Bromate Reduction by Iron(II) during Managed Aquifer Recharge: A Laboratory-Scale Study

Abstract

The removal of bromate (BrO₃⁻) as a byproduct of ozonation in subsequent managed aquifer recharge (MAR) systems has so far gained little attention. This preliminary study with anoxic batch experiments was executed to explore the feasibility of chemical BrO₃⁻ reduction in Fe-reducing zones of MAR systems and to estimate potential inhibition by NO₃⁻. Results show that the reaction rate was affected by initial Fe²⁺/BrO₃⁻ ratios and by pH. The pH dropped significantly due to the hydrolysis of Fe³⁺ to hydrous ferric oxide (HFO) flocs. These HFO flocs were found to adsorb Fe²⁺, especially at high Fe²⁺/BrO₃⁻ ratios, whereas at low Fe²⁺/BrO₃⁻ ratios, the mass sum loss of BrO₃⁻ and Br⁻ indicated intermediate species formation. Under MAR conditions with relatively low BrO₃⁻ and Fe²⁺ concentrations, BrO₃⁻ can be reduced by naturally occurring Fe²⁺, as the extensive retention time in MAR systems will compensate for the slow reaction kinetics of low BrO₃⁻ and Fe²⁺ concentrations. Under specific flow conditions, Fe²⁺ and NO₃⁻ may co-occur during MAR, but NO₃⁻ hardly competes with BrO₃⁻, since Fe²⁺ prefers BrO₃⁻ over NO₃⁻. However, it was found that when NO₃⁻ concentration exceeds BrO₃⁻ concentration by multiple orders of magnitude, NO₃⁻ may slightly inhibit BrO₃⁻ reduction by Fe²⁺.

1. Introduction

Managed aquifer recharge (MAR) has proven to be an effective barrier for many organic micropollutants (OMPs) present in surface water during drinking water production [1,2,3]. However, some highly persistent OMPs can still be detected in MAR filtrate [4] and may reach the drinking water supply [5]. Ozone-based advanced oxidation processes (AOPs) are increasingly being considered as effective alternatives for the removal of OMPs during drinking water treatment [6,7,8]. The combination of MAR with ozonation as a pretreatment has been suggested as a comprehensive multibarrier treatment system to effectively remove various OMPs during drinking water production [9,10,11]. However, bromate (BrO₃⁻) is formed during ozone-based treatment when applied to bromide-containing water [12,13,14]. It has been reported that the BrO₃⁻ concentration in drinking water after ozone-based AOPs typically ranges from 0 to 127 μg/L (1 μM) [15]. BrO₃⁻ is classified as Group 2B, or possible human carcinogen, by the International Agency for Research on Cancer based on its major toxic effects [16,17,18]. The standard of BrO₃⁻ in drinking water regulated by the World Health Organization, the US Environmental Protection Agency, and the European Union is 10 μg/L [19,20,21], demanding water companies to control the BrO₃⁻ concentration in drinking water.

A number of physical, chemical, electrochemical, and biological techniques for BrO₃⁻ removal have already been proposed. With respect to physical techniques, various advanced sorption materials, e.g., ion-exchange resins [22], nanocrystalline akaganeite (β-FeOOH)–coated quartz sand [23], and layered double hydroxides [24,25], have shown the ability to adsorb BrO₃⁻ from aqueous solutions, but so far these have not been applied in drinking water treatment. The use of granular activated carbon (GAC) as a conventional physical sorption technique can successfully reduce BrO₃⁻ [26], but the regenerated GAC loses effectiveness for BrO₃⁻ removal after a certain running time [15]. BrO₃⁻ can be removed by reverse osmosis [27], but this is an expensive process, since membrane fluxes are low and high operating pressures are needed. Electrodialysis reversal (EDR) has been studied in an integrated membrane system for drinking water treatment [28], which showed only limited BrO₃⁻ removal: 64% in a two-stage EDR system and 78% in a three-stage EDR system. BrO₃⁻ removal with catalysts, including zero-valent iron (Fe) [29] and Pd/Al₂O₃ [30], has been found to be limited in the presence of coexisting anions. Different reducing agents, such as ferrous iron (FeSO₄), react with dissolved oxygen (DO), and therefore their practical application during water treatment is quite difficult [31]. UV irradiation successfully reduces BrO₃⁻ but has a high energy demand [15], just like electrochemical methods [32,33]. With respect to biological techniques, biological activated carbon (BAC) filters are capable of reducing BrO₃⁻ effectively, but competitive DO remains a critical factor [34], because it is a challenge to construct a BAC filter with restricted oxygen transfer within the biofilm [35]. Hijnen et al. [36] showed that BrO₃⁻ was removed in a denitrifying bioreactor fed with methanol. However, they demonstrated that this did not seem to be a realistic option for drinking water treatment due to the long contact times required for BrO₃⁻ removal and extensive posttreatment necessary to remove excess methanol and released biomass. Altogether, there are few effective options to remove highly soluble and stable BrO₃⁻ in practice.

In this study, a new approach is being proposed, namely to utilize Fe-reducing zones of MAR as a barrier for BrO₃⁻ after ozonation. This sequence of AOP–MAR has been proposed to effectively remove various OMPs during drinking water production [9,10,11]. It is hypothesized that not only will the removal of OMPs improve with this sequence, but the produced BrO₃⁻ will be removed by MAR. Recently, it was found that BrO₃⁻ is partially biodegraded in NO₃⁻-reducing zones of MAR [37,38]. However, the potential reduction of BrO₃⁻ to Br⁻ in deeper Fe-reducing zones during soil passage has not yet been investigated.

The reduction of BrO₃⁻ by Fe²⁺ [31,39] and the hydrolysis of its product Fe³⁺ under near-neutral pH proceed as follows [40,41]:

BrO₃⁻ + 6 Fe²⁺ + 6 H⁺ → Br⁻ + 6 Fe³⁺ + 3 H₂O

(1)

Fe³⁺ + 3H₂O → Fe(OH)_{3 (s)} + 3H⁺

(2)

BrO₃⁻ + 6 Fe²⁺ + 15 H₂O → Br⁻ + 6 Fe(OH)_{3 (s)} + 12 H⁺

(3)

The reduction rate of BrO₃⁻ by Fe²⁺ is dependent on Fe²⁺ concentration, contact time, pH, and DO [31,42]. In MAR systems, water flows from infiltration ponds through an oxic zone, via an NO₃⁻-reducing anoxic zone and an Mn-reducing anoxic zone, to the Fe-reducing anoxic zone. So, depending on the local geochemical situation of MAR, Fe²⁺ may be released into the groundwater, leading to natural BrO₃⁻ reduction by Fe²⁺ in the Fe-reducing anoxic zone of MAR.

A study by Siddiqui et al. [31] with oxic water (0.22 mM DO) found that an initial BrO₃⁻ concentration of 0.4 µM was lowered to 0.08 µM within 30 minutes following a dose of 0.27 mM Fe²⁺. Dong et al. [42] worked with 0.2 µM BrO₃⁻, a 0.54 mM Fe²⁺ dosage, and 0.07 mM DO, reaching a BrO₃⁻ reduction of 65%. In these studies, the Fe²⁺ dosage was extremely high compared to expected Fe²⁺ concentrations during MAR, when Fe concentrations below 0.03 mM are expected (e.g., the MAR site of Dunea, the Netherlands, shows concentrations ranging from 0.0015 to 0.029 mM Fe). However, little is known about the reaction of BrO₃⁻ and Fe²⁺ in such low concentrations. Additionally, to what extent BrO₃⁻ reduction is possible at such low concentrations of Fe²⁺ is not known, although the extensive residence times in the subsurface do not require fast kinetics for this technology to be effective. Also, competition of BrO₃⁻ with DO is not a problem in these anoxic zones. Fe²⁺ can be formed only when NO₃⁻ as an electron acceptor is exhausted in anaerobic zones of MAR systems [43,44]. However, water containing NO₃⁻ and water containing Fe²⁺ from different pathways have been found to mix in specific zones of MAR [45], so NO₃⁻ and Fe²⁺ can be present simultaneously in anaerobic zones of MAR systems. This was confirmed by Dunea measurements, where NO₃⁻ and dissolved Fe have been detected simultaneously in the effluent of MAR sites (Scheveningen and Monster, the Netherlands). Therefore, NO₃⁻ may compete with BrO₃⁻ for reduction by Fe²⁺ [46,47,48] during MAR. The investigation of BrO₃⁻ reduction by Fe²⁺ in the presence of NO₃⁻ may be an important reference for the feasibility of BrO₃⁻ removal in Fe-reducing zones of MAR systems. Examples of stoichiometric equations for the reaction of NO₃⁻ and Fe²⁺ in which the stable endpoint is nitrogen gas are given below, but less complete reactions may have endpoints anywhere along the reduction pathway [49]:

10Fe²⁺ + 2NO₃⁻ + 14H₂O → 10FeOOH + N₂ + 18H⁺

(4)

15Fe²⁺ + NO₃⁻ + 13H₂O → 5Fe₃O₄ + N₂ + 28H⁺

(5)

The focus of this preliminary study was to investigate the mechanism of chemical BrO₃⁻ reduction by Fe²⁺ and the feasibility of BrO₃⁻ reduction by naturally occurring Fe²⁺ in the Fe-reducing anoxic zones of MAR systems, with an emphasis on the potential competition with or inhibition by NO₃⁻. Microbiological reactions and biochemical reactions were not included in this study.

2. Materials and Methods

2.1. Experimental Design

The research was designed with 2 sets of anoxic batch reactor experiments: (A) high Fe²⁺ and BrO₃⁻ concentrations to investigate reduction mechanisms, and (B) environmentally relevant concentrations of Fe²⁺ and BrO₃⁻ to simulate the concentrations during MAR. As the focus in all experiments was on chemical BrO₃⁻ reduction by Fe²⁺, no soil or sediment was added in the batch reactors. Both sets of experiments were executed in the absence and presence of NO₃⁻. An overview of all experiments is provided in Figure 1.

For the experiments with high Fe²⁺ and BrO₃⁻ concentrations, anoxic batch experiments were performed with 0.03 mM BrO₃⁻ and 0.26 or 1 mM Fe²⁺. The concentration of 0.26 mM Fe²⁺ is close to the required concentration to reduce 0.03 mM BrO₃⁻ according to the stoichiometry of Equation (1). The experiments were executed under 2 pH conditions: pH 7.0, which is realistic for MAR water, and pH 5.2 to slow down the reaction in order to identify potential intermediate species.

To investigate the competition between NO₃⁻ and BrO₃⁻, the same order of magnitude of NO₃⁻ (0.07 mM) and BrO₃⁻ (0.03 mM) was added to anoxic batch reactors, together with the Fe²⁺ (0.26 mM and 1 mM).

To simulate BrO₃⁻ reduction by Fe²⁺ at concentrations similar to MAR, the concentrations were lowered to 0.5 µM for BrO₃⁻ and 0.003–0.033 mM for Fe²⁺. The concentration of 0.003 mM Fe²⁺ was close to the stoichiometric amount to reduce 0.5 µM BrO₃⁻ (Equation (1)). These experiments were conducted at an initial pH of 7.0. The influence of NO₃⁻ was investigated by dosing with 0.16 mM NO₃⁻, which was 3 orders of magnitude greater than the concentration of BrO₃⁻. Our previous study showed that the NO₃⁻ concentration was in the range of 10.7 ± 6 mg/L in MAR influent water [37], so in this study 10 mg/L (0.16 mM) was chosen as a relevant concentration. BrO₃⁻ formation at concentrations ranging from <2–293 µg/L has been reported during ozonation of natural water under normal drinking water treatment conditions [28,50,51,52], but in 100 investigated drinking water utilities, BrO₃⁻ concentration was within the range of <2–60 µg/L after ozonation of water containing 2–429 µg/L Br⁻ [53,54]. For this study, it was decided to investigate the upper value of this range, so 60 μg/L BrO₃⁻ (0.5 μM) was dosed. All experiments were performed in duplicate.

2.2. Anoxic Batch Reactors

Four series of laboratory-scale batch experiments using 250 mL (A experiments) and 1 L (B experiments) glass bottles were carried out under anoxic conditions at a controlled temperature (11.5 ± 0.5 °C). Anoxic conditions were reached by flushing nitrogen gas until a DO concentration below 0.3 µM (0.01 mg/L) was achieved in the batch reactors. The mouths of the batch reactors were sealed with rubber stoppers to prevent DO intrusion. On the rubber stoppers, there were 2 needles with valves, used as a sampling point and a reagent dosing point.

Water samples were collected 8–10 times within 120 h contact time to determine the concentrations of BrO₃⁻, Br⁻, NO₃⁻, and Fe²⁺. In the 0.03 mM BrO₃⁻ experiments (A) and 0.5 µM BrO₃⁻ experiments (B), 3 mL and 50 mL per sample were collected, respectively. After sample collection, several drops of diluted ethylenediamine (EDA) solution (11%) was added to the samples to prevent reactions of residual chemicals [55].

To test the stability of the anoxic system, Fe²⁺ concentrations were monitored in the batch reactors after dosing with 0.033, 0.003, 0.26, or 1 mM Fe²⁺. The Fe²⁺ concentrations remained stable during the 120 h experiment (Figure 2), indicating that the system was well sealed, and therefore no Fe²⁺ oxidation by DO was observed.

2.3. Water and Chemicals

The water used in batch experiments was prepared using chemical reagents and deionized water from a Millipore Milli-Q system. Sodium bromate (NaBrO₃), sodium nitrate (NaNO₃), ferrous sulfate (FeSO₄·7H₂O), sodium bicarbonate (NaHCO₃), and EDA were purchased from Sigma-Aldrich (St. Louis, MO, USA). The amount of 2 mM NaHCO₃ was prepared for use as a pH buffer, and 0.2 M NaOH was prepared to further adjust the pH. To prevent Fe²⁺ oxidation in FeSO₄ solutions, the solutions were always prepared immediately before the experiments and concentrated acid (HCl) was used to acidify the FeSO₄ solutions to pH 2 [55]. All chemicals were of analytical grade.

2.4. Analytical Methods

DO and temperature were measured with an FDO^® 925 optical oxygen sensor (WTW) (Xylem Inc., Weilheim, Germany) and pH was measured with a SenTix^® 940 (WTW) electrode (Xylem Inc., Weiheim, Germany), both using the WTW Multi 3420 meter (Xylem Inc., Weiheim, Germany).

Fe²⁺ was measured by photometry using the Spectroquant^® iron test (Merck, Kenilworth, NJ, USA) with a detection range of 0.0002–0.09 mM. Dilution factors of 4 and 16 were needed to measure the Fe²⁺ in the experiments with dosages of 0.26 and 1 mM, respectively. For the dosages of 0.003 and 0.033 mM Fe²⁺, no dilution was required.

The NO₃⁻ concentration in all experiments was determined by an ion chromatograph (Metrohm 881 Compact IC pro-Anion, Metrohm AG, Herisau, Switzerland) with an A Supp 16-150/4.0 anion column (Metrohm AG, Herisau, Switzerland). For experiments using 0.03 mM BrO₃⁻ (A), BrO₃⁻ and Br⁻ were measured by the same equipment as for NO₃⁻. The detection limits of BrO₃⁻, Br⁻, and NO₃⁻ were 0.008 mM, 0.001 mM, and 0.002 mM, respectively. For experiments using 0.5 µM BrO₃⁻ (B), water samples were analyzed at Het Waterlaboratorium (Haarlem, the Netherlands), where an ion chromatograph (Dionex ICS-300, Thermo Fisher Scientific Inc., Waltham, MA, USA) with IonPac AS9SC column (250 mm × 4 mm ID, Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to measure BrO₃⁻. An ion chromatograph (Dionex ICS-1100, Thermo Fisher Scientific Inc., Waltham, MA, USA) with IonPac AG22 column (4 × 50 mm, Thermo Fisher Scientific Inc., Waltham, MA, USA) and IonPac AS22SC column (4 × 250 mm, Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to measure Br⁻. The detection limits for BrO₃⁻ and Br⁻ were 0.004 µM and 0.125 µM, respectively.

3. Results

3.1. BrO₃⁻ Reduction Rate and Mass Balance

Figure 3 presents the kinetics of BrO₃⁻ reduction and Br⁻ formation within 120 h after the addition of 0.26 and 1 mM Fe²⁺. The experiments were executed at pH 5.2 and 7.0, the latter being most representative of MAR water. For BrO₃⁻ (0.03 mM), the 0.26 mM Fe²⁺ dosage was close to the stoichiometric ratio (1 mol BrO₃⁻:6 mol Fe²⁺) according to Equation (1). For this particular setting, >90% of initial BrO₃⁻ reduced into Br⁻ within 120 h (Figure 3a,c). The BrO₃⁻ reduction fit second-order reaction kinetics well. Moreover, for pH 5.2 and 7.0, the kinetic constant was 2.5 and 3.3, respectively, and the BrO₃⁻ reduction rate was 0.00024 μM/min and 0.00026 μM/min, respectively, indicating that pH 7 promoted BrO₃⁻ reduction compared to pH 5.2. In the case of the 1 mM Fe²⁺ dosage, BrO₃⁻ reduction was accelerated, with almost 100% BrO₃⁻ reduction to Br⁻ within 1 h at pH 5.2 (Figure 3b) and pH 7.0 (Figure 3d). These results indicate that the higher the Fe²⁺ dosage, the higher the BrO₃⁻ reduction rate, which is in line with existing literature [31,42].

Figure 4 shows the consumed Fe²⁺/BrO₃⁻ ratios after 24, 48, and 120 h. In the case of the 1 mM Fe²⁺ dosage (corresponding to an initial ratio of Fe²⁺/BrO₃⁻ = 33), consumed Fe²⁺/BrO₃⁻ ratios were higher than the theoretical ratio of 6 according to Equation (1), which assumes a total reduction of BrO₃⁻ to Br⁻. After 24 h, the ratios were 8.0 and 9.2 for pH 5.2 and pH 7.0, respectively, with BrO₃⁻ reduced to below the detection limit. Between 24 and 120 h, Fe²⁺ continued to be consumed, and correspondingly, Fe²⁺/BrO₃⁻ ratios increased. This may be explained by Fe²⁺ adsorption onto hydrolyzed Fe³⁺ flocs of hydrous ferric oxide (HFO). Interestingly, there was a consumed Fe²⁺/BrO₃⁻ ratio below the stoichiometric ratio of 6 in the case of the 0.26 mM Fe²⁺ dosage (corresponding initial Fe²⁺/BrO₃⁻ = 8). The ratio below 6 could indicate the production of intermediate Br species during the reduction of BrO₃⁻, requiring less Fe²⁺ compared to the total reduction of BrO₃⁻ to Br⁻ as in Equation (1). Additionally, the molar mass sum of BrO₃⁻ and Br⁻ slightly decreased during the experiment by 10–20%, indicating that intermediate products may have formed.

3.2. NO₃⁻, A Competing Electron Acceptor?

NO₃⁻ is known to act as a competitive electron acceptor in the reaction with Fe²⁺ [48]. Figure 5 depicts BrO₃⁻ reduction by Fe²⁺ in the presence of NO₃⁻ at a concentration at the same order of magnitude as BrO₃⁻ (0.07 mM). The rate of BrO₃⁻ reduction in the presence of NO₃⁻ was slightly lower compared to the absence of NO₃⁻ (Figure 3c). NO₃⁻ concentrations in these experiments were steady during the 120 h for both Fe²⁺ dosages (Figure 5a,b), indicating that Fe²⁺ did not reduce NO₃⁻ when BrO₃⁻ and NO₃⁻ were present simultaneously.

Figure 6 shows BrO₃⁻ and Fe²⁺ consumption in the presence and absence of 0.07 mM NO₃⁻. BrO₃⁻ removal in the presence and absence of NO₃⁻ was the same for both Fe²⁺ dosages, while the presence of NO₃⁻ led to higher Fe²⁺ consumption with the 0.26 mM Fe²⁺ dosage. The additional Fe²⁺ removal (62%→71%), 0.02 mM, might have reacted with NO₃⁻, but the change would have remained undetected, given that it would have resulted in a calculated reduction of <0.005 mM NO₃⁻ (NO₃⁻/Fe²⁺ ratio, Equations (4) and (5)). This would not have been noted by our NO₃⁻ analytical methods. Nevertheless, based on the above results, it can be concluded that BrO₃⁻ reduction was hardly affected by NO₃⁻ presence and that Fe²⁺ preferred BrO₃⁻ to NO₃⁻ as an electron acceptor. This was also observed by Westerhoff [56], who suggested that the difference in structure (atomic radii and O-bonds) makes it relatively easier to remove an O atom from a BrO₃⁻ ion than a NO₃⁻ ion. In addition, the preference of Fe²⁺ for BrO₃⁻ reduction vs. NO₃⁻ reduction can also be attributed to the higher ΔΕ [57,58,59] of oxidation-reduction reactions:

Fe²⁺ → Fe³⁺ + e, E₁ = −0.771 V

(6)

2BrO₃⁻ + 12H⁺ + 10e → Br₂ + 6H₂O, E₂ = 1.52 V

(7)

Br₂ + 2e → 2 Br⁻, E₃ = 1.087 V

(8)

NO₃⁻ +3H⁺ + 2e → HNO₂ + H₂O, E₄ = 0.94 V

(9)

HNO₂ + H⁺ + e → NO + H₂O, E₅ = 1.00 V

(10)

Reactions 6 and 7 are preferable over 6 and 9.

3.3. pH Change and Fe³⁺ Hydrolysis

Although the BrO₃⁻ reduction in Equation (1) shows a pH increase, reduction of BrO₃⁻ by Fe²⁺ consequently means that Fe²⁺ is oxidized to Fe³⁺, and subsequently Fe³⁺ will hydrolyze to form flocs of hydrous ferric oxide (HFO) [40]. Therefore, the pH will drop based on Equation (3), the combined BrO₃⁻ reduction with Fe³⁺ hydrolysis. The pH drop was observed in all of the 0.03 mM BrO₃⁻ experiments: 1.5–1.6 drop and 2.6–3.0 drop with initial pH 5.2 and 7.0, respectively. The pH drop was an indicator of Fe³⁺ hydrolysis. Moreover, the observed yellow flocs in the batch reactors also provide evidence of HFO formation. The above two phenomena (pH decrease and visible flocs) are a strong indication that HFO flocs were formed in the reactors. The adsorption of Br⁻ or BrO₃⁻ onto HFO flocs was not expected to have occurred, as BrO₃⁻ and Br⁻ have no affinity for HFO [60]. However, Fe²⁺ adsorption onto the flocs has been frequently reported [31,61,62], which may explain the observed Fe²⁺/BrO₃⁻ removal ratios beyond the stoichiometric ratio of 6 (in Figure 4).

3.4. BrO₃⁻ Reduction under Concentrations Similar to MAR

To investigate the rate of BrO₃⁻ reduction by Fe²⁺ in concentrations similar to MAR, the reduction kinetics were monitored for 0.5 µM BrO₃⁻ after dosing with 0.003 and 0.033 mM Fe²⁺. Figure 7a,b show the BrO₃⁻ and Br⁻ kinetics in the absence of NO₃⁻, while Figure 7c,d show the kinetics of BrO₃⁻ and Br⁻ in the presence of 0.16 mM NO₃⁻. As in the previous experiments with high BrO₃⁻ concentrations, the reduction rate of 0.5 µM BrO₃⁻ also depends on the Fe²⁺ concentration, with a higher rate at a higher concentration. Figure 7a,c show that after 120 h contact time, there was limited BrO₃⁻ reduction (7% in the absence of NO₃⁻ and 12% in the presence of NO₃⁻) at 0.003 mM Fe²⁺, while Figure 7b,d show considerable BrO₃⁻ reduction at 0.033 mM Fe²⁺ (74% in the absence and 58% in the presence of NO₃⁻). Assuming second-order reaction kinetics, it was calculated that the rate constant of BrO₃⁻ reduction at a higher Fe²⁺ dosage (0.033 mM) was 0.049 and 0.023 in the absence and presence of NO₃⁻, respectively. Although the NO₃⁻ concentration was three orders of magnitude higher than the BrO₃⁻ concentration, the NO₃⁻ concentration was steady (Figure 7c,d). It is noteworthy that during these experiments the molar mass sum of BrO₃⁻ and Br⁻ also slightly decreased from 0.50 µM to 0.48 µM and 0.46 µM for 0.003 mM and 0.033 mM Fe²⁺ dosages, respectively, indicating the formation of Br intermediate species.

Figure 8 shows reduction of BrO₃⁻ and consumption of Fe²⁺ in the presence and absence of 0.16 mM NO₃⁻. In the case of the 0.003 mM Fe²⁺ dosage, it appears that the presence of NO₃⁻ did not influence BrO₃⁻ reduction and Fe²⁺ oxidation (Figure 8a). In the case of the 0.033 mM Fe²⁺ dosage, the presence of NO₃⁻ led to lower BrO₃⁻ reduction and lower Fe²⁺ oxidation (Figure 8b). Combining the results in Figure 7 and Figure 8 indicates that Fe²⁺ preferred BrO₃⁻ to NO₃⁻ as an electron acceptor, but it did inhibit BrO₃⁻ reduction to some extent. This could possibly have been set off by considerably higher NO₃⁻ concentrations compared to BrO₃⁻, in combination with the stoichiometric excess of Fe²⁺. One potential reason to explain the inhibition by NO₃⁻ is the hypothesized formation of NO from NO₃⁻ complexed with Fe²⁺ [63], slowing down the reduction of BrO₃⁻.

4. Discussion

4.1. BrO₃⁻ Reduction Mechanism

Figure 9a shows a summary of the mole sum of BrO₃⁻ and Br⁻ in all of the experiments with an initial dosage of 0.03 mM BrO₃⁻. No Br mass loss was observed in the case of a sufficiently high initial Fe²⁺/BrO₃⁻ ratio (33), while the mole sum of BrO₃⁻ and Br⁻ was 78–90% of the initial Br in the case of lower stoichiometric Fe²⁺/BrO₃⁻ ratios (8). It is possible that Br intermediate species formed during the reduction of BrO₃⁻. Equations (11)–(13) show the reduction pathways of BrO₃⁻ to intermediate species requiring less Fe²⁺, as reported by Shen et al. [60] and Siddiqui et al. [31]:

BrO₃⁻ + 2Fe²⁺ + 2H⁺ → 2Fe³⁺ + BrO₂⁻ + H₂O

(11)

BrO₂⁻ + 2Fe²⁺ + 2H⁺ → 2Fe³⁺ + BrO⁻ + H₂O

(12)

BrO⁻ + 2Fe²⁺ + 2H⁺ → 2Fe³⁺ + Br⁻ + H₂O

(13)

The most frequently reported intermediate species is hypobromous acid (HOBr/BrO⁻) [31,64]. Furthermore, the study of Shen et al. [60] showed that the sum of BrO₃⁻, HBrO/BrO⁻, and Br⁻ was 98–101% of the initial Br (as BrO₃⁻) concentration, and therefore almost no other intermediate species except for HOBr/BrO⁻ existed. Taken together, BrO₃⁻ was reduced into the end product Br⁻ most likely via the intermediate species HBrO/BrO during the reaction of BrO₃⁻ and Fe²⁺ in this study.

Figure 9b shows the mole sum change of BrO₃⁻ and Br⁻ in the case of an initial Fe²⁺/BrO₃⁻ ratio of 8 at pH 5.2 and 7.0. More Br loss was observed at pH 5.2 (22%) than at pH 7.0 (12%). Based on the total reaction (Equation (3)), pH 5.2 would slow down the BrO₃⁻ reduction, providing the intermediate species with a longer lifetime and thus a better chance to be detected in this experiment. Moreover, the intermediate species formation requires less Fe²⁺, as shown in Equations (11)–(13). This may be one possible reason for the observation in Figure 4: a relatively low consumed Fe²⁺/BrO₃⁻ ratio at pH 5.2 compared to pH 7.0. However, the lower ratio for pH 5.2 can also be partially explained by the decreased sorption of Fe²⁺ onto precipitating HFO in this experiment, while the higher Fe²⁺/BrO₃⁻ ratio for pH 7.0 correlates with promoted Fe²⁺ sorption at higher pH [61]. The lower Fe²⁺/BrO₃⁻ ratio at pH 5.2 compared to pH 7.0 in Figure 4 may also be related to the surface charge density. It has been reported that the lower the pH of FeOOH formation, the higher the positive surface charge density [65]. So, at pH 5.2, the formed FeOOH presents high positive charge density, potentially promoting BrO₃⁻ adsorption and Fe²⁺ rejection, which in turn may explain the lower Fe²⁺/BrO₃⁻ ratio. In contrast, at pH 7, the neutral to positively charged surface of FeOOH will favor Fe²⁺ adsorption and BrO₃⁻ rejection.

4.2. Feasibility of BrO₃⁻ Reduction by Fe²⁺ during MAR

Based on the results in Section 3.1 and Section 3.4, a preliminary conclusion can be drawn that under anoxic conditions and at a sufficiently high Fe²⁺/BrO₃⁻ ratio, chemical BrO₃⁻ reduction can be achieved. In MAR systems, Fe²⁺ concentrations tend to be 10⁻³ to 10⁻² mM. Fortunately, the same is the case for BrO₃⁻ production after ozone-based AOPs, where concentrations are generally limited to 10⁻⁵ to 10⁻⁴ mM [16,66]. Fe²⁺ concentrations detected in Dunea MAR effluent range from 0.0015–0.029 mM, so the Fe²⁺/BrO₃⁻ ratios in MAR systems are sufficiently high (15–2900). From a drinking water production perspective, the extremely slow BrO₃⁻ reduction shown in Section 3.4 might seem to be a very inefficient process, since treatment technologies most often have contact times of minutes. However, MAR residence times in the subsurface are weeks to months [67,68], making this process a very viable BrO₃⁻ removal pathway. Assuming that Fe²⁺ and BrO₃⁻ concentrations in Fe-reducing anoxic zones and BrO₃⁻ reduction follow second-order kinetics, as in Figure 7b (k₂ = 0.049), the required time to reduce BrO₃⁻ below the drinking water guideline of 10 μg/L (0.08 µM) is on the order of 10–20 days.

As stated previously, the theoretical sequence of MAR infiltration zones follows the sequence of oxic-NO₃⁻-reducing-Mn-reducing-Fe-reducing-SO₄²⁻-reducing [69], but the possible practical cross of different flowlines may result in the joint presence of NO₃⁻ and Fe²⁺. The results in Figure 7 indicate a small negative effect of NO₃⁻ as an inhibitor for BrO₃⁻ reduction by Fe²⁺, though at sufficiently high Fe²⁺ concentrations, bromate reduction is still not inhibited. Although NO₃⁻ reduction by Fe²⁺ is thermodynamically not feasible, in the presence of catalysts this reaction may occur [70]. A previous study reported that the presence of Ni²⁺, Cu²⁺, and Ag²⁺ promoted the reaction of Fe²⁺ with NO₃⁻ [48]. Given the presence of these elements in nature (for example, the concentration of Cu²⁺ at Dunea’s MAR site is 10⁻² mM), these may well set off NO₃⁻ reduction by Fe²⁺. Moreover, previous studies [71,72,73] reported NO₃⁻-dependent Fe²⁺ oxidation mediated by anaerobic ammonium oxidation bacteria, Escherichia coli, and NO₃⁻-reducing bacteria. Therefore, a microbial mediated kinetic reaction of Fe²⁺ and NO₃⁻ could also occur, leading to competition for BrO₃⁻ reduction in these mixing flow paths in MAR systems.

Altogether, this study has shown that chemical BrO₃⁻ reduction by Fe²⁺ is expected to occur in Fe-reducing anoxic zones during MAR and that NO₃⁻ on its own is not a strong inhibitor or competitor; nevertheless, the complexity of subsurface processes may still set off conditions where NO₃⁻ reduction is favored over BrO₃⁻. Therefore, a subsequent study to investigate BrO₃⁻ reduction in simulated Fe-reducing zones, such as a column study, is highly recommended, also to include microbiological and biochemical processes that take place during MAR.

5. Conclusions

Based on anoxic batch experiments, it is concluded that BrO₃⁻ is readily reduced by Fe²⁺. The reaction rate was influenced by the initial Fe²⁺/BrO₃⁻ ratio and by the initial pH, i.e., a higher Fe²⁺ concentration and higher pH accelerated the reaction. The pH dropped considerably during the experiments, set off by the hydrolysis of Fe³⁺ to HFO flocs. These HFO flocs were found to adsorb Fe²⁺, particularly at high Fe²⁺/BrO₃⁻ ratios, whereas at low Fe²⁺/BrO₃⁻ ratios the incomplete BrO₃⁻–Br⁻ mass balance indicated formation of intermediate species. Overall, it can be concluded that the chemical reduction of BrO₃⁻ by naturally occurring Fe²⁺ during MAR can occur, as extensive retention times in the subsurface will compensate for the slow reaction kinetics of low BrO₃⁻ and Fe²⁺ concentrations. In the specific case that Fe²⁺-containing and NO₃⁻-containing water crosses flow paths during MAR, the presence of NO₃⁻ will not compete with BrO₃⁻, as Fe²⁺ is preferred over NO₃⁻ as an electron acceptor. However, it was found that the presence of NO₃⁻ may somewhat inhibit BrO₃⁻ reduction when NO₃⁻ concentrations are far higher than BrO₃⁻ concentrations. The findings in this study show that application of MAR following ozone-based AOPs broadens the application of ozone-based AOPs, as MAR removes the byproduct BrO₃⁻.