Copolymer Informatics with Multitask Deep Neural Networks

Macromolecules, Journal Year: 2021, Volume and Issue: 54(13), P. 5957 - 5961

Published: June 29, 2021

Polymer informatics tools have been recently gaining ground to efficiently and effectively develop, design, discover new polymers that meet specific application needs. So far, however, these data-driven efforts largely focused on homopolymers. Here, we address the property prediction challenge for copolymers, extending polymer framework beyond Advanced fingerprinting deep-learning schemes incorporate multitask learning meta are proposed. A large data set containing over 18 000 points of glass transition, melting, degradation temperature homopolymers copolymers up two monomers is used demonstrate copolymer efficacy. The developed models accurate, fast, flexible, scalable more properties when suitable become available.

Language: Английский

---

RadonPy: automated physical property calculation using all-atom classical molecular dynamics simulations for polymer informatics

npj Computational Materials, Journal Year: 2022, Volume and Issue: 8(1)

Published: Nov. 8, 2022

The rapid growth of data-driven materials research has made it necessary to develop systematically designed, open databases material properties. However, there are few for polymeric compared other systems such as inorganic crystals. To this end, we developed RadonPy, the world-first open-source Python library fully automated all-atom classical molecular dynamics (MD) simulations. For a given polymer repeating unit, entire process modeling, equilibrium and nonequilibrium MD calculations, property calculations can be conducted automatically. In study, 15 different properties, including thermal conductivity, density, specific heat capacity, expansion coefficients, refractive index, were calculated more than 1,000 unique amorphous polymers. properties validated with experimental values from PoLyInfo. During high-throughput data production, eight polymers extremely high conductivities, exceeding 0.4 W/mK, identified, six unreported conductivities. These found have density hydrogen bonding units or rigid backbones. A decomposition analysis conduction, which is implemented in revealed underlying mechanisms that yield conductivity polymers: transfer via bonds dipole-dipole interactions between chains their covalent backbone rigidity. creation massive amounts computational using RadonPy will facilitate development informatics, similar how emergence first-principles database crystals had significantly advanced informatics.

Language: Английский

---

Machine learning for polymeric materials: an introduction

Polymer International, Journal Year: 2021, Volume and Issue: 71(5), P. 537 - 542

Published: Dec. 4, 2021

Abstract Polymers are incredibly versatile materials and have become ubiquitous. Increasingly, researchers using data science polymer informatics to design new understand their structure–property relationships. Polymer is an emerging field. While there many useful tools databases available, not widely utilized. Herein, we introduce the field of discuss some available tools. We cover how share data, approaches for preparing a dataset machine learning recent applications property prediction synthesis. © 2021 Society Industrial Chemistry.

Language: Английский

---

Data-driven algorithms for inverse design of polymers

Soft Matter, Journal Year: 2021, Volume and Issue: 17(33), P. 7607 - 7622

Published: Jan. 1, 2021

The ever-increasing demand for novel polymers with superior properties requires a deeper understanding and exploration of the chemical space.

Language: Английский

---

Featurization strategies for polymer sequence or composition design by machine learning

Molecular Systems Design & Engineering, Journal Year: 2022, Volume and Issue: 7(6), P. 661 - 676

Published: Jan. 1, 2022

In this work, we present, evaluate, and analyze strategies for representing polymer chemistry to machine learning models the advancement of data-driven sequence or composition design macromolecules.

Language: Английский

---