Mystery Peak - after backflush valve off

[Labgurl]

I am testing for Hydrocarbon content in condensate petroleum using 7890A, TCD, Packed column. We recently replaced our column, and had to make some adjustments to the injector & TCD temperatures. The run is fairly short....methane, ethane, propane, iso & n-butane and iso & n-pentane all elute in about 4 mins and then there is a valve switch on and the backflush is started, in which hexane+ is all summed as one peak, and the backflush valve switch off is at about 12mins. It is every time after the backflush is complete and the valve switches off that another peak appears, which we hadn't seen before. I have tried increasing the run times, changing the valve switch times, adding another valve switch, and nothing seems to work.....each time after the last valve switch off the mystery peak appears.

When samples are run in sequence, this mystery peak from the first sample carries into the inital (methane, ethane, propane) peaks of the next sample run.

Could some sample be condensing or be trapped in the valve area?
Injector Temp: 200oC
TCD Temp: 250oC
Column Temp: 100 oC
Valve Temp: 125oC

I have tried baking things out.

Any advice?

Thanks
Labgurl

[Labgurl]

Would it be a terrible thing to remove the backflush entirely for one run and just let the run go until everything elutes?

My understanding of the purpose of the backflush is, to decrease the length of the run and that the sample only enters so far into the column? Is this correct?

[haiedc]

Could you post the schema of columns and valves. It would be easier to figure out. Anyway, the problem can be a small leak. Try to tighten the ferrules and nuts. and see.

[chromatographer1]

It appears that you are doing a C6+ peak using what is commonly called a backflush but it actually a reverse column step or a backflush to detector, a three column arrangement, consisting of a reversal column, an analytical column and a buffer column.

If your column 1 is not well conditioned you may be seeing a phase bleed peak (this will go away with time) but I suspect that you might have a flow upset peak. This might be because your column 3 is too short or you have a leak as an earlier post mentioned.

Since you are injecting as you valve switch, extending your runtime won't help.

Try flushing your sample line with air or helium and do several runs to see if the late peak is still appears. If it is phase bleed peak it will disappear with time as the phase levels begin to saturate the third buffer column. Let it run overnight and see if that helps.

Good luck,

Rodney George
consultant

[AICMM]

Labgurl,

Are you looking at natural gas condensate or processed petroleum condensate? The reason I ask is olefins. If processed gas condensate, then your extra peak may be an olefin although your comment about next run appearance is puzzling. If the new peak is before the C6+ peak, then I would surmise that your new column is not retaining quite as well and that your valve time is a little too late and the new peak is on the buffer column before the backflush valve initiation.

However, your comment about the peak appearing in the new run makes it sound like the new peak elutes after the C6+ peak ?

A couple of questions. First, what is the backflush valve, a 4 port or a 6 port valve? Second, what is the shape of the new peak, gaussian, broad, up-and-down, etc? Third, are you running isothermal and are you running packed or capillary?

You can certainly take the backflush off to see what appears. Just make sure you add enough time to the run to get everything off the column, which could take a while.

Best regards.

[Labgurl]

I have tried tightening the nuts/ferrules & baking out for about 4 hours....tried changing the valve switch time again. The peak still appears. The GC is baking out now overnight.

I will have to check the valve diagram in the morning to see if it is a 4 or 6 port. The peak shape is symmetrical, wide and short. The column is a packed column. And the condensates are processed petroleum, and not natural gas.

I'm not sure how to post my sketch of the chromatogram.....I will try to in the morning and find out the valve information as well.

Thank you everyone for your help.

[Labgurl]

So it appears that the valve is a 6 port. We are only using 1 packed column. The mystery peak still appears, even after bake-out.

We tried running a sample without the back-flush. It took about an hour for everything to elute. It was noticed shortly after the n-C5 peak reaches baseline, that a small peak elutes next in order (possibly around the time of the valve switch on to initiate the back-flush?) Could this small peak getting trapped somewhere and not detected until the valve switches off again at the end of the back-flush? The tubing from the column (detector side) to the valve waas recently replaced, and is much longer than it was before (it is now looped a couple times)...could something such as that being making a difference?

I've checked all the valve connections and fittings and everything appears in place and secure.

[chromatographer1]

It sounds like you have a flow rate change due to your changes.

The TCD is sensitive to flow changes, if your flows are not identical with the different positions of the valve you will see an upset in your baseline which you may interpret to be a peak.

Questions:

You are backflushing the analytical column to vent or to detector? Is there a short backflush column before the main column?

Your change of the tubing (a longer piece of tubing) may be the cause of the problem.

best wishes,

Rodney George

[AICMM]

labgurl,

I would second Rodney's question about the backflush column before the main column. Is it possible for you to post your plumbing diagram?

Best regards.

[Labgurl]

For sure I can post some diagrams..........can anyone help me out with how to post pictures on here. I haven't done this on the forum yet. I can see the 'insert image' button in the post page.....but it looks like it has to be an image from a web address?

I see this:
"Insert image: [img]http://image_url[/img] (alt+p)"[/img][/list]