GC–ICPMS data were exported from the Neptune Evaluation
software into Isodat v. 3.1 (Thermo) for processing. Results are
reported in the conventional d34S notation as permil (‰) deviation
from the VCDT standard. Accurate and precise d34S values were
obtained for analytes producing peak areas of at least 1 Vs, representing
ca. 50 pmol of analyte on-column. Unlike for other light
isotopes (H, C, N), we have not observed chromatographic separation
of sulfur isotopologues, meaning that 34S/32S ratio values are
invariant across even large chromatographic peaks. This is consistent
with previous observations (Amrani et al., 2009, 2012). Poor
chromatographic peak shapes and/or coelution were the dominant
source of measurement uncertainty, particularly during early analyses.
Results were sometimes moderately sensitive to the manually
selected background interval due to the contribution of
unresolved OSCs to the m/z 34/32 ratio of background signals,
and peaks for which d34S values varied by > 1.5‰ for alternative
baseline definitions were discarded. In summary, we conservatively
estimate that the data here have uncertainties of 1.0‰ for
F1 and F2 fractions and 1.5‰ for F3. Uncertainty is greater for F3
because the abundant OSCs and other compounds in this fraction
generated complex chromatograms with poorer baseline resolution.
Although the analytical uncertainties are large compared with
those typically achieved using conventional sulfur isotope analysis,
they are nevertheless much smaller than the scale of isotopic variability
observed in the Cariaco sulfur record and therefore still
permit meaningful interpretation.
3.2. Analysis of bulk organic fractions usi