Kerimasite, ideally is a new calcium zirconium silicate-ferrite member of the garnet group from the extinct nephelinitic volcano Kerimasi and surrounding explosion craters in northern Tanzania. The mineral occurs as subhedral crystals up to 100 μm in size in calcite carbonatites, and as euhedral to subhedral crystals up to 180 μm in size in carbonatite eluvium. Kerimasite is light to dark-brown in colour and transparent with a vitreous lustre. No cleavage or parting was observed and the mineral is brittle. The calculated density is 4.105(1) g/cm3. The micro-indentation, VHN25, ranges from 1168 to 1288 kg/mm2. Kerimasite is isotropic with n = 1.945(5). The average chemical formula of the mineral derived from electron microprobe analyses (sample K 94-25) and calculated for O = 12 and all Fe as Fe2O3 is (Ca3.00Mn0.01Ce0.01Nd0.01)Σ3.03(Zr1.72Nb0.14Ti0.08Mg0.02Y0.02)Σ1.98(Ti0.09)Σ3.00O12. The largest Fe content determined in kerimasite is 21.6 wt.% Fe2O3 and this value corresponds to 1.66 a.p.f.u. in the tetrahedral site. Kerimasite is cubic, space group with a = 12.549(1) Å, V = 1976.2(4) Å3 and Z = 8. The five strongest powder-diffraction lines [d in Å, (I/Io), hkl] are: 4.441 (49) (220), 3.140 (91) (400), 2.808 (70) (420), 2.564 (93) (422) and 1.677 (100) (642). Single-crystal structure refinement revealed the typical structure of the garnet-group minerals. The name is given after the locality, Kerimasi volcano, Tanzania.

Using high intensity beams of fast electrons, the transmission electron microscope (TEM) and scanning transmission electron microscope (STEM) enable comprehensive characterization of rocks and minerals at micrometre to sub-nanometre scales. This review outlines the ways in which samples of Earth and planetary materials can be rendered sufficiently thin for TEM and STEM work, and highlights the significant advances in site-specific preparation enabled by the focused ion beam (FIB) technique. Descriptions of the various modes of TEM and STEM imaging, electron diffraction and X-ray and electron spectroscopy are outlined, with an emphasis on new technologies that are of particular relevance to geoscientists. These include atomic-resolution Z-contrast imaging by high-angle annular dark-field STEM, electron crystallography by precession electron diffraction, spectrum mapping using X-rays and electrons, chemical imaging by energy-filtered TEM and true atomic-resolution imaging with the new generation of aberration-corrected microscopes. Despite the sophistication of modern instruments, the spatial resolution of imaging, diffraction and X-ray and electron spectroscopy work on many natural materials is likely to remain limited by structural and chemical damage to the thin samples during TEM and STEM.

Studies of the inclusions contained in natural diamonds have shown the occurrence of minerals which must have formed at depths below the lithosphere and which may be closely matched with the silicate mineral assemblages determined by high pressure and temperature experimental studies for depths of 300 to 800 km in the Earth's mantle. The inclusions come principally from two main depth zones: (1) the lower asthenosphere and upper transition zone; (2) the Upper Mantle/Lower Mantle (UM/LM) boundary region and the uppermost LM. The inclusions from zone 1 are very largely majoritic garnets (with or without clinopyroxene) which indicate bulk compositions of eclogitic/metabasic affinity. The minerals from zone 2 include Ca-Si and Mg-Si perovskites and ferropericlase and are dominantly of metaperidotitic bulk composition, but include some possible metabasite assemblages. In many of these natural assemblages, the tetragonal almandine pyrope phase occurs rather than the garnet found in experiments.

As natural diamonds are believed to crystallize in fluids/melts, the hypothesis is developed that the restriction of diamonds and inclusions of particular compositions to the above two depth intervals is because they are controlled by loci of fluid/melt occurrence. Attention is focused on subduction zones because both suites of inclusions show some evidence of subducted protoliths. The lower zone (600–800 km) coincides with the region where dehydration may be expected for hydrous ringwoodite and dense hydrous Mg-silicates formed in subducted peridotites. The dehydration of lawsonite in subducted metabasites provides a particular location for melt formation and the inclusion of the shallower (~ 300 km) majoritic inclusions. For the deeper majoritic inclusions in the region of the upper transition zone, melt development may occur as a consequence of the hydrous wadsleyite-to-olivine transformation, and such melt may then interact with the upper crustal portion of a subducting slab. These suggestions offer an explanation of the depth restrictions and the compositional restrictions of the inclusions. The differences in δ13C values in the host diamonds for the two suites of inclusions may also be explained on this basis.

Recent advances in computer simulation at the atomic scale have made it possible to probe the structure and behaviour of the cores of dislocations in minerals. Such simulation offers the possibility to understand and predict the dislocation-mediated properties of minerals such as mechanisms of plastic deformation, pipe diffusion and crystal growth. In this review the three major methods available for the simulation of dislocation cores are described and compared. The methods are: (1) cluster-based models which combine continuum elastic theory of the extended crystal with an atomistic model of the core; (2) dipole models which seek to cancel the long-range elastic displacement caused by the dislocation by arranging for the simulation to contain several dislocations with zero net Burgers vector, thus allowing a fully periodic super-cell calculation; and (3) the Peierls-Nabarro approach which attempts to recast the problem so that it can be solved using only continuum-based methods, but parameterizes the model using results from atomic-scale calculations. The strengths of these methods are compared and illustrated by some of the recent studies of dislocations in mantle silicate minerals. Some of the unresolved problems in the field are discussed.

Pyrometallurgical slags from three Cu-Co smelters (Nkana, Mufulira, Chambishi) in the Copperbelt Province, Zambia, were studied from mineralogical and chemical points of view. The slags were enriched in metals and metalloids, mainly Cu (up to 35 wt.%), Co (up to 2.4 wt.%) and As (up to 3650 ppm). The following primary phases were observed in slags: Ca-Fe silicates (clinopyroxene, olivine) and leucite, oxides (spinel-series phases), ubiquitous silicate glass and sulphide/metallic droplets of various sizes. The presence of glass and skeletal/dendritic crystal shapes indicated rapid cooling of the slag melt. Copper and cobalt were found in low concentrations in the majority of silicates (olivine, clinopyroxene) and oxides, substituting for Fe in their structures (up to 7.15 wt.% CoO in olivine, 4.11 wt.% CuO in spinel). Similarly, up to 0.91 wt.% CoO and 6.90 wt.% CuO were observed in the interstitial glass. Nevertheless, the main carriers of these metals in the slags studied were Cu sulphides (digenite, chalcocite, bornite, chalcopyrite), Co-Fe sulphides (cobaltpentlandite), Co-bearing intermetallic phases ((Fe,Co)2As) and alloys. Weathering features corresponding to the presence of secondary metal-bearing phases, such as malachite (Cu2(CO3)(OH)2), brochantite (Cu4SO4(OH)6) and sphaerocobaltite (CoCO3), were observed on the slag surfaces. They indicate that the slags studied are reactive on contact with water/atmosphere and that their environmental stability and release of potentially harmful metals and metalloids must be evaluated further.

The volumes of a disordered An20 (Qod = 0.15), a disordered An78 (Qod = 0.55) and an ordered An78 (Qod = 0.81) were determined up to 9.569(10) GPa, 8.693(5) GPa and 9.765(10) GPa, respectively, using single-crystal X-ray diffraction. The volume variations with pressure for these samples are described with 4th-order Birch Murnaghan equations of state with V0 = 669.88(7) Å3, K0 = 59.7(7) GPa. K′ = 5.7(5), K″ = −0.8(2) GPa−1 for disordered An20, V0 = 1340.48(10) Å3, K0 = 77.6(5) GPa, K0′ = 4.0(3), K″ = -0.59(9) GPa−1 for disordered An78 and V0 = 1339.62(6) A3, K0 = 77.4(6) GPa, K′ = 4.2(4), and K″ = −0.7(1) GPa−1 for ordered An78. Along with data from previous studies (An0 ordered, An0 disordered and An2o ordered), the volumes for the disordered samples were found to be up to ∼0.3% larger than the ordered samples of the same composition. The disordered samples are softer than the ordered samples of the same composition by 4(1)% for An0, 2.5(9)% for An20 and essentially zero for An78. The relationship between volume increase, density decrease, and decreasing bulk modulus with increasing disorder is in accordance with Birch's Law.

The crystal structure of quintinite-2H-3c, [Mg4Al2(OH)12](CO3)(H2O)3, from the Kovdor alkaline massif, Kola peninsula, Russia, was solved by direct methods and refined to an agreement index (R1) of 0.055 for 484 unique reflections with |Fo| ≥ 4σF. The mineral is rhombohedral, R32, a = 5.2745(7), c = 45.36(1) Å. The diffraction pattern of the crystal has strong and sharp Bragg reflections having h–k = 3n and l = 3n and lines of weak superstructure reflections extended parallel to c* and centred at h–k ≠ 3n. The structure contains six layers within the unit cell with the layer stacking sequence of …AC=CA=AC=CA=AC=CA… The Mg and Al atoms are ordered in metal hydroxide layers to form a honeycomb superstructure. The full superstructure is formed by the combination of two-layer stacking sequence and Mg-Al ordering. This is the first time that a long-range superstructure in carbonate-bearing layered double hydroxide (LDH) has been observed. Taking into account Mg-Al ordering, the unique layer sequence can be written as …=Ab1C=Cb1A=Ab2C=Cb2A=Ab3C=Cb3A=… The use of an additional suffix is proposed in order to distinguish between LDH polytypes having the same general stacking sequence but with different c parameters compared with the ‘standard’ polytype. According to this notation, the quintinite studied here can be described as quintinite-2H-3c or quintinite-2H-3c[6R], indicating the real symmetry.

The change in chemical composition trend of magmatic nepheline through magma evolution has been characterized from the alkaline series of the Messum complex in which nepheline occurs in a succession of different mineral parageneses from mafic-rich (theralites) to strongly evolved felsic-rich rock types (nepheline syenites). The nepheline compositions are dependent on those of coexisting feldspar(s). They record an evolution parallel to that of the melt schematized according to experimental phase diagrams, from initially Ca-rich compositions in equilibrium with calcic plagioclase towards increasingly Ca-poor, Na-rich and Si-rich compositions. The K contents show a maximum that corresponds to the appearance of alkali feldspar in the parageneses. This evolution is qualitatively preserved in spite of the low-T Na/K re-equilibration typical of plutonic nephelines. Although a slight increase in the silica content of nepheline is consistent with the experimentally defined magmatic trend, several high-silica nephelines from the Messum rocks as well as from other reported occurrences, cannot be reconciled with the experimental data. The nepheline solid-solution model available suggests that such ‘abnormal’ compositions might be related to different crystallization mechanisms between natural nephelines and some synthetic analogues.

Geochemical data are presented for five large Palaeogene dykes, members of the Mull swarm in southern Scotland and northern England (the Moneyacres, Hawick-Acklington, Barrmill, Muirkirk- Hartfell and Dalraith-Linburn dykes). The rock types range from basalt through andesite to dacite, although the range in individual intrusions is more restricted. The dykes are divisible into two groups; those where the compositional variation was generated by fractional crystallization of basaltic magmas, and those where it resulted from variable degrees of mixing of basaltic and silicic magmas. Several dykes are composite; the marginal facies can be more or less evolved than the central facies. The dyke magmas are thought to have originated from stratified magma chambers beneath the Mull centre and models are presented to show how the different components were derived from the chambers. Some dykes appear to have been terminated at or near the Southern Upland Fault, perhaps as a result of the chilling of early magma pulses by water in the fault. The Palaeogene dyke swarm is considerably more complex than previously recognized and has a significant input to models of the evolution of the Mull magmatic system.

Accessory monazite-(Ce) with an extraordinarily high proportion of the xenotime component in solid solution of 21–42 mol.% (6.5–14 wt.% Y2O3, 6–11 wt.% HREE2O3) was discovered in a retrogressed Variscan high-pressure, high-temperature granulite from the southern Bohemian Massif, Austria. The grains with the highest proportion of xenotime (XXno ~0.4) should have had a minimum formation temperature of ~1050°C, according to published monazite-xenotime miscibility gap thermometers. This high temperature is consistent with previous petrological studies on the south Bohemian granulites indicating ~1000°C/16 kbar for the peak metamorphic stage.

A chrysotile disc is proposed as the fundamental seed structure for the formation of chrysotile asbestos fibres, polygonal serpentine fibres and polyhedral lizardite spheres. The curvature, fivefold symmetry and hydrogen-bonding alignment of the layers in the seed disc control the formation of the 15 or 30 sectors in polygonal serpentine and the orientations of the planar arrays of 15 or 30 radial crystals in polyhedral serpentine. A polygonized disc precursor to polygonal fibre formation has been observed at an arrested stage of growth in a synthesis experiment.

Concentrations of trace elements in coexisting garnet, clinopyroxene and completely miscible carbonate-silicate melt (formed at 7 GPa from the Chagatai silicocarbonatite rock known to be diamondiferous) were determined using LA-ICP-MS. The partition coefficients for Li, Rb, Cs, Ba, Th, U, Ta, Nb, La, Ce, Pb, Pr, Sr, Nd, Zr, Hf, Sm, Eu, Gd, Tb, Dy, Y, Ho, Er, Tm, Yb, Lu, Sc and Zn were determined. The new experimental data for trace-element partitioning between garnet, clinopyroxene and carbonate-silicate melt have been compared with published data for partitioning between garnet, clinopyroxene and carbonatite melt, and garnet, clinopyroxene and silicate melt. The results show that the trace-element partitioning is not significantly altered by changes in melt composition, with HREE always concentrated in the garnet. Carbonate-silicate melt, as a diamond-forming medium, and carbonatite or silicate melt equilibrated with mantle silicate minerals, behave similarly in respect of trace-element distribution.

Edwardsite, Cu3Cd2(SO4)2(OH)6·4H2O, is a new mineral from the Block 14 Opencut, Broken Hill, New South Wales, Australia. It occurs as druses of tabular and bladed crystals up to 0.06 mm in size, associated with niedermayrite and christelite. Edwardsite is pale blue, transparent with vitreous lustre and has excellent cleavage parallel to {100}. Density was not measured but the calculated density, from the empirical formula, is 3.53 g cm–3 and the Mohs hardness is ∼3. Optically, it is biaxial negative with α ∼ 1.74, β = 1.762(4), γ ∼ 1.77 and 2Vcalc. ∼ +62°. The optical orientation is X = b, Y ∼ a, Z ∼ c. Electron microprobe analysis gave (wt.%): CdO 32.43, CuO 28.06, ZnO 2.26, FeO 0.08, SO3 20.35, H2Ocalc. (from crystal-structure analysis) 14.14, totalling 99.32. The empirical formula, calculated on the basis of 18 oxygen atoms is Cu2.77Cd1.98Zn0.22Fe0.01(SO4)2.00(OH)5.95·4.06H2O. Edwardsite is monoclinic, space group P21/c, with a = 10.863(2) Å, b = 13.129(3) Å, c = 11.169(2) Å, β = 113.04(3)°, V = 1465.9(5) Å3 (single-crystal data) and Z = 4. The eight strongest lines in the powder diffraction pattern are [d (Å), (I/I0), (hkl)]: 9.991, (90), (100); 5.001, (90), (200, 21); 4.591, (45), (20); 3.332, (60), (300, 032); 3.005, (30), (03); 2.824, (40), (2); 2.769, (55), (20, 042, 10); 2.670, (45), (2). The crystal structure was determined by direct methods and refined to R1 = 3.21% using 1904 observed reflections with Fo > 4σ(Fo) collected using synchrotron X-ray radiation (λ = 0.773418 Å). The structure is based on infinite sheets of edge-sharing Cuϕ6 (ϕ: O2–, OH) octahedra and Cdϕ7 (ϕ: O2–, H2O) polyhedra parallel to (100). The sheets are decorated on both sides by corner-sharing (SO4) tetrahedra, which also corner-link to isolated Cdϕ6 octahedra, thus connecting adjacent sheets. Moderate-strong to weak hydrogen bonding provides additional linkage between sheets.

Fluoro-potassic-magnesio-arfvedsonite, ideally AKBNa2c(Mg4Fe3+)TSi8O22wF2, has been found in a dyke ∼25 km southwest of Monte Metocha, Xixano region, northeastern Mozambique. Fluoro-potassic-magnesio-arfvedsonite and low sanidine form a fine-grained mafic, ultrapotassic, peralkaline igneous rock without visible phenocrysts. The amphibole is brittle, has a Mohs hardness of 6 and a splintery fracture; it is non-fluorescent with perfect {110} cleavage and no observable parting, and has a calculated density of 3.174 gcm−3. In plane-polarized light, it is pleochroic, X= pale grey-green, Y = blue-green, Z = pale grey; X ^ c = 23.6° (in β obtuse), Y ‖ b, Z ^ c = 66.4° (in β acute). Fluoro-potassic-magnesio-arfvedsonite is biaxial negative, α = 1.652(2), β = 1.658(2), γ = 1.660(2); 2Vobs = 22.5(7)°, 2Vcalc = 30.2°. The unit-cell dimensions are a = 9.9591(4), b = 17.9529(7), c = 5.2867(2) Å, β = 104.340(1)°, V = 919.73(10) Å3, Z = 2. The nine strongest X-ray diffraction lines in the experimental powder pattern are: [d in Å(I)(hkl)]: 2.716(100)(151), 3.410(70)(131), 8.475(50)(110), 3.178(50)(310), 3.309(30)(240), 2.762(20)(31), 2.549(20)(260), 2.351(10)(51), 2.269(10)(331). Electron microprobe analysis gave: SiO2 54.25, A12O3 0.03, TiO2 1.08, FeO 6.69, Fe2O3 8.07, MgO 13.99, MnO 0.32, ZnO 0.05, CaO 1.16, Na2O 6.33, K2O 5.20, F 2.20, H2Ocalc 0.74, sum 99.18 wt.%. The formula unit, calculated on the basis of 24 (O,OH,F) with (OH+F) = 2−(2 Ti), is AKa0.98B(Na1.18Ca0.18)∑1.99C(Mg3.07Fe0.832+Mn0.04Al0.01Fe0.903+Ti0.12Zn0.01)∑=4.98TSi8O22W[Fi.o3(OH)0.73O0.24]∑2.00 and confirms the usual pattern of cation order in the amphibole structure. The presence of a significant oxo component (locally balanced by Ti at the M(1) site) is related to the crystallization conditions. The presence of Fe3+ at the T sites, originally suggested for the holotype specimen, is discounted for this amphibole composition.

Carlhintzeite, Ca2AlF7·H2O, has been found at the Gigante pegmatite, Punilla Department, Córdoba Province, Argentina. It occurs as colourless prismatic crystals up to 0.8 mm long, ubiquitously twinned on {001}. Electron microprobe analyses provided the empirical formula Ca1.98Al1.02F6.24(OH)0.76·H1.62O. A crystal fragment used for the collection of structure data provided the triclinic, C1̄ cell: a = 9.4227(4), b = 6.9670(5), c = 9.2671(7) Å, α = 90.974(6), β = 104.802(5), γ = 90.026(6)°, V = 558.08(7) Å3 and Z = 4. The crystal structure, solved by direct methods and refined to R1 = 0.0322 for 723 Fo > 4σF reflections, is made up of linkages of AlF6 octahedra, CaF8 polyhedra and CaF6(H2O)2 polyhedra. The AlF6 octahedra are isolated from one another, but share polyhedral elements with Ca polyhedra. Most notably, the Al1 octahedron shares trans faces with two CaF8 polyhedra and the Al2 octahedron shares trans edges with two CaF6(H2O)2 polyhedra. The linkage of the Ca polyhedra alone can be described as a framework in which edge-sharing chains along b are cross-linked by edge-sharing. Edge-sharing chains of Ca polyhedra along b in the carlhintzeite structure are similar to those along c in the structures of gearksutite, CaAlF4(OH)·(H2O), and prosopite, CaAl2F4(OH)4.

Syntheses for the three members of the copper hydroxyl nitrate family – the polymorphs rouaite and gerhardtite, and likasite – are presented along with powder diffraction data and unit-cell parameters. The solubilities, determined in 0.05 M KNO3 solution after equilibration at 25°C for 10 days were used to calculate activity-based solubility product constants. The Gibbs energies of formation, obtained from the solubility products, are –653.2±0.7 kJ/mol, –655.1±1.2 kJ/mol and –1506.4±1.1 kJ/mol, for rouaite, gerhardtite, and likasite (Cu3NO3(OH)5·2H2O), respectively. The values for the polymorphs rouaite and gerhardtite validate the observations of Oswald that gerhardtite is the most stable polymorph at room temperature and that the preparation of predominantly rouaite in syntheses carried out at room temperature must be due to the metastability and low rate of conversion to the more stable gerhardtite.

Quintinite-1M, [Mg4Al2(OH)12](CO3)(H2O)3, is the first monoclinic representative of both synthetic and natural layered double hydroxides (LDHs) based on octahedrally coordinated di- and trivalent metal cations. It occurs in hydrothermal veins in the Kovdor alkaline massif, Kola peninsula, Russia. The structure was solved by direct methods and refined to R1 = 0.031 on the basis of 304 unique reflections. It is monoclinic, space group C2/m, a = 5.266(2), b = 9.114(2), c = 7.766(3) Å, β = 103.17(3)°, V = 362.9(2) Å3. The diffraction pattern of quintinite-1M contains sharp reflections corresponding to the layer stacking sequence characteristic of the 3R rhombohedral polytype, and rows of weak superlattice reflections superimposed upon a background of streaks of modulated diffuse intensity parallel to c*. These superlattice reflections indicate the formation of a 2-D superstructure due to Mg-Al ordering. The structure consists of ordered metal hydroxide layers and a disordered interlayer. As the unit cell contains exactly one layer, the polytype nomenclature dictates that the mineral be called quintinite-1M. The complete layer stacking sequence can be described as …=Ac1B=Ba1C=Cb1A=… Quintinite-1M is isostructural with the monoclinic polytype of [Li2Al4(OH)12](CO3)(H2O)3.

The Li and B isotopic compositions of gem-quality Cu-bearing tourmalines were used (1) to distinguish among Paraiba tourmalines from Brazil and Cu-bearing tourmalines from Nigeria and Mozambique; and (2) to identify the likely source of Li and B for these gem-quality tourmalines. The δ11B values of tourmaline from Paraiba, Brazil, range from –42.4‰ to –32.9‰, whereas the δ11B values of Cu-bearing tourmaline from Nigeria and Mozambique range from –30.5‰ to –22.7‰ and –20.8‰ to –19.1‰ respectively. Tourmalines from each locality have relatively homogeneous δ11B values and display no overlap. There is slight overlap between δ7Li values of Paraiba tourmaline (+24.5‰ to +32.9‰) and Cu-bearing tourmaline from Nigeria (+32.4‰ to +35.4‰), and δ7Li values of Cu-bearing tourmaline from Nigeria and Mozambique (+31.5‰ to +46.8‰). Nevertheless, Cu-bearing tourmalines from each locality can be fingerprinted using a combination of their δ11B and δ7Li values. The very small δ11B values are consistent with a non-marine evaporite source, and are among the smallest reported for magmatic systems, expanding the global range of B isotopicc omposition for tourmaline by 12‰. The corresponding large δ7Li values are among the largest reported, although they are less diagnostic of the source of the Li. The large δ7Li values in conjunction with the small δ11B values suggest a non-marine evaporite or brine as a source for Li and B, either as constituent(s) of the magma source region or, by assimilation during magma ascent. The large range in δ11B and δ7Li values suggests that B and Li isotope fractionation occurred during magmatic degassing and late-stage magmatic-hydrothermal evolution of the granite-pegmatite system.

Several silicate-minerals, such as ardennite – Mn2+4MgAl5[Si5(As5+,V5+)O22](OH)6, Z = 2, tiragalloite – Mn2+4[Si3As5+O12(OH)], Z = 4 and medaite – Mn2+6[Si5(V5+,As5+)O18(OH)], Z = 4 possess (V5+,As5+,P5+)O4 tetrahedra. Using electron-microprobe analysis (EMPA) and single-crystal X-ray diffraction methods, the crystal chemistry of ardennite from Salam-Château, Belgium and the Vernetto mine, Italy, tiragalloite from the Gambatesa mine, Italy, and medaite from the Molinello mine, Italy and the Fianel mine, Switzerland, were studied. Structure refinements converged to R1 values of 2.10–5.67%. According to chemical analysis, the Σ(As+V+P) content increases with decreasing Si content. Thus, Si replaces pentavalent cations in tetrahedral coordination. The (As5+,V5+,P5+,Si4+)O4 tetrahedra are categorized by their connections to SiO4 tetrahedra. The (As5+,V5+,P5+,Si4+)O4 tetrahedron of ardennite is isolated, and those of tiragalloite and medaite terminate a tetrahedral chain. The <T–O> of the isolated (As5+,V5+,P5+,Si4+)O4 tetrahedron shows a positive correlation with the mean ionic radius. For (As5+,V5+,P5+,Si4+)O4 tetrahedra with one T–O–T link, <T–O> and mean ionic radius are also correlated. In addition, the longest bridging T–O bond occurs between (As,V,P,Si)O4 and the adjacent SiO4 tetrahedron. The bridging O atom is over-bonded to satisfy the charge requirement of Σ(As+V+Si).