Figure 1. Flow diagram and description of the used abbreviations
in the process of quantification of the
δ18OSLAP value with respect to
δ18OVSMOW. δ2H and
δ18O in this figure are expressed on the VSMOW-SLAP
scale.
For every experiment, we started with Antarctic water and made a fresh
portion of SLAP-rep-O. After measuring the isotopic values of Antarctic
water, we calculated how much demineralized Groningen tap water should
be added, in order to mimic δ 18O of SLAP. As
the Antarctic water was isotopically ”lighter” than SLAP, we had to add
approximately 18 g of demineralized Groningen tap water (δ2H = -43.5‰, δ 18O = -6.5‰)
to 1 liter of this Antarctic water. In total, we produced 7 portions of
SLAP-rep-O, which were individually measured on the LGR-LWIA along with
aliquots of SLAP.
The next step in the flow diagram shows the mixing of SLAP-rep-O with
highly enriched 18O water in order to obtain
VSMOW-rep-O. As said before, the most critical part of the whole process
is the characterization of the highly enriched18O-water that is added to the SLAP-O replicate. This18O characterization is done by fitting a QMS spectrum
of the enriched water. The steps we took for a careful determination are
described in this section. We did our utmost to avoid memory effects
from natural and highly enriched 18O water in the QMS
and we investigated the influence of several ionization processes in the
ion source of the QMS on this 18O determination. At
the end, we performed a validation of our QMS method by diluting a
H218O water portion with 1% and 2%
H216O. The results of this validation
by comparing the expected abundances based on weights with the measured
abundances and the influence from several ionization processes are
described in the Results section.
To get rid of possible memory effects, the ion source of the QMS was
pumped for more than 48 hours at high vacuum, before measuring highly
enriched 18O water (background pressure was 1.5 x
10-6 Pa). The mass spectrum with this “clean” source
was considered as a background signal and was subtracted from the
spectrum of the enriched water. For this background signal, it hardly
made any difference if the previous injection, before the 48 hours of
pumping, was a water with natural abundances or a water with enriched18O.
Water is very “sticky” and adsorbs to the walls of the injector, dead
volumes and the ion source of the QMS. Therefore, the analyzing QMS
setup needs to be saturated with highly enriched 18O
water in order to reduce memory effects. Thus, more than 20 sequential
identical sample injections were required to reach an equilibrium state.
For every injection, 25 µl water was injected and a scan of m/z 1 to 41
was performed. The measurement pressure was at 2.5 x
10-3 Pa. The QMS exclusively measured highly enriched18O water for several months in a row.
Water molecules in the ion source of the QMS ionize, break and recombine
to produce a combination of peaks corresponding to
[H]+, [H2]+,
[O]+, [OH]+,
[H2O]+,
[H3O]+ and
[O2]+ ions. All of these ions
contain the two different H-isotopes and three different O-isotopes. In
Figure 2, a typical QMS spectrum of a highly enriched18O water is shown. All the main Oxygen-bearing
fragments together produce ion signals from m/z 16 to 24. In the
supplementary material, Table 5 shows a highly enriched18O water (water portion D from Cortec) with the
various isotopologues and fragments for this range of m/z values.
<Figure 2>