column performance calculations with limited data

[Tony]

Hello,
I have a LCMS system that outputs peak data into a spreadsheet as:

1) integrated peak start time
2) integrated peak apex time
3) integrated peak end time
4) peak area
5) peak height

I have an internal standard and sample peak in all my chromatograms, and for QC purposes I want to monitor column performance for my sample peak by calculating the following parameters:

1) theoretical plates (N)
2) resolution (measured relative to the internal standard)
3) peak tailing

Unfortunately, all the calculations I know about for these parameters require peak width data at some % above the baseline (e.g. 5, 10, 50%). I can't get this data - I only the baseline peak width. Can anyone suggest suitable calculations/approximations using only the baseline peak width?

thanks
Tony

[Mark Tracy]

For approximately Gaussian peaks, area/height approximates width at 50%

[Bruce Hamilton]

As well as the above for PW50%, you may get an idea of tailing ( and the validity of the above assumption ) by comparing ( Peak Apex retention time - Integration Start retention time ) and ( Integration Finish retention time - Peak Apex Retention Time ).

The major problem could be the asumptions about the quality of the integration parameters used to detect and define peaks.

I'd be tempted to see if the MS could output some peaks to a printer or plotter, and manually confirm peak shape and any estimated values by visual examination of some expanded chromatograms. You could then keep those to check performance.

Please keep having fun,

Bruce Hamilton

[Tony]

Mark and Bruce,
Thanks for your suggestions. I'm guessing I could use the area/height approximation for 50% peak width estimation, and perhaps multiply it by a sigma value to estimate width at 5 or 10% height. These derived values could then be used to calculate N and R - however, not tailing, as sigma x area/height derives a total width only and not start and end points?

As far as I can see, N and R calculations are very dependent on the peak width values used (at a certain % above baseline). If I use the baseline width, I get terrible resolution and N estimates, as the peaks tail moderately (and hence have a large baseline width value). Standard calculations require width at 50% for R and 5% or 10% for N; what I was really after were "recognised" calculation methods for R and N that use baseline width. I guess these don't exist? I can use the baseline values anyway for our own internal QC monitoring of the method, where I am just monitoring N and R long-term from run to run. I can process several thousand runs we have already done using area/height derived width and peak end - peak start baseline widths to test how the different methods work.

One final question - how relevant is the R calculation if detection is by MS, especially of an extracted ion, rather than UV? i.e. R may be calculated between two peaks from the UV or TIC trace to determine chormatographic resolution; however, if extracted ion values are used from MS data, there may be no overlap at all between traces. So is R still a useful parameter to evaluate in this case? A bit of a philosophical question....

cheers
Tony