The semiconductor industry is embracing the "Chiplet Revolution," transitioning from monolithic integration to chiplet-based architectures to overcome Moore's Law limitations. Chiplets, smaller integrated circuits assembled onto a single package substrate, offer flexibility within System-in-Package (SiP) platforms. While architectures utilizing Si interposers or interconnect bridges have become mainstream for high-performance electronic systems, their reliance on an intermediary organic package for board-level attach has positioned glass-panel substrates as a promising alternative to further extend package sizes owing to its superior dimensional stability, high modulus and tailorable coefficient of thermal expansion (CTE). The latter is particularly desirable to balance chip- and board-level reliability while enabling direct surface mount assembly of the advanced glass packages to the printed circuit board. While this CTE mismatch is manageable with compliant solder-based interconnections, the system-level reliability challenge becomes more complex with emerging all-Cu interconnections. With improved pitch scalability, power handling capability and thermal stability, all-Cu interconnections are expected to replace traditional Cu micro-bump interconnections but introduce unprecedented reliability challenges when applied to non-CTE-matched chip-to-package architectures, requiring exploration of novel strategies to improve the reliability of the fragile on-chip back-end-of-line (BEOL) layers. In this study, finite-element modeling was used to assess the risk of crack propagation in these BEOL layers for a chip-on glass-on board architecture employing 10 μm pitch all-Cu chip-to-glass-substrate interconnections. A global-local model approach offers insights into the impact of glass core CTE and gap fill material selection on system-level reliability based on simulated values for strain energy release rate for an interfacial crack in the BEOL and plastic strain in the package-to-board BGA. This highly flexible model can be used to select combinations of geometric and material parameters that allow for simultaneous chip-level and board-level reliability based on pre-established screening criteria such as maximum allowable crack energy release rate, package warpage, and accumulated plastic strain in the BGAs.
No takes yet. Share an insight, caveat, or question.
Sosa et al. (2024) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: