Data Driven Control of Defect Formation in Solution Deposited Sb ₂ Se ₃ Thin Films

Abstract Semiconductor thin films are foundational to a broad range of optoelectronic technologies. Solution deposition offers a low‐cost, energy‐efficient alternative to vapor‐based methods, but its practical scalability is hindered by poor reproducibility and high defect densities arising from complex interdependencies among processing variables. Statistical design of experiments (DoE) enables critical insight into the non‐intuitive interdependencies that are not accessible by conventional one‐variable‐at‐a‐time (OVAT) approaches. Using Sb ₂ Se ₃ as an example semiconductor, DoE is employed with a central composite design (CCD) to systematically vary six processing parameters encompassing ink formulation and deposition parameters. From only 77 experiments, predictive models are constructed for four macro‐ and microscopic defect types known to interrupt charge transport. All six variables are found to significantly and synergistically influence film quality and are optimized to minimize defect density. Importantly, the optimal film exhibited a threefold enhancement in photoresponse in single‐junction devices with no detectable change in composition, nanostructure, or film stability, implicating defect suppression as the critical driver of improved performance. This work highlights DoE as a powerful methodology for uncovering latent structure–processing–property relationships in thin films and provides a general framework for accelerating the development of optoelectronic‐grade materials via solution processing.