Structure of hydrous aluminosilicate glasses along the diopside-anorthite join: A comprehensive one- and two-dimensional H and Al NMR study

We have taken a systematic approach utilizing advanced solid-state NMR techniques to gain new insights into the controversial issue concerning the dissolution mechanisms of water in aluminosilicate melts (glasses). A series of quenched anhydrous and hydrous (∼2 wt% H₂O) glass samples along the diopside (Di, CaMgSi₂O₆)—anorthite (An, CaAl₂Si₂O₈) join with varying An components (0, 20, 38, 60, 80, and 100 mol %) have been studied. A variety of NMR techniques, including one-dimensional (1D) ¹H and ²⁷Al MAS NMR, and ²⁷Al → ¹H cross-polarization (CP) MAS NMR, as well as two-dimensional (2D) ¹H double-quantum (DQ) MAS NMR, ²⁷Al triple-quantum (3Q) MAS NMR, and ²⁷Al → ¹H heteronuclear correlation NMR (HETCOR) and 3QMAS/HETCOR NMR, have been applied. These data revealed the presence of SiOH, free OH ((Ca,Mg)OH) and AlOH species in the hydrous glasses, with the last mostly interconnected with Si and residing in the more polymerized parts of the structure. Thus, there are no fundamental differences in water dissolution mechanisms for Al-free and Al-bearing silicate melts (glasses), both involving two competing processes: the formation of SiOH/AlOH that is accompanied by the depolymerization of the network structure, and the formation of free OH that has an opposite effect. The latter is more important for depolymerized compositions corresponding to mafic and ultramafic magmas.Aluminum is dominantly present in four coordination (Al^IV), but a small amount of five-coordinate Al (Al^V) is also observed in all the anhydrous and hydrous glasses. Furthermore, six-coordinate Al (Al^VI) is also present in most of the hydrous glasses. As Al of higher coordinations are favored by high pressure, Al^VIOH and Al^VOH may become major water species at higher pressures corresponding to those of the Earth’s mantle.     

CO2 solubility and speciation in intermediate (andesitic) melts: the role of H2O and composition

We determined total CO₂ solubilities in andesite melts with a range of compositions. Melts were equilibrated with excess C-O(-H) fluid at 1 GPa and 1300°C then quenched to glasses. Samples were analyzed using an electron microprobe for major elements, ion microprobe for C-O-H volatiles, and Fourier transform infrared spectroscopy for molecular H₂O, OH⁻, molecular CO₂, and CO₃²⁻. CO₂ solubility was determined in hydrous andesite glasses and we found that H₂O content has a strong influence on C-O speciation and total CO₂ solubility. In anhydrous andesite melts with ∼60 wt.% SiO₂, total CO₂ solubility is ∼0.3 wt.% at 1300°C and 1 GPa and total CO₂ solubility increases by about 0.06 wt.% per wt.% of total H₂O. As total H₂O increases from ∼0 to ∼3.4 wt.%, molecular CO₂ decreases (from 0.07 ± 0.01 wt.% to ∼0.01 wt.%) and CO₃²⁻ increases (from 0.24 ± 0.04 wt.% to 0.57 ± 0.09 wt.%). Molecular CO₂ increases as the calculated mole fraction of CO₂ in the fluid increases, showing Henrian behavior. In contrast, CO₃²⁻ decreases as the calculated mole fraction of CO₂ in the fluid increases, indicating that CO₃²⁻ solubility is strongly dependent on the availability of reactive oxygens in the melt. These findings have implications for CO₂ degassing. If substantial H₂O is present, total CO₂ solubility is higher and CO₂ will degas at relatively shallow levels compared to a drier melt. Total CO₂ solubility was also examined in andesitic glasses with additional Ca, K, or Mg and low H₂O contents (<1 wt.%). We found that total CO₂ solubility is negatively correlated with (Si + Al) cation mole fraction and positively correlated with cations with large Gibbs free energy of decarbonation or high charge-to-radius ratios (e.g., Ca). Combining our andesite data with data from the literature, we find that molecular CO₂ is more abundant in highly polymerized melts with high ionic porosities (>∼48.3%), and low nonbridging oxygen/tetrahedral oxygen (<∼0.3). Carbonate dominates most silicate melts and is most abundant in depolymerized melts with low ionic porosities, high nonbridging oxygen/tetrahedral oxygen (>∼0.3), and abundant cations with large Gibbs free energy of decarbonation or high charge-to-radius ratio. In natural silicate melt, the oxygens in the carbonate are likely associated with tetrahedral and network-modifying cations (including Ca, H, or H-bonds) or a combinations of those cations.     

Solubility mechanisms of fluorine in peralkaline and meta-aluminous silicate glasses and in melts to magmatic temperatures

Structural interaction between dissolved fluorine and silicate glass (25°C) and melt (to 1400°C) has been examined with ¹⁹F and ²⁹Si MAS NMR and with Raman spectroscopy in the system Na₂O-Al₂O₃-SiO₂ as a function of Al₂O₃ content. Approximately 3 mol.% F calculated as NaF dissolved in these glasses and melts. From ¹⁹F NMR spectroscopy, four different fluoride complexes were identified. These are (1) Na-F complexes (NF), (2) Na-Al-F complexes with Al in 4-fold coordination (NAF), (3) Na-Al-F complexes with Al in 6-fold coordination with F (CF), and (4) Al-F complexes with Al in 6-fold, and possibly also 4-fold coordination (TF). The latter three types of complexes may be linked to the aluminosilicate network via Al-O-Si bridges.The abundance of sodium fluoride complexes (NF) decreases with increasing Al/(Al + Si) of the glasses and melts. The NF complexes were not detected in meta-aluminosilicate glasses and melts. The NAF, CF, and TF complexes coexist in peralkaline and meta-aluminosilicate glasses and melts.From ²⁹Si-NMR spectra of glasses and Raman spectra of glasses and melts, the silicate structure of Al-free and Al-poor compositions becomes polymerized by dissolution of F because NF complexes scavenge network-modifying Na from the silicate. Solution of F in Al-rich peralkaline and meta-aluminous glasses and melts results in Al-F bonding and aluminosilicate depolymerization.Temperature (above that of the glass transition) affects the Qⁿ-speciation reaction in the melts, 2Q³ ⇔ Q⁴ + Q², in a manner similar to other alkali silicate and alkali aluminosilicate melts. Dissolved F at the concentration level used in this study does not affect the temperature-dependence of this speciation reaction.     

Ab initio calculation of H, O, Al and Si NMR parameters, vibrational frequencies and bonding energetics in hydrous silica and Na-aluminosilicate glasses

Ab initio, molecular orbital (MO) calculations were performed on model systems of SiO₂, NaAlSi₃O₈ (albite), H₂O-SiO₂ and H₂O-NaAlSi₃O₈ glasses. Model nuclear magnetic resonance (NMR) isotropic chemical shifts (δ_iso) for ¹H, ¹⁷O, ²⁷Al and ²⁹Si are consistent with experimental data for the SiO₂, NaAlSi₃O₈, H₂O-SiO₂ systems where structural interpretations of the NMR peak assignments are accepted. For H₂O-NaSi₃AlO₈ glass, controversy has surrounded the interpretation of NMR and infrared (IR) spectra. Calculated δ_iso¹H, δ_iso¹⁷O, δ_iso²⁷Al and δ_iso²⁹Si are consistent with the interpretation of Kohn et al. (1992) that Si-(OH)-Al linkages are responsible for the observed peaks in hydrous Na-aluminosilicate glasses. In addition, a theoretical vibrational frequency associated with the Kohn et al. (1992) model agrees well with the observed shoulder near 900 cm⁻¹ in the IR and Raman spectra of hydrous albite glasses. MO calculations suggest that breaking this Si-(OH)-Al linkage requires ∼+56 to +82 kJ/mol which is comparable to the activation energies for viscous flow in hydrous aluminosilicate melts.     

Structure of Cl-containing silicate and aluminosilicate glasses: A Cl MAS-NMR study

Chlorine-35 magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectra were collected at 14.1 and 18.8 Tesla fields to determine the atomic scale structural environments of the chloride ions in anhydrous and hydrous silicate and aluminosilicate glasses containing 0.2 to 0.7 wt% Cl. NMR peaks are broad and featureless, but are much narrower than the total chemical shift range for the nuclide in inorganic chlorides. Peak widths are primarily due to quadrupole interactions and to a lesser extent to chemical shift distributions. Peak positions are quite different for the Na- and Ca-containing glasses, suggesting that most Cl⁻ coordination environments contain network modifier cations. Comparison of peak positions and shapes for silicate and aluminosilicate glasses containing either Na or Ca suggests that there is no obvious contribution from Cl⁻ bonded to Al, and relative quantitation of peak areas indicates that there is no systematic undercounting of ³⁵Cl spins in the aluminous vs. the Al-free samples. In Ca-Na silicate glasses with varying Ca/(Ca + Na), the mixed-cation glasses have intermediate chemical shifts between those of the end members, implying that there is not a strong preference of either Ca²⁺ or of Na⁺ around Cl⁻. Hydrous Na-aluminosilicate glasses with H₂O contents up to 5.9 wt% show a shift to higher frequency NMR signal with increasing H₂O content, while the quadrupole coupling constant (C_Q) remains constant at ∼3.3 MHz. However, the change in frequency is much smaller than that expected if H₂O systematically replaced Na⁺ in the first-neighbor coordination shell around Cl⁻. A series of hydrous Ca-aluminosilicate glasses with H₂O contents up to 5.5 wt% show no shift in NMR signal with increasing H₂O content. The C_Q remains constant at ∼4.4 MHz, again suggesting no direct interaction between Cl⁻ and H₂O in these samples.     

Oxygen sites in hydrous aluminosilicate glasses: the role of Al-O-Al and H2O

We describe here high-field ¹⁷O magic-angle-spinning (MAS) and triple-quantum MAS (3QMAS) NMR spectra for several alkali silicate and Na, K, and Ca aluminosilicate glasses containing up to 10 wt.% water. The H₂O site appears to have a large quadrupolar coupling constant, and its chemical shift increases from Na- to K- glasses, suggesting significant cation-H₂O interactions. In ¹⁷O one-pulse MAS and 3QMAS and ²⁷Al one-pulse NMR experiments, major differences were seen between spectra for anhydrous and hydrous calcium aluminosilicate glasses. The changes in the ¹⁷O MAS spectra can be explained by the addition of an H₂O peak and to the disappearance of an Al-O-Al peak from the ¹⁷O NMR spectrum for the hydrous glass. The ²⁷Al results are consistent with this interpretation.     

Water solubility mechanism in hydrous aluminosilicate glasses: information from Al MAS and MQMAS NMR

New ²⁷Al NMR data are presented in order to clarify the discrepancies in the interpretation of the previous ²⁷Al Magic Angle Spinning (MAS) spectra from hydrous aluminosilicate glasses. The ²⁷Al MAS data have been collected at much higher magnetic field (14.1 and 17.6 T) than hitherto, and in addition, multiple quantum (MQ) MAS NMR data are presented for dry and hydrous nepheline glasses and NaAlSi_7.7O_17.4 glass that, according to the model of Zeng et al. (Zeng Q., Nekvasil H., and Grey C. P. 2000. In support of a depolymerisation model for water in sodium aluminosilicate glasses: Information from NMR spectroscopy. Geochim. Cosmochim. Acta64, 883-896), should produce a high fraction (up to 30%) of Al in Al Q³-OH on hydration. Although small differences in the MAS spectra of anhydrous and hydrous nepheline glasses are observed, there is no evidence for the existence of significant (>∼2%) amounts of Q³ Al-OH in these glasses in either the MAS or MQMAS data.     

Microscopic origins of macroscopic properties of silicate melts and glasses at ambient and high pressure: Implications for melt generation and dynamics

Recent development and advances in solid state NMR, together with theoretical analyses using quantum-chemical calculations and statistical mechanical modeling, have allowed us to estimate and quantify the detailed distributions of cations and anions in model silicate glasses and melts with varying pressure, temperature and composition. How these microscopic, atomic-scale distributions in the melts from NMR and simulations affect the thermodynamic and transport properties relevant to magmatic processes has been extensively explored recently. Here, based on these previous studies, we present a classification scheme to quantify the various aspects of disorder in covalent oxide glasses and melts on scales of less than 1 nm. The scheme includes contributions from both chemical and topological disorder. Chemical disorder can further be divided into [1] connectivity, which quantifies the extent of mixing among framework units (often parameterized by the degree of Al avoidance or phase separation) and the extent of polymerization (mixing between framework and nonframework cations), and [2] nonframework disorder, which denotes the distribution of network-modifying or charge-balancing cations. Topological disorder includes the distribution of bond lengths and angles. We use this framework of disorder quantification to summarize recent progress on the structures of silicate melts and glasses, mainly obtained from 2D triple quantum magic-angle spinning (3QMAS) NMR, as functions of temperature, pressure, and composition.Most glasses and melts studied show a tendency for chemical ordering in connectivity, nonframework disorder and topological disorder at ambient and high pressure. The chemical ordering in framework disorder, a manifestation of energetics in the melts and glasses, contributes to the total negative deviation of activity of oxides from ideal solution in silicate melts (reduced activity). While no definite evidence of clustering among nonframework cations was found, these cations tend to form dissimilar pairs upon mixing with other types of network modifying cations. Topological disorder in silicate glasses and melts tends to increase with increasing pressure, as suggested by increasing bond angle and length distribution, while the chemical order seems to be maintained with pressure. We calculate key macroscopic properties, including the activity coefficient of silica and viscosity, based on the quantitative estimation of the extent of disorder from solid-state NMR, in particular ¹⁷O 3QMAS NMR. Structural ordering in melts may strongly affect the composition of partial melts in equilibrium with solids, increasing the silica composition of partial melts as a result. With increasing chemical order, the configurational entropy decreases, which can be correlated to an increase in viscosity of melts.     

Sulfur speciation and network structural changes in sodium silicate glasses: Constraints from NMR and Raman spectroscopy

Information about the state of sulfur in silicate melts and glasses is important in both earth sciences and materials sciences. Because of its variety of valence states from S²⁻ (sulfide) to S⁶⁺ (sulfate), the speciation of sulfur dissolved in silicate melts and glasses is expected to be highly dependent on the oxygen fugacity. To place new constraint on this issue, we have synthesized sulfur-bearing sodium silicate glasses (quenched melts) from starting materials containing sulfur of different valence states (Na₂SO₄, Na₂SO₃, Na₂S₂O₃ and native S) using an internally heated gas pressure vessel, and have applied electron-induced SKα X-ray fluorescence, micro-Raman and NMR spectroscopic techniques to probe their structure. The wavelength shift of SKα X-rays revealed that the differences in the valence state of sulfur in the starting compounds are largely retained in the synthesized sulfur-bearing glasses, with a small reduction for more oxidized samples. The ²⁹Si MAS NMR spectra of all the glasses contain no peaks attributable to the SiO_4-nS_n (with n > 0) linkages. The Raman spectra are consistent with the coexistence of sodium sulfate (Na₂SO₄) species and one or more types of more reduced sulfur species containing S-S linkages in all the sulfur-bearing silicate glasses, with the former dominant in glasses produced from Na₂SO₄-doped starting materials, and the latter more abundant in more reduced glasses. The ²⁹Si MAS NMR and Raman spectra also revealed changes in the silicate network structure of the sulfur-bearing glasses, which can be interpreted in terms of changes in the chemical composition and sulfur speciation.     

Solution mechanisms of H2O in depolymerized peralkaline melts

Solubility mechanisms of water in depolymerized silicate melts quenched from high temperature (1000°-1300°C) at high pressure (0.8-2.0 GPa) have been examined in peralkaline melts in the system Na₂O-SiO₂-H₂O with Raman and NMR spectroscopy. The Na/Si ratio of the melts ranged from 0.25 to 1. Water contents were varied from ∼3 mol% and ∼40 mol% (based on O = 1). Solution of water results in melt depolymerization where the rate of depolymerization with water content, ∂(NBO/Si)/∂X_H2O, decreases with increasing total water content. At low water contents, the influence of H₂O on the melt structure resembles that of adding alkali oxide. In water-rich melts, alkali oxides are more efficient melt depolymerizers than water. In highly polymerized melts, Si-OH bonds are formed by water reacting with bridging oxygen in Q⁴-species to form Q³ and Q² species. In less polymerized melts, Si-OH bonds are formed when bridging oxygen in Q³-species react with water to form Q²-species. In addition, the presence of Na-OH complexes is inferred. Their importance appears to increase with Na/Si. This apparent increase in importance of Na-OH complexes with increasing Na/Si (which causes increasing degree of depolymerization of the anhydrous silicate melt) suggests that water is a less efficient depolymerizer of silicate melts, the more depolymerized the melt. This conclusion is consistent with recently published ¹H and ²⁹Si MAS NMR and ¹H-²⁹Si cross polarization NMR data.     

The distribution of sodium ions in aluminosilicate glasses: a high-field Na-23 MAS and 3Q MAS NMR study

The local configurations around sodium ions in silicate glasses and melts and their distributions have strong implications for the dynamic and static properties of melts and thus may play important roles in magmatic processes. The quantification of distributions among charge-balancing cations, including Na⁺ in aluminosilicate glasses and melts, however, remains a difficult problem that is relevant to high-temperature geochemistry as well as glass science.Here, we explore the local environment around Na⁺ in charge-balanced aluminosilicate glasses (the NaAlO₂-SiO₂ join) and its distribution using ²³Na magic-angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy at varying magnetic fields of 9.4, 14.1, and 18.8 T, as well as triple-quantum (3Q)MAS NMR spectroscopy at 9.4 T, to achieve better understanding of the extent of disorder around this cation. We quantify the extent of this disorder in terms of changes in Na-O distance (d[Na-O]) distributions with composition and present a structural model favoring a somewhat ordered Na distribution, called a “perturbed” Na distribution model. The peak position in ²³Na MAS spectra of aluminosilicate glasses moves toward lower frequencies with increasing Si/Al ratios, implying that the average d(Na-O) increases with increasing R. The peak width is significantly reduced at higher fields (14.1 and 18.8 T) because of the reduced effect of second-order quadrupolar interaction, and ²³Na MAS NMR spectra thus provide relatively directly the Na chemical shift distribution and changes in atomic environment with composition. Chemical shift distributions obtained from ²³Na 3Q MAS spectra are consistent with MAS NMR data, in which deshielding decreases with R. The average distances between Na and the three types of bridging oxygens (BOs) (Na-{Al-O-Al}, Na-{Si-O-Al}, and Na-{Si-O-Si}) were obtained from the correlation between d(Na-O) and isotropic chemical shift. The calculated d(Na-{Al-O-Al}) of 2.52 Å is shorter than the d(Na-{Si-O-Si}) of 2.81 Å, and d(Na-{Al-O-Al}) shows a much narrower distribution than the other types of BOs. ²³Na chemical shifts in binary (Al-free) sodium silicate glasses are more deshielded and have ranges distinct from those of aluminosilicate glasses, implying that d(Na-NBO) (nonbridging oxygen) is shorter than d(Na-BO) and that d(Na-{Si-O-Si}) in binary silicates can be shorter than that in aluminosilicate glasses. The results given here demonstrate that high-field ²³Na NMR is an effective probe of the Na⁺ environment, providing not only average structural information but also chemically and topologically distinct chemical shift ranges (distributions) and their variation with composition and their effects on static and dynamic properties.     

Temperature effects on non-bridging oxygen and aluminum coordination number in calcium aluminosilicate glasses and melts

Configurational changes with temperature are important for the thermodynamic and transport properties of most aluminosilicate melts, but in general are not well understood. Here, we present high-resolution ²⁷Al and ¹⁷O NMR data on several calcium aluminosilicate glasses prepared with varying quench rates and thus with fictive temperatures that span ranges up to about 200 K. In all compositions the content of five-coordinated aluminum increases with fictive temperature, in agreement with recent high temperature NMR data on melts. In a glass of CaAl₂Si₂O₈ (“anorthite”) composition, the content of non-bridging oxygens also increases with temperature; however this effect was not observed in a sample with a much higher CaO/Al₂O₃ ratio. We present a consistent notation for reactions among structural species in these systems that clarify why in some cases, high-coordinated network cations may appear on the same side of the reaction, while in others they occur on the opposite sides: the key difference is in accounting for all coordination changes for oxygens. Mixing of non-bridging oxygens and of high-coordinated aluminum make significant contributions to the overall configurational entropy and heat capacity of the melts, as does the mixing of various bridging oxygens and of tetrahedral network cations. Other, less well known, types of increase in disorder with temperature may be important as well.     

Structure vs. composition: A solid-state H and Si NMR study of quenched glasses along the Na2O-SiO2-H2O join

A suite of six hydrous (7 wt.% H₂O) sodium silicate glasses spanning sodium octasilicate to sodium disilicate in composition were analyzed using ²⁹Si single pulse (SP) magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy, ¹H-²⁹Si cross polarization (CP) MAS NMR, and fast MAS ¹H-NMR. From the ²⁹Si SPMAS data it is observed that at low sodium compositions dissolved water significantly depolymerizes the silicate network. At higher sodium contents, however, dissolved H₂O does not affect a significant increase in depolymerization over that predicted based on the Na/Si ratio alone. The fast MAS ¹H-NMR data reveal considerable complexity in proton environments in each of the glasses studied. The fast MAS ¹H-NMR spectra of the highest sodium concentration glasses do not exhibit evidence of signficantly greater fractions of dissolved water as molecular H₂O than the lower sodium concentration glasses requiring that the decrease in polymerization at high sodium contents involves a change in sodium solution mechanism. Variable contact time ¹H-²⁹Si cross polarization (CP) MAS NMR data reveal an increase in the rotating frame spin lattice relaxation rate constant (T_1ρ*) for various Qⁿ species with increasing sodium content that correlates with a reduction in the average ¹H-²⁹Si coupling strength. At the highest sodium concentration, however, T_1ρ* drops significantly, consistent with a change in the Na₂O solution mechanism.     

Disorder and the extent of polymerization in calcium silicate and aluminosilicate glasses: O-17 NMR results and quantum chemical molecular orbital calculations

Estimation of the framework connectivity and the atomic structure of depolymerized silicate melts and glasses (NBO/T > 0) remains a difficult question in high-temperature geochemistry relevant to magmatic processes and glass science. Here, we explore the extent of disorder and the nature of polymerization in binary Ca-silicate and ternary Ca-aluminosilicate glasses with varying NBO/T (from 0 to 2.67) using O-17 NMR at two different magnetic fields of 9.4 and 14.1 T in conjunction with quantum chemical calculations. Non-random distributions among framework cations (Si and Al) are demonstrated in the variation of relative populations of oxygen sites with NBO/T. The proportion of non-bridging oxygen (NBO, Ca-O-Si) in the binary and ternary aluminosilicate glasses increases with NBO/T. While the trend is consistent with predictions from composition, the detailed fractions apparently deviate from the predicted values, suggesting further complications in the nature of polymerization. The proportion of each bridging oxygen in the glasses also varies with NBO/T. The fractions of Al-O-Si and Al-O-Al increase with increasing polymerization as CaO is replaced with Al₂O₃, while that of Si-O-Si seems to decrease, implying that activity of silica may decrease from calcium silicate to polymerized aluminosilicates (X_SiO2=constant). Quantum chemical molecular orbital calculations based on density functional theory show that a silicate chain with Al-NBO (Ca-O-Al) has an energy penalty (calculated cluster energy difference) of about 108 kJ/mol compared with the cluster with Ca-O-Si, consistent with preferential depolymerization of Si-networks, reported in an earlier O-17 NMR study [Allwardt, J., Lee, S.K., Stebbins, J.F., 2003. Bonding preferences of non-bridging oxygens in calcium aluminosilicate glass: Evidence from O-17 MAS and 3QMAS NMR on calcium aluminate glass. Am. Mineral.88, 949-954]. These prominent types of non-randomness in the distributions suggest significant chemical order in silicate glasses that leads to a decrease in silica activity coefficient and will be useful in modeling transport properties of melts.     

Insight into sodium silicate glass structural organization by multinuclear NMR combined with first-principles calculations

Short and medium range order of silica and sodium silicate glasses have been investigated from a quantitative analysis of ²⁹Si MAS NMR and ²³Na, ¹⁷O MQMAS NMR spectra. The method described enables the extraction of the underlying ¹⁷O NMR parameter distributions of bridging oxygens (BOs) and non-bridging oxygens (NBOs), and yields site populations which are confirmed by ²⁹Si NMR data. The extracted NMR parameter distributions and their variations with respect to the glass chemical composition can then be analyzed in terms of local structural features (bond angles and bond lengths, coordination numbers) with the help of molecular dynamics simulations combined with first-principles calculations of NMR parameters. Correlations of relevant structural parameters with ²³Na, ²⁹Si and ¹⁷O NMR interactions (isotropic chemical shift δ_iso, quadrupolar coupling constant C_Q and quadrupolar asymmetry parameter η_Q) are re-examined and their applicability is discussed. These data offer better insights into the structural organization of the glass network, including both chemical and topological disorder. Adding sodium to pure silica significantly diminishes the Si-O-Si bond angles and leads to a longer mean Si-O bond length with a slight decrease of the mean Na-O bond length. Moreover, the present data are in favor of a homogeneous distribution of Na around both oxygen species in the silicate network. Finally, our approach was found to be sensitive enough to investigate the effect of addition of a small quantity of molybdenum oxide (about 1 mol%) on the ¹⁷O MAS spectrum, opening new possibilities for investigating the Mo environment in silicate glasses.     

The speciation of water in silicate melts

Previous models of water solubility in silicate melts generally assume essentially complete reaction of water molecules to hydroxyl groups. In this paper a new model is proposed that is based on the hypothesis that the observed concentrations of molecular water and hydroxyl groups in hydrous silicate glasses reflect those of the melts from which they were quenched. The new model relates the proportions of molecular water and hydroxyl groups in melts via the following reaction describing the homogeneous equilibrium between melt species: H₂O_molecular (melt) + oxygen (melt) = 2OH (melt). An equilibrium constant has been formulated for this reaction and species are assumed to mix ideally. Given an equilibrium constant for this reaction of 0.1–0.3, the proposed model can account for variations in the concentrations of molecular water and hydroxyl groups in melts as functions of the total dissolved water content that are similar to those observed in glasses. The solubility of molecular water in melt is described by the following reaction: H₂O (vapor) = H₂O_molecular (melt).These reactions describing the homogeneous and heterogeneous equilibria of hydrous silicate melts can account for the following observations: the linearity between f_H2O and the square of the mole fraction of dissolved water at low total water contents and deviations from linearity at high total water contents; the difference between the partial molar volume of water in melts at low total water contents and at high total water contents; the similarity between water contents of vapor-saturated melts of significantly different compositions at high pressures versus the dependence on melt composition of water solubility in silicate melts at low pressures; and the variations of viscosity, electrical conductivity, the diffusivity of “water,” the diffusivity of cesium, and phase relationships with the total dissolved water contents of melts.This model is thus consistent with available observations on hydrous melt systems and available data on the species concentrations of hydrous glasses and is easily tested, since measurements of the concentrations of molecular water and hydroxyl groups in silicate glasses quenched from melts equilibrated over a range of conditions and total dissolved water contents are readily obtainable.     

Iron-bearing silicate melts: Relations between pressure and redox equilibria

Mossbauer spectroscopy has been used to determine the redox equilibria of iron and structure of quenched melts on the composition join Na₂Si₂O₅-Fe₂O₃ to 40 kbar pressure at 1400° C. The Fe³⁺/ΣFe decreases with increasing pressure. The ferric iron appears to undergo a gradual coordination transformation from a network-former at 1 bar to a network-modifier at higher (≧10 kbar) pressure. Ferrous iron is a network-modifier in all quenched melts. Reduction of Fe³⁺ to Fe²⁺ and coordination transformation of remaining Fe³⁺ result in depolymerization of the silicate melts (the ratio of nonbridging oxygens per tetrahedral cations, NBO/T, increases). It is suggested that this pressure-induced depolymerization of iron-bearing silicate liquids results in increasing NBO/T of the liquidus minerals. Furthermore, this depolymerization results in a more rapid pressure-induced decrease in viscosity and activation energy of viscous flow of iron-bearing silicate melts than would be expected for iron-free silicate melts with similar NBO/T.     

Evidence from olivine/melt element partitioning that nonbridging oxygen in silicate melts are not equivalent

Partitioning of Ca, Mn, Mg, and Fe²⁺ between olivine and melt has been used to examine the influence of energetically nonequivalent nonbridging oxygen in silicate melts. Partitioning experiments were conducted at ambient pressure in air and 1400°C with melts in equilibrium with forsterite-rich olivine (Fo >95 mol%). The main compositional variables of the melts were NBO/T and Na/(Na+Ca). In all melts, the main structural units were of Q⁴, Q³, and Q² type with nonbridging oxygen, therefore, in the Q³ and Q² units.For melts with high Q³/Q²-abundance ratio (corresponding to NBO/T near 1), increasing Na/(Na+Ca) [and Na/(Na+Ca+Mn+Mg+Fe²⁺)] results in a systematic decrease of the partition coefficients, K_Ca^ol/melt, K_Mn^ol/melt, K_Mg^ol/melt, and K_Fe2+^ol/melt, because of ordering of the network-modifying Ca, Mn, Mg, and Fe²⁺ among nonbridging oxygen in Q³ and Q² structural units. This decrease is more pronounced the smaller the ionic radius of the cation. With decreasing Q³/Q² abundance ratio (less-polymerized melts) this effect becomes less pronounced.Activity-composition relations among network-modifying cations in silicate melts are, therefore, governed by availability of energetically nonequivalent nonbridging oxygen in individual Qⁿ-species in the melt. As a result, any composition change that enhances abundance of highly depolymerized Qⁿ-species will cause partition coefficients to decrease.     

Vibrational interactions of tetrahedra in silicate glasses and crystals

Normal coordinate calculations have been carried out on partially polymerized simple silicate crystals, including Li and Na di- and metasilicates, Li and Gd pyrosilicates, thortveitite and rankinite. In the antisymmetric Si-O stretching modes which are active at 800–1200 cm^?1 in infrared spectra, Si-Obr vibrations occur at higher frequencies than Si-Onb vibrations if the bonds have equivalent strengths. However, this relationship is usually reversed when bridging oxygens are overbonded and non-bridging oxygens are underbonded in terms of Pauling bond strengths, a situation which is generally more common in crystals. An observed bimodality of the high-frequency envelope in infrared spectra of glasses in the alkali oxide-silica systems may be somewhat fortuitous, with the high frequency component (ca. 1100 cm^?1) representing underbonded non-bridging oxygens and saturated bridging oxygens, and the lower-frequency component (ca. 1000 cm^?1) mainly oversaturated bridging oxygens. Significant differences between crystals and glasses in the number and location of the main high-frequency infrared peaks suggest that there are short-range bonding rearrangements in the glasses, and that crystallite models are not applicable. Mid-frequency (600–800 cm^?1) infrared modes in silicates more polymerized than the pyrosilicate (Si₂O₇) appear to be mostly antisymmetric modes in which Si rattles against bridging oxygens, rather than symmetric stretching modes.     

Structure of silicate melts: Raman spectroscopic data and thermodynamic simulation results

High-temperature Raman spectroscopy was applied to study model silicate melts in the M₂O-SiO₂ system, where M = K, Na, or Li. Structural units of the melts and equilibrium constants of reactions between them are determined. Thermodynamic calculations were conducted for the dependence of the Q ⁿ distribution on the composition and temperature, and the results of thermodynamic simulations were demonstrated to be consistent with the experimental results.