Search

• Na/Mg mixing in aluminosilicates causes non-linear variations in their properties. • Entropy is not modelled by an ideal mixing law but by a sub-regular solution model. • Na and Mg are preferably arranged in sub-networks, reducing structural disorder. • Preferential role-sharing as network modifier and charge compensator between Na/Mg. • Extending sub-regular model and role-sharing to all alkaline/alkaline-earth mixing. This study documents the mixed modifier effect occurring between Na and Mg in aluminosilicate glasses. Rheological, thermodynamic, and structural data were acquired on compositions with 37.5 mol% Na 2 O+MgO, 12.5 mol% Al 2 O 3 and 50 mol% SiO 2. When Na substitutes Mg, melt viscosity decreases non-linearly, as does glass configurational entropy, showing large deviations from the ideal mixing law. This suggests ordering in the glass atomic structure upon Na/Mg mixing, as supported by structural data. Indeed, as Na 2 O/(MgO+Na 2 O) increases, Raman spectra reveal a shift of the 2Q3= Q2+Q4 equilibrium to the left-hand side, while 27Al MAS NMR spectra reveal that the Al coordination tends to 100 % Al[IV]. 23Na MAS NMR data further indicate that the network modifier/charge compensator role of metal cations also changes upon Na/Mg mixing. Those results therefore suggest important changes in melt/glass structure as alkali cations substitute calc-alkaline ones, with important rheological and thermodynamical impacts. [ABSTRACT FROM AUTHOR]

New Raman and NMR spectroscopy data on hydrous Ca aluminosilicate melts and glasses, with eutectic quartz-anorthite-wollastonite composition, are presented here. The glasses were obtained by rapid quench of melts equilibrated at high P and high T in a piston-cylinder apparatus. In situ Raman observations of the structure of the melts were also performed during hydrothermal diamond cell experiments. Using the intensities of the ~860 cm-1 and ~1630 cm-1 Raman signals, respectively assigned to vibrations of T-OH and H2Omol species, we determined the speciation of water in the glasses. T-OH and H2Omol values compare well with those determined from infrared (IR) spectra, except above ~5 wt% total water where IR determinations actually underestimate the proportion of hydroxyl groups. The analysis of the polarized Raman spectra and of the 29Si MAS NMR spectra of the hydrous glasses suggests limited changes in glass polymerization with variations in dissolved water content. However, at high temperatures, in situ Raman spectroscopy observations indicate that the hydrous melt structure differs very strongly from that of a glass containing a comparable concentration of dissolved water. Because of this, this study reinforces the fact that using glass data to try understanding high temperature processes in hydrous melts, like viscous flow or water diffusion toward bubbles during volcanic degassing, may not be very appropriate. [ABSTRACT FROM AUTHOR]

Aluminosilicate glasses and melts are of paramount importance for geo- and materials sciences. They include most magmas, and are used to produce a wide variety of everyday materials, from windows to smartphone displays. Despite this importance, no general model exists with which to predict the atomic structure, thermodynamic and viscous properties of aluminosilicate melts. To address this, we introduce a deep learning framework, 'i-Melt', which combines a deep artificial neural network with thermodynamic equations. It is trained to predict 18 different latent and observed properties of melts and glasses in the K 2 O-Na 2 O-Al 2 O 3 -SiO 2 system, including configurational entropy, viscosity, optical refractive index, density, and Raman signals. Viscosity can be predicted in the 100–1015 log 10 Pa·s range using five different theoretical frameworks (Adam-Gibbs, Free Volume, MYEGA, VFT, Avramov-Milchev), with a precision equal to, or better than, 0.4 log 10 Pa·s on unseen data. Density and optical refractive index (through the Sellmeier equation) can be predicted with errors equal or lower than 0.02 and 0.006, respectively. Raman spectra for K 2 O-Na 2 O-Al 2 O 3 -SiO 2 glasses are also predicted, with a relatively high mean error of ∼25% due to the limited data set available for training. Latent variables can also be predicted with good precisions. For example, the glass transition temperature, T g , can be predicted to within 19 K, while the melt configurational entropy at the glass transition, Sconf(T g) , can be predicted to within 0.8 J mol−1 K−1. Applied to rhyolite compositions, i-Melt shows that the rheological threshold separating explosive and effusive eruptions correlates with an increase in the fraction of non-bridging oxygens in rhyolite melts as their alkali/Al ratio becomes larger than 1. Exploring further the effect of the K/(K + Na) ratio on the properties of alkali aluminosilicate melts with compositions varying along a simplified alkali magmatic series trend, we observe that K-rich melts have systematically different structures and higher viscosities compared to Na-rich melts. Combined with the effects of the K/(K + Na) ratio on other parameters, such as the solubility, solution mechanisms and speciation of volatile elements, this could ultimately influence the eruptive dynamics of volcanic systems emitting Na-rich or K-rich alkali magmas. [ABSTRACT FROM AUTHOR]

Iron oxidation state and environment in magmas affect their phase diagram and their properties, including viscosity and density, which determine magma mobility and eruptive potential. In turn, magma composition, pressure, temperature and oxygen fugacity affect iron oxidation state and coordination, potentially leading to complex feedbacks associated with magma ascent, degassing and eruption. While equilibrium experiments and models have led to a deep understanding of the role of iron in melts, our knowledge of the effects of disequilibrium processes on iron oxidation state and its structural role in lavas and magmas remains limited. Accordingly, we performed a series of dynamic disequilibrium experiments on a natural melt composition (a phonolite lava from Erebus volcano, Antarctica) at atmospheric pressure, in which oxygen fugacity and temperature were controlled and varied. During the experiments, we continuously measured iron oxidation and coordination using Fe K-edge dispersive X-ray Absorption Spectroscopy (XAS). We found that iron oxidation state changes in the phonolite melt are reversible and well reproduced by existing models. Changes in iron oxidation state are driven by joint diffusion of alkali cations and oxygen anions at magmatic temperatures (~ 1000 °C for Erebus phonolite). However, redox diffusion timescales are too slow for any significant oxygen exchange with the atmosphere at the lava/air interface or via air entrainment. Turning to iron coordination, while Fe2+ and Fe3+ are present mostly in an average five-fold coordination, complex coordination variations decoupled from redox changes were detected. The data suggest transitions between Fe3+ in four-fold and six-fold coordination prior to reduction or as a consequence of oxidation. This questions the possible implication of Fe coordination changes in triggering crystallisation of magnetite nanolites upon magma ascent, and, through such crystallisation events, in promoting magma explosivity. [ABSTRACT FROM AUTHOR]

The first version of the machine learning greybox model i-Melt was trained to predict latent and observed properties of K 2 O-Na 2 O-Al 2 O 3 -SiO 2 melts and glasses. Here, we extend the model compositional range, which now allows accurate predictions of properties for glass-forming melts in the CaO-MgO-K 2 O-Na 2 O-Al 2 O 3 -SiO 2 system, including melt viscosity (accuracy equal or better than 0.4 log 10 Pa ⋅ s in the 10−1-1015 log 10 Pa ⋅ s range), configurational entropy at glass transition (≤ 1 J mol−1 K−1), liquidus (≤ 60 K) and glass transition (≤ 12 K) temperatures, heat capacity (≤ 3 %) as well as glass density (≤ 0.02 g cm−3), optical refractive index (≤ 0.006), Abbe number (≤ 4), elastic modulus (≤ 6 GPa), coefficient of thermal expansion (≤ 1.1 10−6 K−1) and Raman spectra (≤ 25 %). Uncertainties on predictions also are now provided. The model offers new possibilities to explore how melt/glass properties change with composition and atomic structure. [Display omitted] • i-Melt predicts various properties for calc-alkaline and alkaline aluminosilicates. • Uncertainties can be estimated via MC Dropout and conformal predictions. • i-Melt predicts cationic partial molar volumes V m and liquid C p l i q u i d • The partial molar volumes and C p l i q u i d of cations change with composition. • Al coordination number correlates with the glass molar volume but not with C p l i q u i d [ABSTRACT FROM AUTHOR]

Following an unprecedented seismic activity that started in May 2018, a new volcanic edifice, now called Fani Maoré, was constructed on the ocean floor 50 km east of the island of Mayotte (Indian Ocean). This volcano is the latest addition to a volcanic chain characterized by an alkaline basanite-to-phonolite magmatic differentiation trend. Here, we performed viscosity measurements on five silicate melts representative of the East-Mayotte Volcanic Chain compositional trend: two basanites from Fani Maoré, one tephriphonolite and two phonolites from different parts of the volcanic chain. A concentric cylinder viscometer was employed at super-liquidus conditions between 1500 K and 1855 K and a creep apparatus was used for measuring the viscosity of the undercooled melts close to the glass transition temperature in the air. At super-liquidus temperatures, basanites have the lowest viscosity (0.11–0.34 to 0.99–1.16 log 10 Pa⸱s), phonolites the highest (1.75–1.91 to 3.10–3.89 log 10 Pa⸱s), while the viscosity of the tephriphonolite falls in between (0.89–1.97 log 10 Pa⸱s). Near the glass transition, viscosity measurements were performed for one phonolite melt because obtaining pure glass samples for the basanite and tephriphonolite compositions was unsuccessful. This is due to the formation of nanolites upon quench as evidenced by Raman spectroscopy. The phonolite viscosity ranges from of 10.19 log 10 Pa⸱s at 1058 K to 12.30 log 10 Pa⸱s at 986 K. Comparison with existing empirical models reveals an underestimation of 1.2 to 2.0 log units at super-liquidus and undercooled temperatures, respectively, for the phonolite. This emphasizes (i) the lack of data falling along the alkaline basanite-to-phonolite magmatic differentiation trend to calibrate empirical models, and (ii) the complexity of modeling viscosity variations as a function of temperature and chemical composition for alkaline compositions. The new measurements indicate that, at eruptive temperatures between 1050 °C and 1150 °C (1323–1423 K), the oxidized, anhydrous, crystal-free and bubble-free basanite melt have a viscosity around 2.6 log 10 Pa⸱s. In contrast, the anhydrous phonolite crystal- and bubble-free melt would have a viscosity around 6–10 log 10 Pa⸱s at expected eruptive temperatures, from 800 to 1000 °C (1073–1273 K). Considering that both basanite and phonolite lavas from the Mayotte submarine volcanic chain contain <6% crystals and a significant amount of water (1-2.3 wt% and 0.8-1.2 wt%, respectively), such viscosity values are probably upper limits. The new viscosity measurements are essential to define eruptive models and to better understand the storage and transport dynamics of Comoros Archipelago magmas, and of alkaline magmas in general, from the source to the surface. • Pure liquid viscosity measurements are performed on submarine melts from the basanite - phonolite trend. • Super-liquidus viscosity (log 10 Pas) is 0.11-1.16 for basanite, 0.89 –1.97 for tephriphonolite and 1.75-3.89 for phonolite. • A discrepancy is revealed by comparing experimental measurements and parametric viscosity models. • The effect of water at the eruptive temperatures for such compositions is explored. [ABSTRACT FROM AUTHOR]

The bonding and speciation of water dissolved in Na silicate and Na and Ca aluminosilicate melts were inferred from in situ Raman spectroscopy of the samples, in hydrothermal diamond anvil cells, while at crustal temperature and pressure conditions. Raman data were also acquired on Na silicate and Na and Ca aluminosilicate glasses, quenched from hydrous melts equilibrated at high temperature and pressure in a piston cylinder apparatus. In the hydrous melts, temperature strongly influences O-H stretching ν(O-H) signals, reflecting its control on the bonding of protons between different molecular complexes. Pressure and melt composition effects are much smaller and difficult to discriminate with the present data. However, the chemical composition of the melt + fluid system influences the differences between the ν(O-H) signals from the melts and the fluids and, hence, between their hydrogen partition functions. Quenching modifies the O-H stretching signals: strong hydrogen bonds form in the glasses below the glass transition temperature T g, and this phenomenon depends on glass composition. Therefore, glasses do not necessarily record the O-H stretching signal shape in melts near T g. The melt hydrogen partition function thus cannot be assessed with certainty using O-H stretching vibration data from glasses. From the present results, the ratio of the hydrogen partition functions of hydrous silicate melts and aqueous fluids mostly depends on temperature and the bulk melt + fluid system chemical composition. This implies that the fractionation of hydrogen isotopes between magmas and aqueous fluids in water-saturated magmatic systems with differences in temperature and bulk chemical composition will be different. [ABSTRACT FROM AUTHOR]

The Adam and Gibbs theory depicts the viscous flow of silicate melts as governed by the cooperative re-arrangement of molecular sub-systems. Considering that such subsystems involve the silicate Q n units ( n = number of bridging oxygens), this study presents a model that links the Q n unit fractions to the melt configurational entropy at the glass transition temperature T g , S conf (T g ) , and finally, to its viscosity η . With 13 adjustable parameters, the model reproduces η and T g of melts in the Na 2 O-K 2 O-SiO 2 system (60 ≤ [SiO 2 ] ≤ 100 mol%) with 1σ standard deviations of 0.18 log unit and 10.6°, respectively. The model helps understanding the links between the melt chemical composition, structure, S conf and η . For instance, small compositional changes in highly polymerized melts generate important changes in their S conf (T g ) because of an excess of entropy generated by mixing Si between Q 4 and Q 3 units. Changing the melt silica concentration affects the Q n unit distribution, this resulting in non-linear changes in the topological contribution to S conf (T g ) . The model also indicates that, at [SiO 2 ] ≥ 60 mol%, the mixed alkali effect has negligible impact on the silicate glass Q n unit distribution, as corroborated by Raman spectroscopy data on mixed Na-K tri- and tetrasilicate glasses. Such model may be critical to link the melt structure to its physical and thermodynamic properties, but its refinement requires further high-quality quantitative structural data on silicate and aluminosilicate melts. [ABSTRACT FROM AUTHOR]

Brillouin spectroscopy is used to investigate the elastic properties of XAlSiO 4 aluminosilicate glasses where X = Li, Na, K, Mg 0.5 , Ca 0.5 , Sr 0.5 , Ba 0.5 , and Zn 0.5 . The Brillouin frequency shifts obtained in two different scattering geometries allow the calculation of the refractive index, the two sound velocities and Poisson's ratio. Measurements of the mass density give in turn the elastic moduli and the Debye temperature. We find that the elastic properties scale with the atomic density of the glassy network or the charge-balancing cation field strength while they negatively correlate with the glass transition temperature. Further, Poisson's ratio depends on the nature of the non-framework cations in this glass series. [ABSTRACT FROM AUTHOR]

The solubility and solution behavior of F and Cl in peralkaline aluminosilicate compositions in the systems Na2O-Al2O3-SiO2 and K2O-Al2O3-SiO2 have been determined for glasses quenched from melts equilibrated at 1400 and 1600 °C in the 1.0-2.5 GPa pressure range. With Al/(Al+Si) increasing from 0 to 0.33 in sodium aluminosilicate melts, F solubility (saturation concentration) increases from 3.3 to 7.4 mol%, whereas Cl solubility decreases from 5.7 to 2.5 mol%. There is no difference in F solubility in sodium or potassium aluminosilicate melts. However, the Cl solubility in potassic aluminosilicate melts is 40-60% lower than in sodic melts with the same Al/(Al+Si) and Na or K mole fraction. Fluorine depolymerizes the silicate melt structure and forms Si-F, Al-F, and Na-F complexes, the proportion of which depends on the melt Al/(Al+Si) ratio. Dissolution of Cl results in a small degree of depolymerization of Al-free silicate melt, whereas Cl has a polymerizing effect in aluminosilicate melts. In both cases, formation of Na-Cl complexes appears to be the driving solution mechanism. The differences in F and Cl solution mechanisms result in contrasting depolymerizing effects that become more pronounced with increasing degree of magmatic differentiation. Through such modifications of melt structure, F and Cl have significant effects on magma properties such as viscosity, compressibility, and element partitioning. [ABSTRACT FROM AUTHOR]

The hydrogen isotopic composition of melt inclusions trapped in phenocrysts during their crystallization and growth in a magma may contribute to a better understanding of the water cycle between the atmosphere, the hydrosphere and the lithosphere. Such understanding relies on the knowledge of the hydrogen isotopic fractionation factors between aqueous fluids, silicate melts, and minerals at temperature and pressure conditions relevant to the Earth's interior. Significant D/H fractionation between silicate melts and aqueous fluids was reported at hundreds of MPa and °C by using in situ measurements in hydrothermal diamond anvil cell (HDAC) experiments ( Mysen, 2013a, 2013b , Am. Mineral. 98, 376–386 and 1754–1764). However, the available dataset is focused on fluids and melts with D/H ratios close to unity. The relevance of such data for natural processes that involve per mil variations of δ D-values may not always be clear. To address such concerns, the effect of the bulk D/H ratio on hydrogen isotope partitioning between water-saturated silicate melts and coexisting silicate-saturated aqueous fluids has been determined in the Na 2 O–Al 2 O 3 –SiO 2 –H 2 O–D 2 O system. To this end, in situ Raman spectroscopic measurements were performed on fluids and melts with bulk D/H ratios from 0.05 to 2.67 by using an externally-heated diamond anvil cell in the 300–800 °C and 200–1500 MPa temperature and pressure range, respectively. In these pressure/temperature ranges, the D/H ratios of fluids in equilibrium with melt barely change with temperature (in average Δ H fluid = 0.47 ± 1.15 kJ / mol ). In contrast, the D/H ratios of coexisting melts display strong dependence on temperature (average Δ H melt = 7.18 ± 1.27 kJ / mol ). The temperature-dependence of the D/H fractionation factor between melt and fluid ( α fluid - melt = D / H fluid / D / H melt ) is comparable in all the experiments and can be written: 1000 ⋅ ln ⁡ ( α fluid - melt ) = 263 ( ± 26 ) ⋅ T − 2 – 126 ( ± 48 ) . Therefore, the α fluid-melt is independent of the bulk D/H ratio of the melt + fluid system. Experimentally determined α fluid-melt using D-enriched fluids, therefore, can be applied to natural systems. It follows that for water-saturated magma strong isotopic fractionation of D and H between water dissolved in magmas and deep aqueous fluids may occur. The δ D-values in melt inclusions in phenocrysts in such water-saturated magma will reflect such fluid/melt fractionation effects. A likely result is underestimation of the δ D isotopic composition of slab fluids based on δ D-values in melt inclusions. The temperature-dependent hydrogen isotope fractionation must be taken into account in the modeling of slab fluid–magma interaction in the mantle wedge. [ABSTRACT FROM AUTHOR]

• Only Ti4+ was found at high temperature in silicate melts at f O 2 between 0.21 and 10−6. • Ti-bearing melts systematically crystallize above 1500 °C under very reducing atmospheres. • Crystallisation might be preferred to reduction. Ti is usually tetravalent in silicate glasses. It may be trivalent under very reduced conditions, but this remains unclear. In this study, in situ Ti K-edge XANES experiments were performed on four different Ti-bearing melts, equilibrated at high temperatures using a micro-furnace with controlled atmosphere. No formation of Ti3+ was observed, even at fO 2 ∼ 10−15 (estimated). Instead, systematic crystallisation of the melts was observed. In experiments at fO 2 ≤ 10−12, the system favours Ti4+ crystallisation instead of reduction to Ti3+ in the studied melts. Fluxing melts with reduced gas mixtures, like Ar-2%H 2 , may be a novel way to induce surface crystallisation and produce ceramic-coated glass materials. [ABSTRACT FROM AUTHOR]

Previous studies of hydrous glasses and melts with infrared spectroscopy have led to the conclusion that the IR combination peaks near 4500 and 5200 cm-1 reflect the existence of OH- (hydroxyl) groups and H2Omol water molecules in those materials. Here, we show that the glass chemical composition can impact profoundly the intensities and frequencies of the fundamental O-H stretching signal and, therefore, potentially those of the 4500 and 5200 cm-1 combination peaks. In alkali silicate glasses, compositional effects can give rise to peaks assigned to fundamental O-H stretching at frequencies as low as 2300 cm-1. This expanded range of Raman intensity assigned to O-H stretch is increasingly important as the ionic radius of the alkali metal increases. As a result, the combination of the fundamental O-H stretch in OH- groups with the Si-O-H stretch located near 910 cm-1 gives rise to a complex combination signal that can extend to frequencies much lower than 4200 cm-1. This combination signal then becomes unresolvable from the high-frequency limb of the band assigned to fundamental O-H stretch vibration in the infrared spectra. It follows that, when O-H stretch signals from OH- groups extend to below 3000 cm-1, the 4500 cm-1 peak does not represent the total OH- signal. Under such circumstances, this infrared peak may not be a good proxy for determining the concentration of OH- hydroxyl groups for glassy silicate materials. [ABSTRACT FROM AUTHOR]

Long-lived active lava lakes are comparatively rare and are typically associated with low-viscosity basaltic magmas. Erebus volcano, Antarctica, is unique today in hosting a phonolitic lava lake. Phonolitic magmas can erupt explosively, as in the 79 CE Plinian eruption of Vesuvius volcano, Italy, and it is therefore important to understand their physical properties. The phonolite at Erebus has slightly higher silica content than that at Vesuvius yet its present activity is predominantly non-explosive. As a contribution to understanding such contrasting eruptive behaviour, we focus on the rheological differences between these comparable magmas. In particular, we evaluate the viscosity of the Erebus phonolite magma by integrating new experimental data within a theoretical and empirical framework. The resulting model enables estimation of the Erebus melt viscosity as a function of temperature, crystal and water concentrations, with an uncertainty of, at most, ± 0.45 log (Pa s). Using reported ranges for these parameters, we predict that the magma viscosity in the upper region of the plumbing system of Erebus ranges between 10 5 and 10 7 Pa s . This is substantially higher than has been hitherto considered with significant implications for modelling the dynamics of the lava lake, conduit and magma reservoir system. Our analysis highlights the generic challenges encountered in calculation of magma viscosity and presents an approach that can be applied to other cases. [ABSTRACT FROM AUTHOR]

Water can form different chemical bonds with the ionic entities composing silicate melts. Because of that, its influence on the physico-chemical properties of magmas can vary with silicate composition and water content, temperature, and pressure. To further our understanding of how silicate chemical composition governs proton distribution in magmas, the environment of protons in hydrous alkali (Li, Na, K) silicate glasses was varied as a function of the type of alkali metal and total water content. From ¹H MAS NMR spectroscopy, H+ are distributed among five different structural environments in alkali silicate glasses. One of these environments is in the form of H2O molecules (H2Omol). The four others are the proton environments associated with Si-OH bonding, and perhaps also with M-OH bonding (with M = Li, Na, or K). Those environments differ in their O-O distance and extent of hydrogen bonding. H2Omol species are located in an environment with an O-O distance of ~290 pm. OH- groups are in environments with O-O distances from 240 to 305 pm. The ionic radius of the alkalis, and hence their ionic field strength, determines the fraction of water dissolved as H2Omol and OH- groups, as well as the distribution of protons in the various OH- environments. The mean volume of the H+ oxygen coordination sphere was calculated using the ¹H+ NMR signal intensity and the mean O-O distance around H+. Increasing ionic radius of the alkali metal in silicate glasses results in a decrease of this mean volume. The partial molar volume of water in the corresponding melts determined through other technics seems to vary in a comparable way. Therefore, the chemical composition of silicate melts may control the partial molar volume of dissolved water because of its influence on the structural environment of protons. This probably also plays a role in determining water solubility. [ABSTRACT FROM AUTHOR]

Abstract: Dissociation of water into protons and hydroxyl ions is a fundamental feature of aqueous solutions. Although it exerts a profound influence on properties of magmas, this autoprotolysis reaction has been hitherto neglected for water dissolved in silicate melts. As made here with an acid–base model, in fact one has to deal with molecular water (H2Omol) and two kinds of hydroxyl groups, bonded or not to network-forming cations (OH and OH−, respectively) in hydrous silicate melts. By mixing cations and anions on distinct sublattices and quantifying the disproportionation of water dissolved in silicate melts into its ionic products, H+ and OH−, we reconcile spectroscopic determinations of water speciation, and highlight the main compositional features involving chemical exchanges between H2Omol, T OH and M OH groups (T and M being a former and a modifier, respectively). In particular, water addition to depolymerized systems, such as basalts, determine a relative predominance of OH− with respect to silicic systems, such that the increase of water concentration tends to immediately limit depolymerization rather than enhance it. This opens new perspectives to the understanding of the chemical control of hydrous magmas and their physical properties, as well as the attainment of saturation in hydrous minerals such as amphiboles or micas, particularly in depolymerized melts. [Copyright &y& Elsevier]

Abstract: Because of their importance in both the geosciences and the glass-making industry, alkali aluminosilicate melts have been the focal point of many past studies, but despite progress many problems remain unresolved, such as the complex behaviour of the thermodynamic properties of aluminium-rich alkali silicate melts. This paper presents a study of Na2O–Al2O3–SiO2 glasses and melts, containing 75mol% SiO2 and different Al/(Al+Na) ratios. Their structure has been investigated by using Raman spectroscopy, as well as, 23Na, 27Al and 29Si 1D MAS NMR spectroscopy. Results confirm the role change of Na+ cations from network modifier to charge compensator in the presence of Al3+ ions. In addition, polymerization increases with increase of the Al/(Al+Na) ratio. These structural changes explain the observed variations in the viscosity of these melts. The viscosity data in turn allow us to calculate the configurational entropy of melts at the glass transition temperature [the S conf(Tg)]. The variations of the S conf(Tg) are strongly nonlinear, with sharp increases and decreases depending on the Al/(Al+Na) ratio. More importantly, a strong increase of the S conf(Tg) is observed when a few Al2O3 is added to sodium silicate melt. A strong decrease is observed after crossing the tectosilicate join, when Al/(Al+Na)>0.5 and when Al3+ ions are present in fivefold coordination, Al[5], in the glass. Furthermore, in situ 27Al NMR spectra of the peraluminous melt show a clear increase of the Al[5] concentration with increasing temperature. When considered in combination with melt fragility and heat capacity, our data demonstrate that Al[5] is clearly a transient unit at high temperature in highly polymerized tectosilicate and peraluminous melts. However, when present in glasses, Al[5] increases the stability of the aluminosilicate network, hence the Tg of glasses. This could be explained by the ability of Al[5] to carry threefold coordinated oxygen atoms in its first coordination shell, as observed in minerals. Localisation of threefold coordinated oxygen atoms on Al[5] implies an increase of the medium-range order in the glass, an hypothesis that is in agreement with the low S conf(Tg) of peraluminous glasses. [Copyright &y& Elsevier]

Abstract: Viscosity and structure of Na/K tectosilicate glasses containing 75 and 83mol% of SiO2 have been investigated. Viscosity increases non-linearly when K+ ions substitute Na+ ions in these melts. The viscosity variations depending on chemical composition cannot be reproduced using an ideal mixing model of the configurational entropy. Consequently, it appears that Na and K elements do not mix randomly in the studied aluminosilicate melts. Raman spectra of glasses show that, during the Na/K substitution, evolutions of both the ring distribution and the T-O-T mean angle, or force constant, occur. The proportion of small-membered (three, four) rings, compared to large-membered rings, is higher in K-rich glasses than in Na-rich glasses. Moreover, Raman spectra features suggest that two different TO2 environments exist and cohabit into the glass. They could represent two populations of six-membered rings differing by their force constant or their T-O-T angles. One of these environments evolves when K substitutes for Na, showing an increase of its mean T-O-T angle and force constant. The other environment remains unchanged. From the observations, we propose that Na/K mixed tectosilicate glasses contained two sub-networks: one composed of the Si, Al, O and K atoms, and another of the Si, Al, O and Na. We suggest that the different size of alkali elements combined to the charge balancing needs of Al3+ ions can be the source of the different clustering of alkali cations into different sub-networks. Furthermore, and as previously inferred by older studies, potassic tectosilicate glasses could present silica-like and leucite-rich regions, explaining notably the incongruent crystallization of the orthoclase liquid. [Copyright &y& Elsevier]

Raman spectroscopy can measure water concentrations of hydrous silicate glasses with several advantages such as: (1) high-spatial resolution of 1-2 µm²; (2) non-destructive character; and (3) easy access, without any specific sample preparation or mounting techniques. The latter reasons render Ra- man highly suitable for studying natural products, such as volcanic pumice and scoriae fragments. Two spectral regions can be distinguished in Raman spectra of hydrated silicate glasses: a low-wavenumber -1 region (15-1500 cm-1), which corresponds to vibrations of the silicate network, and a high-wavenumber -1 region (3100-3750 cm ), corresponding to the OH stretching vibrations of H2O molecules and OH groups. Behrens et al. (2006) have published empirical equations relating the area ratio between these two regions and the water content. However, the proposed internal calibrations depend on chemical composition of the glasses. In this paper, we reinvestigated the previous procedures to improve the background subtraction. Our results allow us to present a more general and linear calibration. Water concentrations up to 13 wt% can be measured for a broad range of natural silicate melts, from basalts to rhyolite (40 up to 80 wt% SiO2), using a single calibration curve with an absolute error of 0.2 wt%. [ABSTRACT FROM AUTHOR]

• Experimental stick-slip on SiO 2 glass surfaces causes localised frictional melting. • Systematic changes in molecular structure of glass recorded following slip. • Structural changes caused by high pressures and fast cooling rates on asperities. • Thermomechanical history alters future strength and behaviour of interface. New insights into the pressure and temperature conditions on fault surfaces during seismic slip are provided by Raman-active vibrational modes of SiO 2 glass. We performed triaxial stick-slip experiments at room temperature and high normal stresses on pre-ground, high-purity silica glass surfaces. During slip, velocities exceed 0.32 m s−1 over durations of less than one millisecond, generating frictional heat and locally melting the fault surfaces. Temperature increases permit structural rearrangement within the melt; these changes are preserved by rapid quenching. Using Raman spectroscopy, we analyse melt-welded regions and show that these areas exhibit systematic changes in the spectra of silica. Changes result from a decrease in the inter-tetrahedral Si-O-Si bond angle and are correlated to increasing silica glass density in the slip regions. Densification results from both rapid cooling rates and exposure to very high pressures at asperity contacts. We use data from other experiments to calibrate these effects, estimating quench temperatures up to 1800 K and pressures of ∼180 MPa. These results provide the first quantitative evidence for the effects of quench rates and high inter-asperity pressures on the physics of melting and quenching during seismic slip and its impact on fault behaviour. [ABSTRACT FROM AUTHOR]

Hydroxylation is a method that allows "decoration" of the pre-existing point defect structure of nominally anhydrous minerals, such as olivine. We tested this method on synthetic forsterite (Fo: Mg 2 SiO 4 ) crystals. To control starting point defect structures, Fo crystals were pre-annealed at different temperatures ( 1100 - 1500 ∘ C ), silica activity conditions (forsterite–enstatite Fo–En and forsterite–periclase Fo–Per) and oxygen fugacity (0.21 and 10 - 6 bars). Then low-temperature hydroxylation (900 °C, 1.5 GPa) of the crystals successfully allowed the decoration with protons of pre-existing point defect structures, as subsequently revealed by infrared spectroscopy. Protons are arranged in three different point defect stoichiometries in Fo, related to Mg and Si vacancies ([Mg] and [Si], respectively) as well as to a trivalent cation-associated substitution mechanism ([triv]). Over the timescale and equilibrium conditions studied, hydroxylation does not reset the point defect structure inherited from pre-anneal. The data further show that the concentrations of [Mg]-, [Si]- and [triv]-hydrated defects are function of pre-anneal silica activity and temperature. Laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analysis of the crystals revealed diffusion of Al and Fe into the crystals during the pre-annealing, a phenomenon clearly promoted at high a SiO 2 . The data confirm a very fast mechanism of Al diffusion in Fo during pre-annealing, and suggest a strong coupling between H + and Al 3 + during hydroxylation. Overall, they show the strong importance of a SiO 2 and temperature in the incorporation of trace cations in forsterite, and the subsequent effects of incorporation of trace cations on Mg- and Si-related point defects in Fo. The dry point defect population of Fo is determined by interactions between the trace trivalent cations and dry Si and Mg vacancies. Without trace elements, T only has a limited effect on Mg- and Si-related point defect populations. Finally, approaching or potentially slightly exceeding the Fo–En solidus leads to strong changes in the trace element concentration and point defect population in Fo, which may be related to either partial melting or pre-melting effects. [ABSTRACT FROM AUTHOR]

We report the abundances of major and lithophile trace elements and volatiles (H2O, Cl and S) in orthopyroxenite veins cutting mantle-derived, spinel harzburgite xenoliths from the active Ritter volcano in the West Bismarck Arc (Papua New Guinea). The veins preserve sulfide-bearing glass coexisting with crystals. The glass formed by the quench of residual liquids left after crystallisation of abundant orthopyroxene, variable amounts of clinopyroxene, and minor olivine and spinel from silicate melts intruding spinel harzburgite. The glass compositions range from low-Ti (≤ 0.2 wt% TiO2), intermediate to magnesian (2-5 wt% MgO) and high-Mg# (0.45-0.6) andesite to dacite, with Mg# = Mg/(Mg + Fet) where Fet indicates all Fe treated as Fe2+. The glass is depleted in moderately incompatible lithophile trace elements and contains 1.2 to 2.5 wt% H2O, 400-1800 ppm Cl and 100-400 ppm S. Thermo-barometric calculations indicate that the original vein-forming melts intruded the shallow sub-arc mantle lithosphere between 0.5 and 1 GPa, where they partially crystallised (50-60%) to form residual liquids during cooling from ~ 1200 °C down to ~ 650 °C. Our data and calculations show that West Bimarck orthopyroxenite veins are of high-temperature magmatic origin and bear no relation to the carrier magmas of the xenoliths. Instead, petrological modelling demonstrates that the veins formed from the intrusion of primitive low-Ca boninite (LCB) melts in the sub-arc mantle lithosphere. These melts were originally produced by low degrees of melting (≤ 5%) of spinel harzburgite at ≥ 1360 °C and ≤ 1.5 GPa. Some key features of the parental LCB vein-forming melts from West Bismarck (e.g. lithophile trace element signatures and moderate volatile abundances and oxygen fugacity) suggest that, contrary to the widely held belief, their formation does not require persistent mantle wedge fluxing by slab-derived components. Instead, we propose that adiabatic decompression melting of spinel harzburgite residues may also form LCB in the sub-arc mantle. West Bismarck veins are very similar to those previously described in mantle xenoliths from the Kamchatka Arc; active boninite magmatism in these mature arcs shows that the generation of these melts is not as restricted as previously thought. In fact, the growing evidence for the prevalence of boninite in sub-arc mantle peridotites is consistent with the presence of boninite-like geochemical components in many island arcs. [ABSTRACT FROM AUTHOR]

CO 2 degassing of mafic silicate melts is an important part of the terrestrial carbon cycle, at mid-ocean ridges, oceanic hot spots, or in the middle of continents. Deeper CO 2 -bearing mafic magmas may also exist, such as those suspected in the D″ zone, and certainly existed in the magma ocean of the early Earth. Knowledge of the CO 2 solubility in mafic melts at high pressure is therefore important but unknown at present. Results from molecular dynamics simulation (e.g., Guillot and Sator, 2011) predict that CO 2 solubility in basalt may be much higher than previously thought at pressures and temperatures relevant to the upper mantle. But some recent models predict low solubility at high pressure (e.g., Eguchi and Dasgupta, 2018). The present study thus experimentally investigates the solubility of CO 2 in basalt and andesite in the pressure range 1.5–8.5 GPa at 1820–2130 K in oxidizing conditions. Up to 4 GPa, the quenched samples are essentially vitreous and CO 2 -saturated. Their CO 2 contents are measured using confocal micro-Raman spectroscopy, where we establish an internal calibration line relating CO 2 content to the area of the Raman band assigned to the ν 1 stretching vibration of carbonate groups. This calibration appears independent from the spectrometer, sample or experimentalist, thus enhancing confidence in this method. At 5 and 8.5 GPa, some of the quenched samples are found partially crystallized. Their CO 2 abundance is estimated at micro-scale from Raman mapping, and at large scale from image analysis and presence/absence of vesicles. Over the 1.5–8.5 GPa pressure range, obtained CO 2 concentrations vary between 1.8 and > 13.6 wt%. At 5 GPa and 1873 K, the CO 2 content in basalt and andesite are similar. These findings experimentally confirm the ability of mafic melts to accommodate large amounts of CO 2 at conditions prevailing in the deep Earth. A consequence is that magmas issued from partial melting of carbonate-bearing silicate rocks may ascend with significant quantities of CO 2 of several wt% and more: when reaching shallower depths, they may degas large quantities of CO 2. Present estimates of the global carbon flux to the atmosphere may thus be underestimated, and implications to early magma ocean degassing may be considered. Other consequences may concern the genesis of kimberlites and carbonatites. We finally speculate that, if silicate melts exist in the D″ zone, significant amounts of carbon may be stored there, and consequences may arise as to carbon sequestration in the core. [ABSTRACT FROM AUTHOR]

In this study, the effect of Ti4+ on the structure of nepheline glass (NaAlSiO 4) is investigated as SiO 2 is systematically replaced with TiO 2. Traditionally, TiO 2 is considered to be a nucleating agent for silicate crystallization but can also be incorporated into the glass network in relatively large amounts as either a network former or modifier depending on its coordination with oxygen. To determine the effect of Ti4+ on the structure of nepheline glass, X-ray and neutron pair distribution function (PDF) analysis paired with Empirical Potential Structure Refinement (EPSR) were conducted and are supplemented with Raman spectroscopy, electron probe microanalysis, and X-ray absorption spectroscopy (including Extended X-ray Absorption Fine Structure, EXAFS). Through these methods, it has been found that up to 15 mol% (16 wt%) TiO 2 can incorporate into the glass network as a four-fold coordinated species, with a minor contribution of higher coordinated Ti. Between NaAlTi 0.1 Si 0.9 O 4 and NaAlTi 0.2 Si 0.8 O 4 , EXAFS suggests a local structure change in the second coordination sphere of Ti, which changes from Ti-Ti to Ti-Al. Raman spectroscopy also suggests that as Ti content increases, the Na environment becomes more ordered. These results suggest that the Ti activity coefficient and its isotopic fractionation for magnetite and other Ti-bearing minerals should be fairly constant in polymerized melts, such as metaluminous and peraluminous rhyolites. [ABSTRACT FROM AUTHOR]

The speciation of water dissolved into and reacted with hydrous alumino-silicate glasses (of NaAlSi 3 O 8 and "rhyolitic" compositions) quenched from high temperature is re-investigated where the predominant species are expected to be "X(Si and Al)-OH" and "H 2 O". Only, two analytical methods are capable of assessing such speciation: Near InfraRed (NIR) spectroscopy and solid-state 1H Nuclear Magnetic Resonance (NMR) spectroscopy. It is observed that the apparent water speciation, as a function of total water content, as determined by NIR spectroscopy is nearly the opposite from what the 1H NMR data reveal. Deuterium (2H) NMR and silicon (29Si) NMR report consistent trends in apparent speciation (depolymerization) with those indicated by the 1H NMR data. Compared with four previous NMR studies of hydrous NaAlSi 3 O 8 glasses it is shown that whereas NIR data always report the same apparent systematic variation in the intensity of the 4500 ("X-OH") and 5200 ("H 2 O") cm−1 bands with total water content, multiple 1H NMR studies of hydrous NaAlSi 3 O 8 report a wide range in OH/H 2 O. The discrepancy between the various NMR studies likely reflects differences in how the various glasses were made. Specifically, quench rate (fast or slow) and synthesis pressure (higher or lower), might impose a strong effect on observed water speciation in glasses via 1H NMR. It is concluded that the application of NIR spectroscopy, specifically the use of the intensities of the 4500 ("X-OH") and 5200 ("H 2 O") cm−1 NIR bands, does not provide an accurate assessment of water speciation in hydrous alumino-silicate glasses. NIR spectroscopy does remain a very valuable analytical tool for determination of total water content. [ABSTRACT FROM AUTHOR]

Evolution of fault strength during the initial stages of seismic slip plays an important role in the onset of velocity‐induced weakening, which in turn, leads to larger earthquake events. A key dynamic weakening mechanism during the early stages of slip is flash heating, where stress concentrations at contacts on the interface lead to the rapid generation of heat. Although potential weakening from flash heating has been extensively modeled, there is little recorded microstructural evidence of its physical manifestations. We present results of a series of triaxial experiments on synthetic faults in quartz sandstone. Samples were subjected to a variety of normal stresses and ambient temperatures, to induce a range of slip event sizes and sliding velocities. We show the microstructural evolution of asperity interactions from the onset of flash heating through to the formation of grain‐scale areas of sheared melt. Using microstructural observations and mechanical data from the experiments, we model temperature and the viscoelastic behavior of the glass. Results suggest that, in the earliest stages of slip asperity contacts melt, but temperatures remain too low for viscous shear to occur within the melt layer. Instead melted asperities behave as glassy solids, facilitating continued frictional heating. With further slip, increased asperity temperatures allow the transition to viscous shear within the melt layer, facilitating weakening. These results highlight the dynamic evolution of the viscoelastic properties of the melt and resulting effects on asperity strength. Such complexity has, to‐date, not been fully addressed in modeling of flash heating. Key Points: A microstructural evolution of asperity interactions is documented from the onset of flash heating to the formation of sheared meltHigh strain rates and long relaxation times cause melted asperity tips to behave as glassy solidsMelt rheology is essential for understanding asperity behavior and strength during the early stages of earthquake slip [ABSTRACT FROM AUTHOR]

To better understand the solution of volatile species in a reduced magma ocean, we identify via Raman spectroscopy the nature of C-O-H-N volatile species dissolved in a series of reduced basaltic glasses. The oxygen fugacity (ƒ O2) during synthesis varied from highly reduced at two log units below the iron-wustite buffer (IW-2.1) to moderately reduced (IW-0.4), spanning much of the magmatic ƒ O2 conditions during late stages of terrestrial accretion. Raman vibrational modes for H 2 , NH 2 –, NH 3 , CH 4 , CO, CN–, N 2 , and OH– species are inferred from band assignments in all reduced glasses. The integrated area of Raman bands assigned to N 2 , CH 4 , NH 3 and H 2 vibrations in glasses increases with increasing molar volume of the melt, whereas that of CO decreases. Additionally, with increasing ƒ O2 , CO band areas increase while those of N 2 decrease, suggesting that the solubility of these neutral molecules is not solely determined by the melt molar volume under reduced conditions. Coexisting with these neutral molecules, other species as CN–, NH 2 – and OH– are chemically bonded within the silicate network. The observations indicate that, under reduced conditions, (1) H 2 , NH 2 –, NH 3 , CH 4 , CO, CN–, N 2 , and OH– species coexist in silicate glasses representative of silicate liquids in a magma ocean (2) their relative abundances dissolved in a magma ocean depend on melt composition, ƒ O2 and the availability of H and, (3) metal-silicate partitioning or degassing reactions of those magmatic volatile species must involve changes in melt and vapor speciation, which in turn may influence isotopic fractionation. [ABSTRACT FROM AUTHOR]

A polycrystalline specimen of liebermannite [KAlSi 3 O 8 hollandite] was synthesized at 14.5 GPa and 1473 K using glass starting material in a uniaxial split-sphere apparatus. The recovered specimen is pure tetragonal hollandite [SG: I4/m ] with bulk density of within 98% of the measured X-ray value. The specimen was also characterized by Raman spectroscopy and nuclear magnetic resonance spectroscopy. Sound velocities in this specimen were measured by ultrasonic interferometry to 13 GPa at room T in a uniaxial split-cylinder apparatus using Al 2 O 3 as a pressure marker. Finite strain analysis of the ultrasonic data yielded K S 0 = 145(1) GPa, K 0 ′ = 4.9(2), G 0 = 92.3(3) GPa, G 0 ′ = 1.6(1) for the bulk and shear moduli and their pressure derivatives, corresponding to V P0 = 8.4(1) km/s, V S 0 = 4.9(1) km/s for the sound wave velocities at room temperature. These elasticity data are compared to literature values obtained from static compression experiments and theoretical density functional calculations. [ABSTRACT FROM AUTHOR]