Andreas:

I routinely use the micro-inlet cold finger on our Isocarb-Optima
combination  for very small carbonates. I have found four tricks that
greatly improve our precision:

(1) The water trap must be optimally cold (a good Cryocool should get down
to less than -90 C). I have found that at -86 there is some loss of
precision, presumably due to inefficiency of the water trap (ours is not
packed with beads).

(2) Use a slightly longer reaction time. I use at least 5 minutes for cold
finger carbonate runs, in which adjacent samples have relatively close
delta values. Sometimes I see a memory effect when there are larger (>1 per
mil) differences between adjacent samples, especially when the samples are
different size (see #4 below).

(3) Make sure the acid is very dry. We load our acid (density about 1.92)
and samples and let the Isocarb pump over it for at least 7 hours before
starting the autorun.

(4) Try to make the samples approximately the same size. A mixed run of
cold fingers and non-cold fingers, and even different sized cold fingers,
tends to run very poorly.

Using these tricks, I typically get precisions of about 0.07 to 0.09 per
mil for d18O and slightly lower for d13C. This precision is one standard
deviation of results for 9 Carrera marbles included in a run of up to 51
carbonates. In general, the precision is slightly worse than that for
typical big carbonates.

For the woofing problem, as others have said pump slowly for a while
through the capillary before opening the main valve to the Isocarb roughing
pump. We sometimes use small glass beads on top of our stainless steel
boats to prevent woofing of very large samples that nearly fill the boats.
But with typical carbonates we never have a problem if you first pump for
about 5 minutes through the capillary.

Hope this helps.
-Dan

Dan Bryant
[log in to unmask]
Tel. (609) 258-2612; Fax (609) 258-1274
Department of Geoscience
Princeton University
Princeton, NJ 08544