The Evolution of 3D Microimaging Techniques in Geosciences
In the analysis of geomaterials, it is essential to be able to analyze internal structures on a quantitative basis. Techniques have evolved from rough qualitative methods to highly accurate quantitative methods coupled with 3-D numerical analysis. The earliest primitive method for "seeing'" what was inside a rock was multiple sectioning to produce a series of image slices. This technique typically completely destroyed the sample being analyzed. Another destructive method was developed to give more detailed quantitative information by forming plastic casts of internal voids in sedimentary and volcanic rocks. For this, void were filled with plastic and the rock dissolved away with HF to reveal plastic casts of internal vesicles. Later, new approaches to stereology were developed to extract 3D information from 2D cross-sectional images. This has long been possible for spheres because the probability distribution for cutting a sphere along any small circle is known analytically (greatest probability is near the equator). However, large numbers of objects are required for statistical validity, and geomaterials are seldom spherical, so crystals, vesicles, and other inclusions would need a more sophisticated approach. Consequently, probability distributions were developed using numerical techniques for rectangular solids and various ellipsoids so that stereological techniques could be applied to these. The "holy grail" has always been to obtain 3D quantitative images non-destructively. A key method is Computed X-ray Tomography (CXT), in which attenuation of X-rays is recorded as a function of angular position in a cylindrical sample, providing a 2D "slice" of the interior. When a series of these "slices" is stacked (in increments equivalent with the resolution of the X-ray to make cubic voxels), a 3D image results with quantitative information regarding internal structure, particle/void volumes, nearest neighbors, coordination numbers, preferred orientations, etc. CXT can be done at three basic levels of resolution, with "normal" x-rays providing tens of microns resolution, synchrotron sources providing single to few microns, and emerging XuM techniques providing a practical 300 nm and theoretical 60 nm. The main challenges in CXT imaging have been in segmentation, which delineates material boundaries, and object recognition (registration), in which the individual objects within a material are identified. The former is critical in quantifying object volume, while the latter is essential for preventing the false appearance of individual objects as a continuous structure. Additional, new techniques are now being developed to enhance resolution and provide more detailed analysis without the complex infrastructure needed for CXT. One such method is Laser Scanning Confocal Microscopy, in which a laser is reflected from individual interior surfaces of a fluorescing material, providing a series of sharp images of internal slices with quantitative information available, just as in x-ray tomography, after "z-stacking" of planes of pixels. Another novel approach is the use of Stereo Scanning Electron Microscopy to create digital elevation models of 3D surficial features such as partial bubble margins on the surfaces of fine volcanic ash particles. As other novel techniques emerge, new opportunities will be presented to the geological research community to obtain ever more detailed and accurate information regarding the interior structure of geomaterials.
A Non-synchrotron Based Nanofocus X-ray Computed Tomography Technique for 3D Microanalysis of Geological Materials
During the last decade, X-ray Computed Tomography (CT) has progressed to higher resolution and faster reconstruction of 3D-volumes. Recent advances also allow for a three-dimensional view inside geological samples with submicron resolution. By utilizing nanofocus tube technology, current commercially available X- ray systems are pushing forward into application fields that were previously exclusive to large-scale synchrotron techniques. CT for geological purposes can lead to a new dimension of understanding of the distribution of rock properties. Our work describes the first commercially available 180 kV nanofocus CT system. We have examined several geological X-ray CT scans that were collected using samples up to 120 mm in diameter and weighing up to 1 kg with voxel-resolutions down to less than 500 nm (0.5 microns). We have studied such problems as resolving spatial distribution of pores, pore-connections and cementation. These are particularly well suited to CT. Moreover rock analysis with the aid of X-ray CT may lead to better analysis and prediction of well stimulation projects. For example, a plug can be scanned before and after being stimulated with acid. The possibility to visualize the whole plug volume in a non-destructive way and to use the same plug for further analysis is currently one of the most valuable features of this new type of rock analysis. Such techniques will be coming to the forefront as X-ray CT is applied to solving a greater variety of geological problems. Our work also describes both the hardware and software requirements for high-resolution non- synchrotron based CT.
Analysis of fracture aperture and roughness using multi-scale computed tomography and numerical modeling
Open and connected fractures, when present, dominate both fluid flow and solute transport in rock bodies. The transport properties of fracture networks are controlled by the aperture and roughness of the individual fractures. Precise measurement and meaningful characterization of these features is typically problematic, particularly in tandem. Furthermore, the empirical equations used to characterize the effect of surface roughness on fluid flow are derived from artificial configurations, and may not be suitable for natural systems. We have undertaken a multi-tiered study utilizing X-ray computed tomography combined with laboratory experiments and numerical modeling to characterize natural fractures and their flow properties from the dm to the µm scale. We have created and continue to enhance a series of calibrations and procedures to evaluate and maximize our ability to determine, in three dimensions, surface location and aperture in randomly- orientated fractures in heterogeneous natural materials such as granites. Our methods are geared to explicitly take account of the innate blurring of CT data through measurement and deconvolution of a point-spread function. These methods developed for fracture characterization apply to any measurement problem in which the feature being analyzed is small relative to the resolution of the scan data. Our experimental program encompasses both exposed fracture surfaces (skins) and paired surfaces within solid samples which are then subjected to flow testing. High-energy (420-450 kV) X-rays are required to image the larger specimens (>10 cm diameter) that are necessary to maintain a measurable hydraulic head gradient during flow testing. Nominal resolution is approximately 250 µm, but accuracy in locating air-rock interfaces is in the 25-50 µm range. Extracted ~2 cm sections of fracture skins were studied with 225 kV microfocal CT, with nominal resolution of 25 µm and surface-location accuracy of 5-10 µm. Finally, ~5 mm sections were extracted and analyzed with an Xradia microCT scanner, providing submicron resolution. Each step of increased resolution reveals details of roughness undetected in coarser imaging. The natural fractures we measure are used as input for computational fluid dynamics models of flow and solute transport. Although it is not clear what scale of roughness appreciably alters flow rates at the field scale, solute transport and contaminant sequestration may be significant at all of these length scales. Our data demonstrate that roughness is frequently not stationary with respect to the one-dimensional statistics typically used to describe it, even at small scales. Consequently, the up-scaling of confined measurements for field-scale network characterization is problematic. At the same time, numerical simulations reveal that slight differences in fracture characteristics, particularly variation in aperture, can have significant effects on flow and transport. Taken together, these findings indicate that considerable challenges lie ahead for obtaining a robust understanding of flow in fractured aquifers.
3D Observation and Evaluation of Induced Damaged Zones in Anisotropic Media
Rift zones, sites for dyke intrusion and supporting roof of a mine under tension, production enhancement and stimulation of hydrocarbon and geothermal reservoirs are good examples to study the nature of interaction of propagating tensile fractures with pre-existing preferably oriented sets of geological structures. Such interactions cause anisotropy in transport and mechanical properties in rocks, eventually affecting the post strength, post-frictional and hydro-geological properties of resultant rock masses. In present study 3D micro CT images are used to understand the relationship between the test crack path propagation and structure of caused damage zones, with microstructural fabric orientation in a granite tested for fracture toughness under mode I along specific directions. X-ray CT scanning images were obtained on a volume of Barre granite when the test cracks were forced to propagate parallel (case 1) and perpendicular (case 2) to the preferably oriented microstructural fabric. These images were binarized to mineral grains and induced cracks by means of a neighborhood-based standard deviation thresholding algorithm. 3D objects counter algorithm was applied on these images to identify and compare the physical properties such as crack induced porosity, induced crack density, generated total surface area and contribution of individual mineral grains within the damaged zones in case 1 and 2 scenarios. Results showed that measured induced crack porosity for case 1 is more than ten times than that of case 2 which further justifies the reason for the fracture toughness in case 1 being almost twice of that of case 2. The type and structure of damage zones in these cases were found geometrically different. This study, linked with observations in field scale helps in better understanding of fracture propagation and its application to stability of underground openings, control of rock fragmentation, prediction of transport properties with divers flow regimes and enhancement of oil and geothermal reservoirs during hydraulic fracturing.
Permeability of Vesicular Stromboli Basalt: lattice-Boltzmann Simulations and Laboratory Measurements
We investigate the permeabilities of Stromboli basaltic foams using lattice-Boltzmann (LB) simulations and laboratory measurements. The samples are vesicular Stromboli basaltic glasses experimentally produced by the degassing of melts with H2O or H2O and CO2, at 162 to 370 MPa and 1 atmosphere. We use the lattice-Boltzmann method to simulate flow through 3D synchrotron X-ray tomographic images of vesicular Stromboli basaltic glasses with porosities between 9.5% and 93.4%. To test the precision of the simulations, the lattice-size and resolution were varied systematically, demonstrating that permeabilities are not sensitive to these parameters at porosities over 65%. In order to evaluate the accuracy of the calculations, we use a gas permeameter to measure the permeability of the experimental products imaged by tomography. The fluid used in the measurement is air. The pressure difference is 10 to 105 Pa, and the flow rate is 10-4 to 10-1 cm3/s. Both our lattice-Boltzmann simulations and measurements provide a power-law permeability-porosity relationship. The permeabilities from lattice-Boltzmann simulations and experimental measurements are in the range 10-18 to 10-14 m2 below the percolation threshold (for spheres) of about 30%, and in the range 10-14 to 10-10 m2 above the percolation threshold. The variance between the LB results and the measured ones is within 1 order of magnitude. However, the permeabilities of vesicular basaltic glasses in our study are about 1 to 2 orders of magnitude higher than permeabilities of dacitic to rhyolitic volcanic rocks at porosities above the sphere percolation threshold. We attribute this difference to the higher bubble connectivity in our basaltic melts, estimated to be about 0.9 to 0.99. Our studies imply that factors affecting permeability are strongly dependent on bubble formation and growth processes. At low porosities, permeability is dominated by bubble connectivity between pairs of bubbles, which is affected by bubble size and distribution. Once a highly connected bubble foam is formed at porosities over 65%, bubble aperture diameters and multiple bubble coalescence dominate the permeability variations.
Opportunities for neutron radiography and tomography in the Earth Sciences
High penetration and large variation in linear attenuation coefficients for thermal neutrons in naturally occurring materials offer unique opportunities in the Earth Sciences to characterize the composition and textures of the geological materials in 3D at macro-, meso- to microscopic scales. Current applications of neutron radiography and computed tomography include non-destructive characterization of the distribution of hydrogen and carbon in silicate and carbonate rocks, quantification of textures in undeformed and deformed crystalline and porous rocks, and multi-phase flow in permeable rocks, sediments and soils. Real-time radiography can be used to monitor fluid flow and reaction progress under controlled laboratory conditions important in both low and high temperature studies of geological systems. Voxel resolution for conventional thermal neutron sources is typically 100-250 µm with opportunities to increase resolution to 10 µm or better. Improvements in spatial resolution, data processing and visualization, coupled with the higher neutron flux at new generation neutron sources will further enhance neutron radiography and computed tomography techniques in the future. Adapting experimental techniques developed at synchrotron x-ray facilities for the study material properties under the extreme pressure and temperature conditions will enable new investigations of materials under stress, carbon-neutral fuel cycles, and complex natural systems with neutrons as the probe.
Development of an ultra-high resolution neutron computed tomography system for the characterisation of geomaterials
The aim of this project is to further develop neutron computed tomography (NCT) as a non-destructive tool for the analysis of drill cores and other geomaterials. NCT highlights the presence of OH bonds (in minerals and/or the glass matrix), which are a strong constituent of magmatic rocks. During extrusion, magma evolution engenders the production of diverse microstructures and textures. Our preliminary work has brought us a great step further in distinguishing between the various petrographical properties of magmas. This initial calibration step has yielded invaluable information of the evolution of magmatic rocks. For example, it is possible to measure the crystal distribution; 3d geometry of crack propagation, and possibly to estimate the relative water content within the sample. The recognition of water present in a volcanic has a strong implication as to the evolution of the magma during ascent. We are particularly interested in estimating the extent of degassing as it is one of the key factors in determining the eruption style: i.e. diffusive or explosive. Magma rheology in terms flow and rupture is strongly influenced by the crystal content. For this reason, much experimental effort has been put into describing the complexities of bubble- and crystal-bearing rheology. To this end we hope to observe the evolution of flow bands through NCT. Preliminary work has revealed the rotation of large crystals during magma deformation experiments. It is hoped that future work will extend this characterisation to micro-crystals and pave the way to a numerical model of complex magma rheology. Failure is the ultimate criterion for explosive eruption. During this process, microscopic cracks grow and link together, promoting the formation of shear fractures. The geometry of crack propagation is very similar to fractures observed around lava domes and shallow conduits. We now propose to couple our unique experimental setup (including acoustic monitoring and a uniaxial press) with a sequence of NCT reconstructions to track in situ the evolution of crystals and cracks during lava deformation. The ultimate goal of this research is to observe the development of shear zones within the samples; study their seismogenic character, and thereby understand the processes involved. It is hoped that this will lead to a deterministic model of faliure, which in turn can lead to a prediction model of explosive eruptions. Such a model could allow for hazard mitigation on a timescale of seconds to minutes, potentially shutting down gas; power, road bridges and other vulnerable systems in the event of an explosive eruption.
3D Image Analysis of Geomaterials using Confocal Microscopy
Confocal microscopy is one of the most significant advances in optical microscopy of the last century. It is widely used in biological sciences but its application to geomaterials lingers due to a number of technical problems. Potentially the technique can perform non-invasive testing on a laser illuminated sample that fluoresces using a unique optical sectioning capability that rejects out-of-focus light reaching the confocal aperture. Fluorescence in geomaterials is commonly induced using epoxy doped with a fluorochrome that is impregnated into the sample to enable discrimination of various features such as void space or material boundaries. However, for many geomaterials, this method cannot be used because they do not naturally fluoresce and because epoxy cannot be impregnated into inaccessible parts of the sample due to lack of permeability. As a result, the confocal images of most geomaterials that have not been pre-processed with extensive sample preparation techniques are of poor quality and lack the necessary image and edge contrast necessary to apply any commonly used segmentation techniques to conduct any quantitative study of its features such as vesicularity, internal structure, etc. In our present work, we are developing a methodology to conduct a quantitative 3D analysis of images of geomaterials collected using a confocal microscope with minimal amount of prior sample preparation and no addition of fluorescence. Two sample geomaterials, a volcanic melt sample and a crystal chip containing fluid inclusions are used to assess the feasibility of the method. A step-by-step process of image analysis includes application of image filtration to enhance the edges or material interfaces and is based on two segmentation techniques: geodesic active contours and region competition. Both techniques have been applied extensively to the analysis of medical MRI images to segment anatomical structures. Preliminary analysis suggests that there is distortion in the shapes of the segmented vesicles, vapor bubbles, and void spaces due to the optical measurements, so corrective actions are being explored. This will establish a practical and reliable framework for an adaptive 3D image processing technique for the analysis of geomaterials using confocal microscopy.