Advances in imaging and experimental techniques have made materials science abundant, increasing ... more Advances in imaging and experimental techniques have made materials science abundant, increasing the potential for new discoveries but requiring new techniques to transform the vast amount of data into information that can be readily interpreted by researchers. Data-driven models, specifically deep and machine learning, are one set of tools that have been applied with noticeable success to some burdensome materials and microscopy problems. However, the ultimate goal with these models has yet to be realized: to automatically process, characterized, and compressed the increasingly common high-resolution, multimodal imaging data on a single system. One crucial component of many workflows is classifying the crystal structure of a material, yet is mostly performed by trained expert who divine structural information from minute diffraction pattern variations. [1, 2, 3] The process of determining a crystal's space group often involves a lengthy process requiring fitting to a series of non-linear equations and intimate knowledge of a sample to be performed properly, including standardized approaches such as Rietveld refinement. [4, 5] The heavy dependence in complex matching, time intensive processes, potential for dimensionality reduction, and the vast amount of data generated makes crystal classification an ideal case for automation.
Abstract Simultaneously capturing material structure and chemistry in the form of accessible data... more Abstract Simultaneously capturing material structure and chemistry in the form of accessible data is often advantageous for drawing correlations and enhancing our understanding of measurable materials behavior and properties. Unfortunately, in many cases, accessing data at the scale required, is highly multidimensional and sparse by the historical and evolving nature of materials science. To mitigate difficulties, we develop and employ methods of data analytics in conjunction with open accessible chemistry and structure datasets, to classify and reduce the amount of data needed for extracting useful descriptors from multidimensional techniques. The construction and systematic ablation of our model highlights the potential for dimensional reduction in data sampling, improved classification, and identification of correlations among material crystallography and chemistry.
While machine learning has been making enormous strides in many technical areas, it is still mass... more While machine learning has been making enormous strides in many technical areas, it is still massively underused in transmission electron microscopy. To address this, a convolutional neural network model was developed for reliable classification of crystal structures from small numbers of electron images and diffraction patterns with no preferred orientation. Diffraction data containing 571,340 individual crystals divided among seven families, 32 genera, and 230 space groups were used to train the network. Despite the highly imbalanced dataset, the network narrows down the space groups to the top two with over 70% confidence in the worst case and up to 95% in the common cases. As examples, we benchmarked against alloys to two-dimensional materials to cross-validate our deep-learning model against high-resolution transmission electron images and diffraction patterns. We present this result both as a research tool and deep-learning application for diffraction analysis.
Deep learning and augmented analysis have begun to disrupt the microscopy and microanalysis commu... more Deep learning and augmented analysis have begun to disrupt the microscopy and microanalysis community with advancements made in material-specific models to solve narrow tasks. Developments in compressive imaging, where the concept of less is more in compressive modalities has begun to set the stage for high throughput and dose fractionation on our latest microscopes and characterization platforms. Crystallographic determination is crucial to many workflows within microscopy and materials research as a whole. At the core, crystallography is deeply rooted in pattern recognition and experts train for years to distinguish minute variations within the data.[1] Determining a crystal's space group often involves a lengthy process requiring fitting series of non-linear equations and intimate knowledge of a sample to be performed properly. This heavy dependence in complex pattern matching and time intensive process makes it an ideal case for automation with deep learning.
A major challenge in the application of nanostructured electrolytes in solid oxide electrochemica... more A major challenge in the application of nanostructured electrolytes in solid oxide electrochemical cells is grain boundary blocking originated from unsatisfied atomic bonding and coordination. The resulting increase in grain boundary resistivity works against the expected benefits from the enhanced ion exchange rates enabled by the extensive interfacial network in nanocrystalline materials. This study addresses this challenge by demonstrating that a reduction in the grain boundary excess energies increases the net ionic conductivity as directly measured by impedance electrical spectroscopy in nanocrystalline yttria-stabilized zirconia. The reduced grain boundary energy was designed by doping the system with lanthanum, leading to local excess energy reduction due to segregation of La to boundaries as observed by scanning transmission electron microscopy-based energy-dispersive spectroscopy. The results suggest rare-earth ions with favorable grain boundary segregation enthalpy can smooth out the energy land...
Origami is the art of folding paper. In the context of engineering, orimimetics is the applicatio... more Origami is the art of folding paper. In the context of engineering, orimimetics is the application of folding to solve problems. Kinetic origami behavior can be modeled with the pseudo-rigid-body model since the origami are compliant mechanisms. These compliant mechanisms, when having a flat initial state and motion emerging out of the fabrication plane, are classified as lamina emergent mechanisms (LEMs). To demonstrate the feasibility of identifying links between origami and compliant mechanism analysis and design methods, four flat folding paper mechanisms are presented with their corresponding kinematic and graph models. Principles from graph theory are used to abstract the mechanisms to show them as coupled, or interconnected , mechanisms. It is anticipated that this work lays a foundation for exploring methods for LEM synthesis based on the analogy between flat-folding origami models and linkage assembly.
Origami is the art of folding paper. In the context of engineering, orimimetics is the applicatio... more Origami is the art of folding paper. In the context of engineering, orimimetics is the application of folding to solve problems. Kinetic origami behavior can be modeled with the pseudo-rigid-body model since the origami are compliant mechanisms. These compliant mechanisms, when having a flat initial state and motion emerging out of the fabrication plane, are classified as lamina emergent mechanisms (LEMs). To demonstrate the feasibility of identifying links between origami and compliant mechanism analysis and design methods, four flat folding paper mechanisms are presented with their corresponding kinematic and graph models. Principles from graph theory are used to abstract the mechanisms to show them as coupled, or inter-connected, mechanisms. It is anticipated that this work lays a foundation for exploring methods for LEM synthesis based on the analogy between flat-folding origami models and linkage assembly.
Advances in imaging and experimental techniques have made materials science abundant, increasing ... more Advances in imaging and experimental techniques have made materials science abundant, increasing the potential for new discoveries but requiring new techniques to transform the vast amount of data into information that can be readily interpreted by researchers. Data-driven models, specifically deep and machine learning, are one set of tools that have been applied with noticeable success to some burdensome materials and microscopy problems. However, the ultimate goal with these models has yet to be realized: to automatically process, characterized, and compressed the increasingly common high-resolution, multimodal imaging data on a single system. One crucial component of many workflows is classifying the crystal structure of a material, yet is mostly performed by trained expert who divine structural information from minute diffraction pattern variations. [1, 2, 3] The process of determining a crystal's space group often involves a lengthy process requiring fitting to a series of non-linear equations and intimate knowledge of a sample to be performed properly, including standardized approaches such as Rietveld refinement. [4, 5] The heavy dependence in complex matching, time intensive processes, potential for dimensionality reduction, and the vast amount of data generated makes crystal classification an ideal case for automation.
Abstract Simultaneously capturing material structure and chemistry in the form of accessible data... more Abstract Simultaneously capturing material structure and chemistry in the form of accessible data is often advantageous for drawing correlations and enhancing our understanding of measurable materials behavior and properties. Unfortunately, in many cases, accessing data at the scale required, is highly multidimensional and sparse by the historical and evolving nature of materials science. To mitigate difficulties, we develop and employ methods of data analytics in conjunction with open accessible chemistry and structure datasets, to classify and reduce the amount of data needed for extracting useful descriptors from multidimensional techniques. The construction and systematic ablation of our model highlights the potential for dimensional reduction in data sampling, improved classification, and identification of correlations among material crystallography and chemistry.
While machine learning has been making enormous strides in many technical areas, it is still mass... more While machine learning has been making enormous strides in many technical areas, it is still massively underused in transmission electron microscopy. To address this, a convolutional neural network model was developed for reliable classification of crystal structures from small numbers of electron images and diffraction patterns with no preferred orientation. Diffraction data containing 571,340 individual crystals divided among seven families, 32 genera, and 230 space groups were used to train the network. Despite the highly imbalanced dataset, the network narrows down the space groups to the top two with over 70% confidence in the worst case and up to 95% in the common cases. As examples, we benchmarked against alloys to two-dimensional materials to cross-validate our deep-learning model against high-resolution transmission electron images and diffraction patterns. We present this result both as a research tool and deep-learning application for diffraction analysis.
Deep learning and augmented analysis have begun to disrupt the microscopy and microanalysis commu... more Deep learning and augmented analysis have begun to disrupt the microscopy and microanalysis community with advancements made in material-specific models to solve narrow tasks. Developments in compressive imaging, where the concept of less is more in compressive modalities has begun to set the stage for high throughput and dose fractionation on our latest microscopes and characterization platforms. Crystallographic determination is crucial to many workflows within microscopy and materials research as a whole. At the core, crystallography is deeply rooted in pattern recognition and experts train for years to distinguish minute variations within the data.[1] Determining a crystal's space group often involves a lengthy process requiring fitting series of non-linear equations and intimate knowledge of a sample to be performed properly. This heavy dependence in complex pattern matching and time intensive process makes it an ideal case for automation with deep learning.
A major challenge in the application of nanostructured electrolytes in solid oxide electrochemica... more A major challenge in the application of nanostructured electrolytes in solid oxide electrochemical cells is grain boundary blocking originated from unsatisfied atomic bonding and coordination. The resulting increase in grain boundary resistivity works against the expected benefits from the enhanced ion exchange rates enabled by the extensive interfacial network in nanocrystalline materials. This study addresses this challenge by demonstrating that a reduction in the grain boundary excess energies increases the net ionic conductivity as directly measured by impedance electrical spectroscopy in nanocrystalline yttria-stabilized zirconia. The reduced grain boundary energy was designed by doping the system with lanthanum, leading to local excess energy reduction due to segregation of La to boundaries as observed by scanning transmission electron microscopy-based energy-dispersive spectroscopy. The results suggest rare-earth ions with favorable grain boundary segregation enthalpy can smooth out the energy land...
Origami is the art of folding paper. In the context of engineering, orimimetics is the applicatio... more Origami is the art of folding paper. In the context of engineering, orimimetics is the application of folding to solve problems. Kinetic origami behavior can be modeled with the pseudo-rigid-body model since the origami are compliant mechanisms. These compliant mechanisms, when having a flat initial state and motion emerging out of the fabrication plane, are classified as lamina emergent mechanisms (LEMs). To demonstrate the feasibility of identifying links between origami and compliant mechanism analysis and design methods, four flat folding paper mechanisms are presented with their corresponding kinematic and graph models. Principles from graph theory are used to abstract the mechanisms to show them as coupled, or interconnected , mechanisms. It is anticipated that this work lays a foundation for exploring methods for LEM synthesis based on the analogy between flat-folding origami models and linkage assembly.
Origami is the art of folding paper. In the context of engineering, orimimetics is the applicatio... more Origami is the art of folding paper. In the context of engineering, orimimetics is the application of folding to solve problems. Kinetic origami behavior can be modeled with the pseudo-rigid-body model since the origami are compliant mechanisms. These compliant mechanisms, when having a flat initial state and motion emerging out of the fabrication plane, are classified as lamina emergent mechanisms (LEMs). To demonstrate the feasibility of identifying links between origami and compliant mechanism analysis and design methods, four flat folding paper mechanisms are presented with their corresponding kinematic and graph models. Principles from graph theory are used to abstract the mechanisms to show them as coupled, or inter-connected, mechanisms. It is anticipated that this work lays a foundation for exploring methods for LEM synthesis based on the analogy between flat-folding origami models and linkage assembly.
Uploads
Papers by Matthew Gong