Five consecutive runs, each with at least 95 % catalyst mass recovery and no more than 3 % rate loss per cycle, is the minimum that supports a stability claim. Three runs only shows the experiment repeats. Report the mean and standard deviation of triplicate rates, not the best cycle.
A cycling test answers one question: does the material that comes back out of the reactor still do the same job? That makes it a mass-balance test first and a rate test second. If 50 mg goes in and 46 mg is recovered, the 8 % rate drop needs no deactivation mechanism to explain it — the catalyst is simply gone, on the filter, on the reactor wall, or down the sampling line.
So every cycle has to close two accounts: how much solid came back, and how fast it produced product. A paper that reports only the second is reporting the first one's error bar.
| Cycle | Treatment between runs | Retention target | What a miss points to |
|---|---|---|---|
| 1 | Fresh catalyst, 4 h run | 100 % (reference) | Check the blank before trusting the baseline |
| 2 | Filter, wash, vacuum dry | ≥ 97 % | Below 90 %: solid lost in filtration or sites blocked by product |
| 3 | Filter, wash, vacuum dry | ≥ 94 % | Below 85 %: progressive deactivation, send the filtrate for ICP |
| 4 | Wash plus regeneration step | ≥ 92 % | A drop only after regeneration: the treatment itself changed the surface |
| 5 | Filter, wash, vacuum dry | ≥ 90 % | Below 85 % cumulative: the claim is “partially stable”, say so |
Run each cycle at the same light intensity, donor concentration and temperature, and record them; a cycling series in which irradiance drifts 10 % between runs is a lamp test, not a catalyst test. Use the same batch of donor solution for the whole series, because a fresh batch changes pH and ionic strength.
A fifth, quieter cause is statistics. Triplicate reactors in one lab typically scatter 3–8 % relative standard deviation. With 8 % RSD you cannot resolve a 5 % loss between two cycles; either run triplicates and report confidence intervals, or extend the series until the trend exceeds the noise.
Filter through a 0.22 µm membrane rather than 0.45 µm, because a fine powder loses a visible fraction through the larger pores. Wash with 20 mL of water three times, then once with the solvent used in the run, and dry at 60 °C under vacuum for 4 h. Weigh on an analytical balance after the sample reaches constant mass; a 50 mg charge needs 0.1 mg resolution to resolve a 0.2 % loss.
For regeneration, choose the mildest step that restores activity: a water wash, then a dilute base wash for carbon residues on TiO2, then vacuum drying. Avoid calcination between cycles unless the phase is known to be stable at that temperature — a 400 °C step will sinter a 20 nm powder and produce a genuine, irreversible loss that then gets attributed to “cycling”.
Confirm the surface came back unchanged with XRD before and after the series, and with XPS if a binding-energy shift is suspected. A retention of 95 % with a new shoulder in the XRD pattern is a phase change that happens to preserve the rate, not stability.
Give the rate of each cycle in µmol h-1 g-1, the recovered mass in mg, and the retention percentage against cycle 1. Add the number of runs, the RSD of the triplicates, and the light and donor conditions. A compact table of five cycles with mass recovery beats a bar chart with no error bars, and a reviewer can audit it.
State the claim at the resolution the data supports. Five cycles at 90–95 % retention justifies “stable over five runs”, not “highly stable” and certainly not “durable”. Long-term claims need tens of runs or a continuous-flow test of hundreds of hours, and those are a different experiment.
Five consecutive runs is the practical floor for a journal figure, and three is enough only to show the experiment is repeatable. Each cycle should reach at least 95 % mass recovery and lose no more than 3 % of the rate, so a five-cycle series ends at 90 % or better.
Usually because something left the reactor. Check the recovered mass first, then the filtrate by ICP for dissolved metal, then the pH and donor concentration. A genuine first-cycle drop of more than 10 % with full mass recovery points to adsorbed product or a change in surface oxidation state.
Report both, and plot the rate per gram of recovered catalyst for the stability trend. Per-initial-mass values mix catalytic decay with handling loss, which is the most common reason a stable catalyst appears to deactivate by 5–8 % over five runs.
It can, in both directions. Washing removes adsorbed products and restores sites, which raises the next cycle's rate; aggressive washing or drying can also strip cocatalyst or sinter the powder. Keep the protocol identical across all cycles and describe it in enough detail to be repeated.