Rinse with DI water, then 0.1 M HNO3 for 10 min in an ultrasonic bath, then the solvent matched to the previous run's organics, then three DI rinses of at least 10 % of cell volume each. Accept the cell when a blank run returns under 1 % of the previous signal and rinse-water TOC is below 0.5 mg/L.
A 500 mL slurry at 0.5 g/L holds 250 mg of catalyst. If 1 % survives cleaning, the next run starts with 2.5 mg of fresh, unpoisoned surface — only 5 mg/L against a nominal 500 mg/L, but fresh, and the effect is a slow upward drift in apparent activity that gets misread as catalyst activation.
Organics behave differently. A monolayer of adsorbed substrate or photopolymer on the catalyst or the quartz window blocks sites and filters light, producing a lag at the start of the next run that is easily reported as an induction period. Grease is worse, because it is hydrophobic, invisible, and changes wetting rather than chemistry: a greased patch scatters light and leaves a dry spot the slurry cannot wet.
| Contaminant | Source | Symptom in the data | Removal |
|---|---|---|---|
| Catalyst fines | 0.2–2 g/L slurry, 21 nm particles | Activity creeps up run to run; cloudy rinse | 0.1 M HNO3, 10 min ultrasonic |
| Adsorbed substrate | Previous reactant, photoproduct, dye | Lag phase, coloured rinse, low initial rate | Matched solvent, then IPA, then DI |
| Silicone grease | Ground joints, stopcocks | Hydrophobic patches, erratic wetting, Si–O–Si at 1020–1100 cm−1 | Toluene wipe, then IPA; fit PTFE sleeves |
| Metal ions | Stainless fittings, stir bar, tap water | ppm Fe, Cr or Ni; shifted selectivity | 0.5 M HNO3 for 30 min |
| Detergent residue | Glassware soap | TOC 5–50 mg/L, foaming, blocked sites | Stop using detergent |
Order matters more than the agent. Bulk removal comes first so the acid is not spent neutralising slurry, and solvent comes after acid so it need not cut through a catalyst film.
The acid step works because titania is amphoteric with a point of zero charge near pH 6.2. Below pH 4 the particles carry net positive charge and repel each other, so 10 min of ultrasonication lifts a film that water leaves behind. Alkali also strips organics but attacks the borosilicate network past about 30 min of soaking and leaves sodium on the surface. Two agents belong nowhere near the sink: HF, which etches fritted discs and quartz, and chromic acid, whose Cr(VI) residue absorbs across 350–450 nm.
Visual inspection fails at exactly the levels that matter, so use three instrumental checks. A blank run under the same light, temperature and purge for 2–4 h should return under 1 % of the previous signal and below three times the GC detection limit; for a previous rate of 200 µmol/h that means under 2 µmol/h. Rinse-water TOC below 0.5 mg/L catches detergent and dissolved organics a blank run may miss. Contact angle on the quartz window below 20° confirms a hydrophilic surface; a greasy film reads 40° or more.
Two consumables should be replaced, not cleaned. Silicone tubing absorbs organics and permeates oxygen roughly two orders of magnitude faster than PTFE, so a cleaned line still bleeds air into a sealed system. Septa lose their seal after about 50 punctures, and a leaking septum looks exactly like a contaminated reactor: falling rate, rising oxygen.
Only inside a sealed, unopened cell with unchanged medium, and no more than two or three times. Each opening admits air and dust, and the slurry segregates as it stands. Once opened, run the full sequence: 45 min against a day of ambiguous data.
Dilute nitric acid, not an organic solvent. Below pH 4 the particles repel each other and a 40 kHz bath lifts the film in about 10 min. Water alone leaves a haze because particles re-deposit as the film dries.
A 2–4 h blank with the lamp on, returning under 1 % of the previous signal and under three times the detection limit; rinse-water TOC below 0.5 mg/L; contact angle under 20°. If any check fails, repeat the acid step.
Not at 60–80 °C, which is all that is needed. Damage comes from drying a sealed cell: internal air gains 0.34 kPa per kelvin, so 25 °C to 120 °C adds about 32 kPa and will lift a septum or crack a window.
