N, efficiency in gradient run

[tilly]

I am trying to find some further information about whether it makes sense to include calculations of effeciency, N when using a gradient run. I have heard it is meaningless to use a measure of efficiency in this context. Can anyone shed some light on this? I'm loking for explanation of why it might not be appropriate to use this measure and possible alternatives. Also, how can efficiency as calculated by N be improved.

This is my first post and welcome all responses. Thanks very much.

[tom jupille]

I have heard it is meaningless to use a measure of efficiency in this context.

N is not meaningless in a gradient, but most people (and most data systems) use the an isocratic formula to calculate N, so the number obtained from gradient data is meaningless. The catch is that N cannot be calculated from a single gradient run; data from two runs with different gradient steepness are required, and even then, the math is far from straightforward.

In general, if you measure N isocratically on a given column, the gradient plate number (officially called N*) will be similar. From a practical "system suitability" perspective, the peak width is just as useful in a gradient as N is in an isocratic separation.

As far as improving efficiency goes, the rules are the same in isocratic and gradient LC: smaller particles, less viscous solvents, higher temperatures, smaller analyte molecules, a well-packed column, and minimal extra-column volume.

[Uwe Neue]

True column efficiency is a way to specify the peak width inside the column. It is useful from a practical standpoint because it permits a judgement about the quality of a column.

However, if you use the same procedure for calculating efficiency in a gradient separation, you get nonsense. You will recognize this yourself, because the so calculated "efficiency" is drastically different at the beginning and the end of the gradient.

Since the calculation is nonsense, it is also not a good means to control column quality. If you want to have a measure of column quality in a gradient separation, you are better off using peak width and asymmetry. These values will still depend on the gradient run and are not good for comparison of different gradient runs, but they are the best measure of column quality for a fixed gradient procedure.

If you want to compare the quality of different gradient separations, say with different particle diameters and column length, the next best thing is to calculate the peak capacity. I just wrote a giant paper on the subject of peak capacity, and I can give more info, if you care.

[tilly]

Thanks for your replies! The calculation is done on the same peak in each run however so could it be argued that the value of N can be used to compare how the column is performing betweeen different runs ??
We have been using this in our method since it was validated so there will some convincing of colleagues required and also more work (revalidation?) to arrive at parameters for peak width and asymmetry. Any ideas on how to approach these issues?
I would also be interested to obtain further information on peak capacity as this is something i know very little about.

Thanks for your help

[adam]

The answer to your last question is yes, but there is a danger.

Even though it doesn't tell you the true efficiency in a gradient method, you can use N as a metric which will tell you if the peak width is changing. This will work if the retention time of the peak is always the same. The danger is if the retention time of the peak changes. And this could happen if you go to an HPLC with a different dwell volume, or if you accumulate "junk" on the guard column, or as the analytical column ages.

Bottom line: you can do it but there are risks. If it has worked historically for your application you may be OK.

[tom jupille]

"If it ain't broke, don't fix it!" is good advice. As Adam pointed out, so long as retention time is constant, then tracking N will tell you if your column efficiency is deteriorating. Tracking the width does the same thing and is simpler, but (to my mind, at least) is not worth the bother of changing in a validated method.

The peak capacity is just what it sounds like: the number of peaks you can fit in your chromatogram. Essentially, the length of the chromatogram divided by the peak width (to a first approximation, all the peaks in a gradient chromatogram have the same width).

[juddc]

. I just wrote a giant paper on the subject of peak capacity, and I can give more info, if you care.

Please let us know where it's been published - I'd like to have a read.

[Uwe Neue]

Journal of Chromatography
"Peak capacity in unidimensional chromatography"
In Press, Corrected Proof, Available online 8 December 2007
Uwe Dieter Neue