Hi Dave,

I think that the most important part of your, and any correction, is the demonstration that it works. If you are running carbonates and you are tied to the VPDB scale, you should have access to NBS19 and LSVEC for d13C and NBS19 for d18O. Many in the carbonate community also have NBS18, IAEA-CO-1, IAEA-CO-8, or IAEA-CO-9 (http://ciaaw.org/Carbon.htm). Whatever correction you settle in on, I believe you should analyze NBS19 and LSVEC as the Calibration Materials along with one or more of the above recognized carbonate Reference Materials as unknowns, run your correction, and see if you obtain the accepted value(s). If you obtain the appropriate values, and you precision is acceptable, the correction works.

andy

Andrew Schauer
ISOLAB
Department of Earth and Space Sciences
University of Washington
Seattle, WA 98195

206.543.6327

----- Original Message ----
Sent: Tuesday, June 10, 2008 3:56:34 PM
Subject: [ISOGEOCHEM] Help with Linearity Corrections

I, and maybe others, would greatly appreciate some feedback on correcting
for linearity for Carbon and Oxygen in carbonates.

We have been struggling for about a year now (since beginning collection of
data on a Delta V using CF) on the best way to correct for linearity for C &
O on carbonates…or even if we should.  Recently I was told by several other
machine users that they made no correction for linearity.  Here are my
concerns/questions below:

Given the following: The “acceptable” linearity max limit for carbon, for
instance, is 0.06 per mille/volt.  Say we are successful in keeping sample
sizes in the range that produce peak amplitudes of 1-8 V.  After running 10
internal stds we end up with an average amplitude for every group of sample
gas peaks, or 10 averages ranging from 1V to 8V for our 10 internal stds.
Assume the cross-plot of average amplitude vs. delta 13-C gives a slope of
0.05 per mille/V.

1.    First Question: before getting into correcting actual samples for
linearity, I have a question on correcting internal stds used in a run.
After using their amp vs delta cross plots to get a linearity slope, we
typically use that linearity to correct them around a 5V “ideal” monitoring
gas amplitude for mass=46.  In other words, if our linearity is +0.05 per
mille/V, then the std at 8V is lessened by 0.15 per mille (3V x 0.05) and
the std at 1V is increased in delta value by 0.2 per mille.  This results in
a tighter grouping of stds when we plot the enriched and depleted stds vs
their real values to make an offset correction.  Any criticisms of this
procedure?

2.    Second Question: say we set 5V as that “ideal” monitoring gas pressure
and we have 2 identical samples on the same run.  If one sample has peaks
averaging 8V and a delta 13-C value of -2.0 and the second sample has peak
amplitudes averaging 1V and the 0.05 per mille/V linearity holds true (I
know this is much more ideal than what really happens) then currently we
would use our linearity slope derived from stds to correct the 8V sample
down from -2.0 to -1.85 per mille.  We correct our low amplitude 1V sample
up from around -1.65 to -1.85 per mille.  Again, for the sake of argument
just assuming this perfect linearity.  Surely to be able to report error
less than ~0.1 per mille, we want to make this linearity correction?

3.    Third and last question:  we are using the mass=46 amplitude of internal
stds throughout for this correction.  Should we instead use mass=45
amplitude for Carbon linearity and 46 for oxygen linearity?

Thank you for any and all thoughts on this; perhaps their has been a short
course on this issue of someone has something written up as a procedure
they’d be willing to share?

Thanks to all,

Dave Mrofka
UC Riverside