Magnesium aluminoborate sinhalite ceramic for millimeter-wave and RF dielectric components

MgAlBO4 — sinhalite — occupies a narrow but strategically valuable position in the dielectric, ferroelectric, and wide-bandgap oxides portfolio as the aluminoborate backup arm of a broader borate sub-ladder. Its role is honest: it is not a lead composition, but a property-pending backup that hedges the family claim by extending coverage to the aluminoborate chemical space. What makes this backup arm genuinely interesting is that sinhalite is not a hypothetical compound. It is a known ambient-pressure-stable olivine-type mineral (orthorhombic, space group Pbnm) present in the Materials Project database (mp-8376) and confirmed to sit on the convex hull, meaning it is thermodynamically stable at zero external perturbation without any entropic stabilization argument required. That ground-state stability is a non-trivial starting point for a ceramic candidate, and it sharply differentiates sinhalite from many computationally proposed dielectrics that require heroic synthesis conditions or remain metastable against decomposition. The commercial logic rests on a forced-substitution pressure building in 5G and emerging 6G millimeter-wave filter hardware. FR2-band (24–52 GHz) and sub-THz filter platforms demand ceramic dielectric bodies with permittivity in the 8–12 range, quality factor-bandwidth products (Q times frequency, Q*f) above roughly 40,000 GHz, and thermal-expansion coefficients matched to alumina or standard low-temperature co-fired ceramic (LTCC) substrates. The borate mineral family — represented by kotoite (Mg3B2O6, Q*f ~230,900 GHz) and forsterite (Mg2SiO4, Q*f ~240,000 GHz) as anchor references — is one of the very few chemistries that delivers Q*f in that range at sub-GHz frequencies, and the literature record on ternary aluminoborate variants is essentially blank for device-grade dielectric applications. Sinhalite, sitting within that same mineral family, inherits the physical motivation. The DFPT-computed static permittivity of approximately 8.47 and CTE near 7.8 ppm/K at 300 K are squarely in the alumina-compatible window. Whether the material can clear the Q*f hurdle remains an open experimental question, and the dossier is structured to be killed cleanly if it cannot. The timing argument is real: the FR2 filter supply chain is currently dominated by a small number of high-purity oxide ceramics, and device integrators are actively seeking co-fireable alternatives that can be processed below the silver melting point (961 °C) to reduce manufacturing cost. Sinhalite crystallizes from MgO-Al2O3-B2O3 glass in the 800–900 °C range, making silver co-firing technically feasible without exotic reducing atmospheres. Combined with a completely clear freedom-to-operate landscape for dielectric device applications of this specific composition, the asset presents a low-encumbrance path to a narrow but defensible position in the RF ceramics market.

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.

MgAlBO4 adopts the olivine-type crystal structure (Pbnm, space group 62, orthorhombic), the same structural family as forsterite and the kotoite-adjacent borate minerals. In the sinhalite ordering, magnesium and aluminum occupy the octahedral M1 and M2 sites in an ordered arrangement, with boron in tetrahedral coordination. This ordered cation distribution is consequential: it eliminates the cation-disorder scattering channels that typically suppress the quality factor in mixed-site ceramics, which is the structural reason the olivine-type borates consistently deliver exceptional Q*f values. The compound is not a computational novelty — it is a naturally occurring mineral with a well-characterized crystallography, and its placement on the convex hull in the Materials Project (EAH approximately 0, mp-8376) means no decomposition driving force exists thermodynamically at ambient conditions. The computational validation program applied to this composition is multi-layer. At the dynamical-stability tier, four independent machine-learning interatomic potentials were run: MACE, CHGNet, and MatterSim each returned phonon dispersions with zero imaginary modes and a positive minimum vibrational frequency across the Brillouin zone, confirming dynamic stability by three independent models. The fourth potential (AlignN universal) returned a softer result that does not reach a stability verdict at the same confidence level, yielding an aggregate three-of-four consensus. This is reported candidly rather than rounded up to full consensus: three potentials agreeing on stability is meaningful but not the four-of-four agreement the portfolio's lead compositions achieve. No DFT phonon calculation has been performed as an independent verification of this consensus, and that remains an open computational step for de-risking before conversion. The dielectric-tensor calculation is more mature. A DFPT static permittivity was computed against the mart.mart_dielectric_static_truth reference set, yielding a static dielectric constant (eps0) of approximately 8.47, decomposed into an ionic contribution of 5.519 and an electronic contribution of 2.953. This ionic-to-electronic ratio is physically consistent with a wide-bandgap oxide ceramic: a large bandgap suppresses electronic polarizability while the relatively soft Mg-O-B sublattice contributes the ionic response. The PBE band gap is computed at approximately 6.0 eV, which is a lower bound (PBE systematically underestimates gaps); the true gap is expected to be higher, reinforcing the classification as a wide-bandgap dielectric. Critically, a quasi-harmonic approximation (QHA) linear thermal-expansion-coefficient calculation was performed and validated against two reference minerals — beta-eucryptite and kotoite — yielding a CTE of approximately 7.819 ppm/K at 300 K. Standard alumina substrates run at 6–8 ppm/K and conventional LTCC tape systems at 5–8 ppm/K, placing sinhalite squarely within the co-sintering compatibility window for both substrate classes. What remains uncomputed and unmeasured is equally important to state. The temperature coefficient of resonant frequency (tau_f) has not been calculated or measured. The Q*f product has not been computed and cannot be reliably extracted from DFPT alone — it requires either a finite-temperature phonon-phonon scattering calculation (a computationally expensive anharmonic treatment) or direct microwave resonator characterization. The dossier specifies a Hakki-Coleman dielectric resonator measurement on a dense, near-stoichiometric sintered MgAlBO4 body as the controlling conversion gate. A Q*f below 40,000 GHz is defined as a kill criterion. The synthesis path identified for producing that test body — crystallization from a MgO-Al2O3-B2O3 glass or direct sintering of Alborex-class aluminum-borate whisker feedstock — is commercially available and does not require custom precursor chemistry, which keeps the experimental gate low-cost relative to its information value.

The addressable market for sinhalite as a dielectric ceramic is bounded by the RF filter and LTCC dielectric component segment of the broader microwave ceramics industry. This is not a commodity market: high-performance dielectric resonator materials for base-station and satellite front-end filters at millimeter-wave frequencies are a specialty segment where price per kilogram can reach several hundred dollars for qualified, characterized bodies, and where the qualification cycle — not raw material cost — is the primary barrier to entry. The RF filter market for 5G infrastructure is currently estimated in the several-billion-dollar range globally, with the millimeter-wave sub-segment at roughly $0.5–2 billion and growing as carrier deployments shift toward higher-frequency bands where smaller, higher-Q resonators are required. 6G planning documents consistently identify sub-THz frequencies (above 100 GHz) as target bands, creating a medium-term pull for dielectric materials with even tighter dimensional tolerances and lower loss than current FR2 ceramics can provide. The buyers in this segment are ceramic component houses and RF module integrators, not end-device OEMs. CoorsTek, Morgan Advanced Materials, and the Murata/TDK-class RF filter manufacturers represent the relevant customer tier — companies that qualify new ceramic bodies through their own internal characterization pipelines and then supply finished filter components to infrastructure OEMs (Ericsson, Nokia, Huawei, Samsung Networks). For these buyers, a new dielectric body must clear internal Q*f and tau_f specifications before commercial relevance, but a composition with a clear FTO landscape, commercially available feedstock, and computable starting-point properties substantially de-risks their own qualification investment. Licensing or co-development agreements, rather than direct ceramic manufacturing, represent the most likely commercial structure for a composition at this stage. The royalty logic for a device-use patent covering a co-fireable sinhalite RF ceramic body would be structured per unit volume of qualified ceramic body or per completed filter assembly, depending on where in the supply chain the license is taken. A conservative royalty rate of 1–3% applied against a specialty ceramic body selling at $100–$500 per kilogram, with addressable volume in the several-hundred-ton range for a niche but qualified RF dielectric, implies a royalty stream in the low-to-mid millions of dollars per year at market penetration. This is a niche-market asset, not a platform play — the investment thesis rests on the combination of clean FTO, low experimental gate cost, and the structural analogy to a mineral family with demonstrated world-class Q*f, not on total-market scale.

known ambient-pressure-stable mineral (sinhalite, mp-8376, on-hull) with DFPT eps0~8.47, CTE matched to alumina/LTCC, Ag co-fireable from commercial Alborex whisker feedstock, and ZERO use-art / zero patents on its dielectric properties (FTO-clean mm-wave whitespace); in a property family with demonstrated Q*f>200k GHz (kotoite/forsterite)

recited as STABLE_3_OF_4, DFPT-eps-computed sinhalite mm-wave/RF dielectric; zero use-art and zero patents on the dielectric properties of MgAlBO4; controlling §1.56 reference is the 2025 multiphase off-stoichiometry low-frequency glass-ceramic (10.1007/s00339-025-09086-6) that claims nothing; laser-host/activator-doped luminophore embodiments excluded per §37.1

The most natural acquirers or licensees for this asset are ceramic component manufacturers already qualified in the microwave dielectric space who are seeking compositional IP coverage ahead of FR2 and sub-THz filter platform build-outs. CoorsTek and Morgan Advanced Materials operate sintering and characterization infrastructure that could run the Hakki-Coleman gate experiment internally within a standard qualification timeline, making them natural co-development or option-license partners at the current stage rather than outright acquirers. The Murata and TDK class of RF filter houses represent a second buyer tier: their interest would be triggered primarily by successful Q*f validation, at which point a license covering the composition-plus-device-use claim in a qualified LTCC tape or co-fired body format carries direct product-line relevance. A defensive buyer — a vertically integrated infrastructure OEM such as Ericsson or Nokia Networks — might acquire the asset as FTO insurance against third-party licensing assertions in the borate ceramic space as FR2 deployments scale, even without an intention to manufacture sinhalite ceramics directly. The asset's relatively low experimental gate cost (commercially available feedstock, standard resonator characterization, no exotic synthesis) makes it accessible to a strategic partner willing to fund the measurement program in exchange for a co-exclusive or field-of-use license.

The primary technical risk is that sinhalite's Q*f on a dense sintered body falls below the 40,000 GHz kill criterion. The olivine-type crystal structure is a favorable structural analogy, but chemistry matters as much as structure: the incorporation of aluminum into the cation sites, relative to pure magnesium borates, introduces the possibility of point-defect scattering and slight structural disorder at grain boundaries that could suppress the quality factor even if the bulk phonon structure is clean. This risk is binary and resolvable with a single Hakki-Coleman experiment — the asset is deliberately structured to be killed cheaply rather than kept alive on structural analogy alone. A secondary technical risk is tau_f: even if Q*f clears the threshold, an unconstrained temperature coefficient outside the ±10 ppm/K window makes the material commercially unusable in most filter applications without a composite formulation, adding a second experimental phase and diluting the composition claim. The prosecution risk is modest but real: sinhalite is a well-known mineral name, and a vigorous examiner may raise enablement objections to a device-use claim where the Q*f and tau_f are uncharacterized at filing. The prosecution architecture addresses this by positioning the Hakki-Coleman data as amendment support at conversion, but there is a window between provisional filing and conversion where a competitor could file a competing claim with measured data. The mitigation is to run the Hakki-Coleman experiment early and use the data to support a rapid conversion with measurement-anchored dependent claims, closing that window before it becomes a material risk.