Wide-bandgap borate and aluminoborate insulating liner alternatives for glass-core vias

The glass-core advanced-packaging substrates portfolio addresses one of the most consequential materials challenges in advanced semiconductor packaging: the need for high-quality electrical insulation on the interior walls of through-glass vias (TGVs). As chipmakers migrate from organic laminates and silicon interposers toward glass substrates — driven by better dimensional stability, lower signal loss, and compatibility with tighter via pitches — the via-wall liner becomes a critical enabler. A liner material must provide genuine electrical isolation between the conductive copper fill and the glass host, withstand the thermal cycling of reflow, remain pinhole-free at atomic layer deposition (ALD) thicknesses, and sit within a manufacturable deposition window. The borate and aluminoborate sub-family described here expands the portfolio's liner coverage meaningfully beyond the primary alumina-borate lead. SrB4O7, Al5BO9, and Ba2Mg2 bring a range of bandgaps — from approximately 4.9 eV up to roughly 7.3 eV — each large enough to function as a genuine wide-bandgap insulator in a TGV liner context. BPO4 is included as a backup arm with a narrower computational validation footprint but a clear freedom-to-operate position on body claims. Together, these four compositions form a coherent defensive and coverage-broadening family that prevents a competitor from designing around the lead compound by substituting a closely related borate or aluminoborate phase. Strategically, this asset is best understood as a supporting extension rather than a standalone flagship. Its value lies in the portfolio architecture: by securing composition and device-use claims across this sub-family, the licensor forecloses the obvious workaround routes for any party seeking to deploy a wide-bandgap borate liner in a TGV stack without engaging with the portfolio. For a prospective acquirer or licensee in the ALD liner supply chain, this breadth makes the portfolio substantially harder to design around and substantially more valuable as a licensing position.

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.

The three primary compositions in this sub-family span two crystal symmetry classes. SrB4O7 adopts an orthorhombic structure and exhibits a computed bandgap of approximately 7.3 eV, placing it in the upper tier of wide-bandgap oxides — a figure comparable to MgO and approaching that of AlN. This is a structurally well-characterized phase with known stability in ambient conditions, making it a credible ALD precursor target. Al5BO9 is a mullite-like aluminoborate with trigonal symmetry; its computed bandgap is approximately 5.0 eV, and its structural relationship to mullite (the canonical high-temperature alumino-silicate ceramic) provides useful prior art in thermal stability, albeit in bulk rather than thin-film form. Ba2Mg2 carries a computed bandgap of approximately 4.9 eV and brings magnesium into the borate network, offering an additional compositional axis for tuning dielectric properties or deposition chemistry. BPO4 — the backup arm — is a borophosphate with known crystallographic characterization and good baseline electrical properties, included to anchor the family at a simpler binary phosphate end member. Dynamic stability screening for the three primary compositions was performed using two independent machine-learning interatomic potentials: MACE and CHGNet. These are distinct ML potential frameworks trained on separate datasets and employing different architectures, meaning their agreement constitutes a genuine independent cross-check rather than a correlated result. For the sub-family as a whole, both potentials return consistent phonon density-of-states results with no imaginary (negative-frequency) phonon modes, confirming that the structures are dynamically stable — that is, they sit at genuine local minima on the energy landscape rather than at saddle points. Minimum phonon frequencies from the two engines span roughly 0.737 THz (MACE) to 0.78 THz (CHGNet), indicating good agreement in the low-frequency acoustic regime where errors in ML potentials are most consequential for predicting mechanical stability. DFT ground-state references from two independent source calculations underpin the energy baseline for these assessments. For via-wall liner applications, the properties that matter most are: bandgap width (directly setting the leakage floor and determining the insulator's usefulness at nanometer ALD thicknesses), structural stability under deposition and thermal cycling conditions, and compatibility with established ALD precursor chemistries. The 7.3 eV bandgap of SrB4O7 is particularly notable — it exceeds that of the commonly used SiO2 liner (~9 eV is theoretical but thin-film effective gap is much lower) in practical contexts, and is substantially wider than HfO2 (~5.7 eV) or Al2O3 (~8.8 eV bulk but lower effective in thin films). Al5BO9 and Ba2Mg2 at ~4.9-5.0 eV remain competitive with HfO2 and superior to TiO2-based options. None of these compositions has been reported in the via-wall ALD liner context in the patent literature, which is the basis for the freedom-to-operate position described below. The next validation gate for all three primary compositions is physical film deposition — specifically, thin-film coupons deposited by ALD or a suitable analog process, followed by measurement of effective bandgap, leakage current density, and interface quality against a glass substrate. This is an open gate that computational work cannot close; it requires laboratory synthesis. Until that coupon data exists, the computational stability results should be treated as strong positive indicators rather than proof of device performance.

The addressable market for via-wall liner materials in glass-core advanced packaging is estimated at roughly $0.5-1 billion annually at maturity, reflecting both the material volumes involved in high-volume TGV fabrication and the substantial value-add of ALD liner processes relative to bulk substrate cost. This estimate should be understood as an order-of-magnitude figure for the materials and process licensing opportunity; the actual realized addressable market will depend heavily on the pace of glass-core substrate adoption across OSDP (Organic Substrate Die Packaging), silicon bridge interposer replacements, and RF/millimeter-wave front-end modules, all of which are currently in active ramp phases at multiple major substrate suppliers. The buyer profile for liner materials or liner-process licenses is primarily ALD equipment and precursor companies, substrate manufacturers integrating TGV processing in-house, and fabless/IDM packaging customers specifying liner chemistry in their supply chain. The relevant customer touchpoint is the ALD process flow for via-wall liner deposition — a step that occurs early in the TGV stack build and sets the electrical isolation baseline for everything built above it. Licensing logic follows a per-wafer or per-substrate royalty model standard in the specialty materials and process IP space, with the alternative being an upfront portfolio acquisition for a party seeking to own the landscape. The composition-plus-device-use claim structure is well-suited to enforcement at the materials supply level (precursor manufacturer) or at the substrate processing level (liner deposition step), giving the licensor flexibility in where along the value chain to enforce or license.

wide-gap insulating liner alternatives to AlBO3

thin-film packaging form; polymer-whisker/CMC + RE-aluminoborate forms excluded

The most natural acquirers or licensees for this asset are parties active in the ALD precursor and liner process supply chain for advanced packaging, or substrate manufacturers building internal IP positions in glass-core TGV technology. This includes companies supplying ALD equipment and precursor chemicals to semiconductor substrate fabs, substrate manufacturers (particularly those with captive glass-core substrate programs), and advanced packaging foundries seeking to differentiate on via performance. Given the clean FTO position and the composition-plus-device-use claim structure, a strategic acquirer could use this sub-family defensively — preventing competitors from freely using borate liner chemistries — or offensively, as a licensing position against substrate suppliers who adopt borate-based liner processes as Al2O3 alternatives. The sub-family is also a natural fit for any party that acquires or licenses the primary alumina-borate lead in the glass-core advanced-packaging substrates portfolio, since the two assets together form a more complete landscape position. Separately, materials companies with existing borate or aluminoborate chemistry programs — particularly those active in nonlinear optics or ceramic processing — may find value in holding this IP even if their primary business is not packaging, as a blocking or cross-licensing asset against packaging-focused competitors who might otherwise enter their adjacent markets.

The primary technical risk is the film deposition gate: none of the three primary compositions has been deposited as an ALD thin film on a glass TGV substrate and electrically characterized. ALD precursor chemistry for strontium borates and magnesium borates is less mature than for Al2O3, meaning the path from computational validation to manufacturable liner process is longer and less certain for these compositions than for the lead alumina-borate compound. It is possible that some compositions prove difficult to deposit at the required thicknesses and conformality, even if they are computationally stable. The market risk is timing: glass-core substrate adoption is accelerating but remains early-stage relative to organic laminates. If the transition to glass-core substrates at high volume takes longer than currently projected, the commercial window for a premium liner IP position compresses. The roadmap to de-risk both dimensions is straightforward in principle: targeted ALD deposition experiments on SrB4O7 and Al5BO9 film coupons (the two highest-bandgap options) would either confirm or falsify the compositions as practical liner candidates within a laboratory program, and the results would sharpen both the technical claim set and the licensing narrative significantly. BPO4's backup status should be maintained rather than elevated until two-engine phonon consensus validation is completed.