I'll get some strange behavior when the Cr is nearing the end of it's
useful life. Some water vapor makes it into the bellows and thus into
the capillary tubing.
hope this helps...
Other stuff we're seeing with our H-device:
1. We're getting up to 400 runs, 0.5uL, per reactor--sometimes more, and
we'll swap out reactor and septa at the same time. Yes, the septa does
stick to everything.
2. Most of the Cr is still Cr, not CrO3, when reactor no longer useful
(longer reaction times won't lengthen reactor life)
3. Like you, we get up to 50 runs per bellows of reference gas. If
bellows compressed too much during run, numbers start to shift a lot
(somewhere below 30% volume)
4. We install our Cr reactor upside down, large opening on top
(suggested by Thermo rep during install). That way we don't have to
worry about Cr fines getting into the valves or bellows.
5. Our readings are done at 3000mV, mass 2, for all runs, and we see
less than 1% memory effect, a 3 to 4 per mil total drift, and a 10 to 12
per mil offset to stds real value with a slope of ca. 0.98. Std.
deviations for multiple runs, after corrections, usually less than 1 per
6. If you're using Isodat2.0, the sample bellows (at least on our 253)
can not always be set to give the correct intensity. Opening up the
tolerance may help (e.g. to 500 mV, which is actually +/-250 mV in the
patch I'm using), but if it doesn't get the correct voltage on the first
try, it will oscillate wildly about the correct value until it times out
(something like 15 minutes). If that happens, that run is shot.
7. One last things, as per discussions on Isogeochem, we've taken to
rinsing our syringe multiple times with water, then acetone in our
case--with the waste vial for the acetone rinses under vacuum to dry the
syringe completely. We have the CTC Leap Tech. PAL.
Geldern, Robert wrote:
>One question concening the drift correction of hydrogen values.
>Since our MS is not equipped with an automatic reference refill system
>I observe a drift during a sequence of 50 samples measured for dD
>using a Finnigan H/Device coupled to a delta S. The drift is the
>result of a fractionation of the reference gas in the standard
>bellow. To correct the data for this drift we carry out repeated
>measurements of the same standard sample during the sequence.
>Theoretically, the drift should follow a Rayleigh fractionation and
>one should observe a (more or less) continuous drop of the measured
>standard sample values. The sequence values can be corrected for the
>drift by fitting a function to the data points of this "drift
>standard". This is exactly what I observe in most of the sequences.
>However, on some days the drift seems not to follow the theoretical
>model. Instrument parameters (filling of the bellow, temperature,
>etc.) are essentially the same as on the other days but in some
>sequences I observe a convex shaped function instead of a concave
>shaped Rayleigh function or a step-like function with a fast drop of
>the values at beginning a stable phase for several hours in the
>middle and a second drop at the end of the sequence. Memory effects
>can not account for the observed trends.
>I expect I am not the only one observing such a phenomena from time
>to time. The question is how you handle this kind of data? The first
>option is to discard the whole sequence. The second would be to fit a
>(polynomial) correction function to the measured values assuming that
>my data points represent the original drift of the gas (even if the
>measured drift differs from theoretical assumptions). The third
>option would be to fit a strict Rayleigh function (or alternatively a
>linear regression) to the data points which results in a very bad
>fitting of the function to the measured values.
>Robert van Geldern
>Leibniz Institute for Applied Geosciences (GGA)
>S3: Geochronology and Isotopehydrology
>Stilleweg 2, 30655 Hannover, Germany
>phone: +49-(0)511-643-2313, fax: +49-(0)511-643-3665
>mailto:[log in to unmask]