Densification of Lithium Disilicate under High Pressure Investigated by XPS

The aim of this work was to investigate by X-ray photoelectron spectroscopy the effect of high pressure on the chemical environments of Si 2p, O 1s and Li 1s in lithium disilicate glass ceramic with stoichiometric composition Li2O·2SiO2 (LS2). A group of samples was processed at 2.5 GPa, 4 GPa and 7.7 GPa at room temperature and a second group was crystallized under high pressure and high temperature. Large shifts of the binding energy toward higher energies were observed in the X-ray photoelectron spectroscopy spectra for samples of the first group after densification at 2.5 and 4 GPa. For samples processed at 7.7 Gpa, the major component of the binding energy for the Si 2p environment remained practically unchanged compared to the pristine sample but new components, with smaller intensities, appeared in the spectra, indicating the existence of distinct Q-species induced by high pressure. This behavior may be related to changes in the number of bridged and non-bridged oxygen atoms in the glass structure. The results for the second group of samples, crystallized under high pressure, showed evidences of three binding energies for the O atoms, one of them related to non-bridged and two of them to bridged O atoms.


Introduction
Structural analysis of glasses under high pressure is important to understand the effect of densification mechanisms on the glass properties [1].Bridgman et al. [2] pointed out that the permanent densification after applying high pressure and high temperature (HPHT) is a phenomenon observed only for glassy materials due to their structural freedom.
Kitamura et al. [1] investigated the densification of lithium disilicate Li 2 O•2SiO 2 (LS 2 ) glasses at 6 GPa and 400˚C.Based on Raman spectroscopy and radial distribution function, they attributed the densification to an increase of the packing density of SiO 4 tetrahedra due to a decrease of the Si-O-Si bond angle between the tetrahedra.They showed the Si-O-Si bond angle decreased from 109.8˚ to 108.8˚ and the Si-O length increased by 0.001 nm.
Schmelzer et al. [3] discussed the pressure dependence of viscosity and free volume of liquids in frozen systems in non-equilibrium states.According to them, the viscosity of simple liquids typically increased with pressure.
The crystallization of LS 2 under HPHT has been stud-ied in the past few years [1,[4][5][6][7][8][9][10][11].In previous works, the effect of densification on the crystallization process and thermal properties of LS 2 glass was investigated [4], as well as the effect of pressure and temperature on the mechanical [5] and optical [6] properties.It was observed that the hardness and elastic modulus of the sample processed at HPHT increased noticeably compared to the samples crystallized at atmospheric pressure.The refractive index of LS 2 was also significantly affected by HPHT.
Nuclear Magnetic Resonance (NMR) results obtained by Fuss et al. [9] indicated the formation of lithium metasilicate solid solution (LS-ss) for samples submitted to 6 GPa and thermal treatment below 750˚C.According to the authors, the difference between lithium metasilicate (MS) and LS-ss was not detectable by X-ray diffraction since there were no changes in the crystal symmetry of both structures.The difference would be related to the number of bridged (BO) and non-bridged (NBO) oxygen atoms.Buchner et al. [6], on the other hand, showed the formation of crystalline MS after processing LS 2 at 7.7 GPa and 455/500˚C for nucleation followed by 610˚C for crystal growing.
Adams & De Jong [12] investigated the early stages of crystallization of a slightly hyper stoichimetric LS 2 sample (34.5 mol% Li 2 O) at 450˚C using X-ray photoelectron spectroscopy (XPS) and they ascribed the presence of an extra Li 1s peak in the XPS spectrum (~51.7 eV) to the formation of a MS phase.Soares et al. [13], however, repeated the XPS analysis at the same conditions and concluded that this extra Li 1s peak was related to contamination.Moreover, they concluded that XPS was not a suitable technique to detect structural changes in LS 2 when the volume fraction of the crystals was very low.On the other hand, the formation of the MS phase in the early stages of nucleation was confirmed by them using TEM analysis [14].
Nesbitt et al. [15] investigated the chemical environment around Na 1s, O 1s and Si 2p atoms in Na 2 O-SiO 2 glass as a function of the glass composition.Six compositions of Na 2 O-SiO 2 glass ranging from 100 to 45 mol% SiO 2 were investigated by XPS and NMR.The Si 2p 3/2 peak maximum shifted to lower binding energy (BE) with increasing Na 2 O content.According to the authors, the BE shift indicated increased valence electron density on the Si nuclei with increased Na 2 O concentration.Each Q-species should yield a peak at slightly different BE since the electron density over each Si atom should be different, depending on the number of BO and NBO bonded to it.
According to Sawyer et al. [16], the BE of the Q-species are strongly affected by the number of NBO bonded to the Si atom, with BE decreasing systematically from Q 4 to Q 0 species.They observed the shift of the peak toward lower BE with increasing K 2 O content in K 2 O-SiO 2 glass.Two doublets in the Si 2p spectra of different glass compositions were identified.According to the authors, the strongly asymmetric nature of the Si 2p spectra of the potasic glasses may result from two or more Q-species contributions to the Si 2p spectra and the intensity of the doublet would depend on the amount of the particular Q-specie.
In this context, the aim of this work was to investigate the effect of high pressure on the chemical environment of Si 2p, O 1s and Li 1s atoms of the LS 2 glass structure using XPS to help elucidate the mechanisms responsible for the early stages of the formation of the metasilicate structure induced by densification.For comparison, the chemical environment of these atoms was also investigated after complete crystallization of the glass structure under high pressure and high temperature.

Sample Preparation
Lithium disilicate glass of stoichiometric composition Li 2 O•2SiO 2 (LS 2 ) was prepared using standard reagent grade Li 2 CO 3 (Aldrich Chem.Co., 99+%) and ground quartz (<99.9%SiO 2 ).The 200 g batch was melted in a Pt crucible at 1450˚C during 2 hours in an electric furnace.The melt was poured on a steel plate, annealed below the glass transition temperature, at 430˚C, during 1 h and cooled down slowly to room temperature.

High Pressure Experiments
The HP experiments were performed in a toroidal high pressure apparatus.The specific configuration and type of high-pressure chamber used in the experiments were optimized for providing quasi-hydrostatic pressure [17].The pressure calibration was carried out by the "fixedpoints" technique using Bi and Yb, which permitted the calibration of the pressure at the following points: Bi with phase transitions at 2.5 GPa and 7.7 GPa, and Yb with a phase transition at 4.0 GPa.A detailed description of the high-pressure calibration method has been provided elsewhere [17].
During the HPHT experiments, a monolithic LS 2 sample was placed inside a hexagonal boron nitride (hBN) capsule which acted as a soft pressure transmitting medium.The hBN capsule was placed inside a graphite cylinder which acted as a heater element.For experiments performed at HP and room temperature (HPRT), the capsule of hBN was replaced by a lead capsule, which is a softer pressure transmitting compared to hBN.
In the first group, the samples were submitted to 2.5 GPa, 4 GPa and 7.7 GPa at room temperature during 5 min inside the lead capsule.In the second group, the samples were submitted to HP (2.5, 4.0 and 7.7 GPa) inside the hBN capsule and, simultaneously, to the following thermal treatment: 500˚C during 2 h for nucleation followed by 610˚C during 0.5 h for crystal growing.For comparison, a pristine glass sample and a sample submitted to the same heat treatment at 1 atm were also investigated.
According to a previous work [6], the processing at 2.5 GPa and high temperature induced the crystallization of the monoclinic phase of lithium disilicate (Li 2 Si 2 O 5 ).At 4.4 GPa it was observed the formation of the orthorhombic phase with the same stoichiometry and, at 7.7 GPa the glass structure crystallized as MS (Li 2 SiO 3 ).
After HPRT and HPHT processing, the surface of the samples was grounded on SiC abrasive paper up to #1200 and polished with CeO 2 slurry for XPS analyses.

Analytical Techniques-XPS
XPS analyses were performed in an Omicron-SPHERA station using Al Kα radiation (1486.6 eV).The anode was operated at 225 W (15 kV, 15 mA).Survey spectra were recorded with a 50 eV pass energy.Si 2p, O 1s, and Li 1s regions were recorded with high resolution (pass energy of 20 eV).The detection angle of the photoelectrons (Θ) with respect to the sample surface (take-off angle) was fixed at 53˚ for all measurements.The C 1s signal from adventitious carbon at 284.6 eV was used as an internal energy reference.All spectra were fitted assuming a Shirley background.Lines were fitted by 70% Gaussian + 30% Lorentzian functions with set values of full width at half maximum for each line.

Densification at Room Temperature (HPRT)
Figures 1-3 show high resolution XPS spectra for Si 2p, O 1s and Li 1s, respectively, after processing at high pressure compared to the pristine sample.It is clearly seen that the BE of the three elements changed irreversible after densification, revealing the strong influence of high pressure in the chemical environment due to the structural freedom of the glassy LS 2 .
According to Figure 1, Si 2p presents two doublets from 2.5 to 7.7 GPa, corresponding to Si 2p 3/2 and Si 2p 1/2 .However, the intensity and BE of each doublet changed with pressure.Figure 2 shows that there is practically only one BE component for O 1s up to 2.5 GPa,   while after processing at 4 and 7.7 GPa there are two components.In these cases, the component with higher intensity corresponds to BO while the lower intensity component corresponds to NBO.For Li atoms (Figure 3), only at 7.7 GPa it is possible to identify a second component of BE.
Table 1 shows the values of the BE obtained in this work by fitting the experimental data compared to the results found in the literature for LS 2 .

Crystallization under High Pressure (HPHT)
Figures 4-6 show the XPS results for the samples crystallized under high pressure.In this case there was practically no shift in the BE as a function of pressure.
Figure 4 shows that the formation of doublets for Si 2p 3/2 and Si 2p 1/2 now happens only after processing at 7.7 GPa. Figure 5 shows that there were three components of BE for O 1s up to 7.7 GPa, one associated to NBO and two associated to BO.The intensity of the component with higher BE decreases with increasing pressure.For Li atoms (Figure 6), only at 7.7 GPa it was possible to identify a component of the BE.Soares et al. [18] already reported that the detection of the Li 1s XPS signal from crystalline samples of LS 2 was difficult.
The BE measured for the pristine glass sample and for the crystalline sample obtained at atmospheric pressure are similar to the results found in the literature [12,13,18], as shown in Table 1.The permanent shift of the BE toward higher energy and the change in the intensity of the peaks for Li 1s, O 1s and Si 2p observed in Figures 1-3 indicated that the densification affects irreversibly the glass structure, even without crystallization.
The large shift of the BE toward higher energies observed for Si 2p after processing at 2.5 and 4 GPa (Figure 1) suggests a stronger interaction with the surrounding oxygen atoms compared to the pristine glass.According to Sawyer et al. [16], the BE decreases systematically from Q 4 to Q 0 species for Si 2p.According to Figure 1, two Q-species were observed in all samples for Si 2p, but for pristine samples the intensity of the Q-species of lower energy was very small.Probably in pristine glass samples the Q 3 specie was dominant and the small peak of lower energy corresponded to residual Q 2 species.In the case of O 1s atoms (Figure 2), the peak at higher energy (~531 eV) for the pristine sample corresponds to BO atoms (Si-O-Si) while the other component (~528.8eV) corresponds to the NBO atoms (Si-O-Li) [13,19].After processing at 2.5 and 4 GPa the BE of the BO atoms shifted irreversibly to higher energies, indicating a larger interaction with the Si atoms, in agreement with the behavior of the Si 2p XPS spectrum.The intensity of the peak related to the NBO atoms was very low for the pristine sample and after processing at 2.5 GPa the intensity of this peak slightly increased after processing at 4 and 7.7 GPa, probably indicating a decomposition of the Q 3 into the Q 2 species at higher pressure.
Figure 3 shows that the BE for Li 1s also shifted after processing at 2.5 GPa and 4 GPa, and a second component at 52.1 eV was detected after processing at 7.7 GPa.According to Adams & De Jong [12,20], this component may be related to a local arrangement that would be favorable to the formation of the metastable phase during heat treatment.
Considering the XPS spectra for crystalline samples produced under high pressure, Figure 4 shows that the Si 2p presents only one chemical environment related to Q 3 species formed during crystallization of lithium disilicate (Li 2 •Si 2 O 5 ).The Si 2p 3/2 spectrum for the sample crystallized at 7.7 GPa, however, presents a doublet indicating the formation of two structures, one of them related to lithium metasilicate Li 2 •SiO 3 and the other, to residual SiO 2 to satisfy stoichiometric restrictions [6].
For O 1s (Figure 5), the smaller BE corresponds to the NBO atoms (Si-O-Li) [13,19] and it was observed for all pressure range investigated.For pressures up to 4 GPa, the intensity and energy of this peak remained practically unchanged, compatible with the formation of crystalline lithium disilicate.At 7.7 GPa, it was observed a small shift of this component toward higher energy and a slight decrease of its intensity.This result is in agreement with the changes observed in the spectrum for Si 2p at the same pressure, indicating a distinct configuration of Q-species at higher pressures.
Two peaks related to BO atoms, of higher BE, were observed for all pressure range in the crystalline samples.The most intense component was probably related to Si-O-Si and the other one may be related to BO-Li.Ching et al. [21] also observed two types of BO atoms in crystalline Na disilicate.According to them, one type was assigned to Si-O-Si as in vitreous silica whereas the other type of BO atom was also bonded to a Na atom with the BO-Na bond length being similar to the NBO-Na bond length [22].Du and Cormack [23] and Mountjoy [24] investigated by molecular dynamics simulations the formation of two BO in sodic glasses.The experimental results obtained by Nesbitt et al. [15] for sodic glass also provided evidences for two types of BO, one bonded to two Si atoms and one bonded to a Na atom as well as to two Si atoms.Sawyer et al. [16] reported that the width of the O 1s XPS peak was highly variable and wider than that of the NBO peak for potasic glass.The variable width would be a consequence of the presence of two types of BO atoms with somewhat different BE; one type bridged to two Si atoms while the other type bridged to three atoms, two Si and one K atom.According to the authors, the abundance of the two types of BO atoms was similar to the abundance of Q 4 and Q 3 species of previous NMR studies.
Fuss et al. [9] observed that at 4.5 GPa and high temperature, LS 2 crystallized as the lithium disilicate phase, consisting only of Q 3 tetrahedra, but the NMR results indicated the presence of a small amount of Q 2 SiO 4 tetrahedra, which could be a glassy or crystalline phase acting as imperfections.In the present work, the crystallization at 4 GPa induced the formation of only the Q 3 species in the Si 2p spectrum, without any evidence of residual Q 2 .At 6 GPa, the results obtained by Fuss et al. suggested that the LS 2 glass densified under pressure and crystallized as a single new phase with a higher density, resembling a slightly deformed MS phase.For this sample, they observed one sharp Q 2 peak in the NMR spectrum associated with crystalline MS, a broad Q 4 band related to glassy SiO 2 and a broad Q 3 band associated with the residual glass.Additionally, they found four more peaks related to Q 3 and Q 4 tetrahedra that could not be related to residual glass.Based on the NMR, Raman and IR results, they suggested the formation of a crystalline lithium metasilicate solid-solution having the same symmetry of MS but formed out of three different tetrahedral species.In this solid solution, four Li ions from the metasilicate structure would be statistically replaced by one Si ion and three vacancies, without changing the crystal symmetry.This replacement would be detected by a shift of the peaks in the XRD pattern related to the change in lattice parameters that was, in fact, observed by the authors and corroborated by the NMR results about the amount of BO and NBO atoms.The XPS results found in the present work for the sample crystallized at 7.7 GPa indicated two chemical environments around Si atoms, associated with Q 3 species of lithium metasilicate (Li 2 SiO 3 ) and Q 2 associated to residual SiO 2 .No evidences of other Q species were observed.Maybe the reason for the differences between the consistent results found by Fuss et al. [9] and the results found in this work is related to the thermal treatment: they investigated temperatures from 608˚C during 20 min up to 753˚C during 60 min at 6 GPa.In this work we considered 500˚C during 2 h for nucleation followed by 0.5 h at 610˚C for crystal growing, both under high pressure.Probably the higher temperatures used by Fuss et al. [9] were not suitable for nucleation of the crystalline phases, resulting in a solid solution favored at higher temperatures where the average Si-O-Si bonding angle would decrease and there would be at least four different Si sites, according to the NMR results.

Conclusion
Densification of lithium disilicate at room temperature under high pressure induced large and irreversible modifications in the chemical environment of Si 2p, O 1s and Li 1s.These modifications would be related to the rearrangement of the atoms induced by high pressure due to the structural freedom of the glass structure, even at room temperature.XPS showed to be a suitable technique to detect the structural changes induced by densification at room temperature.For samples crystallized under high pressure, it was observed that a similar behavior for pressures up to 4 GPa, related to the formation of monoclinic and orthorhombic lithium disilicate.The in-terpretation of the XPS results for Si 2p and O 1s for the sample crystallized at 7.7 GPa was consistent with the formation of lithium metasilicate and SiO 2 .
FAPERGS for the financial support and to LaMaV for the fusion of the samples.

Figure 1 .
Figure 1.Si 2p region of XPS spectrum of the LS 2 pristine sample and after processing at 2.5 GPa, 4 GPa, and 7.7 GPa at room temperature.

Figure 2 .
Figure 2. O 1s region of XPS spectrum of the LS 2 pristine sample and after processing at 2.5 GPa, 4 GPa, and 7.7 GPa at room temperature.

Figure 3 .
Figure 3. Li 1s region of XPS spectrum of the LS 2 pristine sample and after processing at 2.5 GPa, 4 GPa, and 7.7 GPa at room temperature.

Figure 4 .
Figure 4. Si 2p XPS spectra obtained from crystalline samples produced at atmospheric pressure, 2.5 GPa, 4 GPa, and 7.7 GPa.The thermal treatment was: 500˚C during 2 h for nucleation followed by 610˚C during 0.5 h for crystal growing.

Figure 5 .
Figure 5. O 1s XPS spectra obtained from crystalline samples produced at atmospheric pressure, 2.5 GPa, 4 GPa, and 7.7 GPa.The thermal treatment was: 500˚C during 2 h for nucleation followed by 610˚C during 0.5 h for crystal growing.

Figure 6 .
Figure 6.Li 1s XPS spectra obtained from crystalline samples produced at atmospheric pressure, 2.5 GPa, 4 GPa, and 7.7 GPa.The thermal treatment was: 500˚C during 2 h for nucleation followed by 610˚C during 0.5 h for crystal growing.