Département Ressources aquatiques et eau potable

Water treatment with nanofiltration (NF) or reverse osmosis (RO) membranes results in a purified permeate and a retentate, where solutes are concentrated and have to be properly managed and discharged. To date, little is known on how the selection of a semi-permeable dense membrane impacts the dissolved organic matter in the concentrate and what the consequences are for micropollutant (MP) abatement and bromate formation during concentrate treatment with ozone. Laboratory ozonation experiments were performed with standardized concentrates produced by three membranes (two NFs and one low-pressure reverse osmosis (LPRO) membrane) from three water sources (two river waters and one lake water). The concentrates were standardized by adjustment of pH and concentrations of dissolved organic carbon, total inorganic carbon, selected micropollutants (MP) with a low to high ozone reactivity and bromide to exclude factors which are known to impact ozonation. NF membranes had a lower retention of bromide and MPs than the LPRO membrane, and if the permeate quality of the NF membrane meets the requirements, the selection of this membrane type is beneficial due to the lower bromate formation risks upon concentrate ozonation. The bromate formation was typically higher in standardized concentrates of LPRO than of NF membranes, but the tradeoff between MP abatement and bromate formation upon ozonation of the standardized concentrates was not affected by the membrane type. Furthermore, there was no difference for the different source waters. Overall, ozonation of concentrates is only feasible for abatement of MPs with a high to moderate ozone reactivity with limited bromate formation. Differences in the DOM composition between NF and LPRO membrane concentrates are less relevant than retention of MPs and bromide by the membrane and the required ozone dose to meet a treatment target.

Peracetic acid (PAA) in combination with transition metals has recently gained increasing attention for organic micropollutant abatement. In this study, aqueous Co(II), Cu(II), and Ag(I) were compared for their capacity to activate PAA. Co(II) outperformed Cu(II) or Ag(I) and the optimum conditions were 0.05 mM of Co(II), 0.4 mM of PAA, and pH 3. However, due to a wider applicability in water treatment, pH 7 (i.e., bicarbonate buffer) was selected for detailed investigations. The abatement of different micropollutant surrogates could be described with a second-order rate equation (observed second-order rate constants, k_obs were in the range of 42-132 M^-1 s^-1). For the para-substituted phenols, there was a correlation between the observed second-order rate constants of the corresponding phenolates and the Hammett constants (R² = 0.949). In all oxidation experiments, the reaction rate decreased significantly after 1-2 min, which coincided with the depletion of PAA but also with the deactivation of the Co(II) catalyst by oxidation to Co(III) and subsequent precipitation. It was demonstrated that Co(II) immobilized on a geopolymer-foam performed approximately similarly as aqueous Co(II) but without deactivation due to Co(III) precipitation. This provides a potential option for the further development of heterogeneous catalytic Co(II)/PAA advanced oxidation processes utilizing geopolymers as a catalyst support material.

Arsenic (As) is a toxic element, and elevated levels of geogenic As in drinking water pose a threat to the health of several hundred million people worldwide. In this study, we used microfluidics in combination with optical microscopy and X-ray spectroscopy to investigate zerovalent iron (ZVI) corrosion, secondary iron (Fe) phase formation, and As retention processes at the pore scale in ZVI-based water treatment filters. Two 250 μm thick microchannels filled with single ZVI and quartz grain layers were operated intermittently (12 h flow/12 h no-flow) with synthetic groundwater (pH 7.5; 570 μg/L As(III)) over 13 and 49 days. Initially, lepidocrocite (Lp) and carbonate green rust (GRC) were the dominant secondary Fe-phases and underwent cyclic transformation. During no-flow, lepidocrocite partially transformed into GRC and small fractions of magnetite, kinetically limited by Fe(II) diffusion or by decreasing corrosion rates. When flow resumed, GRC rapidly and nearly completely transformed back into lepidocrocite. Longer filter operation combined with a prolonged no-flow period accelerated magnetite formation. Phosphate adsorption onto Fe-phases allowed for downstream calcium carbonate precipitation and, consequently, accelerated anoxic ZVI corrosion. Arsenic was retained on Fe-coated quartz grains and in zones of cyclic Lp-GRC transformation. Our results suggest that intermittent filter operation leads to denser secondary Fe-solids and thereby ensures prolonged filter performance.