Yttrium aluminoborate (YAl3B4O12) huntite-phase dielectric for high-bandwidth-memory packaging

Yttrium aluminoborate — YAl3B4O12 in the huntite phase — is a glass-core dielectric candidate expected, but not yet computed or measured, to remain thermally stable above 800°C, with DFPT-computed permittivity within the 5–9 band required for high-bandwidth-memory redistribution-layer dielectrics. That combination does not exist in organic or polymer RDL dielectrics, which represent the current incumbent technology. As AI accelerator demand forces HBM stacking density higher, packaging thermal budgets tighten correspondingly, and the inability of conventional polymer dielectrics to survive those budgets creates a forced substitution window. YAB could close that window if coupon data confirm that the rigid aluminoborate huntite framework provides the thermal stability that polymers lack while staying within the permittivity range the electrical design requires. This asset stands as a distinct invention from the rare-earth-silicate compositions elsewhere in the critical-mineral recovery and recycling separations portfolio, prosecuted on its own continuation path so that IP in each structural family remains independent. a preliminary FTO screen, two ML potentials in agreement on dynamic stability, and an established sol-gel synthesis route together frame a de-risked starting position — the remaining open gates are experimental rather than conceptual, and the path to closing them is straightforward.

Candidates are screened for imaginary phonon modes with the machine-learning interatomic potentials named above. This is a computed harmonic check, not a measurement; where engines disagree, that is shown.

YAl3B4O12 crystallizes in the huntite structure, a trigonal framework built from corner-sharing AlO6 octahedra and BO3 triangles that locks the yttrium cation into a rigid coordination environment. That rigidity is the proposed structural origin of the expected thermal stability: the computed harmonic phonons show few soft modes, which suggests, but does not establish, that the material stays dimensionally and dielectrically stable below 800°C, where organic dielectrics degrade or decompose. Compositional tunability within the huntite family is achieved by partial substitution of La, Gd, or Lu at the yttrium site (0–30 at%) and Ga at the aluminum site (0–10 at%), providing levers to trim permittivity and CTE without leaving the structural family. Density-functional perturbation theory calculations give a computed dielectric tensor with permittivity in the 5–9 range; the loss figure below 3 × 10⁻³ is an estimate, since harmonic DFPT does not yield a loss tangent. These values sit in the sweet spot for HBM redistribution-layer dielectrics: high enough permittivity to support capacitive decoupling, low enough to avoid parasitic delay, and loss low enough for high-frequency signaling. A second norm-conserving DFPT dielectric-tensor calculation is in progress as an independent cross-check on these values. It should be noted that this DFPT calculation was completed by the invention team on the YAB composition specifically — YAB does not appear in public DFPT reference databases, so the computational result is the team's own, not a literature retrieval. A buyer should weight that appropriately when staging validation milestones. Synthesis follows a sol-gel route using yttrium nitrate, aluminum sec-butoxide, and trimethyl borate, with a final anneal at 900–1100°C. This is a standard ceramic precursor route and a concrete path to densified film fabrication for coupon testing; compatibility of its anneal with glass-core substrate thermal budgets is untested.

We estimate the addressable market at $1–2 billion across advanced semiconductor packaging and high-bandwidth-memory substrates. The primary buyers are HBM makers integrating redistribution layers and glass-core substrate vendors supplying them. The royalty logic is straightforward: value is a function of dielectric content per HBM substrate multiplied by HBM unit volume, a quantity expanding rapidly as AI-accelerator architectures stack more HBM dies per package. Even a modest per-unit royalty on qualified YAB material compounds meaningfully as HBM shipment volumes grow. The addressable set is deliberately narrow — this is not a commodity packaging dielectric play, it is a high-temperature specialist positioned for the thermal-budget-constrained tier of HBM packaging. That narrowness is also what commands premium pricing: a dielectric that survives where polymers fail can capture the value of the thermal headroom it provides, rather than competing on cost with commodity polyimides. The concentration of HBM volume among a small number of memory manufacturers and their glass-core suppliers means licensing or supply discussions can be targeted efficiently.

expected high Tg at the required eps range, to be confirmed on a film coupon -> HBM packaging

packaging-dielectric huntite use (RDL/TGV/glass-core/interposer) distinguishes scintillator YAB; scintillator/luminescent + RE-doped glass scintillator articles expressly excluded in the claim per §7.5.7

Preliminary screen only. This is not a freedom-to-operate opinion.

The natural licensees are HBM makers and glass-core substrate vendors. HBM makers are the most strategically motivated: the high-Tg advantage maps directly onto their packaging thermal budgets, and a field-of-use license scoped to HBM redistribution-layer dielectrics is a clean structure that aligns royalty flow with the value created. Glass-core vendors are candidate licensees for broader glass-core and interposer applications, where the same Tg and permittivity combination is advantageous even outside the strictest HBM thermal tier. A memory or advanced-packaging strategic acquiring the full critical-mineral recovery and recycling separations portfolio would receive differentiated coverage across the dielectric design space, making YAB more valuable as part of a combined transaction than in isolation. Most buyers will prefer a license with milestones tied to the densified-film experimental qualification gate — closing that gate with measured eps, loss, and Tg data substantially de-risks the asset and is the logical point at which license economics firm up. The concentrated HBM supplier base means buyer conversations can be targeted to a short list of strategics whose roadmaps are publicly documented.

The dominant risk is that all three headline properties — permittivity, loss tangent, and glass-transition temperature — are currently computational. The DFPT dielectric tensor was completed by the invention team on a composition not present in public reference databases, so the result has not been independently cross-checked against measurement or prior literature. If the densified-film coupon returns permittivity or loss values outside the claimed range, or if the Tg does not clear 800°C, the entire HBM thermal-budget positioning requires revision. Additionally, the norm-conserving DFPT cross-check calculation is still in progress, meaning even the computational validation is not yet complete. CTE compatibility between a densified YAB film and target glass-core substrates is also unconfirmed and must be established before any packaging qualification can begin. The de-risking roadmap is well-defined and executable. The priority experiment is fabrication of a densified YAB film via the specified sol-gel and 900–1100°C anneal route, followed by coupon measurement of dielectric constant, loss tangent, and Tg. This single experiment closes the most consequential uncertainty. If those measurements confirm the computational predictions, the asset moves from a strong computational case to a validated material ready for packaging integration trials. No blocking patents surfaced in the preliminary screen (not an FTO opinion), and two machine-learned potentials agree on phonon stability; the headline properties are unmeasured, so the remaining exposure is largely experimental.