Symposium CM2 : Quantitative Tomography for Materials Research

While x-ray tomography has been in use for some time, it has traditionally operated under a single absorption-based contrast mechanism. In this operating mode, an image is formed due to spatially varying density within the sample, resulting in varying attenuation of the incident x-ray beam. Reconstruction of the data yields a 3D map of sample density. However, this approach is unable to obtain crystallographic information since even a polycrystalline structure of a single phase exhibits uniform density. In this work, an alternative imaging solution is presented, termed laboratory-based diffraction contrast tomography (LabDCT). The LabDCT imaging modality is implemented on the ZEISS Xradia Versa laboratory X-ray microscope utilizing a polychromatic divergent beam. Rotation of the sample yields a series of diffraction patterns generated by the sample crystallites when the Bragg condition is satisfied. The patterns are then reconstructed to yield crystallographic information including grain orientation, center of mass, and size for a large number of grains. This is used to complement structural data obtained by traditional absorption-based tomography. In this work, results of LabDCT will be presented, with discussion of the results and comparison to alternative techniques including synchrotron-based DCT and EBSD. In particular, merits of the LabDCT method will be highlighted, particularly its non-destructive operation, demonstrated though 4D evolutionary studies obtained by repeating the imaging procedure numerous times on the same sample.

Nanostructured materials for electrochemical energy conversion caused a leap forward in the development of portable power sources, such as Li-ion batteries and fuel cells. In nano-crystalline ceramic/metal thin films over half of all the atoms participate in or are influenced by defects, i.e. the interface, grain boundaries (GB) and triple junctions (TJ). The presence of these defects significantly alters the thermodynamic free energy of the system, allowing for new physical properties that may not comply with expectations. Here we examine a new perspective on the exploitation of planar and line defects as catalytically active sites in energy systems. Microstructural and chemical information with atomic resolution from these sites is gained by a correlative approach combining atom probe tomography, aberration-corrected transmission electron microscopy and Rutherford backscattering spectrometry. By employing these techniques to a model system based on nanocrystalline platinum-zirconia multilayer thin films, we demonstrate that oxygen segregates to the interface and line defects in the nanocrystalline platinum films. These highly abundant oxygenrich line defects can act as fast oxygen transport paths with a diffusivity up to 18 times higher than the surrounding platinum grains. The unconventional transport properties of the material are related to topological disorder in the material defects. These defects show similar chemistry and therefore similar catalytic activity to the materials surface. The results open up the opportunity to design and produce simple scalable structures as catalysts, whose functionality derives from internal defects rather than from the materials surfaces.

Traditional x-ray computed tomography operates under an absorption based contrast mechanism making it well-suited for imaging of dense and high-Z materials. Various phase contrast techniques offer visualization of samples with low-Z materials and differentiation of materials with similar density. However, in these cases the resulting images and tomographic volumes are often prone to artifacts and do not represent the sample quantitatively, making subsequent segmentation and quantification difficult.

Here, we present an overview of phase contrast techniques available on a multi-length scale suite of commercial laboratory x-ray microscopes (ZEISS Xradia Versa and Ultra) and discuss implementations of novel developments to mitigate and remove artifacts in phase contrast tomography.

In spanning a resolution range from tens of micrometers down to 50 nanometers a variety of application cases will be considered that demonstrate the utilization of propagation phase contrast and Zernike phase contrast imaging, as well as ways to deal with the inherent artifacts through experimental methods and analysis techniques.

In order to rapidly quantify key attributes of grain ensembles in structural alloys, a novel 3D characterization instrument has been constructed that couples a SEM with EDS and EBSD detectors to a serial section mechanical polishing system (RoboMet.3D). Importantly, the collection of multi-modal data— crystallographic and spectral x-ray maps, as well as electron-optic and optical images—greatly improves the ability of software codes to segment key microstructural features such as grains compared to conventional image-based segmentation. The talk will discuss the salient features of the serial sectioning system, as well as present results from selected studies of Ni and Ti alloys.

The detection, identification, and quantification of surface and subsurface elements is an important capability in materials science. Often destructive methods such as FIB-SEM are used to locate and identify subsurface materials. Unfortunately, since this method is destructive, the material cannot be used for further measurements or experiments. X-ray-based techniques are preferred in that they are non-destructive. X-ray fluorescence (XRF) uses an X-ray source to illuminate the sample with a detector to measure the emitted X-rays and is useful for elementally specific identification and quantification. However, XRF instruments sold on the market are 2D in nature, that is, there is little to no depth discrimination.

Confocal micro X-ray fluorescence (confocal MXRF) is a 3D analytical technique used to measure the position and density of elemental constituents. This technique is useful in not only identifying the elements present, but also measuring the elemental content (atomic percentage) and the density variations in low density materials (~10 mg/cm³). Confocal MXRF uses a polycapillary optic to focus the X-rays from the X-ray source onto the sample. A second optic, coupled to the detector, transmits the X-rays from the focal spot so that only the X-rays emitted from the sample within the focus of the two optics is detected. Therefore, it is possible to collect 3D point-by-point MXRF data without spectral contamination by the material outside of the focal volume. Previously, we demonstrated that this technique, when coupled to nano X-ray tomography, is capable of quantifying the Ge dopant concentration in one micrometer thick, two atomic weight percentage thin films present within polymer capsules.

We have demonstrated this technology with an instrument initially designed to characterize radioactive materials, but have now constructed a second generation instrument designed specifically for transition metal elements. Design enhancements include: a higher wattage Mo tube; a vacuum chamber to better analyze low atomic number materials; improved coupling of the optics to the detector; higher spatial resolution; and full spectral data acquisition. One other unique capability is the mounting of the detector optic on a manipulator arm so that 2D XRF images can be rapidly collected to identify regions-of-interest followed by the rapid insertion of the detector optic to collect 3D XRF images. Our focal spot size (X ray wavelength dependent) is a spherical volume approximately 30 µm in size. Initial experiments demonstrating the capabilities of the instrument to analyze thin films along with multi-instrument measurements will highlight the non-destructive nature of this technique. Together these enhancements will reduce analysis time (>4X), increase workflow, and improve the sensitivity of the elemental analysis

Next-generation resistive RAM offers efficiency and storage density improvements over memories such as flash, utilising simple stacks of a thin dielectric layers sandwiched between a pair of conductive electrodes. Such devices rely on a phenomenon known as resistance switching, in which sub-breakdown electrical stress is used to reversibly switch the dielectric between states of resistance with a contrast of up to several orders of magnitude. For intrinsic switching materials, whose operation does not rely on metallic dopants, many models for the switching mechanism exist. Generally, these are based around the formation of filamentary conductive pathways through the dielectric. For silica and metal oxides this filament is believed to be a chain of oxygen vacancies that bridges the dielectric and shorts the electrodes. However, a comprehensive understanding of the nature of these filaments has not yet been developed. These filaments have sub-micron diameters, meaning that they are difficult to find and subsequently study. To further develop the knowledge that supports filament-based models we have performed tomography on filaments generated in silicon suboxide, a highly CMOS-compatible material. By performing conductive atomic force microscopy with the scanning tip pressed into the sample we were able to profile through the dielectric layer as material was gradually removed during imaging. Rendering our current map images into a 3D tomogram enabled for the first time a clear visualisation of these filaments in an intrinsic switching material, thus we have been able to discern their size and shape. These results confirm that filamentation is indeed responsible for switching and indicate that their structre conforms to the intrinsic nature of the dielectric. In addition, they suggest that such a tomography technique may be highly effective in mapping conductivity variations in three dimensions through other solid films.

The vast number of 3D characterization tools has enabled acquisition of a wealth of data across length scales. However, there exists a great challenge to efficiently and accurately interpret these data. This paper will present three-dimensional characterization of microstructural features across several structural materials, various acquisition techniques, and multiple length scales. The aim of most of these experiments extended beyond mere 3D visualization and their results have led to unique direct-3D quantification of microstructural features.
Robo-Met.3D™ (iterative mechanical polishing with subsequent optical imaging) has been employed to reconstruct equiaxed-α precipitates in [α+β]-processed Ti-64. Accurate slice alignment, along with novel 3D feature-find and separation algorithms, permitted robust three-dimensional quantification. Comparing such quantification to its stereological complement has improved the understanding of common stereological metrics, the validity of their ideal-shape assumptions for real microstructures, and their role in neural network property-predictive models.
Through DualBeam™ FIB/SEM, α-laths and β-ribs have been segmented from Ti-62222, revealing an α-lath morphology and β-rib connectivity much more complex than previously thought. In addition, a powerful pattern segmentation algorithm based on two-point correlation has been applied to a dataset of Ti-6246 for the rare 3D visualization of an α-colony/α-colony interface morphology. In a more quantitative effort, α-lath thicknesses have been direct-3D quantified and compared to a mean intercept-based thickness approximations (i.e., stereology). These comparisons helped calibrate such stereology, so that its future applications may be more informed.
Through STEM tomography, this paper will present characterization of nano-scale features in high-entropy alloys (HEAs) using tilt-series of HAADF and chemiSTEM™ images acquired from needle-shape specimens, which avoided project-thickness variations when tilted in a 360° Fischione tomography holder. In this work, two microstructures were investigated: one of AlMo_0.5NbTa_0.5TiZr, and one of CoCrNiFeCuAl_2.1. In the first, HAADF images at each tilt (0°-180° in steps of 1°) were sufficient to reconstruct and segment the complex nano-precipitates. A measure of the precipitates’ omega-2 moment invariants revealed distinct cuboid and plate-like morphologies: an observation which was not apparent from 2D microscopy. The second microstructure contained two precipitate phases (Cr-rich and Cu-rich) which were not resolvable from HAADF intensity and hence, chemiSTEM™ mapping was invoked at each tilt. The segmentation of these precipitates revealed each to exhibit an orthogonal platelet morphology. This morphology which was far less complex than expected for the Cr-rich phase, yet significantly different from the artificial cigar-shaped morphology of the Cu-rich precipitates implied from 2D HAADF and chemiSTEM™ imaging.

In recent years, much efforts have been placed on the development of novel techniques which allow deep 3-dimensional insights in morphology, chemistry and functionality of different materials from millimeters down to the atomic scale. In particular, electron microscopes have attracted enormous attention as they combine nanoscale resolution with chemical information. As example, transmission electron microscopy based tomography typically uses high-resolution imaging and electron diffraction together with energy dispersive X-ray spectroscopy (EDXS) and / or electron energy loss spectroscopy to get a deep material insights down to single atom detection within sufficiently small volumes. The drawback, however, is the fabrication which often requires detailed information about the region-of-interest upfront. On a slightly larger scale, dual beam microscopes have evolved into an essential 3D metrology tools as they use a focused ion beam (FIB) for slicing material volumes while accessing material information via EDXS or classical electron beam imaging from the meso- to the nanoscale. Although both electron microscopy based methods have undoubted advantages, increasingly relevant information such as mechanical properties or chemical phases are often inaccessible. However, atomic force microscopy (AFM) can provide such information due to the different principle of operation. So far, several integration concepts have been introduced to the market which combines AFM with electron microscopes and clearly expands their capabilities. However, to realize real 3D tomography with sufficiently high resolution in reasonable time, several key facts would be required which are widely absent in current concepts: 1) high-speed scanning capabilities to provide AFM imaging in the sub-minute range; and 2) a high-resolution tip-scanning concept where no additional stage adaptations are required. In this contribution we present a new approach which fulfill both requirements and provides an easy-to-use character as the optical detection system is entirely eliminated by the introduction of self-sensing cantilever technologies. To demonstrate the new capabilities, we show two application examples. First, we show a FIB based slice-and-view (subtractive) approach where surface imaging is done via AFM to gather 3-dimensionally resolved mechanical properties of soft matter material blends which are inaccessible via electron based methods. The second concept goes a step further and uses focused electron beam induced deposition of nanostructures where the high-speed AFM capabilities are used to enable for the first time additive in-situ (nano)tomography. This is not only highly relevant for fundamental aspects but also needed for rapid prototyping applications in dual beam microscopes. Finally, an outlook is provided which sheds light on current developments towards electric current and thermal measurements which leverages this concept to an entirely new level.

With the introduction of 3D devices and stackable architectures in both logic and memory applications, the physical characterization of 3D nano-sized volumes is becoming of paramount importance. Furthermore, for specific applications the characterization cannot be limited to the observation but it also has to incorporate the electrical features of the sample. Therefore the main requirements for a valuable 3D characterization technique are: (1) nanoscale spatial resolution (e.g. sub-10 nm observation capability). (2) Sensitivity (e.g. electrical and structural) and (3) nanometer accuracy in probing the third dimension. In this work we In this section we propose an approach, termed SPM tomography or scalpel SPM, which is based on extending the 2D-analysis capabilities of scanning probe microscopy (SPM) towards 3D, thereby creating a valuable technique for the electrical 3D characterization of ultra-confined volumes. In essence we combine the high lateral-resolution of conventional conductive atomic force microscopy (C-AFM) to in-house fabricated wear-resistant diamond-tips.[1] We leverage the peculiar behavior of wear at the nanoscale which is often described as a stress assisted chemical reaction to mimic an atom-by-atom material removal[2] in the continuous tip scanning against a solid surface. In essence, with scalpel SPM we slice in a controlled manner through the vertical dimension of our sample collecting C-AFM slices at different heights.[3] The collection of slices is than stacked and interpolated for the final 3D-visualization. We can achieve sub-nm material removal-rate which turns this technique into a powerful approach for the physic/electrical characterization of highly confined volumes. In this paper we are going to discuss its basic principles, its applications in different fields ranging from physical understanding of resistive switching up to failure analysis and reliability of different solid state semiconductor devices.
Ref.
[1] Hantschel, Tet al., Physica Status Solidi (a), 206, (2009), 2077–2081.
[2] Jacobs, T. D. B. and Carpick, R. W., Nature Nanotechnology, 8(2), (2013), 108–12.
[4] Celano, U., et al., Nano Letters, 14(5), (2014), 2401-2406.

In many fields of study, it is imperative to understand the behavior of a system across several length scales in two and/or three dimensions in order to properly interpret the parameters that govern its performance. As characterization techniques have progressed individually, a clear challenge that has emerged has been how to intelligently navigate to and acquire 3D volumes of interest (from centimeter to nanometer), and, subsequently, to fuse multi-scale and multi-modality datasets in a flexible and efficient, sample-centric manner that can also be studied, shared, and extended as further experiments, observations and analyses are undertaken.

We are on the cusp of microscopy technologies that can readily generate petapixels and teravoxels of data; acquiring it sensibly and making sense of what is acquired requires new tools and new approaches.

Here we present the development of a modern microscopy tool for streamlining correlative studies over a range of modalities, scales, and dimensions. ZEISS Atlas 5, a new software environment released in 2015, serves the dual role of an advanced, automated SEM and FIB-SEM acquisition system, as well as a correlative workspace for the combination of data sets acquired across multiple instruments (SEM, LM, FIB-SEM, XRM, etc.), serving as the common hub and interface between various experiments on a given sample. Atlas 5 empowers the user by providing a visualization environment to co-locate, calibrate, register and acquire multiple datasets, simplifying the correlation workflow as well as guiding future data acquisition at specific targeted regions and volumes of interest.

A numerical task of current interest is to compute the effective elastic properties of a random composite material by operating on a 3D digital image of its microstructure obtained via X-ray computed tomography (CT). The 3-D image is usually sub-sampled since an X-ray CT image is typically of order 1000³ voxels or larger, which is considered to be a very large finite element problem. Two main questions for the validity of any such study are then: can the sub-sample size be made sufficiently large to capture enough of the important details of the random microstructure so that the computed moduli can be thought of as accurate, and what boundary conditions should be chosen for these sub-samples? This paper contributes to the answer of both questions by studying a simulated X-ray CT cylindrical microstructure with three phases, cut from a random model system with known elastic properties. A new hybrid numerical method is introduced, which makes use of finite element solutions coupled with exact solutions for elastic moduli of square arrays of parallel cylindrical fibers. The new method allows, in principle, all of the microstructural data to be used when the X-ray CT image is in the form of a cylinder, which is often the case. The appendix describes a similar algorithm for spherical sub-samples, which may be of use when examining the mechanical properties of particles. Cubic sub-samples are also taken from this simulated X-ray CT structure to investigate the effect of two different kinds of boundary conditions: forced periodic and fixed displacements. It is found that using forced periodic displacements on the non-geometrically periodic cubic sub-samples always gave more accurate results than using fixed displacements, although with about the same precision. The larger the cubic sub-sample, the more accurate and precise was the elastic computation, and using the complete cylindrical sample with the new method gave still more accurate and precise results. Fortran 90 programs for the analytical solutions are made available on-line, along with the parallel finite element codes used.

Quantitative tomography opens new dimensions for material research. Beside additional view angles to see the research objects which improve our knowledge about materials structure and their properties, new methods with multi-dimensional analytics based on quantitative tomography aim at specified problems of non-uniform change in volume. It will provide new approaches to answer various questions in different research field and become available analytics for common use. The new methods we are developing focus on typical problems related to localized deformation at crack tip for toughness evaluation and concentration variation around buried interface for diffusion bonding formation.
X-ray tomography (XRT) is an effective methodology to reveal complicate 3D structure around the crack tip. In Principle, fracture toughness (K_IC) is related to plastic zone size in Linear Elastic Fracture Mechanics. The estimate of K_IC will be possible when the plastic zone size and fracture mode can be measured accurately. Multi-dimensional analytics of XRT can provide valuable information for the toughness evaluation, like the size of plastic zone, the area of crack surface etc., which contribute to the fracture toughness. A stainless steel with nanostructured grains was used to check the fracture toughness resulted from the new evaluation method.
Combining 2D imaging in lateral with sputtering in depth, layered Secondary Ion Mass Spectrometry (SIMS) imaging can be reconstructed for the distribution of chemical composition in 3D view. Normal approach in 3D-SIMS can display the species in layered or complex structure. However, detailed analysis of interfaces between phases, such as concentration change and inhomogeneity of diffused species near interface, attract more attention that need to be explored. For example, diffused interface is expected to be a sign of solid bonding for the safety of coating service life. New 3D imaging approach is developed to separate the diffused species through the interface. In terms of the ion imaging of CsAl+ in diffusion region, diffused Al at the interface between the ceramic coating and Zircaloy substrate were investigated in both as-deposited and annealed states. The new analytics toolkit to resolve diffused species is a key evaluation to develop coating layer on alloys. The new method based on 3D-SIMS analytics is expected to provide diffusion information between coating and substrate with new view. It will continue to advance and gives more information about the formation of diffusion interface which facilitate the interface analysis from variety of research interest.

X-ray computed tomography (CT) of materials provides large 3D image data sets (i.e., tomograms), resolving both surface and subsurface features. Tomograms of open-cell polymer foams typically reveal a two-phase material consisting of the supporting polymer material and the void structure. Segmenting the tomograms for the void structure allows for various 3D measurements using image processing software. One common void measurement is Equivalent Diameter, which is the calculated diameter assuming the void volume constitutes a perfect sphere. However, for stochastic and irregular void structures, this singular metric is insufficient when comparing two or more polymer foam samples. Multiple descriptors are typically required, though more than three measurements can lead to difficulty in interpretation. Therefore, a statistically-based pattern recognition technique, termed Principal Components Analysis (PCA), has been implemented to aid in the interpretation of multivariate tomogram data sets of polymer foam systems. PCA transforms N-dimensional data into a reduced number of dimensions which capture most of the data’s variance. PCA is commonly used for pattern recognition in experimental sciences and enables easy visualization of sample groupings based on several descriptors.
30 polymer foam samples were used for this preliminary, proof-of-concept study. Three samples each were punched from five separate locations of two separate larger parts (one larger part has been used; one larger part has been unused). Using image processing software, 11 separate 3D void shape descriptors were calculated for each sample. The corresponding averages and standard deviations of these 3D void shape descriptors (resulting in a 30 x 22 matrix) were used as input into PCA, which then reduced the number of variables from 22 to 2. These two new variables together represent over 75% of the variance in the 22 initial 3D void shape descriptors. Results are promising in that the PCA model adequately differentiates the polymer foam samples based on sample location, part origination, and part use.
A similar analysis has been conducted on 2 different open-cell polymer foam systems using 4D X-ray CT data of the foams undergoing dynamic compression acquired at Argonne National Laboratory’s Advanced Photon Source. Results from the PCA of these 4D tomogram data sets show that stress-strain states for both foams can be differentiated based solely on these 3D void shape descriptors. PCA also reveals which 3D void shape descriptors dominate each stress-strain state.

The final properties of materials composed by a matrix and a filler are highly dependent on the microstructure of the composite: the shape, dimensions, orientation and distribution of the reinforcing element determine, to a large extent, the final mechanical properties. For characterizing the microstructure of highly regular reinforcing materials it is often enough to perform a simple 2D analysis, but in many cases the complexity of the microstructure requires 3D characterization techniques.
The material used for this work is formed by a polymeric matrix (Poly Lactic Acid, PLA) reinforced by laminar nanoparticles (clay particles). The clay platelets can be found within the PLA matrix as individual particles or forming stacks of several clay layers. Moreover, the distribution and orientation of the clays and clay stacks will vary from one sample to another, according to several factors (all of which could be generally described as composition and processing conditions dependent). Traditionally, the microstructure of this kind of sample has been analyzed by x-ray diffraction and/or conventional (2D) transmission electron microscopy (TEM), which imply certain limitations in the obtained data. TEM tomography (TEMT) allows performing a 3D characterization of this kind of materials at the nanoscale, providing a more complete insight into the microstructure.
The mechanical behavior of composite materials can be studied by Finite Element (FE) simulations. In order to perform accurate simulations, models that describe acceptably the microstructure and — as a consequence — the mechanical behavior of the studied material are required. The aim of the present work is to generate such models based on the 2D and 3D microstructural characterization of the studied materials.
3D models of the studied material were generated following several routes, based on conventional (2D) TEM images or based on segmented TEM tomograms. For the 2D characterization based models, a set of images was quantitatively analyzed by Automated Image Analysis (AIA) and several morphological descriptors were obtained, such as dimensions, orientation, degree of dispersion (distance to the nearest neighbors), etc. In addition, the average particle thickness was obtained from x-ray diffraction analysis. Several models of increasing complexity were then generated starting from a model in which the clay platelets were perfectly aligned and ordered up to a model in which the platelets were dispersed and oriented according to the experimental data.
The TEMT based models were generated using the morphological data obtained from segmented tomograms, which avoids many of the limitations that are inherent to 2D characterizations of a complex 3D microstructure. Here again the models were generated with varying degree of complexity and fidelity with the original data.
Finally the FE simulated mechanical properties of the generated models were compared in an attempt to assess the adequacy of each model.

It is well known that the local microstructures of the electrodes greatly affect not only the initial performance but also the long time durability of solid oxide fuel cells (SOFCs). Nickel (Ni) agglomeration due to sintering is one of the major degradation mechanisms of the SOFC anode. In the present study, a phase field method is used to simulate the microstructural evolution of nickel - yttria stabilized zirconia (Ni - YSZ) composite anode. With the help of three dimensional microstructures obtained by focused ion beam-scanning-electron-microscopy (FIB-SEM), the numerical simulation results such as three-phase boundary (TPB) density, Ni surface area and tortuosity factors are quantitatively validated. On the other hand, Ni – YSZ anode is normally prepared by reducing NiO to Ni before initial operation. Therefore, reduction process has a great impact on anode microstructure which strongly affects both initial performance and durability of the SOFC anode as described above. In the present study, the reduction process of sintered NiO - YSZ composite to Ni – YSZ anode is also simulated based on the phase field method. Both reduction reaction and sintering mechanisms are considered in the model. Simulation results are compared to the experimental anode microstructures with different reduction temperatures. The effectiveness of the present approach for investigating the temporal change in SOFC anode microstructure is discussed.

Developments in image acquisition using modalities such as X-ray tomography and FIB-SEM are opening up opportunities for high-quality modeling and characterization of material samples. 3D models created from scans provide the basis for analysis of material properties based on realistic geometries, as well as Finite Element and Computational Fluid Dynamics workflows. Software techniques developed at Simpleware Ltd. (Exeter, UK) are making it increasingly easier to go from 3D scans to simulation-ready models suitable for physics-based solvers.

This presentation will outline the challenges and benefits inherent to image-based materials modeling and characterization, and will describe different software techniques for image processing, reconstruction of 3D microstructure and the calculation of effective material properties using homogenization methods.

Discussion will be made of the steps to create accurate 3D reconstructions from image data, including visualization, segmentation and processing of images when working with complex materials samples and microstructure. Quantitative techniques for obtaining measurements and statistics, such as surface area and porosity, will also be examined in the context of working with image data. The benefits of an image-based approach can be extended here to meshing from images to ensure realistic geometries.

Unique meshing techniques that can handle multiple phases and guaranteed conforming interfaces and shared nodes will be described, with applications to rapid modeling of microstructure with complex topologies. With comparison to other approaches, this will outline smoothing and re-meshing techniques that preserve volume and topology when creating computational models for simulation.

The benefits of these techniques can be extended to the calculation of effective properties from scanned samples; by using Finite Element-based homogenization methods, it is possible to accurately obtain effective properties such as effective stiffness and absolute permeability. Using FE mesh surfaces for a given resolution offers increased accuracy over voxel mesh surfaces in terms of preserving segmented domains when decimating a mesh.

Using image-based modeling techniques for materials characterization has significant potential for increasing accuracy and flexibility when working with complex image datasets. Image-based meshing and homogenization particularly offers a fast and reliable way for materials researchers to obtain high-quality computational models at different length scales, enabling studies of microstructure for composites, rock physics, and any other materials application relying on 3D images.

To compute a physical or geometrical quantity for a random particle, one needs to mathematically know the 3D shape of the particle. For regular particles like spheres and ellipsoids, the mathematics are straightforward - for random particles, with realistic shapes, mathematically characterizing the shape is a harder task. Since about the year 2000, a method has been developed that operates spherical harmonic analysis on X-ray computed tomographic images of particles, resulting in analytical, differentiable mathematical functions for the three-dimensional shape of real particles. Any physical or geometric quantity, for the particle volume or surface, can now be computed with this analysis. The method is briefly described, along with several applications.

3D characterization is of particular importance in the study of mineralized tissues such as teeth and bones due to the presence channels, pores and features that span millimeter, micrometer and nanometer length scales. The major component in human teeth, by weight and volume, is dentin. This hydrated hard tissue encloses the central pulp and has microscopic channels, dentinal tubules that radiate from the pulp to the cementum on the surface of the dentin that connects with the hard outer enamel. The permeability provided by these tubules can cause dental hypersensitivity. The object of our current work is to understand how treatments for dental hypersensitivity act on these tubules. This requires understanding of chemical changes on the nanometer scale using techniques such as STEM combined with analytical methods such as EDX and EELS. In turn, this requires preparation of suitable TEM specimens from site-specific regions with specific orientation using dual beam FIB-SEM methods. In order to reliably and reproducibly identify the specific region and orientation for the FIB-SEM studies we utilize 3-D datasets using XMT methods from dentin specimens with suitable fiducial markers. Ultimately, the 3-D correlative workflow developed will be used in clinical trials to evaluate both qualitatively and quantitatively the effectiveness of interventions on dental hypersensitivity from the atomic to the macroscopic scale.

Since the adoption of X-ray microscopy and tomographic imaging at synchrotron beamlines, continuous improvements in both spatial and temporal resolution have pushed the boundaries of nondestructive 3D imaging. Rapid scanning, affording the potential to collect 3D data sets in the order of seconds, has opened the door to a wealth of new experiments to study rapidly evolving microstructures in their native environment or under stimulus. In addition, as the frequent development center of advanced X-ray techniques, synchrotron imaging beamlines have also offered an increasing variety of imaging modalities (absorption contrast, phase contrast, diffraction contrast, etc).

In recent years, a number of these synchrotron developments have been transferred to analogous laboratory-based instruments, in many cases offering comparable capabilities. For example, the latest lab X-ray microscopes incorporate optical elements to achieve resolution and contrast comparable to many synchrotron experiments, and have even adopted increasing numbers of imaging modalities. In addition, and similarly to the synchrotron community, laboratory X-ray tomography systems have leveraged the nondestructive nature of the technique to foster increasing development of various types of in situ and 4D imaging experiments, albeit at a different time scale than at the synchrotron.

This work will present an evolution of laboratory based X-ray microscopy, with particular focus on the development of in situ sample manipulation and imaging methods. Discussion will include the contrasting capabilities of synchrotron-based systems, which naturally excel when experimental designs require highly dynamic imaging to capture rapid microstructural change, and laboratory systems offering a static or ‘interrupted’ in situ capability. The value of laboratory systems, even in the absence of rapid throughput, will be illustrated by way of several examples from recent work in the materials sciences.

Nanoscale X-ray computed tomography (nano-CT) is a non-destructive imaging method that enables in-operando studies of materials undergoing physical stresses and chemical reactions at sub-micron resolution. The challenge of using non-synchrotron X-ray sources is achieving a sufficient temporal resolution despite exposure times on the order of 1 minute. Current 4D (space+time) reconstruction methods are geared towards systems with exposure times two to four orders of magnitude lower. We propose a 4D dynamic CT method that reconstructs a temporal 3D state every 2 to 16 projections. The first novel addition of this algorithm is a projection angle sequence that features largest possible differences and an even span of 180 degrees. In addition, we employ a central projection sampling, which reconstructs a state from a projection window which also includes its neighbors' projections. Finally, we devise a probabilistic model for the sample change between neighboring states. A model that mimics the sample deformation can allow us to correct for the sample deformation within this projection window and to further improve the reconstruction quality. We characterize this dynamic CT method by generating projections from multiple phantom data sets that are consistent with the reconstructed images. One possible future application of this algorithm is the imaging of Li dendrite growth in Li-ion batteries.

The field of materials science and engineering (MSE) is based on the fundamental principle that microstructure controls properties. Traditionally, the study of material structure has been limited by sectioning and post mortem observations. This approach is often inaccurate or inadequate for solving many fundamental problems. It is also often laborious and time-consuming. Advances in experimental methods, analytical techniques, and computational approaches, have now enabled the development of in situ techniques that allow us to probe the behavior of materials in real-time. The study of microstructures under an external stimulus (e.g., stress, temperature, environment) as a function of time is particularly exciting. Examples include an understanding of time-dependent deformation structures, phase transformations, compositional evolution, magnetic domains, etc.

X-ray synchrotron tomography provides a wonderful means of characterization damage in materials non-destructively. In this talk, I will describe experiments and simulations that address the critical link between microstructure and deformation behavior of metallic materials, by using a three-dimensional (3D) virtual microstructure obtained by x-ray synchrotron tomography. The approach involves capturing the microstructure by novel and sophisticated in situ testing in an x-ray synchrotron, followed by x-ray tomography and image analysis, and 3D reconstruction of the microstructure. Case studies on fundamental deformation phenomena in aluminum alloys under cyclic loading and in a corrosive environment will be presented and discussed. New opportunities for x-ray microtomography, including lab-scale tomography and the next generation of x-ray synchrotron tomography will be highlighted.

Near real-time visualization of complex two-phase flow in porous media is demonstrated with dynamic 4D (3D + time) imaging at 2-BM of Advanced Photon Source (APS), Argonne National Laboratory. The advancing fluid front through tortuous flow paths and its interaction with wetting silica sand particles are clearly captured. Formations bubbles and capillary bridges are visualized. The experiment is carried out to improve understanding of the physics and mechanics associated with two-phase flow, and possible integration with direct numerical simulation is explored. An advanced synchrotron facility provides an intense X-ray photon flux that is several orders of magnitude higher than that of a typical laboratory source. Therefore, synchrotron 4D imaging can possibly capture dynamic evolution of solid and fluid phases. In situ fluid flow experiments through Ottawa sand (dia. = 0.425 ~ 0.85 mm) columns were conducted by using an experimental setup that employs a piezoelectric pump and allows continuous stage rotation. Complete tomograms are collected on two time scales with different spatial resolution: 12sec with spatial resolution of 3.25μm/pixel and in 2sec with spatial resolution of 5.5μm/pixel. A 25μm-thick scintillator (LuAG:Ce) was used for this work. Darter, a HPC machine at National Institute of Computational Science (NICS), was used to reconstruct the data. A change in flow regime (from capillary fingering to stable displacement) was observed with an increase in the flow rate for a similar initial microstructure. The dynamic imaging results will improve the fundamental understanding of multi-phase flow through porous media to isolate the role of fluid rheology, pore and particle morphology, relative proportion of wetting and non-wetting phases, and multi-scale pore network connectivity and its evolution. The obtained results provide motivations for evaluating a numerical simulation code such as Lattice-Boltzmann to model multi-phase flow through porous media. An emerging topic of evaluating Additive Manufacturing (AM) parts using X-ray CT (XCT) will also be introduced. AM is a revolutionary manufacturing technique for creating a complex geometry in a near-net shape configuration. Metal-based AM is now widely used for various applications including aerospace and biomedical implants, and the usage is expected to increase. XCT is found to be a promising technique to qualify AM parts with such complex geometries. An advanced laboratory XCT system (Zeiss Versa 500) was used to characterize AM parts (CoCr), and preliminary analyses related to determining porosity, pore size distribution, pore shape, and elastic modulus will be discussed. This type of research is expected to improve understanding of AM processing-microstructure-property relationship.

In recent years there has been a significant increase in carrying out temporal studies, so-called 4D imaging. This has enabled a very wide range of temporal studies to be carried out from the pupation of butterflies to the degradation of high strength steels. Samples can be monitored continuously in the X-ray scanner or intermittently. In this talk I will discuss the current state of the art focusing on quantitative aspects looking at materials behaviour under a very wide range of environments including tissue regeneration, fatigue, corrosion, self-healing and powder metallurgy stressing the power of correlative tomography.

3D printed materials offer a wealth of possibilities in that unique structures may be printed with geometries that are not possible with traditional molding, extrusion, casting, or machining techniques. Structures may be optimized for weight, strength, or form, to improve their overall function. 3D printed structures are created through the serial laydown of materials, layer-by-layer. Because of this, inherent discontinuities within the microstructure reduce the overall mechanical strength of a part when compared to traditionally formed materials. Interfacial adhesion through polymer chain entanglement is minimized. Mechanical testing indicates that print orientation, as well as the use of recycled print material, can all affect the ultimate mechanical performance. Due to these problems, there are very few demonstrated high performance applications of 3D printed materials, especially polymers. In order to understand the adhesion between the layers and therefore crack initiation, propagation, and ultimately failure, in situ analysis techniques are needed. To further complicate the analysis, these materials are typically hyper-elastic in nature. As such, experiments cannot be paused during data collection, as the material will continue to respond to the applied stress.
Experiments at Argonne National Laboratory’s Advanced Photon Source collected 3D tomographic data during the uniaxial loading of 3D printed tensile specimens. In order to capture the 3D images during loading, high speed radiographs were collected while the sample and load stage were rotated at 2 Hz. Using the fastest continuous 3D imaging yet demonstrated, 901 radiographs were collected in 0.25 s, so that four full 3D images were collected within one second. Eighteen thousand (18,000) radiographs were collected during this dynamic event to collect up to ~20 tomograms succession. Dynamic stretching, cracking, failure, and elastic recovery were imaged in these hyper-elastic materials.
The micro-specimens of glass bead filled nylon (EOS material 3200 GF) were printed on an EOS Formiga printer. This printer uses a powder bed of material that is laser sintered to print the tensile specimens. Specimens were printed in each of the three orthogonal orientations. Additionally, specimens were printed with virgin material, 50% virgin-50% recycled, and 50% virgin-50% doubly recycled material to understand the effect of the thermal history during recycling upon layer adhesion and mechanical performance. 3D data of the reconstructed slices through a specimen shows the delamination of the glass microbeads from the nylon, the elongation of the nylon, breakage, and the elastic recovery. Some blurring of the image is seen due to the high rate of the elastic recovery. Full 3D renderings of the materials show the movement of the glass particles. Due to the elastic nature of the material, stretching is seen in the stress-strain curve and snap-back is seen visually in the images.

X-ray tomography, as a non-destructive 3D imaging technique, has opened up the possibility to follow material evolution in 3D during, for example, mechanical or electro-chemical process. Such 3D+time, i.e., 4D, imaging is becoming increasingly used in various areas of research. Two key components are required in such studies, first an experimental set-up adapted for experiments in-situ in an x-ray tomograph and, second, 4D image processing to extract meaningful information from the time-lapse 3D image sequences. 4D image processing can involve multiple 3D image analyses to characterize certain properties or structures in each image or they can be full 4D analyses of changes from one image to the next. The former could be, for example, determination of porosity at each step to follow its evolution. The latter could be displacement and strain field analyses by Digital Volume Correlation (DVC; also known as 3D Volumetric Digital Image Correlation).

The possibilities for the 4D characterization of material evolution using 4D x-ray tomography and 4D image analysis, including DVC, will be discussed and illustrated using different examples, including one involving the deformation of granular material and another considering the discharge and failure of Si-Li battery cells. In the former, the characterization of structural evolution, including of porosity, grain-contacts and deformation, in terms of continuum strain fields and individual grain kinematics, will be described. For the Si-Li battery cell study, the evolution of the lithiation of the silicon electrode particles during cell discharge is quantified through the change in image intensity, which is correlated with DVC-derived volumetric strain measurements.

Textile ceramic composites represent the enabling materials for several major ultrahigh-temperature structural applications, specifically advanced gas-turbine engines and for leading edges and contact surfaces for future hypersonic flight vehicles. Extreme service conditions of temperatures from 1200°C to potentially well above 1700°C in combination with service loads and hostile environmental atmospheres are well beyond the realm of current commercial structural materials. Using novel strategies in materials and coatings along with integral 3-D architectural design, woven ceramic-matrix composites (CMC’s), such as SiC-SiC materials, make the development of such ultrahigh-temperature structures a feasible proposition. Lifetime prediction and damage assessment for such complex architectures presents a formidable challenge though, as the collection of reliable physical and engineering mechanical data, not to mention the characterization of damage in 3-D, at temperatures above 1200°C, is so difficult. To this end, we have developed a facility to perform such characterization using synchrotron x-ray micro-tomography capable of subjecting samples to tensile loads at temperatures of 2300°C, which has proven to be a tool of choice for non-destructively evaluating the component in three-dimensions at spatial resolutions around a micron. We report on several phenomena governing failure that occur over time in a model SiC/SiC fiber-matrix composite at 1500 to 1800°C ranging from cracking within individual fibers, fracture of entire fiber bundles/tows, and environmental degradation of the BN fiber coatings at these extreme temperatures. This ability to image complex 3D materials undergoing failure under combined extreme physical conditions, opens new possibilities for evaluating ceramic textile materials in real time. Indeed, the technique can be extended in studying not just textile composites but a myriad of new structural materials.

The development and optimization of high performance materials for aerospace applications requires a detailed understanding their microstructural evolution as they approach and exceed their operational limits. 3D imaging in realistic and extreme environments elucidates the details of the failure mechanisms and provides a wealth of data that can be used to validate and refine computational models. Validation of such models is necessary to enable rapid, computationally-based design and optimization of new classes of high performance materials. Imaging of material microstructure and failure mechanisms is achieved by x-ray micro-tomography (μCT) carried out at the Advanced Light Source (ALS). The synchrotron source at ALS provides high x-ray flux necessary to capture material micro-scale failure in near-real time. Custom in-situ cells enable μCT while simultaneously providing high temperature, mechanical loads, and controlled atmospheric conditions representative of realistic extreme environments. Novel intelligent algorithms provide for rapid segmentation and analysis of the μCT data. The failure mechanics of high strength ceramic matrix composites at high temperature (~2000C) is described and compared to existing damage models. The decomposition of spacecraft heat shield materials upon atmospheric entry conditions is imaged and used to validate micro-scale reactive ablation models.

Synchrotron-based x-ray microtomography is being used to study the temperature-dependent microstructural evolution of HMX-based polymer-bound explosives. HMX, a powerful but relatively insensitive high explosive, undergoes a solid-solid β-δ phase transition around 160-180° C. The explosive is much more sensitive to shock or thermal ignition at these elevated temperatures; how the δ phase itself vs. the extensive microstructural damage that occurs with the phase transition play into this increased sensitivity is an intense area of research in need of experimental results. The mechanisms and evolution of microstructural damage, particularly within the bulk, such as whether cracks primarily occur at HMX-binder interfaces or within HMX crystallites, how existing defects play a role in crack initiation and growth, and how HMX crystallite morphology changes or evolves through the transition are still unknown. To resolve these issues, we have been performing synchrotron-based microtomography on HMX-based materials LX-10 and PBX-9501 as they are ramped through the β-δ phase transition. We are using a newly developed, state-of-the-art reconstruction software package called LTT to reconstruct the volume as a function of time/temperature to determine how the microstructure evolves within the bulk as the phase transition occurs at elevated temperature. This work performed under the auspices of the U.S. DOE by LLNL under Contract DE-AC52-07NA27344. This work is funded in part by the Joint DoD-DOE Munitions Program.

Synchrotron x-ray nano-tomography and fluorescence imaging provide unprecedented morphological and chemical information, revealing multi-dimensional details of the interplay between structure and chemistry. These capabilities are particularly powerful when used to study complex structures with morphological and chemical heterogeneity, such as nanoporous materials for energy applications. Nanoporous materials--e.g. nanoporous stainless steel and silicon, fabricated by a recently invented metallic-melt dealloying method-- have great potential applications in energy conversion and storage materials. Their unique properties — including tunable pore size and porosity, high specific surface area, bi-continuous conductivity, and catalytic properties — are of great interest.

To analyze the compositional variation of nanoporous materials, we utilized the scanning fluorescence microscope at the newly commissioned National Synchrotron Light Source II (NSLSII). The 3D morphology of these materials was quantified using full-field x-ray nano-tomography, available at Advanced Photon Source. The process-structure-property correlation of these nanoporous materials will be discussed, in conjunction with show-casing some of the latest fluorescence microscopy results from the new Sub-micron Resolution X-ray Spectroscopy (SRX) beamline of NSLSII.

Lithium ion batteries (LIB) are widely used as rechargeable power sources in the field of portable consumer electronics. Other application fields for LIBs like electric mobility and stationary energy storage are promising and currently subject of strong research. At IMFAA, we focus in the development of new or the improvement of already existing constituent materials of LIBs aiming to improve the battery’s performance in these areas. To accomplish this task different microscopy techniques need to be combined to obtain structural, compositional and crystallographic information across different relevant length scales. In many cases, 2D microscopy studies from mechanically prepared cross sections are not sufficient for a full characterization of the LIB.

As an example, in this talk, we will present multi-scale 3D tomography work on the LiMn₂O₄ cathode material of a commercial 18650 battery. Post-mortem X-ray microscopy (XRM) reveals interesting features in the cathode material, which motivate a more detailed analysis at higher resolutions using FIB-SEM microscopy. In our case, features of interest are rare and located at random locations well-below the sample surface. Therefore, in a correlative approach the XRM data set was used as a reference to – after triangulation – facilitate FIB site specific cross sectioning and tomography.

A natural extension of such 3D studies into more dimensions happens when characterizing battery aging phenomena, possibly under different operating environments. Batches of batteries were subjected to numerous charging-discharging cycles and studied at different stages of this process. Observed changes in the microstructure of the LIB materials can be correlated directly to important parameters like maximum storable energy, remaining life time, and self-discharge rates making targeted and cost-efficient material design possible.

Electrode performance of lithium ion battery cathodes is strongly affected by its microstructure parameters. FIB (focused ion beam) / SEM (scanning electron microscopy) tomography is a powerful tool for determining microstructure parameters. The application of this method and processing of images has been optimized for battery electrodes [1], enabling an accurate detection of the three components (1) carbon black, (2) pore phase and (3) active material. A special challenge is to distinguish between different active materials which are combined in blend electrodes for optimizing their performance and energy density.
In this study, a dual beam modus is applied for analyzing the composition and microstructure of a blended battery cathode consisting of three active materials (NMC, NCA and LMO). By combining the information of InLens and Everhart-Thornley detector it was possible to reconstruct a volume of 35µm x 37µm x 20µm consisting of 1010 x 1070 x 570 voxel (isotropic voxelsize of 35 nm). The pore phase and the carbon black were segmented by evaluating images acquired using the Everhart-Thornley detector, due to lower falsification by local charging, whereas the active material was separated into its three components interpreting information of InLens detector which provides a better contrast of specific material. It was possible to clearly assign the three materials, confirmed by EDX analysis, and to calculate microstructure parameters for every specific component, as particle size distribution, active surface area, porosity and tortuosity.
Limiting factor for an application of this method is the representative volume element size which is dependent of the number of active materials as well as their maximum particle size, as reported in [2] for solid oxide fuel cells. For prospective studies, an increased reconstruction volume is necessary while maintaining high resolution of images, which is challenging with regard to FIB-procedure as well as processing of big data amount. For this purpose, solution approaches are presented and discussed.

[1] M. Ender, J. Joos, T. Carraro and E. Ivers-Tiffee, Journal of the Electrochemical Society, 159, A972-A980 (2012)
[2] J. Joos, M. Ender, T. Carraro, A. Weber and E. Ivers-Tiffée, Electrochemica Acta, 82, 268-276 (2012)

Improving the performance of porous electrodes for fuel cells and batteries is a crucial component of achieving viable energy storage and conversion systems for sustainable energy. This talk will present imaging, modeling, and experimental techniques for characterizing electrodes and identifying opportunities for better performance. Nano-scale resolution X-ray computed tomography (nano-CT) is a non-destructive method of obtaining high resolution 3D images of the complex, internal structure of materials and devices. A high resolution of 50 nm (16 nm 3D voxels) is achieved using advanced X-ray optics. With the use of the Zernike phase contrast mode, the high resolution is even achieved on materials with low atomic numbers; i.e. materials with low absorption contrast, such as the organic materials and polymers used in fuel cell and battery fabrication. This talk will highlight the application of nano-CT in our ongoing fuel cell and battery research. The work includes morphological characterization, degradation analysis, and chemical mapping of fuel cell and battery electrodes and the development of in-operando imaging. These studies include the characterization of emerging non-precious metal catalyst electrodes for polymer electrolyte fuel cells that can reduce electrode costs by two orders of magnitude. Nano-CT is used to resolve the hierarchical electrode structure as well as chemically map the polymer electrolyte additive. Morphological and simulated transport properties from nano-CT images are implemented in cell-level performance models for optimizing future electrode designs for these promising low cost catalysts. For batteries, our recent imaging and analysis of Li-ion graphite anode degradation will be covered as well as the development of cells and high temporal resolution 4D reconstruction algorithms for in-operando imaging of alloying lithium anodes in a lab-scale nano-CT.

A limiting factor for the performance of SOFC anodes is the electrooxidation kinetics of hydrogen in the nickel/yttrium-stabilized-zirconia (Ni/YSZ) anode. Herein, microstructure and material composition at the active triple phase boundary sites play an important role. A high cathodic current density (reverse current treatment, RCT) has proven to be an effective method to create a very high performing nanostructured Ni/YSZ interlayer between the anode and electrolyte [1].
This study analyses the material distribution at the interface anode-electrolyte, which evolved after one RCT cycle to provide an explanation for the anode's excellent performance. A model cell was fabricated by sputtering 400nm Ni thin film onto a 8.5mol% YSZ single crystal. The newly created interlayer is about 200nm thick with an interconnected vermicular structure in the nm-range. Scanning and transmission electron microscopy (SEM and TEM) confirmed the formation of a porous matrix of nanoscaled YSZ and Ni. With atom probe tomography (APT) the atomic 3D distribution of all elements was probed. Y-rich oxide nanoparticles were found within the Ni and the YSZ single crystal partially decomposed into to Y and Zr-rich domains.
A mechanism for the formation of this highly efficient vermicular structure at the Ni/YSZ interface of anode is proposed. These results will help to optimize the fabrication parameters of the interlayer and its performance.

[1] Klotz, D., et al., Journal of the Electrochemical Society, 158 (2011) B587-B595.

Electrochemical device is a term used to describe a group of technologies including fuel cells, batteries, electrolysers and super-capacitors. Whilst many of these technologies are already in common daily usage, for example Li-ion batteries that power our mobile phones, in the future electrochemical devices will play an increasing role in our lives – from fuel cells that can power our homes to high performance batteries for our cars.

At a microscopic length scale, these devices can be considered as one of a general class of porous materials, whereby the physical microstructure will influence a range of phenomena, including diffusion, catalysis and conductivity – our ability to engineer these microscopic features to maximize performance can be translated to substantial improvements in macroscopic device design.

Furthermore, these materials are likely to evolve over time, in response to range of processing and environmental conditions (sintering, corrosion, failure etc); understanding how these changes in microstructure can be linked to understanding of degradation and failure is pivotal to improving device lifetime and safety.

Over the past 10 years the increasingly widespread use of tomography has revolutionized our understanding of these materials; with increasing sophistication researchers have been able to characterize samples over multiple time and length scales from nm to mm and from ms to days. Here we consider examples of our work to explore these materials in three and “four” dimensions, presenting case studies from Solid Oxide Fuel Cells and Li-ion batteries that utilize both laboratory and synchrotron X-ray sources. Furthermore we explore how their application with complementary spectroscopy and correlative microscopy tools can be used to inform a comprehensive understanding of these materials, from the atom to the device.

Carbon composites are being developed for use in a wide variety of applications ranging from sports equipment to aerospace materials. Studying the porous structure in the manufacturing process plays an important role in making better materials with enhanced properties. Using the synchrotron CT capabilities at the Canadian Light Source (CLS) and state of art technology 3D visualization software Avizo, we were able to image the materials non-destructively to visualize and quantify the pore space differences between a cured and uncured sample.

Images were acquired at the BMIT-BM beamline of the CLS, which is a third generation, 2.9 GeV storage ring operating at a ring current of 250 mA. The composites prepared by the Composites Research Network at UBC were mounted onto a piece of polycarbonate and imaged simultaneously at the CLS. The samples were mounted onto a 360 degree rotation stage, where 3750 projections were taken every 0.048 degrees for 180° and captured with a Hamamatsu camera with 4.3 µm pixel resolutions. Avizo 9 was used for image processing. Advanced segmentation and quantification modules were used to analyze the carbon fiber woven structure and pores inside the composite. The combination of BMIT-BM and Avizo techniques are able to accurately demonstrate the 3D composite structure qualitatively and quantitatively. In addition to composite samples, the workflow presented here can be potentially applicable to a wide range of general porous carbon fiber material analysis.

Models of Li-ion battery performance and durability often ignore the conductive carbon-binder domains (CBDs), yet recent experimental and modeling work by Wheeler et al and by Garcia et al have indicated that the manner in which CBD is distributed through the electrode plays an important role in ion transport kinetics. Until now, however, the 3-dimensional distribution of CBDs within the electrode has not been measured. While X-ray microscopy (XRM) has been demonstrated by Shearing et al and Eastwood et al to be a powerful technique for imaging graphite electrodes, the active particles and the CBDs exhibit nearly identical CT numbers in the reconstructed volumes due to their similar mass densities. For this work, we doped small amounts of carbon-coated Fe nanoparticles into the CBDs, taking advantage of the resulting increased effective mass density to elevate the CT numbers of the CBDs in the XRM imaging results. The Fe nanoparticles and the conductive carbon have similar sizes (25 nm) and both have carbon surfaces; thus, they were expected to disperse similarly in the binder. The dispersion was confirmed via X-ray radiography, and the specimen was successfully imaged in 3D using nano-scale XRM. This presentation will describe the results of this imaging experiment, focusing on the analysis for how the CBDs are distributed and their effects on ion transport within a Li-ion battery electrode.

Nuclear grade graphite, essentially a graphite/graphite composite, is a key material in generation IV high-temperature gas-cooled nuclear reactors (HTGR) due to their chemical and mechanical stability at high operating temperatures (~1000°C). Like most commercially produced graphite, the manufacturing process introduces a significant amount of porosity into the bulk material. The porosity and its morphology can drastrically effect a graphite's irradiation, mechanical, thermal, and oxidation performance within an HTGR.

In this study μX-ray tomography was used to investigate differences in the macroporosity of four nuclear graphites manufactured using isomolding, vibrational molding, and extrusion processes. Using various image processing and image analysis techniques, information was obtained from the 3D data sets regarding macropore fraction, interconnectivity, size, shape, preffered orientation, surface area, fractal dimension, and branching of macroporosity in each grade. This data was used to draw correlations between the manufacturing process and the macroporosity formed. In turn, these correlations can be used to better understand the performance of nuclear graphites in the extreme environment of HTGRs.

In recent years, several advancements have been made in available instrumentation for 3D imaging of materials. Amongst these developments is X-ray microscopy (XRM), a rapidly-growing technique for non-destructively imaging materials in 3D. XRM is a highly flexible approach, providing volumetric data spanning 10 orders of magnitude, from 10³ µm³ to 10¹³ µm³, with imaging resolutions routinely down to the sub-50 nm regime. A unique strength of XRM is the ability to collect this data without sectioning, preserving the original specimen for future imaging studies, such as correlative microscopy investigations and studies of microstructure evolution ( “4D” imaging). In parallel with XRM developments, focused ion beam instruments (FIBs) have been paired with scanning electron microscopes (SEMs), producing a destructive 3D imaging platform capable of resolving structures down to single nanometers. These FIB-SEM as well as XRM systems are available as commercial products, providing information across many different length scales. Here, we demonstrate the correlative nature of XRM and FIB-SEM for practical, routine correlative microscopy workflows in a commercial package. We will introduce the ATLAS software platform, which makes use of the 3D X-ray data to map out large volumes of material, producing coordinates that can be passed directly into the FIB-SEM to intelligently navigate to specific regions of interest for higher-resolution imaging. Both the technology and several application examples will be presented.

Tin, being the primary constituent of solder interconnects, strongly influences the reliability of an electronic package. Due to its highly anisotropic nature, the study of Tin grain size and crystal orientation is vital for assessing the mechanical, thermal and electrical performance of a solder joint. With recent advances in electronic packaging, the pitch size of interconnects is extremely small. Under such constraints, the effect of miniaturization on the microstructure is not fully understood. In this study, serial sectioning and Orientation image microscopy (OIM) are employed to study the crystal orientation of Tin grains and the intermetallic distribution in a Cu/Tin/Cu solder joint in 3D. This technique is utilized in solder joints of different sizes to study the ‘volume effect’ on microstructure and mechanical properties. In addition, a reconstructed 3D volume generated computationally, using a two-point correlation function, will be compared with the experimentally obtained 3D volume.

As the production of carbon nanotubes has increased, the development of materials composed mostly or entirely of CNTs has increased as well. Perhaps the best known of these are buckypaper and nanotube arrays (used in field emission); webs, threads, and aligned sheets are also found in many applications. In all of these materials, structure is the key to their properties. Alignment, density, and other morphology features are closely coupled to properties like strength, porosity, thermal and electrical conductivity, and field emission behavior. Critically, these morphology characteristics are known to be hierarchal in nature - e.g., CNTs form local groups, which cluster in bundles, which further aggregate into arrays, with fibers or each hierarchal unit connecting those of the next level up. However, at present, there is a paucity of literature on the subject of this morphology. Further, what literature there is deals mostly in qualitative or aggregate measures, which may miss or obscure the importance of the structural hierarchy.
Part of the reason for the scarcity of data on this subject is the limited application of tomography techniques to its analysis. Which there are Tomographic analyses on materials containing carbon nanotubes, these tend to composites with a high matrix fraction. Such samples can often be treated as monolithic solids during sample preparation, facilitating their study. In contrast, Materials composed mostly of CNTs are fiber aggregates. It can be difficult to extract a sample from these materials without radically altering its structure. They are prone to fracture, slumping and other distortions. Coarse cutting, in preparation for FIB or other fine sampling techniques, may not yield a clean edge from which to pull samples, due to damage, debris, and the tendency of CNTs to stick to most cutting surfaces. Further complicating matters, bulk infiltration of these materials with a polymer or other stabilizer has been shown to change their morphology, obscuring the very structure that we seek to study.
To address these problems, we explore some of these shortcomings in order to develop a sample preparation technique for these fiber aggregate CNT materials. The resulting technique uses a modified FIB lift out, in combination with pre-FIB processing, which enables creation of Tomography samples from materials consisting purely of CNTs. Also contained herein are some preliminary results of studying the morphology if a CNT material, from a sample prepared by this method.

Wood composites are an important renewable building material whose development has recently enabled the construction of cross-laminated timber (CLT) buildings and other wood-framed mid-rise buildings. However, the importance of adhesive morphology in bond strength and durability at the wood-adhesive interface is largely unknown. Until the recent advent of X-ray computed tomography and tagged adhesives, non-destructive investigation of adhesive morphology was not possible. An in situ stepwise loading and scanning process was used to collect 3D images of bonded lap-shear specimens from elastic deformation through plastic deformation into failure. By using digital volume correlation to track the deformation of the samples, adhesive bond morphology will be compared to pre-failure mechanical performance for three wood species and two adhesive types. To determine the accuracy and precision of digital volume correlation software developed in-house at Oregon State University in a cellular composite structure and identify meaningful correlation parameters, the results of digital volume correlation on artificially deformed wood tomography data sets were compared to their applied displacement fields. Based on our tomography data whose resolution is 1.3³ microns per voxel, we have concluded that digital volume correlation is a useful tool for measuring micron scale displacements inside wood structures.

Aluminum alloys are susceptible to corrosion and corrosion fatigue in service, leading to unpredicted failures. To date, most studies of fatigue crack growth in 7075 alloys have used two-dimensional characterization techniques. This work uses X-ray synchrotron tomography to visualize and quantify the fatigue crack initiation and growth from corrosion pits. The samples were heat treated to different aging conditions (peak-aged and over-aged), then mechanically polished and soaked in 3.5 wt.% NaCl for fifteen days to allow for significant corrosion to occur. These samples were fatigue tested in situ in 3.5 wt.% NaCl using synchrotron x-ray tomography to analyze the fatigue crack initiation and growth characteristics (4D). Hydrogen bubbles were observed between the sample and the fluid upon crack initiation, indicating chemical changes in the sample during in situ corrosion fatigue. The mean pit depth and crack growth rate were quantified for the samples tested. The relationships between aging condition, pit size and fatigue crack initiation and propagation from these pits will be discussed.

The physicochemical properties of polymer nanocomposites (such as polymer/clay systems) are dependent on the properties of the polymer and filler, the reinforcement dimensionality, dispersion and orientation, and the nature of the interface between filler and matrix. Hence, thoroughly characterizing the morphology of those materials can lead to a better understanding of the behaviour of the final product and to improved design tools.
The objective of performing TEM tomoghraphy (TEMT) on Poly lactic acid(PLA)/clay samples is to characterize their 3D microstructure, by obtaining the dispersion, distribution and orientation of the laminar shape clays, as well as morphological details, such as their specific area or shape factor. This information cannot be elucidated from a qualitative TEM analysis or from conventional characterization techniques such as X-Ray diffraction.
To carry out this work, nanocomposites obtained by extruding a PLA matrix and Cloisite 30B clays are used. The nanocomposites are obtained by using different extrusion shear rates (120 rpm, 300 rpm and 650 rpm). Shear rate favours the exfoliation of the clay particles and their dispersion, leading to nanocomposites with different microstructures and properties. Then, quantitative TEMT is performed to all the nanocomposites and the resulting 3D quantitative characterization is used for the comprehension of the mechanical behaviour of the nanocomposites.
However, the accuracy of quantitative TEMT relies, among other factors, on the quality of the 3D reconstruction. In this regard, the used angular tilt range (which determines the missing wedge and as consequence the elongation effect), angular increments, the accuracy of image alignment of the tilt series, the chosen reconstruction algorithm and the segmentation of the objects are factors that have a severe impact on the outcome. In this study we focus on the segmentation step, which is essential in order to quantitatively analyze the reconstructed volume. Actually, the tedious and subjective manual segmentation still remains as the prevalent method of choice.
Therefore, one of the main purposes of this study is the formulation and application of a simple, efficient, semi-automated and objective methodology to perform the segmentation. The segmentation threshold is optimized taking into account the rate of variation in the measured dimensions of the segmented objects as a function of the grey level used as threshold. Finally, the segmentation methodology is applied to all the nanocomposites and the 3D quantitative characterization used to explain the composition-processing -microstructure relationship. The effect of the segmentation on the quantified parameters is also evaluated.

Block copolymers (BCP) are considered one of the leading materials for next generation nanolithography of semiconductors devices and magnetic storage media, due to their ability to self-assemble into sub-10 nm features. Directed self-assembly (DSA) of BCP, through chemically or topographically substrate pre-patterning, enables control over the self-assembly process, resulting in highly uniform BCP morphology over macroscopic length scale. However, despite the fact that the three-dimensional (3D) structure of the directed BCP is critical for utilizing these films in patterning and other applications, little is known on the 3D structure of directed self-assembled BCP films.

Here, we have developed methodologies for studying the three-dimensional structure of DSA BCP films using artifact-free sample preparation, and scanning transmission electron microscopy (STEM) tomography. The assembly of solvent annealed poly(2-vinyl pyridine-b-styrene-b-2-vinyl pyridine) (P2VP-b-PS-b-P2VP) directed by chemical pre-pattern was chosen as a model system for this study. BCP DSA was performed on 4” SiN_x/Si/SiN_x wafers using the conventional process, and then back etched to form individual TEM specimens.

STEM tomography characterization revealed the effect of the pre-pattern geometry and chemistry on the 3D morphology. Importantly, it was discovered that at certain geometries, the mismatch between the pre-pattern geometry and the BCP periodicity, leads to the formation of aligned lamellae at the top of the film but hidden defective structures at the bottom of it, which leads to non-optimal pattern transfer. In addition, quantitative study of the aligned lamellae has led, for the first time in the field, to characterization of line edge roughness (LER) variation in 3D. Finally, by combining molecular simulations with STEM tomography characterization, we were able to study the interfacial width between BCP domains in a DSA film.

The insights gained on the 3D assembly of BCP films through this study will enable better design of materials and pre-patterns for future patterning, while the approaches and techniques developed here can be utilized in other soft-matter and inorganic systems.

High efficiency photovoltaic (PV) devices continue to push state-of-the-art conversion efficiency closer to 50%. These high efficiency device structures rely on multiple junctions, with each junction designed to optimally harvest specific regions of the solar spectrum. The availability of high-quality materials that provide the necessary optoelectronic properties for each junction is critical to the success of the designed structure. The next milestone for high efficiency PV performance is to push the state-of-the-art to conversion efficiencies exceeding 50%. In order to achieve these efficiency levels, new materials are being developed to address the optoelectronic needs of the improved multijunction (MJ) devices. Recent work has focused on the development of III-V materials lattice-matched (LM) to InP, which has an available bandgap range of 0.74-1.8 eV. The ability to grow LM materials leads to improved material quality, while this large range of available bandgaps allows for the implementation of near-ideal bandgap combinations in a triple-junction device.

The material of interest for this study is a III-V random alloy that is intended to serve as the top cell in a triple-junction PV device – InAlAsSb lattice-matched to InP with a theoretically-identified bandgap of 1.74 eV. The optoelectronic properties of the grown material do not follow the predicted values, however, and the analyses from multiple material characterization techniques indicate that phase segregation is occurring within the material. In order to identify the magnitude, geometry, and distribution of the segregation, as well as the composition of the segregated alloys, atom probe tomography (APT) is a promising technique for the generation of a 3D compositional profile of the bulk alloy. In this work, we examine the effects of laser pulse energy within the Local Electrode Atom Probe (LEAP) system on InAlAsSb data collection so as to identify the conditions under which the highest resolution with minimal complications to analysis may be achieved. Changing laser pulse energy is shown to affect the uniformity of sample evaporation as well as the features of the mass spectrum – both of which control the ability to accurately reconstruct the 3D composition map of a material sample. In addition to providing a critical foundation for understanding the limits of the InAlAsSb material system survivability during APT analysis, the resulting comparisons identify the trade-offs between data collection challenges such as reduced molecule evaporation versus higher background that have implications for advanced material systems.

Atom probe tomography (APT) has progressively engaged the world of materials characterization with 3-dimensional nanometer-level maps of various dense materials starting from steels reaching now multilayered composites. These atom maps enable new perspectives and analysis of solid materials literally atom by atom. The analysis of organic materials, even aqueous solutions, has however been a long-standing issue as it is impaired by contamination, uncertain phase formation, and questionable observed states.
Here we present the development steps achieved together with the APT results obtained of three aqueous based solutions: a water-based citrate solution, a 1:1 water-ethanol mixture, and a commercially available marginally alcoholic beverage. The aqueous solutions were chosen so as to exhibit differences in their mass-spectrum response. The methodologies enabling these analyses required arresting the liquids so they are stable in vacuum environments, sharpening them to a needle geometry, and transporting them between chambers whilst not altering their structural integrity, all this being done close to liquid N₂ temperatures. All aqueous specimens were analyzed successfully; with the resulting amounts of ROI analyzed being small (only a thin film is probed due to sample geometry), trends and fluctuations in ion concentrations have been interrogated and will be presented.
This work is a first step towards the analysis of organic materials in aqueous solutions in nanometer-level resolution.

FIB-SEMs are used extensively to characterize 3D geometry of composite materials, however, their application to analysis of polymer nanocomposites has been limited. The primary concern that arises with FIB milling of polymers is the possibility of severe thermal damage that occurs in close to the ion beam impact. Recent research has shown that such localized damage can be mitigated through cryogenic cooling of the sample as well as low current milling and intelligent ion beam control. Here we describe unexpected non-localized artefacts that occur during FIB milling of composite polymer coatings with nanoscale pigment particles. Specifically, we show that FIB milling of pigmented polysiloxane coating can lead to formation of multiple microscopic voids within the substrate as far as 5 µm away from the ion beam impact. As such, destructive tomography of polymer nanocomposites with wrong milling conditions can lead to completely wrong conclusions about the materials morphology and porosity. We use further experimentation and modeling to show that void formation occurs because of ion beam heating of the pigment particles that leads to decomposition and vaporization of the surrounding polymer. We also identify FIB milling conditions that mitigate this issue.

X-ray diffraction contrast tomography (DCT) is applied to the analysis of 3D grain structures in strontium titanate during grain growth during successive annealing steps at sintering temperature. The evolution of the 3D grain boundary network is obtained in a multiple time step series with all crystallographic information of the neighbouring grains. The evolution of individual grains is analysed with respect to morphology, topology and crystallographic orientation. This shows a tendency towards faceting of grain boundaries with a {100} orientation of one grain. Some large grains predominantly evolve {100} faces and take a near “cubic” shape. Closer analysis of the individual grain growth rates reveals that a few moderately large grains grow dramatically while most grains do not change much. This is a signature of anomalous grain growth that however is not visible in the grain size distribution. This observation demonstrates the importance of such 3D non-destructive grain growth investigations on the same specimen where grains can be tracked individually and directly be compared to grain growth simulation.

FIB-tomography is used in materials science for 3D-analysis of nanostructured materials [1] and in life science for the analysis of complex structures like brain tissue [2]. This presentation summarizes recent technological improvements, which include advancements in detector technology for electron imaging and elemental analysis, scan generator technology for high throughput imaging, and automated drift correction for reliable 3D reconstruction. New in-column detectors have a higher sensitivity for low energy electrons, which is the basis for a very high resolution down to a few nm voxel size. The low kV imaging can be combined with energy filtering in order to detect a pure signal of backscattered electrons (BSE), which improves the reliability of phase segmentation and quantitative analysis. The quality of the 3D reconstructions can also be improved with refined procedures for drift correction based on reference marks. In addition, with the new scan generators image acquisition and ion milling can be performed synchronously. In this way the acquisition speed increases further. Finally, spectral and elemental mapping (XEDS) based on Silicon Drift Detectors (SDD) provides higher X-ray count rates. Increased acquisition rates open new possibilities in chemical analysis that provide larger data cubes with higher representativeness. The new possibilities of FIB-tomography are illustrated with the following examples: a) Reliable phase segmentation is discussed for a superconducting material with trapped pores that cannot be filled with resin. b) Combined analysis of SE- and BSE stacks reveals the complex microstructure of a Sn-solder with different nano-sized precipitates [3] and c) High throughput elemental analysis is performed of a NiTi stainless steel with a complicated multi-phase microstructure [4]. The examples document the recent advancements in resolution, contrast, stability and throughput, which are necessary for reliable and representative 3D-analysis.

References
1. L. Holzer, M. Cantoni, in Nanofabrication Using Focused Ion and Electron Beams—Principles and Applications, I. Utke, S. Moshkalev, P. Russell, Eds. (Oxford University Press, New York, 2012), pp. 410–435.
2. M. Cantoni, C. Genoud, C. Hébert and Graham Knott, Microsc. & Anal. 24(4): 13-16 (2010)2010.
3. M. Maleki, J. Cugnoni, J. Botsis, Acta Mater. 61 (1), (2013).
4. P. Burdet, J. Vannod, A. Hessler-Wyser, M. Rappaz, M. Cantoni, Acta Mater. 61 (8), 3090 (2013).

Segmentation is a necessary step in the post-processing of 3D tomographic data sets, in order to subsequently calculate quantitative microstructural characteristics of the structure. Multiphase data is often difficult to segment due to high levels of voxel contrast overlap. Further complications arise if the data includes pores. Variation in contrast of unfilled pore backs exacerbates the challenge of segmentation, especially when using intensity-based methods, such as thresholding. Occasionally this problem can be avoided by infiltrating pores, but it is often inescapable. In these cases, either pores are closed and inaccessible to the infiltrate, or filling pores hides a phase of interest, e.g. carbonaceous phases and an epoxy infiltrate.

A contour-based segmentation method, implemented in three dimensions, has been developed to divide the tomogram into super-voxels, or coherent groups of voxels associated with the same region. This method relies on a description of the edge surfaces of the tomogram, based on a 3D Canny edge detector. Watertightness of super-voxels, necessary to separate phase regions, is guaranteed by application of the watershed algorithm (in 3D) on a map of the Euclidean distance of every voxel to the nearest edge voxel. Oversegmentation is mitigated by clipping the distance map – thus limiting the maximum extent of added watershed surfaces – as well as by filtering these surfaces.

Elimination of pores is attempted by exploiting the geometry of data collection for most Focused Ion Beam-Scanning Electron Microscope tomograms. The angle between ion beam and electron column during collection forces micrographs to be collected obliquely. An image-to-image translation of pore backs in aligned images results. When viewed orthogonally to the original imaging surface, this angle manifests as streaking in the tomogram in a characteristic direction. This can be used to directly classify regions as unfilled pores, or to combine oversegmented open pore super-voxels by removing separating surfaces.

Supervised learning from a training set of manually tagged regions in the data is used to describe a classifier of a phase of interest. Similar but untagged regions in the tomogram are then found using the developed classifier, with optional manual filtering of returned regions.

Segmentation has been carried out on a tomogram of a grain from the Tagish Lake meteorite. Of particular interest in this sample are small spherical carbonaceous deposits called nanoglobules, although any included phase could be isolated by segmentation. Nanoglobules have been shown to have a range of C-C and C-H bonding, thought to vary with aqueous processing on the parent body. In this case, volumetric number density, size and shape distributions, and prevalence of inclusions or internal voids are calculated from the segmented regions, in order to elucidate how these nanoglobules form and change on the parent body.

The interaction of engineering alloys with their environment can strongly influence their lifetime – for example, oxidation results in loss of section and environmentally assisted crack growth reduces service life. Research in these areas tends to examine the kinetics and thermodynamics of processes, such as oxidation, but understanding the mechanisms requires detailed microstructural characterisation using a range of complementary techniques that capture data across the length scales. This talk will review strategies that can be used to elucidate the interaction of oxygen with nickel-based superalloys under a range of conditions, including under load.

The diffusion of oxygen into and along the grain boundaries can increase the susceptibility to the formation of surface cracks in polycrystalline Ni-based superalloys at high temperatures. Once cracks are initiated, fatigue tests show that the crack growth rate in the presence of oxygen is about two orders of magnitude higher than that in vacuum. However, even under constant supply of oxygen from the outside of the sample, the crack geometry over the width of the sample suggests that the oxygen pressure distribution varies throughout the sample. A widely branching network of cracks is observed on the outside of a sample with less branching in the bulk. More oxygen reaches the crack tip(s) near an outside surface of a fatigue sample compared to a crack tip in the centre of the sample. This oxygen gradient is assumed to affect the crack propagation path beyond the geometrically imposed stress conditions – plane stress (surface) and plane strain (bulk). Initially, a 2D approach has been adopted to characterise the chemical composition at and ahead of the crack tip (EDS, SIMS and TEM) as well as the stress distribution (HR-EBSD) ahead of the crack tip. As the amount of oxygen accessible to the crack tip influences the stress condition and vice versa, it is crucial to understand this relationship over a 3D volume. To achieve this, a correlative approach has been adopted to understand the cracking behaviour at the macro to micro scale. This involved defining regions of interest within the crack and then studying these in more detail using 3D EBSD based approaches.

Aligned carbon nanotube (A-CNT) based polymer nanocomposites (PNC) are an important class of next-generation advanced materials. Due to the superior, uniaxial properties of the CNTs themselves, composites with tailored, anisotropic properties can be created. However, the experimentally-determined properties of such advanced PNCs, although much higher than the alternatives, still fall short of theoretical predictions. Recently, we have developed the ability to characterize the 3D morphology of such PNCs at the nanoscale using energy-filtered electron tomography. In this presentation, we discuss our novel imaging protocol and an automated segmentation methodology that enables the fast, accurate characterization of statistically-relevant volumes of low-contrast A-CNT PNCs. Further, we discuss image analysis schemes that aid in the extraction of rich, quantitative morphological data (alignment, bundle/network structure, 3D waviness) from the 3D reconstructions. We use this data to re-visit previously-performed experimental measurements of electrical, thermal and mechanical properties in these PNCs. The results show new trends in the dependence of these properties on the CNT volume fraction, and explain previously perplexing phenomena. We observe that the CNT-CNT network connections play a dominant role in transport properties such as electrical and thermal conductivities. Additionally, CNT waviness and proximity are found to strongly influence the PNC stiffness. Electron tomography is shown to be a powerful tool in establishing nanocomposite structure-property relations, making it an essential tool for understanding and tailoring the next generation of advanced materials.

Neutrons and x-rays provide complimentary imaging data due to their differences in elemental sensitivity. X-ray attenuation increases with increasing atomic mass whereas neutrons have a random scattering in attenuation with high sensitivity to hydrogen and lithium among others but low sensitivity to most metals such as aluminum and steel. To utilize this complementarity, a system for simultaneous neutron and x-ray imaging has been developed at NIST. This system orients an x-ray source 90° to a neutron beam to facilitate simultaneous tomography with both imaging systems. Simultaneously imaging with both modes will improve data acquisition times and allow samples that are changing with time or with stochastic processes to be better characterized when compared to sequential imaging with neutrons and x-rays. This work will focus on the implementation and data analysis of the combined tomography sets into quantitative datasets for model validation and direct numerical analysis.

Quantitative neutron tomography is a nondestructive materials characterization method that has several advantages for investigating concrete deterioration. Like X-ray tomography, neutron tomography provides a 3-dimensional image of an object, such as a concrete core. This makes it possible to visualize the spatial distribution of the various phases in the concrete and to compute the volumetric fractions of each. However, unlike X-rays, for which attenuation depends primarily on density, the attenuation of neutrons in concrete is dominated by hydrogen atom concentration. Consequently, hydrous cement reaction product phases such as C-S-H gel and calcium hydroxide can be individually identified. In particular, ettringite [Ca₆Al₂(SO₄)₃(OH)₁₂ 26H₂O] is a calcium aluminate sulfate hydrate mineral that can develop in concrete at times on the order of months to years after casting. This delayed formation leads to expansive stresses that can cause reduction in compressive strength. Ettringite, which is nearly 50% water by weight, has the highest neutron attenuation factor, 4.67 cm^-1, compared to 2.73 cm^-1 for C-S-H gel and 3.19 cm^-1 for calcium hydroxide. In preliminary experiments to evaluate this method, several 5 cm diameter cores drilled from concrete prisms, which had been subjected to different temperature conditions during curing, were scanned at 20 μm resolution by neutron tomography at the NIST Neutron Imaging Facility. The grayscale histogram of attenuation was segmented to develop a classification system that associated each voxel with a specific concrete phase. The ettringite voxels were then summed to determine the total volume fraction in each core. The concrete that showed the greatest physical expansion also had the largest amount of ettringite. This capability for making quantitative estimates of volumetric fractions of phases over time in relatively large concrete samples promises to be transformative for the field of concrete research.

The mixed ionic-electronic conducting cathode La_0.58Sr_0.4Co_0.2Fe_0.8O_3-δ (LSCF) is utilized worldwide for high performing intermediate temperature solid oxide fuel cells (SOFCs). The drawback of using LSCF are secondary phases at the interface to the Zirconia-based electrolyte (8mol% Yttria stabilized Zirconia - YSZ). Although a thin diffusion barrier interlayer of Gd_0.2Ce_0.8O_2-δ (GDC) is introduced in-between cathode and electrolyte, the formation of insulating SrZrO₃ cannot be prevented completely at this interface region. The consequent performance drop may superimpose the outstanding charge transfer properties of LSCF. High resolution detection and 3D quantification of the spatial SrZrO₃ distribution is therefore of fundamental interest.

This contribution will introduce a new methodology of using FIB-SEM tomography to visualize secondary phases at the cathode/electrolyte interface in SOFCs. The core feature is the choice of the microscope parameters using the Everhart-Thornley and Inlens detector at different acceleration voltages that enables a broader spectrum of greyscale information between the primary materials and secondary phases. The correct material assignment is supported by high-resolution STEM-EDXS mappings in an additional analysis. Besides the secondary phase SrZrO₃, also the interdiffusion of GDC and 8YSZ is detected and can be reconstructed individually. The result is a complete 3D reconstructed material data set including primary and secondary phases that enables great possibilities of visualization and modeling of secondary phases – a further step towards understanding one of the most performance deteriorating impacts in SOFCs.