Yield meetings talk overlay. Capacity meetings eventually talk coat uniformity and resist availability. When a track coater starts showing edge-bead anomalies, dispense drift, or lot-to-lot viscosity surprises, the scrap conversation is already late. The conversation that should have happened treats photoresist as a controlled supply chain—cold chain, shelf life, nozzle health, and PEB timing—not a drum that arrives and disappears into folklore.
This brief is that supply-chain map for the coater cell. It is not an EUV pellicle logistics piece, not an AMC detective timeline, and not a dry-pump reliability story.
Symptom → likely owner (before you blame the scanner)
| What you see | Often not the scanner | First place to look | | --- | --- | --- | | Edge bead / thickness map drift | “Scanner focus” | Dispense pump, nozzle wear, cup temperature | | CD swing after resist lot change | “Exposure dose” | Viscosity, filter integrity, lot pre-soak | | Particle adds at coat | “Fab AMC” | Bottle change procedure, dispense line purge | | Soft defects after idle weekend | “Humidity folklore” | Chiller setpoint, resist age in line | | Queue time without tool down | “Planning” | Resist inventory, cold-chain gap, filter lead time |
The point of the table is ownership: materials owns lot pedigree and storage; equipment owns dispense and uniformity; litho owns the process window—but only a named track coater owner connects them before the queue board turns red.
A clean nozzle photo is not a viscosity certificate.
What a track coater program actually controls
Track coaters do not forgive resist abuse. They translate chemistry, temperature, and dispense precision into a film that lithography assumes is truthful. Mature fabs run FIFO with shelf-life alarms, nozzle and filter change counters keyed to dispense volume, uniformity maps trended like SPC, and bottle-change rituals that treat bubbles and skin formation as scrap precursors—not operator superstition.
Supply is not “two weeks on the shelf.” Advanced resists carry cold-chain contracts, filtration specs, and sometimes fab-specific qualification lots that cannot be swapped overnight. A site that optimizes inventory to one day of cover without a named escalation path will discover that “just use the previous generation” is not a plan when the scanner is idle and the bottleneck is a filter lead time measured in weeks.
Uniformity drift is lithography drift with a one-step delay.
An anonymized Thursday that should have been Monday
A logic fab lost sixteen hours of lithography queue after three coaters simultaneously showed thickness map drift following a resist lot rotation. The ticket trail showed bottles warmed outside spec during receipt, deferred nozzle changes because “uniformity was still in spec on the report,” and a materials system that allowed lot swap without forced pre-dispense qualification. Post-mortem was not exposure dose. It was a missing gate: any resist lot change without within-run uniformity check blocks scanner release until disposition—same shift, named owner.
Magnitudes without brochure theater
Exact thickness targets belong to your node and your resist vendor. Ordering is portable: lost scanner hours outrank deferred filter spend; cold-chain proof outranks warehouse convenience; dispense health outranks blaming the last AMC sample. Sites that only monitor scanner CD will keep celebrating “stable overlay” while the coater quietly writes the next defect chapter.
Adjacent topics that stop here
EUV pellicle supply owns mask infrastructure logistics. Lithography AMC hunts own molecular contamination detective work. Dry vacuum pumps own etch/CVD foreline health. CMP slurry owns planarization consumables. None of them own resist viscosity, dispense nozzle wear, or coat uniformity SPC. Do not fund an AMC sample cart and expect track queue time to improve.
Questions that replace hope
Who can swap resist lots without a recorded uniformity disposition? When did you last trend nozzle change intervals against dispense volume—not calendar? Is cold-chain verified at receipt or assumed from the label? If the resist vendor disappeared for a month, would your MES still know qualified lots, filter stocks, and which coaters ran which pre-soak?
The track coater is not a peripheral wet tool. It is the reason your “available” scanner still waits in the queue. Instrument dispense, name the owner, and treat photoresist as a spec product—because lithography already does.