---

Industrial catalysis still plays a major role in the world economy. It suffices to here mention the Haber-Bosh process for ammonia synthesis or the Fisher-Tropsh synthesis for hydrocarbon production. These processes are of great practical interest and of major environmental impact. However, since they take place under extreme conditions of temperature, their experimental study is challenging [1]. Luckly, new machine learning methodologies [2,3] are allowing these phenomena to be studied under realistic conditions and are shedding new light on industrial catalysis revealing an unexpected complexity.  These simulations are showing that industrial catalysis is not simply related to the presence of catalytic sites but results from complex, global dynamical interactions between reagent and catalyst [4-7].  These interactions change significantly the properties of the materials on which the reaction takes place. The changes are not confined to the surface but at times they involve the whole material, sometimes inducing real phase transitions.  We illustrate this behavior in several examples and derive reassurance from favorable comparison with available experiments.

 

[1] Schlögl R., Angew. Chem., Int. Ed., 54 (2015) 3465.
[2] Perego S., Bonati L., npj Comput. Mater., 10 (2024) 291.
[3] Novelli P., Meanti G., Buigues P.J., Rosasco L., Parrinello M., Pontil M., Bonati L., npj Comput. Mater., 11 (2025) 293.
[4] Yang M., Raucci U., Parrinello M., Nat. Catal., 6 (2023) 829.
[5] Tosello Gardini A., Raucci U., Parrinello M., Nat. Commun., 16 (2025) 2475.
[6] Bonati L., Polino D., Pizzolitto C., Biasi P., Eckert R., Reitmeier S., Schlögl R., Parrinello M., Proc. Natl. Acad. Sci., 50 (2023) e2313023120.
[7] Tripathi S., Bonati L., Perego S., Parrinello M., ACS Catal., 14 (2024) 4944.

---