(584z) Seeking Optimal Absorbent for in-Situ Conversion of CO2 to Methanol: A Holistic Approach Combining Experiments, DFT Calculations, and Machine Learning.

In-situ conversion of CO₂ to methanol is an integrated approach combining CO₂ capture and catalytic conversion in a single step, thereby avoiding the energy-intensive regeneration of absorbents. This integration offers significant economic and environmental advantages, yet current in-situ conversion systems suffer from limited reaction performance compared to hydrogenation of pure CO₂ feedstocks. The key challenge is to identify an optimal CO₂ absorbent that performs effectively under in-situ conversion conditions to bridge this efficiency gap.

In this study, we present a holistic approach combining machine learning (ML), density functional theory (DFT) calculations, and experimental evaluation. First, an ML model was developed to explore the space of potential CO₂ conversion pathways under in-situ conditions, providing mechanistic insights into intermediate formation and consumption. Next, DFT calculations were performed to determine the rate-determining step and to identify quantifiable molecular features of amine-based absorbents that correlate with high methanol yield. These features informed the construction of the second ML model for screening a broad library of candidate amines and predicting their performance. Top-ranked amines from this screening were subsequently validated in experiments, confirming improved CO₂-to-methanol conversion performance in line with the model predictions.

This integrated study revealed a notable mechanistic insight: structurally hindered amines (with bulky substituents) do not readily form coordination complexes with the homogeneous catalyst, instead react with CO₂-derived formic acid to efficiently form amides (formamide intermediates). Even more surprisingly, we found that the presence of CO₂ can inhibit formation of amine-cat complexes, thereby enhancing methanol selectivity. This suggests that under in-situ conditions, CO₂ may act not merely as a reactant but as a reagent that governs the reaction selectivity. Our findings shed light on the selectivity mechanisms governing the formic acid–amide–methanol pathway in in-situ thermochemical CO₂-to-methanol conversion, offering a deeper understanding of the process that has not been elucidated in previous studies.