A new methodology of systematic design of new materials for various applications is presented in this paper. In particular, a large number of candidate compounds that are formed by all possible combinations of the targeted elements in the periodic table are first screened and shortlisted by artificial neural network techniques. Then the quantum mechanics computation is employed to evaluate the promising candidates selected from the first step. Finally experiments are performed to further examine the computation results. In the present work, we apply this methodology to the study of semiconductors of binary (III-V and II-VI) and ternary (I-III-VI2 and II-IV-V2) compounds. Firstly, we systematically study all possible binary and ternary compounds by using pattern recognition and perform prediction of two important properties, namely band gap energy and lattice constant, with the artificial neural network model. Candidate semiconductors are then selected. On the basis of the above study, we perform first principles quantum mechanics computation for some promising II-VI binary candidates. The first principles study of the ternary candidates will be conducted in the near future, and the experiment study of the binary compounds is ongoing. The model predicted new compounds as well as the developed design methodology may be of interest to general materials scientists including these of smart materials research.

Date Published: 13 November 2002
PDF: 12 pages
Proc. SPIE 4934, Smart Materials II, (13 November 2002); doi: 10.1117/12.473144

Published in SPIE Proceedings Vol. 4934:
Smart Materials II
Alan R. Wilson, Editor(s)