Machine Learning-Assisted Identification of Superconducting Bilayer Nickelate Thin Films via X-Ray Diffraction Analysis

The recent success in stabilizing ambient-pressure superconductivity in bilayer nickelate thin films grown on compressively strained substrates marks a milestone in nickelate superconductivity research [1]. However, the narrow growth window and the necessity for post-growth oxidation present considerable challenges, making the identification of superconducting samples both time-consuming and complex. It would be helpful to identify, a-priori, which of these as-grown films would become superconducting. Here we present a simple machine learning model that analyzes X-ray diffraction (XRD) data of bilayer nickelate thin films to predict the presence of superconductivity. Our model consistently outperforms experienced researchers in identifying samples exhibiting superconducting signatures in transport measurements, especially in films where the distinguishing features in XRD are subtle. We also discuss the broader applicability of our approach for classifying thin films of novel or metastable materials requiring complex post-growth processing