Construction of high-throughput computation platform for random alloys (HCPRA) and its applications

Abstract

Materials play not only an important role in academic research but also a rather important role in daily life. The discovery of a new material can make a significant contribution to both scientific research and society development. For example, the discovery of graphene in 2004 has attracted enormous attention in academic circle and graphene has shown promising applications in electronic and optical devices. However, the period of discovering a new material is too long to satisfy the demand of society. To speed up the process, Advanced Manufacturing Partnership was proposed in 2011 by US President Obama. This plan has a key part: Materials Genome Initiative (MGI). MGI is made up with three part, (1) construction of high-throughput computation platforms, (2) construction of high-throughput experimental platforms and (3) construction of databases. The high-throughput computation and experimental platforms have promising applications in investigating known materials as well as discovering new materials with high efficiency since they can produce a large amount of data in a short term. Databases can provide convenience for data analysis and machine learning that is a prevailing way to design new materials. In this paper, we present our new designed High-throughput Computational Platform for Random Alloys. The platform is built by using Fortran and Shell languages and its main function is to calculate mechanical and thermodynamic properties of random alloys in large quantities automatically by employing the Vienna ab-initio Simulation Package (VASP). There are two main problems that are needed to be solved in the construction of High-throughput Computational Platform for Random Alloys. The first one is the creation of structures for random binary alloys. Another is to set the comparable computational parameters for all calculations automatically. To create structures for binary random alloys, we employ Alloy Theoretic Automated Toolkit to obtain a set of special quasi- random structure templates. Then, we collect all structure information for elementary substance from literature or by performing the first principles calculations. Finally, we design the platform to create special quasi-random structures (POSCAR) for all kinds of binary random alloys by using the templates and structure information of elementary substance combine with the Neumann-Kopp rule. In order to set comparable computational parameters for the VASP calculations, we write a code to read the structure information from POSCAR and use the information to set a comparable k-mesh for all calculations. The INCAR file is created by using the information from POSCAR and POTCAR. After all input files are generated, the platform employs VASP to perform the first-principles calculations. The platform is also designed to extract and analyze calculation results as well as print a calculation report and save all the calculation data into a database. The platform also has some useful functions such as error information-outputting system and error-skipping system. The function is rather important since convergence problem is inevitable in the first-principle calculations especially in high throughput calculations. To test the efficiency of the platform, elastic constants of 47 pure metals are investigated which indicates a computer with eight cores can finish all calculations in one day. And all calculated results match those in the literature well. By using the platform, the mechanical properties of zirconium based binary alloys are investigated and we find the elastic constants and elastic modulus can be well described by a parabolic function.