Quaternary lithium calcium aluminum fluoride dielectric for superconducting-qubit junction passivation

LiCaAlF6 is the standout member of the colquiriite LiM''AlF6 sub-family within the metal-fluoride qubit dielectric materials portfolio. It combines a bandgap near the top of the enumerated set — 8.09 eV — with a zone-center phonon among the smallest negative values of the leads, -0.06 cm⁻¹. Those two properties together define a clear specialization: junction-adjacent passivation at Josephson junctions, where insulating margin and suppression of two-level-system (TLS) loss channels matter most. No other fluoride candidate in the portfolio occupies this position as confidently. The timing is driven by the DARPA Quantum Benchmarking Initiative 2025-2026 downselect, which is forcing qubit hardware developers to commit to dielectric and passivation strategies. The incumbent — amorphous aluminum oxide grown on the junction — is now the consensus loss bottleneck in state-of-the-art transmon qubits. A crystalline fluoride with an 8.09 eV gap and a near-zero soft mode is a structurally principled alternative, not an incremental reformulation. The quaternary composition (four distinct elements, P-31c colquiriite structure, space group 163) also creates meaningful patent whitespace relative to simple binary fluoride optics, giving the IP a durability that single-element or binary candidates cannot match.

LiCaAlF6 adopts the colquiriite crystal structure, space group P-31c (No. 163), a well-characterized quaternary fluoride framework that distributes Li, Ca, Al, and F across distinct crystallographic sites. The JARVIS database entry JVASP-21143 provides the DFT foundation: density-functional perturbation theory (DFPT) gives a high-frequency dielectric constant (epsilon-infinity) of 2.050, and a computed bandgap of 8.09 eV — among the widest of all fluoride dielectric candidates evaluated in this screening. The bulk modulus of 72.4 GPa reflects adequate mechanical robustness for thin-film deposition processes such as physical vapor deposition. The zone-center phonon of -0.06 cm⁻¹ is the critical dynamical-stability figure. A value this close to zero is a small negative mode, likely within numerical noise and comparable to its colquiriite siblings, meaning the structure sits at the edge of harmonic dynamical stability with little computed soft-mode tendency. This matters for the intended application because soft phonon modes couple directly to two-level-system (TLS) loss mechanisms; suppressing them is a prerequisite for coherence preservation at millikelvin operating temperatures. The 8.09 eV gap simultaneously maximizes the insulating margin across the junction, suppressing quasiparticle injection and defect-state formation at the dielectric-superconductor interface. The epsilon-infinity of 2.050 places LiCaAlF6 mid-range within the portfolio. It is slightly higher than the lowest-epsilon member (cryolite), which creates a modest participation-ratio penalty for bulk capacitor dielectric applications — but for junction-adjacent passivation, where the dielectric volume is small and the gap and stability properties dominate the loss budget, this tradeoff favors LiCaAlF6. The colquiriite sub-family covers three M-site variants: M = Ca, Sr, and Yb, providing a structurally coherent group of tunable candidates within one crystal type.

The addressable market is the passivation and dielectric layer segment of superconducting quantum computing hardware. We estimate this at $1-2 billion, reflecting the premium that qubit manufacturers can realistically attach to coherence-critical materials as qubit counts scale toward fault-tolerant thresholds. The buyers in this market are the handful of organizations building superconducting quantum processors at scale: IBM Quantum, Google Quantum AI, and AWS's Center for Quantum Computing are the named targets, each running active fab programs where junction passivation is a known performance constraint. The commercial logic for LiCaAlF6 specifically attaches to the junction rather than the bulk dielectric. Josephson junctions are the coherence-limiting element in transmon qubits, and the dielectric immediately adjacent to the tunnel barrier — not the shunt capacitor — drives the dominant TLS loss in most state-of-the-art devices. A passivation material positioned precisely here commands a higher per-junction royalty than a bulk capacitor fill, and the qubit node count (rather than chip area) becomes the natural royalty metric. As qubit counts grow from hundreds to thousands, that per-node royalty compounds. The colquiriite sub-family's three M-site variants (Ca, Sr, Yb) support a sub-family field-of-use license rather than a single-composition deal. A licensee gains the ability to tune the M-site for process compatibility — Ca for standard deposition, Sr or Yb for adjusted lattice matching — without stepping outside the licensed scope. That flexibility makes a sub-family license more attractive than a single-member license to a hardware team that needs room to optimize deposition processes across foundry runs. The DARPA QBI 2025-2026 downselect creates an external forcing function: programs that have not committed to a passivation strategy by then risk being locked out of the benchmark timeline.

widest gap (8.09 eV) + highest phonon-stability margin (-0.06 cm^-1) of enumerated members

colquiriite LiM''AlF6 (M''=Ca,Sr,Yb) SG163 in qubit-dielectric use; computationally-predicted patent_count=0; quaternary complexity supports whitespace

Preliminary screen only. This is not a freedom-to-operate opinion.

The natural acquirers and licensees are the organizations running the largest superconducting qubit programs: IBM Quantum, Google Quantum AI, and AWS's Center for Quantum Computing. In our view, all three face pressure from the DARPA QBI 2025-2026 downselect to demonstrate coherence improvements, and junction passivation is a plausible performance lever for each; this is our reading, not their stated priority. LiCaAlF6's widest-gap and highest-stability-margin profile is most attractive to a team prioritizing junction-adjacent coherence over bulk capacitor optimization, which describes the current engineering priority in all three programs as they push toward error-correction thresholds. The deal structure that best matches this asset is a sub-family field-of-use license covering the three colquiriite variants, giving the licensee M-site tuning flexibility within a single licensed scope. A hardware team operating a fab would likely prefer this over a single-composition license, because process optimization across foundry runs may favor one M-site variant over another. For a buyer seeking to own junction-passivation IP as a competitive moat, an exclusive acquisition of the sub-family is the alternative. The natural entry point is a paid option funding the junction-adjacent passivation coupon experiment and T1 measurement, converting to a license on successful demonstration — this sequences the financial commitment to coincide with the first definitive evidence of coherence improvement.

The primary risk is the gap between computed properties and measured device performance. The bandgap, phonon stability, and dielectric constant are all DFT-computed values from a single database source; no thin-film deposition, loss tangent measurement, or qubit T1 measurement has yet been performed. The path from a stable computed structure to a low-loss junction passivation layer involves process development (target selection, deposition rate, substrate compatibility, interface chemistry) that may surface challenges not visible in the DFT data. Until the junction-adjacent passivation coupon experiment is completed, the stability and gap advantages remain projected. The second risk is novelty at the composition level. LiCaAlF6 is a known laser-host crystal, and a validity challenge to the composition claims is possible if an examiner or challenger argues that the qubit-dielectric use does not constitute a patentable distinction from the prior art. The mitigation is the quaternary complexity — the four-element colquiriite structure is meaningfully differentiated from the binary fluoride art that dominates the optics literature — and the specificity of the device context (millikelvin, superconducting, junction-adjacent). A third, lower-probability risk is that the mid-range epsilon-infinity of 2.050 proves penalizing for participation ratio in configurations where the colquiriite layer extends beyond the junction perimeter into capacitor regions; this can be managed by geometric design of the passivation layer, but it limits the use case to junction passivation rather than full-chip dielectric replacement.