Very High Standard Deviation in Excitation Emission Matrix Measurement

[JonasFnr]

TL;DR Summary

I get very high standard deviations for phenolic compounds in EEM measurements of olive oil. Any thoughts/input highly appreciated.

Hi,

I obtain really high standard deviations in Excitation-Emission Spectra mainly for the phenolic compounds in olive oil (Em: 290-350nm).

Method:
I weigh 0.05g of olive oil and dilute it up to 25ml with cyclohexane to remain in the range of linearity for absorbance measurements to correct for filter effects.
To estimate the standard deviation of measurements, I made five equivalent samples and measured each five times.

Both within and between the samples the s.d. is very high in this area (13-20%).
All validity and calibration tests I have tried thus far seem to indicate that the device is working accurately.

I get those results both with a Shimadzu RF-6000 and an Aqualog (which already corrects for filter effects) from 200-800nm.

Settings for the Shimadzu: Datainterval: 2nm each, Scanspeed 6000nm/min, Ex. Bandwith 5nm, Em. Bandwith 3nm, Sensitivity: High.

I don't think the solvent, contamination, scattering, runtime-dependent spectrometer performance or photodecay of the phenols could explain this.
Is the proximity to the rayleigh scattering or are the settings an issue here?
I couldn't obtain a viable spectrum with any other settings for such low concentrations though, which I need to correct for filter effects.

I couldn't find anything on this in the literature, so any thoughts or input would be HIGHLY appreciated.

Thank you and kind regards.

 

[chemisttree]

What do you estimate for S/N?

 

[JonasFnr]

S/N ratio is in the normal range according to the device manufacturer (water raman peak).

 

[chemisttree]

Have you analyzed a blank?