Atomic-scale Study Of Structure Prediction And Phase Transition Of Several Intermetallic Compounds And Metal Oxides

The crystal structure is one of the most fundamental and important information of a material,which directly or indirectly determines the mechanical,physical and chemical properties of the material,and therefore it is crucial to clarify the crystal structure of a material.However,the identification and prediction of crystal structure is a challenging task,because the structural phase space increases dramatically with the increasing number of group elements and the potential energy surface(PES)becomes very complex.In addition,there might exist different stable or metastable phases in the material,and phase transitions may occur between different phases under appropriate conditions,which greatly increases the difficulty of global or local energy minimum structure search and phase transition calculations.Although both structure prediction and phase transition calculations are associated with the PES of the phase space,they focus on different aspectes and face different difficulties.Structural prediction focuses on exploring the entire PES of the phase space and seeking global and local optimal solutions,so the challenge is how to explore the entire phase space efficiently and accurately.In contrast,phase transition simulations are more concerned with the accurate description of the PES along the phase transition path.However,the phase transitions are often thermally activated processes that require to overcome energy barriers,especially in solid-solid phase transitions,where the phase transition dynamics are slow,which places a higher demand on the accurate description of the potential energy surface.In this thesis,we combine high-throughput first-principles calculations,structural search genetic algorithms,molecular dynamics simulations,enhanced sampling of metadynamics,and machine learning potentials to focus on solid-solid phase transition kinetic simulations and thermodynamic calculations of titanium pentoxide,the crystal structure prediction,phase diagram calculation and local structure evolution during liquid-solid transition of a series of binary and ternary compounds.The key scientific issues have been systematically and thoroughly investigated,and the main innovative results are as follows:(1)Using on-the-fly active learning approach combined with the enhanced sampling technique of metadynamics,we constructed a machine learning interatomic potential with first-principles accuracy for the Ti3O5 system.Using the intensity of the X-ray diffraction peak as a collective variable,the reversible solid-state phase transition processes of the β-Ti3O5 and λ-Ti3O5 phases were simulated using metadynamics simulations,and the pressure-temperature phase diagram was constructed by calculating the Gibbs free energy.Through continuous tensile simulations of the βTi3O5 phase and the simulation of nucleation model with large-scale molecular dynamics simulations,we reveal that the phase transition from β-Ti3O5 to λ-Ti3O5 is due to the synergistic motion of atoms within the two-dimensional layers caused by local strain which eventually propagate to the interlayer due to strain accumulation.(2)In order to clarify the experimental structural enigma of the U-Fe-Zr ternary system,we have used genetic evolution algorithms to predict the crystal structures of the U-Fe-Zr ternary system and the U-Fe binary system with variable components structure prediction technique,and have systematically explored and studied the crystal structures,mechanical stability,thermodynamic and dynamical stability of series of binary and ternary compounds obtained from the search.A new ternary phase UFe3Zr2 with space group R3m was successfully predicted and resolved in the U-Fe-Zr ternary system,clarifying the experimentally reported but unknown crystal structure of the χphase(Fe0.5U0.18Zr0.32).In addition,we successfully predicted a variety of novel phases such as mP4-UFe,oC8-UFe,tI12-UFe2,tP6-UFe2,cP4-UFe3,and tI10-U4Fe,which were not found in the phase diagram of the U-Fe system,and the calculations confirmed that they all satisfied the thermodynamic and dynamic stabilities.(3)In addition to the study of the solid-solid phase transition process and the atomic motion mechanism,we also made an intensive study of the evolution of the local structure with the decreasing temperature.The evolution of the local structure of the molten states of the U-Zr system,UFe3Zr2 system and U-Fe system during the cooling down process has been elaborated using first principles molecular dynamics simulations.The pair distribution correlation function,structure factor,HA index,Voronoi tessellation index,local bond orientation order parameter,and bond angle distribution,and the diffusion behavior of the melt has been studied,and the diffusion equation of the molten state was established.(4)To address the problem that the crystal structure of the Pd-Nb binary system is not well understood,we systematically investigated the crystal structure,thermodynamic and dynamical stability,electronic structure and mechanical properties of the Pd-Nb binary system using variable component structure prediction based on genetic evolution algorithm combined with high-throughput first-principles calculations.Our calculations not only reproduce the experimentally observed compounds(oP24-Pd3Nb,tI8-Pd3Nb and oI6-Pd2Nb),but also predict five new ground state phases and nine metastable phases.The elastic constants calculations show a nearly linear relationship between the elastic modulus and Poisson’s ratio and the enthalpy of formation of the structures,while the elastic modulus shows a trend of increasing and then decreasing with increasing Nb content concentration.(5)To clarifiy the unclear parent-phase structure of recently experimentally evidenced near-ambient superconducting N-doped lutetium hydride systems,we have carried out a systematic density functional theory calculation to study the crystal structure,thermodynamic and dynamical stability,and optical absorption spectrum of lutetium hydride.It was clearly established that the phase with the fluorite structure FLLuH2 is the only phase that satisfies all the necessary conditions for experimental observations,i.e.,thermodynamic and dynamical stability,absence of pink-color photon absorption,and matching the position of the main peak of the experimental XRD.The presence of the remaining minor peaks of the experimental XRD is due to the presence of a small fraction of the RS-LuH phase with a rock salt structure in the experimental sample,which is phonon stable at both 0 GPa and 1 GPa pressures.