Dear Bev,


Have a look at the backflush valve that mounted in the top of the GC. Check 
if it open and closes properly when actuated. If not, the o-ring sealing 
the chamber accommodating the spring has hardened and causes compressed air 
to escape thus compromising the action of this valve.

Should the valve as such operate properly, check the valve seat and the 
teflon o-ring around the valve needle.

What you describe sounds to me like the backflush valve (GC side) doesn't 
close properly and, hence introducing a split between column and combustion 
interface.

I see you checked the flow through the backflush exit capillary in 'open' 
mode. Have a look if gas escapes through this capillary when the valve is 
supposed to be 'closed'.

I also note you inject in splitless mode. Given what happened to me on one 
of our Trace GC's, I have to ask 'are you sure'? Recently, we had an 
electronic problem with the valves on the injector controlling split and 
septum purge. So, the system may be telling you the split valve is closed 
where in fact it isn't.

Hope this helps.


Cheers,

Wolfram


On Aug 1 2005, Beverly J. Johnson wrote:

> Greetings All,
> 
> We have a GC (Trace/PAL)- irms (Delta Plus Advantage) with a GC III 
> combustion interface and have been experiencing three problems with d13C 
> measurements of specific compounds over the last few months.
> 
> 1. Selective loss of the low molecular weight compounds (n-alkanes < 
> C24) relative to the higher molecular weight compounds (n-alks> C26). 
> This issue has become increasingly problematic over the last few weeks, 
> and does NOT appear to be related to any dramatic changes in running 
> conditions or instrument mainetenance. Samples are injected in splitless 
> mode. Septa are changed every 20 injections, the inlet has been recently 
> re-plumbed, and flow rates are as expected through the backflush valve 
> (1.5 mL/minute in backflush mode) and the column (1 mL/min). The 
> backflush vent capillary has been replaced and is the recommended 
> dimension. The sample itself is fine (all peaks are as expected and 
> approximately equal in height) as evidenced by analysis on a different 
> system (GC-FID). Furthermore, there do not appear to be any leaks in the 
> system as checked by monitoring mass 40 under a stream of Ar.
> 
> 2. Isotopically enriched d13C values of the sample peaks (2-3 permil too 
> enriched, in fact). 
> 
> 2.  A non-systematic fractionation of +/- 1-2 permil between the first 
> set of CO2 ref peaks, and the last set of CO2 ref peaks on each GC run.
> 
> Any ideas would be greatly appreciated.
> 
> Thanks,
> 
> Bev
> 
>