The Oxygen Isotope Composition of Earth and the Terrestrial Fractionation Line (TFL)

The discovery of non-mass dependent oxygen isotope fractionation effects in terrestrial surface deposits and in trace gas species in the atmosphere has renewed interest in a subject that has been in the province of cosmochemistry for decades. With the discoveries has come the commissioning of new laboratories dedicated to their pursuit. Inter-laboratory comparisons of measurements of both d17O and d18O and the slope of the attendant fractionation line, are imperative to validate rapidly growing data sets from a variety of different laboratories. Because oxygen isotope fractionation is a necessary consequence of the differentiation of planetary bodies, it is not useful to speak of a single, specific oxygen isotope composition for Earth, or any other such body. Planetary processes, however, do produce a diagnostic characteristic that uniquely defines bulk oxygen isotopic composition. A plot of d17O vs. d18O for a given body gives a linear array of data points which is characteristic of the bulk composition of the body. For the Earth the array is termed the Terrestrial Fractionation Line (TFL). Thus, the slope and intercept of a planetary body's oxygen three-isotope fractionation line are definitive. Published values of the slope of the TFL on a plot of d17O vs. d18O range from 0.5164 to 0.5288. The TFL's intercept is defined as zero, relative to VSMOW. Different slopes may be associated with different fractionation mechanisms; in particular, equilibrium vs. kinetic isotope fractionation. Modern analytical techniques should be able to resolve these fractionation mechanisms but the practical capability to do so must be validated. We are conducting an inter-laboratory comparison of silicate mineral samples analyzed at both the Open University (UK) and the Geophysical Laboratory of the Carnegie Institution of Washington. The analyses were performed in both localities by heating samples with a CO2 laser in a reaction chamber filled with BrF5 gas. Two different mass spectrometers were used: a Prism III at the Open University and a Thermo MAT-252 at the Geophysical Lab. Slopes were computed by regression of linearized measured delta values (Miller 2002). Seven samples of quartz extending over a range of d18OVSMOW from +2.40 to +33.28 give a slope of 0.5248 (+/- 0.0005) at Open University. The same samples analyzed at the Geophysical Lab give 0.5281 (+/- 0.0012). Earlier high precision values of the TFL's slope, presented in the same format, include 0.5281 +/- 0.0015 for natural waters (Li and Meijer, 1998) and 0.5263 +/- 0.0008 for the Earth-Moon system (unpublished, Geophysical Lab, CIW). The causes of variations in the TFL slope will be discussed.