Stimulating Nitrogen Biokinetics with the Addition of Hydrogen Peroxide to Secondary E ﬄ uent Bioﬁltration

: Tertiary wastewater treatment could provide a reliable source of water for reuse. Amongst these types of wastewater treatment, deep-bed ﬁltration of secondary e ﬄ uents can e ﬀ ectively remove particles and organic matter; however, NH 4 + and NO 2 − are not easily removed. This study examined the feasibility of stimulating microbial activity using hydrogen peroxide (H 2 O 2 ) as a bio-speciﬁc and clean oxygen source that leaves no residuals in the water and is advantageous upon aeration due to the solubility limitations of the oxygen. The performance of a pilot bio-ﬁltration system at a ﬁltration velocity of 5–6 m / h, was enhanced by the addition of H 2 O 2 for particle, organic matter, NH 4 + , and NO 2 − removal. Hydrogen peroxide provided the oxygen demand for full nitriﬁcation. As a result, inﬂuent concentrations of 4.2 ± 2.5 mg / L N-NH 4 + and 0.65 ± 0.4 mg / L N-NO 2 were removed during the short hydraulic residence time (HRT). In comparison, ﬁltration without H 2 O 2 addition only removed up to 0.6 mg / L N-NH 4 + and almost no N-NO 2 − . A DNA metagenome analysis of the functional genes of the media biomass reﬂected a signiﬁcant potential for simultaneous nitriﬁcation and denitriﬁcation activity. It is hypothesized that the low biodegradability of the organic carbon and H 2 O 2 addition stimulated oxygen utilization in favor of nitriﬁcation, followed by the enhancement of anoxic activity.


Introduction
The tertiary and advanced treatment of wastewater improves water quality, enabling us to meet environmental concerns but also find a potential source of water for reuse [1,2]. The filtration of coagulated-flocculated effluent via high-rate deep-bed filters (velocities between 5 and 20 m/h) is widely depending on substrate availability and biodegradability [15][16][17][18]. Compared to suspended growth systems, nitrification activity can increase significantly in attached biomass systems when oxygen is more available [19]. DO concentration in the attached biomass is limited by the temperature of the solute (~8 mg/L at 25 • C) and decreases along the biofilm depth, resulting in increased denitrification rates [20,21]. To overcome these limitations, a few studies have examined the use of 100% pure oxygen or hydrogen peroxide (hereafter H 2 O 2 ) degradation (Equation (1)) in attached-biomass systems [22][23][24][25].
H 2 O 2 degradation by the enzyme catalase. Stochiometry is by weight (g); K cat -catalytic reaction rate [26].
Catalase is a significant component of the cell defense mechanism against oxidative stress, it scavenges two molecules of H 2 O 2 into water and molecular oxygen in one of the highest documented enzyme reaction rates [27,28].
H 2 O 2 has advantages as a supplemental oxygen source, namely, high solubility in aqueous solution and an improved mass transfer of dissolved H 2 O 2 (liquid-liquid) into the permeable biofilm compared to oxygen that depends also on the oxygen-water transfer into the liquid phase. In addition, rapid H 2 O 2 degradation is bacterium-specific and leaves no persistent residual traces in the water. The disinfection characteristics of H 2 O 2 could limit microbial-based reactors, which depend on concentration and time (C × t). Significantly tolerant degradation rates and oxygen utilization have been observed in biofilms, in comparison to suspended bacteria, in addition to changes in microbial community structure [29][30][31][32][33]. However, the application of H 2 O 2 as a supplemental oxygen source via high-rate media filtration for NH 4 + removal at a short hydraulic residence time (HRT) has never been examined. In this study, we demonstrate the impact of adding H 2 O 2 as a supplemental oxygen source on the operational efficiency of a secondary effluent filtration system with biologically active media. The setup design was aimed at combining the removal of particles, NH 4 + , and, especially, NO 2 − to reduce ozone demand for a subsequent ozonation treatment (detailed in Zucker et al., [34]). In short, the biofilter was part of a multistep pretreatment unit prior to ozonation to allow better soil aquifer treatment (SAT) performance by reducing particle clogging and minimizing oxygen demand during infiltration in the upper-layer SAT. The specific goals of this study were to examine whether significant biotransformation can be achieved with high-rate (relative to standard biofiltration) filtration conditions and H 2 O 2 addition.

Pretreatment System Setup
The biofilter system was operated in direct-filtration at the pilot site of the Shafdan wastewater treatment plant (WWTP), operated by Mekorot. The biofilter system, detailed in Figure 1, was part of the multistage setup specified herein: -Feed flow was 6 m 3 /h (144 m 3 /day) of the secondary effluent from the WWTP (Figure 1a) following a 500-µm wire-mesh filtration (Figure 1b) to remove coarse particles. -Coagulation/flocculation was carried out by injecting polyaluminum chloride (PACl, 18% Al 2 O 3 ) by peristaltic pump (Figure 1c) to achieve a final concentration of~2.7-3.6 mg/L as PACl.
Flocculation was performed in a modified flocculator, which consisted of a pressure filter (Figure 1d) with~15 min hydraulic retention time (HRT). PACl was chosen for this study as it has been shown to be preferable for flocculation and is widely used [35,36]. -Following flocculation, H 2 O 2 was added (10%) to the inlet of the high-rate biofilter ( Figure 1f) using a peristaltic pump (Figure 1e). The final concentration of 27 mg/L was achieved to provide a surplus of DO for full nitrification of 3.5 mg/L-N NH 4 + and 0.5 mg/L-N NO 2 − . The biofilter tank had a surface area of 1.13 m 2 , 1.2 m diameter, 1.1 m 3 media volume, and 36% free headspace (additional characteristics are specified in Zucker et al., [34]).
During the adjustment period with stepwise H 2 O 2 addition, an average measured concentration of 4.29 ± 0.56 mg/L DO at the outlet of the filter was obtained with the addition of~27 mg/L H 2 O 2 . The theoretical calculation of oxygen mass balance is provided in Appendix A.1. During the research period, the loads of NH 4 + in the secondary effluent increased due to changes in the process of the plant, which enabled investigating the system performance for NH 4 + oxidation and N removal under higher NH 4 + loads and O 2 limitation.
Clean Technol. 2020, 2 FOR PEER REVIEW 3 The biofilter system was operated in direct-filtration at the pilot site of the Shafdan wastewater treatment plant (WWTP), operated by Mekorot. The biofilter system, detailed in Figure 1, was part of the multistage setup specified herein: -Feed flow was 6 m 3 /h (144 m 3 /day) of the secondary effluent from the WWTP (Figure 1a) following a 500-µ m wire-mesh filtration (Figure 1b) to remove coarse particles. -Coagulation/flocculation was carried out by injecting polyaluminum chloride (PACl, 18% Al2O3) by peristaltic pump (Figure 1c) to achieve a final concentration of ~2.7-3.6 mg/L as PACl.
Flocculation was performed in a modified flocculator, which consisted of a pressure filter ( Figure  1d) with ~15 min hydraulic retention time (HRT). PACl was chosen for this study as it has been shown to be preferable for flocculation and is widely used [35,36]. -Following flocculation, H2O2 was added (10%) to the inlet of the high-rate biofilter ( Figure 1f) using a peristaltic pump (Figure 1e). The final concentration of 27 mg/L was achieved to provide a surplus of DO for full nitrification of 3.5 mg/L-N NH4 + and 0.5 mg/L-N NO2 − . The biofilter tank had a surface area of 1.13 m 2 , 1.2 m diameter, 1.1 m 3 media volume, and 36% free headspace (additional characteristics are specified in Zucker et al., [34]).
During the adjustment period with stepwise H2O2 addition, an average measured concentration of 4.29 ± 0.56 mg/L DO at the outlet of the filter was obtained with the addition of ~27 mg/L H2O2. The theoretical calculation of oxygen mass balance is provided in Appendix A.1.1. During the research period, the loads of NH4 + in the secondary effluent increased due to changes in the process of the plant, which enabled investigating the system performance for NH4 + oxidation and N removal under higher NH4 + loads and O2 limitation.  Table  500 m, (b) wire filter, (c) the addition of polyaluminum chloride (PACl) from a tank, (d) the coagulator-flocculator tank, (e) the addition of H2O2, (f) the bed biofilter, (g) the storage tank, and (h) the backwash pump. Sampling points were located (1) right after the wire-mesh filter, termed the secondary effluent (SE), (2) after flocculation-coagulation and H2O2 addition, but before the biofilter (BF) and (3) after the biofilter (AF).  Table  500 m, (b) wire filter, (c) the addition of polyaluminum chloride (PACl) from a tank, (d) the coagulatorflocculator tank, (e) the addition of H 2 O 2 , (f) the bed biofilter, (g) the storage tank, and (h) the backwash pump. Sampling points were located (1) right after the wire-mesh filter, termed the secondary effluent (SE), (2) after flocculation-coagulation and H 2 O 2 addition, but before the biofilter (BF) and (3) after the biofilter (AF).

System Specification and Operational Parameters
The biofiltration was operated with a modified active media filter, operated in downflow mode, with a filtration velocity of 5-6 m/h (~5 min hydraulic residence time (HRT, the low range of high-rate filtration) and a backwash cycle every 12 h. The filtered effluent water was stored in a 10-m 3 tank ( Figure 1g) and was used for backwash. Adjustments were performed to avoid the clogging of the filter while maintaining a steady and active biomass to avoid the detachment of biofilm from the media (Appendix A.2). The microbial community of the biomass was indigenous and developed over time by feeding the biofilter with the secondary effluent. It was assumed that the accumulation and growth of viable bacteria in the filter, most of which were fixed to the media, was the basis for the bacterial activity [6,[37][38][39].

Solute Sampling and Analyses
After two months of system operation without H 2 O 2 addition, samples were further taken over for a period of four months with stable performance. Stabilization after H 2 O 2 addition took also about two months and then a sampling campaign was conducted every 2-4 weeks for over two years. At every sampling campaign, samples were collected 6-7 h after the start of the morning filtration cycle (middle of the cycle) from sampling faucet located past the mesh filter, named the secondary effluent (SE), before the biofilter (BF) and after the biofilter (AF). An analysis of water characteristics was conducted at the Shafdan WWTP laboratory specified in the Appendix A.3.

Particle Sampling and Analysis
Effluent particles and aggregates were analyzed by Micro Flow Imaging (MFI) technology (DPA 4100, Protein Simple Inc., Ottawa, ON, Canada). This apparatus employs a digital camera with an illumination and magnification system to capture in-situ images of suspended particles in a flowing sample. A detailed description of this analysis is published elsewhere [32,36]. An analysis was conducted on particles of between~2 and 400 µm.

Media Sampling and Metagenome Analysis
A metagenome analysis was conducted to investigate the functional potential of the microbial community in the media. In the middle of the operation cycle (6 h), the biofilter was emptied and a sample of 50 g anthracite media was taken from a 40-cm depth, thoroughly mixed for homogenization, and immediately stored in dry ice. Sampling was conducted a few days after the sampling campaign designated in star in Figure A1 in the Appendix A. DNA from the media was extracted using an Exgene Soil DNA Kit. The whole-genome (shotgun) libraries were prepared using the TruSeq DNA PCR-Free HT Library Preparation Kit (Illumina, San Diego, CA, USA) and were sequenced using an Illumina MiSeq benchtop sequencer with a target fragment length of 250 bp. After the removal of adapters and low-quality reads,~8.2 M reads were left with an average length of 225 base pairs. (Appendix A Figure A2). High-quality reads were mapped to N-cycle related genes using BWA [40]. The potential for dominant biotransformation of the microbial community was deduced from the relative abundance of specific functional genes. The genes for nitrification (amoA, nxrB, cyt c), denitrification (narG, nirK, nirS, norB, nosZ), and other various metabolic pathways, such as methane (pmoA), hydrogen sulfide (drsEFH), and hydrogen (HiFe) oxidation were analyzed and compared on a log scale [9]. The gene catalog was extracted from NCBI's nucleotide database. Genes aligning to the reads were sorted according to their respective identified organisms, which varied between 6%-19% among all the reads related to each gene ( Figure A2 in the Appendix A).

Results
The goal of this study was to design a biofilter with short HRT to support nitrification activity, by surplus oxygen, for the removal of NO 2 − and NH 4 + . The system showed a significant removal of NO 2 − and NH 4 + when H 2 O 2 was added in comparison to the control. The biofilter was optimized in the matter of H 2 O 2 addition and evaluated as a single step in an integrated treatment with the ultimate goal to produce a streamwith lower oxygen demand for following SAT or non-potable reuse purposes.

Particle Distribution
To determine biofilter performance in removing particles as a function of equivalent circular diameter (ECD), and the particles' size and the size distribution in the secondary effluent, a dynamic image analysis was made. The particle-size distribution (PSD) of secondary effluents based on 13 campaigns is presented in Figure A1 in the Appendix A, taken at the SE sampling point. In general, the PSD was similar in comparison to the high variation in total particle number. Almost all (99%) of the particles were analyzed (<50 µm), with the highest variation in the fraction with ECD = 2-3 µm. The total particle removal performance was found to be highly efficient (95 ± 2%) for all sampling campaigns. No difference was observed in particle removal without H 2 O 2 . A typical PSD analysis of the secondary effluent after 500 µm, post flocculation and after the biofilter, is presented in Figure A1 in the Appendix A. Flocculation before biofiltration increased particle counts due to the flocculation of the macromolecules and colloid particles (<2 µm), which were not analyzed. As expected, particle concentration decreased dramatically after biofiltration for all ECDs.

H 2 O 2 Decomposition
The concentration of DO before filtration varied between 3-5 mg/L. No H 2 O 2 was detected at the inlet or outlet during an operation without H 2 O 2 addition, while the DO values at the outlet were <1 mg/L, which clearly indicated maximum oxygen utilization and reflected anoxic conditions [20]. The preliminary lab-bench experiments revealed no decomposition of H 2 O 2 when mixed with effluents within the interval of the HRT (20 min), as also documented by Lakretz et al. [32]. When H 2 O 2 was added, the measured H 2 O 2 concentration at the inlet was 27 ± 2 mg/L, and <1 mg/L at the outlet, which affirms full H 2 O 2 degradation. Furthermore, DO values after filtration were similar or slightly higher than before filtration. Thus, we conclude that H 2 O 2 decomposition in the system occurs almost solely via bacterial activity and functions as a source of available oxygen.

Reference Measurements-No Addition of H 2 O 2
Sampling campaigns were conducted without the addition of H 2 O 2 as a reference: DO values at the outlet were <1 mg/L, which reflects maximum oxygen utilization and anaerobic conditions inside the reactor [20]. The dissolved organic carbon (DOC)-and chemical oxygen demand (COD)-removal rates were similar (13% ± 0.3%, and 11% ± 2%, respectively) and were mainly caused by flocculation/filtration. and did not exceed 4 ± 1%, which suggests removal by filtration rather than via denitrification.

Organic Carbon Removal with H 2 O 2 Addition
The biofilter removal rates of DOC and UVA were stable at 22% ± 5% and 20% ± 3%, respectively, regardless of the variation in DOC (10.4 ± 1.4 mg/L) and UVA (217 ± 13 m −1 ) at the inlet. Without H 2 O 2 , the ∆DOC, ∆UVA 254 and ∆COD was 1.6 ± 0.4 mg/L, 7 ± 1 m −1 and 3.3 ± 1 mg/L, respectively. With addition of H 2 O 2 , the ∆DOC, ∆UVA 254 and ∆COD increased to 2.3 ± 0.7 mg/L, 44 ± 7 1/m and 14.6 ± 4 mg/L, respectively, indicating additional organic carbon utilization by bacteria (Appendix A Figure A5: concentrations and the removal of DOC and UVA 254 over sampling campaigns before the biofilter and after the biofilter). The oxygen utilization, in favor of nitrification rather than carbon oxidation, reflects the low biodegradability of organic matter, which is typical for effluent post-secondary treatment with flocculation-filtration removal [41][42][43][44]. This is also supported by no removal of most trace organic compounds in the biofilter, besides Acesulfame and Iopromide that were removed by 60% and 30%, respectively [34]. These results show that the biofilter was efficient and robust at removing particles, DOC, and UVA.

NH 4 + and NO 3 − Transformation, Variation, and Nitrogen Mass Balance
To better illustrate the potential nitrogen tranformations in the biofilter due to H 2 O 2 addition, and especially the NH 4 + removal (mg/L-N) ability of the system, the parameters at all the figures were sorted Clean Technol. 2020, 2 58 by NH 4 + concentrations at the inlet (BF) sampling point, as illustrated in Figure 2. It can be clearly observed that the concentration of NO 3 − that increased at the outlet of the biofilter is in parallel to the decrease in NH 4 + concentration ( Figure 2). Full NH 4 + removal with H 2 O 2 addition was obtained when the BF concentration was lower than 4 mg/L-N, whereas above 4 mg/L-N residual NH 4 + was detected at the AF sampling point. The average NO 2 − concentration of 0.64 ± 0.4 mg/L-N was fully removed after the biofilter (AF) for all samples (Supporting data Figure A6: removal of NO 2 − ), indicating rapid NO 2 − oxidation to NO 3 − (nitratation). Thus, H 2 O 2 addition enabled a significantly higher nitrification rate. Given that the nitrification rate in the biofilter was stable and was solely dependent on H 2 O 2 addition, regardless of particle numbers and removal, it is suggested that most of the microbial activity occurred in the biomass attached to the media, perhaps independent of the filtration cake and its size. The ammonium oxidation rate (AOR) (the ratio of formed NO 3 − to the transformed NH 4 + and NO 2 − ), is calculated according to Equation (2) and shown for all sampling points ONT = Ammonification, organic nitrogen transformed to NH 4 + Nit = Nitrification NX = Anoxic/AMX (anaerobic ammonia oxidation) nitrogen transformation to dinitrogen Inorganic nitrogen balance rate to indicate dominant pathway: In most samples, AOR revealed values of less than 100% (Figure 2, green dashed line), indicating the loss of ammoniacal nitrogen as dinitrogen, which will be further detailed.
Clean Technol. 2020, 2 FOR PEER REVIEW 6 trace organic compounds in the biofilter, besides Acesulfame and Iopromide that were removed by 60% and 30%, respectively [34]. These results show that the biofilter was efficient and robust at removing particles, DOC, and UVA.

NH4 + and NO3 − Transformation, Variation, and Nitrogen Mass Balance
To better illustrate the potential nitrogen tranformations in the biofilter due to H2O2 addition, and especially the NH4 + removal (mg/L-N) ability of the system, the parameters at all the figures were sorted by NH4 + concentrations at the inlet (BF) sampling point, as illustrated in Figure 2. It can be clearly observed that the concentration of NO3 − that increased at the outlet of the biofilter is in parallel to the decrease in NH4 + concentration ( Figure 2). Full NH4 + removal with H2O2 addition was obtained when the BF concentration was lower than 4 mg/L-N, whereas above 4 mg/L-N residual NH4 + was detected at the AF sampling point. The average NO2 − concentration of 0.64 ± 0.4 mg/L-N was fully removed after the biofilter (AF) for all samples (Supporting data Figure A6: removal of NO2 − ), indicating rapid NO2 − oxidation to NO3 − (nitratation). Thus, H2O2 addition enabled a significantly higher nitrification rate. Given that the nitrification rate in the biofilter was stable and was solely dependent on H2O2 addition, regardless of particle numbers and removal, it is suggested that most of the microbial activity occurred in the biomass attached to the media, perhaps independent of the filtration cake and its size. The ammonium oxidation rate (AOR) (the ratio of formed NO3 − to the transformed NH4 + and NO2 − ), is calculated according to Equation (2) and shown for all sampling points ONT = Ammonification, organic nitrogen transformed to NH4 + Nit = Nitrification NX = Anoxic/AMX (anaerobic ammonia oxidation) nitrogen transformation to dinitrogen Inorganic nitrogen balance rate to indicate dominant pathway: When AOR > 100%, ONT + Nit >> NX When AOR < 100%, significant NX When AOR ≈ 100%, Nit >> ONT, NX Figure 2. Concentrations of NH4 + before and after the filter, NO3 − after the filter, and the delta of NH4 + and NO3 − (mg/L-N) are presented. The gray dashed line represents the NH4 + oxidation rate, which is the ratio between the introduced NO3 − and the sum of the removed NH4 + and NO2 − (%). The green dashed line represents the complete (100%) calculated mass balance between removed NH4 + and NO2 − and introduced NO3 − . The sampling campaigns were sorted by feed values of NH4 + . The first three values represent the control without H2O2 addition.

Mass Balance
Calculations of the total nitrogen mass balance are presented in Figure 3, considering organic nitrogen values, revealed an average nitrogen loss of 2.24 ± 1 mg-N/L with H 2 O 2 and 0.25 ± 0.06 mg-N/L at the control. In addition, a higher nitrogen loss was calculated in parallel to the increase in NH 4 + concentration at the inlet. Calculations to evaluate the nitrogen loss in favor of assimilation with H 2 O 2 addition revealed that the assimilation rate was around 50% from all nitrogen removed (Table A1). This indicates that significant nitrogen loss was due to anoxic activity in the biofilter, which occurred simultaneously with nitrification. In addition, denitrification credit, the calculated oxygen that was credited due to utilization of NO 3 − as electron acceptor, was significant and varied between 1-7 mg/L of COD and also increased in parallel to the increase in NH 4 + concentrations at the inlet. To explain the higher nitrogen loss under aerobic conditions that was enhanced by the oxygen surpluss as H 2 O 2 , we hypothesized that H 2 O 2 , due to a lack of solubility limitations, better penetrates into the depth of the biofilm, decomposes, and stimulates activity.
Clean Technol. 2020, 2 FOR PEER REVIEW 7 In most samples, AOR revealed values of less than 100% (Figure 2, green dashed line), indicating the loss of ammoniacal nitrogen as dinitrogen, which will be further detailed.
Moreover, the calculated AOR was less than ~70% when NH4 + values at the BF sampling point were higher than 4 mg/L-N. Values over 100% can be explained by the contribution of NH4 + via a higher rate of organic nitrogen (amino acids) decomposition than nitritation. The relatively low removal rates of organic compounds by the biofilter when H2O2 was added, in comparison to NH4 + and NO2 − , suggests that the level of organic carbon degradability was a limiting factor in favoring oxygen utilization for nitrification on the one hand and carbon utilization for denitrification on the other.

Mass Balance
Calculations of the total nitrogen mass balance are presented in Figure 3, considering organic nitrogen values, revealed an average nitrogen loss of 2.24 ± 1 mg-N/L with H2O2 and 0.25 ± 0.06 mg-N/L at the control. In addition, a higher nitrogen loss was calculated in parallel to the increase in NH4 + concentration at the inlet. Calculations to evaluate the nitrogen loss in favor of assimilation with H2O2 addition revealed that the assimilation rate was around 50% from all nitrogen removed (Table A1). This indicates that significant nitrogen loss was due to anoxic activity in the biofilter, which occurred simultaneously with nitrification. In addition, denitrification credit, the calculated oxygen that was credited due to utilization of NO3 − as electron acceptor, was significant and varied between 1-7 mg/L of COD and also increased in parallel to the increase in NH4 + concentrations at the inlet. To explain the higher nitrogen loss under aerobic conditions that was enhanced by the oxygen surpluss as H2O2, we hypothesized that H2O2, due to a lack of solubility limitations, better penetrates into the depth of the biofilm, decomposes, and stimulates activity.

Functional Potential of the Biomass
The Metagenomics preformed on the sample yielded 8.2 M high-quality reads, out of which, 3.6 M were aligned to ~1250 organisms, when >98% are bacteria, both in the abundance and number of strains. The resulting outcome indicated a highly diverse bacterial community, with a calculated Shannon diversity index value of 5.6 as opposed to the theoretical value of 7.13 for a sample with equal diversity. The relative abundance of functional genes in the biofilter biomass is presented in Figure 4. The relative abundance of genes involved in nitrification (AmoA and NxrB) was higher, by 3 to 4 orders of magnitude, than the genes involved in other electron donor pathways (as drsEFH or aprA). In addition, the log relative abundance of the genes involved in denitrification and oxygen

Functional Potential of the Biomass
The Metagenomics preformed on the sample yielded 8.2 M high-quality reads, out of which, 3.6 M were aligned to~1250 organisms, when >98% are bacteria, both in the abundance and number of strains. The resulting outcome indicated a highly diverse bacterial community, with a calculated Shannon diversity index value of 5.6 as opposed to the theoretical value of 7.13 for a sample with equal diversity. The relative abundance of functional genes in the biofilter biomass is presented in Figure 4. The relative abundance of genes involved in nitrification (AmoA and NxrB) was higher, by 3 to 4 orders of magnitude, than the genes involved in other electron donor pathways (as drsEFH or aprA). In addition, the log relative abundance of the genes involved in denitrification and oxygen utilization (cyt c oxidases) was also significantly higher in comparison to the other various metabolic pathways examined.
The comparison of the functional potential of the microbial community reflects clear dominance for N biotransformation by simultaneous nitrification and denitrification. The significant anoxic potential reflected from the relative abundance of functional genes can be explained by the denitrifying community capable of being active at the presence of oxygen (low half-saturation rate of oxygen inhibition, KOI = 0.1-0.2 mg/L) [43].  Moreover, favoring oxygen utilization for nitrification, with a C/N ratio similar to wastewater, can be explained by the selection of organisms that are more competitive and characterized with a low substrate half-saturation rate coefficient (KONitrospira = 0.1-0.2 mg/L, KOAOB = 0.26-0.15 mg/L), which is more typical to nitrifiers in comparison to heterotrophs [51,52]. Jiang et al. (2013) also observed that the increase in oxygen availability in the upflow biofilter resulted in utilization by nitrifiers rather than heterotrophic biomass [19].

Denitrification
We suggest that the typical low biodegradability of the organic carbon at the secondary effluent, which was also reflected in the low removal rate of the COD, BOD, and UVA254, mentioned earlier (Section 3.4, Figure A5a,b in the Appendix A) was a key parameter on the rates of heterotrophic activity [44].
The environmental conditions of a low degradability rate of the organic matter and the addition of oxygen via H2O2, could be a possible explanation for favoring NH4 + rather than organic material oxidation. Such conditions, may promote simultaneous nitrification/nitrogen-removal activity. The significant relative abundance of genes with the potential for NO2 − oxidation on one hand and NO2 − Considering that cyt c is harbored by AOB (nitritation) [45], as well as denitrifying organisms (denitrification) [46], the relative high abundance of this gene is consistent with the experimental results of simultaneous nitrification and denitrification activity (Figure 2). The higher relative abundance obtained for the gene of NO 2 − oxidation (nxrB) over NH 4 + oxidation, might suggest the possible occurrence of a "ping pong" mechanism of cyclic NO 2 − oxidation and reduction, resulting in a higher NO 2 − oxidation rate than NH 4 + [47]. In parallel, the composition of identified genera of each of the N related genes (Figure 4b) revealed a variety of heterotrophs capable of conducting denitrification, when the genera belonging to the family Comamondaceae was the most dominant one. Interestingly, the identified genus related to NH 4 + oxidation was related to AOB and NOB, while the identified genus related to NO 2 − oxidation was of heterotrophs. This might suggest that the involvement of commamox and heterotrophs in nitrification [48][49][50]. The comparison of the functional potential of the microbial community reflects clear dominance for N biotransformation by simultaneous nitrification and denitrification. The significant anoxic potential reflected from the relative abundance of functional genes can be explained by the denitrifying community capable of being active at the presence of oxygen (low half-saturation rate of oxygen inhibition, K OI = 0.1-0.2 mg/L) [43].
Moreover, favoring oxygen utilization for nitrification, with a C/N ratio similar to wastewater, can be explained by the selection of organisms that are more competitive and characterized with a low substrate half-saturation rate coefficient (K ONitrospira = 0.1-0.2 mg/L, K OAOB = 0.26-0.15 mg/L), which is more typical to nitrifiers in comparison to heterotrophs [51,52]. Jiang et al. (2013) also observed that the increase in oxygen availability in the upflow biofilter resulted in utilization by nitrifiers rather than heterotrophic biomass [19].
We suggest that the typical low biodegradability of the organic carbon at the secondary effluent, which was also reflected in the low removal rate of the COD, BOD, and UVA 254 , mentioned earlier (Section 3.4, Figure A5a,b in the Appendix A) was a key parameter on the rates of heterotrophic activity [44].
The environmental conditions of a low degradability rate of the organic matter and the addition of oxygen via H 2 O 2, could be a possible explanation for favoring NH 4 + rather than organic material oxidation. Such conditions, may promote simultaneous nitrification/nitrogen-removal activity. The significant relative abundance of genes with the potential for NO 2 − oxidation on one hand and NO 2 − reduction on the other hand strongly suggest that competition for NO 2 − is a key in the dynamics of the community structure and function.
To address the paradox of excess oxygen availability in the reactor, due to H 2 O 2 addition and degradation, while the relative abundance of genes related to anoxic activity reflects conditions of limited oxygen, we hypothesize that this community structure reflects inner localization in the biofilm: Whereas AOB and NOB are localized at the outer layer of the biofilm, degrade the H 2 O 2 , and utilize the supplemental oxygen for nitrification, in parallel with the depletion of oxygen along the depth of the biofilm, NO 2 − and NO 3 − are reduced. It is also possible that the nitrifiers are located in the pockets inside the biofilm and the better mass transfer of H 2 O 2 enables utilization in a deeper location. We also suggest that the bacterial community that was investigated in this study indicates maximized thermodynamic utilization of electron acceptors and donors, with significant rates of anoxic activity. The limitation of the process performance under increasing loads of NH 4 + might be due to a lack of oxygen, limited HRT, or reactor volume and should be further investigated. As mentioned in previous research [34] the biofilter system succeeded in decreasing the oxygen demand and the drop in the ORP in following SAT. We suggest that the design presented in this study could be beneficial for treating secondary effluents containing residual levels of NH 4 + (few mg/L-N), which is required to be removed under discharge or reuse restrictions.

Discussion
We suggest a model to describe the biokinetics of tertiary wastewater biofiltration systems with the addition of H 2 O 2 . The model focuses on the soluble (not particulate) content, carbon content similar to the secondary effluent (low biodegradability, nitrogen content), and the significance of metabolism over assimilation (Appendix A.1 and Table A1). Generally, although some full reduction of NO 3 − to NH 4 + may occur simultaneously, it is neglected in the model.
The model suggest that the addition of H 2 O 2 ( Figure 5) promoted the rates of the following pathways: (i) nitritation and (ii) nitratation, which occurs at high rates, and (iii) biodegradable COD oxidation, which occurs at minor rates. In parallel, (v) NO 2 − and (vi) NO 3 − reduction via denitrification and (vii) are also stimulated.
metabolism over assimilation (Appendix A.1.1 and Table A1). Generally, although some full reduction of NO3 − to NH4 + may occur simultaneously, it is neglected in the model. The model suggest that the addition of H2O2 ( Figure 5) promoted the rates of the following pathways: (i) nitritation and (ii) nitratation, which occurs at high rates, and (iii) biodegradable COD oxidation, which occurs at minor rates. In parallel, (v) NO2 − and (vi) NO3 − reduction via denitrification and (vii) are also stimulated.

-
The addition of H2O2 to stimulate aerobic activity within a bed filtration process, practiced under the low range of the high rate bed filtration (5-6 m/h), enabled a combined effect of particle filtration and nitrification.The H2O2 was fully degraded, limiting the nitrification rate.

-
The trends in the concentration of NH4 + , NO2 − , and NO3 − with addition of H2O2 demonstrated significant nitirification activity at the bed filter. -Metagenome analysis results were in line with the performance obtained and reflected significant potential for the simultanious nitrification denitrification activity of the attached biomass.

-
The process presented herin may have a techno-economical potential, especially with the development of new and novel methods for the direct production of H2O2 [53,54] and the efficient application of the high-rate filtration.
-In cases of residual NH4 + concentration (<5 mg N-NH4 + /L) in secondary WWTP effluents, the presented technology shows potential for either managing ammonia concentration in the  The process presented herin may have a techno-economical potential, especially with the development of new and novel methods for the direct production of H 2 O 2 [53,54] and the efficient application of the high-rate filtration. -

Conclusions
In cases of residual NH 4 + concentration (<5 mg N-NH 4 + /L) in secondary WWTP effluents, the presented technology shows potential for either managing ammonia concentration in the effluent or for reducing oxygen demand in following processes, such as SAT or direct reuse, under local regulation. Appendix A

Appendix A.1 Theoretical Calculation of and Oxygen Nitrogen Mass Balance
Nitrogen mass balance was calculated by making the following assumptions: (1) ammonia was the only nitrogen source for assimilation for biomass growth, (2) full aerobic utilization of oxygen originated from H 2 O 2 , (3) full nitrification to nitrate as the only electron acceptor for denitrification and (4) all nitrogen loss originated in anoxic denitrification activity. Two values for yields of 0.6 mg COD biomass/mg substrate COD (aerobic) as the upper limit and 0.4 mg COD biomass/mg substrate COD (anoxic) [43] as the lowest limit were used in the calculations as heterotroph activity was probably both aerobic and anoxic. The data presents the delta between the inlet and outlet: including Organic nitrogen (TKN). Oxygen mass balance was calculated by making the following assumptions: (1) All oxygen transformations were biological.
(2) All the degradation of H 2 O 2 was biologically with oxygen and water as the only by products (based on previous lab calibrations).
(3) Excess oxygen, over the dissolved oxygen, in the inlet water was originated from H 2 O 2 degradation.
Degradation Considering the difference between the ammonia removed (theoretically oxidized to the oxygen balance it can be clearly seen that more NH 4 + was consumed than oxygen was available along the increase in NH 4 + concentration (negative values of mass balance).  Generally, the reaction occurs in two distinct stages in the catalytic reaction pathway. First, oxidation of the enzyme by one molecule of H 2 O 2 , while the oxygen-oxygen bond in peroxide is cleaved. At the second stage, another molecule of H 2 O 2 reacts with the bond oxygen ion and forms molecular oxygen and water [28].

Appendix A.5 Secondary Effluent Particle Analysis
Effluent particles and aggregates were analyzed by Micro Flow Imaging. In brief, a fluid sample is drawn through a flow cell, illuminated with a light-emitting diode at 470 nm, and the magnified image is captured by digital camera. This image is then automatically analyzed to determine the particles' equivalent circular diameter (ECD) which represents the diameter of a sphere that occupies the same two-dimensional surface area as the particle. Analysis was conducted on particles of between~2 and 400 µm.

Appendix A.6 Particles Distribution
A typical PSD analysis of the secondary effluent after 500 µm, post flocculation and after the biofilter is presented in Figure A1.
Clean Technol. 2020, 2 FOR PEER REVIEW 14 Figure A1. Particle distribution analysis over sampling campaigns. The Particle size is represented as an equivalent circular diameter (ECD). Total particle concentration (#/mL) is plotted as a red line. The results presented at Figure 2 refer to the campaign marked with a star. No HP is related to control runs without the addition of H2O2.

.7. Media Sampling and Metagenome Analysis
The metagenome analysis was conducted to investigate the functional potential of the microbial media community. At the middle of the operation cycle (6 h) the biofilter was emptied and a sample of 50 g anthracite media was taken from a depth of 40 cm from the top of the media, thoroughly mixed for homogenization and immediately stored in dry ice. DNA from the media was extracted using Exgene Soil DNA Mini Kit. Whole-genome (shotgun) libraries were prepared using the TruSeq DNA PCR-Free HT Library Preparation Kit (Illumina, San Diego, CA, USA) and sequenced using an

Appendix A.7 Media Sampling and Metagenome Analysis
The metagenome analysis was conducted to investigate the functional potential of the microbial media community. At the middle of the operation cycle (6 h) the biofilter was emptied and a sample of 50 g anthracite media was taken from a depth of 40 cm from the top of the media, thoroughly mixed for homogenization and immediately stored in dry ice. DNA from the media was extracted using Exgene Soil DNA Mini Kit. Whole-genome (shotgun) libraries were prepared using the TruSeq DNA PCR-Free HT Library Preparation Kit (Illumina, San Diego, CA, USA) and sequenced using an Illumina MiSeq benchtop sequencer with a target fragment length of 250 bp.
The final outputs of the metagenomics sequencing were modified using Cutadapt [55] to remove adapters and low-quality reads, and contigs were created using IDBA-UD [56]. A total of 8.9 M reads, was produced. The construction of a non-redundant gene catalog and the quantification of reference gene abundance (Gene calling) on the assembled scaftigs was performed using Prodigal [57]. Predicted genes were clustered using UCLUST [58]. High-quality reads were mapped to the reference gene catalog using BWA [59]. Mapped reads were used to form an abundance vector of the number of reads mapped to each gene. This was normalized based on data set size and gene length. Annotation was done using USEARCH [58] against the NCBI database (best hit with Eo1e-5, Bitscore >60 and sequence similarity >30%). Genus annotation was done using BWA aligner against the entire NCBI genome database (~80,000 genomes) and using PathoScope 2.0 [60] to quantify proportions of contigs from individual strains. The diversity of species within the sample indicating low diversity of the community in the sample. The number of reads that were aligned to nitrogen-related species were 43% of the total aligned reads, reflecting significant functional potential for the nitrogen cycle in the sampled bacterial community. Gene catalog was extracted from NCBI's nucleotide database. Mapped reads were subsequently filtered by removing those with a poor alignment (−q30). The remaining mapped reads were used to form an abundance matrix of the number of reads mapped to each gene in every sample. The abundance matrix was normalized based on data set. Flocculation before biofiltration increased particle counts, due to flocculation of the macromolecules and colloid particles (<2 µm) which were not analyzed. As expected, particle concentration decreased dramatically after biofiltration for all ECDs.
Appendix A.10 Organic Carbon Performance Figure A4. Particle-size distribution (PSD) analysis of a representative sampling campaign with H2O2. A comparison between sampling points after 500µ m filtration (1) at the inlet of the biofilter after coagulation and flocculation (2) and at the outlet of the biofilter (3). Flocculation before biofiltration increased particle counts, due to flocculation of the macromolecules and colloid particles (<2 µ m) which were not analyzed. As expected, particle concentration decreased dramatically after biofiltration for all ECDs.