Computer-aided design of polymeric materials: Computational study for characterization of databases for prediction of mechanical properties under polydispersity

Abstract

In Polymer Informatics, quantitative structure-property relationship (QSPR) modeling is an emerging approach for predicting relevant properties of polymers in the context of computer-aided design of industrial materials. Nevertheless, most QSPR models available in the literature use simplistic computational representations of polymers based on their structural repetitive unit. The aim of this work is to evaluate the effect of this simplification and to analyze new strategies to achieve alternative characterizations that capture the phenomenon of polydispersity. In particular, the experiments reported in this work are focused on three mechanical properties derived from the tensile test. The reported results revealed the disadvantages of using these simplified representations. Besides, we contributed with alternative representations for the databases of polymer molecular descriptors that achieved more realistic and accurate QSPR models.

Expert-in-the-loop AI for Polymer Discovery

“…PolyInfo 3 is a database containing information for basic polymer science which have been collected for the last fifteen years, and has recently been made available in RDF [14]. Another dataset of 1,073 polymers and related materials 4 has been released by [13]. This dataset has been created by using first-principles calculations with structures obtained either from other sources or by using structure search methods.…”

“…While this work is in spirit very similar to our proposed approach, there is no indication of how the selection was made and if the SME did not have to manually examine every proposed candidate. Cravero et al [4] propose a method for predicting relevant properties of polymers in the context of computer-aided design of industrial materials, by using an alternative representation for polymer molecular descriptors. Mannodi et al [17] also propose the use of a simplified representation for polymers, which are "fingerprinted" as simple, easily attainable numerical representations, which are mapped to the properties of interest using a machine learning algorithm.…”

“…The goal of their method is to learn the design of polymers with given target properties. Inspired by [4,17], we also propose the usage of a simplified representation for polymers, BRICS [6], which, to the best of our knowledge, has never been used as representation format in the context of automatically learning new polymers.…”

PolyMaS: A new software to generate high molecular weight polymer macromolecules from repeating structural units

“…In this sense, the motivation to develop in silico techniques is related to the significant saving of resources that must be used, considering the economic point of view as well as the time for research and development [9]. One way of carrying out these virtual tests is to use quantitative structure-property relationship (QSPR) models [10]. Traditional synthetic representations, such as monomers, dimers, and trimers, have been used to generate these models.…”

“…However, to obtain models which are closer to actual polymeric molecules, computational representations of very long polymer chains are necessary, and these representations can be obtained by the generation of macromolecules from the structural repetitive unit (SRU). Moreover, in the case of polymeric material databases, in which polydispersity is present, a more realistic representation should include several chain lengths [10].…”