Luminescence behaviour of Eu3+ in hot-compressed silicate glasses

Abstract This paper aims to explore density-driven effects on the luminescence of Eu3+ doped silicate glasses. For this, mildly densified samples were fabricated by quasi-isostatic hot compression at up to 2 GPa from melt-quenched precursor materials. As a result of compression, both density and glass transition temperature of the blank (Eu free) and doped compositions increase. Raman spectroscopy indicates a slight increase in the intensity of vibrational modes assigned to small silicate rings and Si-O-Si bridges. Photoluminescence experiments reveal the creation of new paths of de-excitation, reducing the relative intensity of the transitions in the excitation spectra and the lifetime of the 5D0 → 7F2 emission line. Meanwhile, the luminescence intensity remains unchanged due to enhanced oscillator strength and refractive index at uniform electron-phonon coupling strength. Luminescence spectra also show a slight expansion of some of the energy levels of Eu3+, together with an increase of the bandwidth of the 5D0 → 7F2 emission line due to the crystal-field influence and growing electron population of the lower Stark sub-levels. Finally, increasing symmetry of Eu3+ sites was detected with increasing degree of compaction, resulting in a reduction of the intensity of the forced electric dipole transition of 5D0 → 7F2 relative to the magnetic dipole transition of 5D0 → 7F1.


Introduction
Pressure-induced modifications present an additional degree of freedom to tailor the physical properties of amorphous materials [1]. Depending on the applied temperature during a compression experiment, significant amounts of network compaction and pressure-induced property variations can be obtained by mild treatment, i.e., in the pressure range below 1-2 GPa [2,3]. For oxide glasses, these modifications can be detected on the basis of structural analyses, for example, by monitoring changes in the speciation of the network-forming cations (e.g., Si 4+ , B 3+ or Al 3+ ), modifications of network topology, changes in the proportion of non-bridging oxygens, or variability in the occurrence of superstructural units [1,[4][5][6][7][8][9][10]. In turn, these structural effects affect the macroscopic properties, e.g., Young's modulus, indentation hardness, and chemical durability [1,5,7,[11][12][13][14][15].
High-pressure experiments on glasses are typically done at room temperature, enabling densification (to some extent reversible upon decompression) at very high pressure, often in the range of 10-100 GPa. Hot-compression, on the other hand, enables permanent densification (glasses remain in their densified state unless they are re-heated to around the glass transition temperature, T g ) already at moderate pressure (~100-1000 MPa) when the glass is simultaneously heated in the vicinity of T g [1]. Of particular interest, this latter approach enables the preparation of larger, homogeneously compacted specimens, so that pressure-induced variations of macroscopic properties as well as spectroscopic analyses are more readily accessible [1].
Besides mechanical and chemical properties, densification also impacts the radiative and non-radiative properties of luminescent ions and the phonon energy of their corresponding host lattice [16,17]. This includes a decrease in the distance between active ions and their ligands, or between activator and sensitizer species, thus affecting the statistics of activator clustering and associated quenching effects [18]. Through affecting ligand bond length and geometrical arrangement, pressurization may directly act on the magnitude of the local crystal field surrounding the luminescent centers, and, thus, affect the position of the energy levels, the splitting of the Stark sublevels, or the activation of otherwise forbidden intra-configurational electric dipole transitions by the Laporte selection rule. Indeed, the nephelauxetic effect induced by network compaction may increase the oscillator strength of these transitions, what can be an interesting route to boost the intensity of the absorption lines of rare earth ions [16,19].
Due to these different aspects of sensitivity, the luminescence behaviour of active ions has received much attention for probing pressureinduced structural variations. Such experiments have been performed on various crystalline and amorphous hosts doped with rare earth or transition metal ions, as reviewed in Refs. [16,17] and references therein. However, studies on the effect of hot-compression on the luminescence behaviour of glasses are scarce. For example, even though Eu 3+ is one of the most-studied luminophores at ambient pressure or when embedded in cold-compressed matrices [20][21][22][23][24][25][26], there are no reports on its fluorescence or absorption line variations as a result of hot-compression treatment. Indeed, trivalent europium is often used as an active ion in optical thermometry or photonic devices; its non-degenerate ground state 7 F 0 and excited state 5 D 0 give easily interpreted luminescence and absorption spectra [27,28]. In addition, the characteristic occurrence of the hypersensitive forced electric dipole transition of 5 D 0 → 7 F 2 (~612 nm) has made the Eu 3+ ion a well-studied probe for examining variations on local ligand symmetry in disordered materials [29,30].
Considering all the attributes of Eu 3+ as a probe and active ion, we now investigate hot compression-induced effects on the luminescence behaviour of a Eu 3+ -doped silicate glass. Starting from the pressuredriven changes in density, glass transition temperature and structure (using Raman spectroscopy) of doped and blank samples, we explore the radiative and non-radiative properties of trivalent europium ions using steady state luminescence, time-resolved luminescence, and phonon sideband spectroscopy techniques. This is done in order to obtain insight into the local structure of the dopant and on the electronphonon coupling strength of the host matrix.

Glass preparation
Silicon dioxide, magnesium oxide, calcium carbonate, sodium carbonate, potassium carbonate, and europium oxide with high purity (p.a.) were selected as starting materials for glass preparation. The undoped matrix glass had the composition of (wt%) 71.94 SiO 2 •5.64 MgO•9.23 CaO•13.17 Na 2 O•0.02 K 2 O, while the two doped glasses contained an additional 700 and 2000 ppm of Eu 2 O 3 , respectively. In the following, the samples are denoted SGx, whereby x indicates the dopant concentration of Eu 2 O 3 in wt.-ppm. The specific matrix glass was chosen as representative for the most common types of alkali-alkaline earth silicate glasses. The relatively low europium concentrations were selected to avoid undesirable clustering of Eu 3+ ions and, thus, luminescence quenching phenomena upon densification [18]. The batch materials were ground in a porcelain mortar and transferred to a platinum crucible. The batch was then heated progressively up to 1500°C, melted for 2 h, poured into a graphite mould and annealed at the glass transition temperature of the parent glass.

Hot-compression
Compression procedures have been detailed in previous reports [1,12,13]. Here, glass samples were quenched from high pressures at well-controlled cooling rates in order to be able to distinguish between thermal and pressure-induced effects. To access the onset of pressureinduced structural alterations of the system, experiments were carriedout in the intermediate pressure regime, which also enabled the use of relatively large samples. Compression experiments were performed at pressures of up to 2 GPa in a cold seal pressure vessel. The samples were loaded directly into the vessel, and nitrogen gas was employed as the compression medium. The samples were equilibrated at a temperature around the glass transition temperature of the parent glasses for 30 min so that the equilibration time in these concomitant high-pressure and high-temperature experiments was well above the relaxation time of the glass and/or liquid.
Two nitrogen gas pressure chambers were used: one vertically positioned with an internal diameter of 6 cm for compression at 1 GPa and the other horizontally positioned with an internal diameter of 3 cm for compression at 2 GPa. A multi-zone and a single-zone cylindrical graphite furnace were used for compression at 1 and 2 GPa, respectively. To monitor the temperature during the experiments, PtRh6%-PtRh30% thermocouples were used. They were arranged in the furnaces and coupled with the input power control systems. In both cases, the pressure was measured by manganin gauges. The pressures and temperatures were stabilized with an accuracy of 1 MPa and 0.1°C, respectively. Glass samples were placed in an alumina crucible and then heated in the furnaces inside the high-pressure reactors with a constant heating rate of 600°C/h to the ambient pressure T g value. The systems were kept at these conditions under high nitrogen pressure (1 or 2 GPa) for 30 min. Afterwards, the furnace was cooled down to room temperature at a constant rate of 60°C/h, followed by decompression at a rate of 30 MPa/min.

Characterization
The density of the investigated glasses was obtained by the Archimedes' method at room temperature. Measurements were carried out on bulk samples and using distilled water as the immersion liquid. The glass transition temperatures T g were measured via differential scanning calorimetry (DSC) using a Netzsch STA 449 Jupiter calorimeter with a heating rate of 10°C/min under N 2 atmosphere with an accuracy of ± 2°C.
Raman spectra were recorded using a Renishaw InVia Raman microscope operating at the laser excitation wavelength of 514.5 nm. The spectra were recorded for wavenumbers ranging from 100 to 1450 cm −1 with a resolution of 2 cm −1 .
Luminescence spectra were recorded using a high-resolution spectrofluorometer (Fluorolog, Horiba Jobin-Yvon). Time-resolved luminescence measurements were performed on the same device using a 70 W Xe flash lamp as the excitation source and time-correlated single photon counting (TCSPC). For the luminescence decay analyses, the excitation and emission wavelengths were fixed at 392 and 611 nm, respectively, and the raw data were fitted using a double-exponential function to obtain the average lifetime.

Results
Starting densities and the degree of densification of the blank and doped compositions are presented in Table 1. The densities of the three glasses (i.e., 0, 700, and 2000 ppm Eu 2 O 3 ) increase uniformly from 2.56 g/cm 3 (ambient pressure) to 2.63 g/cm 3 (1 GPa) and then to 2.68 g/cm 3 (2 GPa). The degree of densification is around 2.6% and 4.6% at 1 GPa and 2 GPa, respectively, for both the blank and doped glass series. The variation of T g of the blank and doped samples is presented in Table 1. Corresponding DSC data are given in Fig. 1. For all samples, T g increases by roughly 5-10°C as a result of compaction.
Raman spectra of the blank and doped compositions are presented in Fig. 2. The presented spectra were normalized to the area of the band at around 780 cm −1 . Considering first the low-frequency region between 400 and 700 cm −1 , we observe two bands at~490 and 624 cm −1 , known as the defect lines D 1 and D 2 , which can be related to the symmetric breathing vibration of the four-and three-membered silicate rings, respectively [31]. The band at about 546 cm −1 is assigned to the bending vibration of Si-O-Si bridges [31,32]. In the midfrequency range between 700 and 900 cm −1 , the band at~780 cm −1 corresponds to the symmetric stretching vibrations of Si-O-Si bridges [31,32]. In the high-frequency range between 900 and 1300 cm −1 , the bands at about 955 and 1090 cm −1 are related to the asymmetric stretching vibrations of SieO bonds embedded in the metasilicate (Q 2 ) and disilicate (Q 3 ) species, respectively [31,32].
There are no notable shifts in the Raman band positions of the three base glasses upon densification. However, the bands´relative intensities in the low-and high-frequency ranges are affected by the hot compression treatment. That is, the bands related to metasilicates and disilicates increase slightly with pressure. The same is seen even more pronouncedly for the low-frequency bands assigned to silicate rings and Si-O-Si bridges.
Normalized luminescence spectra (recorded using an excitation wavelength of 392 nm) are also reported in Fig. 3c-d. They show the emission lines of Eu 3+ ions starting from the 5 D 0 excited level to the 7 F j ground manifold, 7 F 0 , 7 F 1 , 7 F 2 , 7 F 3 and 7 F 4 energy levels at~578, 591, 611, 653 and 703 nm, respectively [27].
Time-resolved luminescence data are presented in Fig. 3e-f; lifetimes of the 5 D 0 → 7 F 2 transition are listed in Table 2. For the analyses, the excitation and emission wavelengths were fixed at 392 and 611 nm, respectively. For the SG700 glasses, the lifetime decreases from 2.55 ms to 2.38 and 2.30 ms as a result of hot-compression at 1 GPa and 2 GPa, respectively. Likewise, for the SG2000 glass series, the lifetime is reduced from 2.63 ms in the pristine sample to 2.58 and 2.49 ms, respectively. Table 2 further presents the bandwidth of the 5 D 0 → 7 F 2 transition and the asymmetry ratio R = I( 5 D 0 → 7 F 2 )/I( 5 D 0 → 7 F 1 ) [28] (calculated using the intensity of the 5 D 0 → 7 F 2 over that of the 5 D 0 → 7 F 1 emission lines). The 5 D 0 → 7 F 2 transition is an induced hypersensitive electric dipole transition, whose intensity is influenced by the local symmetry of Eu 3+ cations. On the other hand, the 5 D 0 → 7 F 1 transition is a magnetic dipole transition and its intensity is independent of the local environment of Eu 3+ . The electron-phonon coupling strength g of the host matrix is also listed in Table 2. This latter value was calculated using the areas of the zero phonon line (which appears at 463 nm in the  excitation spectra) and of the phonon sideband associated with the almost pure electric dipole 7 F 0 → 5 D 2 transition (which appears at about 440 nm [33,34]). Bandwidth, R and g parameters all change consistently when the glasses are hot-compressed: an increase occurs in the bandwidth of the 5 D 0 → 7 F 2 emission line with pressure, as well as a decrease in the lifetime and in the asymmetry ratio. However, the electron-phonon coupling strength remains constant for both dopant concentrations.

Discussion
On a macroscopic scale, hot-compression increases the density and the glass transition temperature of the investigated glasses. This is in accordance with previous studies [3]. The overall degree of compaction which is achieved at 2 GPa corresponds to about one third of the maximum degree of compression of a typical soda lime silicate glass with similar former-to-modifier ratio.
On super-structural scale, compression reduces the distances between active ions and their ligands in the host matrix. This enhances the probability of interaction between ligands and 4f electrons by penetration of the ligand electron clouds into the ion configurations and/or Fig. 3. Photoluminescence excitation spectra for λ em = 611 nm (a,b), luminescence spectra for λ ex = 392 nm (c,d) and luminescence decay (e,f) of the annealed and hot-compressed SG700 and SG2000 glasses. by covalent admixture of the 4f electrons with the ligand orbitals [16,17]. These interactions expand the oscillator strengths of the transitions, making them more allowed. In our study, even though the Raman spectra do not exhibit notable variations and the degree of densification is low, these interactions are already visible. This is probably due to the increasing crystal field strength and the nephelauxetic effect caused by compression. Together, these effects contribute to a slight increase in the intensity of the absorption lines of Eu 3+ , reduce the lifetime of the 5 D 0 → 7 F 2 transition as shown in Table 2, and influence the intensity of the emission lines.
In addition to oscillator strength, also the phonon energy of the host matrix plays a role in the lifetime variation. The evolution of the phonon energy can be observed by an uptick in the amplitude of Raman bands versus pressure, as shown in Fig. 2. Indeed, losses by non-radiative processes are expected to increase when the glasses are compressed due to the creation of new de-excitation paths, which alter the photoluminescence quantum yield and simultaneously the transitionsĺ ifetimes [16,17,28]. These new paths of de-excitation can also be detected when observing the declining intensity of the photoluminescence excitation lines for both doped glasses (Fig. 3, a and b).
Meanwhile, the regularity of emission lines for both doped samples versus pressure (Fig. 3c-d) is interesting, since room temperature densification (at high pressure) often reduces the luminescence intensity [16,17].
The fluorescence is also influenced by the expected increase in refractive index upon densification [1], and by the invariance of electronphonon coupling ( Table 2). The fact that the values of g do not change with pressure suggests that the vibration of the host glass perceived locally by the active ions is not strongly pressure-dependent. This corroborates earlier findings according to which hot-compression results in decoupled relaxation reactions on short and longer-range order [3]. Since the ion concentration in the studied glasses is low, the g factor can be considered as a reliable parameter [35]. While Raman spectroscopy probes the average phonon energy of the host, the local vibrational energy of the site in which Eu 3+ ions are embedded can be better assessed using phonon sideband spectroscopy [35]. On the basis of the present observations, we suggest that all the above factors counterbalance the decrease in luminescence intensity (such as commonly observed in room temperature compression experiments) by compensating the losses through non-radiative processes.
Our results can also be used to understand in which sites the active ions are embedded and how hot-compression affects the splitting of the Eu 3+ energy levels and Stark sublevels. Indeed, Fig. 3a-d show a slight blueshift of some of the transitions upon high-temperature densification of the samples. This indicates an expansion of Eu 3+ energy levels, rarely observed during room temperature compression [16]. While such features are not easily understood in the absence of further data, we may discuss them from a simplistic viewpoint of glass structure [16]. Considering that the Eu 3+ ions are smaller than the Ca 2+ ions (1.21 vs. 1.26 Å for eightfold coordination [37]) and that trivalent europium species are embedded in the voids of the silicate network at ambient pressure, the availability of free volume may be larger in the vicinity of europium ions. Upon densification, this excess free volume is consumed without causing a notable change in the silicate backbone (denoted congruent compaction [36]); the glasses are compacted homogeneously as a function of applied pressure. In this case, one would expect increasing site symmetry, such as seen on the decrease in the asymmetry ratio with degree of compaction (Table 2). Finally, the increase in bandwidth of the 5 D 0 → 7 F 2 emission line upon hot-compression can be explained by a growth in the electron population of the lower stark sublevels and, at the same time, increasing crystal field strength.

Conclusion
We investigated the effects of hot-compression at moderate pressure on the luminescence behaviour of Eu 3+ ions embedded in a silicate glass matrix. Density and T g measurements supported by vibrational spectroscopy confirmed densification of the silicate network without distinct structural variations. This densification influences the luminescence of Eu 3+ in multiple ways: the oscillator strength increases, new paths of de-excitation are created, and variations occur in the forced hypersensitive electric dipole transitions. All three effects were observed on the lifetime of the 5 D 0 → 7 F 2 emission line and on the intensities of the transitions as seen in the excitation spectra. However, the overall luminescence intensity of the 5 D 0 → 7 F J emission lines was preserved, partly due to the expected increase in refractive index, and partly because of constant electron-phonon coupling strength. On a more general perspective, these findings suggest that the radiative and non-radiative properties of luminescent centers are not affected similarly by hot-versus cold-compression. Such understanding -if corroborated by further investigations using different matrices and different luminescent ions -can be used when applying Eu 3+ and similar ions as structural probes, but also in the design of photonic devices with enhanced emission bandwidth and spectral coverage.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.