Global Optimization Algorithms In Molten Salt And Metallo-boron Clusters

Energy is one of the major challenges nowadays.The environmentally friendly and nuclear power has been considered as one of the best power supplies ever since it was born.Due to its safety and efficiency,the molten salt reactor（MSR）was elected as one of the six most promising Generation IV nuclear reactors.Theoretical approaches,especially the molecular dynamics（MD）,play a vital role in understanding molten salts（MS）since experiments are dangerous because of the radioactivity.To accurately simulate MS at a long time scale proper force-fields are essential.The polarizable ions model（PIM）can precisely describe the interactions of MS.In this work we developed a molten salt force-field fitting（MSFF）program and studied both the microscopic and electronic structures of ThCl₄-LiCl MS system.In ThCl₄–LiCl MS,Th⁴⁺ions are bridged corner-,edge-and face-sharing Cl^-,the coordination number of Th⁴⁺ranges from 5 to 7,with 6-fold coordinated ThCl₆^2-being the predominant one at all compositions.The lifetime of the Th⁴⁺first solvation shell is shorter compared with the Th⁴⁺in ThF₄–LiF MS.To solve the energy challenge,the development of new materials is also essential.Due to its rich structural properties and the uniqe electron deficiency characteristic,boron and boron-based materials have gained increasingly attractions during the past few years.Global minimum search is crucial to investigate boron clusters.In this article,we introduce an automated global minimum search program named Tsinghua Global Minimum 2（TGMin2）which is the successor of the original TGMin.We have introduced several new features and improvements into TGMin2.We proposed a symmetric structure generation algorithm which can produce good initial seeds for small-and medium-size clusters.The duplicated structure identification algorithm and the structure adjustment algorithm of the original TGMin are further improved.In order to get the simulated photoelectron spectrum（PES）automatically,we also implemented a standalone program named AutoPES,which can simulate the PES and compare it with experimental results automatically.We have demonstrated that TGMin2 and AutoPES are powerful tools to study nanoclusters.With joint photoelectron spectroscopy and theoretical calculations,in this work we report two rare low oxidation states lanthanide-doped boron clusters,PrB^3–and PrB^4–.Chemical bonding and electronic structure analyses in localized coordinate system suggest that denoting two electrons to the B^3–moiety and one electron to the B^4–moiety from Pr are sufficiently enough to form stable compounds,thus making PrB^4–a rare divalent lanthanide compound and PrB^4–the first observed monovalent lanthanide compound.Understanding interactions and structural properties at the atomic level is often a prerequisite to the design of novel materials.Theoretical studies based on quantum-mechanical first-principles calculations can provide this knowledge,but at an immense computational cost.In recent years,machine learning has been successful in predicting structural properties at a much lower cost.In this work we propose a simplified structure descriptor with no empirical parameters,“k-Bags”,together with a scalable and comprehensive machine learning framework that can deepen our understanding of atomic properties of structures.This model can readily predict structure-energy relations that can provide results close to the accuracy of ab initio methods.The model provides chemically meaningful atomic energies enabling theoretical analysis of organic and inorganic molecular structures.Utilization of the local information provided by the atomic energies significantly improves upon the stochastic steps in our evolutionary global structure optimization,resulting in a much faster global minimum search of molecules,clusters,and surfaced supported species.