Run the experiment twice, once with 13CO2 at 99 atom% and once with ordinary 12CO2. A genuine CO2-derived product appears at m/z 29 (13CO) or m/z 17 (13CH4) at more than ten times the 1.07 % natural-abundance baseline, and liquid products show 1JCH coupling of 140–220 Hz. Report atom% enrichment, not just the presence of a peak.
Carbon-13 is already in ordinary CO2 at 1.07 % natural abundance. A run generating 200 µmol of CO from unlabelled CO2 therefore contains about 2.1 µmol of 13CO, and m/z 29 sits near 1 % of m/z 28. That is the reviewer's yardstick: a 13C signal below roughly 5 % of the 12C signal proves nothing, because natural abundance plus fractionation can reach that range, while a 99 atom% cylinder pushes the ratio above 100:1. Report both runs.
Electron-impact ionisation yields a molecular ion plus fragments, and the fragment pattern is where most claims fail. Pick an abundant ion free of isobaric overlap, then confirm by retention time.
| Labelled product | Ion to monitor | What it establishes | Interference to rule out |
|---|---|---|---|
| 13CO | m/z 29 (M+) | Carbon in CO came from the CO2 pool | CHO+ fragment from methanol or formaldehyde at the same m/z |
| 13CH4 | m/z 17 (M+) | Methane carbon came from CO2 | OH+ from residual water vapour, which also gives m/z 17 and 18 |
| 13CH3OH | m/z 33 (M+) | Methanol carbon came from CO2 | None significant; confirm by retention time |
| H13COOH | m/z 47 (M+) | Formate carbon came from CO2 | Poor volatility; better done by NMR or HPLC-MS |
| 13CO2 itself | m/z 45 (M+) | Verifies cylinder and headspace composition | None; check atom% before each campaign |
m/z 28 is unusable for CO here because leaked nitrogen sits on the same mass, so the shift to 29 is a benefit. Dry the loop before injection: water vapour contributes m/z 17 and fabricates a 13CH4 peak in scan mode.
Mass spectrometers do not respond identically to 12CO and 13CO because fragmentation shifts with the heavier nucleus; differences of several percent are normal. Calibrate with a certified 13CO standard in helium and fit in nanomoles. At 101.3 kPa and 298 K a 1 mL loop holds 40.9 µmol, so a 1 % standard delivers 409 nmol per injection. Thermal conductivity detection is more forgiving, with response within a few percent, so a TCD curve transfers if you accept that error. Then state enrichment explicitly, atom% = n(13C) / [n(13C) + n(12C)] × 100.
Formate and methanol do not survive a hot injection port reliably, and their molecular ions overlap solvent fragments. 13C NMR settles it because a labelled carbon couples to its protons: formate appears near 171 ppm and methanol near 49.5 ppm, each a doublet with 1JCH of 140–220 Hz. Natural abundance is far too dilute to show that coupling, so the splitting is direct evidence of enrichment; an 1H–13C HSQC cross-peak makes the same point as a publishable figure. A plain singlet at 171 ppm is not proof: adventitious formate from water or glassware lands on exactly that shift.
A 300 mL reactor with 200 mL headspace holds about 8.2 mmol per fill, and a 1 L cylinder at standard conditions holds roughly 41 mmol, so a small cylinder covers many runs. Budget three to five fill-and-purge cycles per experiment to dilute residual 12CO2 below 1 %.
Yes for aqueous systems, and it is often cleaner because it sets the DIC pool directly. The trade-off is that the carbon source becomes DIC rather than CO2 gas, so report pH and speciation: below pH 6.3 the pool is mostly CO2, above pH 10.3 mostly carbonate.
Three causes, in order of likelihood: residual 12CO2 from incomplete purging, 12C carbonate in the electrolyte, or oxidation of a 12C sacrificial agent. A purge-to-baseline test over three cycles, checking m/z 44 each time, separates the first from the other two.
No. Methanol and formaldehyde both yield CHO+ at m/z 29, and a leak admits nitrogen at m/z 28 that tails into 29. You need retention time plus a 12C reference run before the mass means anything.
