Covalent Organic Frameworks (COFs) are an emerging class of crystalline porous materials with tunable structures and promising applications in gas storage, catalysis, and energy. However, their discovery and synthesis remain challenging due to the vast chemical design space and fragmented data landscape. This project aims to leverage machine learning to accelerate COF design by systematically reviewing and curating existing experimental and computational datasets related to COF synthesis, structure, and performance. The student will identify and organize available databases, assess their quality and coverage, and develop predictive models for both synthesis feasibility and material properties (e.g., stability, porosity, adsorption performance). The project will combine data-driven insights with domain knowledge to uncover structure–property–synthesis relationships, ultimately contributing toward more efficient and guided discovery of COFs.

- A structured review of the literature on COF datasets, including experimental and computational sources, with a clear comparison of their scope, quality, and limitations - A curated and well-documented dataset (or collection of datasets) of COF structures, synthesis conditions, and properties, including preprocessing and standardization steps - Development and evaluation of machine learning models for (i) predicting synthesis outcomes/feasibility and (ii) predicting key COF properties (e.g., stability, porosity, adsorption performance) - A final scientific report summarizing the dataset curation process, modeling approaches, results, and key insights into structure–property–synthesis relationships in COFs