Detectors and wavelength resolution

[shaun78]

I have been struggling with this one for a while. Something is just not sitting well with me about it. So, please let me know if I am either nuts or having the woll pulled over my eyes.

We have some old detectors (Perkin Elmer 235C's if anyone cares). So, the wavelength resolution/bandwidth on these detectors is +/- 5 nm, meaning that I can enter in 250 nm, 255 nm, 260 nm, etc. but not something like 253 nm as the detector will round the number up to 255 nm.

Now, when these instruments are calibrated against an anthracene standard I am being told that the calibration specification is 250 nm +/- 1 nm. I understand the 250 nm when the lambda max of anthracene is actually 251 nm is due to the bandwidth of the 235C. However, what I do not unbderstand is how we can state +/- 1 nm, when the detector only "sees" in 5 nm incraments.

Like I said, what am I missing or is someone trying to trick me?

[AICMM]

shaun78,

Just a guess, but it could be that the grating driver you are using (especially since they are older) uses an optical encoder so it can only find the 5 nm marks on the encoder wheel but that the grating has a resolution of 1 nm. Like I say, just a guess.

Best regards.

[Bruce Hamilton]

I don't know the instrument in question, but i'd expect the manual would clarify some of the issues you have. If anthracene is typically used, details should be in the manual.

Firstly, I'm surprised anthracene is used. If I recall correctly, the peak is asymmetric. Secondly, I suspect the issue may be the difference between the optical performance testing and the normal output.

For example, some instruments allow setting 1 or 2 nm spectrum slices for routine use so data handling is manageable, however a calibration mode is provided that uses the highest resolution and may even iterpolate between several points on a symmtrical or known peak profile to calculate the reported peak apex to 0.1 nm resolution or better, and report deviations in the same units.

[shaun78]

AICMM: that sounds very close to what a service tech told me when I asked. However, if this were the case, then shouldn't it be possible for me to collect spectra from a wavelength range and obtain a lambda max of 253? When spectra is collected and maximums are determined these 235C's always give a maximum that is a multiple of 5. Again, which makes me think that the resultion must be limited to 5 nm.

Bruce: As far as I can tell, the minimum spectrum slice for the detector is that 5 nm. There is no sort of enhanced calibration mode that I am aware of. There is going to be another tech out here today from Perkin Elmer, so I can ask to be sure...

Thanks for both your replies!

[tom jupille]

I suspect that the writer copied that part of the manual over from the documentation for another detector that *does* resolve to +/- 1 nm and nobody at PE noticed the discrepancy.

[Alex Buske]

As far as I remember, one could run the 235 as a tuneable UV detector on any wavelength.
It is just that the old boxes had (or still have) a limited number of diodes (e.g. each every 5 nm). They are simply interpolating. I still can run the Waters 2996 on any wavelength (249, 250, 251 nm), even though the distance between the diodes is 1.2 nm.
The spectra delivered by the 235 had always peak maxima different from the 5 nm steps (interpolated). If that was accurate to +/- 1 nm I am really not sure.

Alex

[DR]

I think that they are fairly accurate as a result of decent interpolation (curve fitting), so even when you're lambda max is between the pegs, you can get a pretty close approximation of the true max wave number for a particular spectrum.