2
$\begingroup$

Are there any generic rules / heuristics to identify a good GC vs a bad one?

The context is that during manufacturing-troubleshooting I often get shown GCs of various fine chemicals and I often wonder whether a problem is truly process-related or merely just an artifact of a bad GC operator / column / instrument etc.

Are there any benchmarks or signs to look for that will give a hint that a particular GC wasn't very reliable etc?

For example I've added one such GC graph below. Are there things too look for? e.g. I suppose non flat baselines or clamped off peaks are obviously bad? Merging peaks too? What else?

enter image description here

enter image description here

$\endgroup$
  • $\begingroup$ I used to intern at a GC laboratory so my knowledge is very limited. Nevertheless, what data is available to you? $\endgroup$ – CoffeeIsLife May 13 '16 at 10:52
  • $\begingroup$ @QuantumMOCHACCINO I've added one such graph to my post. Typically this is available & the heights / Areas & respective percentages. $\endgroup$ – curious_cat May 13 '16 at 13:10
  • $\begingroup$ What you need is a standard mixture of compounds that you can store indefinitely and inject every day or every week or every 20 samples (some frequency that you find gives you confidence). The chromatogram for this sample should have well-spaced peaks for your particular column and temperature gradient program, and it should look the same every single time. Here's one possibility: sigmaaldrich.com/catalog/product/sial/… $\endgroup$ – Curt F. May 13 '16 at 16:28
  • $\begingroup$ I would first take a look at the background scan. Make a run with the GC with the same solvent and conditions previous to the run of the sample. Also, if you happen to possess the pure chemical in the sample, analyze it with the same GC instrument (under same conditions) and compare the retention times. You could also try different detectors. Bear in mind that any GC will have noise. So don't worry too much about small peaks. $\endgroup$ – CoffeeIsLife May 14 '16 at 3:54
  • $\begingroup$ @QuantumMOCHACCINO Thanks. Couple of points: These are solvent free GCs. i.e. Liquid products. Injected neat. Problem is, the small peaks are still on the Quality Control specs. e.g. Some specs. on impurity peaks are less than 0.5% by GC. Is the GC the wrong instrument for such accurate specs? $\endgroup$ – curious_cat May 14 '16 at 5:40
1
$\begingroup$

For any instrument that quantifies via integration of a peak (GC, NMR, etc.), sample concentration (peak height) is important especially with low signal/noise ratios. Obviously if the peak isn't in the sweet spot - so to speak - (where its area is significantly above the baseline, but not too big that it saturates the detector) the quantification may be unreliable. Also peak shape is a consideration in integration. If the peak is too broad, much of the area under it can be obscured by the baseline. On GC/MS or GC/FID in particular I used to have problems with peaks 'tailing', that is a broad lingering asymmetry after the peak max which would often overlap with species with greater retention times. No es bueno for quantification. Baking out the column, finding a better heating profile/method usually did the trick.

Optimal methods are going to vary widely based the instrument you use, the solvent, what components you have in your mixture, there relative amounts and concentration etc.. A lot of trial and error. As a general rule though typically when I had 2 components that were close in rt, but wouldn't resolve, I'd move to a method with a much slower ramp for the column temp, upping the carrier gas if possible on your equipment might help, too.

$\endgroup$
  • $\begingroup$ Thanks! Is there some method / heuristic you used for the "finding a better heating profile/method " bit? $\endgroup$ – curious_cat May 14 '16 at 5:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.