Erbium oxyorthosilicate radiation-hardened interlayer dielectric for space and defense ICs

Radiation-hardened interlayer dielectrics represent one of the most persistently underserved material slots in advanced IC design for space and defense. Modern rad-hard processes use SiO2, Si3N4, or Al2O3 as their ILD and passivation stacks — materials whose total ionizing dose (TID) behavior has been characterized for decades but whose fundamental radiation-trapping mechanisms at the silicon interface are essentially baked in. No oxide in that incumbent trio achieves TID retention at the level demanded by next-generation missions: multi-year deep-space transit, high-earth orbit, and proximity to nuclear environments where dose levels reach or exceed 10^5 rad(Si). There is, consequently, a real and growing tension between mission requirements and what the ILD can offer. Er2SiO5 — erbium oxyorthosilicate, P2_1/c monoclinic — enters that gap with a computed bandgap of 4.83 eV and a TID retention benchmark of at least 95% at 10^5 rad(Si). That performance is framed as approximately four times better than Lu2SiO5 in non-scintillator ILD configurations. The four-times figure is the operative competitive differentiator: Lu2SiO5 has attracted attention from the rare-earth silicate research community primarily through its scintillator heritage, but its non-scintillator ILD radiation-hardness ceiling is substantially lower than Er2SiO5. The broader rare-earth oxyorthosilicate family — La2SiO5, Sm2SiO5, Gd2SiO5, Tm2Si2O7, Yb2SiO5 — rounds out the dependent scope, while Lu2SiO5 and Gd2SiO5 carry explicit scintillator disclaimers that preserve clearance from dense prior art. The timing matters. Patent-whitespace analysis across more than 300,000 materials-relevant filings shows the rad-hard RE-silicate ILD lane is open as of 2025-2026, with no in-force patent addressing this device-use of rare-earth oxyorthosilicates outside scintillator applications. That whitespace is perishable: as mission profiles expand and rare-earth oxide thin-film deposition matures, the window for a clean composition-plus-device-use filing will narrow. Lattice Graph's computational pipeline has validated the anchor composition ahead of that closure.

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.

Er2SiO5 belongs to the rare-earth oxyorthosilicate family, crystallizing in the monoclinic P2_1/c space group. The oxyorthosilicate framework is structurally distinct from the more common pyrosilicate (RE2Si2O7) arrangement: isolated SiO4 tetrahedra are charge-balanced by rare-earth coordination polyhedra, producing a dense, relatively open-shell electronic structure. For dielectric applications, the critical consequence is a wide bandgap — 4.83 eV computed for Er2SiO5 — which places it comfortably above the threshold where hot-carrier injection into defect traps is kinetically suppressed. High-bandgap oxides in general create a larger energy barrier for the electron-hole pairs generated by ionizing radiation, reducing the density of trapped charge that accumulates at the dielectric-semiconductor interface over a device lifetime. The combination of a rigid silicate lattice, low oxygen-vacancy formation energy, and the 4f electronic shielding of the Er3+ centers distinguishes this class from simpler binary oxides like Al2O3. The computational validation of Er2SiO5 used two independent machine-learning interatomic potentials, and both returned dynamically stable results — meaning no imaginary phonon modes were identified in the phonon dispersion, confirming the structure is a true local minimum on the energy landscape rather than a saddle point. This two-potential consensus is the internal threshold for the controlled anchor composition. For context on how this compares within the broader family: the previous lead, Y2SiO5, was put through a more demanding four-potential adjudication (MACE, CHGNet, MatterSim, and ORB) and only one of the four potentials agreed it was stable — a 1-of-4 outcome that fell short of the consensus threshold. Y2SiO5 was accordingly demoted from anchor to a candor-flagged dependent composition, with an open validation gate for DFPT Gamma-point adjudication. Sm2SiO5 and Gd2SiO5 were also evaluated during structure relaxation; their positive lowest phonon branch values (+0.258 THz and +0.214 THz, respectively) indicate marginal stability without imaginary modes, qualifying them as dependent claims but not as anchors. Two independent DFT reference calculations support the Er2SiO5 structural and electronic property data. The radiation-hardness mechanism is grounded in the high bandgap and the crystallographic regularity of the P2_1/c structure. TID-induced degradation in conventional ILDs proceeds through the buildup of interface trap density (Dit) and oxide trap charge (Qot) that shift threshold voltages and degrade carrier mobility. In a high-bandgap, high-density oxyorthosilicate matrix, the recombination rate of radiation-generated electron-hole pairs is faster, the lateral diffusion of trapped holes toward the Si interface is geometrically constrained, and the energy depth of trapping sites is larger — all of which reduce the net oxide-trapped charge density surviving at dose levels in the 10^5 rad(Si) range. The target specification of at least 95% TID retention at that dose, with a claimed factor-of-four advantage over Lu2SiO5 in ILD use, is what positions this material for high-reliability applications rather than laboratory curiosity. The principal open validation gate is a physical Co-60 dose coupon irradiation experiment. Computational phonon stability and electronic structure are necessary but not sufficient for a TID retention claim: the 95%-at-10^5-rad figure requires confirmation against actual ionizing radiation under device-realistic conditions (substrate type, bias state, deposition method). Y2SiO5 has its own separate gate — a DFPT Gamma-point adjudication to resolve the phonon ambiguity before it could be elevated back toward anchor status. These are honest open items. The computational work de-risks the crystal structure and the electronic property baseline; the irradiation coupon is the straightforward next step for any development partner or licensee.

The addressable market for radiation-hardened ILD materials is estimated in the range of $1-2 billion, spanning custom rad-hard IC foundries, defense electronics primes, and the expanding commercial space segment. The customer set is well-defined: U.S. government and allied space agencies (including NASA programs structured around SBIR Z2 funding mechanisms), defense prime contractors sourcing rad-hard ASICs for satellites, missile guidance, and tactical electronics, and increasingly commercial new-space operators who are building medium-earth and geostationary platforms that must survive multi-year dose accumulation. This is not a consumer market — unit prices are high, qualification cycles are long, and the technical bar to substitute an ILD in a rad-hard process is significant. That means established relationships and demonstration data carry enormous weight, which is both a barrier to entry and a moat for anyone who clears it first. Royalty and licensing logic in this space tracks the qualified-process model. A foundry that qualifies a new ILD material into a radiation-hardened CMOS process node is making a multi-year commitment: the qualification runs, the reliability data, and the customer design-ins are all tied to that specific material stack. A composition-plus-device-use patent that covers Er2SiO5 (and the broader RE2SiO5 family) in a rad-hard ILD configuration creates a licensing lever that is most naturally exercised at the foundry or process-IP level, with royalties captured per wafer-start or per qualified process node. The more common alternative — an outright acquisition of the patent family by a defense electronics prime or a specialty fab — is also plausible given the strategic nature of the supply chain. At $1-2B in addressable market, even a modest royalty rate on rad-hard IC wafer revenue would represent a meaningful return on a patent position that currently has no competition in-force.

>=4x Lu2SiO5 TID retention in non-scintillator configurations

rad-hard ILD device-use of RE2SiO5; scintillator use disclaimed

The most strategically aligned acquirers or licensees are rad-hard IC foundries and defense electronics primes who are already qualified for military and space programs and who need to extend their process nodes' TID ceilings without re-qualifying a fundamentally different device architecture. In the U.S., this means companies operating within the DMEA-accredited trusted foundry framework, specialty fabs serving the high-reliability IC market, and the aerospace and defense divisions of large system integrators who control their own ASIC design and sourcing. NASA SBIR Z2 program structures are specifically relevant: that mechanism funds early-stage material-to-process integration work that could carry the Co-60 coupon gate as a funded milestone. International equivalents in allied space programs (ESA, JAXA, and allied defense ministries) represent secondary licensing channels where U.S. export control considerations would apply but where demand for TID-hardened ILD materials follows similar mission profiles. Rare-earth materials and thin-film deposition companies with existing erbium or rare-earth silicate synthesis capability represent a second buyer category — less likely to acquire the patent outright but natural candidates for a co-development or supply-chain licensing arrangement in which they provide qualified Er2SiO5 precursor or deposited film, and the patent family provides their downstream IC-fab customers with freedom-to-use coverage. The backup asset classification reflects that this is not the single flagship position of the broader dielectric, ferroelectric, and wide-bandgap oxides portfolio, but in the specific rad-hard ILD niche it is the most advanced composition-plus-device-use position that currently exists with clean FTO, which is a meaningful distinction for any buyer assessing strategic value in space and defense electronics supply chains.

The most immediate technical risk is the gap between computational stability and physical radiation-hardness performance. Two-potential phonon stability is a necessary precondition for material viability but does not directly measure TID retention. If the Co-60 coupon experiment shows that as-deposited Er2SiO5 films accumulate interface traps at rates closer to Al2O3 than to the claimed target, the commercial thesis weakens substantially. This risk is mitigated by the high bandgap (4.83 eV is a strong a priori indicator) and by the structural arguments about RE oxyorthosilicate trap dynamics, but it has not yet been experimentally closed. The Y2SiO5 demotion also serves as a reminder that computational predictions within the RE silicate family are not uniformly reliable — the four-potential disagreement on Y2SiO5 is a calibration point that the broader family evaluation should take seriously. The prosecution and commercialization risk is the race window. The 2025-2026 whitespace is real but not permanent: as thin-film deposition of rare-earth oxides becomes more routine in advanced CMOS processes and as mission-driven demand for better ILD radiation hardness becomes more visible, other filers will enter the space. The scintillator-use disclaimers protect clearance from dense existing art but do not prevent future competitors from filing ILD-specific claims on other RE silicate compositions. The de-risking roadmap is clear and bounded: execute the Co-60 coupon experiment, complete the Y2SiO5 DFPT adjudication, and advance the filing before the window closes. For a buyer, the cost of those two experiments is modest relative to the value of a clean first-mover patent position in a market where ILD qualification is a multi-year process that rewards whoever gets there first.