Formation de sous-produits de désinfection lors de la chloration de l’eau potable : identification des précurseurs de matière organique et des dangers associés

Disinfection by-products (DBPs) form when dissolved organic matter (DOM) in water reacts with disinfectants, leading to regulated compounds such as trihalomethanes (THMs) and haloacetic acids (HAAs). However, unregulated DBPs like iodinated THMs/HAAs, and nitrogenous DBPs, such as haloacetonitriles (HANs) and haloacetamides (HAMs), pose higher health risks, as studies with Chinese Hamster Ovary cells have shown. To control DBP formation in tap water, researchers focus on identifying DBP precursors by fractionating DOM using resins or membranes, though results vary widely. In this thesis, a meta-analysis compares the reactivity of DOM fractions towards (un)regulated DBPs. The meta-analysis reveals that hydrophobic compounds exhibit 10-20% higher reactivity to both THM and HAA formation compared to hydrophilic compounds in waters with high specific ultraviolet absorbance at 254nm (SUVA254 >2). In waters with low SUVA254, both hydrophobic and hydrophilic compounds have equal reactivity. However, hydrophilic compounds are 20-80% more reactive towards unregulated DBPs. Neither chlorination time nor dose alters this reactivity ratio, but an increase in column capacity factor does. Additionally, dead-end ultrafiltration membranes may not always separate fractions sharply, leading to inconsistent results for DBP formation across groups. To address this, a novel membrane fractionation method was developed. This method was first validated using a size exclusion chromatographic approach (HPSEC) to ensure reliability, considering key factors such as detection limits (19.0µgC dm-3) and the importance of preserving samples for a maximum of two weeks at 4°C. Inorganic carbon removal was achieved by acidifying the sample to pH 6 and purging it with N2. The HPSEC-TOC method proved reliable for DOM fraction analysis in both treated and untreated water sources. Using this method, DOM was separated into three molecular weight (MW) fractions: high (>20kDa), medium (0.3-20kDa), and low (<0.3kDa). A mathematical tool was developed to optimize the protocol and predict the diafiltration factor for each fraction, ensuring consistent results. The method was applied to surface water across three seasons, yielding a fraction up to 50% high MW compounds, a fraction over 80% medium MW compounds, and a fraction containing only low MW compounds. Chlorination and chloramination tests (24h, 1mgCl2 L-1 residual) on these fractions assessed the formation of THMs, HAAs, HANs, HAMs, and nitrosamines. Results showed that the medium MW fraction had the highest overall reactivity. However, the low MW fraction was highly reactive toward di-iodinated THMs, and its reactivity increased with increasing chlorine atoms in the HAA, HAN, and HAM families. The high MW fraction showed the lowest reactivity towards most DBPs, except iodoacetic acid and chloroacetonitrile. Fluorescence spectroscopy further revealed that DBP precursors' chemical properties vary by fraction and disinfection method, suggesting complex reaction mechanisms. A site-specific strategy is recommended to identify the most concerning DBPs, considering both their toxicity and concentration. Finally, the thesis explores the impact of reactive oxygen species on inflammation and oxidative stress pathways induced by HAAs and HAMs. In human colorectal adenocarcinoma (Caco-2) cells, lethal concentrations ranged from 12µM for iodoacetamide and iodoacetic acid to 47mM for dichloroacetamide. Primary intestinal epithelial cells from donors with colorectal cancer, Crohn's disease, and healthy controls were also exposed to HAAs and HAMs. HAAs triggered oxidative stress and inflammatory responses in cells adjacent to cancerous tissue, while HAMs induced these effects across all cell types. These findings suggest a potential link between DBP exposure and inflammatory bowel disease development, although further validation through RNA sequencing is needed.