Regarding peak purity

<p><span> I taken peak purity plot main peak in noise+solvent mode.</span></p><p><span>My main peak purity angle less than that purity threshold. Purity flag No. But in peak purity plot similarity curve and threshold curve interference is there.</span></p><p><span>is it correct? Please give any reference.</span></p><p><span>Regards</span></p><p><span>MSK</span></p>


  • Hi msk,

    can you report peak angle and threshold, maybe post the peak purity plot?

    If you have a small peak, purity angle can be greater than 1, but lower than purity threshold and that is okay.

    some examples:

    Peak hight: 344 /purity angle: 6.646 /purity threshold: 7.382 -> that's okay, peak is not much higher than baseline noise.

    Peak hight: 3735 /purity angle: 0.845 /purity threshold: 1.011 -> that's okay, too.

    Peak hight: 38538 /purity angle: 0.084 /purity threshold: 0.268 -> that's okay, too.

    Peak hight: 38538 /purity angle: 0.646 /purity threshold: 1.382 -> there is something wrong with baseline noise, the value is too high. Check out noise interval time, it is likely that there are some up and downs in the baseline. Try another time window.

    Peak hight: 441238 /purity angle: 1.646 /purity threshold: 0.382 -> At first the peak seems to be not to great an the choosen wavelenght, but peak lambda max at another wavelenght exceeds linearity. Second: again baseline is not smooth enough at noise interval time.

    BTW: Try to read all the manuals waters has published so far. If you got the chance, get the old millennium manuals, there you can read a lot about peak purity. If you can't get those old manuals, contact your local waters people and ask them , beg them , put some pressure on them , what ever nessecary to get the manuals, it's worth the effort.

  • the significance of a noise angle is entirely dependent upon the selection of the noise interval. if the noise interval contains peaks, the spectral contrast angle effectively becomes a library match to the unknown peak. be sure to use the maxplot for peak purity measurements. this will ensure that you can see all of the "hidden" peaks in the noise interval and ensures that you see the integrated peaks at their full height. since a combination of changes in photometric noise and non linearity effects (a detector which has 5% deviation from linearity at 2.5 AU does NOT have a 0 deviation from linearity at 1 AU) limits the rigor of comparing spectra for peak heights greater than 1 AU. most anomolies in peak purity results are associated with (1) spectra containing absorbances greater than 1 and (2) unknown peaks in the noise interval.

    it is also important to avoid oversampling when there are more than about 100 points across a peak, the differences between successive spectra can be smaller than the noise. if sampling rates which are typical of UPLC (20 - 40 Hz) are used with peaks that have HPLC peak widths that should be sampled at 5 Hz, peak purity results can be compromised.

  • A link to user guides and manuals is located in the "Featured Resources" widget on the bottom right-hand corner of the Overview tab of the community. When you select the link, you will be directed to our web site ( where you will be presented with a predefined search of all UPLC guides/manuals.