How 17O excess in clumped isotope reference-frame materials and ETH standards affects reconstructed temperature

Abstract Carbonate clumped isotope measurements (∆47) have been widely applied as a paleothermometer, but various sources of error typically limit their application to anomalies greater than about ±3 °C. Calculation of ∆47 requires a correction for mass interference from 17O, which traditionally assumes all analytes follow a linear relationship between 17O/16O and 18O/16O with a slope of 0.528 and a 17O excess (∆17O) of zero. Here, we evaluate these assumptions by measuring ∆17O in ∆47 reference-frame gases and waters, our mass spectrometer working gas, and ETH standards 1, 2, 3 and 4. We systematically evaluate how the ∆17O of these materials influence ∆47 values, and the magnitude of ∆47 error introduced by assuming a ∆17O of zero. We find most ∆47 reference-frame materials and ETH standards have negative ∆17O values, ranging from −300 to +38 permeg, which can largely be explained by equilibration with surface water at Earth surface temperature. Exceptions include CO2 equilibrated with evaporatively enriched water, which has a more negative ∆17O than other reference-frame materials, and CO2 derived from fossil fuel combustion, which has a less negative ∆17O and greater variance. CO2 heated to 1000 °C in quartz glass tubes shows a small, 5–10 permeg, increase in ∆17O that may reflect contamination in the quartz. CO2 evolved from ETH standards exhibits mean ∆17O values of −151 to −123 permeg, similar to most natural carbonates, but ETH 2 and 4 have greater variance. Our results show that assuming ∆17O = 0 can overestimate or underestimate measured ∆47 values, with the direction of change dependent on whether sample ∆17O falls above or below the ∆17O of the working gas. Similarly, assuming ∆17O = 0 can also overestimate or underestimate ∆47 in the carbon dioxide equilibration scale (CDES), with the direction of change dependent on whether sample ∆17O falls above or below reference-frame materials. The magnitude of this effect is significant, and is equivalent to an error on reconstructed temperature of 0.6–1.7 °C for Earth surface conditions and 1.9–5.8 °C in the shallow crust. While analytical and calibration errors are also important, the effect of ∆17O needs to be considered for ∆47 thermometry to achieve its highest possible accuracy. The same is true for the recently introduced dual clumped isotope thermometer. We provide suggested laboratory protocols that best account for ∆17O when measuring ∆47.