This book deals with an information-driven approach to plan materials discovery and design, iterative learning. The authors present contrasting but complementary approaches, such as those based on high throughput calculations, combinatorial experiments or data driven discovery, together with machine-learning methods. Similarly, statistical methods successfully applied in other fields, such as biosciences, are presented. The content spans from materials science to information science to reflect the cross-disciplinary nature of the field. A perspective is presented that offers a paradigm (codesign loop for materials design) to involve iteratively learning from experiments and calculations to develop materials with optimum properties. Such a loop requires the elements of incorporating domain materials knowledge, a database of descriptors (the genes), a surrogate or statistical model developed to predict a given property with uncertainties, performing adaptive experimental design to guide the next experiment or calculation and aspects of high throughput calculations as well as experiments. The book is about manufacturing with the aim to halving the time to discover and design new materials. Accelerating discovery relies on using large databases, computation, and mathematics in the material sciences in a manner similar to the way used to in the Human Genome Initiative. Novel approaches are therefore called to explore the enormous phase space presented by complex materials and processes. To achieve the desired performance gains, a predictive capability is needed to guide experiments and computations in the most fruitful directions by reducing not successful trials. Despite advances in computation and experimental techniques, generating vast arrays of data; without a clear way of linkage to models, the full value of data driven discovery cannot be realized. Hence, along with experimental, theoretical and computational materials science, we need to add a fourth leg to our toolkit to make the Materials Genome'' a reality, the science of Materials Informatics.
Index Terms
- Information Science for Materials Discovery and Design
Reviews
Anthony Joseph Duben
The development of new materials for manufacturing and in processing (such as new catalysts) has undergone a major transformation from the Edisonian hit-or-miss try-many-guesses approach that hopes to stumble onto the serendipitous discovery of a substantially improved material. Even though there is substantial documentation of applying classical factor-factorial statistical approaches in material formulations, the body of work ignores the problem of discovering which properties of a class of materials are important and how then to use or manipulate these properties to achieve the improvements desired. The approaches described in this book combine Bayesian statistical analysis, big data informatics analysis including visualization of correlations among properties, machine learning to uncover hidden properties, and combinatorics. Theoretical calculations of electronic properties of materials (often quantum mechanical using density functional theory) play an important role. This book presents reports given in a workshop sponsored by Los Alamos National Laboratory in 2014. The methods described originated in the human genome project and have been used in organic chemistry for prescreening candidate molecules in drug development. The application of these techniques in materials development is a much more challenging problem because these materials are usually solids in different crystal habits, consist of several elements often in nonstoichiometric proportions, and sometimes are limited to surface behavior or thin films on a solid substrate. This volume documents the progress made in applying these techniques. There are 14 chapters divided into three parts. The first part, with six chapters, is called "Data Analytics and Optimal Learning." After an introductory chapter on the state of the art and challenges, the remaining chapters in this part describe the theoretical basis and applied mathematics of the approaches used. Experimental design based on Bayesian inference is presented in chapters 2 and 3. The Bayesian approach is used to infer both parameters and appropriate models at the molecular/solid state phase level of material composition. Chapter 4 is devoted to small sample analysis since often it is impractical to prepare and test many different variations of the materials. Chapters 5 & 6 employ data visualization and multiscale modeling techniques to reveal correlated properties, coupled structural parameters, and hidden properties. Although the emphasis in this part is on theory, the contributors include substantial experimental evidence to illustrate their arguments. Parts 2 and 3 emphasize more directly the computational supports applied to examples taken from the laboratory. The two parts differ in the type of computations. Part 2, on high-throughput calculations, also has six chapters. Chapter 7 applies data mining techniques to improving the manufacture of parts by laser fusion of powders. Chapter 8 shows how the techniques are used in selecting the best dopants in cerium oxide catalysts used to split water into hydrogen and oxygen gases. Chapters 9 and 10 reverse the approach by creating databases using top-down first principles by computing the properties of materials using density functional theory calculations, which are then used in conjunction with experiments. Chapters 11 and 12 focus on solid state structures. In the first of the two, properties of materials with perovskite structures are correlated against the distortion modes of the solid state structures. In the second, machine learning is used to discover electronic signatures associated with phase stability of solids. The computational emphasis in Part 3 is combinatorics. Only two chapters are in this part. Combinatorics applied to solid state materials is much more challenging than in organic chemistry because of the possibility of large variations and continual variations in composition and the changes in properties that these variations can cause. In order to manage the computational complexity associated with combinatorics, the complexity of the materials must be reduced by clustering composition-property data and using simplex and statistical methods to achieve simplification. This book impressed me personally because four decades ago I worked for a company developing catalysts for processing petroleum using Edisonian techniques with only the slightest amount of experimental physical and chemical data and virtually no theoretical computations for guidance. There was nothing upon which to base a systematic protocol for improving the materials and processes. This book takes the theory and practice of experimental design of new materials to a new level. Online Computing Reviews Service
Access critical reviews of Computing literature here
Become a reviewer for Computing Reviews.