Issues with baseline on Varian-450 GC

[JWF239]

Hello all,

I am using a varian-450 GC with system control version 6.9.3. The channel I am having issues with uses a PDHID.

I have never had any issues with baseline until a few days ago.



Both spectra are runs of the same gas with the same method. As you can see the spectrum from a few weeks ago has a great background and distinct peaks while the newer scan has an absolutely terrible S/N ratio. I have been told that it looks like the injector is not injecting the gas but you can clearly see the same peaks in both spectra.

I was told to turn off auto zero and then clear auto zero but I am not sure how to do this.

Does anyone have an idea what the problem may be and how to auto zero if that is what looks like needs to be done?

[JWF239]

So I found the auto zero function.

I turned it off and then cleared autozero and my signal in that detector is around 5 V. I was told it should be less than 2. Any ideas what may be the issue?

[rb6banjo]

5 V is a lot. 2 V is a lot. Most of the time, detector signals are in the millivolt range. Are you sure it's not 5 volt full-scale?

Are those chromatograms on the same scale? Sometimes, overlay features autoscale both chromatograms - independent of each other. It almost looks to me like you've lost your sensitivity. Are you able to inject more sample? If yes, up your injection volume and see if the H2, O2 and N2 peaks get larger.

My HDPID is from Valco. One time, I saw something similar to this. The column (molecular sieve) must have had a stress in it. Unbenknownst to me, the column broke in the oven. I reinstalled the column up into the detector and the problem vanished.

[JWF239]

Not really sure what you mean by 5 V full scale. When I had auto zero on it was in the mV range but then I cut it off and I get the 5 V. The chromatography are on the same scale.

I am using a packed column, since the PDHID is supposed to have a capillary column I have a packed column insert for the detector. I removed it and it looked fine so I replaced it. Then I removed the ground pen and that had a little black ring on it but for the most part it is clean.

Did you replace the column, or just put the one that broke back in it?

After all this diagnoses, the signal is still at 5 V. I am starting to think the detector is bad.

[rb6banjo]

The break was after the last part of the coil so I only lost about 20 cm of column (HP-MOLSIV, 30 m x 0.53 mm x 25 µm).

Do you have a way to inject more sample to see if you get any increased response or the column fed by a rotary injection valve/fixed volume loop?

[aidnai]

You said the chromatograms are the same scale, so it looks like the noise has increased a huge amount and the signal has been decreased somewhat, not quite half it looks like.

I think rb6banjo's hunch about broken column is a good one. I would check for leaks elsewhere as well -- ferrules can loosen after thermal cycles, columns can break, any number of things.

Personally I don't see the autozero thing getting you anywhere.

Good luck figuring it out!

[Peter Apps]

Smaller peaks and a noisy baseline point to a leak, but the retention times are not affected so it must be somewhere that does not affect carrier flow. Between column and detector is the best candidate, followed by the injector valve.

Check with a leak seeker.

Peter

[JWF239]

Smaller peaks and a noisy baseline point to a leak, but the retention times are not affected so it must be somewhere that does not affect carrier flow. Between column and detector is the best candidate, followed by the injector valve.

Check with a leak seeker.

Peter

Thanks for the help everyone. I have a new theory. I baked the column oven but it was only for two hours before I ran a sample. I believe that I may have not given the column enough time to bleed everything out and now the column and the detector are saturated and need to be baked out. Would this potentially cause the issues I am seeing?

[rb6banjo]

Perhaps. You can see "ringing" on the red trace. That's what I would consider the noise from your detector. Tyically, true noise is much more random than the "blobby" sort of stuff that you're monitoring at the detector. A good bakeout of the column might be wise.