Amorphous aluminum oxychloride ALD film with deliberately retained bound chlorine

The glass-core advanced-packaging substrates portfolio includes a process-defined dielectric film that is differentiated not by a new element or exotic chemistry, but by what it deliberately keeps: bound chlorine. Conventional atomic-layer deposition of alumina using trimethylaluminum and water drives off essentially all chlorine from the precursor cycle, yielding dense Al2O3. This invention intentionally retains 2–25 atomic percent chlorine in the film as Al-O-Cl bonds, producing an amorphous aluminum oxychloride phase whose dielectric constant is meaningfully lower than dense alumina and which simultaneously passivates copper interconnects against corrosion under bias and humidity. The differentiation is thermally verifiable — the retained chlorine is measured by depth-profiling (XPS or equivalent) after the full thermal budget — which means the claim boundary is analytically sharp and hard to design around by accident. The timing is driven by a structural shift in advanced packaging. As glass substrates replace organic laminate in high-density redistribution-layer (RDL) stacks, the dielectric and passivation films covering copper traces face more demanding flatness, temperature, and hermeticity requirements than organic processes tolerated. Dense alumina meets hermeticity but comes with a dielectric constant that penalizes capacitance; silicon nitride passivates but requires plasma conditions incompatible with some glass-core flows. An amorphous Al-O-Cl film tunable between these extremes — low-k, conformal, and deposited by standard ALD hardware — addresses a real gap in the materials palette available to packaging engineers today. This is a lead asset within the family, meaning the core commercial embodiment and the broadest process claims sit here. The supporting defensive filings and method variants are separate. The investment case rests on the process claim breadth covering multiple chlorine-bearing ALD precursors, a clean freedom-to-operate position established by the post-anneal bound-chlorine measurement limitation, and a total addressable market in advanced-packaging dielectrics that is already measured in billions of dollars and growing as AI accelerator packaging complexity increases.

The material is an amorphous aluminum oxychloride, written AlOxCly, where x and y are determined by ALD process parameters rather than by a fixed crystal stoichiometry. The critical structural feature is that chlorine is retained as covalent or ionic Al-Cl bonds rather than as surface contamination or physisorbed HCl — a distinction that is confirmed spectroscopically by the binding-energy signature of chlorine in XPS depth profiles taken after the device thermal budget. The 2–25 at% chlorine window is broad enough to encompass multiple ALD precursor chemistries and dose/purge conditions, which is intentional: the family covers the range rather than a single composition, giving commercial flexibility while keeping the analytical test (bound-Cl post-anneal) as the definitive membership criterion. The computational validation program for this asset follows a specific logic dictated by the amorphous nature of the target phase. Standard crystal-stability workflows — phonon calculations on a periodic unit cell, cross-validated by multiple machine-learning interatomic potentials — are not directly applicable to an amorphous film. Instead, the computational work was directed at chemically relevant proxy calculations. A chlorine-vacancy formation energy of +2.6 eV was computed using two independent simulation engines, establishing that once chlorine is incorporated into the Al-O-Cl network it is thermodynamically disfavored to leave at moderate temperatures — a quantitative rationalization for the thermal stability of the retained-chlorine phase. Separately, a semiempirical extended tight-binding (xTB) precursor ranking was performed across the chlorine-bearing ALD precursor space (TMA+HCl, methylaluminum dichloride, dimethylaluminum chloride, and AlCl3) to rank their tendency to leave residual Al-Cl bonds after the purge half-cycle, providing process-chemistry guidance that directly informed which precursors appear in the claim set. A particularly important computational result is the explicit confirmation that crystalline AlClO — the ordered phase one might naively propose as a model for this material — is dynamically unstable according to two independent simulation engines. This is not a weakness of the invention; it is a feature of the claim strategy. The commercial embodiment is explicitly amorphous, and the instability of the crystalline polymorph validates why process-defined amorphous structure is the correct phase space to occupy. The negative limitations in the claims (crystalline AlClO excluded) are computationally grounded. This kind of negative-result data, drawn from the portfolio's atlas of failed and unstable compositions, directly sharpens claim scope and supports prosecution arguments distinguishing this material from prior crystalline oxide literature. On the dielectric and passivation properties: dense amorphous Al2O3 deposited by TMA+H2O ALD typically shows a dielectric constant in the range of 8–9, substantially higher than the 3.5–4.5 window targeted by low-k interlayer dielectrics in advanced packaging. The incorporation of Al-Cl bonds into the amorphous network introduces lower-polarizability units and increases free volume in the amorphous structure, both of which reduce the dielectric constant. The copper passivation function is attributed to the Al-O-Cl network's ability to form a stable, conformal barrier against Cu+ migration under bias, supplemented by the Cl atoms' tendency to block oxidative attack pathways at the Cu/dielectric interface — a mechanism consistent with the high Cl-vacancy formation energy computed above. These property hypotheses remain to be directly validated on coupon hardware, which is the primary open proof gate described below.

The immediate commercial context is the RDL passivation and interlayer-dielectric market within advanced semiconductor packaging. Redistribution layers in high-density fan-out and glass-core packages require thin-film dielectrics that are conformal, low-leakage, chemically compatible with copper, and manufacturable on the same ALD equipment already installed in leading-edge packaging lines. The total addressable market for advanced-packaging dielectric films — spanning passivation, IMD, and barrier applications across OSAT, IDM, and substrate suppliers — is estimated at $1–3 billion annually, a figure that is growing as AI accelerator packages push toward higher interconnect densities and tighter dielectric performance specifications. The buying logic for this technology is primarily licensing into existing ALD process flows rather than materials supply. A packaging house running TMA-based ALD already owns the hardware; adopting an Al-O-Cl process requires only a precursor qualification, a recipe change, and a process integration study. That low switching cost for hardware — combined with meaningful performance gain — creates a licensing model where royalties can be tied to wafer starts processed using the claimed recipe. Glass-core substrate suppliers (a growing segment as companies like Corning, AGC, and their packaging partners develop glass interposers for HBM and chiplet stacks) represent a secondary and arguably more captive customer base, because glass-core RDL is a newer process flow where there is no entrenched incumbent recipe to displace. Adjacent applications extend the market beyond packaging. Amorphous Al-O-Cl films with tunable dielectric constant are relevant anywhere ALD alumina is currently used as a gate dielectric, capacitor dielectric, or encapsulation layer — including DRAM capacitor stacks, MEMS encapsulation, and thin-film photovoltaic moisture barriers. These adjacent segments are not the primary claim target but represent optionality for licensees who qualify the process for packaging and then extend it laterally.

lower-k + Cu passivation vs dense alumina

deliberate bound-Cl retention measured post-anneal distinguishes conventional TMA+H2O alumina

The most direct acquirers or licensees are advanced-packaging OSATs and substrate suppliers who operate ALD tools in their passivation or IMD process flows and who are actively developing glass-core or high-density fan-out packages for AI accelerator and HBM customers. Companies with significant ALD-based dielectric capability in packaging — including major OSATs integrating glass interposer stacks and substrate manufacturers developing next-generation RDL — represent the natural first-tier licensing targets. The process-claim structure makes licensing straightforward: royalties can be tied to wafer starts under a qualifying recipe, and the analytically specific claim boundary (post-anneal depth-profile Cl measurement) makes compliance auditing tractable. A second tier of strategic buyers includes semiconductor capital-equipment and precursor companies who would value the invention as a product-enabling IP block — an ALD precursor supplier offering MADC or DMACl, for example, has a direct commercial incentive to license or acquire claims that anchor the value of those chemistries to a measurable performance outcome in packaging. Materials suppliers in the specialty ALD precursor space, and the tool vendors who qualify new dielectric processes on their platforms, would find this family valuable as either a defensive holding or an offensive licensing position against customers moving to chlorine-bearing alumina processes.

The primary technical risk is the transition from simulation-supported process rationale to demonstrated coupon results. The Cl-vacancy formation energy and precursor ranking are computationally credible, but the actual retained-chlorine concentration achievable in a real ALD cycle — and whether it falls within the 2–25 at% window claimed across all four precursor chemistries — depends on reactor geometry, purge times, deposition temperature, and substrate surface chemistry in ways that simulations do not fully capture. There is a non-trivial probability that some precursor-condition combinations produce either too little retained chlorine (falling outside the claim window) or too much (causing reliability issues). The coupon XPS measurement is the critical gate to de-risk this, and it should be the first experimental milestone for any party taking the asset forward. A secondary risk is commercial adoption speed. Even if the technical case is validated, ALD process qualification at packaging fabs is a 12–24 month cycle involving equipment qualification, materials compatibility studies, and reliability testing. The competitive window narrows if alternative low-k passivation approaches — fluorine-doped alumina, engineered SiOCH, or new plasma-enhanced nitride processes — advance to production qualification before this family is licensed and validated. The mitigation is to prioritize the XPS coupon experiment now, which can be accomplished at relatively low cost and converts the asset from a computational-case invention to an experimentally grounded one, substantially improving licensing attractiveness and valuation.