Chromatography Forum

Exactly how peak purity is calculated?
Does bad peak shape will give less peak purity?

Waters uses spectral contrast you would have to read the PDA guide....peak shape should not matter. Unless your poor peak shape is due to coelution (not pure)

Agilent uses a very different algorithm to arrive at a peak purity factor - best bet: read your manual.

Actually, at the mathematical level, all the algorithms used by the various vendors are similar. They use either correlation or vector product (which are effectively the same thing for this application). Now the way that it gets interpreted can be either as a vector angle (Waters) or some multiple of correlation (everyone else). The only real difference between the vendors is in how the thresholds get set.

What happens is the peak apex is selected as the reference spectrum. Then for each point across the peak, a correlation is calculated versus the reference. This gets plotted over the peak. You will need to read your manual to determine how to set the threshold.

Keep in mind this is spectral purity, and not chemical purity. Two isomers with nearly identical UV spectra may give good peak purity. A small amount of a highly colorful, chemically unrelated coelution may give a poor peak purity. Either way, your quantification is at risk.

Mark

Can you recommend any good articles - or book chapters - on this topic.

Thanks
Adam

Unfortunately for you, most of my knowledge on this subject comes from reading manuals, discussions with peers, and internal documents. There is a useful article on correlation on wikipedia.org.

You can read the operator's manual for your chromatography data system, and also the online help files. For Dionex's Chromeleon, the good stuff is in the online help.

Here is an article:
http://www.jstatsoft.org/v18/i09/v18i09.pdf

It uses R for the calculations, but the theoretical background is the same.
There are 4 different methods mentioned in the article to determine peak purity.


Bart