RRF using different wavelengths ??

[Mattias]

Hi,

I am working with an impurity method where the UV(max) for the main peak and the impurities (also between the impurities) are very different. The main peak has a nice UV max at 300 nm whereas one of the impurities does not absorb above 210-220 nm. One impurity has single UV max of 242 nm. There is no single wavelength where I can get good LOQ of all peaks.

I don't want to use external standards for every impurity. Could an acceptable solution be to record at many wavelengts (DAD), and then compare the impurity peak areas with the standard peak (recorded at a different wavelength using RRF factors)?

Thanks,
Mattias

[Alex Buske]

Yes, if your PDA is sensible enough.

Alex

[Mattias]

This might be a trivial question:

Will the peak area be dependant on the settings for each channel (e.g. bandwidth, reference channel etc)? These things probably needs to be controlled carefully.

[HW Mueller]

Wow! You have sensible PDAs now? It doesn´t make sense to me what Mattias asked. Is this the same question we had somewhere else recently, whether one can quantitate without calibration? I tried to be sensible and it told me strictly speaking: No. If you state what values you assumed for the specific absorbtivity you may present such an assumption? ("My impurities are 20% assuming the absorbances are..... , but I really don´t have the foggiest idea what the absorbances are.")

[tom jupille]

Hans, I interpreted Mattias' situation a bit differently: the impurities are known and he has standards, he just doesn't want to use them every time. What he's proposing (I think!) is to determine for each impurity the RRF relative to the parent at λmax of the impurity, then simply use those RRFs in the analysis runs. (Mattias, correct me if I'm wrong!)

That can work, but it would be risky because you would be working on sloping portions of the spectrum for the parent. In addition to the pitfals Mattia mentioned (bandpass, reference wavelength, etc.) I'd be concerned about spectral shifts with temperature or mobile phase composition.

Variable λ detectors are still more sensitive than PDAs by a bit less than a factor of 10 (but then again, I've seen more variation than that on a single instrument!).

[Basil]

Tom,

With all this kind of differences in UV absorption would you be sure that you see ALL impurities...

What happens if an unknown impurity arise ? How wolud you control it ?, using UV at 300 nm which is maximum for product ....

Mattias:
Could be possible to prepare a spiked sample at known level that could be ussed as an external standard ??


Last option: Would it be possible to use GC (molecular weight, properties,..)

[DR]

Basically, I'd say "Yes" but you have to keep in mind that your absorbance ratios can be scrutinized by any audit inclined people. This means that it would be a good idea to make sure your detector is calibrated using holmium oxide filters or potassium permanganate solutions so you know what your relative absorptivities are over several wavelengths for one of these standard materials. Min/Max for caffeine won't cut it.

[Mattias]

Tom Jupille> You have interpreted everything correctly. It is an old product, and I have standards for all known impurity peaks.

I guess the best solution would be to use external standards anyway, then there will be no discussion about my quantitation.

The old method is using fluorescence detection for the impurities (with all these differences in UV spectra!!). Quite interesting choice of detection, since 50% of the impurities show no flourescence

[HW Mueller]

Glad to hear that, but after several suggestion to guess my sensitivity is up and sensibility down. But still I don´t understand something. If you calibrate everything at the respective maximum wavelengths with the detector programmed to change to these wavelength at the right time (rt window), then run your samples the same way...., what´s the diff. to any other run with external standard?

[tom jupille]

With all this kind of differences in UV absorption would you be sure that you see ALL impurities...

First of all, you can never be sure on the basis of chromatographic information alone (there is always the possibility of coelution), and spectral data can be of limited utility for impurities with similar chromophores, but that's a general problem.

What happens if an unknown impurity arise ? How wolud you control it ?, using UV at 300 nm which is maximum for product ....

The PDA lets you aquire spectral and chromatographic data simultaneously. The ones I'm familiar with will let you go back and plot a "maxplot" chromatogram of the maximum absorbance at any wavelength as a function of time (sort of the equivalent of TIC on a mass spec). That will at least let you see that there is a "new" peak present (subject to coelution issues, of course!). Quantitating is a different matter; you would have to either assume a RRF at some wavelength or identify the compound and obtain a standard. Again, this is a general problem, not limited to this situation.

[Alex Buske]

Matthias,

Yes, wandwidth and reference can matter, however, if you collect all (detemination and use of the rrfs) the information on one detector with channels with identical settings i wouldn't see a problem.

Hans,
For some impurities it is hard to get standards. they have to be synthsised, tested..... that can be a few kEUR for few mg.
On the other hand, the drug substance is readily availible, also as standard.
If you specify in your method that external calibration is to be used with reference standards someone has to organise and pay for a continous supply of of properly qualified standard substances.

Using different UV traces is in my point of view nothing different than using deifferent mass traces in LC;S.

Alex