Databases
The voluminous data generated by combinatorial synthesis teamed with rapid characterization must be organized, structured, and annotated in order to provide the basis for enhanced materials discovery and development. Furthermore, legacy data from prior investigations must be integrated into this picture to make the most of existing materials knowledge, which raises questions not only of data volume but particularly of consistent, manageable data structures.
Once data is organized in structured databases which enable its integration, the challenge is to extract value from it in the form of fundamental insights and/or in quantitative relationships. These goals define an emphasis on data filtering, to categorize data sets, to organize and categorize them, and to prioritize the data. Once this is accomplished, powerful methods of data mining can be applied to identify correlations and relationships which benefit the materials engineering task, e.g., by revealing subtle relationships that simplify models and/or convey fundamental mechanistic insights as guidance for subsequent materials discovery and optimization activities.
With databases of even moderate size, data itself delivers little value unless the existing in the data are made explicit and usable through generation of models which reflect the behavior of the data. The filtering and mining operations which are part of data analysis lead naturally to model generation. Validation and refinement of models constitute another important component of the modeling process. Modeling is also essential in order to enable optimization of materials properties, and particularly when the optimization involves multiple metrics and their generally nonlinear contributions to a value metric for the materials and associated synthesis processes.
Materials Development Workstation
The materials engineer carries the responsibility for materials discovery and development, and as such needs means to manage the methods of combinatorial experimentation and materials informatics efficiently. Essentially, this means that the engineer needs a powerful design and analysis environment which enables not only experimentation, database generation and analysis, and modeling, but also visualization of complex data, multivariate optimization, and iteration of the experiment-analysis cycle in a way which exploits the learning achieed in each iteration.