Copper tin intermetallic stress-redistribution interlayer for glass-core package reliability

The advanced-packaging industry is in the middle of a forced transition from organic to glass cores. Glass offers dramatically lower coefficient of thermal expansion (CTE), superior planarity, and the electrical properties needed for next-generation chiplet integration — but it brings a reliability liability that organic packages largely avoid: the CTE mismatch between the glass core (roughly 3-6 ppm/°C) and the copper redistribution layers and vias (17 ppm/°C) generates thermomechanical stress that drives warpage, delamination, and fatigue cracking across thermal cycles. Solving that mismatch at the interface is not optional; it is the gating reliability problem for glass-core adoption at scale. This asset, a member of the glass-core advanced-packaging substrates portfolio, addresses that problem through a different mechanism than the family's primary arm: rather than engineering a material whose CTE sits between glass and copper, Cu6Sn5 works through modulus-based stress redistribution. With a Young's modulus of approximately 112 GPa — intermediate between glass (~70 GPa) and bulk copper (~130 GPa) — a Cu6Sn5 interlayer inserted at the glass-copper interface acts as a mechanical compliance gradient, spreading and redistributing thermomechanical stress rather than eliminating the CTE differential directly. This is an important honest distinction from CTE-matching claims made elsewhere in the family; the mechanism is modulus-management, and the specification discloses it as such. The asset is classified as a backup filing within the portfolio, playing a defined defensive and breadth role. The primary claim of direct CTE matching is carried by other family members, including graded-modulus interlayer constructions and intermediate-CTE oxide variants. Cu6Sn5 and its sister intermetallics (Cu3Sn, Ag3Sn, Ni3Sn4) broaden the family's protected composition space, complicate design-arounds, and position the portfolio to capture value across multiple assembly-process pathways. A buyer of this portfolio should understand Cu6Sn5's role precisely — it is not the lead composition, but it is computationally validated, cleanly cleared on freedom-to-operate, and contributes real claim coverage for a composition class that is directly relevant to solder-derived intermetallic layers already present in flip-chip and advanced packaging assembly flows.

Each candidate is validated by multiple independent machine-learning interatomic potentials. A material advances only when the engines agree on phonon (dynamic) stability — disagreement is surfaced, not hidden.

Cu6Sn5, also called the eta-phase of the copper-tin binary system, crystallizes in a monoclinic structure with space group C2/c. In the context of glass-core packaging, it is not a synthetic curiosity — this phase forms naturally and unavoidably at copper-solder interfaces during reflow, meaning it is already present in the assembly flows of every major packaging house. What this asset claims is the deliberate, ordered placement of Cu6Sn5 (or its compositionally related intermetallics) in the stress-bearing region between the glass core and the copper metallization, rather than treating it as an incidental reaction product to be minimized. The key property enabling the stress-redistribution function is Young's modulus. Computational evaluation places Cu6Sn5 at approximately 112 GPa. This sits between the modulus of common glass substrates (~70 GPa) and bulk copper (~130 GPa), creating a mechanical compliance gradient at the interface. By interpolating stiffness across the glass-Cu junction, the interlayer distributes stress concentrations that would otherwise localize at a hard interface. It is worth being precise about what this mechanism is not: Cu6Sn5 has a CTE that is broadly Cu-like, meaning it does not provide CTE matching to glass. The warpage and delamination relief this layer provides comes from modulus management, not CTE interpolation. The specification is written with this candor, which is appropriate — it makes the claims more defensible and avoids conflation with the family's CTE-matching arms. Computational validation of Cu6Sn5 stability was performed using two independent machine-learning interatomic potentials: MACE and CHGNet. Both potentials were used to compute phonon dispersion, and both returned results indicating dynamic stability — there are no imaginary (negative-frequency) phonon modes in either calculation. The lowest-frequency acoustic branch in the MACE calculation sits at approximately 0.227 THz, and the CHGNet result returns 0.138 THz. The agreement between two independent ML potential frameworks, trained on distinct datasets and using distinct architectures, provides a meaningful consensus on structural validity. Additionally, two DFT source calculations support the structure. The package of evidence — dual-MLIP phonon consensus plus DFT — clears the dynamic stability gate that Lattice Graph requires before advancing a material to disclosure. Open validation work remains. The primary experimental gate is a warpage and four-point bend coupon test, which would directly demonstrate stress redistribution in a fabricated glass-core stack incorporating the Cu6Sn5 interlayer. This is the expected next step before any licensing discussion could cite physical demonstration data. Two candidate compositions were evaluated and formally excluded from the family during prosecution: La2O3 and Cu2O were assessed as interlayer candidates and not advanced. Their exclusion is captured as negative limitations, which contributes to the portfolio's atlas of labeled failed-experiment results and provides prosecution-history support for the claimed compositions' distinctiveness. The broader intermetallic family — Cu3Sn, Ag3Sn, and Ni3Sn4 — shares the structural role and is included in the composition claims, reflecting the fact that multiple intermetallic phases accessible from standard soldering processes can serve the same stress-redistribution function.

The direct addressable market for this asset is the glass-core substrate reliability segment of advanced packaging — specifically, the suppliers of glass-core substrates, the semiconductor assembly houses integrating those substrates, and the chipmakers specifying reliability requirements for next-generation advanced packaging. The total addressable market for glass-core substrate adoption is estimated at $0.2 to $0.5 billion over the near-to-medium term horizon, reflecting the segment's current status as an early-commercial and rapidly scaling technology rather than a mature commodity. These estimates are necessarily approximate given the technology's trajectory. The purchasing logic for a licensee or acquirer differs by segment. A substrate manufacturer integrating this interlayer approach into their glass-core product benefits from a reliability-differentiated product and a defensible IP position against competitors. An assembly house benefits from a process solution that leverages intermetallic phases they already form incidentally — turning a reaction byproduct into a designed functional layer requires process adjustment but not a fundamentally new chemistry. The royalty model most naturally fits a per-unit substrate license, where the relatively modest per-substrate cost of IP access is justified by the reliability improvement and the reduction in warranty exposure from thermomechanical fatigue failures. Given that glass-core substrates are expected to carry high-value AI accelerator and advanced chiplet packages, the cost-per-substrate justification for reliability-enabling IP is favorable. The timing of market entry matters here. Glass-core substrates are transitioning from pilot to production at major substrate manufacturers including players in Japan, South Korea, and the United States, driven by TSMC, Intel, and Samsung roadmaps for next-generation packaging. Thermomechanical reliability is a primary qualification gate for these transitions. IP covering stress-redistribution interlayer architectures — particularly those claiming the specific intermetallic compositions that already appear in assembly flows — is most valuable in the window before design-freeze and manufacturing qualification, when substrate developers are still selecting architectural approaches. That window is open now and will narrow as design choices lock in over the next two to three years.

converts asserted CTE match into a positively-claimed structure

ordered placement in stress-bearing region

The most direct strategic buyers for this asset, as part of the glass-core advanced-packaging substrates portfolio, are glass-core substrate manufacturers and OSATs (outsourced semiconductor assembly and test providers) that are actively qualifying glass-core architectures for AI accelerator and advanced chiplet packaging. Companies in this category include the major Japanese and Korean substrate manufacturers with active glass-core development programs, as well as Intel Foundry Services and TSMC-affiliated substrate suppliers that have publicly announced glass-core roadmaps. For these buyers, the interlayer family provides IP coverage for a reliability mechanism that they will likely need to address regardless of whether they license this specific asset, making it more valuable as a defensive acquisition than as a licensing royalty stream. Secondary buyers include advanced materials companies supplying copper-tin solder systems, intermetallic preforms, or structured interlayer films for packaging applications, for whom this asset provides both offensive licensing capability and freedom-to-operate confidence. A strategic acquirer of the full glass-core advanced-packaging substrates portfolio would receive this asset as part of the family's breadth coverage — its value is additive to the portfolio rather than standalone, and pricing should reflect that complementary role. The $0.2-0.5 billion addressable market estimate implies a licensing revenue ceiling that positions this as a portfolio-bundled asset rather than a primary standalone licensing target.

The primary risk for this asset is its backup classification within the family. The modulus-management mechanism, while technically honest and computationally supported, is less intuitively compelling to a commercial audience than CTE-matching, and CTE-matching is what the market narrative around glass-core reliability typically emphasizes. If the primary CTE-matching arms of the family achieve strong commercial uptake, this backup arm becomes valuable as defensive breadth; if the family's commercial traction is limited, the backup arm's standalone value is modest. The absence of experimental warpage data is a material validation gap — the four-point bend coupon test remains the key de-risking milestone, and its completion would substantively improve the asset's licensing position. A secondary risk is that the very ubiquity of Cu-Sn intermetallic phases in standard solder assembly — the same property that makes them commercially natural — could complicate claim enforcement if competitors argue their processes form Cu6Sn5 incidentally rather than deliberately in the stress-bearing position. The ordered-placement limitation in the claims is the primary defense against this argument, and the strength of that limitation will be tested if enforcement action is ever pursued. The recommended de-risking path is sequential: complete the warpage coupon demonstration to establish experimental proof-of-concept, then pursue licensing discussions with glass-core substrate manufacturers during the current qualification window before design choices lock in.