The Exploration Of Preparation Of High Quality FeSe Thin Films

In the iron-based superconductor family,the chemical composition and structure of FeSe are simplest.But FeSe system's T_c can be remarkably enhanced,indicating FeSe system serves as an excellent candidate for studying the mechanism of iron-based superconducting.In order to study and optimize the preparation conditions of FeSe thin films,various growth conditions are tried based on our group's previous work.The growth parameters of FeSe thin film PLD preparation were systematically and carefully explored.Moreover,machine learning method is used to build model describing relationship between growth conditions and T_c.The main contents include:1.The optimum deposition temperature of FeSe film was explored.The suitable growth temperature of the FeSe/Ca F₂ is 350?.The scanning electron microscope images of the samples grown at different temperatures are compared.2.The growth of FeSe films on different substrates was studied.High-quality thin films were successfully grown on CaF₂,LiF,Sr TiO₃,MgO,BaF₂,TiO₂?100?,LaAlO₃,MgF₂,Nb:SrTiO₃,LSAT,?Sr,La?.The T_c of the films on the CaF₂,LiF and SrTiO₃substrates is superior to those of the other substrate films,which are 14 K,12 K and11.5 K,respectively,which are all higher than the T_c of the FeSe bulk.It is found that not the epitaxial stress but the proportion of Fe,Se leads to different superconductivity of FeSe films with varying thickness.3.Through systematic statistical analysis of the sample data of more than 1500thin films,we found that there is a positive correlation between the film sample T_c,the lattice constant c and RRR.4.The application of machine learning in materials science is introduced.The model analyzing the relationship between growth conditions and T_c of FeSe superconducting thin films are established by machine learning.SVR and GBDT models were tried out respectively.There are many growth conditions in PLD preparation of samples and the control is complicated.It always takes a long time to find suitable growth conditions.Machine learning is often used for data mining and pattern recognition.If the relationship between growth conditions and T_c can be modeled,the model can help to accelerate the exploration of the best growth conditions when growing new materials.As a preliminary attempt,the growth conditions of more than 1500 FeSe films are studied.The relationship between the growth conditions and the T_c was analyzed by using the SVR and GBDT models in machine learning.The GBDT model has the highest score.The prediction on which substrates are more suitable for growth is consistent with many reports that CaF₂ is easier to grow the iron-based superconducting thin film than the oxide substrate.