Discerning the underlying trends in electrocatalytic hydrogen evolution activity of MOF/MOF-derived materials using data-driven approach

• A manually curated metal organic framework-database (MOF-DB) pertaining to electrocatalytic hydrogen evolution reaction (HER) activity of 566 HER catalyst entries was created. • Electrocalytic features, such as overpotential, Tafel slope, exchange current density, and R ct – alongside structural descriptors such as metal centres, linkers, and synthesis methods, were extracted from the SCOPUS database. • Inferential statistical methods, supervised and unsupervised machine learning (ML) models, including Decision Tree (DT), Support Vector Machine (SVM), Naive Bayes (NB), Principal Component Analysis (PCA), and K-Means clustering, were used to reveal underlying mechanistic patterns and identify high-performance catalysts. • From the dataset, 62 promising catalyst entries were shortlisted based on insights from the modelling approaches, among which 10 demonstrated superior HER performance competing with Pt/C reference catalyst. Developing alternative green energy sources is vital for addressing climate change driven by deforestation and excessive resource exploitation. Among the emerging clean energy strategies, the use of metal–organic framework (MOF) or MOF-derived materials as catalysts for electrocatalytic hydrogen evolution reaction (HER) presents a promising route for sustainable green hydrogen production. These materials have attracted considerable attention as HER electrocatalysts due to their structural versatility and tunable electronic properties. However, identifying top-performing candidates remains challenging due to the complexity and redundancy of existing experimental data. In this study, we present a manually curated database of 566 HER catalyst entries, each characterized by 26 features extracted from SCOPUS-indexed publications. The database includes key electrochemical parameters–overpotential, Tafel slope, exchange current density, and R ct – alongside structural descriptors such as metal centres, linkers, and synthesis methods. Inferential statistics, supervised and unsupervised machine learning (ML) models, including Decision Tree (DT), Support Vector Machine (SVM), Naive Bayes (NB), Principal Component Analysis (PCA), and K-Means clustering, were used to reveal underlying mechanistic patterns and identify high-performance catalysts. It enabled the classification of the catalysts based on dominant HER mechanisms (Mechanistic Limiting Cases, Volmer–Heyrovsky, Volmer–Tafel) and benchmarking of catalysts against a commercial Pt/C reference. From the dataset, 62 promising catalyst entries were shortlisted based on the insights derived from the modelling approaches. Among these, 10 catalyst entries have demonstrated HER performance that competes with Pt/C in terms of energy demand (low overpotential) and rapid kinetics (low Tafel slope). Four catalyst entries in this group are Ni or Co-containing nonprecious metal catalysts. The majority of the catalyst entries in the group reveal Pt-incorporation, for instance, incorporation of Pt in nonprecious metal catalysts or in MoS 2. Discovery of high performing HER catalyst from MOF-DB using employing ML models