A Review of Multiscale Thermal Modeling in Heterogeneous 3D ICs
It provides a comprehensive framework for researchers and engineers to navigate thermal modeling challenges in advanced 3D ICs, but is a review without new experimental results.
This review synthesizes multiscale thermal modeling methods for heterogeneous 3D ICs, analyzing trade-offs among compact models, FEM/FDM, Green's function, reduced-order methods, and physics-informed ML, and provides design guidelines and a research agenda for thermally aware co-optimization.
Thermal behavior has become a first-order constraint in advanced 2.5D/3D integrated circuits (ICs) and heterogeneous packages. As power densities rise and multiple active dies are vertically integrated, heat removal paths become constricted, elevating junction temperatures, magnifying temperature gradients, and exacerbating reliability risks. This review synthesizes the physical mechanisms, modeling assumptions, and analysis methods that govern multiscale thermal transport in 3D ICs, with emphasis on interface-dominated conduction, material anisotropy, and strong electrothermal coupling. We unify device-to-system scales into a coherent framework, analyzing trade-offs among compact thermal models (CTMs), finite element/finite difference methods (FEM/FDM), Green's function and semi-analytical techniques, reduced-order and multi-fidelity methods, and physics-informed machine learning (PIML), while highlighting the central role of thermal boundary resistance (TBR) and variability in thermal interface materials (TIMs), the pitfalls of decoupled electrical/thermal analyses, and the need for rigorous validation against measurements. Finally, we outline practical design guidelines and a forward-looking research agenda that integrates physics-based modeling, data-driven surrogates, and in situ sensing to enable thermally aware co-optimization across the IC-package-system hierarchy.