Large-scale expansion of the Cu catalyst CO2 hydrogenation pathway through ML reaction-network exploration

AI editorPublished 2026.09.17Posted 2026.09.18 00:01 KST

A reaction-network for CO2 hydrogenation on Cu catalysts was expanded using ML activation-energy prediction based on reaction/reagent fingerprints and an exploration algorithm, constructing a microkinetic model containing 9,389 elementary reaction. This approach predicted CO2 conversion up to 40 times higher than a manually screened 152 reaction model, and highlighted the importance of methanol and CO formation pathways and intermolecular hydrogen-hopping. [1][2]

1/ AI utilization method: machine-learning activation-energy prediction based on reaction/reagent fingerprint
2/ Applied task: prediction of activation barriers and reaction-network exploration for Cu catalyst CO2 hydrogenation reactions
3/ Reported result: constructed a microkinetic model at the scale of 9,389 elementary reaction and predicted CO2 conversion up to 40 times higher than a manually selected 152 reaction model
4/ This study targeted the CO2 hydrogenation reaction on Cu catalysts, predicted activation energies with an ML model using fingerprints of reactions and reagents, and combined reaction-network exploration with automated microkinetics (MKM) to expand the reaction network on a large scale.
5/ As a result, a network containing 9,389 elementary reaction was constructed, and this model predicted CO2 conversion up to 40 times higher than a model based on 152 reactions selected manually.
6/ It also reported that the pathways for methanol and CO formation, as well as intermolecular hydrogen-hopping, are important for understanding the reaction route.
7/ From a materials perspective, it shows that data-driven reaction-network expansion in a complex catalytic network can change macroscopic prediction and mechanism interpretation relative to a limited reaction set.
8/ However, these results are based on a specific Cu-CO2 hydrogenation system and the reported network construction, and cannot be taken as a direct generalization to other catalysts or conditions. [1][2]

How to apply this

Use reaction/reagent fingerprints to predict activation barriers and explore catalytic reaction networks more broadly.

Collect elementary reaction candidates and fingerprints for each reaction from existing literature and computational results.
Train an ML regression model on a subset of known barrier values and check prediction error with cross-validation.
Use the predicted barriers to expand reaction-network exploration and compare product distributions with automated microkinetic calculations.
Revalidate promising pathways through independent calculations or experiments to strengthen mechanism interpretation.

Data needed: Need elementary reaction candidates for the target catalyst and reactant combination, structure- or composition-based fingerprints for each reaction, and, if possible, some computed or experimental rate constants or barrier values.
Checking the result: Compare newly found pathways with independent DFT calculations or experimental kinetics to verify whether the barriers and product distributions are consistent.
What can go wrong: If the training data are too sparse or the reaction space is biased, accumulated errors in barrier prediction can lead to selection of an incorrect dominant pathway.

이 가이드는 원문의 방법을 바탕으로 AI가 제안한 적용 방향입니다. 원문에 적힌 내용이 아니며 같은 결과가 재현된다는 보장도 아닙니다.

웹 자료를 바탕으로 AI가 자동 요약한 글입니다. 연구의 세부 조건과 결과는 연결된 원문을 확인해주세요.

Large-scale expansion of the Cu catalyst CO2 hydrogenation pathway through ML reaction-network exploration

How to apply this

Related concepts and practice

Sources