Fluorescence spectroscopy: considerations for highly absorbing dissolved organic matter samples

Fluorescence spectroscopy is a robust method for characterizing organic matter (OM). However, proper collection and correction of spectra are necessary to provide useful data. One important correction is the inner-filter correction, which primarily accounts for the inner-filter effect by adjusting for the wavelength dependent attenuation of emitted light by the solution prior to detection by the fluorometer. The most commonly used correction is based on an assumption that light is emitted at the center of the pathlength. Thus, the inner-filter effect is more pronounced in highly absorbing samples, and has the potential to skew the fluorescence spectra. For this study, the terrestrially derived Suwannee River fulvic acid (SRFA) and microbially derived Pony Lake fulvic acid (PLFA), from the International Humic Substances Society (IHSS), were diluted to incremental absorbances at a wavelength of 254 nm from 0.05 to 1.0 at pH 4 and 7. Three dimensional fluorescence spectra were measured and modeled with the Cory and McKnight (2005) parallel factor analysis (PARAFAC) model which resolves the fluorescence spectra into 13 components, including quinone-like and protein-like components. In the absence of inner-filter effects, plots of absorbance vs. loadings should be linear. Using the data from absorbance of 0.05 to 0.3, where the inner-filter affect is least pronounced, a linear regression was created and used as a baseline to predict component loadings at higher absorbance values in the absence of inner-filter effects. Results indicate that at absorbance values greater than 0.3, the commonly-used inner-filter correction is not able to remove the inner-filter effect. Therefore, in order to obtain reliable component loadings and correctly interpret the spectra, samples should be diluted to absorbance values less than 0.3 at 254 nm prior to collection of three dimensional fluorescence scans. The recommendation of a maximum absorbance of 0.3 agrees with the results of a study by Ohno (2002), which investigated several simpler fluorescence metrics. Suwannee River fulvic acid, pH 4, absorbance at 254nm vs. loading of component Q2. At the higher absorbances the component loadings are over-predicted compared to the linear regression, estimated from data with absorbance below 0.3, where the inner-filter effect is minimal.