Selective and Efficient Reduction of Nitrate to Gaseous Nitrogen from Drinking Water Source by UV/Oxalic Acid/Ferric Iron Systems: Effectiveness and Mechanisms

Abstract

Nitrate (NO₃⁻) reduction in water has been receiving increasing attention in water treatment due to its carcinogenic and endocrine-disrupting properties. This study employs a novel advanced reduction process, the UV/oxalic acid/ferric iron systems (UV/C₂O₄²⁻/Fe³⁺ systems), in reducing NO₃⁻ due to its high reduction efficiency, excellent selectivity, and low treatment cost. The UV/C₂O₄²⁻/Fe³⁺ process reduced NO₃⁻ with pseudo-first-order reaction rate constants of 0.0150 ± 0.0013 min⁻¹, minimizing 91.4% of 60 mg/L NO₃⁻ and reaching 84.2% of selectivity for gaseous nitrogen after 180 min at pH_ini. 7.0 and 0.5 mg/L dissolved oxygen (DO). Carbon dioxide radical anion (CO₂^•−) played a predominant role in reducing NO₃⁻. Gaseous nitrogen and NH₄⁺, as well as CO₂, were the main nitrogen- and carbon-containing products, respectively, and reduction pathways were proposed accordingly. A suitable level of oxalic acids (3 mM) and NO₃⁻ (60 mg/L) was recommended; increasing initial iron concentrations and UV intensity increased NO₃⁻ reduction. Instead, increasing the solution pH decreased the reduction, and 0.5–8.0 mg/L DO negligibly affected the process. Moreover, UV/C₂O₄²⁻/Fe³⁺ systems were not retarded by 0.1–10 mM SO₄²⁻ or Cl⁻ or 0.1–1.0 mM HCO₃⁻ but were prohibited by 10 mM HCO₃⁻ and 30 mg-C/L humic acids. There was a lower reduction of NO₃⁻ in simulated groundwater (72.8%) than deionized water after 180 min at pH_ini. 7.0 and 0.5 mg/L DO, which meets the drinking water standard (<10 mg/L N-NO₃⁻). Therefore, UV/C₂O₄²⁻/Fe³⁺ systems are promising approaches to selectively and efficiently reduce NO₃⁻ in drinking water.

1. Introduction

Nitrate (NO₃⁻) naturally exists in some geological formations and groundwater. It should be noted that NO₃⁻ contamination mainly results from anthropogenic activities, such as fertilizer runoff in farmland, rainwater runoff on the urban surface, and the discharge of sewage or treated wastewater [1]. NO₃⁻ causes adverse effects to human health, such as known methemoglobinemia, carcinogens, and endocrine disruptors [2,3,4]. To minimize its adverse health effects, the World Health Organization has set the guideline of 50 mg/L NO₃⁻ (~11 mg/L as N-NO₃⁻) in drinking water [1]. The United States Environmental Protection Agency (US EPA) and China, however, have regulated more stringent levels of NO₃⁻ at 10 mg/L N-NO₃⁻ [2]. Furthermore, China has also promulgated an acceptable level of ≤10 and ≤20 mg/L of N-NO₃⁻, in level I and level II water sources, respectively. As reported, NO₃⁻ concentrations can reach up to 300 mg/L in the drinking water and groundwater of Northern China [5]. Since NO₃⁻ is a stable, highly mobile, and highly soluble oxyanion, it further underscores the importance of NO₃⁻ reduction.

Currently, there are several major technologies for minimizing NO₃⁻, including biological methods [6,7], catalytic reduction [8], electrocatalysis [9,10], photocatalysis [11,12], and membrane technologies [13]. Although reverse osmosis has already been implemented in a practical process [14], its development has been limited due to its large investment costs and high operating expenses. The other existing technologies mentioned above have been barely promoted and applied in actual water treatment systems. On the contrary, photocatalysis has received much attention due to its high performance, excellent stability, and easier combination with ultraviolet illumination, primarily including homogeneous and heterogeneous photocatalytic processes [15,16,17,18]. Furthermore, reducing radicals, such as aqueous electrons (e_aq⁻) and carbon dioxide radical anions (CO₂^•−), is responsible for the photocatalytic system [19]. Admittedly, e_aq⁻-based systems can highly effectively decompose various contaminants at nearly diffusion-limited rates (10⁹–10¹⁰ M⁻¹ s⁻¹) [20,21,22] because of their very high reduction potential (e.g., −2.9 V vs. standard hydrogen electrode (SHE) for e_aq⁻), including bromate [20,23], perchlorate [24,25], chlorate [26], nitrate [27], and halogenated organic compounds [28,29]. However, they have low selectivity and thus can be largely affected by competing background compounds in real water. These further suggest their low potential for practical applications.

On the other hand, CO₂^•− is generally formed from different hole scavengers and is a very strong one-electron reductant with a high reduction potential of E⁰(CO₂/CO₂^•−) = −1.81 V vs. SHE [19]. The CO₂^•−-related process has recently received considerable attention for water environmental remediation, i.e., efficiently removing a wide range of pollutants, including trichloroacetic acid [30], carbon tetrachloride [31], hexavalent chromium [32], divalent mercury [33], nitrate [34], etc. Although the formate-radical-induced photochemical process could efficiently remove NO₃⁻, it requires larger formic acid doses and also likely produces toxic and harmful products, such as formic acid. Reportedly, An et al. [35] found that CO₂^•− generated from Fe(III)/oxalate/UV systems played a predominant role in the effective reduction of nitrite to N₂. Additionally, there is a high likelihood that the Fe(III)/oxalate/UV systems could selectively reduce NO₃⁻ without decreasing the decontamination rate. To the best of our knowledge, there has been no report of the UV/C₂O₄²⁻/Fe³⁺ process reducing NO₃⁻ in water under neutral conditions. Therefore, we employed the CO₂^•−-associated process, Fe(III)/oxalic acid/UV systems (UV/H₂C₂O₄/Fe³⁺ systems), in the treatment of NO₃⁻ in drinking water, mainly because they have less involvement with reactants, efficient NO₃⁻ removal, and promote further formation of innocuous products, including CO₂ and iron precipitates.

In this work, the efficiency of NO₃⁻ reduction and gaseous nitrogen conversion was first carried out in circumneutral environments in the UV/H₂C₂O₄/Fe³⁺ process. We further investigated the effects of important operating parameters, i.e., initial concentrations of oxalic acids and iron dosage, initial nitrate levels, solution pH, UV intensity, and dissolved oxygen, as well as background compounds, i.e., chloride, sulfate, hydrocarbonate, and organic matters. The mechanism of NO₃⁻ reduction was carried out in the UV/H₂C₂O₄/Fe³⁺ system. Reduction pathways of NO₃⁻ were proposed accordingly. Finally, the reduction kinetics of NO₃⁻ was investigated in simulated groundwater to verify the effectiveness of UV/H₂C₂O₄/Fe³⁺ systems in actual water.

2. Experimental Section

2.1. Materials

Potassium nitrate (KNO₃, ≥99.0%), iron(III) chloride (FeCl₃, ≥97.0%), nitrate nitrite (NaNO₂, ≥99.0%), oxalic acid (H₂C₂O₄, 98.0%), sodium hydroxide (NaOH, ≥96.0%), hydrogen chloride (HCl, 36.0–38.0%), sodium sulfate (Na₂SO₄, ≥99.0%), sodium chloride (NaCl, ≥99.8%), sodium hydrogen carbonate (NaHCO₃, ≥99.5%), calcium chloride (CaCl₂, ≥96.0%), magnesium sulfate (MgSO₄, ≥98.0%), potassium iodide, (KI, ≥99.0%), potassium iodate (KIO₃, ≥99.8%), humic acid (HA), and Nessler’s reagent were obtained from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). Most chemicals were at least of analytical grade and used as received. All solutions were prepared in Milli-Q ultrapure water (18.2 MΩ cm, Millipore). The simulated groundwater was utilized for testing the performance of UV/H₂C₂O₄/Fe³⁺ systems. The major characteristics of simulated groundwater, which were nearly the same as the components of realistic groundwater, are summarized in Table S1.

2.2. Experimental Procedures

The photoreactor and methodology for irradiation experiments have been depicted elsewhere [23,36]. Generally, we selected a medium-pressure mercury UV lamp (UV-M) (500 W, Xujiang Electromechanical Plant, Nanjing, China). The UV irradiation intensity was estimated to be 10.2 mW/cm².

Prior to photocatalytic reduction experiments, the photoreactor was preheated for about 15 min to achieve stabilization. Meanwhile, stock solutions of KNO₃ (6 g/L), FeCl₃ (5 mM), H₂C₂O₄ (300 mM), Na₂SO₄ (10 or 100 mM), NaCl (10 or 100 mM), NaHCO₃ (10 or 100 mM), and HA (100 mg/L) were prepared. Magnetic stirring was utilized for the complete mixing of the solution. Reductive reactions were initiated by adding KNO₃, FeCl₃, and H₂C₂O₄ in aqueous solutions to a quartz tube. Samples were then taken at regulated time intervals, including 0, 20, 40, 60, 80, 100, 120, and 180 min, and immediately analyzed for N-containing compounds (i.e., NO₃⁻ NO₂⁻, NH₄⁺, and Total N (TN)), residual DOC, and iron ion levels. Furthermore, the impact of important variables on NO₃⁻ reduction was determined for the UV/C₂O₄²⁻/Fe³⁺ systems by varying initial concentrations of oxalic acid (1–6 mM), ferric iron (0–0.1 mM), and NO₃⁻ (20–100 mg NO₃⁻/L) and varying the solution pH (3–11) and UV lamp power (100, 300, and 500 W), as well as dissolved oxygen level (0.5 and 8.0 mg/L). Moreover, to determine the application potential, the reduction of NO₃⁻ was investigated in simulated groundwater for an initial concentration of 120 mg NO₃⁻/L in the UV/H₂C₂O₄/Fe³⁺ process. Additionally, a CO₂^•−-quenching experiment was conducted to confirm the predominant species in the process. All experiments were performed in a photoreactor, as illustrated in Figure 1.

Unless otherwise noted, the initial NO₃⁻ concentration was 60 mg NO₃⁻/L, representing NO₃⁻ levels (60–70 mg NO₃⁻/L) in water from a reservoir in Qingdao, China. The solution temperature was maintained at 25 (± 0.5) °C. The initial solution pH was adjusted using either 0.1 mM HCl or NaOH to 3.0, 5.0, 7.0, 9.0, and 11.0 in different experiments. All experiments proceeded without adding a buffer solution to simulate practical water treatment processes. The corresponding variation in solution pH was monitored over time, which indicated an actual water treatment process. Dissolved oxygen (DO) was used in concentrations of 0.5 and 8.0 mg/L with and without purging using nitrogen gas for about 60 min. All experiments were repeated in triplicate independently, and average values along with the standard deviation are presented.

2.3. Chemical Analysis

Solution pH was determined with a pH meter (LA-pH 10, HACH, Loveland, CO, USA), and DO concentrations were measured with a DO meter (HQ30d, HACH, Loveland, CO, USA). Dissolved organic carbon (DOC) and the total nitrogen (TN) were analyzed by a TOC analyzer (TOC-L, Shimadzu, Kyoto, Japan). The ferrous ion concentration was immediately measured without filtering by the 1,10-phenanthroline colorimetric method at 510 nm using an ultraviolet-visible spectrophotometer (HACH DR1900, Loveland, CO, USA) [37]. According to our previous study, there were no significant differences in Fe(II) levels without and with filtration [38]. Inorganic anions (NO₃⁻ and NO₂⁻) were quantified using an ion chromatograph (Dionex ICS-1000, Sunnyvale, CA, USA) equipped with a Dionex AS19 analytical column (4 × 250 mm) and an AG19 guard column (4 × 50 mm) for analysis of NO₃⁻ and NO₂⁻. Potassium hydroxide at a concentration of 20 mM was utilized as an effluent in an equal washing mode. Furthermore, a flow rate of 1.0 mL/min, a constant suppressor current of 50 mM, a column temperature of 30 °C, an inlet ring of 25 μL, and a duration of 25 min were set during the measurement. Of note, all samples were filtered using 0.22 μm membrane filters (Nylon, Titan) before the analysis of ion chromatography. Ammonium ion (N-NH₃) was further measured by UV-vis spectrophotometer (DR6000, HACH, USA), with a measurement range from 0.02 to 2.50 mg/L N-NH₃.

3. Results and Discussion

3.1. Photocatalytic Reduction Efficiency of Nitrate and Gaseous Nitrogen Selectivity

Figure 2 shows the reduction efficiency of NO₃⁻ in UV/C₂O₄²⁻/Fe³⁺ systems. As expected, UV irradiation alone and UV/Fe³⁺ systems reduced NO₃⁻ by 24.3% and 25.0%, respectively, within 180 min. In contrast, NO₃⁻ was effectively reduced (removal efficiency = 41.3%) after a reaction time of 180 min in the UV/C₂O₄²⁻ process as shown in Figure 2a. Comparatively, a more rapid reduction in NO₃⁻ levels was also observed within 180 min with a removal efficiency of 91.4% for an initial NO₃⁻ concentration of 60 mg/L in the UV/C₂O₄²⁻/Fe³⁺ systems (Figure 2a), which was larger than in previous studies (i.e., removal efficiency = 92.4% in 6 h) [39,40]. Furthermore, the reduction of NO₃⁻ follows the pseudo-first-order decay kinetics as presented in Figure 2b, where the ln(C₀/C) value is proportional to the reaction time. Figure 2c shows that the reaction rate constants for 180 min in the UV/C₂O₄²⁻/Fe³⁺ process, UV/C₂O₄²⁻ systems, UV/Fe³⁺ systems, and UV alone were 0.0150 ± 0.0013, 0.0034 ± 0.0007, 0.0015 ± 0.0003, and 0.0015 ± 0.0002 min⁻¹, respectively. The first-order reaction rate constant for the UV/C₂O₄²⁻/Fe³⁺ process was 3.4 and 5.2 times greater than that for the UV/ SiW₉/TiO₂/Cu and UV/TiO₂ systems, respectively (0.0044 and 0.0029 min⁻¹, respectively) [39]. At the same time, conversion of gaseous nitrogen was carried out in the following experiments. As illustrated in Figure 3, the removal of total nitrogen (TN) reached 84.2%, implying that a large amount of NO₃⁻ was selectively converted to gaseous nitrogen, thus showing its high application potential.

Notably, C₂O₄²⁻ addition probably led to subsequent pollution problems. Thus, we further investigated variations in TOC with reaction time to confirm the effect of C₂O₄²⁻ levels on this process. Results show TOC approaching zero with removal efficiencies of 100% after 120 min in Figure S1, indicating that the C-containing secondary contamination could be nearly completely removed by managing doses of oxalic acids. Taking together, UV/C₂O₄²⁻/Fe³⁺ systems could be an alternative technique for selective and efficient reduction of NO₃⁻ in water.

3.2. The Effect of Important Parameters

Figure 4a shows the effect of initial oxalic acid concentrations. Obviously, an initial dose of oxalic acids facilitated the reduction of NO₃⁻ in the UV/C₂O₄²⁻/Fe³⁺ systems. For example, there was a minor removal observed even for an initial C₂O₄²⁻ level of 1 mM after 180 min, which we attributed to UV irradiation, as evidenced by the results shown in Figure 2a. The efficiency of NO₃⁻ reduction was enhanced by increasing the initial H₂C₂O₄ concentration, and then declined after a further improvement in H₂C₂O₄ levels. The optimal reduction efficacy was obtained at 3 and 4 mM H₂C₂O₄ during the UV/C₂O₄²⁻/Fe³⁺ process, probably because an appropriate addition of H₂C₂O₄ led to the largely formed reductants and thus increased reduction. On the contrary, excessive H₂C₂O₄ doses resulted in a faster reducing radical-quenching reaction (Equations (1)–(3)) [41] and inhibitory NO₃⁻ reduction in water accordingly. These together suggest reducing CO₂^•− radicals probably facilitated NO₃⁻ reduction during the UV/C₂O₄²⁻/Fe³⁺ process. Furthermore, to confirm the effect of oxalic acid concentrations on NO₃⁻ reduction, we investigated the variation in its decomposition products over time in the UV/C₂O₄²⁻/Fe³⁺ systems. As shown in Figure S2, efficiencies of NO₃⁻ reduction reached 28.7%, 62.1%, 91.4%, 95.0%, 92.3%, and 71.3% with doses of 1, 2, 3, 4, 5, and 6 mM H₂C₂O₄, respectively; simultaneously, the conversion efficiency of gaseous nitrogen was 22.8%, 51.1%, 84.2%, 44.3%, 36.9%, and 24.0% for initial H₂C₂O₄ levels of 1, 2, 3, 4, 5, and 6 mM, respectively. Clearly, the maximum selectivity to gaseous nitrogen was achieved at a dose of 3 mM H₂C₂O₄, which was due to the greater formation of reducing radicals. The results, therefore, imply that reducing species predominated in the selective conversion reaction of gaseous nitrogen. A total of 3 mM oxalic acid was selected in the following experiments in terms of reduction efficiencies and gaseous nitrogen selectivity.

Figure 4b shows the effect of the initial iron dosage. Generally, the higher the initial iron dosage is, the faster the reductive reaction is. After dosing Fe³⁺, NO₃⁻ reduction was obviously improved during the process compared to that in the absence of Fe³⁺, as illustrated in Figure 4b. For instance, the efficiency of NO₃⁻ reduction was 42.0%, 84.5%, 91.4%, 91.4%, and 93.9%, in the absence and presence of 0.017, 0.025, 0.05, and 0.1 mM, respectively, with a respective gaseous nitrogen selectivity at 43.4%, 53.4%, 62.4%, 84.2%, and 74.9% (Figure S3). Further, degradation kinetics of NO₃⁻ at various Fe³⁺ doses could also be well-fitted using pseudo-first-order kinetics models. The results show that the reduction rate constants were 0.010, 0.013, 0.015, and 0.017 min⁻¹ at the initial Fe³⁺ levels of 0.017, 0.025, 0.05, and 0.1 mM, respectively, as shown in Table S2. These were highly consistent with the results of Equations (4)–(7), where higher Fe³⁺ doses led to a greater formation of the iron(III)–oxalate complex and thus increased production of reducing radicals such as C₂O₄^•− and CO₂^•−. Accordingly, NO₃⁻ reduction would be improved in the UV/C₂O₄²⁻/Fe³⁺ systems. It was interesting to find that as Fe³⁺ concentrations further increased to 0.1 mM, although NO₃⁻ reduction was slightly improved, the conversion of gaseous nitrogen was decreased. This might be because excessive reducing-radical (i.e., CO₂^•−)-induced reactions tended to proceed towards the NH₄⁺ product, as substantiated by the results shown in Figure S3. Therefore, 0.05 mM Fe³⁺ was utilized for NO₃⁻ reduction in UV/C₂O₄²⁻/Fe³⁺ systems.

$$F{e}^{3+}+O{H}^{-}\stackrel{hv}{\to }F{e}^{2+}+•OH$$

(1)

$$F{e}^{3+}+H_{2}O\to F{e}^{2+}+•OH+H^{+}$$

(2)

$$C_2{O}_4^{2-}+•OH\to C_2{O}_4^{•-}+O{H}^{-}$$

(3)

$$F{e}^{3+}+n{H}_2{C}_2{O}_4\to F{e}^{III}{\left(C_2{O}_4\right)}_n^{3-2n}$$

(4)

$$F{e}^{III}{\left(C_2{O}_4\right)}_2^{-}\stackrel{hv}{\to }F{e}^{II}\left(C_2{O}_4\right)+C_2{O}_4^{•-}$$

(5)

$$F{e}^{III}{\left(C_2{O}_4\right)}_3^{3-}\stackrel{hv}{\to }F{e}^{II}{\left(C_2{O}_4\right)}^{2-}+C_2{O}_4^{•-}$$

(6)

$$C_2{O}_4^{•-}\to C{O}_2+C{O}_2^{•-}$$

(7)

Figure 4c presents the effect of the initial solution pH. Increasing the initial pH contributed to a slower NO₃⁻ reduction. NO₃⁻ reduction reached 96.0%, 95.5%, 91.4%, 91.0%, and 87.0%, at pH_ini. 3, 5, 7, 9, and 11 (initial NO₃⁻ levels = 60 mg/L), respectively, in the UV/C₂O₄²⁻/Fe³⁺ process. The reduction kinetics of NO₃⁻ follows the pseudo-first-order kinetics. As presented in Figure S4 and Table S3, the reaction rate constants were determined to be 0.019, 0.018, 0.015, 0.014, and 0.012 min⁻¹ at initial solution pHs of 3, 5, 7, 9, and 11, respectively. According to previously reported literature, iron–oxalate complexes varied with the initial solution pH [42]. Figure 4d further presents the distribution of iron (III)-containing compounds as a function of the solution pH. At a solution pH below 2.9, Fe(C₂O₄)₂⁻ was the dominant species, and Fe(C₂O₄)₃³⁻ became the main complex at a pH ranging from 2.9 to 6.3. As the solution pH further increased to above 6.3, the fraction of Fe₂O₃ predominated. The mole fraction of Fe₂O₃ was nearly 100% at a pH greater than 7.0 (Figure 4d). As is known, Fe(C₂O₄)₂⁻ and Fe(C₂O₄)₃³⁻ can form CO₂^•−, as shown by Equations (5)–(7), while Fe₂O₃ cannot produce CO₂^•−. Therefore, the slower NO₃⁻ reduction kinetics at a higher pH was probably the consequence of the less formation of CO₂^•− resulting from the decreased fraction of Fe(C₂O₄)₂⁻ and Fe(C₂O₄)₃³⁻ and increased generation of Fe₂O₃.

Figure 4e further shows the effect of UV intensity. As expected, increases in irradiation intensities led to rapid production of reducing radicals, and thus the increased reduction of NO₃⁻. Specifically, the NO₃⁻ reduction efficiency was measured as 80.6%, 82.9%, and 91.4% for a UV power of 100, 300, and 500 W, respectively, which corresponded to 67.4%, 71.6%, and 84.2%, respectively, of gaseous nitrogen selectivity in Figure S5. The degradation kinetics of NO₃⁻ further follows the pseudo-first-order kinetics. As shown in Table S4, reaction rate constants were estimated to be 0.009, 0.010, and 0.015 min⁻¹, with 100, 300, and 500 W, respectively, of the UV lamp. Both CO₂^•− reduction and UV illumination itself contributed to NO₃⁻ reduction with a strong UV light from the medium-pressure UV lamp (UV-M), which is evidenced by the results, as seen in Figure 2. Accordingly, 500 W of UV-M was selected throughout the experiment.

Figure 4f presents the effect of initial NO₃⁻ levels. The higher the initial NO₃⁻ level is, the greater the reduction kinetics of NO₃⁻ is. At initial NO₃⁻ levels of 20, 40, 60, 80, and 100 mg/L, the NO₃⁻ removal efficiency was determined to be 92.5%, 91.8%, 91.4%, 80.9%, and 53.6%, respectively. Meanwhile, the conversion of gaseous nitrogen was further determined. From Figure S6, gaseous nitrogen selectivity was 28.5%, 44.4%, 84.2%, 70.6%, and 54.0% in the presence of 20, 40, 60, 80, and 100 mg/L NO₃⁻, respectively, after 180 min in the UV/C₂O₄²⁻/Fe³⁺ system. Interestingly, the conversion of gaseous nitrogen increased with initial NO₃⁻ concentrations and declined with increases in the initial level. The highest rate was achieved at an initial NO₃⁻ concentration of 60 mg/L. This was probably attributable to an insufficient iron(III)–oxalate complex and thus the limited formation of CO₂^•− at higher initial NO₃⁻ doses. Comparatively, although a faster reduction of NO₃⁻ was obtained at lower initial NO₃⁻ levels, there was still a lower conversion rate of gaseous nitrogen in the UV/C₂O₄²⁻/Fe³⁺ process. This is in accordance with the results seen in Figure S3 as the excessive CO₂^•−-induced NO₃⁻ reduction reaction favorably produced products other than gaseous nitrogen. Therefore, an appropriate initial NO₃⁻ concentration (i.e., 60 mg/L) was recommended for the UV/C₂O₄²⁻/Fe³⁺ systems.

Figure 4g shows the effect of dissolved oxygen (DO). Clearly, DO slightly affected the reduction of NO₃⁻. Specifically, the reduction efficiency of NO₃⁻ was measured as 92.4% and 91.4% at DO levels of 8.0 and 0.5 mg/L, respectively, in the process. Accordingly, a DO of 0.5–8.0 mg /L only had a slight impact. The results could imply that oxygen was not involved in the reduction reaction in the UV/C₂O₄²⁻/Fe³⁺ systems, and/or oxygen played a role in circular reactions in the process. According to Scheme 1, which will be discussed later, DO reacted with CO₂^•− to produce O₂^•−, which was shown to further oxide Fe²⁺ and form Fe³⁺, thus reparticipating in the formation of the iron (III)–oxalate complex and reducing radicals. Therefore, it has been confirmed that oxygen is involved in the UV/C₂O₄²⁻/Fe³⁺ systems in a circular reaction, indicating a great application potential of UV/C₂O₄²⁻/Fe³⁺ systems in real water.

3.3. The Photocatalytic Reduction Mechanisms

3.3.1. Involved Reducing Species

It has been reported that the carbon dioxide radical anion (CO₂^•−) is most likely responsible for the reduction reaction in this process [32]. To determine the role of CO₂^•−, methyl viologen (MV²⁺) was selected as the scavenging compound to perform its inhibitory effect on NO₃⁻ reduction. As known, MV²⁺ exhibited a good reactivity with CO₂^•− [43,44]. Figure 5a presents the efficiency of NO₃⁻ reduction in the presence of 0.01, 0.1, 0.2, 0.4, 5, and 10 mM MV²⁺, which reached 87.0%, 86.7%, 75.7%, 57.5%, 33.2%, and 24.8%, respectively, and decreased by 4.4%, 4.7%, 15.7%, 33.9%, 58.2%, and 66.6%, respectively, compared to when the MV²⁺ addition was not added (91.4%). The reduction reaction was significantly inhibited by a higher MV²⁺ concentration (i.e., ≥10 mM MV²⁺), as demonstrated by Equation (8) [44]. Thereby, CO₂^•− did play a major role in the reduction of NO₃⁻ in UV/C₂O₄²⁻/Fe³⁺ systems. Notably, 24.8% of NO₃⁻ reduction in 180 min at 10 mM MV²⁺ was primarily due to the UV-M illumination, consistent with the results of UV alone (removal rates of ~24.3%), as seen in Figure 2.

$$C{O}_2^{•-}+M{V}^{2+}\to M{V}^{•+}+C{O}_2$$

(8)

As stated before, in addition to CO₂^•−, iron (III)–oxalate complexes can produce Fe²⁺ with UV irradiation, which would effectively reduce Cr(VI) [32]. Accordingly, Fe²⁺ might also have an impact on NO₃⁻ reduction. Figure 5b presents Fe²⁺ levels at different reaction times. A sharp increase in Fe²⁺ concentrations was obtained for the first 10 min, and the generated Fe²⁺ remained at a relatively stable level with further improving reaction time. For instance, Fe²⁺ concentrations increased from 0 to 2.6 mg/L over a period of 0–10 min and reached 2.8 mg/L (0.05 mM) after a further increase in time to 60 min. Additionally, the initial Fe³⁺ dosage in the system was 2.8 mg/L (0.05 mM). Thus, iron (III)–oxalate complexes with UV illumination were completely converted into Fe²⁺ and CO₂^•− within the first 10 min. As reported, a negligible reduction of NO₃⁻ was achieved with Fe²⁺ alone [45], thereby excluding the role of Fe²⁺ alone in the system. Collectively, it was highly probable that CO₂^•− was the major reducing species in the UV/C₂O₄²⁻/Fe³⁺ process.

3.3.2. Formation of Products

To further investigate reduction pathways of NO₃⁻ in the UV/C₂O₄²⁻/Fe³⁺ system, the speciation of nitrogen (N) was characterized accordingly. Figure 3 presents the formation of primary products and total masses of nitrogen during NO₃⁻ reduction by the UV/C₂O₄²⁻/Fe³⁺ system. As can be seen in Figure 3, NO₃⁻ levels decreased significantly with increasing reaction time. Meanwhile, the level of NH₄⁺ increased over reaction time, and NO₂⁻ concentrations first increased over time but declined after a further rise in reaction time. It should be noted that the generated NO₂⁻ was nearly completely minimized (100%) after 180 min, which is primarily attributable to its reaction with reducing species, i.e., reactions of NO₂⁻ with the products NH₄⁺ and CO₂^•−. Additionally, the total N content (TN) decreased over time, primarily because gaseous nitrogen was formed during the reaction process, for which N₂ was considered to be the main product due to its high reduction capacity of CO₂^•− (E⁰(CO₂/CO₂^•−) = −1.9 V). Additionally, the TN reached 3.07 mg N/L within 180 min, which was in good accordance with the total amount of the product NH₄⁺, primarily, and the remaining NO₃⁻. These indicate that NH₄⁺ was the predominant NO₃⁻ reduction product in aqueous solutions. Furthermore, as described above, almost 100% of DOC removal could be achieved after 180 min reactions, as seen in Figure S1, so carbon dioxide (CO₂) was the only oxalic acid degradation product. Accordingly, NH₄⁺, N₂, CO₂, and Fe²⁺ were the primary reaction products in the UV/C₂O₄²⁻/Fe³⁺ systems.

3.3.3. Proposed Reduction Pathways

The transformation pathway for NO₃⁻ reduction was proposed in the UV/C₂O₄²⁻/Fe³⁺ systems in Scheme 1. In the UV/C₂O₄²⁻/Fe³⁺ systems, the irradiation of Fe³⁺ and C₂O₄²⁻ finally led to the production of oxalate radicals, C₂O₄^•− (Equations (1)–(3)). Additionally, C₂O₄²⁻ reacts with Fe³⁺, producing the iron(III)–oxalate complex, Fe^III(C₂O₄)_n³⁻²ⁿ, as shown in Equation (4), including two main species, Fe^III(C₂O₄)₂⁻ and Fe^III(C₂O₄)₃³⁻. Subsequently, under UV irradiation, these complexes decomposed and then generated C₂O₄^•−, as shown by Equations (5) and (6). Furthermore, C₂O₄^•− decomposed into CO₂ and CO₂^•− according to Equation (7) [46], which is in high accordance with the results shown in Figure 4b. In addition, •OH could also react with C₂O₄²⁻ to produce C₂O₄^•− and finally CO₂^•−, as shown in Equations (7) and (9) [46]. Of note, excessive CO₂^•− most likely recombined in pairs to generate C₂O₄²⁻, as shown in Scheme 1 [41].

$$C{O}_2^{•-}+C{O}_2^{•-}\to C_2{O}_4^{2-}$$

(9)

$$N{O}_3^{-}+6H^{+}+5C{O}_2^{•-}\to \frac{1}{2}{N}_2+3H_{2}O+5C{O}_2$$

(10)

$$N{O}_3^{-}+10H^{+}+8C{O}_2^{•-}\to N{H}_4^{+}+3H_{2}O+8C{O}_2$$

(11)

$$N{O}_2^{-}+4H^{+}+3C{O}_2^{•-}\to \frac{1}{2}{N}_2+2H_{2}O+3C{O}_2$$

(12)

Three major pathways for NO₃⁻ reduction are proposed. On the one hand, NO₃⁻ was photolyzed to NO₂^• and •OH, and the former (NO₂^•) was further converted to NO₂⁻ under photolysis and finally to N₂ after reaction with CO₂^•− (Equations (3), (6), (7), (9), and (10)). On the other hand, NO₃⁻ was directly reduced to NH₄⁺ or N₂ by CO₂^•− (Equations (11) and (12)) as shown in Scheme 1. Additionally, to confirm the reaction mechanisms of NO₃⁻, we further investigated the variation in solution pH over time in UV/C₂O₄²⁻/Fe³⁺ systems. The results show that the solution pH declined rapidly after the addition of oxalic acids, followed by a gradual increase in the pH over time. For instance, the solution pH was decreased from 7.0 to 2.4 after a dose of oxalic acid was added but increased from 2.4 to 5.7 with increasing time for a total reaction time of 180 min, as illustrated in Figure S7. The consumption of H⁺ was consistent with the proposed reduction pathway, as shown in Equations (10)–(12) of Scheme 1. Although CO₂^•− reacted with a small amount of dissolved oxygen in the solution, the generated O₂^•− further oxidized Fe²⁺ to Fe³⁺ (Scheme 1), thus substantiating a minor effect of DO, as shown in Figure 4g.

3.4. The Reduction of Nitrate from Stimulated Groundwater

3.4.1. The Effect of Water Background Compounds

Figure 6a–c show the effect of common anions, such as Cl⁻, SO₄²⁻, and HCO₃⁻, during the UV/C₂O₄²⁻/Fe³⁺ process. The presence of 0.1–10 mM Cl⁻ or SO₄²⁻ negligibly affected NO₃⁻ reduction after 180 min, demonstrating the effectiveness of UV/C₂O₄²⁻/Fe³⁺ systems in the Cl⁻- and SO₄²⁻-containing water matrices. In contrast, HCO₃⁻ exhibited an inhibitory effect on NO₃⁻ reduction. There was little impact on NO₃⁻ reduction at 0–1 mM HCO₃⁻, as illustrated in Figure 6c. However, as the HCO₃⁻ level increased to 10 mM, the reduction efficiency of NO₃⁻ was decreased by ~10%, which is consistent with previous studies by Gu et al. [31], because HCO₃⁻ can compete for CO₂^•− with NO₃⁻. On the other hand, Figure 6d shows the effect of humic acids (HA). The HA concentration inhibited NO₃⁻ reduction. With HA concentrations of 0, 10, 20, and 30 mg-C/L, the removal efficiency of NO₃⁻ was 91.4%, 78.3%, 76.5%, and 61.6%, respectively, after 180 min in the UV/C₂O₄²⁻/Fe³⁺ process. The inhibitory results are probably due to competitive adsorption of UV light by HA, as well as possible quenching reactions with CO₂^•−. In particular, the former prohibited both direct photolysis and radical-induced photocatalysis, and the latter inhibited photocatalysis. Therefore, UV/C₂O₄²⁻/Fe³⁺ systems could be inhibited in the presence of high HA and HCO₃⁻ concentrations.

3.4.2. The Reduction of Nitrate from Simulated Groundwater

To evaluate the effectiveness of UV/C₂O₄²⁻/Fe³⁺ systems for NO₃⁻ reduction in real water, NO₃⁻ reduction was further performed in simulated groundwater, the characteristics of which were based on inorganic and organic levels in groundwater in Northwest China in Table S1. Figure 7 further presents the reduction of NO₃⁻ by UV/C₂O₄²⁻/Fe³⁺ systems for an initial nitrate concentration of 120 mg NO₃⁻/L in simulated groundwater (SGW). There was a slower reduction of NO₃⁻ in SGW (removal rates = 72.8%) in comparison with deionized water (removal rates = 91.4%) after 180 min at a low DO of 0.5 mg/L (Figure 2 and Figure 3f). This was mainly due to the decreased information of CO₂^•− resulting from its reaction with inorganic anions (i.e., 122 mg/L HCO₃⁻) and strong competitive adsorption of UV light by organic matters, as illustrated in Figure 6. Despite this fact, the NO₃⁻ concentration decreased to 44.1, 32.6, and 31.0 mg NO₃⁻/L after a reaction time of 120, 180, and 240 min, respectively, meeting the drinking water standard (<10 mg/L N-NO₃⁻, that is, 44.3 mg NO₃⁻/L). In conclusion, the UV/C₂O₄²⁻/Fe³⁺ systems are a promising treatment technology as a novel photocatalytic process for efficient and selective decontamination of NO₃⁻ from complex water matrices.

4. Conclusions

The UV/C₂O₄²⁻/Fe³⁺ systems have been demonstrated to be an alternative technology for efficient and selective reduction of NO₃⁻ in water. This system reduced 91.4% of 60 mg/L NO₃⁻ with a gaseous nitrogen selectivity of 84.2% in 180 min, compared to UV alone, UV/C₂O₄²⁻ systems, and UV/Fe³⁺ processes. The quenching experiment demonstrated that CO₂^•− was primarily responsible for NO₃⁻ reduction. Further, the mass balance calculation of nitrogen, carbon, and iron showed that NH₄⁺, N₂, CO₂, and Fe²⁺ were the primary reaction products. Reduction pathways were proposed accordingly. Appropriate oxalic acids levels (i.e., 3 mM) and initial NO₃⁻ doses (i.e., 60 mg/L) facilitated NO₃⁻ reduction and gaseous nitrogen conversion. The higher the Fe³⁺ concentration and UV intensity are, the greater the reduction of NO₃⁻ is. A higher initial pH inhibited NO₃⁻ reduction due to the lower proportion of iron (III)–oxalate at higher pHs. Of note, 0.5–8.0 mg/L of dissolved oxygen did not have a significant inhibitory effect. Comparatively, the UV/C₂O₄²⁻/Fe³⁺ systems were not inhibited by 0.1–10 mM Cl⁻ or SO₄²⁻ or 0.1–1.0 mM HCO₃⁻ but was retarded by 10 mM HCO₃⁻. Additionally, 30 mg-C/L humic acid (HA) led to the decreased NO₃⁻ reduction, probably because of competitive UV adsorption by HA, and its possible reactions with CO₂^•−. As expected, NO₃⁻ reduction in simulated groundwater was not as fast as in deionized water, mainly due to the presence of competing anions (i.e., HCO₃⁻) and natural organic matters (i.e., HA) in complex water matrices. Moreover, considering its cost-effectiveness and the wide availability of reagents in nature (i.e., iron and oxalic acid), UV/C₂O₄²⁻/Fe³⁺ systems are particularly attractive in the selective and effective conversion of NO₃⁻ to N₂. Overall, the UV/C₂O₄²⁻/Fe³⁺ process is an alternative selective reduction technology in the field of environmental water treatment.