Integration of fluorescence quenching correction into trihalomethane formation prediction models

Despite there being numerous models of trihalomethane (THM) formation, they are limited by high estimation errors, which can be close to the regulatory limits for THMs, due to the fluorescence quenching effect. In this research, the estimation error for THM formation was reduced by correcting the quenching effect. The trihalomethane formation potential (THMFP) test was conducted in the presence of chlorine and bromine, individually and in mixtures. The THM precursors used in this study were protein (bovine serum albumin; BSA), amino acids (tryptophan and tyrosine), chlorine, bromine, and Suwannee River natural organic matter (SWNOM). BSA tended to form bromodichloromethane (BDCM) rather than trichloromethane (TCM) during chlorination in the presence of bromide (Br^-). In contrast, SWNOM tended to form chlorinated THMs (TCM) rather than brominated THMs (BDCM and dibromochloromethane; DBCM), and no TBMs were formed in these processes. BSA with SWNOM decreased the formation of TCM due to the decrease in the amount of TCM precursor in SWNOM through binding with BSA. The concentration of each THM species was predicted from the fluorescence intensity of peak C, corrected fluorescence intensity of peak T, and Br^- concentration. The use of humic-like and corrected protein-like fluorescence in the excitation-emission matrix model for predicting THM species reduced the prediction error. In this research, correction of the fluorescence quenching decreased the mean percentage estimation error for TCM, BDCM, and DBCM from 47%, 35%, and > 100% in classical approaches to 6.6%, 26.9%, and 2.0%, respectively. This study is expected to make contributions in reporting the relationship between the concentration of natural organic matter compositions and the formation of THM species.