EUV Scanner Optics

Every lithography generation before EUV used refractive lenses to focus light, and EUV is the generation where that approach simply stopped being physically possible: at a wavelength of 13.5nm, light is absorbed by essentially every optical material, including the fused silica and calcium fluoride glasses that focused every DUV and ArF scanner before it, so an EUV scanner cannot use a lens anywhere in its optical path and instead builds the entire illumination and projection system out of precisely figured mirrors. Mirrors alone are not enough, because no single reflective material or simple metal coating reflects 13.5nm light efficiently either; the solution ASML and its optics supplier Zeiss settled on is the molybdenum-silicon (Mo/Si) multilayer mirror, a stack of 40 to 50 alternating Mo and Si layer pairs each roughly 7nm thick, engineered as a Bragg reflector so that the reflections off each interface add constructively at 13.5nm even though no single layer boundary reflects much on its own. That multilayer trick is the only reason an EUV mirror reflects usable light at all, and even so each individual Mo/Si mirror only achieves roughly 65 to 70 percent reflectivity, a number that becomes the central economic and engineering constraint of the entire scanner once enough mirrors are chained together to steer and focus the beam from source to wafer.

A full ASML EUV scanner's illumination and projection optics path uses on the order of 10 to 11 of these Mo/Si mirrors in sequence, and because reflectivity losses multiply rather than add, that mirror count is the real reason EUV power at the wafer is such a hard-won, expensive result compared to any DUV system. Each mirror the beam bounces off keeps only about two-thirds of the light that hit it, so ten mirrors at 65 percent reflectivity apiece leave only a small single-digit percentage of the original EUV power reaching the wafer, before the EUV source's own tin-droplet conversion-efficiency limits are even factored in; this compounding loss is why EUV source power and mirror reflectivity have been pursued as aggressively as any other variable in EUV scanner development, and why every fractional percentage-point improvement in a single mirror's coating quality is worth chasing at the full system level. The entire optical path, mirrors included, also has to run in high vacuum, because EUV light is absorbed by air and most ordinary gases just as thoroughly as it is absorbed by glass, so there is no "purge with an inert gas and keep some ambient pressure" option the way DUV scanners have; every mirror surface, every mask, and the wafer itself sit inside a vacuum chamber for the entire exposure.

EUV source generates 13.5nm light
Tin-droplet plasma source; conversion-efficiency limits already cap usable power before optics losses begin
Illumination optics
Several Mo/Si mirrors shape and direct the beam
Each mirror ~65-70% reflective; losses multiply mirror to mirror
Reflects off the patterned EUV mask
Mask is itself a Mo/Si multilayer mirror with an absorber pattern on top; protected by euv-pellicle-process-and-requirement
Projection optics
Several more Mo/Si mirrors demagnify and relay the pattern
Final mirrors in the ~10-11 mirror path; cumulative reflectivity loss is steepest here
Pattern exposed onto photoresist-coated wafer
Final image lands on the photoresist-process stack; this is what all the preceding optical loss was spent to deliver

That flow runs inside a single constraint the diagram above only hints at: the entire path, source chamber through wafer stage, is held in high vacuum from end to end, because any gas in the beam path would absorb EUV light the same way glass does. The patterned EUV mask sitting in the middle of that path is not a passive stencil the way a DUV photomask is; it is built on the same Mo/Si multilayer principle as every other mirror in the system, with an absorber material patterned on top of the multilayer stack to define the circuit pattern, which is exactly why a single particle landing on that mask surface is such a severe defect risk and why euv-pellicle-process-and-requirement exists as a dedicated standoff membrane to keep particles out of the mask's focal plane. Once the pattern reflects off the mask and passes through the projection mirrors, it lands on a wafer coated through photoresist-process, and every photon that survived the illumination mirrors, the mask reflection, and the projection mirrors is what that resist layer actually has available to react with.

DUV/ArF refractive lens opticsEUV reflective Mo/Si mirror optics
Optical material approachTransmissive glass lenses (fused silica, CaF2); light passes through the materialReflective Mo/Si multilayer mirrors; light never passes through bulk material, only reflects off engineered layer stacks
Per-element efficiency / lossLens transmission losses are comparatively small per element, even across many lens elementsEach mirror reflects only ~65-70% of incident light; losses multiply across a ~10-11 mirror illumination + projection path
Vacuum requirementOperates in air or with modest inert-gas purge; no absorption problem from ambient atmosphereRequires full high-vacuum along the entire optical path; EUV is absorbed by air and most gases
Cost / complexity consequenceMature, comparatively lower-cost optical trains; refractive lens manufacturing is a well-established disciplineDramatically more expensive scanners; cumulative mirror loss plus source conversion-efficiency limits make usable EUV power at the wafer a hard-won engineering result, and the vacuum requirement adds system complexity DUV never needed
EUV Scanner: All-Reflective Mo/Si Mirror Optical Path No lens survives 13.5nm light, so every optical element in the path is a precision mirror, in high vacuum HIGH VACUUM CHAMBER — entire path, source to wafer Sn EUV source 13.5nm light M1 M2 Illumination optics (shapes + directs beam) Mask Mo/Si mirror + absorber pattern M3 M4 M5 Projection optics (demagnify + relay pattern) … ~10-11 mirrors total Photoresist-coated wafer pattern exposed here Every green mirror above is a Mo/Si multilayer, not a plain reflective coating — no single-material surface reflects 13.5nm light usefully The mask (purple) is protected from particles by a standoff pellicle membrane; it is a mirror with an absorber pattern, not a transmissive photomask EUV source (tin droplet plasma) Mo/Si multilayer mirror Patterned EUV mask Wafer / photoresist Mo/Si Bragg stack, and why loss multiplies mirror by mirror Mo/Si stack (zoom) 40-50 Mo/Si pairs ~7nm per pair tuned as a Bragg reflector for 13.5nm Power remaining after each mirror (~65-70% kept per bounce) M1 after 1 mirror: ~67% after 5 mirrors: ~13% after 10-11: low single digits Reflectivity loss compounds multiplicatively, not additively, across the ~10-11 mirror path Combined with EUV source conversion-efficiency limits, this is why usable power at the wafer is so hard-won Mo/Si multilayer mirrors are the physical reason EUV scanner cost and throughput differ so sharply from DUV

The reflective Mo/Si mirror stack is not an incremental optics upgrade carried over from DUV, it is a completely different optical architecture forced into existence by one physical fact: nothing transmits 13.5nm light well enough to build a lens out of it, so EUV lithography had to abandon refraction entirely and rebuild its illumination and projection system as a chain of precision-figured, Bragg-tuned multilayer mirrors operating in full vacuum. That architecture is also the direct cause of EUV scanner economics: each Mo/Si mirror's roughly 65 to 70 percent reflectivity multiplies against every other mirror in the roughly 10 to 11 mirror illumination-plus-projection path, compounding into a steep cumulative power loss long before the beam reaches the wafer, and that loss, stacked on top of the EUV source's own conversion-efficiency ceiling, is the root engineering reason EUV scanners cost what they do, why EUV throughput has always lagged behind DUV on a wafers-per-hour basis, and why the entire industry remains dependent on ASML, the sole manufacturer capable of building and qualifying this all-reflective optical system at production precision, with Zeiss as its optics supplier.

Take EUV scanner optics further

Ask the copilot about this term, or have our engineers assess it against your process.