GC - Ghost Peaks - small hump at 10 mins second time around

[swheelz]

Dear All

I have just spent a very frustrating week trying to find the source of some very annoying 'hump-like' ghost peaks in my fatty acid profiles.

I am using an Agilent 6890 GC machine with an autosampler. I am analysing EDTA plasma for fatty acids using direct transesterification preparation and then a 30min split injection (1 or 5ul) GC run starting at 160 degrees for 10 minutes, then 200 degrees, then baking off at the end for 5 minutes at 240 degrees C. The matrix is a mixture of the lipid fraction of human plasma, toluene and methanol.

I'm a bit of a newcomer to GC and chromatography in general but my colleague in the dept is a biochemist specialising in separation science and he is totally stumped too. I would really appreciate some help if anyone has any ideas.

Problem Description:

I can run a plasma sample through once and it is fine. The column is new (I forget the model I'm afraid) - it gives cracking peak resolution and flat-as-a-pancake baselines. However, if I run the same plasma sample through again the second run shows a slightly asymmetrical hump (gentle slope followed by sharp fall back to baseline - about 25 pA in size lasting approx 1min) at about 10-11 minutes, though it can appear later than this. The rest of the chromatogram is perfect.

If I put a blank hexane or methanol after the sample, I get a flat line but with the same ghost peak at 10-11 minutes - the next plasma sample will then be fine. However, I obviously can't be putting a rinse after every sample!

It is almost like each sample is contaminated by the previous one. I have left the GC running over the weekend on a run where each plasma sample is followed by both a methanol and then a hexane rinse - nearly 50hrs runtime for about 30 plasma samples which is pretty OTT but I wanted to get some results out of the batch that I had prepared and they were already getting a bit old. At least the results might tell me more about the source of contamination when I go in tomorrow morning.

Attempted Fixes So Far:

We have (a) changed the water filter (dessicated beads were 'full' so refilled with fresh beads), (b) changed the liner (goose-neck) twice - having found black particles in the old liners though not sure what they were or where they came from, (c) cleaned the inlet port (not sure about the terminology - basically the bit below the autosampler) with methanol, acetone and hexane, and (d) taken a few inches off the column at the injection end. We have also changed the source of the distilled water for the H2 and changed the air canister, though only because it was empty. We have also changed the septum.

I have made new batches of internal standard with totally clean glassware - all reagents are new and unlikely to be contaminated. The problem remains though.

If it is always the second run that contains the ghost hump, I am assuming that this is 'carryover', which I read is caused by overdosing the column with sample. Yet the phenomenon remains even when injecting only 1ul of sample (split from 5ul). Surely this is not too much?

If anyone can help, please please help as I am about to seriously lose it and throw the GC out of the window.

Regards

Simon

[Peter Apps]

Hi Simon

From your description of the shape of the ghost peak, and assuming that you are using a non-polar column (please post the column description - there should be a label on the column with all the details), the ghost is a free fatty acid, and its disapperance after you do a blank run confirms that it is coming from the previous sample (a genuine ghost peak).

The easiest solution is to extend the hold time at the end of the temperature programme until you see the peak come out. However, having a free acid in the sample might point to a problem with the sample prep.

Peter

[swheelz]

Dear Peter

Thanks for your speedy advice. I am using a BPX-70 25m x 0.22ID x 0.25um capillary column. from SGE Forte. I am told this is relatively polar.

The massive run I put on over the weekend is consistent with what has happened thus far. I have got good results for my fatty acids, but the methanol rinse is always contaminated afterwards with the same hump at 10-11mins. The hexanes after the methanols are clear.

I will try the extended run idea while I am making new samples, but I know that this 30min method works and is well proven in my lab.

I am going to test my reagents, prepare new internal standard master stocks and find out where this is coming from. If anyone else has any ideas, please let me know.

Thanks,

Simon

[KC]

Yes, it certainly sounds like a late eluter to me. The component stops on your column when GC temp comes down for next injection. If you were to increase your hold even a little and the next injection this peak comes off a little sooner than when you typically see it, its definately the problem. However, if you say that this method has been run previously with the same matrix by others without this problem. it points to either your sample processing or there is truly something different with this matrix.

[pi3832]

I will try the extended run idea while I am making new samples, but I know that this 30min method works and is well proven in my lab.

Just extending the hold time at the end of the run will tell you if the hump is due to a late eluter.

If it is, there may be ways of dealing with it that don't involve the long hold times that it takes to identify the problem.

Mike

[swheelz]

Dear All

Thanks for your advice. I have ruled out reagent contamination through a process of elimination.

I have also extended the runtime to 50 minutes (from 30mins). At 40 minutes I get a large peak (about 100pA - about the same as one of the major fatty acids). This prevents the following sample from being contaminated so no need for any more rinses!

However, there must still be something wrong as no fatty acid elutes at 40mins - certainly not one with such a large peak. My colleague thinks there may be something wrong with the detector. I have ruled sample prep out as the problem and just about everything else has been changed.

Any more thoughts?

Thanks for all your help.

Regards

Simon W

[chromatographer1]

If your peaks are coming out with the proper retention time I do not expect a detector problem.

You have something in your sample that elutes much later than the fatty acid methyl esters you are measuring.

This could be a monoglyceride, a highly oxidized fatty acid or hydroxy fatty acid, an fatty acid anhydride, or simply some underivitized free fatty acid.

If you are measuring unsaturated fatty acids you might have derivitized them too long or at too high a temperature.

You could have also derivitized at too low a temperature or too short a time.

Or you may have found an previously unknown component in the plasma which you separated and included in your fatty acid sample.

There are other possibilites but that is enough to start your trouble shooting.

Good luck,

ROd

[swheelz]

I don't think it's the sample itself as I am using two different types of plasma: my own plasma which I put in the freezer for test runs such as these, and the subjects' plasma which comes from pregnant teenagers. The reagents are definitely not contaminated and I have triple-checked each stage of sample preparation. All glassware was scrubbed and rinsed in deionised water by me this morning and dried in the oven before use.

The method that I am using (modified version of Lepage & Roy, 1986) gives me the option of immersing the samples in a water bath either at 40 degrees overnight or at 60 degrees for 2 hours.

Both periods have been used successfully for several years by others in the lab and by me since I started doing this assay a couple of months ago (though admittedly with a previous column - this one is new). I prefer the overnight option if possible but sometimes time does not permit.

That's a point - could it be the column?

Could it be the toluene reacting with the plastic Gilson pipette tips used to add internal standard to the plasma? I've not had this problem before but perhaps the store has changed brands?

Regards

Simon W

[Peter Apps]

Hi Simon

The change in column might have something to do with the appearance of the problem. If the old column had developed activity towards free acids, hydroxy acids etc then the peak that you now see as a ghost could have been tailing so badly that it was never recognised as a peak - just a vague hump on the baseline. Your methyl esters would still have eluted as nice sharp peaks, even on a column that was chewing up free acids - that is the whole rationale for derivatising the acids in the first place.

Is there any possiblity that you could run the samples on GC-MS to identify the mystery peak, without knowing what it is any troubleshooting is just guesswork.

I would be very surprised if anything was coming off the Gilson tips - I torture tested Gilson tips a few years ago and they were remarkably clean.

Peter

[chromatographer1]

Certainly I agree with Peter, you could have a better column than you had before.

You could have had this late eluting peak previously and never saw it due to the column absorbing or flattening the peak so it was not noticeable.

The good news is you know it is there. The bad news is you will have to wait for it to elute each time.

It could be worse.

best wishes,

Rod

[swheelz]

Dear All

I have successfully brought forward the 40min peak to 30mins by introducing the 240 degree ramp-up earlier. All-in-all, I have only extended the total runtime by 3 minutes over the way it was before the new column, from 30 to 33 minutes - I thought it wise to put in a 10% margin of error on the elution time.

I do have access to GC-MS though it is being used mostly for isoprostanes at the moment. Perhaps I will satisfy my curiosity later - it would be good to learn how to use GC-MS. I thought that one had to know roughly what one is looking for when using GC-MS? I shall go and look it up when I have time.

I just wanted to say a big thank you to all those who have helped me, especially Peter from South Africa who is a total star. There is no way that I would have resolved this problem without your expertise and I would still be tearing my hair out.

Many thanks and best wishes,

Simon W

PhD Student
King's College London

[HW Mueller]

To avoid this sort of problem we did a "heart cut"....published ~1982. I would be surprised if you didn´t have such a problem.