Catalog of computationally excluded compositions supporting non-obviousness of disclosed TIM fillers

Every patent prosecution involving a broad composition claim faces the same adversarial question: why these materials and not the thousands of others that seem structurally analogous? The honest answer — that most candidates fail on fundamental physical grounds — is enormously more persuasive when it is documented, systematic, and computational rather than asserted. This asset is that documentation. It is a formally assembled catalog of approximately 70 compositions that were surveyed during the development of the high-power thermal-interface materials portfolio and found to be disqualified — either harmonically unstable (imaginary phonon modes, meaning they would spontaneously distort or decompose at any finite temperature), prone to phase decomposition, or failing multi-criteria kill events in the computational screening pipeline. The catalog spans a wide chemical space: refractory diborides (molybdenum diboride, tungsten diboride, chromium diboride, manganese diboride, scandium diboride), beryllium nitrides, mixed-anion phases including germanium pyrophosphate, osmium dinitride, and calcium tungsten nitride, as well as Heusler intermetallics and fluoride-based ceramic candidates. The strategic value of this catalog is explicitly prosecutorial and defensive rather than commercial in the conventional sense. When an examiner or a validity challenger argues that the claimed filler families were obvious to try — an argument that gains traction precisely because the disclosed materials often look similar in stoichiometry or crystal class to the excluded ones — the existence of this catalog provides a factual foundation for rebuttal. The selection of the disclosed compositions was not arbitrary or intuitive; it emerged from a computation-intensive elimination process that specifically tested and rejected the alternatives. Documenting approximately 70 such rejected alternatives, with four independent machine-learning potentials applied consistently, turns what might otherwise be an attorney's argument into a matter of scientific record. The timing of this catalog is also relevant. Materials patent litigation has become increasingly sophisticated, with defendants and inter partes review petitioners routinely mobilizing computational chemists to argue that prior art structures render claims obvious. A portfolio that can point to its own pre-filing computational screen — showing that the examiner's hypothetical "obvious variants" were in fact tested and found physically untenable — occupies a meaningfully stronger defensive position than one that cannot. This catalog was built with that adversarial dynamic in mind.

The engines did not fully agree here — the asset carries that uncertainty openly rather than overstating confidence.

The catalog documents harmonic instability — the most fundamental and unambiguous form of materials failure — across approximately 70 compositions. In crystalline solids, harmonic stability is assessed through phonon dispersion calculations: if any vibrational mode at any wavevector in the Brillouin zone has a negative (imaginary) frequency, the structure is unstable at the harmonic level, meaning it will spontaneously distort toward a lower-energy configuration and cannot exist as a stable phase at any practical temperature. This is not a marginal or threshold-sensitive result; a composition that fails harmonic stability is physically excluded from the candidate space regardless of what other properties it might exhibit. The computational workflow applied to these excluded compositions used four independent machine-learning interatomic potentials: MACE, CHGNet, MatterSim, and ORB. Each of these potentials was developed independently, trained on partially overlapping but distinct datasets, and implements different architectural choices for representing atomic interactions. Requiring that all four agree on instability is a conservative, high-confidence standard — if even one potential had found a stable basin, the composition would not have been recorded as unambiguously excluded. In practice, across this catalog, all four potentials consistently identified harmonic instability, phase decomposition tendency, or other disqualifying behavior, constituting a robust multi-source consensus. A cross-engine reconciliation matrix was generated to document agreements and any edge cases where potentials required additional reconciliation before a final determination. For certain compositions — specifically the diboride and nitride batch — stability was additionally confirmed through a dedicated batch-level analysis that treated these structurally related compounds as a coherent group, allowing systematic comparison of how minor chemical substitution (Mo vs. W vs. Cr vs. Mn vs. Sc in MeB2 stoichiometries) uniformly failed to yield a stable candidate. The excluded compositions span several distinct chemical failure modes that are worth understanding separately. The refractory diborides (MoB2, WB2, CrB2, MnB2, ScB2) fail primarily on harmonic stability grounds; despite their appealing thermal-conductor-like stoichiometry and the known success of related phases such as hexagonal boron nitride, these specific compositions exhibit soft phonon modes that signal structural instability in the target polymorph. Mixed-anion phases such as GeP2O7, CaWN3, and N2Os fail through a combination of phase decomposition — the total-energy minimum is not the target phase but rather a mixture of simpler binary or ternary compounds — and, in some cases, kill events flagged by the computational pipeline before full phonon calculation. AlBO3 and AlF3, which might superficially appear attractive as aluminum-containing ceramics with potential dielectric compatibility, were found to be four-engine unstable even at converged supercell sizes, meaning the instability is not a convergence artifact. Beryllium nitride (Be3N2) was excluded for combined reasons including inherent toxicity-related regulatory disqualification and computational instability in the phases of interest. Heusler and intermetallic compositions in the ScZn2Cu family were found to decompose preferentially into competing binary phases under the thermodynamic conditions relevant to thermal interface applications. Together, these failure modes illustrate that the chemical space near the disclosed thermal interface material filler compositions is not uniformly hospitable; it contains large regions of structural inviability. The catalog captures this topography of failure with enough resolution and chemical breadth to support a credible scientific narrative that the disclosed materials represent genuine islands of stability in a landscape where instability is common.

This catalog does not address a product market in the conventional sense; it is a prosecution-defense instrument whose value accrues to the owner of the high-power thermal-interface materials portfolio rather than to an end-user customer. Its commercial significance is therefore best understood through the lens of patent value protection rather than product revenue. The thermal interface materials market itself — which is the market the portfolio as a whole addresses — is substantial and growing. Advanced semiconductor packaging, high-density server processors, electric vehicle power electronics, and solid-state power conversion all generate heat fluxes that conventional thermal greases and pads cannot manage reliably over product lifetimes. Estimates of the global thermal interface materials market vary, but the segment relevant to high-performance and industrial applications (where novel filler chemistries are most likely to command premium pricing and where intellectual property ownership is most consequential) is broadly estimated in the range of several hundred million to over one billion dollars annually and growing at rates consistent with the expansion of data center and power electronics deployment. These are estimates rather than audited figures, but they situate the strategic importance of the portfolio's patent protection. Within that context, the value of this negative-control catalog is most cleanly expressed as a multiplier on the defensibility of the portfolio's core claims. A portfolio whose granted claims are successfully invalidated through obviousness arguments forfeits its licensing leverage entirely; a portfolio that can rebut those arguments with systematic computational evidence preserves that leverage. The licensing or sale value of the high-power thermal-interface materials portfolio is therefore partially a function of how well the prosecution record supports the non-obviousness of the claimed filler selections, and this catalog is a direct contribution to that record. Potential licensees or acquirers conducting due diligence on the portfolio will assess not just the scope of the claims but the durability of those claims under adversarial challenge, making this kind of documented prosecution support a tangible component of portfolio valuation.

demonstrates deliberate non-obvious selection (~70 documented exclusions) for prosecution defense

The primary buyers or licensees for this catalog are not acquirers of the catalog in isolation but parties acquiring or licensing the high-power thermal-interface materials portfolio as a whole, for whom this catalog represents an important component of prosecution integrity and validity durability. That population includes specialty materials companies, advanced semiconductor packaging manufacturers, thermal management system integrators, and power electronics component suppliers who would benefit from owning or exclusively licensing the portfolio's claims against growing competitive pressure in the high-power TIM space. For these acquirers, the presence of a systematic, computationally grounded negative-control catalog is a due-diligence positive — it signals that the prosecution record was built with adversarial validity challenges in mind rather than assembled opportunistically. Law firms and patent assertion entities evaluating the portfolio for licensing campaigns would similarly treat this catalog as a risk-reduction asset that increases the expected value of assertion by reducing the probability of a successful obviousness challenge. A secondary buyer category is any party seeking to establish or defend a broad composition patent in the advanced ceramics or refractory materials space more generally, who might look to this catalog as a model for prosecution-support documentation. In that context, the methodological approach — systematic multi-potential computational screening with formal cross-engine reconciliation, applied to a broad chemical space and documented at a granularity sufficient for prosecution use — has value as a template or precedent, independent of the specific compositions cataloged here.

The primary risk specific to this catalog is that its utility is entirely contingent on the validity and breadth of the claims it supports elsewhere in the portfolio. If the core claims in the high-power thermal-interface materials portfolio are narrowed substantially during prosecution or post-grant review — to the point where the non-obviousness of the specific claim scope no longer requires distinguishing the excluded compositions documented here — the catalog's prosecution value diminishes accordingly. It has no standalone commercial value and cannot be monetized independently of the portfolio it supports. A secondary risk is that the field of machine-learning interatomic potentials is advancing rapidly; a validity challenger with sufficient resources could commission new calculations using more recent or higher-accuracy potentials and argue that the instability results documented here are artifacts of the potentials used. This risk is partially mitigated by the four-potential consensus design of the screening workflow and by the fact that the most clearly unstable compositions (those with strongly imaginary phonon modes across all four potentials) would remain unstable under essentially any credible computational method. However, for compositions near the stability boundary — if any in the catalog fall into that category — the risk of a technical dispute is real. The roadmap to managing these risks is straightforward. The catalog should be formally cross-referenced in the prosecution history of the portfolio's patent applications at the earliest practicable opportunity, creating a documented record of its role before any invalidity challenge arises. For any compositions where the instability result is less decisive — closer to the stability boundary — supplementary DFT-level phonon calculations would provide an additional layer of evidence that is substantially harder to challenge than machine-learning potential results alone. Maintaining the cross-engine reconciliation matrix as a living document, updated if new potentials are applied, would also provide a basis for responding to future technical challenges with contemporaneous computational evidence.