This is the authors' abstract. We don't add key points for this paper.
Polyolefin elastomers (POE) are widely used in strategic fields such as automotive lightweighting and medical devices because they combine excellent rubber-like elasticity with thermoplastic processability. As a key technological enabler for producing high-performance POE, the catalytic activity of constrained geometry catalysts (CGCs) is governed by coupled effects of ligand steric hindrance, electronic structure, and reaction conditions. Conventional trial-and-error strategies and purely linear models struggle to resolve these nonlinear structure-condition-performance relationships, leading to long development cycles for high-performance catalysts. In this study, we integrated experimental data with DFT calculations to build a database covering the electronic structure, geometric structure, and process parameters of CGC catalysts, and developed a machine-learning framework for catalytic activity prediction based on multifeature synergistic optimization. The gradient boosting regression (GBR) model showed the best predictive performance, with R 2 = 0.9403 and RMSE = 0.98 kgPOE/(mmol cat·h) on the test set; 10-fold cross-validation and external validation on three independent CGCs further supported its robustness within the chemical domain represented by the data set. SHAP analysis, together with DFT-derived descriptors, identified the buried volume of the cyclopentadienyl unit (%V_Bur(Cp)) and the NBO charge of the donor atom (J-nbo) as the two dominant factors controlling activity. We further propose a nonlinear steric-electronic matching mechanism: sufficient Cp steric shielding suppresses chain-transfer pathways, whereas a moderate donor-electronic window promotes olefin activation without excessive coordination stabilization. Guided by this mechanism, two new CGC candidates were computationally designed and predicted to exhibit enhanced activity relative to representative literature CGCs. This work provides a data-driven strategy for structure-level directional regulation of CGC activity, while also clarifying current limitations related to data set size, reaction-condition coverage, and the need for future experimental synthesis and kinetic validation.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Xu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: