When an experiment fails and the chemistry was fine, the reactor was usually not. Most "failed runs" are equipment states in disguise, and a short daily checklist catches most of them before they cost data.
Seals: inspect and seat clean, replace on schedule, not on failure. Windows: inspect for fouling and scratches, clean with the right solvent and material. Stirring: verify coupling before loading. Temperature path: probe present, reading sane against a second reference. Light path: quick intensity spot-check against the log.
Weekly: full intensity measurement at the working plane, pressure-hold verification on pressurized reactors, and a look at consumable stock, seals, septa, lamps. Monthly: deep clean, window polish or replacement evaluation, and calibration verification of the temperature chain. Log everything, a maintenance record is a troubleshooting aid in advance.
When results go wrong, check in this order: light (intensity, spectrum, alignment), then temperature (actual versus setpoint), then seals (leaks change chemistry silently), then stirring (suspension failure), then sampling (quench and timing), and only then the chemistry. The order is cheap-to-expensive and fast-to-slow, most faults are found in the first three.
Falling rates over days: lamp aging. Selectivity drift with time of day: ambient or bath temperature drift. Sudden oxygen sensitivity: seal aging. Replicates disagreeing: stirring or position. Diagnose the signature, fix the cause, log the fix.
For Perfectlight reactor support, see instrument maintenance and service notes.
