1. Introduction

Oxide glasses have several advantages, such as high transparency, good formability, and good chemical durability. The wide transparent window from the UV to the near-infrared (NIR) makes them essential for use in optical devices [1]. On the other hand, glass is a type of liquid whose structural ordering is much poorer than that of conventional stoichiometric crystals. This poor structural ordering without grain boundaries is the source of its good formability and also an important factor in its potential as a host material for various types of activators. Rare earth (RE)-doped glass is a typical example of an optical material containing activators [2–6]. Except for trivalent RE cations exhibiting the 4f–4f transition, the emission center of the glass is generally affected by the coordination field, which depends on the chemical composition of the glass [2]. Because the dopant cation is an ionic species, the covalency or ionicity of the host glass network is crucial to the dispersion of activators. However, glass-forming oxides generally possess a covalent bonding network, which is why the dispersibility of dopant cations in oxide glasses is inferior to that in halide glasses. However, P₂O₅ differs from other network-forming oxides such as SiO₂ or B₂O₃ because the P = O bond allows the inclusion of non-bridging [PO₄]³⁻ (ortho-) or [P₂O₇]²⁻ (pyro-) phosphate units, which have delocalized electrons [7]. Thus, P₂O₅ has been added to silica glass to improve the dispersibility of the dopant cation [8–10].

Among several emission centers, the ns²-type emission center is among the promising activators in practical devices, such as Sb³⁺-doped white fluorescent lamps [2]. This emission center is characterized by the s-p parity-allowed transition and the electrons in the outermost shell, which are the origin of the wide and tunable emission wavelength [2, 11–24]. Recently, we have focused on oxide glasses containing the Sn²⁺ center, which is an ns²-type emission center [17–24]. It is notable that Sn²⁺-doped phosphate glasses exhibit high photoluminescence (PL) efficiency comparable to that of conventional phosphors. In recent papers, we also reported that the emission properties of the Sn²⁺ center depend on the network-forming oxide [17–19] and are little influenced by the network modifier cations [20]. These findings suggest that the local coordination state of Sn²⁺ can be tailored by changing a glass's network-forming oxides. In other words, the emission of Sn²⁺-doped glasses consisting of two network-forming oxides is expected to vary depending on the network formation [21]. In terms of the homogeneity of the host glass matrix, it is worth examining the optical and emission properties of Sn²⁺ in oxide glasses, which have a tendency toward phase separation.

Here, we examined the relationship between the glass network structure of ZnO–P₂O₅–B₂O₃ glasses [25–27] and the emission properties of the Sn²⁺ center. ZnO–P₂O₅–B₂O₃ glass reportedly exhibits phase separation behavior at an intermediate composition of B₂O₃ and P₂O₅ [27]. On the other hand, it was reported that this glass is a host glass for activators [28-29]. However, these studies discussed either the structure [25–27]. or the emission properties [28-29], and there is no study on the correlation between the structure of these glasses and their emission properties as a luminescent material. In addition, no quantitative investigation of the network structure of these glasses based on NMR results has been reported. Considering these previous reports, we assume that the emission properties of the Sn²⁺ center should depend on the chemical composition of ZnO–P₂O₅–B₂O₃ glasses. We focus on whether the spatially neighboring units, i.e., the local coordination state, around the Sn²⁺ center depends on the P₂O₅–B₂O₃ ratio. To remove the interaction between Sn²⁺ centers, both the host glass network and the PL properties were examined using 0.1 mol% Sn-doped ZnO–P₂O₅–B₂O₃ glasses.

2. Experimental

Pap The 0.1SnO-60ZnO-(40-x)B₂O₃-xP₂O₅ (xSZBP) glasses were prepared according to a conventional melt-quenching method using a platinum crucible [24]. Batches consisting of SnO (99.5%), ZnO (99.99%), (NH₄)₂HPO₄ (99%), and B₂O₃ (99.9%) were mixed and melted at 1100°C for 30 min in air and Ar (99.999%) atmosphere. The glass melt was quenched on a stainless steel plate maintained at 200°C and then annealed at the glass transition temperature T_g, as measured by differential thermal analysis, for 1 h. After cutting (10 mm × 10 mm × 1 mm), the glass samples were polished with aqueous diamond slurries. Because the sample size for measurement was fixed, we can quantitatively compare the luminescence intensity among these samples.

The absorption spectra were measured using a U3500 UV–vis–NIR spectrophotometer (Hitachi, Japan). The PL and PL excitation (PLE) spectra were measured at room temperature using an F7000 fluorescence spectrophotometer (Hitachi, Japan). In the PL measurements, band pass filters were used for the excitation (2.5 nm) and emission (2.5 nm). The emission decay at room temperature was measured using a Quantaurus-Tau instrument (Hamamatsu Photonics, Japan). The excitation light source for emission decay measurement was a light-emitting diode operated at a photon energy of 4.43 eV and a frequency of 10 kHz. The internal quantum efficiency was evaluated using a Quantaurus-QY instrument (Hamamatsu Photonics, Japan). X-ray diffraction (XRD; RINT-2100, Rigaku) measurements were performed to assign the precipitated crystal phase. ¹¹B magic-angle spinning (MAS) NMR spectra were measured (Delta, JEOL) using a magnetic field strength of 600 MHz (14.1 T), a frequency of 192.56 MHz, a spin rate of 16 kHz, a relaxation delay of 3 s, and a pulse width of 0.1 μs. The reference was a 1 M H₃BO₃ aqueous solution (19.6 ppm). Further, ³¹P NMR measurement was performed (CMX-400, JEOL) using a spin rate of 14 kHz and a pulse delay of 15 s. The frequency used in this study was 161.80 MHz, and the accumulation number for each measurement was 3600–4500. Each chemical shift is estimated in δ (ppm) with respect to the reference H₃PO₄ (in D₂O) solution (0 ppm).

¹¹⁹Sn Mössbauer spectra, i.e., the spectra of γ-ray absorption by the ¹¹⁹Sn nuclei in the samples, were measured in a conventional transmission geometry using a Ca^119mSnO₃ source at room temperature. The energy of the γ-rays from the source was modulated by the Doppler effect using a velocity transducer with a constant acceleration mode, and the abscissa of the spectra represents values in units of the Doppler velocity, as in the literature. The valence states of the Sn atoms, which are sensitively reflected in the peak positions in the ¹¹⁹Sn Mössbauer spectra [30]. were deduced by fitting the measured spectra using the standard software Normos (developed by R. A. Brand, commercially available from WissEl GmbH).

3. Results and discussion

3.1 Host glass network

The glass-forming region of the xSZBP glasses is shown in the photographs of several typical samples in Fig. 1(a). Although transparent glasses were obtained in the B₂O₃-rich and P₂O₅-rich regions, opaque or crystallized samples were obtained at intermediate compositions (x = 10–25). Figure 1(b) shows the powder XRD pattern of the 20SZBP glass, which appeared opaque. Among the prepared samples, the 20SZBP glass showed clear precipitation of the Zn₃(PO₄)₂ phase [Joint Committee on Powder Diffraction Standards (JCPDS) card #29-1390] [31]. On the other hand, because no diffraction peak was observed for the 10SZBP and 25SZBP glasses, the translucency is attributed to phase separation. The present tendency is similar to that of previously reported Sn-free glasses [27].

 

Fig. 1 (a) xSZBP glass system and photographs of several glasses. (b) XRD patterns of 0.1SnO-doped 60ZnO-20B₂O₃-20P₂O₅ glass and Zn₃(PO₄)₂ (JCPDS # 29-1390).

Download Full Size | PDF

Figure 2(a) shows ³¹P MAS NMR spectra of the xSZBP glasses. These spectra are normalized using each P₂O₅ fraction. Several peaks are observed and assigned to Q⁰ (~2 ppm), Q¹ (~−10 ppm), and Q² (~−30 ppm), respectively [32–38]. The superscript n in Qⁿ indicates the number of bridging oxygens of the phosphorus cation. Phosphorus cations exist as Q⁰ or Q¹ units in the B₂O₃-rich composition, and Q² units are generated in the P₂O₅-rich composition. The chemical shift of the Q⁰ peak is almost constant at P₂O₅ contents below 10 mol%, suggesting that the neighboring cation of the Q⁰ phosphate unit is boron. In contrast, the chemical shift of the Q¹ and Q² peaks change at higher magnetic fields with increasing P₂O₅ fraction. The chemical shift indicates that the neighboring cations of the Q¹ and Q² units change from boron to phosphorus. Figure 2(b) shows the peak area ratio of the Qⁿ units after peak deconvolution using a Gaussian function as a function of the amount of substituted P₂O₅. The Q⁰ ratio decreases linearly, whereas the Q² ratio increases with increasing P₂O₅ content. The decrease in Q⁰ means that instead of a borate glass network forming, the generated phosphate chains form the glass network. Figure 2(c) shows the number of each type of Qⁿ unit, which is obtained by multiplying the ratio of borate units by the B₂O₃ fraction, (40 − x) mol%, as a function of the amount of substituted P₂O₅. It is notable that the Q⁰ and Q¹ units are isolated phosphate units possessing delocalized electrons and are inherently unable to form the main network. The number of Q⁰ unit is maximum in the phase-separated composition (x = 10), suggesting that Q⁰ units are localized in the zinc borate network. The number of Q¹ units is maximum around 33 mol%, which corresponds to the chemical composition Zn₂P₂O₇. It is notable that the substitutional fraction of B₂O₃ in the P₂O₅-rich (P₂O₅ = 30–40 mol%) region corresponds to the increase in Q² units in the same compositional region. BO_3/2 and BO_4/2, which are discussed later, are three-dimensional (3D) structural units, whereas the Q² (chain) unit is two-dimensional. The increase in Q² units instead of borate units follows a decrease in the network dimension of the SZBP glasses if the structural ZnO plays the same role in both B₂O₃- and P₂O₅-rich glasses. Because ZnO can act as an intermediate [39], i.e., it can serve as either a network former or a network modifier, it is expected that the role of structural ZnO in B₂O₃-rich glass is not equal to that in P₂O₅-rich glass [40].

 

Fig. 2 (a) ³¹P MAS NMR spectra of the xSZBP glasses. (b) Qⁿ unit ratios of xSZBP glasses as a function of P₂O₅ content. (c) Number of Qⁿ unit ratios of xSZBP glasses as a function of P₂O₅ content.

Download Full Size | PDF

The ¹¹B MAS NMR spectra of the xSZBP glasses are shown in Fig. 3(a). The spectra are also normalized using each B₂O₃ fraction. The broad peak at around 15 ppm corresponds to three-coordinated boron (BO_3/2), whereas the sharp peak at around 0–5 ppm is attributed to four-coordinated boron (BO_4/2) [38-39]. The three-coordinated boron is further deconvoluted into BO_3/2 ring and BO_3/2 non-ring structures. Although both BO_3/2 and BO_4/2 exist in the B₂O₃-rich glasses, only BO_4/2 is observed in the P₂O₅-rich glass. The BO_4/2 peak shift moves towards higher magnetic field when P₂O₅ is substituted for B₂O₃ owing to a change in the average coordination state due to formation of B–O–P bonds. It is notable that the peaks around 0 ppm attributable to BO_4/2 units are split. Because the peak at a higher magnetic field shifts with increasing P₂O₅ content, the observed change suggests the formation of BO_4/2 units with different covalent bonds, i.e., B–O–B–O–B, B–O–B–O–P, and P–O–B–O–P. Figure 3(b) shows the unit ratios of BO_4/2 and BO_3/2 as a function of the amount of substituted P₂O₅. In the B₂O₃-rich region, the BO_3/2 ratio is almost constant. BO_4/2 increases continuously, and the ratio of BO_4/2 exceeds that of BO_3/2 around the phase-separated region (around x = 30). Figure 3(c) shows the number of BO_3/2 and BO_4/2 units as a function of the amount of substituted P₂O₅. The numbers of BO_3/2 and BO_4/2 units decrease linearly with increasing P₂O₅ fraction, and the slope of the number of BO_4/2 units changes gradually. When BO_4/2 units are formed, a counter cation is needed for charge compensation. Because Zn²⁺ cations act as a counter cation to the BO_4/2 units, the number of residual ZnO units that can form a covalent bonding network is increased. We think that such ZnO can form a subnetwork by interaction with phosphate chains in P₂O₅-rich glass, as suggested above.

 

Fig. 3 (a) ¹¹B MAS NMR spectra of the xSZBP glasses (x = 0, 2, 4, 6, 8, 10, 30, and 35). (b) Ratios of borate units as a function of P₂O₅ content. (c) Number of BO_3/2 and BO_4/2 units of xSZBP glasses as a function of P₂O₅ content.

Download Full Size | PDF

From the ¹¹B and ³¹P MAS NMR data, it is expected that PO₄ units form phosphate chains in P₂O₅-rich glass, in which BO_4/2 units are chemically bonded with the phosphate units. In contrast, PO₄ units are isolated without a long P–O–P chain structure in B₂O₃-rich glasses, in which the network consists of both BO_3/2 and BO_4/2 units.

3.2 Local coordination of Sn²⁺

In this section, we determine whether local coordination in the Sn²⁺-P₂O₅ region is stable in ZnO–B₂O₃ glass so that Sn²⁺ exhibits the same PL properties as it does in ZnO–P₂O₅ glass, considering the result that adding P₂O₅ to silica glass improves the dispersibility of the dopant cation [8–10].

First, we examined the valence state of the Sn cation using ¹¹⁹Sn Mössbauer spectroscopy. Figure 4(a) shows ¹¹⁹Sn Mössbauer spectra of the xSZBP glasses (x = 0, 4, 8, 10, and 40). The peaks around 0 mm s⁻¹ correspond to the Sn⁴⁺ species, whereas those around 2 and 4 mm s⁻¹ correspond to the Sn²⁺ species [30]. These spectra show that the Sn⁴⁺ species exists in the 0SZBP glass; that is, Sn²⁺ is oxidized to Sn⁴⁺ even though the samples were melted in an inert atmosphere (99.999% Ar) [41]. However, with a small increase in the P₂O₅ fraction, the oxidation reaction was effectively prevented as the 4SZBP in Fig. 4(a). It is known that the Sn²⁺/Sn⁴⁺ ratio is also dependent on the redox-behavior of the starting material. Since the starting material of P₂O₅ was (NH₄)₂HPO₄, it is expected that (NH₄)₂HPO₄ worked as a reducing agent. Although calcination before melting was performed in order to prevent damage to Pt crucible during the melting [24], we assume that the residual species act to increase the Sn²⁺ ratio. If the difference in recoilless fraction between the Sn⁴⁺ and Sn²⁺ sites is assumed to be the same at room temperature, the Sn²⁺/Sn⁴⁺ ratio can be estimated. Figure 4(b) shows the Sn²⁺ ratio, isomer shift (IS), and quadrupole splitting of the Sn²⁺ species in the xSZBP glasses as a function of P₂O₅ content. To discuss the IS, the position of the peak splitting is used as a standard by averaging the two peaks. Although the IS depends on the P₂O₅ fraction, the slope of the increase in Sn²⁺ in the B₂O₃-rich region differs from that in the P₂O₅-rich region. Because the IS is affected by the surrounding Coulombic interaction, this result clearly suggests that the local coordination state depends on the chemical composition of the glass [42–46].

 

Fig. 4 (a) ¹¹⁹Sn Mössbauer spectra of the xSZBP glasses (x = 0, 4, 8, 10, and 40). (b) Sn²⁺ ratio in xSZBP glasses as a function of P₂O₅ content. The isomer shift (IS) and quadrupole splitting (QS) of Sn²⁺ are also shown.

Download Full Size | PDF

Figure 5(a) shows optical absorption spectra of the xSZBP glasses. Since the observed absorption edge is due to the s-p transition of Sn²⁺, the values of (hνα)² is plotted as the left axis. The optical absorption edge of Sn²⁺ is red-shifted with increasing P₂O₅ content. In zinc phosphate or zinc borate glasses, a red-shift in the absorption edge of Sn²⁺ is observed when the amount of added Sn²⁺ increases [19]. On the other hand, optical absorption due to Sn⁴⁺ exists in the UV region [47]. If the shift in the optical absorption edges is correlated only with the valence state of tin, the change in the spectra should be nonlinear, similar to that in Fig. 4(b). Considering that the IS of Sn²⁺ depends on the chemical composition, it is natural that the optical absorption is also affected by the chemical composition. To discuss the optical properties, the optical absorption edge of tin is determined by extrapolation of these absorption curves to the 3 × 10⁴ to 6 × 10⁴ eV² cm⁻² region. Figure 5(b) shows the optical band gap E_gopt of the xSZBP glasses as a function of P₂O₅ content. Like the IS, the slope of E_gopt in the B₂O₃-rich glasses differs from that in the P₂O₅-rich glasses. The absorption shift, therefore, is due mainly to a change in the coordination state of the Sn²⁺ centers, which was observed in previous studies [19].

 

Fig. 5 (a) Optical absorption spectra of the xSZBP glasses. Dashed line is an extrapolation line for determination of the optical band gap E_gopt. (b) E_gopt of the xSZBP glasses as a function of P₂O₅ content.

Download Full Size | PDF

PL-PLE 3D contour plots of the xSZBP glasses (x = 0, 8, 30, and 40) are shown in Fig. 6(a). In terms of the excitation energy, the PLE peak energy of SZP glass is higher than that of SZB glass. With increasing P₂O₅ content, a higher excitation band emerges, and the emission intensity increases. The PLE spectra consisting of at least two components change in a complex manner. Figure 6(b) shows PL and PLE spectra of the 0SZBP and 40SZBP glasses. When the P₂O₅ fraction increases, the excitation band becomes higher with increasing emission intensity, i.e. the 40SZBP glass exhibited the highest emission intensity. Figure 6(c) shows the peak energies of the PL and PLE after peak deconvolution. Both peak energies are blue-shifted with increasing P₂O₅ fraction. This behavior may be correlated with the basicity of the glass [48], which decreases with increasing P₂O₅ fraction. Thus, the PL energies are affected by the chemical composition of the glass with different B₂O₃/P₂O₅ ratios.

 

Fig. 6 (a) PL-PLE contour plots of the xSZBP glasses (x = 0, 8, 30, and 40). (b) PL and PLE spectra of the 0SZBP and 40SZBP glasses. (c) Peak energies of the PL and PLE of the xSZBP glasses as a function of P₂O₅ fraction.

Download Full Size | PDF

Figure 7 shows the PL decay curves of the xSZBP glasses (x = 0, 2, 4, 6, 8, 10, 30, 35, and 40) at an excitation energy of 4.43 eV (280 nm). The decay curves are monitored at the photon energy of each PL spectrum. In the B₂O₃-containing glasses, the decay curves consist of two components: one is faster (the lifetime, τ′_1/e, is of nanosecond order), and the other is slower (τ′′_1/e is of microsecond order). As P₂O₅ is substituted for B₂O₃, the faster component, whose decay constant is several tens of nanoseconds, disappears. The observed decay in the B₂O₃-containing glasses, which was also observed in another borate glass system [18], is thought to be associated with the borate glass network. The PL decay constant is on the order of microseconds owing to relaxation from the triplet state and singlet state. It may seem that the fast decay component of nanosecond order is associated with Sn⁴⁺, because only Sn²⁺ centers exist in ZnO–P₂O₅ glass exhibiting no fast decay component. However, there is no relationship between the Sn²⁺ content (see Fig. 4(b)) and the fast (ns)/slow (μs) component ratio. Therefore, it is natural that the Sn²⁺ species in the B₂O₃ region are the main influence on the decay profile in Fig. 7.

 

Fig. 7 PL decay curves of the xSZBP glasses (x = 0, 2, 4, 6, 8, 10, 30, 35, and 40). Excitation energy was 4.43 eV (280 nm).

Download Full Size | PDF

Here, we consider the relationship between the emission properties and the glass network structure. With increasing P₂O₅ content, the main glass network changed from three- and four-coordinated borates to a chain phosphate network. This structural change causes changes in the IS, E_gopt, the PL peak, and the PL decay constant of Sn²⁺. However, the degree of change in the P₂O₅-rich glasses differs from that in the B₂O₃-rich glasses, although tin takes the divalent state in most of the glasses. Because the local coordination state of tin changes continuously depending on the chemical composition, we conclude that no localized tin exists around the phosphorus units in the glass; i.e., tin species are homogenously dispersed in the glass network. It is notable that the present glasses were prepared by melt-quenching method, although quenched glass samples were annealed at the T_g. Spatial distribution of Sn²⁺ centers might be changed depending on the cooling condition from the melt [21]. Since homogeneity of activators governs the performance of the phosphor, such transient state of glass material will be important for future functionalization.

4. Conclusion

We examined the Sn²⁺ emission center in zinc borophosphate glasses possessing different random networks. The glass network changed continuously depending on the chemical composition. Although the Sn²⁺ center is an ionic species, a localized distribution of Sn²⁺ around the phosphate region was not found. With increasing P₂O₅ content, the glass structure consisting of BO_3/2 and BO_4/2 changed into a phosphate chain structure via a phase separation region. The optical properties, such as the E_gopt value of Sn²⁺, PL-PLE, and PL decay constant, varied continuously depending on the chemical composition. The emergence of a faster decay component of Sn²⁺ observable in the B₂O₃-rich glass is associated with this structural change. The homogenous dispersion of the Sn²⁺ center in the zinc borophosphate glass is also confirmed by ¹¹⁹Sn Mössbauer spectra, in which the IS of Sn²⁺ is affected by the chemical composition.