Sodium aluminoborate (Na2Al2B2O7) glass for EUV pellicle membranes

Na2Al2B2O7 — sodium aluminoborate — is an alkali aluminoborate glass/glass-ceramic composition claimed in a highly specific embodiment: a thin film under 10 microns thick, with a target range of 30–70 nm, for use as an EUV pellicle membrane or radiation-hardened optical substrate. The core commercial proposition is that silicon and silicon-nitride pellicles, the current industry standard, carry intrinsic absorption penalties and thermal management challenges at 13.5 nm extreme-ultraviolet wavelengths. An oxide glass system built around light elements — boron, aluminum, sodium, oxygen — offers a credible alternative pathway to high EUV transmission through careful compositional tuning rather than the crystalline-perfection demands of silicon membranes. This asset sits within the PFAS-free dielectric and process fluids portfolio and occupies a narrow but genuinely defensible position: the claim is deliberately scoped to the thin-film device embodiment rather than the bulk composition, because the bulk crystalline phase was found, through rigorous multi-method simulation, to be dynamically unstable. That candor is a feature of the filing strategy, not a weakness — it makes the granted or pending claim far less vulnerable to prior-art challenge than an overbroad bulk-composition claim would be, while still covering the commercially relevant product form. The EUV pellicle market is small by semiconductor standards (estimated $200–500 million addressable range), but strategic value to ASML-ecosystem participants, leading logic foundries, and radiation-hardening specialists is disproportionate to its dollar volume, because pellicle performance directly gates lithography yield at the most advanced process nodes. The timing context is important. EUV adoption is accelerating across logic and DRAM manufacturers. High-NA EUV systems from ASML place even more demanding power loads on pellicle membranes, raising the thermal and radiation-tolerance bar beyond what current Si/SiNx pellicles were designed to handle. An early filing on a non-silicon oxide-glass thin-film embodiment with a credible EUV-transmission prediction creates a licensing or co-development anchor at precisely the moment when pellicle materials diversification is commercially motivated.

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.

Na2Al2B2O7 is a ternary oxide glass in the alkali aluminoborate family, combining network-forming boron-oxygen and aluminum-oxygen units with sodium as a network modifier. The EUV-transparency argument rests on atomic composition: boron, oxygen, sodium, and aluminum all have relatively low atomic numbers and correspondingly low photoabsorption cross-sections at extreme-ultraviolet wavelengths near 13.5 nm, the primary emission line of tin-plasma EUV sources. Silicon, by comparison, has a non-trivial absorption edge in this spectral region, requiring extraordinary purity and film thinness to achieve acceptable pellicle transmission. A boron-rich oxide glass system trades crystalline perfection for compositional flexibility, and at film thicknesses of 30–70 nm the bulk absorption is predicted to allow greater than 80% transmission at 13.5 nm — the threshold generally cited for viable EUV pellicle performance. The computational validation trajectory for this compound is instructive about both its promise and its current limitations. Three independent machine-learning interatomic potentials — representing distinct model architectures trained on distinct ab-initio datasets — were all applied to relax the Na2Al2B2O7 structure, and all three reached consistent relaxed geometries, which is a meaningful finding. Majority agreement across three ML potentials on the relaxed structure gives reasonable confidence that the energy landscape near that geometry is not an artifact of any single potential. However, when phonon calculations were subsequently carried out using Phonopy at a fully converged supercell size, the bulk crystalline phase exhibited pronounced dynamic instability: 32 imaginary phonon modes were identified, including 4 at the Gamma point, with the softest mode reaching approximately -2.17 THz. This means the ideal crystalline bulk structure is not a true ground-state minimum — it would spontaneously distort or amorphize under real conditions. For a glass, this finding is not disqualifying. Oxide glasses by definition lack long-range crystalline order; the relevant embodiment here is an amorphous or glass-ceramic thin film, not a crystalline bulk solid. The phonon instability of the hypothetical crystalline parent structure actually reinforces the expectation that the material prefers a disordered or locally ordered glass phase. The computational verdict that shaped the claim strategy is therefore materially correct in its reasoning: because a broad "bulk composition" claim anchored in a specific crystal structure cannot be defended when that structure is dynamically unstable, the filing narrows to the thin-film device embodiment where the glassy nature is inherent and the relevant performance metrics — EUV transmission, mechanical integrity at 30–70 nm, thermal conductivity under EUV dose — do not require long-range crystalline order. The claim family also extends coverage to alkali and alkaline-earth aluminoborate analogs, providing compositional breadth within the glassy materials space. Key open technical questions include independent DFT phonon recomputation on both crystalline and amorphous model structures, and — most critically — direct 13.5 nm EUV bench transmission measurements on actual deposited thin films. The predicted >80% transmission figure is compositionally motivated and internally consistent with published EUV optical constants for constituent elements, but it has not yet been confirmed by synchrotron or EUV-source measurement. Thermal stability under repeated EUV pulse loading, mechanical robustness of a free-standing 30–70 nm oxide glass membrane, and deposition process compatibility (sputtering vs. ALD vs. CVD routes) are additional engineering questions that a pellicle development program would need to address.

The EUV pellicle market is a specialized sub-segment of the broader semiconductor photomask and process equipment consumables space. Current estimates for the addressable pellicle market — spanning EUV pellicle membranes and associated frames, predominantly for logic and memory manufacturers operating EUV scanners — fall in the $200–500 million range. This figure reflects the relatively small number of installed EUV scanners globally (measured in hundreds of units, not thousands), high per-unit pellicle cost, and replacement cycles driven by EUV dose accumulation and contamination. High-NA EUV tools, which are entering early production at leading logic nodes, carry even higher thermal loads on pellicles and are widely expected to accelerate materials development investment, potentially expanding the addressable market as tool counts grow through the late 2020s. Buyers in this space are not commodity purchasers. EUV pellicle qualification is an extraordinarily high-stakes process: a single pellicle failure can destroy a photomask worth hundreds of thousands of dollars and contaminate a scanner. Customers — principally ASML's pellicle program, Mitsui Chemicals (the current dominant pellicle supplier), S&S Tech, and the in-house development arms of TSMC, Samsung, and Intel Foundry — pay premium prices and move slowly through qualification. The licensing or partnership model for a novel pellicle material would most likely take the form of a materials supply agreement with a royalty component, or a co-development arrangement in which the IP holder receives milestone payments and a per-unit royalty after qualification. At pellicle prices of several thousand dollars per unit and annual EUV scanner output of potentially millions of wafers per node, even a modest royalty on a qualified alternative material creates meaningful recurring revenue. The radiation-hardened substrate application — windows and substrates for space and defense electronics exposed to particle radiation — represents a separate, smaller, but strategically distinct customer set less sensitive to the extreme qualification demands of lithography.

non-Si EUV-pellicle / rad-hard option in narrow thin-film form

narrow thin-film <10 um EUV-pellicle / rad-hard embodiment; broad bulk composition not asserted (phonon-unstable)

The primary strategic acquirers and licensees for this asset are companies with active EUV pellicle development programs or direct commercial interest in qualifying alternative pellicle materials. Mitsui Chemicals, as the largest current EUV pellicle producer, has clear motivation to hold IP on alternative materials that could either extend their product line or defensively block competitors. ASML, which co-develops pellicle specifications with customers and has its own materials research program, is another natural counterparty. Semiconductor manufacturers operating EUV fabs at scale — TSMC, Samsung Foundry, and Intel Foundry — have internal pellicle qualification programs and IP strategies designed to reduce single-supplier dependency; a narrow but defensible filing on an oxide-glass pellicle approach could be attractive as a portfolio addition. Specialty glass manufacturers such as Corning or Schott, who already have expertise in thin-film oxide glass deposition and optical coatings, could see a licensing opportunity in a defined application space where their existing process capabilities apply. For the radiation-hardened substrate application, the buyer set shifts toward defense and space electronics integrators: BAE Systems, Teledyne, Raytheon (RTX), and boutique rad-hard substrate suppliers serving satellite and launch vehicle programs. These customers value material novelty combined with defensible IP that can be incorporated into long-duration supply agreements. The asset's value in both segments is enhanced by its embeddedness in the broader PFAS-free dielectric and process fluids portfolio, which provides context for cross-licensing discussions where a counterparty's interest in one asset in the portfolio creates negotiating leverage across others.

The most significant technical risk is the gap between predicted and demonstrated EUV transmission. The greater-than-80% transmission figure at 13.5 nm is a prediction derived from elemental optical constants and film thickness modeling; it has not been confirmed on a physically deposited film under real EUV illumination conditions. Free-standing aluminoborate glass membranes at 30–70 nm present substantial process challenges — achieving film uniformity, adhesion, and mechanical integrity across pellicle-scale areas (tens of centimeters in diameter) in an amorphous oxide is non-trivial, and the thermal conductivity of oxide glasses is generally lower than silicon, which creates dose-handling risk under high-power EUV sources. The phonon instability of the bulk crystalline phase, while consistent with glassy behavior, means that standard crystalline-phase stability metrics cannot be used to benchmark this material and that glass-state simulations (molecular dynamics of amorphous models, for example) would be needed for a more complete computational picture. The patent scope is deliberately narrow, which limits upside but also limits defensive utility against competitors who file on compositionally adjacent but technically distinct glass formulations. The roadmap to de-risk the asset follows a clear sequence: amorphous thin-film deposition by sputtering or ALD, EUV transmission measurement at a synchrotron or EUV metrology tool, thermal cycling and dose-accumulation testing, and mechanical robustness characterization of free-standing membranes. DFT-level recomputation of the phonon landscape for both crystalline and amorphous model structures would strengthen the computational case and potentially reveal additional structural analogs worth claiming. If transmission and mechanical results are positive, a co-development agreement with a pellicle manufacturer or a fab-sponsored materials qualification program would be the natural commercial vehicle, with the IP serving as the entry point for that partnership.