Clean an optical window when a measurement shows transmission has fallen, not when it looks dull. A sapphire window on a PLR-HTSR 2.0 is specified above 90 % transmission when new; a thin film or a haze moves you below your own baseline quietly, with no alarm and no change on the lamp power display. Measure, clean, measure again.
Window contamination reduces irradiance at the sample while every control reading on the bench stays where it was, so the loss is absorbed into your results rather than flagged. The lamp power supply holds its setpoint, the chiller holds its temperature, the shutter opens on time, and the reactor simply receives less light than the last campaign did. Because the loss is gradual, it arrives inside normal scatter and is easy to attribute to the catalyst, the sacrificial agent or the season.
The optics make it worse. A window is a short, high-angle path, and a film of a few tens of nanometers is enough to shift transmission noticeably while remaining hard to see under room light. Deposits also scatter rather than absorb, so they change the beam profile as well as its intensity, and the edge of the illuminated area dims first. That is why the verification step should include both a transmission reading and a look at the spot.
Identify the deposit before choosing a solvent, because the wrong solvent can bake a film into the surface or attack the coating. The table maps the common cases in photocatalytic work to a cleaning route and a check.
| Deposit | Appearance | Cleaning route | Verification |
|---|---|---|---|
| Catalyst film or photodeposit | Grey or colored patch matching the illuminated area | Flush with water, then wipe once with a lint-free tissue wetted with ethanol; repeat rather than rubbing harder | Transmission back to baseline; illuminated area even on a target |
| Condensation or water spotting | Fine droplets or ring-shaped marks after cooling | Dry with warm air first, then a single pass with a water-wetted tissue and a dry pass | No rings under oblique light; transmission stable |
| Salt creep at the seal | White crystalline fringe at the window edge | Dissolve with deionized water, wipe inward to outward, dry fully before refitting | No crystal residue in the groove; seal seats cleanly |
| Baked organic residue | Brown or amber tint that does not respond to water or alcohol | Soak in the solvent specified for the window material, then a single gentle wipe; never scrape | Transmission and color both restored; check for surface etching |
| Handling marks | Oily smears at the edges where fingers touched | Isopropanol on a lens tissue, one direction only, fresh tissue for each pass | No smear under oblique light |
The routes differ less in chemistry than in patience. Every entry assumes multiple light passes with fresh tissue rather than one firm rub, because the damage that ends a window's life is almost always mechanical and almost always caused by trying to finish the job quickly. If a film does not lift after three passes, stop and identify it rather than increasing pressure.
Treat the window as a coated optical component with a defined transmission, because that is what it is. The practical rules are short and they are not negotiable under time pressure.
Windows differ in what they tolerate, so the material matters. A sapphire window rated above 90 % transmission and used from room temperature to 500 °C in a PLR-HTSR 2.0 will take solvents that a coated transmission window in a 0.19–5 μm band will not, and a window specified above 97 % transmission in a PLR-GPTR50 deserves more care than either, because a small loss is a larger fraction of its margin.
Verification is a comparison, so it only works if the geometry is identical before and after. Mount the window in the same holder at the same distance from the source, use the same meter and range, and record the reading with the source at the same power setting and the same warm-up state. A hand-held check at a new distance measures your hand, not the window.
Take three readings and keep the median rather than the best one; the spread tells you whether the setup is stable enough to detect a real change. If the window is mounted in a reactor that cannot be removed, measure through the whole optical path with a clean reference window in place of the sample, and compare the difference between the two configurations. That method also catches misalignment, which produces the same symptom as contamination and is fixed differently.
Record the result against the value measured when the window was new, not against the last measurement alone. A window that has lost transmission slowly over a year will look stable campaign to campaign while sitting well below its original specification, and the drift will be attributed to everything except the window.
Replace the window when cleaning no longer restores transmission, when the surface shows scratches or crazing, or when the coating has a visible discontinuity. Scratches scatter light permanently and no solvent removes them; crazing indicates the surface has been attacked, and it will worsen with each thermal cycle. A chipped edge is a mechanical weakness as well as an optical one, and it can propagate under clamping load.
Replace also when the seal groove or the window seat shows corrosion, because a window that cannot be seated cleanly will leak regardless of how well it transmits. Keep one spare window per instrument with its transmission value written on the box, so a replacement can be verified immediately rather than after the next campaign.
Clean when the transmission check shows a drop, and inspect at every reactor opening. For a system running liquid-phase reactions with a suspended catalyst, that often means cleaning after each campaign; for gas-phase work with clean feed, inspection may be enough for several. Fixed calendars are less useful than the measurement.
Only if the coating and the window material are confirmed compatible with it, and never as a first attempt. Acetone attacks some coatings and dissolves some seal materials, and it leaves a residue if it evaporates from a dirty surface. Start with water, then ethanol, and move to stronger solvents only with the manufacturer's confirmation.
Yes, if it reduces the light reaching the sample. The change may sit inside run-to-run scatter, which is exactly why it goes unnoticed: it shifts every rate by a small, consistent amount. Measuring transmission before and after cleaning tells you the size of the effect on your own setup, and that number is the only one worth arguing about.
