Estimation of transition temperature for doped iron-based superconductors based on crystal cell structure

From the experimental dataset on the superconducting transition temperatures Tc for 31 different superconductors of the doping iron-based oxy-arsenide systems, rough set theory data preprocessing methods and back propagation neural network(BPNN) combined with genetic algorithm(GA) for its parameter optimization were proposed in this paper. A model for estimating the Tc´s of the superconductors of the doping iron-based oxy-arsenide system was established by using their crystal cell structures, including lattice parameter a, lattice parameter c, crystal volume V and bonding parameter topological index H31. The results show that the predicted errors by leave-one-out cross validation (LOOCV) of GA-BPNN models are small. This study suggests that the crystal cell structure is an effective descriptor and GA-BPNN can be used as a powerful approach to foresee the Tc in development of doping iron-based oxy-arsenide superconductor, which can provide theoretical guidance for physical experiments and reducing the experimental blindness.