Data Assimilation Between Experimental and Crystal Growth Simulation on REBCO Thin Films

We have developed a Monte Carlo (MC) simulation for REBa2Cu3Oy(REBCO) + BaMO3(BMO) thin film crystal growth. However, identifying parameters that accurately reproduce the experimentally determined crystal grain size has remained a challenge. The data assimilation between the simulation and experiment is crucial as BMO nanostructures significantly impact in-field critical current density. In this study, we utilized Definitive Screening Design (DSD), a powerful type of Design of Experiments, to efficiently search for optimal simulation parameters. We targeted three key variables: REBCO-REBCO bonding energy (ERR), desorption energy from the substrate (Edes), and deposition rate (vDR). We conducted MC simulations based on the DSD. The DSD successfully identified an optimal condition in only nine simulation runs, which is in strong agreement with the experimental target. The optimized parameter set was determined to be (ERR,vDR,Edes) = (1530 K, 100 nm/h, 5000 K). Furthermore, the REBCO surface morphology simulated under these conditions confirmed the expected 2D island growth mode, consistent with experimental observations. This result demonstrates that DSD is an effective tool for the optimization and data assimilation of multi-parameter crystal growth simulations.