Informatics for Combinatorial Experiments: Accelerating Data Interpretation *
- PDF / 332,630 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 31 Downloads / 187 Views
0894-LL09-01.1
Informatics for Combinatorial Experiments: Accelerating Data Interpretation* M. Stukowski, C. Suh, K. Rajan, P. D. Tall1, A. C. Beye2, A. G. Ramirez3, W. O. Soboyejo2, M. L. Benson4, P.K. Liaw4 Department of Materials Science and Engineering, Iowa State University, Ames, IA 50011, U.S.A. 1 Department of Physics, Université Cheikh Anta Diop, Dakar, Sénégal 2 Department of Mechanical and Aeronautical Engineering, Princeton University, Princeton, NJ 08544, U.S.A. 3 Department of Mechanical Engineering, Yale University, New Haven, CT 06520, U.S.A. 4 Department of Materials Science and Engineering, University of Tennessee, Knoxville, TN 37996, U.S.A. ABSTRACT Combinatorial experiments provide a means of generating large amounts of experimental data; however that does not necessarily lead to high throughput interpretation of that data. In this paper we provide a brief summary of how one can use informatics techniques to accelerate data interpretation from high throughput experiments. We provide examples from high throughput nanoindentation and diffraction experiments. INTRODUCTION Using modern combinatorial techniques to develop large materials databases is rapidly becoming essential in advance materials design [1-3]. In order to interpret the data produced in large experiments in a useful fashion, it is necessary to not only assess one attribute from a combinatorial experiment but multiple attributes that collectively contribute to the final “signal” of a material characteristic. A good example is spectral screening (such as diffraction) of combinatorial libraries (Figure 1). This essentially poses the problem of applying multivariate analysis techniques to high throughput screening data. In this paper we demonstrate the use of Principal Component Analysis (PCA) as an important tool for accelerated screening of combinatorial data. We shall provide two examples. One deals with the nanoindentation studies of combinatorial libraries of thin films, where we are trying to quickly assess correlations between disparate parameters such as alloy chemistry, modulus and surface roughness. The second example deals with rapidly screening hundreds of diffraction spectra from in-situ experiments to quickly identify specific sets of spectra that capture information on microstructural changes during deformation.
*
Joint contribution from NSF International Materials Institute (IMI) program: CoSMIC-IMI: Iowa State University; US/Africa - IMI: Princeton University and ANSWER-IMI: University of Tennessee
0894-LL09-01.2
Figure 1. Schematic of research overview and the research approach for an informatics approach to the interpretation of very large multivariate spectral data sets. EXPERIMENTAL DETAILS AND DATA ORGANIZATION Since there are many variables used in nanoindentation experiments for shape memory alloys [4], the appropriate use of multivariate analysis is crucial to easily identify property-composition relationships in the field of materials informatics. Here we apply PCA to rapidly screen the multivariate nanoin
Data Loading...
