Computational materials design aims at accelerating the discovery of materials with desired properties, thereby going beyond the standard trial-and-error approach in research and development. In this seminar, I will first describe the application of first-principles methods to searching for new p-type transparent conducting compounds as an example of design of materials with target functionality. I will then turn to the application of electronic structure schemes, along with optimization algorithms, to address the problem of predicting the crystal structure of new materials just starting from the chemical composition.
Optimum p-type transparent conducting materials must satisfy several design criteria: (a) possess a large concentration of hole-producing defects, (b) have high hole conductivity, and (c) have an optical absorption edge above the transparency threshold. We selected the Cu3VO4 and Ag3VO4 oxovanadates from a pool of candidate noble-metal oxides and applied ab initio methods, based on density functional theory, to assess whether they meet the above design principles. We predicted that Ag3VO4 (i) is p-type with ~1014 cm-3 hole concentration at room temperature along with a very low concentration of hole-killing defects, (ii) has a hole effective mass lower than that of the prototypical p-type TCO, CuAlO2, and (iii) is on the verge of transparency (i.e., transparent to red light).
Crystal structure prediction, given only the elemental components of a solid and without constraints on the lattice vectors and atom positions, is a central problem in solid state physics. This problem requires a global space-group optimization (GSGO) of the total energy of a solid as a function of the crystal degrees of freedom. Evolutionary algorithms are powerful global optimization methods and I will describe a procedure based on an evolutionary algorithm to address the GSGO problem. I will first illustrate the application of this optimization scheme to selected binary systems with fixed compositions, including Cd-Pt, Al-Sc, and Pd-Ti. I will then extend this evolutionary algorithm to predicting the compositions, as well as the crystal structures of the thermodynamically stable phases of a solid system, starting from the elements that compose it. As an example, I will apply this variable-composition GSGO algorithm to predicting the stable compounds of the Al-Sc alloy.
