Fluorescence Approach for the Determination of Fluorescent Dissolved Organic Matter

Excitation–emission
matrix (EEM) fluorescence spectroscopy
coupled with parallel factor (PARAFAC) analysis has been widely applied
to characterize dissolved organic matter (DOM) in aquatic and terrestrial
systems. However, its application in environmental samples is limited
because PARAFAC is not able to handle nontrilinear EEM data, leading
to the overestimated number of components and incorrect decomposition
results. In this work, a new method, parallel factor framework-clustering
analysis (PFFCA), is proposed to resolve this problem. First, simulated
data with different signal-to-noise ratios and intensities of nontrilinear
structure were tested to confirm the robustness of PFFCA. The residual
sum of squares (RSS) of PARAFAC was significantly higher than that
of PFFCA (p < 0.037). Second, a set of data originating
from a synthetic mixture of humic acid and bovine serum albumin was
applied to compare with PARAFAC with known samples. PFFCA provided
an estimation (R2 > 0.92) closer to
actual
EEM than PARAFAC (R2 > 0.81). Finally,
to confirm the feasibility of PFFCA in analyzing natural samples,
DOM-containing samples collected from both a polluted lake and river
were tested, indicating that PFFCA provides a more precise estimation
than PARAFAC. The results clearly indicate that PFFCA offers a robust
approach for the unique decomposition of complex synthetic and natural
samples, which is of great significance in understanding the characteristics
of DOM in aqueous systems.