Dear Gerald, Gareth and Community

I seriously doubt that the composition of your standard is "drifting."
Rather, I suspect your CO2 working standard is compositionally quite
different (deltas > 10 per mill) from the composition of the CO2 from the
Merck standard.  In our MAT 252, we have found that high-differential
measurements, especially after many runs, tend to "contract" slightly.  We
believe this is due to the gradual buildup of ultratrace H20 on certain
inlet/source surfaces, and subsequent absorption and slow outgassing of CO2
during analysis.  Reproducibility usually remains good.  Running at higher
voltages (i.e. using more gas) alleviates the problem somewhat by
overwhelming the sources of memory.  We've found that heating the source up
to 110 deg. C. after the fact helps only marginally, but it may be
worthwhile to maintain a hot source after cleaning. 

If your sample and working standard are compositionally quite different, try
increasing the time you allow between changeover of sample/working standard
and voltage ratio acquisition.  If it's a memory effect, your deltas should
widen with an increase in allowed time.  We (and others) have seen the
effect on several instruments even after 3-5 minutes.

Mike 

At 04:20 PM 3/1/96 +0000, you wrote:
>
>from: Gerald Ganssen and Gareth Davies
>subject: drifting standards
>
>Dear colleagues,
>we are running a Finnigan Mat 251 mass spectrometer in combination with an
>automated carbonate preparation line (Kiel type, latest Bremen
>modification, individual droplets of acid on samples).
>After having run quite a time with acceptable reproducibility of our
>carbonate standard (100% calcite, Merck) -8.1 per mil and -21.5 per mil del
>18O and del 13C vs. PDB, respectively with a standard deviation of 0.09 and
>0.04, values gradually decreased.
>At high voltages (6-7V signal) the effect is small 0.1 per mil in del 13C
>and 0.2 per mil in del 18O, low signal values (3-4V) decreased to -21.6 and
>-8.7!
>Individual runs even at low voltages still give a good st. dev.of 0.02-0.04
>per mil.
>Heating the source helps for just a few runs (about 50); after having
>cleaned the source the fractionation reoccured after about 200 samples.
>There is no leak in either the inlet system of the mass spec. or the carbo
prep.
>The spectrum does not point to organic compounds from oil around masses 38
>to 42.
>Heating of the capillaries did not set free remarkable amounts of rubbish
>and did not improve the measurements.
>We run at an acid temperature of 70 degrees centigrade and normally use two
>droplets of acid. Adding 20 droplets instead did not increase the effect.
>Has anyone of you experienced an effect like this, detected the reason and
>eliminated it?
>Any suggestions are very welcome.
>
>Kind regards
>
>Gerald Ganssen  Gareth Davies
>
>
>Gerald Ganssen
>Institute of Earth Sciences
>Free University
>De Boelelaan 1085
>1081 HV Amsterdam
>The Netherlands
>tel.: 31 20 4447369 (office)
>      31 20 4447308 (mass-spectrometer lab.)
>fax:  31 20 6462457
>
>
>
>
====================================
Dr. R. Michael Verkouteren
Atmospheric Chemistry Group
B364, Chemistry Building (222)
National Institute of Standards & Technology
Gaithersburg, Maryland   20899   USA
301-975-3933 (telephone)                                      
301-216-1134 (facsimile)                                       
====================================