Neural network and experimental thermodynamics study of YCrO3-δ for efficient solar thermochemical hydrogen production

Abstract

ABO3-type perovskites have been demonstrated as promising redox materials for solar thermochemical H2 production. In this work, a new energy material screening method based on the neural network system is designed as a feasible way to search for promising H2 production materials. The predicted oxygen vacancy formation energy of the selected YCrO3-δ is 4.199 eV, hinting excellent H2 production potential. Thermogravimetric analysis shows that the doping of Zr into YCrO3-δ improves oxygen formation capacity, leading to the maximum δ of 0.106. The molar enthalpy and entropy of YCr0.75Zr0.25O3-δ have the positive relationship with δ, and the maximum values of which are 273.7 kJ mol−1 and 164.9 J mol−1 K−1 respectively. Based on the equilibrium thermodynamic principle, the peak H2 yield is predicted to be 444.6 μmol g−1. Considering material kinetic limitation, gas-solid heat recovery and parameter sensitivity, the maximum H2 production efficiency of YCr0.75Zr0.25O3-δ is 17.3%. The combination of neural network and material thermodynamics provides a new pathway to design promising H2 production materials, and the screened YCrO3-δ presents excellent solar thermochemical H2 production capacity.