EUV Lithography
# EUV Scanner Availability and Fab Capacity Planning
An EUV scanner's effective output isn't just its rated wafers-per-hour exposure speed — it's that speed multiplied by the fraction of time the tool is actually running production wafers rather than down for maintenance, source servicing, or recovery from an unplanned fault, and that availability fraction has historically been a bigger lever on real fab output than raw throughput numbers suggest. A scanner rated at a certain wafers-per-hour figure only delivers that output during the hours it's actually exposing wafers. Every hour spent on scheduled preventive maintenance, droplet-generator servicing, collector mirror cleaning cycles, or recovering from an unplanned stoppage is an hour of zero output from one of the most expensive and capacity-constrained tools in the entire fab. Because EUV scanners cost on the order of hundreds of millions of dollars each and a leading-edge fab typically has relatively few of them, the gap between theoretical throughput and actual delivered wafers matters enormously to a fab's real production capacity.
EUV scanners have historically required more maintenance attention than their DUV predecessors specifically because the source itself — tin droplets, dual-pulse lasers, collector mirror cleaning, hydrogen radical chemistry — is mechanically and chemically more complex than any light source a DUV scanner ever needed, and that complexity is the primary driver behind availability being a harder problem for EUV than for the lithography generations before it. A DUV scanner's light source is comparatively simple and robust by comparison; an EUV source involves a precision droplet generator firing tens of thousands of times per second, a high-power laser hitting a moving target, and a cleaning chemistry balance that has to hold steady continuously. Every one of those subsystems is a potential source of unplanned downtime that simply didn't exist in prior lithography generations, which is part of why availability improvement has been as central to EUV scanner development roadmaps as resolution or throughput specifications.
Fab capacity planners model EUV scanner fleets with availability as an explicit, tracked variable — not an afterthought — because with relatively few EUV scanners per fab compared to the hundreds of DUV tools a mature fab line might have, a single scanner's availability swing has an outsized effect on total fab throughput compared to losing equivalent availability on a more abundant tool type. This concentration risk is why fab operations teams treat EUV scanner maintenance scheduling, spare-parts logistics, and source-servicing cadence as capacity-planning problems with direct revenue impact, not purely as equipment-engineering concerns — losing a few percentage points of availability on a handful of EUV tools can constrain an entire fab's output in a way that a similar availability dip on a widely-replicated DUV tool type never would.
Projection-optics carbon contamination and collector-mirror tin debris are two of the ongoing maintenance burdens that directly eat into scanner availability — every contamination-control cleaning cycle this session's rows have described is, from the fab capacity planner's perspective, exactly the kind of scheduled downtime that availability modeling has to account for, tying EUV's deepest physics problems directly to a fab's bottom-line wafer output.