New Perspectives on the Interactions between Adsorption and Degradation of Organic Micropollutants in Granular Activated Carbon Filters

The removal of organic micropollutants in granular activated carbon (GAC) filters can be attributed to adsorption and biological degradation. These two processes can interact with each other or proceed independently. To illustrate the differences in their interaction, three 14C-labeled organic micropollutants with varying potentials for adsorption and biodegradation were selected to study their adsorption and biodegradation in columns with adsorbing (GAC) and non-adsorbing (sand) filter media. Using 14CO2 formation as a marker for biodegradation, we demonstrated that the biodegradation of poorly adsorbing N-nitrosodimethylamine (NDMA) was more sensitive to changes in the empty bed contact time (EBCT) compared with that of moderately adsorbing diclofenac. Further, diclofenac that had adsorbed under anoxic conditions could be degraded when molecular oxygen became available, and substantial biodegradation (≥60%) of diclofenac could be achieved with a 15 min EBCT in the GAC filter. These findings suggest that the retention of micropollutants in GAC filters, by prolonging the micropollutant residence time through adsorption, can enable longer time periods for degradations than what the hydraulic retention time would allow for. For the biologically recalcitrant compound carbamazepine, differences in breakthrough between the 14C-labeled and nonradiolabeled compounds revealed a substantial retention via successive adsorption–desorption, which could pose a potential challenge in the interpretation of GAC filter performance.


INTRODUCTION
−5 Empty bed contact time (EBCT) has been proposed as a key operating parameter to allow sufficient time for the adsorption of micropollutants in GAC filters 6 but could also influence the biodegradation of micropollutants by GAC biofilms.EBCTs typically range from 10 to 45 min, 7 which is considerably lower than the hydraulic retention time (HRT) for most biological wastewater treatment processes (4−20 h). 8 Thus, the EBCT appears to be insufficient to obtain any significant biodegradation of organic micropollutants in GAC filters. 9−12 These observations suggest that adsorption may be central to achieving high biodegradation in GAC filters.Yet, the potential interactions between the adsorption and biodegradation of micropollutants in GAC filters remain poorly understood due to the challenges in separating biodegradation from adsorption.
−20 In wastewater, the use of this approach has indicated a positive contribution of GAC biofilms to the removal of several biodegradable micropollutants, such as diclofenac and sulfamethoxazole. 13,15,20However, comparing biologically active and sterilized filters has inherent limitations.The sterilization methods that are applied might be unable to inhibit biological activity�completely or selectively�without affecting adsorption conditions. 21Further, the observed differences between a biologically active and sterilized filter cannot be used to discern whether the mechanism by which micropollutant removal increases is attributed to direct biodegradation of the micropollutant or to greater adsorption from the liberation of adsorption sites through the biodegradation of competing dissolved organic matter.
The use of 14 C-labeled organic micropollutants constitutes a direct method for separating biodegradation from adsorption when the 14 C-labeled moiety is mineralized to 14 CO 2 , which adsorbs poorly to activated carbon and can be extracted with CO 2 traps. 22,23By measuring the radioactive decay (ßparticles) from disintegration of 14 C, the labeled carbon(s) can be tracked between different liquid phases by liquid scintillation counting.In particular, measurements of 14 CO 2 formation with CO 2 traps can be used to quantify the extent of degradation that occurs via mineralization ( 14 CO 2 formation) of the labeled moiety but no other moieties.This 14 C-approach has been used to study the biodegradation and bioregeneration of selected drinking water contaminants in small-scale GAC columns 24,25 but has not been applied in column-based experiments for treating organic micropollutants in wastewater.
The objective of this study was to examine the interactions between adsorption and biodegradation with regard to the removal of organic micropollutants in the GAC filters.We selected 3 micropollutants with varying adsorption affinities to activated carbon and differing potential for biodegradation: diclofenac, a biodegradable and adsorbable compound; Nnitrosodimethylamine (NDMA), a biodegradable and weakly adsorbable compound; carbamazepine, a nonbiodegradable and adsorbable compound.We hypothesized that the biodegradation of adsorbable and biodegradable compounds (represented by diclofenac) is less sensitive to changes in EBCT than that of non-adsorbable and biodegradable compounds (represented by NDMA).To test this hypothesis, we studied 14 C-labeled and nonradiolabeled micropollutants in laboratory-scale column experiments with adsorptive GAC media and non-adsorptive sand media that had been retrieved from a full-scale treatment process.

MATERIALS AND METHODS
To examine the biodegradation of previously adsorbed micropollutants and the effect of EBCT, 3 separate experiments were performed with GAC and sand filter media from Degeberga WWTP, collected at ∼25,000 bed volumes (BVs).Each experiment comprised 8 columns that were operated for up to 1000 BVs to study the adsorption and biodegradation of the selected compounds.An overview of these micropollutants, their 14 C labeling, and the experimental setup is shown in Figure 1 and detailed in Sections 2.1−2.5.

Environmental Science & Technology
Hartmann Analytics, Germany).Carbamazepine is a nonbiodegradable 26 anticonvulsant drug with high adsorptive potential to activated carbon. 27Diclofenac is a nonsteroidal anti-inflammatory drug that undergoes low removal in biological wastewater treatment 28 but is potentially biodegradable in GAC filters, 10 with medium adsorptive affinity to activated carbon. 29NDMA is a small, polar, and biodegradable carcinogen, 30 with low adsorptive affinity to activated carbon. 31,32The 14 C-labeled forms of these micropollutants are hereafter referred to as 14 C-diclofenac, 14 C-carbamazepine, and 14 C-NDMA, respectively.All 14 C positioning was selected based on commercial availability, and the radiochemical purities were >98%.Nonradiolabeled micropollutants of analytical grade were also used for the experiments with subsequent UPLC-MS/MS analysis.
2.2.Experimental Design.Three experiments were performed in the dark at 21 ± 2 °C and a pH of 7.3 ± 0.5 but under different redox conditions and EBCTs (Figure 1).Experiment 1 was used to study whether previously adsorbed micropollutants could be degraded and was divided into two phases: 6 days of anoxic conditions (DO < 0.1 mg/L) with feeding of 14 C-labeled micropollutants (experiment 1a), followed by 5 days of oxic conditions (DO > 8 mg/L) without feeding of 14 C-labeled micropollutants (experiment 1b).The objective of the anoxic conditions in experiment 1a was to limit biological degradation, thus accumulating micropollutants in the GAC column via adsorption.The EBCT was ∼15 min in both phases.In the anoxic phase, lowoxygen conditions were established by sparging the feed with nitrogen gas overnight and performing the experiment in a glovebox with continuous flow of nitrogen gas.To reduce the risk of concentration-induced desorption of 14 C in the oxic phase (experiment 1b) due to the termination of 14 C in the feed, nonradiolabeled micropollutants were added individually to the feed at concentrations that corresponded to the 14 C added during the anoxic phase (experiment 1a).
Experiments 2 and 3 were designed to study the effect of the empty bed contact time on micropollutant adsorption and biodegradation (Figure 1).Experiment 2 was conducted with long EBCT (85 min), and experiment 3 was conducted with short EBCT (15 min).Both experiments were performed over 5 days under oxic conditions (DO > 8 mg/L).

Experimental Setup. 2.3.1. Full-Scale Treatment at Degeberga WWTP and Media for Column Experiments.
GAC and sand were retrieved from the top of two full-scale filters at Degeberga WWTP, Sweden, on three separate occasions (Figure S1 and Table S1).GAC from the top of the filter bed was used to best resemble long-term operation in terms of adsorbed micropollutants 33 and biofilm development. 34The sand and GAC filters are operated in series, downstream of an activated sludge process with nitrification and partial denitrification, followed by postprecipitation and final clarification.The sand filter, the first to be established in Sweden, has been in operation since 1975 and has a current hydraulic surface loading of 1 m/h, a bed depth of 1.4 m, a mean EBCT of 87 min, and high dissolved oxygen (DO > 8 mg/L) in the influent.
The GAC filters, installed in April 2020, constitute the first full-scale GAC treatment of municipal wastewater in Sweden, and they were scaled up from a previous pilot-scale study�the FRAM Project 22,35 �with sequential sand and GAC filters that have been operated for over 40,000 BVs.The GAC filters consist of open basins that are aerated by overfall, ensuring high dissolved oxygen concentrations (DO > 8 mg/L) in the influent.The sampled GAC filter is operated with the same bituminous coal-based activated carbon as in the FRAM Project (Aquasorb 5000, 8′30 mesh, 0.6−2.4mm, Jacobi), with a specific Brunauer−Emmett−Teller (BET) surface area of 1200 m 2 /g.The GAC used was selected based on adsorption studies with 9 different commercial GAC types using an LC-UV-based technique, 36 where it showed the highest adsorption ability, highest removal of negatively charged compounds, and fast adsorption kinetics.The filter has a bed depth of 1 m and is operated with a mean EBCT of 45 min and a hydraulic surface loading of 1.3 m/h.At the time of media retrieval for the column experiments, the GAC filter had been running for over 2.5 years and over 25,000 BVs without any backwashing events.
In the full-scale filters, the removal of carbamazepine was 30−50% in the GAC filter and ±5% in the sand filter at the time for this investigation (Table S1).For diclofenac, on the other hand, the removal was 80−90% in the GAC filter and 15−40% in the sand filter.Detailed information on the GAC filter performance and the removal of the 24 monitored organic micropollutants can be found elsewhere 2 (please contact Ola Svahn (ola.svahn@hkr.se) for details regarding the FRAM Project and the advanced treatment in Degeberga).
After media retrieval, the GAC and sand were stored at 8 °C for <72 h before the start of the experiment.Before use, suspended solids that might have originated from previous treatment steps were removed from the GAC and sand media through careful, repeat washes with tap water.In addition, medium size fractions >1.5 mm were eliminated by carefully passing the media through a 1.5 mm sieve under water.The effective size of the resulting GAC and sand was determined via sieving per Swedish standards (SS-EN ISO 17892-4:2016).The resulting ratios of the inner diameter (16 mm) of the column to the mean particle diameter ranged from 14.8 to 15.4 for the GAC media and from 17.4 to 19.3 for the sand media (Table S2), which exceed the minimum ratios (5−10) that have been recommended to avoid wall effects. 37,38.3.2.Column Setup.Gas-tight glass columns (length of 200 mm, inner diameter of 16 mm; Cytiva) with adapters for adjustable bed heights were filled with media (GAC or sand) to a volume of 10 cm 3 , corresponding to a bed height of ∼50 mm and dry weights of ∼3 g GAC and ∼15 g sand (Table S3).The feed consisted of biologically and chemically treated wastewater that was retrieved downstream of the sand filter at the Degeberga WWTP and spiked with radiolabeled or nonradiolabeled micropollutants.Before use, the water was filtered (0.45 μm cellulose nitrate, Whatman), aerated (dissolved oxygen, DO > 8 mg/L), and adjusted to pH 7 with 1 M NaOH or HCl.
Eight columns were used in each experiment and divided into 4 pairs, each of which consisted of one GAC filter and one sand filter.Three pairs of columns were fed with 14 C-labeled micropollutants (one micropollutant per column pair) at 14 C activities of 0.1 μCi/L, corresponding to concentrations of 100−1100 ng/L (Table S4).The last pair was used to study the nonradiolabeled compounds and was spiked with the same amounts of the corresponding nonradiolabeled micropollutants as with the 14 C-labeled micropollutants, and the resulting concentrations were thus the sum of the spiked and background levels (Table S4).The last pair was also used to follow the concentrations of standard parameters.

Column Operation.
Before the start of the experiments, the filters were adapted to the experimental conditions (EBCT, T, DO) by feeding them wastewater without spiking of 14 C or nonradiolabeled micropollutants for 12−24 h before 14 C-labeled micropollutants were added to the feed.The effluent from each filter was collected over time into several sets of sealed glass bottles through a hypodermic needle (Figure 1).A secondary needle was used to purge excess air through a CO 2 trap (25 mL, 1 M NaOH) to monitor 14 CO 2 loss from the bottle during sample collection.To avoid 14 CO 2 loss, the bottles contained small, predetermined amounts of 1 M NaOH to achieve a final pH of 9−10, at which dissolved inorganic 14 C exists primarily as 14 CO 3 2− and H 14 CO 3 − .On complete retrieval of the sample for each interval, a glass-tube CO 2 trap (30 mL, Ø 1.5 cm) that contained 20 mL of 1 M NaOH was installed carefully in the glass bottle, after which predetermined amounts of 1 M HCl were added to lower the pH to 3 ± 0.2, favoring CO 2 release.The bottles were then sealed immediately and incubated for 40−48 h in the dark at 20 °C and 130 rpm to allow complete absorption of CO 2 by the trap.After incubation, the samples were retrieved from the trap and liquid phase and transferred to 10 mL Falcon tubes for subsequent analysis of 14 C activity.At the end of the experiment, all remaining GAC and sand from each column were frozen immediately for a later analysis of adsorbed 14 C by combustion and subsequent CO 2 trapping.
2.4.Analysis.2.4.1. 14C Measurements. 14C activity was measured in disintegrations per minute (DPM) by liquid scintillation counting on a Tricarb 4910 TR (PerkinElmer) using a scintillation cocktail (Ultima Gold, PerkinElmer).Quenching was accounted for using transformed Spectral Index of the External Standard values. 39For the column experiments, 0.6 mL of the 1 M NaOH traps was added to a final volume of 4 mL in 6 mL polypropylene vials.For the liquid wastewater samples, 4 mL was added to a total of 16 mL in 20 mL polypropylene vials.For the combustion experiments, 0.6 mL of the 4 M NaOH traps was added to a total of 16 mL in 20 mL polypropylene vials.The background radiation (n = 3) in NaOH and wastewater was subtracted from the value of each measurement.
To measure 14 C in the solid phase, portions of the remaining GAC (1.00 ± 0.05 g wet weight) or sand media (2.00 ± 0.05 g wet weight) were combusted in triplicate in a Pyrolyzer genIII (Raddec International) that was equipped with downstream CO 2 traps (2 traps in series with 25 mL of 4 M NaOH each).Dry weight:wet weight ratios were determined, based on triplicate measurements, before and after drying at 105 °C.The combustion program was run at an air flow rate of ∼0.1 L/min, with stepwise heating and dwelling to 900 °C over ∼5 h and the catalyst zone maintained at 900 °C during the entire combustion run, as detailed in Table S5.The combustion method had a 14 C recovery of >90% for 14 C-diclofenac and 14 C-carbamazepine and >80% for 14 C-NDMA for adsorbed compounds to virgin GAC.
2.4.2.Micropollutant Analysis.Wastewater samples (50 mL) that contained nonradiolabeled micropollutants were concentrated by solid-phase extraction (SPE) and analyzed by ultraperformance liquid chromatography (UPLC), coupled with tandem mass spectroscopy (MS/MS), as detailed elsewhere. 40,41Briefly, the samples were extracted on SPE columns (Oasis HLB 200 mg).After being dried, eluted, and evaporated, the samples were reconstituted and injected (1 μL) into the UPLC MS/MS instrument (Waters Acquity UPLC H-Class, Xevo TQS Waters Micromass, Manchester, UK).Limits of quantification (LOQs) and relative standard deviations (RSDs) for diclofenac and carbamazepine are provided in Table S6.Nonradiolabeled NDMA was not measured in the experiments.2.5.Data Presentation.The results from the column experiments are presented as the mass flow of 14 C per unit time or accumulated fractions (eq 1).Accumulated fractions were used to graphically visualize data even after the addition of 14 C-labeled micropollutants to the influent had been terminated (Experiment 1b) and was used to present the data from all experiments.
where t is the sampling time point, n is the time point at which the accumulated fraction is calculated, c in is the influent concentration, c eff,t is the effluent concentration between time point t and t − 1, and V t is the volume collected between time point t and t − 1.

RESULTS AND DISCUSSION
A total of 24 columns with GAC and sand filter media were run over 5 or 11 days to evaluate the biodegradation and adsorption of diclofenac, NDMA, and carbamazepine under various redox conditions and EBCTs.The experimental conditions and influent/effluent concentrations of the standard wastewater parameters are provided in the Supporting Information for the initial adaptation period (Table S7) and the corresponding analysis of diclofenac and carbamazepine in the nonradiolabeled columns (Table S8).Low removal of DOC (<16%) and UVA 254 values were observed, as expected, based on the high number of treated BVs (>25,000 BVs) in the full-scale GAC filter.The flow through the columns was generally stable (Figure S2), with similar contact times, in the 8 columns in the experiments with anoxic/oxic conditions (16.2 ± 0.7 min) and oxic conditions with long EBCT (85 ± 2.6 min) and short EBCT (15.4 ± 0.5 min).

Biodegradation of Previously Adsorbed Micropollutants.
To determine whether micropollutants could be adsorbed and degraded at a later stage, 14 C-labeled micropollutants were allowed to adsorb during an initial anoxic phase (Experiment 1a), with a subsequent oxic phase, without 14 C being fed (Experiment 1b).The EBCT was ∼15 min in both phases.The anoxic conditions in the initial phase were confirmed through stable NH 4 + -N concentrations compared with those in the oxic phase (Tables S7 and S8).Limited 14 CO 2 formation was observed in the anoxic phase for 14 Cdiclofenac (<10%), 14 C-NDMA (<5%), and 14 C-carbamazepine (<1%) in the GAC and sand columns (Figure 2).−44 Anoxic removal of NDMA has been observed at high residence times (days) in soil, 45 and dealkylation of the 14 C-labeled methyl carbon has been confirmed through the formation of 14 CO 2 . 46or GAC, 14 CO 2 formation from 14 C-diclofenac increased in the oxic phase (Figure 2), which must have originated from previously adsorbed 14 C and might have consisted of both 14 Cdiclofenac and 14 C-labeled products that resulted from anoxic transformation.Transformation products that are generated from the cleavage of the labeled carboxylic carbon in diclofenac have also been reported for aerobic biofilms.However, among the primary transformation pathways reported for diclofenac (hydroxylation, decarboxylation, and amidation), only decarboxylation directly leads to a cleavage of the 14 C-labeled carboxylic carbon. 47Further transformation of the other primary transformation products may, however, also lead to cleavage of the labeled carbon.
The rate of 14 CO 2 formation rose immediately after the introduction of dissolved oxygen, with its peak exceeding the inflow of 14 C-diclofenac per time unit in the anoxic phase.These results suggest that previously adsorbed diclofenac can Please note that the data for the nonradiolabeled and 14 C-labeled compounds were obtained in separate columns operated in parallel.

Environmental Science & Technology
be degraded under oxic conditions and that the biofilm can attain biodegradation rates that are sufficiently high to remove substantial amounts of diclofenac.For the sand column, however, negligible 14 CO 2 formation was observed in the oxic phase, as expected, based on the limited adsorption to sand in the anoxic phase.
The adsorption of 14 C-NDMA was low (<10%) in the GAC and sand columns during the anoxic phase, as expected, based on the polar nature of NDMA and its low adsorptive affinity to activated carbon. 31,32Therefore, the potential for the biodegradation of previously adsorbed NDMA in the oxic phase was limited.In contrast, 14 C-carbamazepine partially adsorbed to the GAC column during the anoxic phase, but its biological recalcitrance 28 prevented its subsequent degradation (or potential 14 CO 2 formation) under oxic conditions.
Figure 3 shows the accumulated fractions (eq 1) of 14 C in the effluent liquid phase and the CO 2 trap and those of the nonradiolabeled compounds (diclofenac and carbamazepine) in the effluent liquid phase.Low accumulated fractions in the effluent liquid phases can be interpreted as high removal over the column (or low breakthrough), whereas high accumulated fractions of 14 CO 2 can be interpreted as high biological degradation of the 14 C-labeled moieties.The removal of nonradiolabeled diclofenac and carbamazepine was stable over time but differed between compounds and columns.The removal of 14 C from 14 C-diclofenac was lower than the removal of nonradiolabeled diclofenac in the GAC and sand columns, suggesting that biodegradation of diclofenac occurs in the anoxic phase and that some of the 14 C in the liquid phase effluent (∼25% for the sand filter and ∼20% for the GAC filter) consisted of 14 C-labeled transformation products.These observations contrast the results of previous studies on anoxic transformation of diclofenac in sludge 42,43 and moving bed biofilm reactors. 44,48Anoxic transformation of diclofenac has however been indicated during managed aquifer recharge, 49 but there are no such data for rapid sand filtration.
For 14 C-NDMA, the level of 14 CO 2 formation was ∼5% in the anoxic phase for both the sand and GAC columns.However, the low adsorption of 14 C-NDMA in the anoxic phase for the sand column (∼5%) and the GAC column (∼10%) limited the degradation in the oxic phase when the addition of 14 C-NDMA had been terminated.
Limited removal of carbamazepine (<5%) was observed in the sand column based on the analysis of both the 14 C-labeled and nonradiolabaled compounds.However, a clear difference (50−70%) was observed between the removal of 14 C-labeled and nonradiolabeled carbamazepine in the GAC column.For 14 C-carbamazepine, the breakthrough increased slowly but steadily, even after the addition of 14 C-carbamazepine had been terminated in the oxic phase.These results show that 14 C-carbamazepine was substantially retained on its travel through the GAC column via successive adsorption− desorption and that the adsorption was at least partially reversible.The higher breakthrough of nonradiolabeled carbamazepine can be explained by the desorption of carbamazepine that had been adsorbed prior to the experiments during the operation of the full-scale filter (∼25,000 BVs).Whereas nonradiolabeled carbamazepine was preloaded throughout the GAC bed and could have desorbed from all areas of the column, newly introduced 14 C-carbamazepine had to travel through the entire GAC bed to reach the effluent.Although it may appear as a large portion of carbamazepine that passed through the bed without being adsorbed, our results with 14 C-labeled carbamazepine show that the compound is substantially retained on its passage through the GAC bed via adsorption−desorption interactions.
After these experiments, 14 C mass balances were established via combustion of the GAC and sand filter media and subsequent 14 CO 2 trapping, resulting in >75% recovery of the 14 C for all compounds (Figure S3).Higher uncertainties were generally observed with higher degrees of biological mineralization ( 14 CO 2 formation); one explanation is that it is easier to recover stable and recalcitrant compounds in one or two phases compared to biodegradable compounds with potentially volatile transformation products in three phases.

Influence of EBCT.
To further investigate the influence of the EBCT on micropollutant adsorption and biodegradation, we performed column experiments at short (∼15 min; experiment 2) and long (∼85 min; experiment 3) EBCTs with GAC and sand columns (Figure 4 and Figure S3) without an initial anoxic phase. 14C-diclofenac and 14 C-NDMA were partially degraded at their respective 14 C-labeled moieties, as evidenced by the 14 CO 2 formation in both columns; further, the extent of 14 CO 2 formation was higher than for the anoxic phase in the previous experiment (Figure 3).
There was substantial biodegradation of 14 C-diclofenac in the GAC column at both the short and long EBCTs (∼60 and ∼70% 14 CO 2 formation at EBCTs of 15 and 85 min, respectively), accompanied by low breakthrough (0−10%) of liquid-phase 14 C. Low breakthrough (<5%) was also observed for nonradiolabeled diclofenac compared with the full-scale GAC filter during GAC sampling (10−20% at an EBCT of ∼45 min; Table S1).Increasing the EBCT from 15 to 85 min only resulted in a slightly higher 14 CO 2 formation, a slightly lower 14 C breakthrough in the liquid phase, and a lower breakthrough of nonradiolabeled diclofenac.
The formation of 14 CO 2 in the GAC columns (∼60% at a 15 min EBCT) is high compared with the values in the literature for nonradiolabeled diclofenac removal in biofilm systems, in which hydraulic retention times of 4−16 h have been required for ∼80% removal in high-performing systems. 47,50In addition, the observed formation of 14 CO 2 from diclofenac is higher than expected from previous comparisons between biologically active and sterilized GAC; 13,51 biodegradation, however, might have been masked by high adsorption in the sterilized column. 20Overall, our results support previous indications that biodegradation of diclofenac underlies the sustained removal of diclofenac in full-scale GAC filters at higher bed volumes (>25,000) and short EBCTs (16−25 min). 10,52n the sand columns, 14 CO 2 formation from 14 C-diclofenac (∼5% at a 15 min EBCT and ∼10% at an 85 min EBCT) was much lower than for GAC.Lower effluent fractions were observed for nonradiolabeled diclofenac compared with liquidphase 14 C (similar to the anoxic experiment), suggesting that biodegradation also proceeded via pathways that resulted in the formation of 14 C-labeled transformation products, such as hydroxy-diclofenac. 10,47The removal of nonradiolabeled diclofenac in the sand columns with similar biofilm content (expressed as volatile solids, Table S3) ranged from ∼20% (15 min EBCT) to ∼40% (85 min EBCT), which is comparable with the values in the literature, ∼ 20% at a 120 min EBCT 11 and 20−30% at an EBCT of 8−16 min, 53 and consistent with what has been observed for the full-scale sand filter at Degeberga WWTP (15−40% in Table S1 and a 3 year average of 35% 2 ).

Environmental Science & Technology
In contrast to 14 C-diclofenac, more 14 CO 2 was formed from 14 C-NDMA in the sand (35%, short EBCT; 70%, long EBCT) versus GAC columns (15%, short EBCT; 50%, long EBCT).Aerobic transformation pathways of NDMA are well documented in mammalian cells 54 and have been proposed to proceed via similar pathways in bacteria expressing specific monooxygenase enzymes. 55The transformation is expected to occur either through an initial hydroxylation of one of the methyl groups 55 or via an initial oxidation of the nitroso group into a nitro group, followed by a hydroxylation of one of the methyl groups. 56Subsequent demethylation of the hydroxylated intermediates has been reported to result in formaldehyde formation, where further transformation is expected to result in mineralization of one of the labeled carbons.Aerobic mineralization of the 14 C-labeled carbon in 14 C-NDMA has also been demonstrated for aqueous and soil systems. 57imilar removal of nonradiolabeled NDMA (50−90%) has previously been reported for sand and GAC filters. 30By increasing the EBCT from 15 to 85 min, the fraction of 14 CO 2 formation from 14 C-NDMA rose from 35 to 70% for sand and from 20 to 50% for GAC (Figure 4).These results suggest that GAC filter media do not necessarily have higher overall biodegradative capacity for organic micropollutants than sand filter media.This could to some extent also limit the possibilities for predicting the biodegradation capacity of organic micropollutants in GAC filters based on degradation data from other biological wastewater treatment processes.For the GAC columns, the liquid-phase 14 C fractions in the effluent from 14 C-NDMA were higher than for diclofenac (∼60% at a 15 min EBCT and ∼25% at an 85 min EBCT), illustrating that degradation, rather than adsorption, governs the removal of NDMA.Overall, the removal of poorly adsorbing NDMA was more sensitive to changes in EBCT compared with welladsorbing diclofenac.These results strengthen the theory that by prolonging the time for biodegradation for adsorbable and biodegradable substances in GAC filters, adsorption can improve the conditions for biodegradation.
A longer EBCT decreased the breakthrough of nonradiolabeled carbamazepine in GAC columns from ∼70 to ∼50%.In contrast, the breakthrough of 14 C-carbamazepine in the GAC columns was negligible at a long EBCT but increased slowly but steadily over time at the short EBCT.Decreased removal and earlier breakthrough are expected at shorter EBCTs due to mass transfer limitations of the adsorption process at EBCTs <20 min 6 but also to the higher number of BVs passing the columns (∼480 BV at the short EBCT and ∼85 BV at the long EBCT) in the experiments with 14 Clabeled carbamazepine.As shown in the previous anoxic−oxic experiment, the large differences between the fractions of liquid-phase 14 C and nonradiolabeled carbamazepine highlight the slow transport of carbamazepine through the columns via continuous adsorption−desorption.
3.3.Implications.The observed degradation of previously adsorbed diclofenac illustrates that the retention in GAC filters prolongs the time available for biodegradation, thereby decoupling the biodegradation time from the hydraulic residence time.This retention increases the micropollutant concentration locally (adsorbed + liquid phase), which may allow for higher micropollutant biodegradation rates in terms of mass 58,59 and/or increase micropollutant availability, potentially promoting the selection of microorganisms targeting their removal.Thus, by prolonging the micro-pollutant residence time through adsorption in GAC filters, it may be possible to achieve high removal of certain micropollutants at hydraulic residence times that are normally considered insufficient for substantial biodegradation in biological wastewater treatment processes. 28till, not all potentially biodegradable micropollutants may be available for biotransformation in GAC filters due to the expected adsorption in micropores and mesopores that are considered too small for microorganisms or their extracellular enzymes to reach. 60Desorption is considered a prerequisite for subsequent biodegradation, 24,61 so compounds that adsorb too strongly, with high adsorption energies, may not be available for subsequent biodegradation. 62Thus, synergistic effects between adsorption and degradation in GAC filters may be mainly possible for compounds with moderate affinity for adsorption such as diclofenac.
Successive adsorption−desorption was demonstrated for the biologically active compound carbamazepine through the observed differences in breakthrough between the 14 C-labeled and nonradiolabeled compounds.The retention of compounds with moderate-to-strong adsorption can also have an implication for the evaluation of GAC filters, since the effluent concentrations of micropollutants can be affected by historical variations in the influent, not necessarily captured by conventional sampling campaigns over a few days or a week.
Figures showing wastewater treatment plant configuration, flow measurements, and 14 C mass balance calculations; tables showing micropollutant removal in full-scale filters, GAC and sand particle size distribution and column aspect ratios, total and volatile solids, micropollutant concentrations, combustion settings for analysis of solid-phase 14 C, nonradiolabeled micropollutant LOQ and recovery, and experimental conditions during the initial adaptation period and in reference columns throughout the experiments (PDF) ■

Figure 2 .
Figure2.Flux of14 C in the influent (green diamonds) and effluent (liquid phase; blue circles) and to the CO 2 trap (orange squares) of the columns during the anoxic/oxic experiment at an EBCT of 15 min.After the anoxic phase (DO < 0.1 mg/L) (shaded), oxic (DO > 8 mg/L) conditions were introduced but without the addition of14 C.

Figure 3 .
Figure 3. Accumulated fractions of 14 C in the liquid phase (blue rings) and the CO 2 trap (orange squares) and nonradiolabeled micropollutant (yellow diamonds) in the anoxic−oxic experiment (experiment 1).Please note that the data for the nonradiolabeled and 14 C-labeled compounds were obtained in separate columns operated in parallel.

Figure 4 .
Figure 4. Accumulated fractions of 14 C in the liquid phase (blue rings) and CO 2 trap (orange squares) in the columns with 14 C-labeled micropollutants and nonradiolabeled micropollutant (yellow diamonds) in the experiments with short and long EBCTs (experiments 2 and 3).Please note that the data for the nonradiolabeled and 14 C-labeled compounds were obtained in separate columns operated in parallel.

AUTHOR INFORMATION Corresponding Author Alexander
Betsholtz − Department of Process and Life Science Engineering.Division of Chemical Engineering, Lund University, Lund 221 00, Sweden; orcid.org/0000-0002-3023-8293;Email: alexander.betsholtz@chemeng.lth.seDepartment of Process and Life Science Engineering.Division of Chemical Engineering, Lund University, Lund 221 00, Sweden Ola Svahn − School of Education and Environment, Division of Natural Sciences, Kristianstad University, Kristianstad 291 88, Sweden Michael Cimbritz − Department of Process and Life Science Engineering.Division of Chemical Engineering, Lund University, Lund 221 00, Sweden Åsa Davidsson − Department of Process and Life Science Engineering.Division of Chemical Engineering, Lund University, Lund 221 00, Sweden Complete contact information is available at: AuthorsPer Falås −