Measurement of 18O18O and 17O18O in atmospheric O2 using the 253 Ultra mass spectrometer and applications to stratospheric and tropospheric air samples

Rationale The doubly substituted isotopologues (e.g., 18O18O, 17O18O) in atmospheric O2 are potential tracers for ozone photochemistry and atmospheric temperatures. Their low abundances and isobaric interference are the major analytical challenges. The 253 Ultra high‐resolution stable isotope ratio mass spectrometer is suitable for resolving isobaric interferences. Methods O2 from air is purified using gas chromatography on a packed column filled with molecular sieve 5 Å and cooled to −78°C. The δ17O, δ18O, Δ17O, Δ35 and Δ36 values are measured on the extracted O2 with the 253 Ultra at medium mass resolution (M/ΔM ~10000) using Faraday detectors for the singly substituted isotopologues and ion counters for the doubly substituted isotopologues. Results Interferences from isobars, mainly 35Cl for 17O18O and H35Cl and 36Ar for 18O18O, are sufficiently resolved to enable high‐precision determination of Δ35 and Δ36. The Δ35 and Δ36 values of O2 after photochemical isotope equilibration at −63°C and heating to 850°C agree with the theoretical prediction. The stratospheric Δ35 and Δ36 values are close to isotopic equilibrium at the ambient temperatures. However, the values for tropospheric O2 differ from those expected at equilibrium. Conclusions The 253 Ultra allows interference‐free clumped isotope measurements of O2 at medium mass resolution. The Δ35 and Δ36 signatures in atmospheric O2 are mainly governed by O3 photochemistry, temperature and atmospheric transport. Tropospheric O2 is isotopically well mixed and retains a significant stratospheric signature.

also 17 O 17 O, also called clumped isotopes, can be useful to independently constrain the cycling of oxygen in the Earth's atmosphere. 7 Clumped isotope signatures in O 2 , expressed as Δ 35  Interestingly, the biological recycling of O 2 , which is of primary importance for its bulk isotopic composition, has a negligible effect on the clumped isotopic composition because of the slow resetting time scale compared with photochemical isotope exchange. 1,7 Here we report a method for measuring clumped and bulk stable isotope ratios in atmospheric O 2 using the 253 Ultra mass spectrometer. After establishing and validating the purification and the mass spectrometric methodology, an empirical transfer function was developed to convert the measured isotope compositions into the absolute reference frame (ARF) by isotopically equilibrating O 2 at low and high temperatures. We also report Δ 35 and Δ 36 values measured in stratospheric and tropospheric air O 2 samples. In addition, we discuss the challenges and possibility of measuring the rarest isotopologue, 17 O 17 O, of O 2 using the 253 Ultra.

| Conventional and clumped isotopes in O 2
The conventional singly substituted isotopic composition of a gas is characterized by the delta value, i.e., δ 17 O = ( 17 R sam / 17 R std -1) and  13 The doubly substituted isotopologues of O 2 , also called clumped isotopes, are 18

| Measurement of O 2 clumped isotopes with the 253 Ultra
Isotopic measurements including the clumped isotopes (Δ 35 and Δ 36 ) were carried out with the 253 Ultra isotope ratio mass spectrometer at Utrecht University, which is one of the first instruments produced by Thermo Fisher Scientific on the basis of the prototype described in Eiler et al. 16 It is a double-focusing mass spectrometer with an electrostatic analyzer followed by a magnetic sector. It can be operated at three different mass resolutions, set by one of the three slits between the source and the electrostatic analyzer.

| Preparation of equilibrated gases for clumped isotope calibration
We

| Preparation of O 2 for testing isotopic reordering
There is a possibility of isotopic reordering in O 2 during sample storage, purification (e.g. GC column) and analysis in the mass spectrometer. 7 To quantify potential reordering effects, we prepared  The separation between the Ar and O 2 peaks is 5 min when the GC column is kept at −77.8°C. Under these conditions N 2 is trapped on the GC column. To release the N 2 and observe the N 2 peak on the chromatogram, the GC column was taken out of the dry ice/ethanol bath when O 2 collection was finished, and kept at 25°C (indicated by the arrows below the detector trace) FIGURE 1 O 2 purification system. An air sample or gas mixture is admitted to a sample loop (10-mL volume) via a six-port valve. O 2 is separated from the other air constituents using a packed molecular sieve 5 Å (length: 3.05 m, OD: 1/8 inch, ID: 2 mm, Molsieve 5 Å) column cooled to dry-ice temperature (−77.8°C). The effluent from the GC column is monitored using a thermal conductivity detector (TCD, see Figure 2) and the sample is collected on silica gel at liquid nitrogen temperature (−196°C)

| Tropospheric and stratospheric air sampling and measurements
To determine the clumped isotope composition in lower tropospheric air O 2 , and to check the reproducibility of measurements, we have  Table 1 presents the results from the zero-enrichment measurements. is the limiting factor here. A slightly positive value is observed for δ 36 , but it is within the error associated with the individual measurements. Therefore, a zero-enrichment correction is not applied for the sample measurements.

| Zero enrichment and counting statistics
The measured errors for all the isotopic ratios including the δ 35 and δ 36 values are compared with the errors expected from counting statistics and found to be similar (EECS, see Table 1 and supporting information). This proves that the 253 Ultra is very stable over the 7-h duration of these measurements. The variation in the EECS is due to different signal strength and measurement duration. As the measurement uncertainty closely follows the counting statistics, the

| Ar and other isobaric interference
It is difficult to achieve full separation of Ar from O 2 by passing the gases through a GC column. Some traces of Ar from the tail of the The mass difference between these two isobars is 0.0308 u and the resolving power required to separate their peaks is~1160, which is easily achieved with the 253 Ultra at medium mass resolution.  Figure 3D) on the low-mass side of the 17 Figure 3D). However, for the Ar-O 2 mixture ( Figure 3B), it is barely visible because it is superimposed on (and partly contributes to) the tilted slope of the 36 Ar peak. Note that mass interferences conceptually cause "step changes" when they enter and exit the detector, and not tilted peak tops. The interferences from these isobars can easily be avoided in the 253 Ultra by measuring the isotopologue ratios at the right position of the peaks as the mass resolving power used for the present configuration (~10 000) is much higher than that required to resolve most of these isobars (e.g., 1664 for resolving 18 Table 2 shows the isotopic ratios measured on pure O 2 and the Ar-O 2 mixture at different positions of the peaks, as indicated in Figure 3B. A clear 36 Ar interference to 18 O 18 O is observed when measurements are made near the 36 Ar peak, i.e., position 3 in Figure 3B and last line in  Figure 3B).

| Isotopic reordering in the GC column and the source of the mass spectrometer
To test isotopic reordering in the GC column and in the source of the mass spectrometer, isotopically spiked O 2 (Δ 36~6 84‰ and Δ 35~3 1‰) is used ( Table 3). The spiked O 2 sample is analyzed after preparation and then again after 4 days of storage in a stainless-steel canister of the same type as used for the stratospheric air samples presented below. The results agree within the analytical errors, suggesting that isotopic reordering in this storage canister is insignificant (Table 3).
To investigate reordering in the purification system (in particular the GC column), an aliquot of the isotopically spiked O 2 is mixed with pure helium to prepare an O 2 concentration of~20%, similar to atmospheric O 2 , and the mixture is passed through the purification system following the same procedure as for the samples. The measured isotope ratios before and after passing through the purification system are presented in Table 3. From the changes in difference between the spiked and unspiked O 2 is 0.02‰ (Table 3) which is similar to the analytical error of 0.02‰ (   Figure 3. The 36 Ar isobaric interference for Δ 36 is very prominent when the measurement is made near the right edge of the Ar peak (position 3 in Figure 3)  12,19 in Figure 4 (the numerical values are provided in Table S1, supporting information). The linear fit to the data presented in Figure 4 provides For the samples analyzed, the Δ 35 and Δ 36 values before and after application of the transfer function are provided in Table S2 (supporting information

| Dependence of clumped isotope ratio (Δ) on bulk isotope ratio (δ value)
Clumped isotope measurements with isotope ratio mass spectrometers may exhibit a dependence of the clumped isotope ratio Δ on the bulk isotopic signature δ. A corresponding correction is found to be necessary for most of the CO 2 clumped isotope measurements using low-resolution isotope ratio mass spectrometers. 15

| Possibility of measuring 17 O 17 O in the 253 Ultra
We also investigated the possibility of measuring the rarest isotopologue 17

| Stratospheric and tropospheric clumped isotope ratios
The Δ 35 and Δ 36 values along with the conventional isotope ratios measured in the stratospheric and upper tropospheric air O 2 are presented in     Figure 5B shows the difference between the measured and the calculated equilibrium smaller. Another potential point of concern is that the effective temperature, however determined, is an average over very different temperature conditions in the reactor, changing from hot near the illuminated Suprasil finger to cold at the outer surface. Thus, reactor geometry and illumination conditions may also play a role. Although