Glasses and Melts : Linking Geochemistry and

Glasses and Melts: Linking Geochemistry and Materials Science Georges Calas,1 Grant S. Henderson,2 and Jonathan F. Stebbins3 S ilicate melts are major components of magmatic activity and of its most spectacular expression, volcanic eruptions. The "hidden part" is even more fascinating, as silicate melts are directly involved in matter and heat transfer within the Earth and planets. Silicate glasses, often investigated as a frozen picture of their molten counterparts, are also materials of major importance in technology. Despite the difficulties in rationalizing physical and chemical properties of glasses and melts, due to an incomplete knowledge of their structure, major progress has been made recently in synthetic and natural systems. This issue of Elements reviews the properties of silicate glasses and melts from the molecular to the field scale. It includes insights into their technological applications and describes some recent advances in this fast-evolving field. KEYWORDS: magmas, volcanology, geochemistry, glasses, silicate materials INTRODUCTION Silicate glasses and melts have been of interest to mineralogists, geochemists and igneous petrologists for many decades. Silicate melts are essential components of igneous processes, participating actively in Earth differentiation, heat transfer and global geodynamics, and geochemical cycles. Melt viscosity can decrease some 20 orders of magnitude below that of corresponding solids, and timescales for transport properties are dramatically reduced in the presence of a liquid-phase. In addition, melts relevant to volcanological processes contain various amounts of volatiles, mainly H2O, CO2, and sulfur-bearing species. Models for Earth’s formation propose a period of substantial melting of the planet during the later stages of accretion, due to the extremely high energies involved in a probable Moon-forming “giant impact” event. As properties of deep magmas must be experimentally constrained, researchers studying melts and glasses participate actively in current experimental developments in Earth sciences, such as in situ experiments at high temperature (T) and high pressure (P) and numerical modeling of complex systems. Many of these experiments are designed to aid in developing an understanding at the atomic level of the structure and dynamics of silicate melts under the P–T conditions of Earth’s crust and mantle. Current issues range from the conditions of formation of planetary systems to volcanological monitoring and Earth’s dynamics. 1 Institut de Minéralogie et de Physique des Milieux condensés, CNRS UMR 7590, Université Pierre et Marie Curie–Paris 6, Université Paris 7, Institut de Physique du Globe de Paris, 75005 Paris, France. E-mail: calas@impmc.jussieu.fr 2 Department of Geology, University of Toronto, 22 Russell Street, Toronto, Ontario M5S 3B1, Canada. E-mail: henders@geology.utoronto.ca 3 Department of Geological and Environmental Sciences, Stanford University, Stanford, California 94305-2115, USA. E-mail: stebbins@stanford.edu ELEMENTS, VOL. 2, PP. 265–268 27Al NMR spectra of Ca-aluminosilicate glasses quenched from melts at 1 bar, 5 GPa, and 10 GPa. Peaks for 4-, 5-, and 6coordinated Al are shown (ALLWARDT ET AL. 2005). Silicate glasses and melts have been of interest in ceramic science, glass technology, and metallurgy for centuries, and more recently in the management of wastes (municipal, industrial and nuclear). A glass plant typically produces 600–1000 tons/day of silicate melts. These applications represent a major field of connection between Earth sciences and industry, and provide job opportunities for students. In addition, studies of archeological and historical glasses make them interesting as analogs for the longterm evolution of waste matrices and give insights into ancient technologies and sources of trade. This broad interest in glasses and melts stems not only from their roles in Earth sciences and technology, but from their peculiar physicochemical properties, which are correlated to structural organization at the atomic scale. Terms used to describe glass/melt structure are defined in the glossary. The atomic structure and dynamic processes that occur at high temperature provide a basis for understanding the relationships between melt structure, thermodynamic properties, and rheology in these liquids. These fundamental relations permit extrapolation from laboratory-scale simplified melt compositions to multicomponent geological systems. For instance, atomic motion, favored by the high temperatures of silicate melts, is at the origin of the rheological properties of magmas. Element diffusion and gas dissolution are also enhanced relative to solid rocks; this gives magmas their unique geochemical properties. This issue of Elements presents some aspects of glass and melt science and shows that defining structure–property relationships (Calas et al. 2003a; Mysen and Richet 2005) and linking microscopic to macroscopic phenomena (Stebbins et al. 1995) are especially intriguing fields of research. THE STRUCTURE OF SILICATE GLASSES AND MELTS: SOME BASICS Information about the molecular-scale structure of silicate glasses and melts and the dynamics of silicate melts is needed to understand their properties at a fundamental level. This task is a challenge because silicate melts possess no long-range structural periodicity or symmetry. Even today the structure of a simple network glass such as SiO2 remains controversial, with no consensus as to its precise nature (cf. Henderson 2005). However, silicate glasses and melts do possess short-range order, usually manifested as polyhedral units such as tetrahedra, 5-, or 6-coordinated 265 O CTOBER 2006 sites. The most detailed information on the local atomic environments comes from a combination of spectroscopic and diffraction measurements and bulk property determinations. These data are used to construct a working model of the glass and melt structure. In addition, their structure has some medium-range order, depending on bulk composition and temperature. Insight into the structural organization beyond this limit is mostly obtained from numerical modeling. represents the simulated structure of a borosilicate glass and illustrates the heterogeneous distribution of cations (Henderson et al. 2006). The structure of glasses and melts depends on their chemical composition. To a first approximation, a silicate glass or melt can be understood as a network of tetrahedra (FIG. 1A) bonded by non-bridging oxygens to network-modifying cations, such as Na+ and Mg2+, which depolymerize the silicate network. However, in the presence of trivalent tetrahedral cations, such as Al3+ and B3+, cations are charge compensators and then contribute to increasing the connectivity of the network. This connectivity, defined by the relative abundances of Qn species, has a strong influence on the viscosity of silicate melts and other melt properties. The lack of periodic structure and disorder effects are a major limitation for the structural description of silicate glasses and melts. FIGURE 1B shows how disorder can act in the absence of a periodic structure: silicate tetrahedra are free to rotate, which causes a loss of information over the medium-range structure scale. Numerical modeling provides the possibility of describing the structure of multicomponent glasses. It confirms recent experimental observations that the structure of most glasses and melts, in contrast to silica and other network glasses, e.g. glasses of feldspar composition, is not simply the continuous random network illustrated in the majority of textbooks. FIGURE 1C MELTS AND GLASSES A direct visualization of the existence of well-defined sites in silicate glasses is provided by the magnificent coloring of medieval stained glass: the optical absorption bands arising from crystal field effects indicate the oxidation state and site geometry of transition elements (FIG. 2). Any liquid might become glassy if cooled rapidly enough. The glass transition is characterized by a change in thermodynamic properties (e.g. heat capacity), from solid-like to liquid-like values, with changing temperature. The glass transition temperature depends on thermodynamic and kinetic parameters (Wong and Angell 1976). Glasses provide a frozen image of the melt “structure” at Tg; in the reverse direction, a glass gradually softens to a molten state above Tg. This property is at the origin of glass making. Blowing the supercooled liquid remains a dramatic aspect of the glass craft industry. FIGURE 3 depicts the progressive cooling of an industrial silicate melt. As a melt is cooled, its structure rearranges more and more slowly. At some point (if crystallization is avoided by slow kinetics), the rate of this ordering process can no longer keep pace with the rate of cooling. The melt falls out of equilibrium, forming a glass, with the melt structure “frozen in” at a certain “fictive temperature” (Tf). Faster cooling thus leads to a higher Tf, allowing the melt structure to be sampled over a range in temperatures and studied conveniently and precisely at ambient temperature. This method is limited to the experimentally obtainable GLOSSARY Amorphous material – A solid that lacks a periodic crystalline Medium- or intermediate-range order (MRO, IRO) – Structural arrangement out to a radial distance that incorporates nextnearest neighbors. This arrangement can also incorporate polyhedral organization such as rings. structure, i.e. it lacks translational symmetry, or a liquid Bridging oxygen (BO) – Oxygen atoms that are shared between two network formers Charge compensator – A cation or anion that is in close proximity to another cation or anion in order to provide an overall neutral charge to a polyhedral unit, e.g. Na+ charge compensates Al3+ near [AlO4]- tetrahedra. Melt – Hot molten liquid found in magmas Network – A three-dimensional structure composed of network formers. It may or may not be continuous, ordered or disordered. Network former – Cations that bond to bridging oxygens to form Clusters – Small groupings of atoms or polyhedra that have some the network. They usually have coordination numbers of 3 or 4. sort of directionality or non-randomness associated with them Network modifier – Cations that disrupt the network and pro- Disorder (structural) – Applied to arrangements of atoms that duce non-bridging oxygens. They usually have a coordination number of 6. cannot be described by well-defined bond angles and distances and/or more-random distributions of adjacent cations or anions Non-bridging oxygen (NBO) – Oxygen atoms that are bonded to only one network former Fictive temperature (Tf) – The temperature at which a liquid was last in thermal equilibrium during cooling to form a glass. It therefore characterizes the thermal history of the glass and is higher for more rapid cooling. Order (structural) – Applied to arrangements of atoms that can Glass – An amorphous material that exhibits the glass transition Polyamorphism – Two liquids or glasses with the same compo- be described by well-defined bond angles and distances and/or non-random distributions of adjacent cations or anions sition but different structures Glass transition – A relatively narrow temperature region over which second-order thermodynamic properties, such as heat capacity and thermal expansivity of a glass, decrease from liquidlike to solid-like values Glass transition temperature (Tg) – A specific point on the temperature curve of a glass property measured at a standard heating rate; often associated with a viscosity of 1012 Pa Intermediate cation – Cations that may act as either a network former or a network modifier depending upon coordination Polymerization – The bonding together of polyhedral units to form two- and three-dimensional structures Qn species – A network former that has 0 to 4 bridging oxygens (n = 0–4) bound to the cation Short-range order (SRO) – Structural arrangement out to a radial distance that incorporates nearest neighbors Topology – The interconnections between atoms that describe Magma – A naturally occurring high-temperature melt, generally the network containing dissolved gases and often mixed with bubbles or crystals or both ELEMENTS 266 O CTOBER 2006 B A C (A) (Left) Schematic of two SiO4 tetrahedra containing 3 NBO (blue) and bound together through a BO (red). Each tetrahedron is a Q1 species. (Right) Schematic of four SiO4 tetrahedra containing 2 NBO (blue) and 2 BO (red). Each tetrahedron is a Q2 species. (B) Structural disorder occurs through variation in the intertetrahedral angle (α) and the two torsional angles δ1 and δ2. The tetrahedral angle (φ) and Si–O bond lengths (d) remain relatively unchanged. (C) Schematic representation of the local structure of a borosilicate glass derived from an initial molecular dynamics simulation and showing heterogeneous distribution of cations (AFTER CALAS ET AL. 2003B). FIGURE 1 range of cooling rates (six to eight orders of magnitude), which fixes the obtainable range of Tf to generally only about 150 to 250K above the “normal” (slow cooled) Tg (Dubinsky and Stebbins 2006). An important property is the excess volume of silicate melts relative to their crystalline counterparts, as this property is at the origin of magma ascent through the mantle and crust. However, at high pressures, the high compressibility of melts may lead to important density inversions. For example, perovskite may be neutrally buoyant in a chondritically derived liquid at the core–mantle boundary in a deep magma ocean (Miller et al. 1991). A schematic diagram of an oxygen-diffusion mechanism is shown in FIGURE 4A (Farnan and Stebbins 1994). The diffusion mechanism is initiated when a non-bridging oxygen of the SiO4 tetrahedron A (probably with three other bridging oxygens) impinges on a tetrahedron B, where all oxygens are bridging oxygens (e.g. to tetrahedron C among others). In this situation, B transforms to an intermediate five-foldcoordinated species ([5]Si, step 1). Diffusion occurs when [5]Si dissociates, and an oxygen that was initially bonded to B (e.g. green circle) is now associated with a neighboring tetrahedron such as C (step 2). The increase of higher-coordinated species is consistent with increasing diffusivity in the liquid, which implies that viscosity decreases, since diffusivity and viscosity are inversely correlated. These atomicscale mechanisms are at the origin of important transport properties, such as diffusivity or viscosity (FIG. 4B). Understanding melt thermodynamics is important, as the free energy is needed to calculate phase equilibria. However, changes of melt structure with increasing temperature are indicated by the observation that liquid-state heat capacities and thermal expansions are generally considerably higher than the solid-like values of glasses, requiring significant “configurational” components (Mysen and Richet 2005). These must be considered to develop accurate structure-based models of melt properties. The Dynamics of Silicate Melts Silicate melts are not static mixtures of silicate polymer species and cations, but contain dynamic, rapidly exchanging chemical environments, with bond-breaking and ionic exchange between structural units occurring on a timescale comparable with those of viscous flow and chemical diffusion. In particular, the Adam-Gibbs theory of relaxation allows a quantitative link to be made between the timescale for viscous flow and the configurational entropy of the melt (Mysen and Richet 2005). Silicate melts are viscoelastic liquids, with a finite relaxation time related to dissipation of the strain induced by chemical, mechanical or thermal stress. Relaxation time increases with decreasing temperature down to the glassy state. ELEMENTS Detail of a stained glass, Notre Dame de Chartres, France (early thirteenth century). Glass coloration originates from crystal field transitions of minor transition elements. PHOTO COURTESY OF THE DATABANK PALISSY, MINISTÈRE DE LA CULTURE ET DE LA COMMUNICATION, FRANCE. FIGURE 2 IN THIS ISSUE In the articles that follow, the unique structures and properties of glasses, melts, and magmatic systems are further explored. Grant Henderson, Georges Calas and Jonathan Stebbins give insight into the peculiar aspects of the atomic-scale structure of glasses and melts, including the major changes observed at high temperature and at high pressure. Harald Behrens and Fabrice Gaillard discuss the 267 O CTOBER 2006 A B A silicate melt emerging from an industrial furnace and being transformed into a bottle. PHOTO COURTESY OF SAINT-GOBAIN COMPANY FIGURE 3 influence of volatiles on the properties of silicate melts. They show how the speciation of these volatiles affects their solubility, the rheology and redox state of magmas, and the stability of minerals. The parameters controlling the transport properties of magmas, mostly diffusion and rheology, are addressed in the article by Don Dingwell. He brings attention to the major experimental and theoretical advances in viscous behavior and diffusion, which give a new perspective on magma migration and volcano dynamics. Bruce Marsh discusses the dynamic aspects of magmatic systems, setting this question in a historical perspective. Field studies, including investigations on the fascinating Sudbury complex, provide the real-scale validation of laboratory experiments and outline pertinent parameters, such as marginal solidification fronts and magma ascent. The final contribution presents examples of properties of industrial and natural glasses. Laurence Galoisy emphasizes the role of transition elements in defining glass coloration and stability, and in defining the formation conditions of volcanic and impact glasses. Silicate glasses and melts play a critical role in many geological and technological processes. From the molecules to the planetary scale, they constitute a broad though still REFERENCES Allwardt JR, Stebbins JF, Schmidt BC, Frost DJ, Withers AC, Hirschmann MM (2005) Aluminum coordination and the densification of high-pressure aluminosilicate glasses. American Mineralogist 90: 1218-1222 Calas G, Cormier L, Galoisy L, Jollivet P (2003a) Structure-property relationships in multi-component oxide glasses. Compte-Rendus Chimie 5: 831-843 Calas G, Le Grand M, Galoisy L, Ghaleb D (2003b) Structural role of molybdenum in nuclear glasses: an EXAFS study. Journal of Nuclear Materials 322: 15-20 ELEMENTS (A) A cartoon of a possible atomistic mechanism for viscous flow based on the presence of an intermediate 5-fold Si species. Large circles: oxygens; smallest circles: network forming cations; intermediate circles: network modifier cations. (B) Fluid lavas flowing during the October 4, 2005, eruption of the Piton de la Fournaise, Réunion Island. PHOTO COURTESY OF INSTITUT DE PHYSIQUE DU GLOBE DE PARIS FIGURE 4 largely unknown universe, in which the mineralogist, the geologist, and the materials scientist must often find original ways to solve the many questions that are still open. This issue of Elements shows the rapid progress that has been made recently. We hope it will encourage young scientists to enter this fascinating field! ACKNOWLEDGMENTS We thank the authors of the articles presented in this issue, who give a synthetic and interdisciplinary view of the complex world of glasses and melts. Editors Pierrette Tremblay and Bruce Watson are thanked for their efforts in bringing this issue together. This is IPGP contribution. . Dubinsky EV, Stebbins JF (2006) Quench rate and temperature effects on framework ordering in aluminosilicate melts. American Mineralogist 91: 753-761 Farnan I, Stebbins JF (1994) The nature of the glass transition in a silica-rich oxide melt. Science 265: 1206-1209 Henderson GS, Calas G, Stebbins JF (2006) The structure of silicate glasses and melts. Elements 2: 269-273 Henderson GS (2005) The structure of silicate melts: a glass perspective. Canadian Mineralogist 43: 1921-1958 Miller GH, Stolper EM, Ahrens TJ (1991) The equation of state of a molten komatiite 2: Application to komatiite 268 petrogenesis and the Hadean mantle. Journal of Geophysical Research 96: 11849-11864 Mysen B, Richet P (2005) Silicate Glasses and Melts: Properties and Structure, Elsevier, Amsterdam, 560 pp Stebbins JF, McMillan P, Dingwell B (eds) (1995) Structure, Dynamics, and Properties of Silicate Melts. Reviews in Mineralogy 32, Mineralogical Society of America , Washington DC, 616 pp Wong J, Angell CA (1976) Glass Structure by Spectroscopy, Dekker, New York, 864 pp . O CTOBER 2006