Multiscale Simulation Framework for Predicting Cu-Cu and SiO2-SiO2 Hybrid Bond Quality in Semiconductor Packaging

Presented at the 2026 IEEE Hybrid Bonding Symposium, Jan 21-22, 2026 in Silicon Valley.  More information below.
(29:42 + Q&A) Ang Gao and Scott Burlison, University of Texas at Austin
Summary: Cu-Cu hybrid bonding is a transformative technology for high-density 3D integration in semiconductor packaging, offering superior performance in Through-Silicon Vias (TSVs) and other advanced interconnects. However, its adoption remains challenging due to the sensitivity of bond quality to void formation, interfacial roughness, misalignment, and oxide layers—all of which impact mechanical reliability, thermal conductivity, and electrical performance. Conventional in-line inspection techniques provide limited insight into the buried interfacial quality after bonding, restricting the ability to evaluate bond reliability or mechanical and electrical performance post-bonding. As hybrid bonding is a costly and irreversible process, there is an urgent need for predictive tools that estimate bond quality before dies are permanently joined. Such tools would enable optimal die placement and real-time yield optimization during manufacturing.
This work introduces a metrology-integrated, physics-based multiscale simulation framework to predict Cu-Cu hybrid bond quality prior to bonding. The proposed approach synergistically combines pre-bond surface characterization (e.g., Atomic Force Microscopy, AFM) with multiscale physics-based simulations and machine learning models to evaluate bonding potential for a given die pair. The framework integrates Molecular Dynamics (MD), Phase Field Modeling (PFM), and Finite Element Analysis (FEA) to capture bonding phenomena across atomic, mesoscopic, and die-level scales. Surface metrology data, including topography and dishing profiles, provides the initial conditions for PFM simulations that model thermal annealing and interface evolution. These simulations predict critical metrics such as bonded area fraction, void geometry, and surface roughness. The results are then used to guide localized MD simulations, which resolve atomic-scale interactions and predict key bonding properties including interfacial thermal boundary conductance, electrical resistivity, and adhesion energy. These atomic-scale outputs serve as inputs for FEA models, which compute die-level properties such as thermal and electrical conductivity, stress distributions, and temperature fields. Simultaneously, the FEA results are fed back to refine PFM simulations by modeling stress-assisted diffusion and void evolution—closing the multiscale feedback loop.
Bio: Ang Gao is a Ph.D. student in Mechanical Engineering at the University of Texas at Austin, conducting research in Dr. Michael Cullinan’s Nanoscale Design & Manufacturing Lab.

To access the full HBS program, and videos/slides for most of the presentations, visit https://attend.ieee.org/hbs/?page_id=779 

To hear about future HBS events, subscribe to our IEEE ListServ Dlist: http://listserv.ieee.org/cgi-bin/wa?SUBED1=ieee-hbs&A=1