GC/MS FAMES

[lmh]

I'm really ashamed to have to ask about such a basic method, but I'm an LC-MS person who knows nothing about GC.

I have a co-worker who runs occasional FAME analysis on GC-MS (EI). Until recently he's done everything by integrating the total ion chromatogram and assuming the area percent profile of peak areas reflects the amount percent distribution of different fatty acids in the sample. The problem is that there are small but completely repeatable differences between his values and the expected fatty acid distribution in a commercial standard FAME mix, and the differences are causing us problems. The differences are consistent between several independent batches of the standard mix.

As an LC-MS person I immediately assume this means that not all FAMES have the same response curve, and get the urge to use the commercial standard to prepare a set of calibration curves. He has included C17 standard, so I would probably set it up as an internal standard method with C17 as internal standard for all my calibrated peaks from the commercial mix. I would also probably integrate extracted ion chromatograms chosen for each analyte, and quite probably a few spare masses for each as qualifier ions.

I know this "forces" results that match the commerical mix, so it will certainly get rid of our problem, but I'm unhappy about making changes in a protocol that belongs to a technology I don't use myself.

Am I barking up completely the wrong tree? I'd be grateful if someone could let me know (a) am I misunderstanding the problem, and (b) am I diverging from accepted FAMES practice? I don't want to be unfair to my co-worker and push him towards unsuitable methods.

Many thanks for any help.

[Ron]

You're on the right track, using the TIC is using the MS as an FID, without taking into account differences in ionization efficiency. Extracted ion chromatograms and calibration curves are the way to go. Even with the FID calibration curves should be used, but the response factors for the individual components are similar and the errors are not very large.

[Peter Apps]

I would go with the TIC peak areas rather than single ions and qualifiers, apart from that I don't see any problem.

How big is the discrepancy ? - before you start it might be a good idea to check the uncertainty on the composition of the standard mix, and the purity of anything that you are making up in house.

Peter

[Consumer Products Guy]

We use GC-FID for FAME assays, we find response factors for saturated C8 to C18 pretty similar (v. commercially purchased equal amount mix). For soaps and such, C8 to C18 covers it for us.

I agree that for GCMS to use a commercial or in-house standard and develop response factors. Also be aware that many "99% oleic" fall substantially below that level, if including that in a house-made mix...

[lmh]

Dear Peter and Ron,

Thanks lots for replying so speedily.

The discrepancy is in the range 1% to 10% of total peak area, unsurprisingly worse for the most abundant peaks. It's reproducible and consistent between different batches of the standard. Although the errors don't sound huge, a 1-2%-of-total error on a component that is only 6-8% of total represents an error in the 10-30% range for that particular component, which isn't acceptable to the client.

The standards are Sigma's oil reference standards, which come with a certificate of composition for that particular batch (I'm inclined to believe Sigma!). So far as I'm aware, the only things that have happened to the standard are that it's been diluted, and that an internal standard has been added; I don't think either action could favour one of the existing components over another.

I'm appreciating this site. GC-MS is outside my area. Even the TIC versus EIC thing is something I have to rethink in the context of analytes producing lots of ions instead of just the one or two I expect in LC-MS!

[Peter Apps]

Hi lmh

That the error is larger for bigger peaks suggests that your colleague is running off the top of the linear range for the MS. The solution might be something as simple as an increased split ratio, or more dilution of the stock. Another possible problem is molecular weight discrimination in the inlet - if the heavier components are more abundant in the standard you will get results that are biased high for the lighter, less abundant compounds. Splitless injections are generally more robust to discrimination than split injections.

Peter

[lmh]

Sounds like calibration curves would be useful on two counts; they'd let us know whether we've gone beyond detector linearity and also deal with variable response curves between the different fatty acids.

Ron and ConsumerProducts, you are right: my co-worker comes from a GC-FID background and would have liked to do FAMES by FID, but ironically we don't have that detector available. I come from an LC background and can't help him much.

Thanks also for the comments on injection problems.

[Ron]

One of the negatives of GCMS compared to GC FID is that the MS has a narrower linear range than an FID, thus it can be more difficult to quantitatively determine trace levels and high levels in the same run. I suspect that Peter is right, the high concentration peaks are exceeding the dynamic range of the MS. You should be able to look at extracted ion chromatograms of the major mass peaks and see if they are flat topped. The TIC may not look flat topped, but if the major peaks are clipping the area will be too low.

[lmh]

Ron, thanks for that tip. A useful way to look retrospectively at data where things might have gone wrong, but we can no longer check by running a more concentrated standard to check for linearity.