2.2. Photocatalytic Degradation Experiments

Dead-end filtration and degradation experiments were performed in a custom photocatalytic membrane reactor (PMR) consisting of a TiO₂-coated alumina membrane within a PMMA holder, placed inside a cupboard to protect the micropollutants from the reaction with ambient light (see Heredia Deba et al. [16] for details on the setup configuration and the PMR). For the cross-flow experiments, a stainless steel PMR with the same structure but without a feed reservoir was employed. In the stainless steel module, the space on top of the membrane is limited by the o-ring (EPDM 25 × 1.5, Eriks) thickness after closing the module.

For the experiments with the MPs independently (singles), the aqueous solutions were pumped into the setup with fluxes in the nanofiltration range, 1.6, 3.3, 6.5, 9.7, 13.0, and 16.2 L·m−2·h−1 and additionally for DCF 19.5 and 21.1 L·m−2·h−1; these experiments were repeated three times to analyze the reproducibility of results. For the experiments with the mixtures, three fluxes (1.6, 6.5, and 16.2 L·m−2·h−1) and two repetitions were investigated, as the reproducibility of the experiments was high.

The feed concentration varied across experiments and micropollutants. For the experiments with the single MP and for those in a mixture, the initial concentration was 2 mg·L−1 DCF, 1 mg·L−1 INN, 4 mg·L−1 MB, and 1 mg·L−1 MTP (molar concentration [μmol·L−1] ratio 6.8:1.3:12.5:3.7 DCF:INN:MB:MTP). For the experiments with low concentrations (LC) the feed was 2 μg·L−1 DCF, 6 μg·L−1 INN (3 μg·L−1 in TW), 4 μg·L−1 MB, and 2.5 μg·L−1 MTP (molar concentration [nmol·L−1] ratio 6.8:7.7(3.9):10:9.4 DCF:INN:MB:MTP). For experiments regarding the effect of background water constituents, the initial concentration in the mixture was 45 mg·L−1 (0.7 mM) and 215 mg·L−1 (3.5 mM) of bicarbonate ions, and 61 mg·L−1 (1.7 mM), and 607 mg·L−1 (17.1 mM) of chloride ions. A concentration of 142 mg·L−1 (1 mM) of sodium sulfate was used as background in all the singles experiments and the experiments with bicarbonate to avoid corrosion in the system. For the experiments with sodium chloride and low concentrations of micropollutants, sodium bicarbonate with a concentration of 23.4 mg·L−1 (0.3 mM) was used. In the experiments with TW, no extra background was added. The chemical composition of the mixture in TW can be found in Appendix A. The natural pH of the system was used without further adjustment. For the experiments with sodium sulfate in the background, the pH was between 6 and 7, except for the experiments with bicarbonate, in which the pH was approximately 8. For the experiments with bicarbonate in the background, the pH was between 7 and 7.5. In the experiments with tap water, the pH was 8. A table summarizing the measured feed solution and permeate pH can be found in Appendix B.

Two photocatalytic membranes were used for the experiments named A and B. Membrane A was utilized for the singles experiments, and membrane B for the experiments with the mixtures. In order to compare the photocatalytic properties of both membranes, the experiments with MB and MTP were reproduced, and the data confirmed that the performance of both membranes was similar. More data about the membrane fabrication can be found in our previous work [16], as we used the same titanium dioxide suspension (Evonik, VP Disp. W 2730 X) and deposition technique (dip coating). Results on the morphology of membranes A and B can be found in Appendix C.

Before each experiment, the membranes are equilibrated with the feed solution for at least 120 min at 16.2 L·m−2·h−1 to ensure adsorption equilibrium before the degradation measurements. This is because the pre-adsorption of reactants on the surface of the TiO² membrane may lead to a more efficient electron-transfer process [17,18]. After equilibrating, the LED was turned on, and samples were taken from the permeate every 30 min, except for MB, which was continuously monitored. The experiments were finalized when the outlet concentration reached a steady value for each filtration rate.

The input radiation level was set before each run to 210 W·m−2 and measured by using a power meter (Thorlabs) with a thermal power sensor head (S310-C). Control experiments with MB were carried out with a membrane without the TiO₂ layer to rule out effects other than photocatalytic oxidation, e.g., bulk photolysis. It should be noted that none of the used MPs absorb photons in the used wavelength (λmax = 366 nm) (Appendix D), and hence no direct photolysis of the MPs is expected to take place in our system (Grotthuss–Draper law). Photolysis of DCF is generally reported in studies utilizing direct sunlight [19,20] or lamps emitting polychromatic light, including those using filters restricting the transmission of wavelengths below 290 nm [21,22]. The overlap with the absorption spectrum of DCF (λmax = 194 nm) with a shoulder absorbance of up to ∼320 nm could explain this effect. Martínez et al. [23] reported photolysis upon near-UV-Vis irradiation (mainly at 366 nm), and Calza et al. [24], who used a xenon arc lamp and special glass filter to restrict the emissions below 290 nm, and Rizzo et al. [25], who used a black light fluorescent lamp emitting radiation between 300 and 420 nm, did not report any significant DCF degradation via photolysis.