Chemometric Organization and Structure–Property Relationships in an Industrial Polypropylene Product Portfolio

Industrial polypropylene portfolios comprise multiple commercial grades differentiated by molecular architecture, phase morphology, processability, and performance. In this study, 81 industrial polypropylene grades, including 38 homopolymers, 21 random copolymers, and 22 impact copolymers, were analyzed to evaluate the chemometric organization and structure–property relationships of a complete commercial portfolio. The dataset integrated melt flow index, xylene-soluble fraction, total ethylene content, ethylene content of the rubber phase, rubber-phase fraction, and mechanical, thermal, and optical performance variables obtained from routine industrial quality-control and product-certification activities. Principal component analysis, partial least squares discriminant analysis, and variable importance in projection analysis were used to examine portfolio organization, evaluate consistency with the predefined polypropylene families, and identify the descriptors contributing most strongly to family-level discrimination. The first two principal components explained 83.8% of the total variance. They revealed a low-dimensional organization consistent with the molecular and morphological differences among homopolymer, random copolymer, and impact copolymer grades. The full-descriptor PLS-DA model achieved 98.8% cross-validated accuracy and correctly classified 80 of the 81 grades using two latent variables. This performance reflects the internal consistency between the descriptor matrix and the existing industrial family classification rather than independently validated predictive capability for unknown grades. Homopolymer differentiation was mainly associated with molecular-weight-related flow behavior, random copolymer organization with ethylene-induced modification of crystallinity, and impact copolymer differentiation with heterophasic rubber-phase characteristics. The results provide a portfolio-specific chemometric workflow for grade organization and structure–property interpretation. However, the numerical domain boundaries and their transferability require validation using independent polypropylene portfolios from other producers.