EUV Flare
EUV flare is the stray light that reaches the wafer from everywhere a mirror isn't supposed to send it — scattered off sub-nanometer surface roughness and residual particles on the multilayer mirrors — and it washes out image contrast even when the intended Bragg-reflected beam is perfectly focused. At a high level, the light budget splits like this:
A few things are worth knowing about why flare is treated as its own budget line, separate from both resolution and the stochastic-roughness problem:
It is a background-light problem, not a focus problem. A perfectly aligned, perfectly focused EUV scanner can still print low-contrast or CD-shifted features because every one of the six-to-eight mirror reflections in the optical train scatters a small fraction of the beam diffusely across the exposure field, rather than reflecting it along the intended ray path. That scattered light lands everywhere on the field as a slowly-varying background exposure — it doesn't blur a single edge the way aberration or diffraction does; it raises the floor under the whole image, which is why flare is specified and budgeted as a percentage (roughly 10-20% of total dose in practice) rather than a nanometer tolerance.
It depends on what's elsewhere on the field, not just the feature itself. Because scattered light is collected from the whole illuminated area, a dense array of features contributes more flare to its own neighborhood than an isolated line does — the same design can print differently depending on local pattern density, which is exactly the kind of context-dependent effect that computational lithography and OPC/ILT models must correct for alongside the usual proximity effects. Mask absorber design and reticle cleanliness both feed directly into the flare budget, since any particle or defect on the mask scatters light the same way a rough mirror does.
Mirror roughness control is the real lever, and it's brutally tight. The multilayer mirrors that make EUV imaging possible at all (ordinary refractive lenses absorb 13.5 nm light) must hold surface roughness to a few tenths of a nanometer across the full mirror aperture — any deviation scatters a disproportionate amount of EUV light because scatter scales steeply with roughness at this short a wavelength. That specification sits upstream of nearly every other EUV performance number: tighter mirror roughness buys back dose margin, throughput, and the contrast budget that lithographers would otherwise have to spend fighting flare with brute-force dose or resist tricks.
Read EUV flare through a *field-wide light-budget* lens rather than a *single-feature-printing* lens: it is the cost of building an imaging system entirely out of reflective optics at a wavelength where even near-perfect mirrors scatter a measurable fraction of every bounce, and it must be managed at the mirror-polishing and mask-cleanliness level upstream, because no amount of per-feature optical proximity correction downstream can remove a background that depends on the whole field, not just the edge being printed.