How can the equilibrium and non-equilibrium thermodynamics of minerals be understood from their atomic-scale structural features? How can they be predicted, simply from models for the forces between atoms? Advances in analytical theory, statistical mechanics, experimental solid-state science, computational power, and the sophistication of a mineralogical approach that brings all of these together, means that these questions, once imponderable, are now realistically tractable. These questions have been exercising the minds of mineralogists over the last decade or so, and have motivated many developments in the science. Acting as way-markers along the path, there are a number of publications which have followed from meetings where these questions have been addressed. It is now twelve years since the publication of Microscopic to Macroscopic, an edition of Reviews in Mineralogy (Kieffer and Navrotsky, 1985) that sought to identify the fundamental controls on the bulk properties of minerals in terms of their atomic-scale characteristics.

The equilibrium structures of cubic, tetragonal and orthorhombic phases of magnesium silicate perovskite are found from first principles electronic structure calculations. Zone centre and zone boundary phonons of each phase are also calculated from ab initio forces from finite displacments, and phase transitions between the phases are analysed in terms of phonon instabilities, and coupling between modes. Both the cubic and tetragonal phases have strongly unstable modes dominated by rotation of the SiO6 octahedra, which freeze in to ultimately form the orthorhombic phase. First priniciples molecular dynamics simulations at finite temperatures are used to further investigate the stability of the intermediate tetragonal phase and the coupling between participating phonon modes. The implications for a transition temperature between orthorhombic and tetragonal phases are discussed.

Computer simulation techniques have been used to investigate defect formation and the diffusion of Ca and Mg in diopside. It was found that isolated, non-interacting CaO and MgO Schottky defects had the lowest formation energies (3.66 and 3.97 eV respectively); oxygen Frenkel defects are the most favourable oxygen defects (formation energies 3.93 eV). Magnesium and calcium self-diffusion in the c-direction of diopside is easiest by a vacancy mechanism involving either direct jumps along the c-direction, or double jumps in the b-c plane. In the extrinsic regime, diffusion activation energies for Mg are predicted to be 9.82 eV (direct route) and 1.97 eV (double jump route); for Ca diffusion, activation energies are predicted to be 6.62 eV (direct route) and 5.63 eV (double jump route). If additional vacancies (oxygen or magnesium) are present in the vicinity of the diffusion path, Ca migration energies fall to 1.97–2.59 eV. At elevated temperatures in the intrinsic regime, diffusion activation energies of ⩾ 5.95 eV are predicted for Mg self-diffusion and 9.29–10.28 eV for Ca self-diffusion. The values for Ca diffusion are comparable with published experimental data. It is inferred that a divacancy mechanism may operate in diopside crystals.

Time-of-flight neutron powder diffraction has been employed to determine precise occupancies of the M1 and M2 metal cation sites in synthetic olivines of compositions (Fe0.3Mn0.7)2SiO4, (Fe0.5Mn0.5)2−SiO4, and (Fe0.7Mn0.3)2SiO4. The distribution coefficient for Fe-Mn exchange in these samples has values of 4.864, 3.976, and 4.078, reflecting the preference of Mn2+ for the M2 site, over Fe2+. These results, and the behaviour of the difference in mean bond lengths of the two sites, indicate that, while showing a tendency towards ordering, the composition-dependence of the solid solution is near ideal.

Kinetic models and rate equations for polymorphic reconstructive phase transformations in polycrystalline aggregates are usually based on the assumptions that (a) the product phase nucleates on grain boundaries in the reactant phase and (b) growth rates of the product phase remain constant with time at fixed P-T. Recent observations of experimentally-induced transformations between (Mg,Fe)2SiO4 olivine (α) and its high pressure polymorphs, wadsleyite (β) and ringwoodite (γ), demonstrate that both these assumptions can be invalid, thus complicating the extrapolation of experimental kinetic data. Incoherent grain boundary nucleation appears to have dominated in most previous experimental studies of the α–β–γ transformations because of the use of starting materials with small (<10–20 µm) grain sizes. In contrast, when large (0.6 mm) olivine single crystals are reacted, intracrystalline nucleation of both β and γ becomes the dominant mechanism, particularly when the P-T conditions significantly overstep the equilibrium boundary. At pressures of 18–20 GPa intracrystalline nucleation involves (i) the formation of stacking faults in the olivine, (ii) coherent nucleation of γ-lamellae on these faults and (iii) nucleation of β on γ. In other experiments, intracrystalline nucleation is also observed during the β-γ transformation. In this case coherent nucleation of γ appears to occur at the intersections of dislocations with (010) stacking faults in β, which suggests that the nucleation rate is stress dependent. Reaction rims of β/γ form at the margins of the olivine single crystals by grain boundary nucleation. Measurements of growth distance as a function of time indicate that the growth rate of these rims decreases towards zero as transformation progresses. The growth rate slows because of the decrease in the magnitude of the Gibbs free energy (stored elastic strain energy) that develops as a consequence of the large volume change of transformation. On a longer time scale, growth kinetics may be controlled by viscoelastic relaxation.

At any set of thermodynamic conditions a mineral will have some well defined equilibrium crystal structure. However, this structure can be locally disturbed by crystal defects, such as domain walls or solute atoms. This distorted structure will only affect a finite volume within the crystal, but the need to retain continuity within the crystal means that this volume must be non-zero. This means, for example, that the boundary between two domains will include a transition zone from one domain's crystal structure to that of the other domain. Thick twin domain walls can be studied quantitatively, by measuring the intensity of diffuse diffraction between pairs of twin-related Bragg peaks. In alkali feldspar (Or30) at room temperature, these walls are approximately 25 Å thick. Similarly, a single solute atom in a mineral will only affect a small region within a crystal. As a result, chemical mixing will only occur in a substitutional solid solution once there is significant overlap between the strain fields around individual solute atoms. This causes the ‘plateau effect’, where the properties of a phase transition are independent of composition. In alkali feldspar, this plateau extends from albite to 2% Or, which corresponds to a strain field radius of 10 Å.

These phenomena can be modelled using Ginzburg-Landau theory, which predicts that the range of these strain fields will increase as the temperature is raised to Tc. This has been confirmed by measuring the thickness of twin walls as a function of temperature.

The thermal dehydration of naturally occurring Ca-montmorillonite has been studied by in situ X-ray diffraction at temperatures between 60–120°C. The time-temperature-dependence of the position of the basal (001) reflection reveals that interlayer water loss on isothermal dehydration occurs in two stages. After an initial rapid decrease in interlayer spacing (on shock heating to an isothermal soak temperature) the reaction proceeds towards equilibrium more slowly. Furthermore, the width of the (001) reflection changes with time, reflecting transformation-dependent changes in homogeneity perpendicular to (001) with a maximum in peak width at the point where the rate of the reaction appears to change. This suggests that, as the interlayer spacing collapses, a local change is induced in the structure, affecting the means of movement of the water from the interlayer.

Monte Carlo and molecular dynamics computer simulations are now able to provide detailed information concerning the structure, dynamics, and thermodynamics of pore fluids in 2:1 clays. This article will discuss interparticle interaction potentials currently available for atomistic simulations of clay-water systems, and will describe how computational techniques can be applied to modelling of clay systems. Some recent simulation studies of 2:1 clay hydration will then be reviewed. Comparison with experimental data promotes confidence in the molecular models and simulation techniques, and points to exciting future prospects.

First-principles quantum techniques based on density functional theory (DFT) have made important contributions to the understanding of oxide surfaces over the last four years. Important features of these calculations include: the use of periodic boundary conditions, which avoid the edge effects associated with the cluster approach; plane-wave basis sets, which make the calculation of ionic forces straightforward, so that both static relaxation and dynamical simulation can be done; and the approximate inclusion of electron correlation. A short introduction to DFT techniques is given, and recent work on the structure and energetics of a variety of oxide surfaces is presented. It is shown how the techniques can be used to study molecular and dissociative adsorption of molecules on oxide surfaces, with the emphasis on water and simple organic molecules. The growing importance of dynamical first-principles simulation in the study of surface chemical reactions is illustrated.

Extremely high abundances of TiO2 (up to 12.57 wt.%) and BaO (up to 7.11 wt.%) are reported from dark reddish-brown phlogopite-biotite rims around paler phlogopite cores in a lamprophyric dyke from Ilha da Trindade, South Atlantic. Both TiO2 and BaO contents increase towards the rims of the micas and in groundmass crystals. This is accompanied by corresponding increases in FeO*, MnO and Mg number; whereas SiO2, Al2O3, Cr2O3, MgO and K2O abundances all decrease. The substitution of Ba into the biotite structure can be explained by the scheme Ba + Al = K + Si. Ti-substitution schemes however are much more complex, as no general equation can account for all the Ti incorporated into the structure and therefore several schemes are proposed.

A vitric fiamme containing BaO-bearing sanidine (BaO = 11 wt.%) was sampled (via di Fioranello, Rome) in the pyroclastic products of the Lower Flow Unit (LFU) belonging to the Villa Senni Eruption Unit in the Colli Albani Volcanic District (Roman Comagmatic Province, Central Italy). The fiamme of the LFU is essentially made up of glass, in which scarce phenocrysts of BaO-bearing sanidine, leucite and Al2O3-bearing Ca-clinopyroxene occur. In this paper the various parameters controlling barium partitioning between alkali feldspar and LFU liquid, are discussed.

Okayamalite, Ca2B2SiO7, tetragonal, P21m, a = 7.116, c = 4.815 Å, Z = 2, is a new member of melilite group, the boron analogue of gehlenite. Electron microprobe analysis gave CaO 46.28, B2O3 28.50, SiO2 24.24, Al2O3 0.36, total 99.38 wt.%, corresponding to Ca2.01B2.00Si0.98Al0.02O7, a natural counterpart of Ca2B2SiO7 known only synthetically. The strongest lines in the X-ray powder diffraction pattern are 3.479 (40)(111), 2.862 (55)(201), 2.654 (100)(211), 2.129 (20)(301), 1.920 (35)(212), 1.644 (29)(312), very close to those of the synthetic material (a = 7.115, and c = 4.812 ). It is creamy white in colour with an earthy appearance due to the fine grain size. Streak white, cleavage not observed. Hardness ∼5½. Density calculated on the ideal formula is 3.30 g/cm3. It is optically uniaxial negative with ω = 1.700, and ɛ = 1.696. It occurs as patches of a few millimetres across in grey homogeneous-looking aggregate of wollastonite, vesuvianite, calcite and johnbaumite from Fuka mine, Bicchu-cho, Okayama Prefecture, Japan. The patches consist of very fine grains of the mineral up to 30 μm. Okayamalite is considered to be a product after the reaction formula: CaCO3 + CaSiO3 + B2O3 = Ca2B2SiO7 + CO2, arising from boron metasomatism of a wollastonite-calcite aggregate. The name is for the prefecture.

The rare iron-tin hydroxide minerals natanite and jeanbandyite occur as small pseudo-octahedral crystals in high-temperature Sn-Cu-As veins at Penberthy Croft Mine and Hingston Downs Quarry, Cornwall. The lattice parameters for jeanbandyite and natanite are reported and compared to previously calculated values. The chemical composition of jeanbandyite is discussed and a significant proportion of divalent Fe and Mn is shown to be present.

Synthetic andradite (Ca3Fe2Si3O12) has been compressed to loading pressures >21 GPa and heated to ∼1000°C by a YAG laser in a Diamond Anvil Cell (DAC). After quenching to room temperature, X-ray diffraction of the sample, still held at 21 GPa, showed that andradite had transformed to a cubic perovskite type polymorph with a = 3.460(4) Å. Upon decompression the perovskite phase transformed into an amorphous phase. The density of the perovskite polymorph (Ca3Fe2Si3O12) is ∼13.6% greater than that of isochemical andradite at 21 GPa. Ferric iron replaces Ca2+ and Si4+ in the perovskite structure (Fe3+ + Fe3+ = Si4+ + Ca2+), giving a formula unit: (Ca,Fe3+)(Si,Fe3+)O3. The new Fe3+-rich Ca-perovskite may provide new insight into the controls on the electrical conductivity of the lower mantle.