In order to optimize the performance of dense polymer networks, which are widely utilized for high-performance applications, quantitative structure-activity relationships (QSARs) have proven to be an invaluable tool. The reliability of QSARs, and the chemoinformatics approaches that are used to develop them, however, can be compromised by issues such as structural bias, which is the unequal prevalence of molecule having certain structural features in the datasets utilized to generate QSARs. For instance, many of the monomers utilized to form dense polymer networks are derived from petroleum-based sources, which often involve the coupling of phenol during synthesis. As a result, certain phenol-based motifs are over-represented compared to other structural possibilities in many of the datasets on which QSARs are built, decreasing the robustness of the predictions. Interestingly, the recent availability of new monomers from bio-based sources has introduced a much greater variety of structural motifs into the available database of monomer properties, helping to mitigate the issue. This presentation will utilize multiple examples to illustrate how the availability of bio-based structures has led to an overall improvement in the predictive power of QSARs for cyanate esters, a class of dense polymer networks.

Predictive methods for dense polymer networks: Combating bias with bio-based structures

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.