Machine learning simulations reveal how plastic transforms into hard carbon

Researchers from the University of Oxford have utilized machine learning simulations to understand how a common polymer, polyvinylidene chloride, transforms into hard carbon. This hard carbon, found in sodium-ion batteries, possesses an intricate, disorderly structure that resists conversion to ordered graphite, unlike traditional graphite.

Prior to this study, the atomic-scale mechanisms behind hard carbon formation remained elusive. The Oxford team employed machine learning-driven atomistic simulations to study the thermal decomposition of PVDC, a polymer that serves as a precursor to non-graphitizing carbon. By tracking individual atoms and observing the formation and breaking of chemical bonds, the simulations revealed how hydrogen and chlorine atoms from PVDC were removed, generating reactive carbon sites.

These sites subsequently formed new carbon-carbon bonds, progressively connecting polymer chains into an extended network. Unlike graphite, hard carbon includes not only six-membered carbon rings but also five- and seven-membered rings, introducing curvature. This curvature, resembling the pentagonal and hexagonal structure of a soccer ball, prevents the carbon network from forming flat, ordered layers.

The study's lead author, Litong Wu, expressed excitement in uncovering hidden patterns amidst the chaos of high-temperature atomic movements. Professor Volker Deringer emphasized the significance of the findings in advancing the understanding of hard-carbon formation and its potential applications in battery anodes. The research data is publicly available for further exploration, reflecting the team's commitment to open science and collaboration.