Abstract:A machine learning strategy is presented for the rapid discovery of new polymeric materials satisfying multiple desirable properties. Of particular interest is the design of high refractive index polymers. Our in silico approach employs a series of quantitative structure–property relationship models that facilitate rapid virtual screening of polymers based on relevant properties such as the refractive index, glass transition and thermal decomposition temperatures, and solubility in standard solvents. Explorati… Show more
“…Their origin dates back to the pioneering work by Venkatasubramanian et al 19 Most such studies have focused on the use of a limited number of chemical fragments and their stochastic recombination to sequentially transform starting compounds into desired targets. 20,21 However, this approach significantly narrows the design space. To broaden the search space, more advanced ML techniques using probabilistic language models have appeared in recent years.…”
The use of machine learning in computational molecular design has great potential to accelerate the discovery of innovative materials. However, its practical benefits still remain unproven in real-world applications, particularly in polymer science. We demonstrate the successful discovery of new polymers with high thermal conductivity, inspired by machine-learning-assisted polymer chemistry. This discovery was made by the interplay between machine intelligence trained on a substantially limited amount of polymeric properties data, expertise from laboratory synthesis and advanced technologies for thermophysical property measurements. Using a molecular design algorithm trained to recognize quantitative structure-property relationships with respect to thermal conductivity and other targeted polymeric properties, we identified thousands of promising hypothetical polymers. From these candidates, three were selected for monomer synthesis and polymerization because of their synthetic accessibility and their potential for ease of processing in further applications. The synthesized polymers reached thermal conductivities of 0.18-0.41 W/mK, which are comparable to those of state-of-the-art polymers in non-composite thermoplastics .
“…Their origin dates back to the pioneering work by Venkatasubramanian et al 19 Most such studies have focused on the use of a limited number of chemical fragments and their stochastic recombination to sequentially transform starting compounds into desired targets. 20,21 However, this approach significantly narrows the design space. To broaden the search space, more advanced ML techniques using probabilistic language models have appeared in recent years.…”
The use of machine learning in computational molecular design has great potential to accelerate the discovery of innovative materials. However, its practical benefits still remain unproven in real-world applications, particularly in polymer science. We demonstrate the successful discovery of new polymers with high thermal conductivity, inspired by machine-learning-assisted polymer chemistry. This discovery was made by the interplay between machine intelligence trained on a substantially limited amount of polymeric properties data, expertise from laboratory synthesis and advanced technologies for thermophysical property measurements. Using a molecular design algorithm trained to recognize quantitative structure-property relationships with respect to thermal conductivity and other targeted polymeric properties, we identified thousands of promising hypothetical polymers. From these candidates, three were selected for monomer synthesis and polymerization because of their synthetic accessibility and their potential for ease of processing in further applications. The synthesized polymers reached thermal conductivities of 0.18-0.41 W/mK, which are comparable to those of state-of-the-art polymers in non-composite thermoplastics .
“…Venkatraman et al. also used a QSPR model built on linear partial least squares regression and ensemble tree‐based random forests to rapidly test polymers for high refractive indices . The researchers used geometrical quantum chemistry‐based descriptors consisting of the highest occupied and lowest unoccupied molecular orbital energies, charges, polarizabilities, superdelocalizabilities, and radial distribution function indices.…”
Section: Machine Learning For Polymer Systemsmentioning
confidence: 99%
“…The researchers used geometrical quantum chemistry‐based descriptors consisting of the highest occupied and lowest unoccupied molecular orbital energies, charges, polarizabilities, superdelocalizabilities, and radial distribution function indices. The authors mention that while their model is sufficiently predictive, there are limitations on the extrapolation that can be performed resulting in high variance between the predictive model and experimental observations . This highlights the importance and difficulty associated with choosing the appropriate descriptors for data‐driven techniques.…”
Section: Machine Learning For Polymer Systemsmentioning
confidence: 99%
“…In addition to glass transition temperatures, researchers have focused on predicting properties such as refractive indices and dielectric constants . Jabeen et al.…”
Section: Machine Learning For Polymer Systemsmentioning
The number of applications of informatics or data-driven discovery is growing in many fields, including materials science. The large amount of data that is readily available, combined with high-level statistical algorithms, is proving to be extremely useful in developing complex predictive models with little to no human supervision or bias. However, in the field of soft matter, which includes complex materials such as polymers, liquids, emulsions, colloids, and gels, there is a slower adoption of informatics strategies than in adjacent fields. Here, the current state of soft matter informatics is discussed. Challenges specific to soft materials, including data classification, various degrees of organization at multiple length scales, and process-dependent properties require unique approaches by researchers in order to develop robust informatics approaches in soft matter. The current ability to extract and analyze the information from the PoLyInfo database is demonstrated by the fitting of the Flory-Fox equation for glass transition temperature for several polymers. This Progress Report serves to introduce and excite the scientific community about the remarkable potential of informatics for exploring the properties of soft materials.
“…The computer-aided design and virtual selection of new materials can take advantage of these computational models [10][11][12][13]. However, the QSPR modeling in Polymer Informatics is particularly complex [14][15][16][17]. In this field, a careful computational modeling of polymeric materials is required.…”
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.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.