Thanks for your many suggestions. Some may hit the mark, others don't.
Perhaps I didn't stress that zero's all valves open are fine. The ratio
traces during measurement are absolutley flat with no stepping and internal
precisions of better than +/-0.01 (2 sigma (n-1)). There is no sign of a
leak to air looking at both 32 and 40 peaks, even after individual sections
of the inlet, capillaries, change over valves have been isolated and left
for a 72 hour period. The sytem has been extensively baked as well.
The drift only appears when running a series of zero tests with the
reference and sample sides isolated. The experiment is simple. Balance the
major beam signals, close sample and reference bellows then measure the
sample wrt to the reference. Perfect traces, flat, no zig-zagging (seen with
water contaminated samples). Immediately after the run, open sample and
reference bellows re balnace the beams. Close the bellows and remeasure.
Between single measurements that are not separated by more than 15 minutes
there is a drift of about 0.08 per mille, i.e. 4 per mille over a repeat of
48 measurements. During this time the bellows hardly change their volume,
thus I don't think its a pressure dependent exchange between water and CO2.
If you repeat this test, but instead use either a glass flask to store the
reference gas rather than the bellows, or the volume in the vacuum line
between sample and reference sides and balance sample against reference
using the sample bellows. Then hey presto perfect results. Absolutely no
drift between sample and reference!
Ann-Lise Norman's suggestion of ethanol is interesting and one which we may
come back to. A component - the connecting tube between reference inlet
block and reference bellows was recently replaced and I need to check on the
manufacturers cleaning arrangements.
I'll keep everyone informed of progress. It's re-assuring that others have
seen similar phenomena, but not so rer-assuring that the cause is truly
elusive. If its a leak it is very, very small..yet the effect is so
Many thanks for your efforts,
Paul F. Dennis
Head of Stable Isotope Laboratory
School of Environmental Sciences
University of East Anglia
NORWICH NR4 7TJ
Tel: 01603 593105
Fax: 01603 507719
email: [log in to unmask]
From: Attila Demeny <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>
Date: Friday, November 19, 1999 07:32
>I had a similar problem with our delta S. I drift appeared after about
>10 measurements and the composition changed significantly if the gas was
>stored in the bellow. I don't know the exact reason, but after changing
>the capillary and baking the inlet system for some days (that's a
>routine with new capillaries) with H2 gas "rinsing" the problem
>disappeared. I would suspect that there was a contaminant. Now I can use
>a batch of working standard until it's amount gets really too low and
>can leave it in the bellow overnight without any change.
>Dr. Attila Demeny
>Laboratory for Geochemical Research
>Hungarian Academy for Sciences
>Budapest, Budaorsi ut 45,
>e-mail: [log in to unmask]