Isotope-enriched cubic boron arsenide particulate thermal interface material

The material at the center of this asset is isotope-enriched cubic boron arsenide (cBAs, space group F-43m), formulated as boron-11-enriched particles dispersed in a polysiloxane matrix for use as a thermal interface material at the die-to-lid and lid-to-cooler junctions of premium AI accelerators and high-bandwidth memory stacks. The specific claimed form is a bimodal particle distribution — coarse particles (5–50 µm) and fine particles (50 nm–5 µm) combined — loaded at 0.40–0.70 volume fraction, with the boron enriched to greater than 90 at% boron-11 (preferred: at least 97 at%). At 0.60 volume fraction, effective-medium analysis projects composite thermal conductivity of 30–46 W/m·K, anchored to a bulk boron-11 conductivity of 1500 W/m·K (Kim 2025). That is well above commodity alumina and aluminum nitride composites and competitive with diamond-filled premium compounds. The commercial forcing function is straightforward: AI accelerator thermal design power has outrun what commodity thermal interface materials can handle, and the premium-tier parts — think NVIDIA B200/B300, AMD MI400, Samsung HBM4 — represent a price-inelastic segment willing to pay for thermal headroom. Isotope enrichment of cBAs directly addresses the root cause of natural-abundance cBAs's shortfall: the random mix of boron-10 and boron-11 isotopes in unprocessed material scatters phonons, suppressing the intrinsic conductivity that makes cBAs attractive in the first place. This asset claims the specific, package-placed, enriched, particulate embodiment with a near-empty prior-art field — exactly one cBAs hit in a broad patent search — creating an unusually wide runway for durable exclusivity at the premium tier. The high-power thermal-interface materials portfolio of which this is the lead asset spans multiple filler chemistries, but this filing is the strongest commercial entry point within that portfolio.

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

Cubic boron arsenide belongs to the zincblende pnictide family (space group F-43m, bandgap 1.8 eV). Its theoretical thermal conductivity — predicted to rival diamond and cubic boron nitride — arises from a fortunate combination of light atomic masses, strong covalent bonding, and the near-cancellation of three-phonon scattering processes that ordinarily dominate heat transport in semiconductors. The critical practical problem is that natural boron is a 20/80 mixture of boron-10 and boron-11 by atomic fraction, and the mass contrast between those isotopes scatters phonons energetically enough to meaningfully suppress conductivity. Purifying to at least 97 at% boron-11 eliminates that mass-disorder channel and allows the lattice to approach its intrinsic phonon-transport limit, which Kim (2025) anchors at 1500 W/m·K for the bulk crystal. That controlling value underlies all effective-medium projections in this asset. The composite architecture is designed to maximize filler connectivity at practical loadings. A monomodal particle size creates voids that cap achievable volume fraction; the bimodal distribution — coarse 5–50 µm primary particles with fine 50 nm–5 µm secondary particles filling the interstices — allows packing to 0.40–0.70 volume fraction while maintaining matrix processability. Effective-medium modeling using the Agari-Uno framework, with supporting Bruggeman and Hashin-Shtrikman bounds, projects composite thermal conductivity of 30–46 W/m·K at 0.60 volume fraction. It is important to be clear about what that means: the Bruggeman and Hashin-Shtrikman calculations produce theoretical upper bounds of 600–1000 W/m·K, which represent idealized geometric limits, not physically realistic coupon performance. The 30–46 W/m·K Agari-Uno result reflects the percolation factor (0.85–0.98) appropriate for a real dispersion, calibrated against the nearest published measured analog of 21 W/m·K at 40 volume percent (Cui 2021). The gap between the theoretical bounds and the Agari-Uno projection is expected and disclosed explicitly — it is not an inconsistency. Surface passivation of the particles is specified to manage oxide layer formation and filler-matrix interface thermal resistance, both of which would otherwise erode composite conductivity at high loading. An isotope mass-substitution simulation confirms a 0.91% shift in optical phonon frequency relative to natural-abundance cBAs, providing a spectroscopic corroboration of the enrichment physics that is consistent with the expected conductivity lift.

The addressable market is estimated at $5 billion and growing, but the relevant slice for this asset is narrower and more defensible than the full thermal interface materials market. The target is the premium tier of AI accelerator and high-bandwidth memory packaging — specifically the TIM-1 (die-to-integrated heat spreader), TIM-1.5, and TIM-2 (heat spreader-to-cooler) junctions on parts where thermal design power is the binding constraint and the bill of materials can absorb a premium filler. NVIDIA's B200 and B300 series, AMD's MI400, and Samsung's HBM4 stack are the named reference customers. These are exactly the parts whose per-unit value is high enough that a thermally superior TIM is an engineering necessity rather than a cost trade-off. The royalty and licensing logic follows from that premium positioning. A value-based per-package royalty — tied to the thermal headroom delivered by the enriched filler relative to a natural-abundance or nitride baseline — is the natural structure, potentially combined with a supply-linked component given the specialized boron-11 feedstock requirement. Because isotope-enriched cBAs would be used selectively at high-flux zones rather than across the entire package, volumes are smaller than commodity TIM, but per-unit economics are correspondingly richer. That geometry favors a higher royalty rate on a qualified, premium base rather than a high-volume, thin-margin commodity program. The near-clean prior-art position — one cBAs hit in a broad search — supports strong exclusivity claims, which in turn support pricing power in any licensing negotiation. The supply-chain constraint on boron-11 feedstock (isotope separation via electromagnetic, thermal diffusion, centrifuge, or chemical-exchange processes) creates an additional structural barrier to competition that reinforces the IP position rather than undercutting it.

higher composite k_eff than natural-abundance cBAs and far above commodity nitride/oxide composites; near-clean prior-art (breadth scan: cBAs=1 hit)

four-distinguisher: (i) particulate bimodal form (not bulk crystal); (ii) >=97 at% 11B enrichment; (iii) thermal-interface use class (not power device/crystal growth); (iv) realistic composite k_eff 30-46 W/m/K Agari-Uno

The most strategically aligned acquirers and licensees are the premium accelerator and HBM packaging strategics whose parts are already named as target customers: NVIDIA (B200/B300 premium tier), AMD, and Samsung. Each of these organizations faces thermal design power constraints that are tightening with each hardware generation, and each has the purchasing power to absorb a premium TIM at the die-to-lid junction where the thermal conductivity gain is most valuable. For these buyers, a field-of-use license restricted to premium-tier accelerator and HBM packaging is the natural structure, potentially combined with a supply agreement for boron-11-enriched cBAs feedstock given the isotope-separation supply chain's specialization. The supply-chain component is not incidental — it creates a recurring, supply-linked revenue stream that complements the IP license. A premium thermal interface materials supplier or specialty materials house with existing isotope-separation capability represents a different but equally compelling buyer profile — one that could own both the IP position and the constrained supply chain, an unusually defensible vertical combination. Given the near-clean prior-art position and the feedstock chokepoint, an exclusive or semi-exclusive license commands substantially better terms than a non-exclusive program would. Buyers evaluating license-versus-acquire should factor that acquisition also captures the defensive value of the four-distinguisher carve-out and the optionality across TIM-1, TIM-1.5, and TIM-2 placements — a broader product coverage than a single field-of-use license would typically convey.

The performance risk is the most immediate. The headline 30–46 W/m·K is an Agari-Uno effective-medium projection, and the nearest measured physical analog — a cBAs composite at 40 volume percent — was measured at 21 W/m·K (Cui 2021), roughly half the modeled projection. The percolation factor applied (0.85–0.98) is an extrapolation at high bimodal loading, not a calibrated experimental value at the target loading. The Bruggeman and Hashin-Shtrikman upper bounds (600–1000 W/m·K) are geometric limits, not achievable coupon targets. Real composite thermal conductivity could land below the modeled band, and that gap is the primary financial risk for a buyer pricing the asset before the coupon measurement is complete. The path to resolving this risk is a single experiment: a measured composite coupon at production-representative bimodal loading, which is the stated next validation action. The cross-potential disagreement on phonon stability — MACE reporting dynamic stability, a universal potential reporting three imaginary modes — is a secondary risk. For a structurally known material with existing synthesis reports, this is less alarming than it would be for a novel theoretical prediction, but it does mean that the computational stability validation is not fully converged, and additional DFT phonon calculations on the boron-11-enriched structure would strengthen the position. The boron-11 isotope-separation supply chain is a cost and concentration risk: the specialized production infrastructure required (electromagnetic separation, centrifuge, or chemical exchange) is available at few facilities globally, which could constrain production scaling and expose the commercial program to feedstock pricing volatility. A buyer should assess supply-chain redundancy before committing to a program that depends on a single-source material. Making any license or acquisition terms contingent on the coupon measurement result is the most direct de-risking mechanism available at this stage.