Laser oxygen isotope analysis of silicate/oxide grain separates: Evidence for a grain size effect?

285 oxygen isotope analyses of grain separates of quartz, biotite, muscovite, aegirine, actinolite, olivine and chromite were carried out by laser fluorination using BrF5. Two of the eleven samples analyzed are oxygen isotope standards (NBS-28 and NBS-30) with internationally agreed delta(18)O-values. The delta(18)O-values of all other samples were previously determined by conventional fluorination. Laser oxygen isotope analyses were performed on chips, calibrated size separates and powdered aliquots of each sample. The results indicate that fine-grained (<250 mu m) and powdered samples of most minerals yield laser delta(18)O-values that are systematically and reproducibly lower than the conventional delta(18)O-values. A continuous series of five size fractions, ranging from >500 to <75 mu m, was analyzed for two quartz samples. The results show a progressive decrease of laser delta(18)O-values with decreasing grain size. The delta(18)O-value of the finest size fraction is, on average, 0.8-1.0 parts per thousand lower than the conventionally measured delta(18)O-value. Also, 22 analyses of quartz standard NBS-28, which grain size is 120-250 mu m, yielded an average laser delta(18)O-value of +8.8 +/- 0.2 parts per thousand, i.e. 0.8 parts per thousand lower than its true value. Similarly, powdered aliquots (grain size <75 mu m) of all minerals but actinolite and aegirine gave laser delta(18)O-values that were 0.5-2.0 parts per thousand lower than conventional values. For all minerals, except muscovite and chromite, laser analyses of size fractions >250 mu m yielded correct delta(18)O-values, suggesting that the grain size of 250 mu m may constitute a threshold. Detailed examination of the data indicates that this unexpected depletion cannot be attributed to contamination during powder, loss of material during fusion (low yields), intragrain heterogeneity of the samples, atmospheric contamination, oxygen isotope fractionation during the fluorination reaction or during O-2-CO2 conversion. In addition, laser beam size, which is shown to have an influence on the laser delta(18)O-values measured for some minerals, is not responsible for the observed depletion. At this time, we do not have a complete understanding of the cause(s) of the observed grain size effect. As a consequence, it is advisable that investigators performed preliminary tests before analyzing fine-grained or powdered minerals by the laser fluorination technique.