>What happens when CO2 spills over to the next sample?
>1st a standard peak is detected (I assume a run with reference gas
>injection) that has a proprietary background declining over the peak.
>2nd follows the sample peak, again with its own, this time smaller
I should have remarked on our system. We are running C&N on a Europa
Geo 20/20 and we are not using reference gas injection. We normally
calibrate to a reference material every 6th position in a run. We
did test runs to describe the effects I reported yesterday by running
sets of three samples with the first being a blank, the second being
a sample material loaded at various masses, and the third being the
same sample material loaded at a constant and relatively low mass.
References were also preceded by a blank during this run.
Only if the third position was a small mass of a sample with low
nitrogen content (say 1 or 2 mg of pine leaf with 1% total N) did we
see the described memory effect attaining any substantial importance.
So the practical conclusion was, as others have discussed, to first
load a sufficient amount of material to have a decent sized nitrogen
peak for ALL samples. We do not vary weights sample by sample
however, as some labs do if they have prior knowledge of nitrogen
contents. We find that the basic linearity of nitrogen once in a
range of decent peak magnitude is very good (better than carbon for
us), and that having adjacent samples of unequal size can cause more
harm than good in a dual measurement of C&N.
We actually stumbled on this behavior while trying to do a simple
linearity test varying sample size in a sequentially increasing and
repeated series. The smallest weight was often preceded by the
largest, and showed a noticeable deviation from all the other
samples. This often proved to be more strongly related to the
imbalance of the 'carbon' interference at the point where the series
started over than due to an inherent linearity response in nitrogen.
As mentioned yesterday this effect is seen primarily in 15N, and is
almost undetectable in 13C or % content of either C or N. We do see
inherent linearity effects in % contents as described by others this
past week.
>A quick comment to Jonathans remarks:
>Jonathan wrote
>'As noted by Willi, CO+ is +500 per mil when measured with respect to a
>nitrogen
>reference, and yet the apparent memory effect is a negative shift in
>15N'
>What happens when CO2 spills over to the next sample?
>1st a standard peak is detected (I assume a run with reference gas
>injection) that has a proprietary background declining over the peak.
>2nd follows the sample peak, again with its own, this time smaller
>background. It still is declining and the sample peak is wider than the
>reference injection peak. Thus, from the reference a portion of the
>nitrogen ion current is subtracted with a CO signature background making
>the remaining isotope value too negative. The same happens to the sample
>peak. Because the relative ramounts subtracted depend on the steepness
>of the CO+ decline and on the relative width of the peaks, the resulting
>effect can have either sign. The next sample or standard will have a
>different spill over. It is the ratio of the two that determines the
>final sign. Hence, a memory with a highly positive isotope signature can
>give rise to a negative shift in 15N.
>Willi
>
>--
>‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰
>Willi A. Brand, Stable Isotope Laboratory [log in to unmask]
>Max-Planck-Institute for Biogeochemistry
>Carl-Zeiss-Promenade 10, 07745 Jena, Germany Tel +49-3641-643718
>P.O.Box 100164, 07701 Jena, Germany Fax: +49-3641-643710
>‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰‰
Jonathan Comstock
Associate Research Scientist, BTI
http://bti.cornell.edu/btihome.html
Director, Cornell University and Boyce Thompson Institute Stable
Isotope Laboratory
http://bti.cornell.edu/CoBSIL/CoBSILHOME.HTML
Boyce Thompson Institute
Tower Road
Ithaca, NY 14853
USA
phone:607-254-1214
fax:607-254-1242
|
