Process for depositing amorphous aluminum oxychloride dielectric films with controlled chlorine retention

The semiconductor packaging industry is undergoing a structural transition driven by the end of conventional transistor scaling. Advanced packaging — heterogeneous integration, chiplets, fan-out wafer-level packages — is the primary vehicle through which the industry is extracting continued performance gains. At the center of that transition sits the redistribution layer (RDL), the fine-pitch dielectric and metal stack that routes signals between dies within a package. The dielectric material used in RDL formation is a chokepoint: it must be deposited as a conformal, defect-free amorphous film at temperatures compatible with back-end-of-line (BEOL) thermal budgets, and it must deliver adequate dielectric properties for sub-micron line-space structures. Aluminum oxide (Al2O3) deposited by atomic layer deposition has become the de facto amorphous RDL dielectric, but it carries well-known integration challenges including slow deposition rates in thick-film regimes and limited tunability of dielectric constant. This asset addresses a process gap that the incumbent approach does not: how to incorporate chlorine into the aluminum oxide network in a controlled, reproducible manner to yield an amorphous aluminum oxychloride (AlOxCly) film whose Cl content sits within plus or minus five atomic percent of any chosen target value. The commercial significance of controlled Cl retention is that chlorine in an oxide network acts as a network modifier, reducing bridging-oxygen connectivity and thereby softening the dielectric constant while maintaining amorphous film morphology. For RDL applications this offers a potential handle on permittivity that pure alumina does not readily provide. The process claim protected in this family covers the two practical deposition routes capable of achieving this: a melt-quench route that takes an Al/O/Cl melt to approximately 1,500 K and quenches rapidly to room temperature at a rate sufficient to preserve the amorphous structure with the target Cl content locked in, and a chlorine-retaining vapor or solution deposition route as a broader alternative pathway. Sitting within the portfolio of integrated packaging, storage, and PFAS-treatment systems, this asset is specifically the process anchor for the retained-chlorine RDL dielectric family, making it a complement to any composition claims in the same family rather than a standalone composition patent. The timing of this filing is relevant. The global push toward chiplet architectures — driven by AI accelerator demands that are reshaping packaging supply chains — is pulling packaging foundries to qualify new BEOL-compatible dielectric processes at a pace not seen in the prior decade. Qualification windows for novel dielectric processes at major packaging fabs typically open during transitions in node or package architecture; once a process is designed in and qualified, switching costs are high. Capturing a defensible process position now, while the RDL dielectric market is actively being re-evaluated, is strategically sensible.

Aluminum oxychloride is not a well-studied material in the thin-film context. The binary Al2O3 system has been exhaustively characterized both experimentally and computationally, but the ternary Al-O-Cl space — particularly in amorphous form — has received far less systematic attention. Aluminum chloride (AlCl3) and its hydrolysis intermediates are encountered in chemical vapor deposition (CVD) and atomic layer deposition (ALD) precursor chemistry, but in those processes chlorine is generally regarded as a contaminant to be minimized rather than a compositional variable to be controlled. This asset inverts that framing: chlorine retention at a defined, targeted level is the desired outcome, not a processing defect. The amorphous phase is essential because crystalline AlOxCly phases would be both structurally ill-defined for dielectric use and potentially subject to prior art; the amorphous network provides the homogeneous, defect-tolerant morphology required in RDL dielectric applications. The computational backbone of this asset is a melt-quench molecular dynamics simulation of the AlOxCly ternary system (internal reference: melt-quench MD simulation PKG-ALCLO-MQ-001). In this protocol, a simulation cell representing the Al/O/Cl composition is equilibrated at approximately 1,500 K — a temperature sufficient to achieve a well-mixed, liquid-like structural ensemble — and then quenched at a controlled rate to approximately 300 K. The key output of the simulation is confirmation that the amorphous structural character is preserved through the quench and, critically, that the chlorine content in the quenched structure falls within plus or minus five atomic percent of the starting composition. This directly validates the core process claim: that a melt-quench protocol of this type is capable, in principle, of producing an amorphous AlOxCly film with specified Cl retention. The simulation provides atomic-level structural evidence — bond angle distributions, radial distribution functions, coordination statistics around Al — consistent with a disordered network rather than any crystalline phase, and demonstrates that Cl atoms are retained in the network as Al-Cl bonds rather than volatilizing during quench. It is important to characterize what the MD simulation is and what it is not. Molecular dynamics at this level is a structural proof-of-concept: it establishes that the proposed quench pathway is physically plausible and that Cl retention within the claimed tolerance window is achievable under the simulated conditions. It does not constitute a phonon-stability calculation (phonon analysis is not meaningful for amorphous systems in the conventional sense), and accordingly the standard multi-potential consensus stability workflow — which validates crystalline or quasi-crystalline candidates by requiring agreement across MACE, CHGNet, MatterSim, and ORB potentials on the absence of imaginary phonon modes — is not applicable here. The amorphous AlOxCly system sits outside that crystalline-validation pipeline by design. The relevant computational validation is the MD trajectory itself, and the physical observables extracted from it (composition, structure factor, coordination environment) are the proof artifacts. Dielectric properties (bandgap, dielectric tensor from DFPT) have not yet been computed for this specific composition, and no experimental bandgap data is provided; these remain open validation gates. The process scope is deliberately broad across two manufacturing-compatible routes. The melt-quench route most directly maps to the MD simulation and provides the mechanistic foundation. The vapor-route and solution-route alternatives are claimed as additional embodiments, extending coverage to CVD, ALD, or wet-chemical deposition approaches in which precursor chemistry and post-deposition annealing are tuned to achieve the same Cl retention outcome. This route-agnostic framing is strategically valuable: it means a packaging fab could implement the claimed process through whichever deposition tool set is already qualified in their line, without being constrained to a single deposition modality.

The addressable market for RDL dielectric processes sits within the broader advanced packaging materials and equipment segment. Reliable third-party estimates place the advanced packaging materials market (including dielectrics, copper metallization, underfill, and encapsulants) in the range of tens of billions of dollars globally by the late 2020s, with the dielectric sub-segment representing a meaningful fraction. The specific sub-market for RDL dielectric deposition processes — encompassing both the materials consumed and the process IP embedded in deposition tool qualification agreements or licensing arrangements — is estimated at one to five billion dollars in addressable annual revenue, reflecting the concentration of high-volume advanced packaging activity at a relatively small number of major packaging foundries and IDM packaging lines. This is an estimate, not a certified market figure, and the actual capturable market will depend on adoption pace and licensing structure. The customer base is concentrated. Packaging fabs — OSAT houses (outsourced semiconductor assembly and test) and IDM-internal packaging lines at companies with leading-edge packaging programs — are the direct process customers. There are fewer than twenty facilities globally running at the volumes and pitches where an advanced RDL dielectric process claim is relevant, but those facilities are high-throughput and supply the entire AI accelerator, high-bandwidth memory, and mobile SoC ecosystem. Licensing economics in this segment often take the form of a process patent license tied to wafer starts or per-package royalties rather than materials sales, which means even a modest royalty rate applied to multi-billion-unit packaging volumes can produce substantial annualized revenue. Alternatively, an acquirer with direct relationships at packaging foundries could use this asset to control process qualification and extract value through supply-chain positioning rather than pure IP licensing. The race window for this process family is tied to the RDL dielectric qualification cycles underway now. Packaging roadmaps at the leading fabs typically lock in process decisions two to four years ahead of high-volume manufacturing introduction. A process patent in this space filed and granted before those qualification windows close has structural leverage; one filed after a process is already widely qualified faces a harder adoption argument. No explicit race-window flag was identified in the available context, suggesting this asset is in a phase where qualification-cycle timing should be tracked actively but has not yet closed.

retained-chlorine amorphous RDL dielectric process

The natural acquirers or licensees for this asset fall into two categories. The first is advanced packaging foundries and OSAT companies with active RDL technology development programs — entities such as ASE Group, Amkor Technology, JCET, and the packaging divisions of IDMs running in-house fan-out or 2.5D/3D packaging lines. These buyers would value the asset as a process license that enables or protects a differentiated RDL dielectric offering for their AI accelerator and HBM packaging customers. The second category is thin-film dielectric materials and equipment suppliers — companies like Applied Materials, ASM International, or specialty precursor suppliers — for whom a process patent covering a novel chlorine-retaining deposition approach could provide competitive differentiation in tool or chemical supply agreements with packaging fabs. Strategic fit is strongest for buyers who are already navigating the transition from ALD-alumina to next-generation RDL dielectrics and who have the process engineering resources to execute the experimental validation steps needed to advance this asset from MD-simulation-backed to experimentally confirmed. A buyer with an existing ALD tool portfolio and a precursor chemistry group is positioned to run the deposition experiments that would convert this asset from a process claim with simulation support to a claim with full experimental reduction to practice — substantially increasing its defensive and offensive value in the RDL dielectric landscape.

The most material risk for this asset is the gap between computational proof and experimental validation. The melt-quench MD simulation establishes plausibility but does not constitute reduction to practice in the experimental sense, and process patent claims can face enablement challenges if the claimed process has not been experimentally demonstrated. A challenger could argue that the plus-or-minus five atomic percent Cl retention tolerance, while achievable in simulation, has not been shown achievable in an actual thin-film deposition process at industrially relevant scales and conditions. This risk is addressable through targeted deposition experiments, but those experiments represent a near-term investment requirement for any buyer seeking to strengthen the asset's enforceability position. A secondary risk is competitive inertia at packaging fabs. ALD-alumina is deeply entrenched, and packaging foundries are conservative about introducing new dielectric chemistries into qualified process flows. Even a patented process with compelling properties requires a clear performance advantage — lower dielectric constant, better step coverage, higher throughput, or equivalent reliability at lower cost — to motivate fab qualification. If those advantages are not demonstrated experimentally, the market pull for the process claim may remain limited regardless of the patent position. The roadmap to de-risk this is straightforward: experimental deposition, analytical characterization, and dielectric property measurement to generate the performance data that creates fab qualification motivation alongside the IP position.