On 22.07.2010 18:31, Daniel H Doctor wrote:
>  Finally, there are those among us who run hundreds
> if not thousands of samples with unknown yields in an automated CF mode,
> creating a very complex situation when dealing with the simultaneous
> correction of both linearity and drift, over and above considering
> chromatography and possible memory effects.

We run a couple of standards in every seqence to deal with this. (1) a 
set of 5 to 10 standards with increasing signal for correction of the 
linearity of the source. (2) a drift standard measured at every 8th oder 
10th position with identical peak heights (or areas) (3) 2 reference 
materials (or lab standards) for normalization (4) a quality control 
standard treated as a sample. Memory effects can be either removed by 
mathematical calculations (by using the results of multiple analysis of 
the 2 reference materials) or by ignoring the first (or more) analysis 
after a large change in delta values. This depends on your peripheral 
(e.g. Gasbench memory effect is small, H-Device is large).

Make sure to calculate the corrections in a logical sense of order, 
otherwise you will not know what effect has already been corrected for: 
(1) linearity (2) drift (3) normalization (e.g. to VSMOW) (4) check your 
control standard if it is within your analytical limit. No. (4) will 
also give you the standard deviation over time and therfore your 
external (includes the effect of sample preparation and so on) 

> My question is this: would it be better to calculate and correct isotope
> ratios for samples run in CF mode using the peak height (or maximum
> voltage response) rather than the peak area? Of course, this would only
> apply to the standards and the samples in the run that result in
> chromatographic peaks, not the reference gas peaks, which are flat-topped.

Wolfram already explained why the isotope ratio should be calculated 
from the area.
However, for any correction of the final delta isotope value (e.g. 
linearity: delta value vs. signal) we use the intensity since the 
correlation area:height usually is 1:1 in 99% of the cases. For other 
things, eg. concentrations it should be better to use the area instead 
of the maximum signal height to account for possible GC effects.

Isotope people who started with dual inlet analysis tend to think in 
"intensities" (mV) - and we still do in the CF era  whereas people from 
organic geochemistry using GC/MS analysis always think in areas. :-)


Dr. Robert van Geldern
GeoZentrum Nordbayern / Applied Geosciences
University of Erlangen-Nuremberg
Schlossgarten 5
91054 Erlangen, Germany

mailto:[log in to unmask]
fon: +49(0)9131-85-24780 (office)
fax: +49(0)9131-85-29294
Room: O 0.106 (office)