The absence of an agreed-upon δ17O value for the primary reference water SLAP leads to significant discrepancies in the reported values of δ17O and the parameter 17Oexcess. The accuracy of δ17O and 17Oexcess values is significantly improved if the measurements are normalized using a two-point calibration, following the convention for δ2H and δ18O values.
New measurements of the δ17O values of SLAP2 and GISP are presented and compared with published data. Water samples were fluorinated with CoF3. Helium carried the O2 product to a 5A (4.2 to 4.4 Å) molecular sieve trap submerged in liquid nitrogen. The O2 sample was introduced into a dual-inlet ThermoFinnigan MAT 253 isotope ratio mass spectrometer for measurement of m/z 32, 33, and 34. The δ18O and δ17O values were calculated after 90 comparisons with an O2 reference gas.
We propose that the accepted δ17O value of SLAP be defined in terms of δ18O = −55.5 ‰ and 17Oexcess = 0, yielding a δ17O value of approximately −29.6968 ‰. Using this definition for SLAP and the recommended normalization procedure, the δ17O value of GISP is −13.16 ± 0.05 ‰ and the 17Oexcess value of GISP is 22 ± 11 per meg. Correcting previous published values of GISP δ17O to both VSMOW and SLAP improves the inter-laboratory precision by about 10 per meg.
The data generated here and compiled from previous studies provide a substantial volume of evidence to evaluate the various normalization techniques currently used for triple oxygen isotope measurements. We recommend that reported δ17O and 17Oexcess values be normalized to the VSMOW-SLAP scale, using a definition of SLAP such that its 17Oexcess is exactly zero. Copyright © 2013 John Wiley & Sons, Ltd.