Data-Driven and Molecular Simulation Insights into the Synergistic Mechanisms of Pore Structure and Surface Chemistry in CO2 Capture by Biochar.

The CO2 adsorption performance of biochars is governed by its pore structure, surface chemistry, and adsorption conditions, yet the relative contributions of these factors across different pressure ranges remain unclear. Here, we developed an integrated framework combining machine learning, molecular simulation, and Experimental observations consistent with the proposed mechanism to systematically investigate these contributions. Using 4774 CO2 adsorption data points for biochars, predictive models were constructed, revealing that XGBoost achieved high accuracy and effectively captured the nonlinear interactions among pore structure, N/O-containing functional groups, adsorption conditions, and CO2 uptake. SHAP and PDP analyses demonstrated that at low pressure, adsorption is more strongly influenced by N/O-containing functional groups, whereas the contribution of ultramicropore volume gradually increases with pressure and becomes dominant over the representative pressure range above approximately 0.15 bar owing to the pore-filling effect. GCMC and DFT calculations further showed that ultramicropores of 0.35-0.5 nm strongly enhance confined CO2 adsorption, whereas N/O-containing surface features are associated with enhanced low-pressure adsorption, while DFT calculations further illustrate the distinct adsorption behaviors of representative functional groups. Experimental measurements of lignin-derived biochars were consistent with these findings. This study reveals a pressure-dependent mechanism of CO2 adsorption by biochars, highlighting surface-chemistry control at low pressure and ultramicropore-filling control at higher pressures, providing a theoretical basis for the rational design of high-performance carbon adsorbents.