| 초록 |
While crystal structure of most unary and binary compounds were thoroughly investigated by X-ray crystallography, only about 16% and 0.6% have been revealed within the ternary and quaternary (multinary) inorganic materials, respectively. This means that at least millions of multinary compounds are yet to be revealed. Experimental data on the crystal structure accumulate at a pace of 6,000 per year and so it will take many decades to uncover a large portion of multinary domain in the structure database. On the other hand, theoretical prediction of crystal structures can expedite discovery of novel materials from uncharted chemical space. In this presentation, we develop crystal structure prediction of completely unknown compounds by employing neural network interatomic potentials as a high-fidelity surrogate model of the density functional theory. Since the crystal structure is unknown, we construct the training set using disordered phases such as liquids and amorphous structures. We confirm that the neural network potentials can rank various metastable structures properly. By combining the neural network potential and evolutionary algorithm, we search the equilibrium structure of various test compounds. |
