by
lmh » Mon Dec 08, 2014 2:02 pm
Personally, I use the ion that I consider to be most selective. If the most abundant ion also gives 15 other peaks over a few minutes when used on real samples, then I don't like it much, because I struggle to believe that it runs no risk of picking up a wrong, nearby peak by mistake if anything goes wrong with chromatography, and because I struggle to convince myself that there's a 0% chance that there isn't a 16th unknown peak lurking directly under the one I'd like to quantify.
I can see the idea behind using the weakest ion so that you can be sure the qualifiers are good, but that seems to me to be unnecessarily pessimistic, and means you're actively making sure that the random errors in your measurements are as large as you can make them - a very odd thing to do. My personal take on this is to prefer an abundant ion, but accept that there is a threshold considerably higher than the LOD for the quantifying ion below which I will fail to find the qualifier ions. In fact, there is a higher threshold still, below which the qualifiers will become too contaminated by random error to give an "adequate" match to their expected values, and the peak will be rejected. In fact, I've noticed this is a common area of misunderstanding for users in our lab: they see a peak, and can't understand why chemstation is giving no data for it, and the reason is that the integrator threshold has been set high enough that the qualifier ions aren't being found, but no one ever checks the EIC of the qualifier...
The idea of using a qualifier that is less abundant than the quantifier means that the method as a whole has a set of ranges over which it can be used. As well as the LOD and the LOQ of the quantifying ion, which tell us whether the data are too little to detect, detectable but too little to quantify, or quantified accurately, we have in parallel a LOD and LOQ of the qualifier ions, which tell us how certain we should be of the identification: yes, we have a matching spectrum; no, we can't match the spectrum quantitatively because the spectrum is too weak; no, we have no supporting evidence because the supporting peaks would be undetectable.
I almost never use the TIC, but I'm not sure why. I think it's because I've almost always got at least one qualifier with other random peaks and dubious selectivity, and I worry that the TIC is the least selective chromatogram of all.
But note: I'm in an academic lab, not a regulated environment.