Ziegler–Natta catalysts, discovered in the 1950s, are still central to the industrial production of polyethylene and polypropylene. Despite being the workhorse of the polymer industry and years of extensive research, the improvements in Ziegler–Natta catalysts have been mostly empirical. In particular, the coordination surroundings of the Ti sites in precatalysts, key for the formation of active sites upon AlEt3 activation, have been highly debated. Notably, quantification of different Ti sites on the MgCl2 surface has not been possible thus far, hindering the development of quantitative structure–activity relationships. In this work, we prepared a series of precatalysts with increasing concentrations of the BCl3 modifier during the catalyst synthesis, known to affect the number of Cl and alkoxo ligands in the Ti local surroundings, as well as the activity in ethylene polymerization. We developed a methodology, relying on a library of theoretical X-ray absorption spectroscopy (XAS) lines computed from the density functional theory (DFT)-optimized structural models, benchmarked on a series of well-defined molecular compounds. A quantitative analysis of the XAS data allows us to evaluate the relative fractions of Ti sites in the Cl surrounding, as well as containing alkoxide and O-donor ligands. The analysis shows that the number of Ti–O bonds (initially mostly present in their alkoxo form) decreases upon treatment with BCl3 up to the B/Ti ratio of 2, in agreement with earlier proposals, and increases afterward, due to O-donor ligands, likely related to B-alkoxo species coordinated to Ti. The catalytic activity of these precatalysts after activation with AlEt3 in ethylene polymerization passes through a maximum at a B/Ti ratio of ∼2, pointing to its relation to the detrimental effect of O-based ligands on the activity. This approach allows one to address and quantify the metal speciation in the complex Ziegler–Natta precatalysts and relate this to their catalytic activity, a first step toward establishing quantitative structure–activity relationships.
Yakimov et al. (Tue,) studied this question.