Calculating limit of detection for nonresolved peaks

[degenchem]

If you are unable to change the chromatographic conditions you must change the integration method to give a more accurate idea of the area. The current area will be about double the true area of created by that substance.

PS why are you unable to change the conditions? If it is important to quantitate that the conditions need changing.

A fair amount of work has been put into achieving a better sepraration, though I haven't been involved myself. The two compounds are structurally almost identical, the only difference being a double bond.

I will have an accuracy test done. I will also look at alternative methods of integrating the peaks.

Thanks again all for valuable input.

[Bruce Hamilton]

Your baseline appears fairly smooth, so my first suggestion would be to inject far less sample, maybe 5 - 10%. That should improve resolution and still keep small peak above LLOQ.

My second suggestion is to submit your structures and chromatograms to the technical departments of column manufacturers, with details of your trials, and politely request guidance on columns or conditions that would effect a superior separation.

It's not an area I know anything about ( so experts can laugh and correct me ), but C30 columns appear useful for some lipid isomers.

Bruce Hamilton

[AdrianF]

If you have a reliable variable volume injector you could inject say between 5 and 50ul.
Plot area vs volume injected and experiment with the integration to get the best line with intercept close to zero. Then you could think about LOQ.

[HW Mueller]

What good would that do?

[bartjoosen]

What good would that do?

You can find out how to minimise the bias due to the overlapping peaks.
This can be done by spiking with know concentrations or by injecting different volumes. Integrate them all the "same" way, and look where the intercept is ~0. Then you are sure that there is no bias caussed by overlapping.

If you use the different volumes injection method, you have to be sure that there isn't a bias from the injector.

Bart

[HW Mueller]

How does different injection change the "bias"? Are you assuming that he is overloading the column (with the larger component) and that lower amounts will lift this overloading, thus get a better resolution?

[AdrianF]

My point was that if you have varying amounts of the minor peak you could experiment with the integration parameters to get the intercept as near as possible to zero. I am assuming that it is impossible to get the main substance free of the minor impurity. If it is possible to get it impurity free then the problem is much easier.

All contributors are of the view that it is best to improve resolution which is clearly the ideal solution but we don't live in an ideal world!

Another question is - to what accuracy do you want to analyse the impurity? Is it just a limit test?

[Peter Apps]

Sorry, there is no way that you are going to get useful quantitation from a separation this poor. Dropping to baseline you have an unmeasured contribution to area from the preceding peak, tanget skimming you have an unmeasured loss of area. The proportion of area that is added or lost will vary with the size of both the analyte peak and the interference peak, which will compromise linearity and the repeatability of recovery.

Peter

[HW Mueller]

Furthermore, the world might not be ideal, but it hits back mercilessly when one deviates from facts, truth, or whatever you want to call it.

[AdrianF]

I have never had the opportunity to use them but there are peak deconvolution programs which can improve the area estimates.

Exponential skim is available with some integration packages which should also be better than tangential skim.

Again I ask the question what are you aiming for? How vital is it to estimate this small peak. In a years time will you still be losing sleep if the problem is not perfectly solved!

[Jumpshooter]

"Chromatography conditions in your method". At this point you have no justification for continuing to waste resources in this pursuit. Please change your method conditions--starting with your mobile phase modification.