Ethylene Polymerization Activity vs Grafting Affinity Trade-off Revealed by Importance Learning Analysis of In Silico Phillips Catalyst

Many aspects of the single-atom Cr/SiO₂ catalyst for ethylene polymerization, i.e., the Phillips catalyst, remain enigmatic due to the amorphous structure of the silica support. Previous computational studies have examined various mechanisms on cluster models of a single site or a small collection of sites, but none have examined the site-averaged kinetics for the ensemble of sites on a slab model of amorphous silica. This work uses the importance learning algorithm to probe the structural heterogeneities of the Phillips catalyst. In this way, we obtain site-averaged kinetics in quantitative agreement with several experimental studies. Moreover, we show that the catalytic rates are anticorrelated with the grafting probabilities and examine the structural features which modulate this trade-off. Our results explain why only a fraction of the Cr atoms appear to be active in ethylene polymerization and why calcination at higher temperatures yields more active catalysts. Furthermore, we show that the Bell–Evans–Polanyi (BEP) relation does not explain the site-to-site activity variations in the Phillips catalyst due to different influences of strain on the intermediates and the transition states.