First-principles Calculation Study Of Novel Multicomponent Semiconductors Based On Materials Genome Database

In the past 70 years,functional devices based on semiconductors such as solar cells,luminescent devices,photodetectors,field-effect transistors have attracted wide attention.Recently,to improve the performance of these devices,numerous studies have focused on the design of novel semiconductor materials,among which multicomponent semiconductors are of great importance.Compared with elemental and binary semiconductors,multicomponent semiconductors have a high degree of freedom in composition and structure,which enriches their properties but also increases the complexity of the study.The traditional "trial and error" method often needs a higher cost and longer period to study such materials.Since 2010,the emerging materials genome database and high-throughput calculations make up for this deficiency and greatly accelerate the design of new materials.Based on the emerging methods,we can design a series of novel multicomponent semiconductor materials with excellent performance.Meanwhile,based on the high-throughput calculation data,we can train the relationship model between component elements,crystal structure,and semiconductor properties,which provides a more efficient method for screening stable semiconductors with specific properties.The thesis is composed of seven chapters.In the first chapter,we briefly introduce the development of materials genome database,first-principles high-throughput calculations,perovskite,zinc-blende-and wurtzite-derived multicomponent semiconductors.The second chapter briefly introduces the basic theories and computational methods involved in this thesis,including the density functional theory,methods for measuring thermodynamic stability,correction in formation energy calculations,and metrics used for evaluating classification modelsThe third chapter introduces the software developed by me,named HTMP(Highthroughput-calculation of Thermodynamic-stability of Multicomponent-semiconductor Python Package).The software includes three modules,including automatic VASP management,thermodynamic stability calculation,and data visualization.The modules of HTMP are low coupling and can flexibly perform the material design and screening based on the target properties.Besides,we introduce machine learning algorithms for the generation of the reasonable lattice constants of the initial structure,which can reduce the time-consuming of structure optimization.The fourth chapter discusses the first-principles high-throughput calculations of about 60,000 ABX3 single perovskites,980 A2B+B3+X6 halide double perovskites,and 78 layered double perovskites.Firstly,980 A2B+B3+X6 double perovskites are constructed through the combination of different component elements.The energies above the convex hull show that only 112 of 980 double perovskites are stable and 27 double perovskites that have been predicted to be stable in the literature are actually unstable after considering more competing compounds.The stability of these double perovskites is determined mainly by A,X,and B+elements and increases gradually as A becomes heavier(from Li to Cs)and X becomes lighter(from I to F).The bandgaps are determined mainly by X,B+,and B3+elements,decreasing monotonically as X becomes heavier while changing non-monotonically as B+and B3+change.The bandgaps of the 112 stable halide double perovskites are in the range of 2.02 eV-9.48 eV,and most of them are greater than 3 eV.Therefore,these halide double perovskites are not suitable for photovoltaic devices.However,there are many halide double perovskite pairs with lattice mismatch less than 1%and bandgap difference greater than 0.5 eV.These double perovskite pairs can form highly miscible alloys to realize band engineering.The 980 double perovskites mentioned above only consider a small part of the element combinations that satisfy the valence balance.We try to figure out other element combinations that may form stable double perovskite.It is infeasible to solve the problem directly through high-throughput calculations.Therefore,we first calculate about 60,000 ABX3 single perovskites.A highly accurate machine learning classification model is trained using these data.The model shows accurate predictions of the thermodynamic stability of experimentally synthesized single and double perovskites(96%accuracy).Finally,68,278 thermodynamically stable double perovskites are screened out from about 4 million element combinations.Besides,we explore a series of<111>oriented layered halide double perovskites and find Cs4MnSb2Cl12,Cs4CuSb2Cl12 and Cs4ScSb2Br12 are thermodynamically stable.The fifth chapter turns to the study of 1008 Ⅰ-Ⅲ-Ⅳ2-Ⅴ4 quaternary pnictides.Their crystal structures include two kinds of zinc-blende-derived structures(Kesterite and Stannite)and two wurtzite-derived structures(Wurtzite-Kesterite and WurtziteStannite).The total energy calculations show that the ground-state structures of most nitrides are wurtzite-derived structures,and those of phosphides and arsenides depend on the Ⅰ elements.Thermodynamic stability analysis shows,the stability is mainly determined by Ⅰ,Ⅴ,and Ⅳ elements,while the influence of Ⅲ elements is small;20 of 1008 compounds are thermodynamically stable,and 16 compounds have never been reported previously.The analysis of the electrical and optical properties of 20 stable compounds shows that 13 compounds are direct bandgap semiconductors,four of them have weak optical absorption coefficients due to forbidden transition,and the other nine have strong optical absorption coefficients.Further,we find 23 kinds of Ⅰ-Ⅲ-Ⅳ2-Ⅴ4/ⅢⅤ heterojunction combinations with lattice mismatch less than 2%.These heterojunctions are expected to be useful for optoelectronic devices.The sixth chapter systematically explores the thermodynamic stability,groundstate structures,and bandgaps of the quaternary chalcogenide compounds Ⅰ2-Ⅱ-Ⅳ-Ⅵ4.These compounds have diverse crystal structures,such as diamond-like structures and non-traditional structures.Thousands of Ⅰ2-Ⅱ-Ⅳ-Ⅵ4 may be designed through choosing different elements,but the thermodynamic stability and ground-state structure of these compounds are unclear.Therefore,we select seven different crystal structures which are predominant in the experiment and consider 2079 compounds with different element combinations.The results show that 472 of 2079 compounds are thermodynamically stable.The ground-state structures of these stable compounds are mainly determined by Ⅰ and Ⅱ elements.For example,when Ⅰ = Cu and Ag,Ⅱ=Be and Mg,the ground-state structures of these compounds are diamond-like structures;whenⅡ=Sr and Ba,the ground-state structures of these compounds are non-traditional structures.The electronic structure calculation shows the bandgaps can be tuned in the range of 0 eV-4.2 eV through combining different Ⅰ and Ⅱ elements.The seventh chapter reviews all the chapters and gives prospects of potential improvement.