Fluorescence linearity problems at Low concs?

[tayzyboy]

We are trying to calibrate over a large range (1-10,000ng/ml) using HPLC with fluorescence. However the curve seems to be non-linear in a way I don't understand. The curve appears to be sigmoidal (but we dont go high enough to see the upper levelling off).

I've seen the levelling off at high concentrations before, but never the sigmoidal effect at the bottom end of the curve.
Does this normally happen with flourescence? or should I be investigating my preparation techniques (serial dilutions 1 in 2).
I use an internal standard, but as the peak size of this is very nearly identical on each injection it doesnt really make any difference to the data if I use it or not.

Thanks for any help you can give.

James

[ivanvins]

What is your signal to noise ratio at the low concentration range? I have no much experience with fluorescence detectors linearity, however as they tend to be noisy, this effect may be created by the peak integration parameters. Other common problem is fluorescence quenching (by oxygen in mobile phase most often) - however I have no idea of its effect on response linearity.

[Mark Tracy]

Check your blank. You might have a carryover problem, a coeluting interference from your sample prep, instrument contamination. It is not specific to fluorescence, but since fluorescence is so sensitive, one tends to work in the concentration range where this kind of problem is more common. I have had this problem myself with UV.

[tayzyboy]

our signal to noise is over 10 to 1 at the lowest point.
Blanks are very clean - no carryover is detected

Further experiments with independant manual integration has shown this effect to be consistant.

We then went on to weigh each dilution (to check pipettes and human error) but all dilutions were within 1% of expected values, so that should be ok too.

We then tried diluting the calibration curve 1in3. This obviously gives less data points, but for some reason increases linearity greatly.

Very odd problem, Id like to spend more time on it, but I think due to deadlines we will have to stick with the 1in3 dilution.

[HW Mueller]

If your "sigmoidal" means that the lower values were relatively high and since dilution improved the situation it would seem to indicate that you had quenching of the fluorescence, especially at higher concentrations. It could be selfquenching or another substance in the St/sample which is "co-diluted"..