Changing Elution Order of Peaks on FID

[la_donz27]

Hi, all;

Early last December I posted a problem with late-eluting peaks switching order on a GC we were setting up. We finally resolved that by swapping injectors as a last resort.

I am setting up another 7890 GC, and am seeing the same swap in elution order of the last six peaks. The configuration is simple, really - Hydrogen carrier, Split/Splitless injector, HP-1 20m x 0.18mm ID column, into an FID. Gas hydrocarbon sample is introduced into the injector via a 6-port Gas Sampling Valve. An easy analysis I've set up/run a hundred times.

I get great, fast chromatography for all compounds at the front/middle of the chromatogram. However, by Ethylbenzene, sensitivity drops to very poor, m/p-xylene looks like Bart Simpson, and the last several peaks elute in the wrong order.

This is the "bad" chromatogram:


this is what it looks like on the eight working GCs:


The net amount of sample on column is approximately the same for both chromatograms; I've tried fiddling with the split ratio on the new system, but it doesn't increase my back-end sensitivity at all, just overwhelms the Methane/Ethane and causes them to co-elute.

I tried different liners, glass wool, no glass wool, re-installing column into injector, tightening the injector weldment nut, leak checking (ok), flow checking (ok), and even swapped out the injector/EPC entire unit, since that worked last time . Nothing has made things better (maybe even worse sensitivity with the new SSI).

I've ordered another column from a different manufacturer, but not very hopeful that will make much difference. Has anyone else ever run into this strange problem? Any ideas what else I can try to resolve this?

I'd really appreciate any suggestions, I can't think of much else to try.

[aidnai]

I assume you have already eliminated differences between working GCs and non-working, such as different column or injection type.

You mentioned you have a GSV but also liners, i.e. injection port? This is a nice flexible setup but does perhaps offer more opportunities for inlet discrimination -- also, is your sampling valve hot enough (verified with thermometer if possible)?

Beyond that, I don't have any good guesses, but I suppose if I were in the same situation I would be substituting 1 part from a working GC in at a time -- probably starting with column and then injection zone parts.

Good luck!

[GasMan]

You seem concerned about sensitivity, but what about the 100% difference in retention times.

Gasman

[Peter Apps]

Discrimination between low and high boilers and peak distortions are clear red flags for a cool spot somewhere. Also possible, but much less likely is transfer by diffusion rather than bulk flow.

If you have an FID (i.e. non-selective detector) how do you know that the order of elution is changed ? Are you going only by relative peak size ?

Peter

[la_donz27]

Good Morning;

Gasman – Good catch! I didn't mention in my original post that we are using a “fast gc” column with the same phase & phase ratio (0.40um df). This column and standard have been verified as “identical” in resolution of this gas standard on other GCs, just gives us much faster run times and sharper peaks. That is why the RTs are vastly different. Otherwise, analyses have an identical method translation.

Peter & Aidnai – I think you're both on to something with the cold spot in the valve. We had to modify the valve heater on this instrument as we have three channels, two injectors, & three valves, so only have enough heated zones for one valve plate, not two. We had to make our own valve plate double the usual size, so it heats very slowly. Maybe there are cold spots in the valve oven, as the 6-port valve is the furthest away from the heat source. I will stuff some insulation on top of the valve and see if I can measure the temperature on that side of the valve box with a temperature probe.

Thanks for the ideas!