z www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.