Page 1 of 1

UV detection on the slope of the absorption curve

Posted: Wed Nov 28, 2018 12:42 pm
by Mike H.
What are the potential problems of setting the UV detection wavelength on the slope of the curve as opposed to a maxima?

Would a successful method validation at a wavelength on the slope prove that the method and generated data is scientifically sound?

Re: UV detection on the slope of the absorption curve

Posted: Wed Nov 28, 2018 1:23 pm
by Consumer Products Guy
Seems OK to me, if everything checks out. Sometimes there could be an interference at the maximum/apex.

Re: UV detection on the slope of the absorption curve

Posted: Thu Nov 29, 2018 10:21 am
by lmh
It shouldn't be a problem. In fact Shimadzu have built a couple of slope-related things into their quantitative software. They have an optional scheme to widen the dynamic range of PDA-based assays by deliberately taking the absorbance further down the slope, away from the peak, in over-concentrated samples. The system seamlessly moves down the slope until it's happy that the absorbance isn't too high, and then compares the expected absorbance at its new, chosen wavelength with that at the original (probably maximum) wavelength in a standard where both values are sensibly in the linear range of the detector. Clever scheme! They wouldn't have put it in the software if measurement away from the maximum was unsafe.
(they also have another clever little trick, where they plot a chromatogram of the slope of the spectrum at the chosen wavelength, rather than the actual value of the absorbance spectrum at that wavelength. This is good where you have coeluting peaks; by choosing a wavelength that is a maximum or minimum for one of the compounds but half way up a slope for the other, the chromatogram excludes the first compound.)
Basically provided you can demonstrate in validation that your signal is a true, precise, specific and accurate measure of the amount of stuff in the sample, it doesn't matter in the faintest from where the signal is derived.

Re: UV detection on the slope of the absorption curve

Posted: Thu Nov 29, 2018 1:37 pm
by Mike H.
I'm surprised that detection on the slope is non issue. In this LC separation, there are 3 compounds. All 3 have nearly identical UV absorption curves with maximas at 238nm and 287nm. One of our vendors developed and validated a method at 252nm right on the slope. I know from experience that there are no spectral interference that between these compounds and they are well resolved. We have a meeting this morning to ask why 252nm was selected.

That said I've looked at the UV spectra for the compounds over a 4nm bandwidth. The mAu spread is between 350nm- 550nm. Each compound has a different min/max mAu. What I'm worried about is with this much variability inherent to the method to begin with could minor LC to LC variability pose a potential problem?

Re: UV detection on the slope of the absorption curve

Posted: Thu Nov 29, 2018 3:38 pm
by Andy Alpert
A number of inexpensive absorbance detectors consist of a module with an LED that emits a single wavelength. You can use any LED that's commercially available for these. An LED below 240 nm (cf. your 238 nm) is at the cutting edge of the field, but LED's around 255 nm are used routinely. Maybe your vendor happened to have a module with a 252 nm LED, so that's what they used when developing the method.

Re: UV detection on the slope of the absorption curve

Posted: Fri Nov 30, 2018 2:52 am
by tom jupille
Would a successful method validation at a wavelength on the slope prove that the method and generated data is scientifically sound?
My take on this is that the answer is "yes". Operating on a sloping part of the absorbance spectrum increases the *potential* for error if the wavelength or the spectrum shift. Proper validation should demonstrate whether or not that potential problem is an actual problem.

Re: UV detection on the slope of the absorption curve

Posted: Fri Nov 30, 2018 11:52 am
by lmh
Tom's got a point, but he's right, it depends on validation. My experience of modern PDA detectors is that they have very, very good wavelength stability. Andy's point is good too; back in the old days, a lot of people used peaks taken from a mercury lamp's emission spectrum (which is not continuous), so there are still applications around that specify wavelengths from mercury rather than the true maximum of the target compound. Of course the mercury spectrum is something very reliable, so it would never have been an issue.

In terms of whether the variability between your compounds will lead to variability in your results between labs and different instruments: if you have stable instruments and each lab makes its calibration curves on the instrument it uses, then really the variability shouldn't translate to variable results. We already have different lamps in each instrument, with wildly different light intensities and emission spectra, and that doesn't prevent labs from setting up robust assays.

Your application has well-resolved peaks; if there were near-coeluting peaks with different spectra I would definitely, if possible, choose a wavelength that was suboptimal for my target compound but zero absorbance for the coeluter, rather than struggle with a mixed peak (so, in effect, choice of wavelength can also depend on matrix)