An informatics software stack for point defect-derived opto-electronic properties: the Asphalt Project
- PDF / 747,719 Bytes
- 7 Pages / 612 x 792 pts (letter) Page_size
- 50 Downloads / 142 Views
rtificial Intelligence Prospective
An informatics software stack for point defect-derived opto-electronic properties: the Asphalt Project Jonathon N. Baker, Preston C. Bowes, Joshua S. Harris, and Douglas L. Irving, Department of Materials Science and Engineering, North Carolina State University, 911 Partners Way, Suite 3002, Raleigh, NC 27695, USA Address all correspondence to Douglas L. Irving at [email protected] (Received 14 January 2019; accepted 6 August 2019)
Abstract Computational acceleration of performance metric-based materials discovery via high-throughput screening and machine learning methods is becoming widespread. Nevertheless, development and optimization of the opto-electronic properties that depend on dilute concentrations of point defects in new materials have not significantly benefited from these advances. Here, the authors present an informatics and simulation suite to computationally accelerate these processes. This will enable faster and more fundamental materials research, and reduce the cost and time associated with the materials development cycle. Analogous to the new avenues enabled by current first-principles-based property databases, this type of framework will open entire new research frontiers as it proliferates.
Introduction In 2011, the National Science and Technology Council (NSTC) announced the Materials Genome Initiative for Global Competitiveness.[1] This white paper outlined the seven stages of materials development, which has traditionally taken 10 to 20 years from the initial investigation of a material to its first industrial use. These stages are shown in Fig. 1. The NSTC envisioned a massive collaboration to accelerate all stages, which would enable data and algorithms to be shared among computational and experimental researchers and industry. Thus far, the vast majority of effort has yielded tools aimed at computationally accelerating the first three stages with a general focus on bulk property optimization. These efforts resulted in the creation of repositories like the Materials Project,[2] OQMD,[3] and AFLOW,[4] which have built up vast databases of bulk material properties. These repositories can be screened to find good candidate materials for a given application, or data mined and used with artificial intelligence (AI) algorithms to predict new materials with desired properties.[5,6] This has resulted in a number of significant advances in areas such as rechargeable batteries, thermoelectrics, and photovoltaics.[7,8] Nevertheless, much of the original vision and promise of the materials genome initiative has yet to materialize in systems where the properties of interest depend on dilute concentrations of point defects, as in opto-electronics. This is even more problematic considering that stages 2 and 3 are the most timeconsuming and expensive steps for optimizing such properties, and are most critical for taking a candidate material system from the laboratory to commercial products and devices.
This is partially an architectural problem, and partially a cost-benefits pro
Data Loading...
