Research On QSPR Of Perovskites Compounds Using Data Mining And Quantum Chemistry Calculation

QSPR (Quantitative Structure-Property Relationships) is an important part of material science. The researchers extrapolate the properties of materials from their structures. QSPR comes from an auniversal postulation in chemistry:the molecular properties are determined by their structures. Therefore, how to establish quantitaive structure-property relationships model is one of the fudamental and hot issues in chemistry and material science. In principle, there are two parallel strategies for the solution of this problem:first-principles calculation and semi-empirical approach. Data mining, a multi-disciplinary research area, is a semi-empirical technology to find the unknown, hidden and interesting knowledge from the massive data. It has been recognized as a key research topic in database and machine learning. It has also aroused wide interest of scientific or industrial circle for its large potential application. In this work, the programe CASTEP based on the first principle and common data mining algorithms were used to calculate the properties of some perovskite-type materials:perovskite-type ionic conductor and organic-inorganic hybrid perovskites.Perovskite-type ionic conductors are attracting much interest in recent years because of their applications in oxygen sensors and solid oxide fuel cells (SOFCs). Oxygen sensors are widely used in metallurgy, automotive industry, medical and biological researches, etc. SOFCs have many advantages such as wide variety of fuel and long life among the various types of fuel cells. Furthermore, the development of SOFCs is highly important to develop environmental friendly power generator system. The perovskite-type oxides have been considered to be a potential material for solid electrolytes of SOFCs and have been widely studied.The oxide ion conductivity is an indispensable property of the electrolyte material used in SOFCs. Except for exterior influences such as oxygen pressures and operating temperature, the conductivity of the perovskite-type oxide is determined by its composition and structure. It is necessary to study the relationships between the molecular composition parameters and the oxygen ion conductivity of the perovskite-type oxides. Atomic properties and ionic conductivity data of perovskite-type oxides were collected from literatures and experiments. The relationship between the electrical conductivity and the atomic property was examined. The oxide ionic conductivities were predicted by using two approaches based on CASTEP first-principles calculations and three data mining methods, such as Partial Least Squares (PLS), Back Propagation Artificial Neural Network (BP-ANN), and Support Vector Regression (SVR). It was found that P/L (the ratio of O-O charge population to the O-O band length in B06 octahedron of ABO3 perovskites) has a quadratic curving relationship with Ln(σ) (logarithm of oxide ion conductivity) in some undoped and doped perovskite-type oxides. The results of machine learning indicate that the generalization ability of SVR is better than those of BP-ANN and PLS models for predicting Ln(σ).New materials with organic-inorganic hybrid perovskite structure are novel molecular-scale composites through organic and inorganic compounds' self-assembling. The layered perovskite framework comprised of metal halide octahedra provides a distinct set of advantages including good electrical mobility, mechanical and thermal stability, while organic component offers a number of useful properties including structural diversity and ease of processing et al. It's self-assembling hybrid perovskite structure of semiconducting inorganic sheets alternating with organic layers result in unique electronic, optical and magnetic properties. Recently, it has received considerable attention due to its promising application prospect.Optical property is one of indispensable characteristics of the organic-inorganic hybrid perovskite structure materials. It is determined by its composition and structure of the hybrid perovskites. It is necessary to study the relationships between the molecular composition parameters and the optical property of the hybrid perovskite. First-principles calculation and semi-empirical approaches were both used to compute the optical properties. The program CASTEP based on the first principle and the data-mining algorithms were applied in calculation of the optical property and lattice constants. The materials with organic-inorganic hybrid perovskite structure ((C4H9NH3)2SnI4 and (C6H5C2H4NH3)2SnI4) had been prepared and the optical properties had been studied by UV-Vis absorption. Compared with the experimental and calculated values, the advantages and disadvantages of two strategies were studied.