An Acquisition Parameter Study for Machine-Learning-Enabled Electron Backscatter Diffraction.

Methods within the domain of artificial intelligence are gaining traction for solving a range of materials science objectives, notably the use of deep neural networks for computer vision for the analysis of electron diffraction patterns. An important component of deploying these models is an understanding of the performance as experimental diffraction conditions are varied. This knowledge can inspire confidence in the classifications over a range of operating conditions and identify where performance is degraded. Elucidating the relative impact of each parameter will suggest the most important parameters to vary during the collection of future training data. Knowing which data collection efforts to prioritize is of concern given the time required to collect or simulate vast libraries of diffraction patterns for a wide variety of materials without considering varying any parameters. In this work, five parameters, frame averaging, detector tilt, sample-to-detector distance, accelerating voltage, and pattern resolution, essential to electron diffraction are individually varied during the collection of electron backscatter diffraction patterns to explore the effect on the classifications produced by a deep neural network trained from diffraction patterns captured using a fixed set of parameters. The model is shown to be resilient to nearly all the individual changes examined here.

Efficient few-shot machine learning for classification of EBSD patterns.

Deep learning is quickly becoming a standard approach to solving a range of materials science objectives, particularly in the field of computer vision. However, labeled datasets large enough to train neural networks from scratch can be challenging to collect. One approach to accelerating the training of deep learning models such as convolutional neural networks is the transfer of weights from models trained on unrelated image classification problems, commonly referred to as transfer learning. The powerful feature extractors learned previously can potentially be fine-tuned for a new classification problem without hindering performance. Transfer learning can also improve the results of training a model using a small amount of data, known as few-shot learning. Herein, we test the effectiveness of a few-shot transfer learning approach for the classification of electron backscatter diffraction (EBSD) pattern images to six space groups within the [Formula: see text] point group. Training history and performance metrics are compared with a model of the same architecture trained from scratch. In an effort to make this approach more explainable, visualization of filters, activation maps, and Shapley values are utilized to provide insight into the model's operations. The applicability to real-world phase identification and differentiation is demonstrated using dual phase materials that are challenging to analyze with traditional methods.

Bulk high-entropy nitrides and carbonitrides.

High-entropy ceramics have potential to improve the mechanical properties and high-temperature stability over traditional ceramics, and high entropy nitrides and carbonitrides (HENs and HECNs) are particularly attractive for high temperature and high hardness applications. The synthesis of 5 bulk HENs and 4 bulk HECNs forming single-phase materials is reported herein among 11 samples prepared. The hardness of HENs and HECNs increased by an average of 22% and 39%, respectively, over the rule-of-mixtures average of their monocarbide and mononitride precursors. Similarly, elastic modulus values increased by an average of 17% in nitrides and 31% in carbonitrides over their rule-of-mixtures values. The enhancement in mechanical properties is tied to an increase in the configurational entropy and a decrease in the valence electron concentration, providing parameters for tuning mechanical properties of high-entropy ceramics.

Phase Mapping in EBSD Using Convolutional Neural Networks.

The emergence of commercial electron backscatter diffraction (EBSD) equipment ushered in an era of information rich maps produced by determining the orientation of user-selected crystal structures. Since then, a technological revolution has occurred in the quality, rate detection, and analysis of these diffractions patterns. The next revolution in EBSD is the ability to directly utilize the information rich diffraction patterns in a high-throughput manner. Aided by machine learning techniques, this new methodology is, as demonstrated herein, capable of accurately separating phases in a material by crystal symmetry, chemistry, and even lattice parameters with fewer human decisions. This work is the first demonstration of such capabilities and addresses many of the major challenges faced in modern EBSD. Diffraction patterns are collected from a variety of samples, and a convolutional neural network, a type of machine learning algorithm, is trained to autonomously recognize the subtle differences in the diffraction patterns and output phase maps of the material. This study offers a path to machine learning coupled phase mapping as databases of EBSD patterns encompass an increasing number of the possible space groups, chemistry changes, and lattice parameter variations.

Deep Neural Network Enabled Space Group Identification in EBSD.

Electron backscatter diffraction (EBSD) is one of the primary tools in materials development and analysis. The technique can perform simultaneous analyses at multiple length scales, providing local sub-micron information mapped globally to centimeter scale. Recently, a series of technological revolutions simultaneously increased diffraction pattern quality and collection rate. After collection, current EBSD pattern indexing techniques (whether Hough-based or dictionary pattern matching based) are capable of reliably differentiating between a "user selected" set of phases, if those phases contain sufficiently different crystal structures. EBSD is currently less well suited for the problem of phase identification where the phases in the sample are unknown. A pattern analysis technique capable of phase identification, utilizing the information-rich diffraction patterns potentially coupled with other data, such as EDS-derived chemistry, would enable EBSD to become a high-throughput technique replacing many slower (X-ray diffraction) or more expensive (neutron diffraction) methods. We utilize a machine learning technique to develop a general methodology for the space group classification of diffraction patterns; this is demonstrated within the $\lpar 4/m\comma \;\bar{3}\comma \;\;2/m\rpar$ point group. We evaluate the machine learning algorithm's performance in real-world situations using materials outside the training set, simultaneously elucidating the role of atomic scattering factors, orientation, and pattern quality on classification accuracy.

Crystal symmetry determination in electron diffraction using machine learning.

Electron backscatter diffraction (EBSD) is one of the primary tools for crystal structure determination. However, this method requires human input to select potential phases for Hough-based or dictionary pattern matching and is not well suited for phase identification. Automated phase identification is the first step in making EBSD into a high-throughput technique. We used a machine learning-based approach and developed a general methodology for rapid and autonomous identification of the crystal symmetry from EBSD patterns. We evaluated our algorithm with diffraction patterns from materials outside the training set. The neural network assigned importance to the same symmetry features that a crystallographer would use for structure identification.

Novel remapping approach for HR-EBSD based on demons registration.

In this study, the possibility of utilizing a computer vision algorithm, i.e., demons registration, to accurately remap electron backscatter diffraction patterns for high resolution electron backscatter diffraction (HR-EBSD) applications is presented. First, the angular resolution of demons registration is demonstrated to be lower than the conventional cross-correlation based method, particularly at misorientation angles >0.157 rad. In addition, GPU acceleration has been applied to significantly boost the speed of iterative registration between a pair of patterns with 0.175 rad misorientation to under 1 s. Second, demons registration is implemented as a first-pass remapping, followed by a second pass cross-correlation method, which results in angular resolution of ~0.5 × 10-4 rad, a phantom stress value of ~35 MPa and phantom strain of ~2 × 10-4, on dynamically simulated patterns, without the need of implementing robust fitting or iterative remapping. Lastly, the new remapping method is applied to a large experimental dataset collected from an as-built additively-manufactured Inconel 625 cube, which shows significant residual stresses built-up near the large columnar grain region and regularly arranged GND structures.

High-entropy high-hardness metal carbides discovered by entropy descriptors.

High-entropy materials have attracted considerable interest due to the combination of useful properties and promising applications. Predicting their formation remains the major hindrance to the discovery of new systems. Here we propose a descriptor-entropy forming ability-for addressing synthesizability from first principles. The formalism, based on the energy distribution spectrum of randomized calculations, captures the accessibility of equally-sampled states near the ground state and quantifies configurational disorder capable of stabilizing high-entropy homogeneous phases. The methodology is applied to disordered refractory 5-metal carbides-promising candidates for high-hardness applications. The descriptor correctly predicts the ease with which compositions can be experimentally synthesized as rock-salt high-entropy homogeneous phases, validating the ansatz, and in some cases, going beyond intuition. Several of these materials exhibit hardness up to 50% higher than rule of mixtures estimations. The entropy descriptor method has the potential to accelerate the search for high-entropy systems by rationally combining first principles with experimental synthesis and characterization.

Lightweight Open-Cell Scaffolds from Sea Urchin Spines with Superior Material Properties for Bone Defect Repair.

Sea urchin spines (Heterocentrotus mammillatus), with a hierarchical open-cell structure similar to that of human trabecular bone and superior mechanical property (compressive strength ∼43.4 MPa) suitable for machining to shape, were explored for potential applications of bone defect repair. Finite element analyses reveal that the compressive stress concentrates along the dense growth rings and dissipates through strut structures of the stereoms, indicating that the exquisite mesostructures play an important role in high strength-to-weight ratios. The fracture strength of magnesium-substituted tricalcium phosphate (ß-TCMP) scaffolds produced by hydrothermal conversion of urchin spines is about 9.3 MPa, comparable to that of human trabecular bone. New bone forms along outer surfaces of ß-TCMP scaffolds after implantation in rabbit femoral defects for one month and grows into the majority of the inner open-cell spaces postoperation in three months, showing tight interface between the scaffold and regenerative bone tissue. Fusion of beagle lumbar facet joints using a Ti-6Al-4V cage and ß-TCMP scaffold can be completed within seven months with obvious biodegradation of the ß-TCMP scaffold, which is nearly completely degraded and replaced by newly formed bone ten months after implantation. Thus, sea urchin spines suitable for machining to shape have advantages for production of biodegradable artificial grafts for bone defect repair.

Calcium phosphate-bearing matrices induce osteogenic differentiation of stem cells through adenosine signaling.

Synthetic matrices emulating the physicochemical properties of tissue-specific ECMs are being developed at a rapid pace to regulate stem cell fate. Biomaterials containing calcium phosphate (CaP) moieties have been shown to support osteogenic differentiation of stem and progenitor cells and bone tissue formation. By using a mineralized synthetic matrix mimicking a CaP-rich bone microenvironment, we examine a molecular mechanism through which CaP minerals induce osteogenesis of human mesenchymal stem cells with an emphasis on phosphate metabolism. Our studies show that extracellular phosphate uptake through solute carrier family 20 (phosphate transporter), member 1 (SLC20a1) supports osteogenic differentiation of human mesenchymal stem cells via adenosine, an ATP metabolite, which acts as an autocrine/paracrine signaling molecule through A2b adenosine receptor. Perturbation of SLC20a1 abrogates osteogenic differentiation by decreasing intramitochondrial phosphate and ATP synthesis. Collectively, this study offers the demonstration of a previously unknown mechanism for the beneficial role of CaP biomaterials in bone repair and the role of phosphate ions in bone physiology and regeneration. These findings also begin to shed light on the role of ATP metabolism in bone homeostasis, which may be exploited to treat bone metabolic diseases.

Cancer cell migration within 3D layer-by-layer microfabricated photocrosslinked PEG scaffolds with tunable stiffness.

Our current understanding of 3-dimensional (3D) cell migration is primarily based on results from fibrous scaffolds with randomly organized internal architecture. Manipulations that change the stiffness of these 3D scaffolds often alter other matrix parameters that can modulate cell motility independently or synergistically, making observations less predictive of how cells behave when migrating in 3D. In order to decouple microstructural influences and stiffness effects, we have designed and fabricated 3D polyethylene glycol (PEG) scaffolds that permit orthogonal tuning of both elastic moduli and microstructure. Scaffolds with log-pile architectures were used to compare the 3D migration properties of normal breast epithelial cells (HMLE) and Twist-transformed cells (HMLET). Our results indicate that the nature of cell migration is significantly impacted by the ability of cells to migrate in the third dimension. 2D ECM-coated PEG substrates revealed no statistically significant difference in cell migration between HMLE and HMLET cells among substrates of different stiffness. However, when cells were allowed to move along the third dimension, substantial differences were observed for cell displacement, velocity and path straightness parameters. Furthermore, these differences were sensitive to both substrate stiffness and the presence of the Twist oncogene. Importantly, these 3D modes of migration provide insight into the potential for oncogene-transformed cells to migrate within and colonize tissues of varying stiffness.

Three-dimensional scaffolding to investigate neuronal derivatives of human embryonic stem cells.

Access to unlimited numbers of live human neurons derived from stem cells offers unique opportunities for in vitro modeling of neural development, disease-related cellular phenotypes, and drug testing and discovery. However, to develop informative cellular in vitro assays, it is important to consider the relevant in vivo environment of neural tissues. Biomimetic 3D scaffolds are tools to culture human neurons under defined mechanical and physico-chemical properties providing an interconnected porous structure that may potentially enable a higher or more complex organization than traditional two-dimensional monolayer conditions. It is known that even minor variations in the internal geometry and mechanical properties of 3D scaffolds can impact cell behavior including survival, growth, and cell fate choice. In this report, we describe the design and engineering of 3D synthetic polyethylene glycol (PEG)-based and biodegradable gelatin-based scaffolds generated by a free form fabrication technique with precise internal geometry and elastic stiffnesses. We show that human neurons, derived from human embryonic stem (hESC) cells, are able to adhere to these scaffolds and form organoid structures that extend in three dimensions as demonstrated by confocal and electron microscopy. Future refinements of scaffold structure, size and surface chemistries may facilitate long term experiments and designing clinically applicable bioassays.

Evolution of iridium-based molecular catalysts during water oxidation with ceric ammonium nitrate.

Organometallic iridium complexes have been reported as water oxidation catalysts (WOCs) in the presence of ceric ammonium nitrate (CAN). One challenge for all WOCs regardless of the metal used is stability. Here we provide evidence for extensive modification of many Ir-based WOCs even after exposure to only 5 or 15 equiv of Ce(IV) (whereas typically 100-10000 equiv are employed during WOC testing). We also show formation of Ir-rich nanoparticles (likely IrO(x)) even in the first 20 min of reaction, associated with a Ce matrix. A combination of UV-vis and NMR spectroscopy, scanning transmission electron microscopy, and powder X-ray diffraction is used. Even simple IrCl(3) is an excellent catalyst. Our results point to the pitfalls of studying Ir WOCs using CAN.

Effects of age and loading rate on equine cortical bone failure.

Although clinical bone fractures occur predominantly under impact loading (as occurs during sporting accidents, falls, high-speed impacts or other catastrophic events), experimentally validated studies on the dynamic fracture behavior of bone, at the loading rates associated with such events, remain limited. In this study, a series of tests were performed on femoral specimens obtained post-mortem from equine donors ranging in age from 6 months to 28 years. Fracture toughness and compressive tests were performed under both quasi-static and dynamic loading conditions in order to determine the effects of loading rate and age on the mechanical behavior of the cortical bone. Fracture toughness experiments were performed using a four-point bending geometry on single and double-notch specimens in order to measure fracture toughness, as well as observe differences in crack initiation between dynamic and quasi-static experiments. Compressive properties were measured on bone loaded parallel and transverse to the osteonal growth direction. Fracture propagation was then analyzed using scanning electron and scanning confocal microscopy to observe the effects of microstructural toughening mechanisms at different strain rates. Specimens from each horse were also analyzed for dry, wet and mineral densities, as well as weight percent mineral, in order to investigate possible influences of composition on mechanical behavior. Results indicate that bone has a higher compressive strength, but lower fracture toughness when tested dynamically as compared to quasi-static experiments. Fracture toughness also tends to decrease with age when measured quasi-statically, but shows little change with age under dynamic loading conditions, where brittle "cleavage-like" fracture behavior dominates.

Loading rate effects on the R-curve behavior of cortical bone.

Rising resistance curve (R-curve) behavior in bone during quasi-static experiments has demonstrated the importance of microstructural toughening mechanisms in resisting fracture. However, despite clinical bone fracture primarily occurring under dynamic loading and the significant changes in material behavior observed with increasing strain rates, there have been no previous investigations into whether crack growth resistance is maintained during dynamic fracture. Using a novel modified split-Hopkinson pressure bar coupled with a high-speed camera to measure crack propagation, we present the first evidence of rising R-curve behavior in bone under dynamic loading (∼2 × 10(5)MPam(1/2)s(-1)). Results indicate that rising R-curve behavior is maintained, although with lower crack initiation toughness and propagation resistance than observed in quasi-static experiments. Observations of crack initiation and propagation in double-notched specimens using confocal fluorescence microscopy and electron microscopy suggest that this is due to subtle differences in toughening mechanisms between quasi-static and dynamic fracture.

Preparation, characterization and mechanical performance of dense beta-TCP ceramics with/without magnesium substitution.

Beta-tricalcium phosphate (beta-TCP) powder was prepared by a two-step process: wet precipitation of apatitic tricalcium phosphate [Ca(9)(HPO(4))(PO(4))(5)(OH)] (beta-TCP 'precursor') and calcination of the precursor at 800 degrees C for 3 h to produce beta-TCP. Magnesium-substituted tricalcium phosphate (beta-TCMP) was produced by adding Mg(NO(3))(2) . 6H(2)O into Ca(NO(3))(2) solution as Mg(2+) source before the precipitation step. The transition temperature from beta-TCP to alpha-TCP increases with the increase of Mg(2+) content in beta-TCMP. beta-TCMP with 3 mol.% Mg(2+) has beta-TCP to alpha-TCP transition temperature above 1,300 degrees C. Dense beta-TCMP (3 mol.% Mg(2+)) ceramics ( approximately 99.4% relative density) were produced by pressing the green bodies at 100 MPa and further sintering at 1,250 degrees C for 2 h. The average compressive strength of dense beta-TCP ceramics sintered at 1,100 degrees C is approximately 540 MPa, while that of beta-TCMP (3 mol.% Mg(2+)) ceramics is approximately 430 MPa.

Conversion of bulk seashells to biocompatible hydroxyapatite for bone implants.

Strombus gigas (conch) shells and Tridacna gigas (Giant clam) shells have dense, tailored structures that impart excellent mechanical properties to these shells. In this investigation, conch and clam seashells were converted to hydroxyapatite (HAP) by a hydrothermal method at different temperatures and for different conversion durations. Dense HAP structures were created from these shells throughout the majority of the samples at the relative low temperature of approximately 200 degrees C. The average fracture stress was found to be approximately 137-218MPa for partially converted conch shell samples and approximately 70-150MPa for original and converted clamshell samples, which is close to the mechanical strength of compact human bone. This indicates that the converted shell samples can be used as implants in load-bearing cases. In vivo tests of converted shell samples were performed in rat femoral defects for 6 weeks. The microtomography images at 6 weeks show that the implants did not move, and untreated control defects remain empty with no evidence of a spontaneous fusion. Histological study reveals that there is newly formed bone growing up to and around the implants. There is no evidence of a fibrosis tissue ring around the implants, also indicating that there is no loosening of the implants. In contrast, the untreated controls remain empty with some evidence of a fibrosis ring around the defect hole. These results indicate good biocompatibility and bioactivity of the converted shell implants.

Experimental investigation of dynamic effects in a two-bar/three-point bend fracture test.

A Hopkinson pressure bar has been modified to measure the dynamic fracture properties of materials at loading rates greater than approximately 10(6) MPa ms. Some fundamental dynamic effects associated with the incident stress pulse, such as stress wave propagation characteristics along the Hopkinson bar and within the cracked specimen, the specimen's dynamic response excited by the stress pulse, and the specimen contact situations with the impactor and supports, need to be understood. To better comprehend these fundamental issues, an experimental investigation of these dynamic effects with the emphasis on "loss of contact" was first performed on a two-bar/three-point dynamic bend fracture test setup using a voltage measurement circuit across the specimen/loading-pin interfaces and high-speed photographs. It was demonstrated here that the three-point bend specimen employed with the current two-bar/three-point bend test setup remains in contact with the impactor and supports throughout the first loading duration and that "loss of contact" does not occur. A further improvement using a pulse-shaping technique was employed for achieving a tailored incident pulse. The effect of pulse shaper on the rise time and duration of the incident pulse as well as the dynamic stress equilibrium in the cracked three-point bend has been investigated, for the first time here, with profound implications for significantly improved dynamic three-point bend fracture testing.

Conversion of sea urchin spines to Mg-substituted tricalcium phosphate for bone implants.

The skeleton of sea urchin spines is composed of large single crystals of Mg-rich calcite, which have smooth, continuously curved surfaces and form a three-dimensional fenestrated mineral network. Spines of the echinoids Heterocentrotus trigonarius and Heterocentrotus mammillatus were converted by the hydrothermal reaction at 180 degrees C to bioresorbable Mg-substituted tricalcium phosphate (beta-TCMP). Due to the presence of Mg in the calcite lattice, conversion to beta-TCMP occurs preferentially to hydroxyapatite formation. The converted beta-TCMP still maintains the three-dimensional interconnected porous structures of the original spine. The main conversion mechanism is the ion-exchange reaction, although there is also a dissolution-reprecipitation process that forms some calcium phosphate precipitates on the surfaces of the spine network. The average fracture strength of urchin spines and converted spines (beta-TCMP) in the compression tests are 42 and 23MPa, respectively. In vivo studies using a rat model demonstrated new bone growth up to and around the beta-TCMP implants after implantation in rat femoral defects for 6 weeks. Some new bone was found to migrate through the spine structural pores, starting from the outside of the implant through the pores at the edge of the implants. These results indicate good bioactivity and osteoconductivity of the porous beta-TCMP implants.

Local heating of discrete droplets using magnetic porous silicon-based photonic crystals.

This paper describes a method for local heating of discrete microliter-scale liquid droplets. The droplets are covered with magnetic porous Si microparticles, and heating is achieved by application of an external alternating electromagnetic field. The magnetic porous Si microparticles consist of two layers. The top layer contains a photonic code and it is hydrophobic, with surface-grafted dodecyl moieties. The bottom layer consists of a hydrophilic silicon oxide host layer that is infused with Fe3O4 nanoparticles. The amphiphilic microparticles spontaneously align at the interface of a water droplet immersed in mineral oil, allowing manipulation of the droplets by application of a magnetic field. Application of an oscillating magnetic field (338 kHz, 18 A rms current in a coil surrounding the experiment) generates heat in the superparamagnetic particles that can raise the temperature of the enclosed water droplet to >80 degrees C within 5 min. A simple microfluidics application is demonstrated: combining complementary DNA strands contained in separate droplets and then thermally inducing dehybridization of the conjugate. The complementary oligonucleotides were conjugated with the cyanine dye fluorophores Cy3 and Cy5 to quantify the melting/rebinding reaction by fluorescence resonance energy transfer (FRET). The magnetic porous Si microparticles were prepared as photonic crystals, containing spectral codes that allowed the identification of the droplets by reflectivity spectroscopy. The technique demonstrates the feasibility of tagging, manipulating, and heating small volumes of liquids without the use of conventional microfluidic channel and heating systems.