Organic micropollutant removal during river bank filtration

This study investigated the factors influencing the main removal mechanisms (adsorption and biodegradation) for organic micropollutant (OMP) removal during river bank filtration (RBF) and the possibility of developing a predictive model of this process for OMP removal during RBF. Chapter 2 analysed the sorption and biodegradation behaviour of 14 OMPs in soil columns filled with technical sand (representative of the first meter of oxic conditions in RBF systems. Breakthrough curves were modelled, based on the advection-dispersion equation, to differentiate between OMP sorption and biodegradation. Retardation factors (indicators for OMP sorption) for most compounds were close to 1, indicating little sorption of these compounds and thus the mobile behaviour of these compounds during passage in soils. The influence of active and inactive biomass (bio-sorption), sand grains and the water matrix on OMP sorption was found to be negligible under the conditions investigated in this chapter. Although trends were observed between charge or hydrophobicity of charged OMPs and their biodegradation rates, a statistically significant linear relationship for the complete OMP mixture could not be obtained using these physico-chemical properties. However, a statistically significant relationship was obtained between OMP biodegradation rates and the functional groups present in the molecular structure of the OMPs. The presence of ethers and carbonyl groups increased biodegradability, while the presence of amines, ring structures, aliphatic ethers and sulphur decreased biodegradability. Chapter 3 examined relationships between functional groups present in the molecular structure of a mixture of 29 OMPs (of which 11 were the same as in Chapter 2) and their biodegradation rates obtained from lab-scale soil columns and constructed a multi-linear regression model for biodegradation rate prediction based on this. This model was then validated with field data. In contrast to Chapter 2, where technical sand was used to fill the columns, lab-scale columns here were filled with soil from an operational RBF site. A statistically significant relationship was found between OMP biodegradation rate and the functional groups present in the molecular structures of the OMPs. OMP biodegradation rate increased in the presence of carboxylic acids, hydroxyl groups, and carbonyl groups, but decreased in the presence of ethers, halogens, aliphatic ethers, methyl groups and ring structures in the molecular structure of the OMPs. Differences between the predictive models obtained in Chapter 2 and 3 could be explained by the different soil types and water qualities used (Schie Canal water and technical sand in Chapter 2, Lek River water and soil from an operational RBF site in Chapter 3). The predictive model obtained from the lab-scale soil column experiment in Chapter 3 gave a good indication of biodegradability for approximately 70% of the OMPs monitored in the field (80% excluding the glymes). The model was found to be less reliable for the more persistent OMPs (OMPs with predicted biodegradation rates lower or around the standard error = 0.77 d-1) and OMPs containing amide or amine groups. These OMPs should be carefully monitored in the field, to determine their removal during RBF. Water quality was reported to be an important factor in OMP removal during soil passage, however it is unclear if this is the only important factor and therefore Chapter 4 explored the effect of soil type on OMP removal. Sorption and biodegradation behaviour of 20 OMPs was investigated in lab-scale columns filled with two different soil types and fed with the same water quality - the columns were simulating RBF under oxic conditions. Differences in retardation factors and OMP biodegradation rates were statistically not significant between the two soil types, although these soil types were characterized by a different cationic exchange capacity, organic matter and sand/silt/clay content. This result was supported by the microbial community composition (richness, evenness) of the two soils that became more similar during the course of the experiments as a result of feeding both columns with the same water quality. This indicates that microbial community composition and thereby OMP removal in soils is primarily determined by the aqueous phase (organic matter quantity and quality, nutrients) rather than the soil phase. These results imply that different RBF sites located along the same river may show similar OMP removal (in case of similar water quality and residence time). Chapter 5 investigated the effect of the water quality in more detail, and more precisely the effect of different organic carbon fractions (hydrophilic, hydrophobic, transphilic and the complete river water organic carbon) obtained from river water on the OMP biodegradation rate. Additionally, the effect of short-term OMP and DOC shock-loads (e.g. quadrupling the OMP concentrations and doubling the DOC concentration) on OMP biodegradation rates was investigated to assess the resilience of RBF systems to, for example, climate change. The results imply that – in contrast to what is observed for soil systems operating on wastewater effluent - OMP biodegradation rates during RBF are not affected by the type of organic carbon fraction (obtained from river water) fed to the soil column, in case of stable operation. No effect of a short-term DOC shock-load on OMP biodegradation rates was observed, for none of the different organic carbon fractions dosed. This means that the RBF site investigated in this chapter is resilient towards transient higher DOC concentrations in the river water (e.g. following a decrease in river discharge due to seasonal effects). However, a temporary OMP shock-load increased OMP biodegradation rates for the river water organic matter and hydrophilic organic carbon fractions. These increased biodegradation rates could not be explained by any of the parameters investigated in this chapter (ATP, DOC removal, SUVA, richness/evenness of the soil microbial population or OMP category (hydrophobicity/charge). The effect of redox condition on OMP biodegradation rate as well adaptive behaviour of a mixture of 15 OMPs (largely similar to the OMP mixtures used in Chapter 3, 4, and 5) in laboratory-scale soil columns fed with river water was analysed in Chapter 6. Dimethoate, diuron, and metoprolol showed redox dependent removal behaviour with degradation rates larger for the oxic zone compared to the suboxic/anoxic zone. OMPs that showed persistent behaviour in the oxic zone (atrazine, carbamazepine, hydrochlorothiazide and simazine) were also not removed under more reduced conditions. Adaptive behaviour was observed for five OMPs: dimethoate, chloridazon, lincomycin, sulfamethoxazole and phenazone. Newly developed, or existing, RBF sites exposed to these OMPs for the first time may require up to 9 months following start-up to reach full removal capacity. For some chemicals, such as dimethoate, even longer start-up times could be required since full removal capacity was not reached in our tests even after 15 months. The adaptation time observed for some OMPs could not be explained by their physico-chemcial properties (hydrophobicity, charge, molecular weight) or functional groups. Finally, adaptive behaviour of the biomass towards OMPs was found to be an important factor that should be incorporated in predictive models for OMP removal during RBF. Chapter 7 presents the conclusions, implications for the practical application and recommendations for future research