Revealing the Carbon Hierarchical Porosity and Biomass Conversion Relationship Using a Combined Experimental and Data‐Driven Approach

ABSTRACT

This study explores the importance of hierarchical porosity in controlling the catalytic conversion of biomass. Four carbon materials (powder (SA), granulated (WV), nanotube (COH), and nanofiber (CFC)) were used to hydrolyze cellulose into glucose under mild conditions. It focuses on linking key physicochemical properties, surface chemistry, and porosity with catalytic efficiency using a combined experimental and machine‐learning (ML) approach. Characterization and catalytic tests showed that SA and COH carbons achieved the highest glucose production. Selectivity is linked to a combination of high BET surface area, moderate weak acid density, and increased porosity. The ML model, trained on curated experimental data with the Random Forest algorithm, achieved R ² train = 0.98; R ² test = 0.93; R ² (cross‐validation) = 0.93; RMSE test = 6.33. This ML model independently identified BET surface area (24.3%) and total acid density (20.5%) as the most influential features affecting glucose yield. These trends were confirmed through experiments for SA, COH, and CFC. Moreover, integrating ML into this experimental approach underscores the importance of accessible surface area and hierarchical porosity in driving catalytic activity. This led to the development of more efficient ML‐based computational tools to aid in the design of tuned hierarchical carbon materials for biomass conversion.