Physical properties of materials, such as melting point, glass transition temperature, and solubility parameters, are ubiquitous in many different materials applications. Developing fast and reliable computational approaches for these properties has been a challenging avenue for computational researchers for a long time. Molecular dynamics (MD) has shown potential for most of such properties, however, they are computationally intensive, and performing simulations for a large library of candidates is not a viable option. The recent development in advanced machine learning techniques, especially graph convolutional neural networks, has shown tremendous promise in the accurate predictions of material properties. However, most of the work in the past has focused on electronic properties. In this work, we apply graph convolutional neural networks for predicting the physical properties of organic molecules, with a focus on melting point and solubility parameters. We demonstrate that graph-based models are highly efficient, and in some cases better than physics-based models, in predicting the physical properties of materials. We also present the learning curve analysis of the models and evaluate their performance with the data size. Additionally, we use the best models from this work to predict the properties of materials from existing databases whose properties are unknown.

Learning the physical properties of organic molecules using graph-based models

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

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