Rules-Driven Materials Design Using an Informatics-Based Approach

  • PDF / 340,179 Bytes
  • 6 Pages / 612 x 792 pts (letter) Page_size
  • 71 Downloads / 176 Views

DOWNLOAD

REPORT


0894-LL03-04.1

Rules-Driven Materials Design Using an Informatics-Based Approach Joan T. Muellerleile1, Kim F. Ferris2, Dumont M. Jones3, and Roger W. Hyatt1 1 Battelle Memorial Institute, Columbus, OH 43201; 2Pacific Northwest National Laboratory, Richland, WA 99352; 3Proximate Technologies, LLC, Columbus, OH 43209, U.S.A. ABSTRACT A rules-driven, informatics-based approach to multiply-constrained materials design is outlined, employing the example of polymer coating design for silica fibers. This approach to the inverse mapping problem of structure generation from design constraints and quantitative structure-property relationships (QSPRs) emphasizes design rule generation and analysis. Using this approach addresses several issues in new materials discovery: 1) factoring a larger design problem into tractable components, 2) integrating physical and non-physical requirements (such as cost), 3) identifying information gaps that must be resolved to complete a design, and 4) identifying situations in which a solution consistent with known information is not feasible. INTRODUCTION Our rules-driven approach to materials design, which we call Application-Driven Chemistry (ADC), encompasses tools and procedures for mapping chemical system requirements onto corresponding system compositions and/or molecular structures. ADC is a chemical information analysis platform in a semi-empirical framework; its purpose is to ascertain whether existing chemical information can lead to useful, and possibly new, chemical systems, and to direct the gathering of new knowledge as required. Specifically, our approach provides a database for storing generic information related to chemical system design, evolution tools for creating new chemical systems from requirements, and modeling tools for construction of composition-requirement relations. These capabilities are accessed through a scripting language interface designed for compact algorithmic expression. ADC does not replicate computational-chemistry calculation methods or visualization software, as both are readily available elsewhere. No a priori constraints are placed on the type of requirement or system under consideration, provided that sufficient information exists for rule generation. Requirements or imposed constraints may involve conventional physical properties such as boiling point, practical constraints such as cost, or structural constraints. ADC at present favors conventional organic or inorganic molecules including polymers, with limited coverage of multi-component systems and composite materials. The problem of chemical system design, which describes the intent of ADC, can be stated as: given a set of system property or performance requirements, determine the chemical system(s), whether currently existing or not, which meet those requirements. Conversely, experimental or “modeling” activities estimate properties given system definition(s). The degree of difficulty of chemical system design depends upon two factors, namely 1) scope, i.e. the range of potential chemistries