19th International CODATA Conference
Category: Knowledge Discovery

Investigation of Regularity in Materials by Comprehensive Material Database

Y. Kaneta (kaneta@q.t.u-tokyo.ac.jp), Y. Chen, A. Moteki, S. Wada, H. Kobayashi, School of Engineering, The University of Tokyo, Japan
S. Iwata, Graduate School of Frontier Sciences, The University of Tokyo, Japan
P. Villars, Materials Phase Data System, Switzerland

By employing a comprehensive material database, Pauling File[1], regularities between properties of compounds and their constitute elements for binary systems are investigated aiming to develop a new method to predict new materials. Current achievement is summarized as follows.  On the basis of the study of structural regularity in 1:1 (AB type) compounds by using structure map with atomic environment type (AET) and Mendeleev number (MN)[2], we extend the research to various compositions of binary compounds and investigate the distribution of AETs systematically. The correlation between MN and AETs types is found throughout all main compositions, especially well manifested in 1:3 (AB_3) compounds.  Furthermore, various atomic radius data in PF are examined by comparison with lattice parameters of crystal structures, and new radius data set for some elements group are proposed. Within many factors affecting crystal structure, lattice parameters, nearest neighbor distance in a structure and elemental radius are re-examined as geometrical factors. A clear relationship between these parameters in certain structure types of AB_3 compounds is investigated. The correlation between material properties is attempted by using regression analysis and artificial neural network. It is found some property such as melting point of a compound can be predicted from properties of its constituent elements. A series analysis on a sample data set of melting points, bulk modulus and energy gap for certain structure types (CsCl and NaCl type) shows the possibility to predict unknown properties based on the relationship among properties derived from known data set.

References

[1] P. Villars, M. Berndt, K. Brandenburg, K. Cenzual, J. Daams, F. Hukkiger, T. Massalski, H. Okamoto, K. Osaki, A. Prince, H. Putz, S. Iwata, J. Alloys and Compounds, 367 (2004), p.293-297. [2] P. Villars, K. Cenzual, J. Daams, Y. Chen, S. Iwata, J. Alloys and Compounds 367 (2004), p.167-175.