Backscattered electrons (BSE) are incident electrons reflected back from a target specimen and imaged with the scanning electron microscope (SEM). Three distinct BSE signals exist: atomic number or Z-contrast, in which composition determines image contrast; orientation contrast, in which specimen crystal structure determines image contrast; and electron channelling patterns (ECP), which are unique for a particular crystal orientation. The origins of these three signals are described, with particular attention being given to the necessary SEM operational and specimen preparation requirements. Z-contrast images are relatively simple to obtain and also have a familiar appearance such that their usage should become commonplace. ECP in comparison require subsequent interpretation which depends on the crystal structure and the relationship between crystal and specimen coordinate systems. A general solution to ECP interpretation is therefore presented, involving the construction of reference ‘ECP-maps’ over the surface of a sphere. A brief summary of the applications and potential use of the three BSE signals in the geological sciences is also given.

The modern analytical SEM, which can provide high-quality imaging facilities together with quantitative elemental analysis using an energy-dispersive spectrometer, is finding wide application in the investigation of archaeological problems. Many of these investigations involve the study of silicate and carbonate-based artefacts which may be relatively unmodified from their original geological parent raw materials so that mineralogically based interpretations are often appropriate. In this paper we present a series of examples illustrating the role of the analytical SEM in the mineralogical investigation of archaeological problems, including the characterization and provenancing of geological raw materials, the elucidation of the processes used to transform those raw materials into useful objects and the recognition and characterization of changes which archaeological artefacts may have undergone during burial or during storage.

In convergent-beam electron diffraction (CBED) a highly convergent electron beam is focussed on to a small (⩽50 nm) area of the sample. Instead of the diffraction spots that are obtained in the back focal plane of the objective lens with parallel illumination in conventional selected-area electron diffraction, CBED produces discs of intensity. The point group can be determined uniquely from the symmetry within the individual discs and the overall pattern. In order to determine the point group, it is usually necessary to record a number of CBED patterns with the electron beam aligned along different zone axes, but sometimes only one, high-symmetry pattern is required. The positions of reflections in higher-order Laue zones can be used to identify the crystal system and lattice type and to detect the presence of certain glide planes. The repeat along the zone axis that is parallel to the beam can be calculated from the diameters of the Laue zones. Hence the presence ofpolymorphs can be detected. Doubly-diffracted discs in CBED often contain a ‘line of dynamic absence’, the orientation of this line with respect to the symmetry seen in the bright field disc allows the symmetry element responsible for it (glide plane or screw diad) to be identified. This allows 191 of the 230 space groups to be uniquely identified. The measurement of specimen thickness, extinction distance and cell parameters are also briefly discussed.

In a thin specimen X-ray absorption and fluorescence can, to a first approximation, be ignored and the observed X-ray intensity ratios, IA/IB, can be converted into weight fraction ratios, , can be converted into weight fraction ratios, CA/CB, by multiplying by a constant , by multiplying by a constant kAB;

kAB values can be calculated or determined experimentally. The major correction which may have to be made to the calculated weight fraction ratio is for X-ray absorption within the specimen. The activated volume for analysis in a thin specimen is approximately 100 000 × less than in a bulk sample. Beam spreading within the specimen can be estimated using a simple formula based on a single elastic scattering event at the centre of the specimen. Examples are given of the application of the technique to obtain both qualitative and quantitative analyses from thin mineral specimens. The minimum detectable mass and the minimum mass fraction which can be measured using the technique are estimated.

Iron ore sinters prepared for the blast furnaces at Scunthorpe, Humberside, comprise iron oxides bound in a matrix of aluminium- and silicon-rich calcium ferrites with some calcium orthosilicate and a residual silicate glass. A super-silicon-deficient clinopyroxene with up to half of the Si in the chain replaced by Fe3+ and Al3+ precipitates from the glass in some areas. The iron oxides are principally magnetite and hematite with some wüstite in highly reduced areas. The calcium orthosilicate has been previously reported as the mineral larnite, β-Ca2SiO4. This study found a range of Fe for Ca substitution and three polymorphs were recognised: the range Ca2SiO4 to Ca1.85Fe0.15SiO4 has the structure of larnite; the range Ca1.85Fe0.15SiO4 to Ca1.6Fe0.4SiO4 has the structure of bredigite, α′-Ca2SiO4. Iron substitution beyond this, and up to a maximum of about Ca1.5Fe0.5SiO4, was not recognisable as a polymorph of Ca2SiO4, but an orthorhombic cell is tentatively proposed. The complex ferrite SFCA (silico-ferrite of Ca and Al) is variable in composition, but has a minimum of about 3 wt. % Al2O3 and contains 20 wt. % CaO and SiO2 with a Ca: Si ion ratio of about 2 : 1. A C-face centred monoclinic cell has been deduced: a 15.70, b 9.70, c 8.48 Å; β 105°30′. The Al appears critical for the formation of SFCA in sinter; without it, a eutectic of magnetite and bredigite precipitates in preference.

Recent TEM observations of two-phase microstructures and associated crystal defects in selected, rare dolomites have been extended to calcite-structured (R3̄c) carbonates and to other natural and synthetic carbonates that crystallize with the dolomite (R3̄) structure. The samples included siderites (FeCO3), smithsonites (ZnCO3), ankerites (Ca[Mg,Fe](CO3)2), and kutnahorites (Ca[Mn,Fe](CO3)2).

TEM methods show that the forms of second phases which result from the presence of common, divalent, metallic impurities are morphologically similar in R3̄c and R3̄ carbonates and occur more widely than hitherto realized. The most common form consists of thin ribbons of second phase which are coherent with and have the same crystallographic orientation as the host carbonate. Another form of microstructure, manifest as modulations in diffraction contrast, appears to be associated with incipient breakdown of single-phase carbonate. The results of extensive TEM/EDS microanalyses show that in siderite and ankerite the formation of ribbons is promoted by Ca impurity or Ca excess (with respect to R3̄c stoichiometry). In smithsonite, Cu and Ca impurities can play similar roles in relation to modulated microstructures. In kutnahorites, the perfection of grains and the absence of second-phase effects is strongly dependent on the ratio of Fe to Mn but is also affected by Ca in excess of the stoichiometric requirement. Electron diffraction results from several of the minerals show c-type spots, which can be interpreted as indicating ordering within basal layers of cations.

The results show that, by correlating analytical TEM data with the study of second phases and incipient two-phase microstructures, it should be possible to determine the limits of solid solubility in carbonate systems.

The form and thickness of the ferrihydrite coating on kaolinite basal surfaces was studied in mixes containing 1–10% ferrihydrite through a combination of TEM, shadowed replicas, selected area electron diffraction, and determinations of surface area and distribution of charge on the surfaces. At pH3 the ferrihydrite was found to attach initially as single particles to certain sites on the kaolinite. These particles grew laterally and vertically until with 4% ferrihydrite they were aggregates of from 8 to 25 layers of particles. With more ferrihydrite the aggregates coalesced. Despite the irregular cover on the kaolinite satisfactory electron diffraction patterns were obtained and as little as 2% ferrihydrite was detected in a mixture at pH3.

Transmission electron microscopy has been used to investigate the mechanism of natural biotite breakdown under pyrometamorphic disequilibrium conditions. Biotite in a xenolith of pelitic gneiss collected from a Tertiary dolerite sill, Isle of Mull, Scotland, shows evidence of an incipient reaction, characterised by a darkening in colour and the appearance of areas of fine-grained reaction products. TEM and analytical electron microscope data show that the reaction can be described as:

The orientations of the product phase are controlled by the crystallography of the reacting biotite, demonstrating that the transformation proceeds by a topotactic mechanism. An empirical method, based on the Mg/(Fe2+ + Fe3+) ratios of coexisting spinel and biotite from experimental data, is used to deduce that the reaction occurred above ∼ 770 °C. A comparison of the natural reaction microstructures with those produced experimentally suggest that the xenolith was probably above 800 °C for less than 48 hours and cooled to temperatures of 770 °C after ∼ 150–200 hours.

Phengite and chlorite have undergone decomposition during pyrometamorphism caused by the intrusion of a dolerite feeder pipe into Dalradian greenschists in Argyllshire, Scotland. All reaction products are extremely fine grained. Transmission electron microscopy has revealed that phengite pseudomorphs consist of biotite, spinel, mullite, sanidine and phengite, and that chlorite pseudomorphs consist of combinations of chlorite, spinel, orthopyroxene, magnetite, cordierite and biotite. Although the reactions were short-lived and did not go to completion, microprobe analysis and phase diagram analysis have revealed that there has been significant chemical interaction between the phyllosilicates and the surrounding rock. Numerous orientation relationships exist between the original minerals and their reaction products; the close-packed planes in the precursor phyllosilicates were inherited by their reaction products.

The transmission electron microscope is now used in a great variety of mineralogical and petrological contexts, The development of such applications over the past thirty years is illustrated by reference mainly to studies on serpentine and amphibole minerals.

Palagonites from the Red Sea consist of two zones: an orange palagonite which is a mixture of Mg-Al double hydroxide, Al-hydroxide and an undetermined Si-K phase, and a white palagonite, similar to motukoreaite, a Mg-Al hydroxy-sulphate-carbonate. This mineral, frequent in experimental alteration of glass by seawater, is discovered for the first time in natural palagonite. Hydroxysulphates and hydroxides are the precursors of phyllosilicates, generally found in palagonites. The very young palagonites of the Red Sea are the first link between the natural and experimental observations.

Loveringite-davidite members of the crichtonite group were synthesized at high pressure and temperature (7.5 kbar, 1000–1050 °C) from a melt of TiO2 and rare earth element (REE) enriched basaltic andesite composition. Four sets of partition coefficients for La, Srn, Ho, Lu and Sr (analogue for Eu2+) were obtained. These show that light and heavy REE are readily accommodated, but the intermediate REE are discriminated against in the loveringite—davidite structure. This confirms the previously proposed two sites (A and M) for REE substitution in the crichtonite group. Additional experiments verified the stability of REE-rich crichtonite group minerals to 20 kbar, 1300 °C and 30 kbar, 1000 °C, and indicate that this phase may be an important accessory repository for the light and heavy REE in the upper mantle.

Camptonites from the Taourirt region of Morocco contain kaersutite crystals that exhibit clear sector zoning. The electron microprobe reveals systematic variations between sectors, with preferential enrichment of Mg and Si in sector (011) and higher values for Fe, Ti and Al in sectors (110) or (100). These differences are similar to those recorded in clinopyroxenes with hour-glass zoning where this type of crystallization has been known for more than a century.

We use an approach based upon the atomistic or Born model of solids, in which potential functions represent the interactions between atoms in a structure, to calculate the infrared and Raman vibrational frequencies of forsterite. We investigate a variety of interatomic potentials, and find that although all the potentials used reproduce the structural and elastic behaviour of forsterite, only one potential (THB1) accurately predicts its lattice dynamics. This potential includes ‘bond-bending’ terms, that model the directionality of the Si-O bond, which we suggest plays a major role in determining the structural and physical properties of silicates. The potential was derived empirically from the structural and physical data of simple oxides, and its ability to model the lattice dynamics of forsterite is a significant advance over previous, force-constant models, which have been simply derived by fitting to the spectroscopic data that they aim to model. The success that we have had in predicting the lattice dynamics of forsterite indicates that the potential provides the previously elusive yet fundamental, quantitative link between the microscopic or atomistic behaviour of a mineral and its macroscopic or bulk thermodynamic properties.