Machine learning molecular dynamics advances thermal modeling of graphene oxide

Graphene oxide is a material with remarkable potential in electronics and composites, and poses a major challenge for researchers seeking to understand how its complex chemical structure affects its ability to conduct heat. Bohan Zhang, Biyuan Liu, and Penghua Ying from their respective institutions are leading a team that uses advanced computational techniques to successfully link the reduction process of graphene oxide to its thermal properties. Researchers have developed new machine learning possibilities trained on precise quantum mechanical calculations. This makes it possible to simulate the behavior of graphene oxide on a scale previously unattainable. This study reveals that reduced graphene oxide exhibits significantly lower thermal conductivity than the pristine state and, importantly, demonstrates a predictable relationship between its chemical structure and heat transfer, providing a powerful new framework for designing carbon materials with tailored thermal properties.

Relationship between reduction chemistry and heat transport

Graphene oxide exhibits complex chemical properties that influence its structural, thermal, and mechanical behavior, but quantitatively linking reduction chemistry and heat transport has proven difficult. In this study, we investigate this relationship using a combination of controlled chemical reduction, Raman spectroscopy, and molecular dynamics simulations to characterize how the structure and thermal properties of graphene oxide evolve. This study demonstrates that systematic reduction of oxygen-containing functional groups predictably increases thermal conductivity and provides a quantitative understanding of how reduction chemistry affects heat transfer in these two-dimensional materials. The findings reveal that the repair of the sp2 carbon network is a key mechanism to enhance thermal conductivity during reduction and provide insights for tailoring the thermal properties of reduced graphene oxide for specific applications.

Neural evolutionary potential of molecular dynamics simulations

Researchers are leveraging machine learning interatomic potentials (MLIPs), particularly neuroevolutionary potentials (NEPs), to enable large-scale molecular dynamics (MD) simulations. These potentials are refined to increase their accuracy for first-principles calculations and applied to the study of a variety of materials, with an emphasis on carbon-based systems such as graphene and graphene oxide. Research includes the thermal conductivity of raw graphene, the effects of oxygen functionalization and reduction levels on graphene oxide, and heat transport in various carbon allotropes. The team also studied 2D materials beyond carbon, such as MoS2 and h-BN, and developed accurate models of the thermal and thermodynamic properties of water.

The main focus is on understanding the thermal conductivity of disordered materials such as amorphous silicon and graphene oxide, which poses challenges for traditional modeling techniques. Researchers decompose thermal conductivity into the contributions of various vibrational modes and analyze the phonon mean free path to reveal the mechanisms governing heat transfer. This research spans applications such as thermal management, flexible electronics, and energy storage, investigating anisotropic heat transport and temperature-dependent behavior.

Neuroevolution accurately models graphene oxide dynamics

Scientists have created a new computational technique, the neuroevolutionary potential (NEP), to accurately and efficiently model the thermal reduction of graphene oxide. This advance enables large-scale molecular dynamics simulations, overcoming the limitations of previous computational approaches and providing a means to understand heat transport within this complex material. NEP models trained on density functional theory data show strong agreement with high-fidelity calculations and achieve significant computational speed improvements compared to existing methods. Simulations revealed that the reduced graphene oxide exhibits suppressed thermal conductivity ranging from a few watts to tens of watts per meter Kelvin, which is significantly lower than the pristine graphene.

The researchers found that thermal conductivity increases modestly with increasing hydroxyl content, but reverses at the highest oxidation levels. Analysis of gaseous byproducts during thermal reduction confirmed the production of water, carbon dioxide, and carbon monoxide, providing insight into the structural evolution of the material. These results provide a computationally tractable framework for investigating the relationship between chemical structure and heat transport in heterogeneous carbon materials.

Reduction pathway and heat transfer of graphene oxide

This study establishes a computational framework for understanding how chemical changes affect heat transfer in graphene oxide, a material with potential for thermal management. Scientists have developed a highly efficient model trained using basic calculations that accurately simulates the thermal reduction of graphene oxide on scales previously unattainable. Simulations revealed that the thermal conductivity of reduced graphene oxide is strongly influenced not only by the overall oxidation level but also by the specific chemical composition of the starting material. The research team discovered two important pathways governing structural evolution during reduction. One is the enhanced lattice recovery and enhanced heat transfer with increasing hydroxyl content, and the other is the pathway leading to defect formation and suppressed conductivity with increasing oxygen content.

Quantum effects were found to reduce the thermal conductivity by suppressing high-frequency vibrations, with values ​​ranging from 1.28 to 13.71 watts/meter Kelvin. These values, although lower than pristine graphene, suggest that reduced graphene oxide is promising for thermoelectric applications where minimizing heat transfer is beneficial.