Computional Method For Ternary M/B/Si Funcitonal Compounds Within First-principles Calculations

Based on big data technologies,our focus is to develop functional materials databases serving for scientific materials research and highly improve the efficiency of data mining(DM),comprehensive modeling analysis,structure screening and computations,which is also the urgent project of computer and materials scientists.Meanwhile,with the developments and applications in the field of solid electrode materials,solar cell material and photoelectric conversions,the attentions for efficiency improvement has been paid to the studies of the ternary lithium compounds and explore the new materials for energy storage and conversion.Combining the density of functional theory(DFT)with big data analysis methods,in this dissertation,the ternary lithium compounds are investigated to understand the structural stability,mechanical property and electronic characteristic based on first-principles calculations.Parallel computations are realized for a variety of structures candidates and properties jobs in order to reduce the complexity and computational cost of screening and processing mode of data analysis.The method of high-throughput calculations is suggested to reduce the computational cost of structure searching and properties calculations.Structural selections from materials database is improved by the optimal strategy within parallel computing framework.The method for computing cost reducing is optimized by the periodic policy improvement and parallel computing framework.Within the framework of density functional theory and high-throughput computational methods,the algorithms of STRUC_VECT and VACA_OPT for structure screening is proposed for alkaline metal boron silicides.The results show that the model of high-throughput method we used here can effectively improve the efficiency of calculations and reduce I/O resource consumption.A simple and effective method for effective selections of electronic properties of Li-based materials from materials database is developed by the analysis of material genome and localized orbital optimizations and the model of element permutation and combination.The algorithms of GENE_TRANS and HAMI_OPTI are developed via the technique of decomposing domain computation,optimized solution of genetic characteristics and high-throughput functional simulations,which can help to resolve the massive calculations and reduce the computational cost for structural screening.Using the data acquisition method of high-throughput calculations,alkaline metal silicon nitrides were investigated systematically based on the structural model constructing of Li-based compounds,density functional theory and material Genome database.The modified algorithms of ATOM_OPT and ELECT-OPT are suggested by algorithm optimizations to reduce the physical cycle index of traditional algorithm and computational cost.This method can quickly locate the needed structures by the enforcement of localization algorithm optimizations,which greatly helps the structural searching to avoid the presence of local energy minimization.This method also improves the feasibility and effectiveness of solution optimization algorithms and simplifies the computational flow.