High‐resolution X‐ray computed tomography (HRXCT) is a nondestructive method used to study the interiors of opaque solid objects. Here we present the results of a first application of the HRXCT method to imaging the interior of impactites, in particular, suevites (glass‐bearing impact breccias) from the Bosumtwi (Ghana) and Ries (Germany) craters and a Muong Nong–type tektite from Thailand. The aim of these studies was to determine the three‐dimensional (3‐D) distribution of clasts within the matrix of the suevites, to test this technique with respect to its suitability for the recognition of different clast types of different densities, and to determine textural characteristics of the tektite. The main part of the study concentrated on a large suevite sample (about 10 × 7 × 5 cm) from the Bosumtwi impact structure. Target rock fragments in the Bosumtwi sample consist of greywacke and sandstone/quartzitic rocks, shale and phyllite, and granites. Another large clast component is composed of impact melt and glass fragments. Macroscopic petrography and thin section petrography were used to identify the clast types in the specimen for correlation with its HRXCT signatures. The results show that HRXCT allows the easy discrimination of the relatively frothy inclusions of glassy melt in the suevites, as they are darker than the matrix in the raw X‐ray scans and can be traced through the whole sample. Color or gray scale applications allow the distinction of at least four different clast types based on density differences. The size of the smallest discernable clasts (about 0.5 mm) is determined by the resolution of the measurements, which, in turn, is a function of scan slice thickness and field of view. In the case of the tektite sample, we were also able to image the 3‐D distribution of vesicles, possibly indicating glass flow. The HRXCT method allows us to determine the three‐dimensional distribution of clast populations in impact breccias by image processing techniques and to quantify their abundances in volume percent.

Abstract We have determined the three-dimensional form of leucosome in two migmatites produced by syntectonic anatexis of different protoliths: (1) stromatic migmatite derived from pelite, which comprises sheets of leucosome (quartzofeldspathic layers with Grt) with walls of melanosome (Bt-rich selvages) in mesosome (schistose layers of Bt + Pl + Qtz ± Grt ± Crd) and (2) a migmatitic garnetamphibolite derived from basalt, which is composed of spindle-shaped leucosomes (Qtz + Pl), spatially associated with peritectic Grt, in melanosome (Hbl + Qtz ± Cpx). Three-dimensional images were generated from two-dimensional representations of spatial data obtained by two methods: (1) serial grinding and (2) high-resolution X-ray computed tomography (HR X-ray CT). Projections of three-dimensional images of stromatic migmatite derived using data from either method show the planar nature of leucosome throughout the sample; melt transport through this rock when it was partially molten could be modeled as flow in parallel conduits. In the image derived from HR X-ray CT data, garnet in leucosome is only rarely in contact with melanosome, which suggests these garnet grains were suspended in melt during flow. Projections of three-dimensional images of the migmatitic garnet-amphibolite do not reveal the full extent of leucosome connectivity, due to the irregular geometry of leucosome. Connectivity in this sample can be shown, however, by virtual slicing of the threedimensional images perpendicular to the plane of the two-dimensional representations (approximately parallel to the lineation defined by the leucosome), and by using three-dimensional projections of a single leucosome connectivity “tree” constructed by projecting leucosome patches from slice to slice and noting the overlap. Based on leucosome geometry and volume, we estimate effective porosity for flow in this rock to have been 20 vol% at stagnation. Leucosome in the migmatitic garnet-amphibolite occurs in strain shadows around garnet, which are inferred to have been obstacles to flow along linear paths. Blocking of inferred flow channels by garnet contributes to the high degree of pathlength tortuosity in this sample (τ = 2-6), which is expressed visually by the complex form of the leucosome in three dimensions. Cross-sectional areas for individual inferred melt flow paths are highly variable (over 2-3 orders of magnitude) and minimum channel radius is changeable (by ~1 order of magnitude), meaning there was large variability along the channels and implying strong local flow divergences. Based on these data, unusually straight and uniform channels would have dominated the mesoscopic melt flux through this rock when it was partially molten.

Plagioclase crystals in the slowly cooled interior of the thick Holyoke flood-basalt flow of Connecticut linked to form monomineralic chains at an early crystallization stage. Partial melting experiments reveal that when the quartz tholeiite was only 25% crystallized the chains had already linked to form a continuous 3-D network. At such an early stage of crystallization, the network was weak, highly permeable, and easily deformed. Consequently, the mush of plagioclase-chains and interstitial pyroxene crystals underwent compaction in the lower third of the flow with the expelled liquid rising to the center of the flow where it crystallized to form coarse-grained sheets of fractionated basalt.Plagioclase chains are most easily seen in the basalt after it has been partly melted and the late crystallizing minerals converted to glass. The chains are several crystals wide. The crystals, which are similar to 0.5 mm long, are attached together randomly. Normal zoning patterns indicate crystals had a brief period of growth before linking together. The chains branch every few millimeters to form the 3-D network, which was mapped using serial polished sections and X-ray CT scans. The chain frequency measured along vertical and horizontal traverses decreases toward the center of the flow. In the compaction zone, the frequency in the vertical direction is greater than in the horizontal. Making the reasonable assumption that these frequencies were initially the same, the difference is used to calculate the degree of compaction. The resulting pattern through the flow matches almost exactly the pattern indicated by variations in the incompatible elements. Plagioclase chains are also found in some coarser-grained plutonic rocks. If they are common, their fabric may provide a new, direct means of measuring the degree of compaction in crystal mushes.

High-resolution X-ray computed tomography (HRXCT) is a completely non-destructive means of examining the interiors of opaque solid objects. It produces two-dimensional images (slices) that show the interior of an object as if it had been sliced open along the image plane for viewing. Contrast in an X-ray CT image is generated by differences in X-ray absorption that arise principally from density differences. For tomographic images, X-rays pass through the object along several different paths in several different directions, resulting in an image that displays differences in density at each of several thousand points in a two-dimensional slice through the object. By stacking equidistant slices, it is possible to obtain a continuous three-dimensional map of the density variations in the object. This technique was developed for medical diagnosis, but new CT instruments with significantly higher X-ray intensity and spatial resolution allow to image the interiors of geological samples. Here we present the first results of a study of an application of HRXCT to imaging the interior of impactites, especially suevite, impact glasses, and impact melt rocks and breccias. The aim of our study is to determine the three-dimensional distribution of clasts within the suevite matrix, and to obtain information on the different clast types (having different densities). We used a massive suevite sample (about 10 × 7 × 5 cm) from the 11 km diameter, 1.07 Ma old, Bosumtwi impact crater in Ghana. Target rocks are dominated by graywackes and sandstone/quartzitic rocks, with some shale, mica schist, and granite. Suevite occurs as large blocks of up to several meters width and as patchy massive deposits outside and mainly to the north of the crater rim. We used a Pantak tungsten high-energy (420 kV) X-ray source coupled with a high-energy P250D X-ray detector system, which is a solid-state linear array of 512 cadmium tungstate crystals. Each 500 μm-thick “slice” required about 2 minutes scanning time. Images were reconstructed from the raw data using a filtered back-projection algorithm. Macroscopic and thin section petrological studies are used to identify the clast types in the hand specimen and to cross-correlate them with the HRXCT observations. The results allow the easy discrimination of the relatively frothy inclusions of glassy melt in the suevites (see Fig. 1). These melt clasts appear darker than the matrix in the raw Xray scans and can be traced through the whole sample. Usage of color or gray scales allows the distinction of at least four different clast types based on density differences. The size of the smallest discernable clasts (about 0.5 mm) is determined by the resolution of the measurements, which, in turn, are a function of slice thickness. We are currently attempting to quantify the abundance of the various clast populations by image processing techniques. Acknowledgments: The measurements were performed at the University of Texas at Austin HRXCT facility. This research was supported by the Austrian FWF, project Y-58 (to C.K.).

Three‐dimensional quantitative textural analysis coupled with numerical modelling has been used to assess the dominant mechanisms governing crystallization of garnet porphyroblasts in rocks from diverse regional metamorphic environments. In every case, spatial dispositions, crystal size distributions, and compositional zoning patterns of porphyroblasts indicate the dominance of diffusion‐controlled nucleation and growth mechanisms.Nine samples from three geological areas were studied: a suite of semi‐pelitic rocks from the Picuris Mountains, New Mexico (USA); a suite of mafic samples from the Llano Uplift, Texas (USA); and a kyanite schist from Mica Dam, British Columbia (Canada). The semi‐pelitic suite exhibits post‐deformational garnet growth, whereas garnet in the mafic suite and in the kyanite schist grew synkinematically in rocks displaying weak and strong penetrative fabrics, respectively.For each sample, the centres and radii of thousands of garnet crystals were located and measured in three dimensions, using images produced by high‐resolution computed X‐ray tomography. Statistical measures of the degree of ordering and clustering of nucleation sites, and estimates of crystal isolation for each porphyroblast, were then computed from the measured spatial dispositions. These measures can be reproduced in simple numerical models only by diffusion‐controlled nucleation and growth mechanisms. Normalized radius‐rate relations computed from compositional zoning patterns in the garnets require thermally accelerated diffusion‐controlled growth, providing independent confirmation of the conclusions based on textural analysis. The unexpected similarity of results from all samples indicates that diffusion‐controlled nucleation and growth mechanisms may govern porphyroblast crystallization in many metamorphic regimes.

Macroscopic textures resulting from different atomic‐scale mechanisms for metamorphic crystallization display different degrees of order, clustering, intergrowth and relative isolation of porphyroblasts. Data on the sizes and locations of thousands of crystals in a three‐dimensional volume are required to identify reliably the mechanisms governing nucleation and growth of porphyroblasts from these textural features. These data can now be acquired by means of high‐resolution computed X‐ray tomography. Numerical models that simulate porphyroblast formation governed by either interface‐controlled or diffusion‐controlled reaction mechanisms indicate that quantitative textural analysis can discriminate between these possibilities. These numerical models also allow a comparison between textures predicted for different crystallization mechanisms and textures measured in natural samples, from which inferences can be drawn concerning the relative importance of these mechanisms in nature. An independent test of the validity of such inferences is possible for porphyroblasts such as garnet that may preserve prograde growth zoning and allow the examination of normalized radius–rate relations.

Quantitative textural analysis of garnetiferous rocks from diverse metamorphic environments has demonstrated that diffusionally influenced nucleation and thermally accelerated diffusion-controlled growth are commonly among the dominant controls on porphyroblast crystallization. A new numerical model of diffusion-controlled nucleation and growth that incorporates time as an explicit variable is capable of replicating the essential features of many of these natural porphyroblastic textures using geologically reasonable values for the kinetic parameters governing crystallization. This model has been used to interpret quantitative textural data from four rocks in which garnet grew during prograde regional metamorphism, by assuming constant heating rates and by accepting a previously published estimate for the activation energy for intergranular diffusion. From fits of the model to natural textures, estimates have been extracted for three previously undetermined kinetic quantities that control the nucleation and growth of garnet porphyroblasts, namely the activation energy for nucleation and the pre-exponential rate constants for nucleation and intergranular diffusion. The uncertainty in derived values for the pre-exponential rate constant for intergranular diffusion is probably not much larger than a factor of two because it is proportional to the uncertainty in prograde heating rates, given a defined temperature interval over which crystallization occurs. The uncertainty in estimates of the activation energy and pre-exponential rate constant for nucleation is larger, perhaps a factor of ten, because values for these quantities are derived from fits of the model to natural crystal size distributions; the precision of these estimates appears to be limited by the fact that the crystal size distributions are affected by factors not encompassed within the model (probably variable heating rates and inhomogeneity in the distribution of nutrients in the rock's precursor).

Quantitative three-dimensional analysis of rock textures is now possible with the use of high-resolution computed x-ray tomography. When applied to metamorphic rocks, this technique provides data on the sizes and positions of minerals that allow mechanisms of porphyroblast crystallization to be identified. Statistical analysis of the sizes and spatial disposition of thousands of garnet crystals in three regionally metamorphosed rocks with diverse mineralogies, in conjunction with simple numerical models for crystallization, reveals in all cases the dominance of crystallization mechanisms whose kinetics are governed by rates of intergranular diffusion of nutrients.