Enhancing the Quality of MOF Thin Films for Device Integration Through Machine Learning: A Case Study on HKUST-1 SURMOF Optimization

Abstract

Metal–organic Frameworks (MOFs), especially as thin films, are increasingly recognized for their potential in device integration, notably in sensors and photo detectors. A critical factor in the performance of many MOF-based devices is the quality of the MOF interfaces. Achieving MOF thin films with smooth surfaces and low defect densities is essential. Given the extensive parameter space governing MOF thin film deposition, the use of machine learning (ML) methods to optimize deposition conditions is highly beneficial. Combined with robotic fabrication, ML can more effectively explore this space than traditional methods, simultaneously varying multiple parameters to improve optimization efficiency. Importantly, ML can provide deeper insights into the synthesis of MOF thin films, an essential area of research. This study focuses on refining an HKUST-1 SURMOF (surface-mounted MOF) to achieve minimal surface roughness and high crystallinity, including a quantitative analysis of the importance of the various synthesis parameters. Using the SyCoFinder ML technique, thin film surface quality is markedly enhanced in just three generations created by a genetic algorithm, covering 30 distinct parameter sets. This method greatly reduces the need for extensive experimentation. Moreover, the results enhance the understanding of the vast synthesis parameter space in HKUST-1 SURMOF growth and broaden the applications of MOF thin films in electronic and optoelectronic devices.