Size the headspace from the gas you will make, not from a fixed 1:1. Hydrogen dissolves to only 0.79 mM in water at 1 atm and 25 °C, so 50 µmol of H2 in a 100 mL headspace is already 1.2 vol%. Holding the predicted H2 fraction under 1 vol% at the end of a run usually means 2:1 to 4:1 headspace-to-liquid.
Gas-phase concentration. This is what your GC measures. The Ostwald coefficient of H2 in water is about 0.019 at 25 °C, so at equilibrium the gas phase holds roughly 50 times the concentration of the liquid. With any reasonable headspace, more than 90 % of the produced hydrogen ends up in the gas; the ratio decides how diluted it is, not whether it gets there.
Total pressure. A sealed vessel accumulates gas and the pressure rises in inverse proportion to headspace volume. Glass reactors with silicone septa start leaking at 20–40 kPa gauge, and the leak rate of a septum rises steeply with pressure, so a small headspace does not only concentrate the product, it also makes the product leak out.
The flammable envelope. The lower flammability limit of hydrogen is 4.0 vol% in air and in oxygen, with an upper limit near 75 vol% in air. Water splitting makes H2 and O2 in a 2:1 ratio inside the same vessel, which is the worst possible mixture to concentrate.
The dissolved inventory. Saturation at 1 atm is 0.79 mmol/L, so 50 mL of slurry holds about 39 µmol of dissolved H2 before any appears in the gas. That dissolved pool is what Pt can re-oxidise, and it makes headspace sampling lag the true rate by tens of minutes at low production.
| Headspace : liquid | H2 at 50 µmol, vol% | Pressure rise at 50 µmol H2 + 25 µmol O2 | Suitable for |
|---|---|---|---|
| 1 : 5 (10 : 50 mL) | 12.3 | 18.6 kPa | Nothing routine: flammable and septum-limited |
| 1 : 1 (50 : 50 mL) | 2.5 | 3.7 kPa | Low-rate catalysts, runs under 2 h |
| 2 : 1 (100 : 50 mL) | 1.2 | 1.9 kPa | Default for H2 screening |
| 4 : 1 (200 : 50 mL) | 0.6 | 0.9 kPa | High-rate catalysts, longer runs |
All values at 25 °C and 1 atm start, using 24.5 L/mol for an ideal gas. The pressure column is the reason 1:5 is not a conservative choice: the mixture crosses the lower flammability limit while the vessel is still far from any pressure alarm.
Take 50 mL of 1 g/L slurry in a 50 mL headspace, a modest rate of 50 µmol/h H2, and a 4 h run. Total product is 200 µmol, or 4.9 mL at 25 °C, which is 9.8 vol% of the headspace and more than twice the lower flammability limit. The same run ends at 2.4 vol% in a 200 mL headspace and 1.2 vol% in a 400 mL one.
That is why 1:1 circulates in the literature: much of it was written for catalysts producing a few µmol per hour, where 1:1 stays near 0.5 vol% all day. Adopt the same ratio with a catalyst ten times faster and the safety argument inverts. State the headspace volume, the liquid volume and the run length, and the reader can check the arithmetic themselves.
For CO2 reduction the balance shifts, because CO2 solubility is 34 mM at 1 atm, forty times that of H2. A liquid-rich reactor keeps more carbon available at the catalyst, and the products of interest (CO, CH4) are formed in µmol quantities that stay far below any flammability limit. Pressurised systems go further and use a small headspace deliberately, since the total pressure is set by the CO2 feed rather than by product accumulation.
For overall water splitting without a sacrificial agent the opposite applies: the stoichiometric H2/O2 mixture is the hazard, and a large headspace with periodic venting, or a gas bag held at ambient pressure, is the safe configuration.
2:1 to 4:1 headspace-to-liquid for hydrogen work, chosen from the expected gas production rather than copied from a paper. Below 1:1 the H2 fraction can exceed the 4 vol% flammability limit within a few hours even at modest rates.
It changes the measured concentration and can change the true rate. A large headspace keeps the H2 partial pressure low and limits back-reaction on Pt; a small one raises the dissolved H2 and can suppress net production by 10 to 30 % in sensitive systems.
Gas accumulates and cannot leave. At 25 degrees Celsius, 50 micromol of H2 plus 25 micromol of O2 adds 1.9 kPa to a 100 mL headspace and 18.6 kPa to a 10 mL one. Both numbers are far below the burst pressure of glass, but both are enough to make a silicone septum leak.
Withdraw the sample and immediately inject an equal volume of argon at the same pressure. Record the dilution factor, or use a gas bag so the total pressure stays at ambient and the correction stays simple.