UV spectrum saturation affecting Accuracy

[mwa]

Hello everyone,

While verifying a method for assay & purity of two coumpounds, we obtained bad accuracy results, with recoveries too high at 80% (>102%) and too low at 120% (~95%). Calibration is performed at 100% on the same material, so there is no matrix effect.

(The analysis is performed on a CSH C18 with gradient elution of water/actonitrile with 0.1% TFA, UV detection at 215/220 nm)

I suspected peak saturation, but the peak height is < 1AU and no flattening is observed at apex. I tried to reduce sample concentration (3X dil), but obtained the same results.

We couldn't find a proper root-cause for this, until we looked at peak purity: it appears the main peak is saturated on some portions of the spectrum, but not at the analysis wavelength or within UV bandwidth (2-4 nm).

We perform the analysis at 215/220nm and observe saturation below 205/210 nm (depending of the compound). It takes a dil 10X to avoid saturation.

Could this explain the results we obtained for accuracy ?

We have a second max in UV spectrum at 260 nm that is much lower in absorption, and when the analysis is performed at this wavelength, there is no accuracy issue, so I would say yes, but I don't really understand why...

I also observed that the current wavelengths are in a "slope" of the spectrum, which I think is also problematic, although it doesn't explain why accuracy results are systematically skewed...

Does anyone have insight on this problem ?

Thank you

[Consumer Products Guy]

Whenever I had a choice to use 260nm or so as a more-specific wavelength than 205-220nm, I used 260nm. Always.

[Multidimensional]

Using any method at 210nm (or close to it) is going to result in many problems. Often an invalid method as just about EVERYTHING absorbs in that range so the HPLC method is NOT selective for the sample and will usually fail validation (due to not following good chromatography fundamentals). It sounds like you inject a lot of material to detect the sample. The good news is that based on your post, you have an option.

If you are fortunate enough to have a second lambda max at 260nm, then please consider developing the method around that signal (with an appropriate, narrow bandwidth too). Be sure to use a DAD/PDA, not a single wavelength detector too. The use of a scanning detector will allow you to monitor ALL wavelengths and develop a method which shows one dimension of selectivity for the sample.

[Multidimensional]

"Proper Wavelength Selection for HPLC Method Development (or Purity Determination)"; https://hplctips.blogspot.com/2013/08/w ... ethod.html