Removal of disinfection by-product precursors in drinking water treatment processes: Is fluorescence parallel factor analysis a promising indicator?

Abstract

This work investigated the removal efficiency of disinfection by-product (DBP) precursors by different drinking water treatment processes and evaluated the feasibility of using fluorescence components removal as an indicator. A four-component (including tryptophan-like, protein-bound, tyrosine-like, and humic-like components) parallel factor analysis model was developed basing on 288 fluorescence excitation-emission matrices. Among all treatment processes, coagulation-sedimentation process showed the best performance, with mean removal ratios of 30% in total fluorescence intensity and 31% in total formation potential (FP) of DBPs, respectively. It preferentially removed humic-like component C4 (43%). Advanced treatment processes were less effective in comparison. Ozone and biological activated carbon (BAC) combined process reduced 20% of total fluorescence intensity, while ultrafiltration process reduced < 3%. Ozonation and BAC filtration preferentially removed free amino acids (i.e., C1 and C3) and protein-bound (i.e., C2) components, with mean removal ratios of 12% and 17%, respectively. Significant correlations (p < 0.01, double-tailed) were observed between four fluorescence components removal and FPs reduction of three trihalomethanes, dichloroacetonitrile (DCAN), and 1,1-dichloropropanone (1,1-DCP). Specifically, the correlation coefficients for three trihalomethanes and 1,1-DCP followed the order of C4 > C1 > C2 > C3, while the order for DCAN was C2 > C4 > C1 > C3.