Calcium zirconate (CaZrO3) supply-advantaged high-permittivity gate dielectric

Calcium zirconate (CaZrO3) occupies a specific and well-reasoned position in the PFAS-free dielectric and process fluids portfolio: it is the pure-zirconate endmember that closes the hafnate-to-zirconate design-around genus. The strategic logic is supply-chain driven. Hafnium — the incumbent high-k dielectric metal at the leading edge of semiconductor logic — is geochemically tethered to zirconium, extracted almost entirely as a byproduct of zircon sand processing, and is acutely supply-concentrated. Zirconium, by contrast, is extracted directly and at scale from a far broader base of mineral sources. A gate-dielectric material that delivers comparable permittivity using Zr rather than Hf is therefore not just a technical backup; it is a forced-substitution candidate that becomes commercially relevant precisely when supply stress tightens. CaZrO3 crystallizes in the orthorhombic Pnma perovskite structure, an endmember of the alkaline-earth zirconate family, and has been validated through a multi-stage computational protocol that reaches consensus across three independent machine-learning interatomic potentials before any DFT resource is expended. The headline dielectric constant — approximately 48.7 (total permittivity, from a DFPT calculation via the Materials Project) — sits comfortably above HfO2's typical range in its monoclinic or orthorhombic phase, and the bandgap of approximately 3.83 eV sustains the leakage floor required for gate dielectric duty. The asset is honest about its stage: the permittivity figure is currently single-source and the thin-film MOS coupon has not yet been fabricated. What is already secured computationally, however, is rigorous enough to anchor a credible patent claim and to support licensing conversations with MIM capacitor and gate-dielectric integrators. The timing of this filing is deliberate. As foundries push beyond HfO2 and its Hf-Zr solid-solution variants, the design-around space for pure-zirconate perovskites is not yet heavily patented. The portfolio closes that whitespace by asserting use and property claims across the hafnate-to-zirconate claimed family, with CaZrO3 as the highest-fidelity arm validated to date.

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.

CaZrO3 adopts the GdFeO3-type orthorhombic perovskite structure (space group Pnma), the ground-state polymorph stable under ambient conditions for the Ca-Zr-O system. The Ca2+ cation is large enough relative to Zr4+ to tilt the ZrO6 octahedra cooperatively, locking the structure into this distorted variant rather than the ideal cubic perovskite. This tilt pattern is directly relevant to dielectric response: it suppresses the soft phonon modes that would produce a Curie-type ferroelectric anomaly but still permits a meaningfully polarizable Zr-O framework, yielding a total dielectric constant near 48.7. The bandgap of approximately 3.83 eV positions CaZrO3 as an electrical insulator with sufficient conduction-band offset relative to silicon to plausibly serve as a gate oxide, though that interface-offset requirement still awaits explicit interface-MD and band-alignment simulation. Computational validation followed the portfolio's consensus-gating protocol. Three independent machine-learning interatomic potentials — MACE, CHGNet, and MatterSim — each independently relaxed the CaZrO3 structure and then contributed to a joint phonon calculation executed through Phonopy. The critical output is the phonon dispersion and density of states across the full Brillouin zone: all three potentials agree the structure is dynamically stable, with no imaginary phonon modes anywhere in reciprocal space. The lowest-frequency acoustic modes reach approximately 0.53 THz (CHGNet) to 0.71 THz (MACE and MatterSim), establishing a consistent and physically reasonable acoustic floor. Requiring three-way consensus before proceeding is meaningful quality control: a structure that appears stable under one empirically fitted potential but not others is flagged as uncertain, and CaZrO3 passes that filter cleanly. The Materials Project identifier mp-4571 anchors the structural reference. The dielectric tensor was computed via Density Functional Perturbation Theory at the DFPT level, also referenced through the Materials Project. The total permittivity of approximately 48.7 reflects both electronic and ionic contributions, with the z-component contributing approximately 19.1 — indicating a degree of anisotropy that is physically expected given the orthorhombic symmetry. This is a meaningful number for gate-dielectric engineering: it corresponds to an effective oxide thickness (EOT) roughly four to five times thinner than thermal SiO2 for the same physical thickness layer, which is the fundamental driver of interest in high-k dielectrics at advanced nodes. The bandgap of 3.83 eV is sufficient to suppress Fowler-Nordheim tunneling at operating biases typical of logic gate stacks, though the conduction-band offset to Si has not yet been independently simulated. On thermodynamic stability, the structure sits on the convex hull of the Ca-Zr-O system, meaning it is the ground-state phase at that composition and is not expected to decompose spontaneously into competing phases. This is a necessary (though not sufficient) condition for practical deployment: a material that is dynamically stable but thermodynamically metastable would dissolve into competing oxides at processing temperatures. CaZrO3 clears both checks. The open validation gates — an independent DFPT total-permittivity confirmation from a second source, and physical thin-film MOS coupon fabrication — are the honest remaining milestones before the computational case becomes a device-level case.

The addressable market for high-k gate dielectrics and metal-insulator-metal capacitor dielectrics in advanced logic and memory is substantial and growing. Current estimates for the high-k dielectric materials and deposition ecosystem range from roughly $0.5 billion to $1 billion annually, encompassing precursor chemicals, ALD targets, and IP licensing tied to the HfO2/ZrO2 family. The relevant buyers are not bulk commodity customers but rather a small set of highly capital-intensive foundries and IDMs — TSMC, Samsung, Intel, and their advanced-packaging supply chains — plus a broader tier of analog and mixed-signal fabs integrating high-k into MIM capacitors for RF, automotive, and power-management applications. Royalty and licensing logic in this sector follows IP-stacking precedents set by the original HfO2 gate-dielectric patents: a process or device integrator who adopts a new dielectric composition for production cannot avoid IP covering that composition's properties and use in MOS structures. Licensing rates for foundational dielectric compositions have historically settled in the low single-digit percentage of wafer-level value-added, which at advanced-node wafer economics translates to material per-unit IP yield. The CaZrO3 claim is use- and property-anchored rather than bare composition — a structure that is useful specifically because it reaches permittivity near or above 48.7 in a gate-dielectric application — which means the claim tracks the value-creating use rather than blocking the material in all contexts. The supply-chain forcing function is the most concrete near-term demand signal. Hafnium is produced almost entirely as a byproduct of zirconium extraction, at a mass ratio of roughly 1-2% Hf in zircon-derived feedstocks, with primary refining concentrated in a small number of facilities in France, the United States, and China. Any supply disruption — export controls, geopolitical friction, or simple demand surge from leading-edge node ramp — creates immediate incentive for foundries to qualify a non-Hf alternative. CaZrO3 and its zirconate-family peers are the logical landing zone, and establishing IP position now, ahead of that qualification wave, is the timing rationale for this filing.

supply-advantaged Zr endmember closing Hf-to-Zr design-around; highest single-source eps_r (~48.7)

genus-completion zirconate endmembers claimed within the hafnate/zirconate claimed family on use/property/combination, not bare composition novelty

The most direct acquirers or licensees for this asset are semiconductor foundries and IDMs who are actively evaluating Hf-free high-k alternatives for logic or DRAM capacitor applications, particularly those with supply-chain risk management programs explicitly targeting hafnium exposure. TSMC, Samsung Foundry, and Intel Foundry are the obvious tier-one names; below them, specialty analog and RF foundries — GlobalFoundries, Tower Semiconductor, X-Fab — integrate MIM capacitors into their process design kits and have shorter qualification cycles and less institutional inertia around HfO2. Advanced packaging integrators building capacitor-in-substrate or deep-trench capacitor structures for chiplet interconnect are an adjacent and potentially faster-moving market segment. A second natural buyer category is the ALD precursor and deposition-equipment supply chain. Companies like Merck KGaA (EMD Electronics), Entegris, and Air Liquide Electronics who supply organometallic precursors for high-k deposition would have strategic interest in licensing IP that covers novel compositions they intend to support with new precursor products. For that buyer, the IP functions as a freedom-to-operate umbrella enabling them to market CaZrO3 ALD precursors without downstream IP exposure. The asset is also relevant as a defensive holding for any company that expects to be operating in the zirconate-perovskite dielectric space and wants to ensure the genus is either owned or licensed before a supply-chain forcing event accelerates adoption timelines.

The primary technical risk is single-source permittivity. The ~48.7 figure is a single Materials Project DFPT calculation, and if an independent calculation returns a materially different value — either because of pseudopotential choices, k-point convergence, or functional selection — the claim as drafted would need amendment. This is a known and bounded risk: CaZrO3 is a well-studied ceramics compound and independent DFPT calculations are feasible at modest computational cost; running one is the obvious next step and should be completed before any licensing discussion cites the specific permittivity value as a guarantee. The second technical risk is integration unknowns: band offsets to silicon, interface trap density, and thermal stability under CMOS processing temperatures are all uncharacterized, and at least one of these (the conduction-band offset) could disqualify the material for gate-oxide duty regardless of bulk permittivity. On the commercial side, the forced-substitution thesis depends on hafnium supply remaining tight or becoming tighter; if the Hf supply situation eases (new refining capacity, recycling programs, or reduced advanced-node demand), the urgency for a Hf-free alternative diminishes and the asset reverts to a longer-horizon defensive position. The roadmap to de-risk is clear: commission an independent DFPT permittivity calculation, initiate a thin-film deposition and electrical characterization collaboration with an academic or contract fab partner, and complete the band-offset simulation (band-alignment DFT at the CaZrO3/Si interface). Those three steps, none of which requires production-scale resources, would convert this asset from a computationally grounded composition claim into a materially stronger technology demonstration suitable for a tier-one licensing approach.