Graphene provides a foundation for exploring thermal transport in low-dimensional materials and engineering carbon-based thermal management materials. Its practical performance depends on phonon transport, structural order, and interfacial thermal resistance. Relating these mechanisms with material configuration and device architectures motivates the integration of thermal characterization, atomistic simulation, and artificial intelligence (AI).
Graphene- and graphite-based assemblies have emerged as versatile platforms for thermal management, enabling efficient heat spreading, thermal interface engineering, and thermal energy storage[1]. For instance, the thermal performance of graphene thermal-spreading films can be enhanced through precise control of wrinkles and interlayer stacking. Highly aligned graphene fibers and mechanically compliant thermal interface materials (TIMs) facilitate efficient axial and through-plane thermal conduction, respectively. These highlight the critical roles of structural alignment, network connectivity, and interfacial contact in governing thermal transport and ultimately determining device-level thermal management performance. For energy conversion and storage, a self-adaptive photothermal system combines a moss-like graphene/polydopamine/MXene surface with a phase-change storage base [2]. Coordinating photon absorption, thermal conduction, latent heat storage, and heat-loss suppression enables efficient solar-energy utilization. Thermally conductive phase-change composites extend this materials strategy to temperature regulation and thermal buffering.
AI for science research extends the analysis from material structures to atomic interactions. A series of neuroevolution potentials (NEPs) expands both chemical coverage and accessible simulation scales. NEP89 covers 89 elements across the periodic table, supporting inorganic and organic materials and adaptation through fine-tuning[3]. Independent validation of stability, structure, and thermal transport remains essential for reliable predictions. NEP-based atomistic simulations reach device-scale field-effect transistors (FETs) containing millions of atoms[4], enabling spatially resolved predictions of temperature distributions, local heat fluxes, and thermal stresses. Physics-constrained AI further enables inverse design of multifunctional temperature-control materials. PHICS combines multiscale physical relationships encoded in a directed acyclic graph (DAG) with multi-objective optimization[5]. In graphite-fiber/alkane composites, amorphous alumina interlayers promote thermal conduction and electrical insulation while retaining latent heat storage.
Future thermal materials should combine high conductivity and low contact resistance with mechanical compliance, scalable fabrication, and device reliability. In parallel, embedding physical knowledge into AI architectures could enable interpretable inverse design and materials world models linking processing, structure, interfaces, and properties [6].
Acknowledgements The work is in part supported by Research Grants Council of Hong Kong, particularly via Grant Nos. AoE/P-701/20, 14206721, 2151231, and 14203623.
Selected References:
[1] H. Zheng, P. He, S. Yang, G. Ding, Thermally conductive graphene-based films for high heat flux dissipation, Carbon, 233 (2025).
[2] H. Liang, D. Ding, H. Wang, T. Liang, C. Wong, P. Zhang, W. Liang, L. Ji, Q. Zhang, C. Song, T. Deng, R. Sun, J. Xu, Y. Chen, Self-Adaptive Solar-Thermal System With Moss-Like Surface for Efficient Energy Utilization, Advanced Materials, 38 (2026) e73376.
[3] T. Liang, K. Xu, E. Lindgren, Z. Chen, R. Zhao, J. Liu, E. Berger, B. Tang, B. Zhang, Y. Wang, K. Song, P. Ying, N. Xu, H. Dong, S. Chen, P. Erhart, Z. Fan, T. Ala-Nissila, J. Xu, NEP89: universal neuroevolution potential for inorganic and organic materials across 89 elements, Nature Computational Science, 6 (2026) 789-801.
[4] K. Xu, G. Wang, T. Liang, Y. Xiao, D. Ding, H. Guo, X. Gao, L. Tong, X. Wan, G. Zhang, J. Xu, Device-Scale Atomistic Simulations of Heat Transport in Advanced Field-Effect Transistors, arXiv, (2025) 2511.18915.
[5] D. Ding, M. Zou, R. Huang, T. Liang, H. Shi, K. Xu, C. Wong, Y. Chen, M. Han, L. Yu, G. Zhang, Y. Chen, Y. Yao, X. Zhang, Y. Zhou, C. Song, X. Zeng, Y. Peng, T. Deng, J. Xu, R. Sun, Multi-objective and cross-scale inverse design of temperature-control materials via physics-constrained machine learning, National Science Review, 13 (2026) nwag503.
[6] T. Lyu, J. Xiong, T.-Y. Zhang, Embedding science in AI for inverse materials design, National Science Review, (2026) nwag615.