shark fins and overload solvent peaks - OK???

[JHZR2]

Hi,

Say youre running tests and youre using a small amount of a component in a solvent. Say youre testing for the presence and if so, concentration of the component in the solvent via FID.

Youre going to get a 'solvent peak', and its going to be real high, wide, look like a shark fin, and obviously be overloading the column... but to split it will prevent you from seeing really low concentrations of the other component.

Does it really matter that the solvent peak is so abnormally shaped, so long as you get a nice gaussian when the other component eludes in quantity??? I guess, in other words, no matter how ugly your solvent peak is, so long as the baseline goes back to zero prior to the other thing eluding, can you just automatically ignore the solvent peak, no matter what?

Thanks for the help and clarification!

JMH

[Bruce Hamilton]

For a minute there I thought I'd have to skip this thread because it was dealing with real shark fins...

Keep life simple. As long as the peak of interest meets all your criteria, you can ignore the behaviour of peaks that don't infer with you analysis. What you have to do is convince youself, and possibly any auditors or clients, that the solvent peak shape is irrelevant.

My normal practice would be to have an internal standard at a concentration similar to that of the analyte expected in the samples. Then all you have to worry about is the internal standand and analyte peaks. I highly recommend using IS in methods.

The choice of IS is up to you, if you can use a molecule with similar similar to your analyte, you should get an indication of your chromatography, which can be helpful. Alternatively, you can choose a compound like a hydrocarbon as IS that will almost always give nice peaks, even after the column should have been moved to a hospice.

Bruce Hamilton

[Russ]

Bruce: Quick, run for the hills! There are many people who will say an internal standard (IS) is not necessary, if not wrong. I had the misfortune to run across someone who claimed an IS was improper procedure. You might want to brace yourself for some "responses" to your comments.

[Victor]

JMH,

It is possible that an overloaded peak, if it eluted very close to your component of interest, might produce some effect on your peak of interest. But in the case of your solvent peak, it will be running ahead of your analyte for the majority of the separation, and so your analyte is seeing a mostly "clean" surface. The solvent peak is overloaded in most GC determinations. Sometimes, this has been shown to be beneficial-e.g. in splitless injections where it gives a focusing effect.

I would go with Bruce and say if your analyte peak is good, and your analysis works, then forget about the solvent peak.

[Bruce Hamilton]

Russ, Fortunately, I'm far enough away

I should clarify my position on the use of an Internal Standard. I find that use of an IS can often simplify sample preparation. Adding an IS usually means I don't have to worry about other material that may be in the sample and affect the analysis.

For quantification of a single peak or two in a clean sample ( lots of flat baseline ) that hasn't been through lots of extractions etc., I find it easier to add an internal standard just to cover chromatographic aspects, and confirm that the system is working fine.

For clean samples with exhaustive extractions etc., some sort of assurance about extraction efficiency is usually required anyway, and Internal standards added prior to any processing can be helpful, at least until the method is shown to be robust.

For unclean ( lots of peaks ) and diverse ( different sources ) where relative concentrations of components are only required, then an Internal Standard can be more trouble than it's worth. An obvious example is gasoline, where a 50 metre column may yield over 500 peaks in 45 minutes, and different refineries and crudes give differing profiles. For general characterisation, it's easier to monitor total area to confirm injection consistency, even after allowing for FID response factors.

My recommendation is that people should start with the simplest solution that works for them and their customer, only adding complexity as required.

Please keep having fun,

Bruce Hamilton

[JHZR2]

Thanks for all the advice, i greatly appreciate it!

Best regards,

JMH