Computational Studies On The Structure And Properties Of Multi-system Based On Adaptive Genetic Algorithm

Guiding experimental synthesis through theoretical predictions can speed up the research and development of new materials,which is of great significance to the search for new functional materials.Crystal structure prediction starting from the chemical composition alone has always been one of the long-term challenges of theoretical solid physics,chemistry,and materials science.To focus on this problem,this thesis perfects an adaptive genetic algorithm(AGA)for predicting the structure of crystals/surfaces/interfaces,and summarizes our computational research on the structure and properties of materials.The focus is on the application of AGA in ternary Li-Ni-B,Fe-Ni-B and rare-earth-free magnetic materials.We proves that AGA can accurately predict the crystal structure and quickly obtain the structure and phase stability of different components in a multi-component system.We also introduces a Li substitution method to study the complex crystal structure of NaFeP04 cathode material.This article contains the following seven chapters:The first chapter first introduces the paradigm of material design and important concepts of crystal structure,and lists the calculation algorithms that have been widely used for crystal structure prediction in the past two to three decades.Secondly,two methods for calculating energy are introduced:density functional theory(DFT)and empirical potential,which pave the way for the research of this article.In the second chapter,it focuses on the AGA developed by the Kai-Ming Ho's research group and the fixed symmetry genetic algorithm led and perfected by myself.AGA combines the structural relaxation rate of classical potential with the accuracy of DFT calculation in an adaptive and iterative manner.While maintaining the accuracy of the DFT,AGA is much faster than the complete DFT-GA and provides a useful tools to study the structure of complex materials containing a large number of atoms.Subsequently,two methods for judging structural similarity are introduced:atomic cluster alignment method and fingerprint function method.Then,some machine learning tools and methods are briefly introduced.Finally,several methods for judging structural stability based on first-principles calculations are introduced.In the third chapter,the use of AGA to search for new materials of alkali metal-transition metal borides(A-T-B)with different motifs is introduced.The structural diversity of rare-earth and transition metal borides indicates that A-T-B have tremendous promise to exhibit a variety of crystal structures with different dimensionalities of T-B frameworks.On the other hand,the A-T-B ternary systems are severely underexplored because of the synthetic challenges associated with their preparation.In this chapter,we have computationally discovered several new phases(stable phase and metastable phase with lower energy)in the Li-Ni-B ternary system.The newly-discovered LiNiB,Li2Ni3B and Li2NiB phases expand the existing theoretical database,and re-constructed the convex-hull surface of Li-Ni-B system.The lowest energy structure of LiNiB compound with layered motif predicted by AGA was successfully synthesized experimentally.According to our electrochemical calculations,LiNiB and another predicted layered Li2NiB compound have great potential as anode materials for lithium batteries.The Li2Ni3B compound with the space group P4332 is predicted to crystallize in a cubic structure composed of distorted octahedral units of BNi6,which is isostructural to two non-centrosymmetric superconductors Li2Pd3B and Li2Pt3B.In the fouth chapter,by searching the composition space of Fe-Ni-B,we predict an energetically and dynamically stable FeNiB2 compound.This system belongs to the class of highly responsive state of material,as it is very sensiti ve to the external perturbations.This state is also characterized by a high level of spin fluctuations which strongly influence possible magnetic long-and short-range orders.Furthermore,we demonstrate that these antiferromagnetically dominating fluctuations could lead to the appearance of spin mediated superconductivity.The obtained results suggest a promising avenue for the search of strong spin fluctuation systems and related superconductors.In the fifth chapter,using previously developed structural LiFePO4 database,we examine the low-energy crystal structures in NaFePO4 system after replacing Li with Na.The known maricite and olivine NaFePO4 phases are reconfirme,and an unreported phase with energy between them is identified by our calculations.Experimentally,LiFePO4 has been highly successful as cathode in Li-ion batteries because its olivine crystal structure provides a stable framework during battery cycling.However,in NaFePO4,maricite replaces olivine as the most stable phase.And the maricite phase is experimentally found to be electrochemically inactive under normal battery operating voltages(0-4.5 V).We find that partial substitutions of Na with Li can stabilize the olivine structure and may be a way to improve the performance of NaFePO4 cathodes.Besides,the thermodynamic stability of olivine type LixNa1-xFePO4 can be further improved at finite temperatures.In the sixth chapter,we develop an open-access database that provides a large array of datasets specialized for magnetic compounds as well as magnetic clusters.Our focus is on rare-earth-free magnets.Available datasets include(?)crystallography,(?)thermodynamic properties,such as the formation energy,and(?)magnetic properties that are essential for magnetic-material design.Our database features a large number of stable and metastable structures discovered through AGA searches.Many of these AGA structures have better magnetic properties when compared to those of the existing rare-earth-free magnets and the theoretical structures in other databases.Our database places particular emphasis on site-specific magnetic data,which are obtained by highthroughput first-principles calculations.Such site-resolved data are indispensable for machine-learning modeling.We illustrate how our data-intensive methods promote efficiency of the experimental discovery of new magnetic materials.Our database provides massive datasets that will facilitate an efficient computational screening,machine-learning-assisted design,and the experimental fabrication of new promising magnets.In the last chapter,I summarize the work of the dissertation and make an outlook for future study.