Single-site catalysts are revolutionizing polyolefin production, the largest by volume polymer produced in the chemical industry. Their impact could be enhanced ever further, however, by a rational method for design of new catalysts to achieve a specific molecular architecture of the polymer. In our view, such a rational design procedure must begin with fundamental knowledge of the micro kinetic mechanism for a given catalyst system and the associated rate constants; information that is difficult to obtain because the number of different mechanistic possibilities for individual steps in the polymerization (i.e. activation, initiation, propagation, chain transfer/termination) can result in potentially thousands of different polymerization models. Our strategy for solving this problem is to take full advantage of the number of experimental techniques (e.g. ¹H-NMR, ¹³C-NMR, UV-VIS, and GPC) used in analyzing both the polymerization and the resulting polymer, to rapidly and robustly use that information in determining which candidate polymerization models can describe the experimental data, and to then design new experiments to discriminate between candidate models.

We will show that batch polymerization data is ideal for modeling single-site polymerization. Since a single experiment samples the rate of polymerization over a range of monomer concentrations, fewer experiments are required for model building and discrimination, although model building is more complex. Kinetic rate constants are primarily determined by fitting rich multi-response data sets consisting of (i) monomer concentration versus time profiles and (ii) the time evolution of the molecular weight distribution for batch polymerizations with (iii) constraints imposed by other measurements like end group analysis by NMR. This integrated analysis approach is unique to our group. Special computational tools have been developed for the population balance models of the polymerization that can include up to 100,000 ODEs, where a new kinetic model can be formulated, computer code automatically generated and the model parameters optimized with a set of experimental data – all within a few hours. An example of this method of kinetic analysis for the polymerization of 1-hexene with [rac-(C₂H₄(1-Ind)₂)ZrMe][MeB(C₆F₅)₃] will be shown. During the course of this analysis it was discovered that new mechanisms beyond those considered in the literature for this catalyst are required to fit the molecular weight distribution. Study of this catalyst system exemplifies the power of this quantitative modeling strategy in conjunction with batch polymerization data by significantly aiding the discovery of new mechanisms that would otherwise not have been considered. Finally, I will discuss how this approach can be extended to develop quantitative descriptions for other polymerization processes for producing novel and technologically interesting polymers.