I work in a contract lab and we frequently get questions from sponsors about the sampling rate we use on our TurboChrom system. I was always taught that as long as you have 20 points across the peak, you should be OK, and I have tried to argue that with a sponsor. Unfortunately, I have no reference I can point to to convince the sponsor that I'm right (or learn that I'm wrong). Can anyone help?

I believe the correct sampling rate depends on the interaction between the chromatography and the data system's integration alogrithm. In most cases the sampling rate recommended by the data system manufacturer (normally dependent upon the peak width) is adequate. However, if you're dealing with very broad peaks as might be found in GPC/SEC or an unusual detector, you may need to play with the sampling rate to obtain proper integration. You can always collect data with a fast sampling rate and re-integrate it with a slower rate to see if the change has any effect. Of course, you can't do the reverse. Finally, the greater the sampling rate, the larger the data file and the longer processing is going to take, which may or may not present problems depending upon your system resources.

Overall, I don't think there is one simple answer; the choice of the correct sampling rate is dependent upon many factors. If you're using something besides the manufacturer's reccomended settings, you probably should have some supporting documentation for your choice.

Regards,

Michael Hinsberg

http://www.hinsbarlabs.com

I appreciate the feedback. However, I am certain that there is some statistical basis for defining the minimum number of points (something along the lines of 2 sigma relative to peak width). Other input?

Okay!!! Read this:

Snyder, Kirkland, and Glacjh, "Practical HPLC Method Development", p.665 under the topic entitled 'data system effects'.

There is no one answer to this question....you just have to use good logic during the development phase.