Wavelength accuracy test

[anirbanrc]

We have a certain querry ,raised by one of auditors of one of our clients. This is regarding the wavelength range for which Waters HPLC. As far as we know during the IQ,OQ,PQ of HPLC instrument (as done by Waters ) calibration using erbium perchlorate is done at wavelengths at 255nm,379nm and 522nm.
So what about the wavelenghts below 255 nm. Most of the compounds we are working are below the above specified wavelength. So what is the validity of these analysis.

In other systems similarly it uses caffiene , at wavelengths 205 and 274 nm .But what about the other ranges.

Thanks in advance for your reply.

[danko]

Hi Anirbanrc,

Under the PQ (I would place it there anyway, but it can be done under the OQ) one needs to validate the instrument’s performance in the context of its relevance to ones specific requirements. In other words; if you use the detector at wavelengths around 210 nm (or something like that), you have to choose an external reference material for testing the accuracy that absorbs in that region – caffeine is one possibility, uracil is another etc. If I’m not mistaken one can purchase certified reference materials for almost all thinkable wavelength regions.


Best Regards

[grzesiek]

Interpolation is usually more apropiate than extrapolation

This kind of test (for example with caffeine) can be performed by your staff and results attached to HPLC documentation.

The auditor is in my opinion right (how I don't like to agree with them ), and as danko suggested your results are not valid.

[lmh]

If you've got a DAD, the spectrum of your standards is a good demonstration that wavelength calibration hasn't drifted from day to day. Of course you still need to prove that the calibration was right in the first place.

[anirbanrc]

Thanks everyone for the reply.
If interpolation check is required to be done, then wavelength range to be done should be within the calibration range.
Which implies that for a particular compound , which i am currently working has working wavelength of 207 nm, I have to do a calibration of lower than that.
And one of my future project has working wavelength of 195 nm.Any suggestions what compound to use for this calibration.

[danko]

Which implies that for a particular compound , which i am currently working has working wavelength of 207 nm, I have to do a calibration of lower than that.

Or at the actual wavelength (i.e. 207 ± a couple of nm).

And one of my future project has working wavelength of 195 nm.Any suggestions what compound to use for this calibration.

195 nm is almost “looking for troubleâ€

[fsistere]

Hi
I remember that i use Holmium Oxide for the accuracy of wavelenght. In the spectrum there are a lot of thin lines in the visible and UV region.

[tom jupille]

While I'm generally in favor of "better safe than sorry", I think we may be worrying excessively in this case. First of all, UV spectra in solution are notoriously information-poor. For most compounds, there's not much structure, just a broad blob or two.

Secondly, most HPLC detectors have a fairly broad bandpass (compared to spectrophotometers) because S/N is more important than spectral accuracy. USP and the FDA allow +/- 3 nm error in wavelength as within the allowable "adjustment" range.

Third, you will be running the calibrators and samples on the same instrument, so any small wavelength errors will affect both in the same proportion.

Fourth, assuming your method(s) have good system suitability requirements, passing system suit effectively provides a final qualification test of the overall system.

My take on PQ is that it is designed to check the general functioning of the module to provide a benchmark and reference when problems arise (like failing system suitability). To put this in perspective, when you PQ your pump, do you check flow accuracy at multiple flow rates?

[DR]

Caffeine, Holmium oxide have been mentioned, so I'll toss one more option in (that we used to use) - make a series of standard potassium permanganate solutions (0.0200, 0.0600, 0.100mg/mL) and check molar absorptivities at 235, 257, 313 and 345nm.

[anirbanrc]

thanks again for your replies.
Taking the system suitability test, should have satisfied me, or any analytical person.But was difficult to digest for an auditor ( who perhaps was not so analytical) .I think concept of interpolation is becoming the consensus because perhaps , we equate the detectors are working in the same way as a say an pH meter works( milli volt is converted to pH and the linear relationship between the actual pH and the reading is interpolated).
But i wonder whether on what principle the HPLC detector works.Is it not so that , if you check one particular wavelength , the diffraction grating adjusts itself to give correct wavelengths at all other wavelengths.
We have a Malvern particle size analyser , which works on the same principle of laser diffraction . Malvern claims that 3 calibration runs( covering a particular range) suffices for the entire range.

Secondly the compounds mentioned namely caffiene , holmium oxide, pottasium permnaganate,and for that matter Pottasium Dichromate etc, does not cover below 200 nm range.

[danko]

Taking the system suitability test, should have satisfied me, or any analytical person.

Depends on what you’re looking for. Some people are satisfied with seeing the expected retention time/s. Other people are satisfied with seeing the expected peak pattern etc. In this particular case it would be of relevance to look for certain peak heights for instance. Or certain small impurities’ Area %. So, you’ll need to find suitable arguments that make your SST adequate in the context.

if you check one particular wavelength , the diffraction grating adjusts itself to give correct wavelengths at all other wavelengths.

Only if the grating moves to the required position. And that is what you need to show/validate. Actually I’ve seen (not that infrequently) problems with wavelength accuracy and especially in the lower end of the UV range e.g. 200 nm or so

Best Regards

[DR]

re: <200nm range - Just about everything absorbs at such low wavelengths, so it is usually impractical to run that low. Even oxygen absorbs below 200nm, so unless you're willing to nitrogen sparge your detector's optical bench and have to deal with lots of extraneous peaks, you probably don't want to run your detectors below 200nm anyway.

[HW Mueller]

Just wonder what sort of crap is in some laboratories. I have been doing UV since the early 1960´ies and have never seen any apparatus which varies any near as much as the max wavelenths of compounds. I go completely with Tom on this.

[juddc]

I'm with Tom as well. Linearity and reproducibility are, in my humble opinion at least, significantly more important than absolute wavelength accuracy. Furthermore, if the machine is easily within specification at 3 disparate wavelengths, I wouldn't expect error to increase exponentially in other regions. It's possible, I suppose, but I would think that a well validated method run on a machine that's appropriate for the analysis and has been qualified according to manufacturer's instructions should suffice. One can ALWAYS find nits to pick and the relevant issue to me is whether or not the source of error is likely to cause an OOS result. If a UV detector reads an absorption max at 212 nm that is actually 210 nm, does it really matter as long as it does so consistently? I think not...Still, if you want to toss another test into the mix, I suppose it can't hurt.

[danko]

I have been doing UV since the early 1960´ies and have never seen any apparatus which varies any near as much as the max wavelenths of compounds.

I, on the other hand, have never seen max wavelengths varying spontaneously. Unless of course, some alterations of the compounds’ environment are introduced – e.g. pH, temperature, solvent polarity etc. - which of course is the opposite of spontaneous.
Another possibility I can think of is instrument performance instability
In order to offset such sources of variation, there are these wonderful guidelines; GLP/GMP, which are usually helpful in ensuring reliable results etc. The same guidelines address the reliability of instruments’ performance as well. Some of the procedures, recommended in that context are different types of qualification (IQ, OQ and PQ) as well as periodical control (annual, or semiannual or whatever).
Otherwise, I would agree that, from time to time, one can experience appalling exaggerations of the above mentioned points. That’s why we are condemned to eternal need for common sense.

Best Regards