Interference Using PDA and fluorescence detectors in series

[Gregory McLaughlin]

To whom it concerns,

I am working with a drug that we analyze using UV detection by PDA and fluorescence detectors in series. The normal analytical wavelength is 228 nm. The fluorescence excitation wavelength is 228 nm with emission at 305 nm. I noticed that when using analyzing the UV signal at 228 nm for this compound (a strong fluorescent molecule) that the UV response curve was very non-linear (almost a quadratic response). I also reanalyzed the same mixture on a simple variable wavelength detector and the UV response curve was very linear (R2=1.0000 with a linear data fit). I believe the native fluorescence of the molecule is causing the UV to go non-linear using the PDA.

I want to understand this better. Do you know of a technical brief or reference to papers that go into details on this phenomena- specifically how native fluorescence impacts linearity of UV detection with a diode array detector?
Thanks for your help!

Gregory McLaughlin
gmclaughlin@pearltherapeutics.com

[HW Mueller]

My interpretation is that there is something grossly wrong with your PDA, or that it is set up wrong (ref. wavelength?). With an exceedingly strong fluorescence (high quantum yield) you may get a wrong absorbance, but I don´t see why it should be strongly nonlinear, unless there was quenching at the higher concentrations. If that was the case then it would show up in any UV spectrometer/detector.

[danko]

Could it be that the bandwidth on the PDA-detector is set too large? If it is (i.e. some of the florescence radiation at 305 nm is measured by the diode array in addition to the radiation at 228 nm) then the absorbance measured would be reduced progressively, following the increasing analye concentration.
The test of this hypothesis would be: Reduce the bandwidth by half or something like that and see if it results in a (more) linear fit.

Best Regards

[HW Mueller]

Danko, could you please explain your reduction of absorbance?

[danko]

Hans, absorbance signal is actually less light reaching the detection unit (in this particular case the diode array) compared to a reference light beam that does not transverse the flow cell. Fluorescence signal is the opposite of absorbance (i.e. more light reaching the detection unit).

So, the PDA detector is reading absorbance, which is increasing with analyte concentration increase.
But the incident light beam is also exciting the molecule which results in light emission (fluorescence). So if the bandwidth is large enough, the fluorescence (in this case light at 305 nm) will be detected as well – but as more light (i.e. less absorbance) even though the detection wavelength is set to 228 nm. And because more light means less absorbance, the absorbance readings will be reduced with increasing analyte concentrations and the resultant standard curve will be nonlinear.

Best Regards

[Uwe Neue]

Is your response curving upward (positive squared term) or downward (negative squared term)?

[Gregory McLaughlin]

To begin with thanks for some of the suggestions. They are all interesting to consider. To clarify a few things the PDA bandwidth was set at 8 nm,
The reference wavelength of the PDA is set at 400 +/-20 nm. The absorbance spectrum of the fluorescent molecule has a UV maxima near 280 and drops off to near 0 at about 300 nm. I measure at 228 nm because I need more sensitivity for a number of other related substances and another API of interest in the product.

The absorbance response curve for the fluorescent molecule is concave shaped and fits fairly well with a quadratic function.

I will test of this hypothesis to reduce the bandwidth by half or something like that and see if it results in a (more) linear fit.

I think that the explanation that Dancho suggests concerning absorbance and fluorescence signals seems to fit what I see.

[Uwe Neue]

Based on your description, I suspect that the fluorescence adds to the reference signal, and that this is causing the curvature.

I do not know what you can do with the reference signal... Can you turn it off? Select a higher wavelength?

[Gregory McLaughlin]

Today I found a good description of these phenomena in an Agilent UV-visible spectroscopy book listed under photochemical problems - Fluorescence. Part of what it describes is summarized, as follows: "As the excitation wavelength is scanned, absorption occurs, initiating the fluorescence process that emits light at longer wavelengths. Because the detector cannot differentiate among the individual wavelengths, the absorbance measured at the excitation wavelength is too low. As the emission wavelength range is scanned, no fluorescence occurs, and the adsorption measurements are accurate. ....An additional factor that effects the magnitude of the error is the so-called acceptance angle of the detector...The fluorescent light is emitted in all directions. If the acceptance angle is wide, a significant portion of the fluorescent light will reach the detector."

[Alex Buske]

Hi,

Generally I support Uwes statement. Did you ever measure a reference WL chromatogramm?
I think 8 nm bandwith is o.k. Our VWDs have the same and team up perfectly with FLDs.
To the text: In contrast to UV spectrometers PDAs emit at all WLs and detect the different WLs simultaniusly. So the excitation WL is NOT scanned and the detector CAN differentiate among the individual wavelengths.
It looks like some of the FL emission goes up to 400 nm, causing a decrease in the reference signal and intraPDA correction leads to signal disturbance.

Alex

[HW Mueller]

Sorry, I don´t follow Danko nor Gregory on this. As I stated before, unless there is considerable quenching the ratio of fluorescence to absorbance is constant, otherwise one could forget about fluorescence for quantitation. Also, all UV apparati would give very similar results, unless one uses ref wavelength, another not. So it is back at what Uwe and I said about the ref. wavelength.
Is there a max at 228nm? What is the absorbance relative to that at 280nm? What does the emission spectrum look like when you excite at 228 as compared to 280nm?

[HW Mueller]

Darn, got "forbiden"again, so came in after Alex, anyway that makes three going for ref. wavelength.

[danko]

As I stated before, unless there is considerable quenching the ratio of fluorescence to absorbance is constant, otherwise one could forget about fluorescence for quantitation

Nice try Hans. The ratio between fluorescence and absorbance is far from constant. Both the UV and fluorescence curves might be linear but their slopes are most frequently far from equal.
That’s why the situation I described is quite probable.
At the time I posted my thoughts, I didn’t have all the details and certainly not the detector brand and model. Agilent PDA 1100/1200 is different from others and one of the differences is the the specific to Agilent reference beam that goes through the flow cell in addition to the traditional reference beam, that corrects for energy fluctuations only and is just a split fraction of the detection beam.
The Agilent corrective reference beam is going, as mentioned before, through the flow cell and should correct for eluent absorbance or refractive index changes. If the light signal increases (i.e. absorbance decrease), the detector will correct this by "adding" some “absorbanceâ€

[HW Mueller]

Danko, ever heard of quantum yield?

[danko]

Yes I have. But what is your point?
If you’re wondering what causes the slope differences I mentioned in my previous post, it’s easy to explain. Actually you almost had it, when you mentioned quenching. The only thing is, you assume that quenching is an on/off phenomenon. In fact quenching is an integral part of fluorescence. There are mainly two reasons for inherited quenching, that can not be offset – completely anyway – regardless of concentration etc. (but they affect the response increasingly with concentration increase). Although low concentrations and other initiatives will be helpful, there will always be some quenching due to:
1. The inner-filter effect, caused by re-absorption of the emitted radiation due to excitation and emission wavelength overlap.
2. Collisional quenching, due to energy loss upon intermolecular collisions.
There could be other factors, but they are easier to compensate.
Needless to say, these factors will cause slope differences between UV (absorption) and Fluorescence (emission). And that is also the main reason for the smaller linear range in fluorescence quantitation, compared to the UV absorption mode.

Best Regards