Good enough is three measurements, not a cycle count: a vacuum that holds steady with the pump isolated, a pressure-rise rate inside your leak criterion, and background hydrogen below the analyzer's limit. On a 65 mL line the leak criterion is 0.1 μmol/h as oxygen; degassing ends when two consecutive evacuation–purge cycles stop changing the background.
Because the endpoint depends on four variables that a fixed count ignores: the liquid volume, the headspace and line volume, the solubility of the gas being removed, and the pressure the pump can actually reach. A 65 mL line with a shallow liquid charge arrives at a different background after three cycles than a tall vessel with a deep charge, and neither number tells you whether the rig is ready.
Each evacuation removes a fraction of the dissolved and headspace gas, and that fraction depends on how far the pressure falls before you stop. A count of cycles is therefore a count of dilutions of unknown ratio. The measurement that replaces it is the background itself: once two consecutive cycles return the same reading, further cycles are removing nothing you can detect, and the gas that remains is coming from somewhere else.
Write all three into the run sheet; a rig that passes one and fails another is not ready for a rate measurement.
| Criterion | How to measure it | Pass criterion | What failure means |
|---|---|---|---|
| Vacuum hold | Isolate the pump at working vacuum and watch the gauge | Steady once outgassing settles, with no continuous rise | A real leak, or a virtual leak from wet walls and trapped volumes |
| Pressure-rise rate | Timed rise over a fixed interval, converted to a molar flow | At or below 0.1 μmol/h measured as O₂ on a 65 mL line | Ingress at a joint, valve seat or septum |
| Background hydrogen | Analyzer reading after the final purge, before the lamp is switched on | Below 0.05 μmol for H₂, and below 0.0005 μmol for CH₄ and CO | Residual gas, carryover from a previous run, or a contaminated line |
The first two criteria test the boundary and the third tests the interior, which is why they are not interchangeable. A perfectly tight rig can still show a hydrogen background from the previous experiment, and a clean rig that leaks will show a background that grows with time rather than staying flat. Read the third row together with the first: flat and low, the rig is ready; flat and high, purge more; rising, find the leak.
Degas the liquid as well as the headspace, and do it in this sequence.
The order is not arbitrary. Purging before the pressure has dropped leaves gas in the dead volumes at the far end of the line, and filling before the liquid has equilibrated leaves dissolved gas to desorb during the run, where it will be mistaken for a product.
Oxygen consumes photogenerated electrons and re-oxidizes hydrogen on the cocatalyst, so residual oxygen appears as a low initial rate that recovers over the first hour — an apparent induction period that has nothing to do with the catalyst's own activation. In sacrificial systems it also oxidizes the hole scavenger, which changes the stoichiometry you calculate from the gas measured.
Nitrogen is less reactive but not harmless: it dilutes the headspace and lowers the hydrogen partial pressure, so a calibration that assumes a pure hydrogen atmosphere overestimates the molar quantity in the loop. Both effects are largest in the first sample after start-up, which is exactly the sample most people report.
If the background returns at the same rate after every cycle, the rig has ingress rather than dissolved gas, and more cycles will not help. Test at operating temperature as well as cold, because a joint that holds at 25 °C can weep at 60 °C, and re-torquing cold will not fix it. Wet walls are the other false positive: water desorbing from glass and tubing reads as a slow pressure rise for hours, so dry the line before you conclude that a seal has failed.
As many as it takes for two consecutive cycles to return the same background reading within the analyzer's noise band. That is usually three to six cycles on a small rig with a shallow liquid charge, and more when the liquid volume is large, because the liquid holds far more dissolved gas than the headspace. Record the background value each time, not just the count.
Both, but the liquid matters more because it holds the larger reservoir of dissolved gas. Evacuating only the headspace and then filling leaves the liquid to release its dissolved gas into the headspace over the first hours of the run. Stir during evacuation and allow an equilibration period after filling, so that the exchange happens before you start measuring.
Most often carryover from a previous run adsorbed on the line walls, or a septum and valve that were not purged, rather than incomplete evacuation. Purge with the reactant gas for longer, replace the septum, and compare a blank with the lamp off against one with the lamp on. If the lamp-on blank is higher, the source of hydrogen is in the reactor, not in the line.
