Run four blanks before you report any photocatalytic rate: dark, photolysis without catalyst, catalyst without donor, and an empty-reactor lamp run. Match the volume, headspace and sampling interval of the real test. The dark run must stay below 3 % of the light-driven yield over 30–60 min; anything above that is thermal background or stray light, not photocatalysis.
A control experiment does not prove the mechanism. It removes alternative explanations, one at a time, until the only remaining source of product is light absorbed by the catalyst. That is a narrower claim than most papers make, and it is the one a reviewer will test first.
The four blanks below cover the four ways product appears without photocatalysis: heat, direct photolysis, the catalyst's own surface chemistry, and the apparatus. Each takes one reactor run of 1–4 h, and together they cost less than repeating the main experiment once.
| Run | Light | Catalyst | Donor | What a positive signal means |
|---|---|---|---|---|
| Full test | On, 100 mW/cm2 | 0.5 g/L | 10 vol% | Reference rate, µmol h-1 g-1 |
| Dark control | Off, reactor wrapped | 0.5 g/L | 10 vol% | Above 3 % of the light rate: thermal contribution or a light leak at the septum |
| Photolysis | On | None | 10 vol% | Product without catalyst: direct photolysis of the donor or solvent |
| Catalyst-only | On | 0.5 g/L | None | Product without donor: water splitting, or carbon desorbed from the powder |
| Empty reactor | On | None | None | Signal from housing, grease or septum: an instrument artifact, not chemistry |
Keep irradiance, temperature and stirring identical across all five runs. If the dark control is measured at 25 °C but the illuminated run reaches 42 °C, the comparison answers a different question, because a 17 °C rise alone can shift a thermal background by a factor of two.
Start with the dark run and hold it for 60 min, sampling at 0, 15, 30 and 60 min. A flat baseline over that window also tells you the reactor is gas-tight and the GC or sensor has no drift, which you need before interpreting anything else.
Then run photolysis and catalyst-only for the same duration as the main test, not shorter. A 30 min photolysis run cannot exclude a slow uncatalysed path that contributes 15 % over 4 h. Finally, run the empty reactor once per campaign, not once per sample: it characterises the rig, and it only needs repeating after you change septa, grease or tubing.
Use the same batch of donor solution for the blanks and the test. A fresh batch moves pH by 0.3–1.0 units and changes the ionic strength, which is enough to shift a marginal rate by 10 % and make the blank look better than it is.
Give each blank its own rate in the same units as the test, plus the ratio to the test in percent. A single sentence — dark 0.08, photolysis 1.9 and catalyst-only 0.4 against a test rate of 42 µmol h-1 g-1 — is more convincing than a paragraph claiming the blanks were clean.
Report the net rate as the test minus dark minus photolysis, and state the correction explicitly. If the correction exceeds 15 % of the raw rate, the experiment needs better light confinement or a purer donor before the number is worth publishing. Record irradiance, temperature and pH for every run, because the blanks are only valid at the conditions they were measured at.
Sixty minutes with sampling at 0, 15, 30 and 60 min is enough to show a flat baseline and to confirm gas tightness. For slow reactions, extend to the full duration of the main test, since a 30 min blank cannot exclude a background that accumulates over 4 h.
Yes, because the donor absorbs too. Methanol, TEOA and sulfide solutions all show measurable H2 under UV in the absence of catalyst, typically 1–5 % of a catalysed rate but occasionally 20 % with high donor concentration and short wavelengths.
Run it once per campaign and repeat it only after changing septa, grease, tubing or the lamp. It takes one afternoon and it is the only blank that catches contamination from the apparatus itself, which otherwise appears in every sample at a constant rate.
Treat it as a real background, not noise. Check for light leaks around the window and septum, measure the reactor temperature in the dark and under illumination, and re-run the dark control after wrapping the vessel in two layers of aluminium foil. If it persists, the reaction has a genuine thermal component.