Modélisation de la synergie entre nanoparticules et promoteurs lors de processus catalytiques : cas de la synthèse de l'ammoniac

Ammonia is a promising vector for green hydrogen storage, yet its current synthesis via the Haber-Bosch process remains unsuitable for sustainable applications. An alternative catalytic route involving cobalt nanoparticles encapsulated in barium oxide has attracted growing interest. This thesis investigates, through atomistic modeling, the interactions between cobalt nanoparticles and potential promoters (Ba, Ca), by developing machine learning potentials (MLPs) and robust analytical methodologies. A high-performance Spectral Neighbor Analysis Potential (SNAP) was developed for pure cobalt, accurately capturing its static and dynamic properties from the bulk phase to the nanoscale. After numerous unsuccessful attempts, a stable Co-Ba potential was finally obtained at the end of the thesis. In parallel, an advanced method was designed to characterize atomic environments in complex systems, leading to quantitative insights into surface activation and melting phenomena in pure cobalt nanoparticles. These findings, along with the developed methodology, open new avenues for the dynamic analysis of nanoparticles under catalytically relevant temperatures.