Automation of first-principles calculations for screening functional oxides

Abstract

Through the last decade, numbers of technologies in electronic applications have faced the stagnation due to the constrained properties of conventional materials. Especially for semiconductor devices, fabrication approaches using existing materials no longer guarantee further improvement of device performance, so demands for new superior materials have emerged. Developing new materials design that has unusual properties is also the key to realize future technologies such as transparent and/or high energy-efficient devices. First-principles calculation based on density functional theory is the most powerful tool to predict physical properties of unexplored materials. In addition, as computing power is greatly increased by Moores law, the high-throughput screening using first-principles calculation became practicable recently. However, a high-throughput calculation study still requires huge human resources if one uses typical approaches. To conduct reliable and highly efficient high-throughput calculations, a well-established automatization of computation procedures is inevitable. In this dissertation, we present three individual high-throughput studies assisted by in-house automation codes which are optimized for each targeted applications. First, we screen novel high-k dielectric oxides by calculating band gap and static dielectric constant of most binary and ternary oxides from ICSD structure database. As a result, we find new candidate materials such as c-BeO for next-generation high-k dielectrics. Second, we built single-dopant property database of doped ZnO through calculating all possible dopant configurations of 61 elements from the Periodic Table. We screen dopants for n-type and magnetic applications and identify the trends of dopants behaviors considering their chemical variance. Lastly, we explore p-type transparent conducting oxides by examining band gap, hole effective mass, thermodynamic stability and behaviors of intrinsic defects. We identify the constraint of former studies and suggest new promising candidates satisfying strict criteria. By developing reliable automation codes, we successfully explore massive materials that is unreachable by experiments. The suggested new materials and large property database will serve as a useful reference to make breakthroughs in various applications.