Topology-optimized microchannel cooling for heterogeneous logic-HBM heat sources: thermal-hydraulic optimization and thermo-mechanical assessment
Efficient cooling of heterogeneous 2.5D packages remains a critical challenge due to the spatially non-uniform heat generation and distinct thermal requirements of logic and High Bandwidth Memory (HBM) regions. However, existing microchannel heat sink (MCHS) studies predominantly focus on thermal-hydraulic enhancement under nominal operating conditions, leaving the combined effects of non-uniform logic-HBM heat loads, operating-condition variations, inter-die temperature differences, and thermally induced stress insufficiently addressed. To bridge this gap, a density-based thermo-fluid topology optimization is performed for the MCHS of a heterogeneous logic-HBM system. The optimized 2D layouts are reconstructed into three-dimensional (3D) microchannel heat sinks and systematically evaluated by CFD and thermo-mechanical simulations under different operating conditions. Results indicate that the inlet/outlet arrangement mainly dictates the primary flow path, whereas secondary branches shorten solid-conduction paths to suppress local hotspots. Compared with the straight microchannel baseline, the optimized layouts reduce maximum thermal resistance by ∼30%, cut pumping power by 57%. Further evaluations under varying flow conditions and logic power ratios confirm that the topology-optimized layouts retain their cooling advantages beyond the nominal design point. Crucially, the post-optimization thermo-mechanical evaluation of the MCHS reveals a 70% reduction in maximum thermally induced von Mises stress (from 35.59 to 10.75 MPa). These results demonstrate that topology optimization provides an effective strategy for enhancing the thermo-hydraulic performance, while inherently mitigating thermally induced stress within the MCHS.
Authors
- Boxiang Wang
- Meihong Zhao
- Shuai Gong
- Siping Gao
Institutions
- Shanghai Jiao Tong University (CN)
- Chinese Academy of Sciences (CN)
- Shanghai Institute of Microsystem and Information Technology (CN)
Publication Details
- Journal
- Applied Thermal Engineering
- Published
- 2026-10-07
- DOI
- https://doi.org/10.1016/j.applthermaleng.2026.133538
- Primary Topic
- Heat Transfer and Optimization
- Type
- article
- Field-Weighted Citation Impact
- 0.00