An EUV mask is reused for every wafer in a production lot, so a single defect on the mask repeats on hundreds of wafers — one of the most expensive failure modes in the fab. Preventing mask defects is therefore high-value yield engineering. Taiwan Semiconductor Manufacturing Co., Ltd.'s mid-2022 grant claims a defect-prevention approach.

US11402743B2, "Mask defect prevention" (issued 2022-08-02), is classified across the EUV-mask subclasses G03F 1/24 with G03F 1/36, 1/38, 1/54, and 1/64, plus G03F 7/2004. The breadth of mask-related CPC codes signals a claim spanning mask design and handling aimed at preventing defects before they reach the wafer.

“A photolithographic mask assembly according to the present disclosure accompanies a photolithographic mask. The photolithographic mask includes a capping layer over a substrate and an absorber layer disposed over the capping layer.”— U.S. Patent No. 11,402,743 source

Construe the prevention limitation, which turns out to be a very specific structural idea rather than an aspiration. Independent claim 1 recites a mask with a capping layer over a substrate and an absorber layer that is divided into a first main feature area, a second main feature area, and a venting feature area between them — the venting area holding a plurality of "venting features." These venting features are the mechanism: sub-resolution openings placed in the otherwise-blank regions of the absorber so that reflected EUV intensity is locally bled off where no pattern is meant to print. Claim 4 makes them a set of parallel elongated bars, and claim 5 is the key constraint — the bars are dimensioned so they do not print and do not affect printing of the main feature areas. They relieve a defect-causing condition without adding anything to the wafer.

The materials are conventionally claimed (claim 2: capping layer of ruthenium, silicon oxide, and/or amorphous carbon; claim 3: absorber of Cr, TaN, TaO, TaBN, TiN, TaBO, Cr₂O₃, SiO₂, or SiN), and claims 6—8 add a pellicle frame with an opening and a black-border area, plus a second venting feature area within that black border made of more parallel bars. The point is that venting is placed both between main features and in the border region where stray reflected energy and heat accumulate.

The method claims describe how the venting is inserted and why it prevents defects. Claim 9 receives a mask design with two main feature areas and a divider, determines a venting-feature insertion area in each, inserts venting features, and fabricates the mask. Claim 10 conditions insertion on a template shape fitting a pattern-free area — identifying where there is room to vent and excluding where there is not — and claim 11 sizes that rectangular template at roughly 4—6 µm per side. Claim 12 runs optical proximity correction afterward. Most tellingly, claim 14 exposes the mask and uses radiation reflected from the main and venting areas to expose the wafer while keeping the intensity reflected by the venting area below the photoresist's exposure threshold — the defect being prevented is the printing or thermal artifact from concentrated reflection off large blank regions. Claim 18 forms the venting features by completely removing the EUV absorber to expose the underlying multilayer mirror.

The third independent claim, claim 16, broadens the insertion rule away from the two-main-feature framing: it receives a design with printing features and pattern-free areas, identifies a venting-feature insertion area wherever a template shape fits within the pattern-free areas, inserts venting features there, and fabricates the mask. Claim 17 again runs optical proximity correction after insertion and before fabrication, and claim 18 ties the fabrication to a real EUV blank — a substrate with a multilayer mirror, a capping layer over it, and an EUV absorber over the cap — with the venting features formed by completely removing that absorber over part of the insertion area. The mechanism, read across all three independent claims, is consistent: blank, unpatterned stretches of absorber reflect (or absorb and re-emit as heat) EUV in ways that disturb the nearby printed pattern and stress the mask, so the method peppers those blank stretches with sub-resolution openings that locally relieve the condition while the bar dimensioning (claim 5) and the below-threshold reflected-intensity limit (claim 14) guarantee nothing extra prints. The defect "prevented" is thus a reflection- and heat-driven artifact, and the cure is a layout transformation applied at mask-design time, made manufacturable by the absorber-removal step.

The design-around space is the prevention mechanism. EUV mask defectivity is attacked many ways — pellicles, inspection-and-repair, mask-handling protocols, design-for-manufacturability rules — and a competitor using any of those instead of inserted sub-print venting features reaches comparable yield outside a claim tied to TSMC's venting-area scheme, the non-printing-bar dimensioning, and the below-threshold-reflection requirement.

The breadth of the CPC list reflects that the claims straddle mask design and the exposure process. G03F 1/24 is the EUV mask itself; G03F 1/36, 1/38, 1/54, and 1/64 cover absorber design, phase-shift considerations, mask inspection/repair, and mask blanks; and G03F 7/2004 reaches the EUV exposure step where the venting features do their work below the resist threshold. That spread is consistent with three independent claims that run from a static mask assembly (claim 1) through a design-time insertion method (claim 9) to a generalized pattern-free-area insertion rule (claim 16), all converging on the same sub-print venting mechanism. A competitor's EUV program is near this claim only if it relieves blank-region reflection by inserting non-printing features into the absorber — not if it solves defectivity through pellicles, repair, or handling.

This is operational yield IP — the kind a foundry generates from running the technology at the highest volume in the world. TSMC's EUV mask patents from this period (anti-contamination, defect prevention, radiation modification) collectively encode hard-won fab experience into defensible claims, and identifying the specific venting mechanism here — not the generic title — is what determines whether a competing EUV program is anywhere near the claim.