Treatment of the insensitive munitions compound, 3-nitro-1,2,4-triazol-5-one (NTO), in flow-through columns packed with zero-valent iron

The need for effective technologies to remediate the insensitive munitions compound 3-nitro-1,2,4-triazol-5-one (NTO) is emerging due to the increasing use by the US Army and environmental concerns about the toxicity and aqueous mobility of NTO. Reductive treatment is essential for the complete degradation of NTO to environmentally safe products. The objective of this study is to investigate the feasibility of applying zero-valent iron (ZVI) in a continuous-flow packed bed reactor as an effective NTO remediation technology. The ZVI-packed columns treated an acidic influent (pH 3.0) or a circumneutral influent (pH 6.0) for 6 months (ca. 11,000 pore volumes, PVs). Both columns effectively reduced NTO to the amine product, 3-amino-1,2,4-triazol-5-one (ATO). The column treating the pH-3.0 influent exhibited prolonged longevity in reducing NTO, treating 11-fold more PVs than the column treating pH-6.0 influent until the breakthrough point (defined as when 85% of NTO was removed). The exhausted columns (defined as when only 10% of NTO was removed) regained the NTO reducing capacity by reactivation using 1 M HCl, fully removing NTO. After the experiment, solid-phase analysis of the packed-bed material showed that ZVI was oxidized to iron (oxyhydr)oxide minerals such as magnetite, lepidocrocite, and goethite during NTO treatment. This is the first report on the reduction of NTO and the concomitant oxidation of ZVI in continuous-flow column experiments. The evidence indicates that treatment in a ZVI-packed bed reactor is an effective approach for the removal of NTO.


Introduction
Insensitive munitions compounds (IMCs) are increasingly being applied by the US Army to replace legacy munitions such as 2,4,6-trinitrotoluene (TNT) and 1,3,5-trinitro-1,3,5-triazine (Royal demolition explosive, RDX) and prevent unintentional explosions. The nitroaromatic compound, 3-nitro-1,2,4-triazol-5-one (NTO), is an important IMC that has a comparable explosive power to legacy explosive compounds while also being insensitive to heat and mechanical shock (Jun et al. 2010;Powell 2016). The ever-increasing use of NTO in IMC formulations has raised concerns about its potential environmental fate and impact. NTO residues released in firing ranges and munitions storage areas can be solubilized by rainfall and contaminate the surrounding environment due to its higher aqueous solubility (16640 mg L −1 at 25 ℃) and lower soil sorption coefficients (K OC = 4.0-61.5) compared to legacy explosive compounds Responsible Editor: Weiming Zhang (Mark et al. 2016;Taylor et al. 2015). The off-site migration of NTO may cause adverse impacts on aquatic and terrestrial organisms. For example, NTO is toxic to a marine bacterium, Allivibrio fischeri (20% inhibiting concentration or IC 20 = 19.2 mM) and methanogenic archaea (50% inhibiting concentration or IC 50 = 1 mM) (Madeira et al. 2018). In addition, NTO causes abnormalities on the swimming behavior of zebrafish embryos (Danio rerio) at low concentrations (7.5 µM). NTO can also exert toxicity on mammals. Acute oral exposure of rats to NTO (5000 mg kg −1 ) caused 50% lethality whereas oral exposure to this chemical (500 mg kg −1 ) for 4 weeks resulted in hypospermia (Johnson et al. 2017). Chronic oral exposure to NTO (> 250 mg kg −1 ) in male rats interfered with the development of a reproductive organ (Lent et al. 2015). Therefore, the high mobility and reported ecotoxicity of NTO suggests that the environmental release of this IMC has the potential to result in adverse effects on the ecosystem.
NTO can be degraded abiotically to environmentally benign products through sequential reduction and oxidation reactions (Khatiwada et al. 2018a;Madeira et al. 2017). Reduction is the first step in the NTO degradation process. The reduction of NTO can be achieved by reactive reduced iron minerals such as zero-valent iron (ZVI), sulfate green rust (Fe II 4 Fe III 2 (OH) 12 SO 4 •2H 2 O), and iron sulfide minerals (i.e., mackinawite and a mixture of pyrrhotite and troilite), which transform nitro groups in NTO into amino groups (Khatiwada et al. 2018b;Le Campion and Ouazzani 1999;Menezes et al. 2021). Reducing the nitro group to amino group in NTO forms 3-amino-1,2,4-triazol-5-one (ATO). ATO can be subsequently oxidized to environmentally benign products such as urea, carbon dioxide, and nitrogen gas by biotic and abiotic ways (Khatiwada et al. 2018b;Madeira et al. 2017). Therefore, the conversion of NTO to ATO is an essential step for the complete degradation of NTO. Previous studies indicate that reducing Fe-based chemicals can effectively degrade NTO. However, little is known about whether the reduction of NTO by reactive minerals is sustained over the long term and how the reduction of NTO influences the surface reactivity of the minerals.
ZVI is a promising material to apply in remediation technology targeting NTO in the long-term. Previous studies have proven that ZVI is a strong reducing material for organic contaminants such as trichloroethylene (TCE) and nitrobenzene (Xin et al. 2018;Yin et al. 2016). The proven effectiveness of ZVI led to its practical application as an effective and inexpensive medium for in situ treatment technologies for the remediation of hazardous contaminants, such as permeable reactive barriers (PRB) (Henderson and Demond 2007;Phillips et al. 2010). Our research group recently optimized the reduction of NTO using acid-pretreated ZVI in batch assays, providing complete conversion of NTO to ATO within 10 min at pH 3. However, accumulation of iron oxide minerals resulting from the oxidation of the ZVI surface has been reported to diminish the reactivity and hydraulic conductivity of ZVI in flow-through remediation technologies (Johnson et al. 2005;Moraci et al. 2016;Santisukkasaem and Das 2019;Xin et al. 2018). Therefore, it is essential to evaluate the longevity of ZVI in reducing NTO and investigate the oxidation of ZVI during the long-term treatment of water contaminated by NTO.
The objective of this study was, therefore, to assess the feasibility of applying ZVI in a continuous packed bed reactor (PBR) to treat NTO-contaminated water. To investigate the effect of pH on the longevity of the ZVI bed, we monitored the performance of two PBRs fed with a synthetic NTO influent with an initial pH of 3.0 or 6.0 for 6 months (i.e., 11,000 pore volumes). Additionally, we analyzed the solid phase transformation of ZVI and identified the main iron species formed in the continuous flow PBRs.

Batch experiments
Batch experiments to investigate the reduction of NTO by ZVI were performed under anoxic conditions at initial pH values ranging from 3.0 to 8.0. Acidic to circumneutral pH conditions were selected because previous studies have reported that the reducing capacity of ZVI with contaminants such as nitrobenzene decreases with increasing pH (Dong et al. 2010;Yin et al. 2012). The initial pH in the NTO assays was adjusted using diluted NaOH or HCl, as needed. Prior to mixing ZVI with the NTO solution, ZVI was pretreated to remove iron (oxyhydr)oxides on the surface. Briefly, 2.5 g of ZVI was incubated in 10 mL of 1 M HCl and washed with 10 mL of deionized (DI) water three times. All the solution used in the experiments were sparged with N 2(g) to remove the dissolved oxygen in the liquid. The washed ZVI was dried overnight using a lyophilizer (LAB-CONCO, Kansas City, Mo, USA). Pretreated ZVI (100 mg) was reacted with 10 mL of 0.5 mM NTO solution in 10-mL Luer-lock syringes at solid to solution ratio (SSR) of 10 g L −1 . Aliquots of the reaction medium (1 mL) were taken periodically over 1 h to monitor the reduction of NTO. The samples were filtered immediately through 0.22 µm membranes (FZFlow, Foxx Life Sciences, DE, USA) to stop the reaction and, subsequently, stored in the dark at 4 °C. All the batch experiments were conducted in triplicate in an anoxic vinyl glove box (COY Inc., Grass Lake, MI, USA).

ZVI-packed flow-through column experiment
Two sets of upflow columns packed with ZVI were prepared, and each was operated at a different initial pH (3.0 or 6.0). Figure 1 shows a schematic representation of the experimental setup. Each column (2-mL Supelco SPE tube, Supelco, Bellefonte, PA, USA) was packed with 3 g of ZVI pre-sieved using a 325-mesh for a uniform particle size. A frit was inserted below and above the ZVI bed to prevent the escape of ZVI particles. The packed tube was connected to a Minipuls 3 peristaltic pump (Gilson Inc., Middleton, WI, USA) using 0.25 mm peristaltic pump tubing (Thermo Fisher Scientific, Waltham, MA, USA). Additional details of the configuration and operation of the two columns are provided in Table 1.
Column experiments were conducted using an influent containing 1 mM NTO under two different influent pH conditions to investigate the effect of pH on NTO removal. An influent had a pH value of 3.0, i.e., natural pH of the NTO solution, whereas the other was adjusted to pH 6.0 with diluted NaOH. The influents were sparged with N 2(g) for 10 min to remove dissolved oxygen, and stored in closed containers to avoid exposure to the atmosphere and maintain anoxic conditions. The influent bottle was sealed with a rubber stopper and an aluminum cap and a 1/2-inch needle was inserted to connect the tubes in and out of the influent bottle. A N 2(g) filled syringe was attached to the influent bottle to avoid negative pressure due to the decrease in the liquid volume with time.
To remove the passivating layer on the ZVI surface, prior to the column start-up, both columns were pretreated with 1 M HCl for 30 min at 24 mL h −1 and then conditioned with deionized (DI) water for 1 h at 24 mL h −1 . After this pretreatment, the reactors were fed with the NTO influent at a flow rate of 2 mL h −1 , which is equivalent to a hydraulic retention time (HRT) of 1 h. The concentrations of NTO and ATO in the effluent were monitored to evaluate the NTO reducing capacity of the ZVI-packed bed reactors. In the current experiments, the breakthrough point is defined as the time when the NTO concentration in the effluent reached 15% of the initial NTO concentration (C/C 0 = 0.15).
Exhausted columns (C/C 0 ≥ 0.90) were regenerated by injecting 1 M of HCl for 100 pore volumes (PVs) at 24 mL h −1 , followed by DI water for 40 PVs at 24 mL h −1 . The wastewater generated from the reactivation process was collected to analyze the mass of dissolved iron released from the ZVI-packed column. The effectiveness of regeneration was assessed by restoring feeding of the NTO influent to the columns. At the end of the experiments, the reacted materials in the PBRs were analyzed to investigate the morphological change and the products of solid phase transformation.

Dissolved iron species (Fe II /Fe III )
Total dissolved iron was measured at 238 nm by inductively coupled plasma-optical emission spectrometry (ICO-OES, ThermoFisher, Waltham, MA, USA). Dissolved Fe II was analyzed by the ferrozine method (Stookey 1970). The filtered liquid samples (1 mL) were diluted in 4 mL of nitric acid (2% v/v) to stabilize Fe II and Fe III . The mass of Fe III was calculated by subtracting the mass of Fe II quantified by the ferrozine method from the mass of total dissolved iron measured by ICP-OES.

X-ray diffraction
Unreacted-and reacted ZVI from the column experiments were analyzed by X-ray diffraction (XRD) at the University of Arizona. The reacted samples were dried overnight in a lyophilizer (LABCONCO, Kansas City, MO, USA). Diffractograms were collected using Cu K-α radiation (λ = 1.5406 Å) on a Panalytical X'Pert PRO with ultra-fast X'Celerator detector (Malvern Panalytical, Malvern, UK) between 5° and 80° (2θ) at 0.02-degree 2θ steps. The generator voltage was 45 kV, and the tube current was 40 mA. Data analysis, peak identification, and assignment of crystalline phase were performed using the X'Pert HighScore Plus software (Malvern Panalytical, Malvern, UK) with ICCD PDF-2 diffraction reference files (Degen et al. 2014).

SEM analysis
Morphology changes on the surface of the reacted ZVI from the columns were examined by Hitachi S-4800 field emission scanning electron microscopy with integrated energydispersive X-ray spectroscopy (Fe-SEM-EDS, Chiyoda City, Tokyo, Japan). The accelerating voltage was 15 kV with 10 µA beam current.

Reductive transformation of NTO in batch experiments
Batch experiments were performed to evaluate the NTO reducing capacity of ZVI at initial pH conditions of 3.0 to 8.0 prior to flow-through column experiments (Fig. 2). NTO reduction by ZVI was favored at lower initial pH conditions. NTO was completely converted to its amine daughter product, ATO, within 30 min at an initial pH of 3.0. In contrast, incomplete NTO reduction was observed at higher pH conditions after the same time (Fig. 2). For example, only 15.1% of NTO was reduced after 30 min at pH 8.0 (Fig. 2). Peaks other than those corresponding to NTO and ATO were not detected in any of the HPLC chromatograms collected in the batch experiments. The pseudo first-order rate constants (k 1 ) of NTO reduction by ZVI were calculated based on the first five data points (Fig. 3). The results indicated that NTO reduction was fastest at pH 3.0, followed by pH 4.75, 6.0, and 8.0. The value of k 1 determined in assays with an initial of pH 3.0 was 2.7-fold, 4.2-fold, and 20-fold higher compared to the k 1 values determined in tests with less acidic pH conditions, i.e., 4.75, 6.0, and 8.0, respectively. After 1 h, the average recovery in moles of ATO plus unreacted NTO was 93.6% in the batch experiments at pH ranging 3.0 to 8.0 (Table S1 in Supplementary Information). A high recovery indicates that the NTO removal resulted from transformation rather than adsorption. Figure 4 presents the normalized concentrations of NTO and ATO in the effluent of the ZVI-packed bed reactors treating pH-3.0 and pH-6.0 NTO influent (expressed as a fraction of the molar concentration of NTO in the influent, C/C 0 ). The acidic influent was more beneficial than circumneutral influent to treat the NTO-contaminated wastewater over the long term. The column treating the pH-6.0 influent encountered earlier breakthrough (C/ C 0 ≥ 0.15) compared to the column treating the pH-3.0 influent (Fig. 4). The effluent concentration of NTO in the circumneutral influent started to increase rapidly after 250 PVs. In contrast, breakthrough was observed after 2930 PVs in the PBR treating the acidic influent and, as a result, it converted 16-fold more NTO than the reactor treating the pH-6.0 influent until the breakthrough point.

Performance of the ZVI-packed flow-through columns
In addition, ZVI exhaustion (C/C 0 ≥ 0.9) took fivefold longer in the reactor with the pH-3.0 influent than for the reactor with the pH-6.0 influent (Table 2). Thus, feeding the acidic influent helped extend the longevity of the ZVIpacked column, enabling extended treatment of the NTOcontaminated water. The columns eventually exhausted their NTO reducing capacity (C/C 0 ≥ 0.9) at ca. 9200 PVs for the pH-3.0 influent and at ca. 2100 PVs for the pH-6.0 influent (Fig. 4). The exhausted columns regained their NTO reducing capacity after reactivation with 1 M HCl for 100 PVs, and they were able to convert NTO fully to ATO (Fig. 4). However, the column treating pH-6.0 influent was exhausted again at ca. 6020 PVs. The second reactivation of the column treating circumneutral influent at ca. 7380 PVs also showed a similar exhaustion pattern.
Moderate decrease of hydraulic conductivity was observed from the reactor treating pH-3.0. The effluent flow rate, indicative of the hydraulic conductivity, decreased by 19% loss after ca. 9360 PVs (150 days) and 29% loss after ca. 10900 PVs (180 days) in the reactor treating pH-3.0 ( Table 2). The average hydraulic conductivity of the column fed with pH-6.0 was relatively stable during the duration of the experiment ( Table 2). The greater oxidation of ZVI in the column treating acidic influent likely contributed  Table 2).

Iron released from the columns
ZVI was consumed and released as Fe II and Fe III during the PBRs operation and reactivation (Fig. 5, Table 2). The actual amount of iron released was greater than the estimated consumption of ZVI by NTO reduction ( Table 2). The total iron loss during the column experiments was 41.0% of the initial ZVI loading in the column treating pH-3.0 influent and 29.6% in the column treating pH-6.0 influent (Fig. 5). Most of dissolved iron was released by dissolution of the oxidized iron deposited on the ZVI surface during the reactivation with 1 M HCl for 100 PVs. The percentages of dissolved iron released during the reactivation from the PBR treating pH-3.0 and pH-6.0 influent were 37.5% and 29.4%, respectively (Fig. 5). However, ZVI consumption by NTO reduction was calculated to be 25.2% with the pH-3.0 influent and 13.6% with the pH-6.0 influent (Table 2), which was about half of the total iron loss. In terms of a balance of electrons, it can be estimated that 69.1% and 52.7% of the electron equivalents released from the oxidation of ZVI were consumed by the reduction of NTO in the PRBs treating pH-3.0 and pH-6.0 influent, respectively (Table S2). The anoxic oxidation of ZVI is likely to account for the gap in the electron equivalent balance. ZVI can react with water under anoxic conditions causing the reduction of water to H 2 (g) (Dong et al. 2010), as follows: Increased pH in the effluent from the PBRs (Fig. S1) resulted from the iron corrosion by anoxic water. The effluent pH increased up to 8.2 with pH-3.0 influent and 9.1 with pH-6.0 influent since the anoxic corrosion of water releases OH − [Eq. 1]. Figure 6 shows the morphological changes in the ZVI particles obtained from the PBRs at the beginning and end of the experiment. A smooth surface of the unreacted ZVI (Fig. 6a) turned into a roughened surface by the HCl pretreatment (Fig. 6b). The SEM micrographs of the reacted ZVI (Fig. 6  c and d) show neoprecipitates with acicular and prismatic habit and partial aggregation, attributed to the formation of the ferric oxides (α-and γ-FeOOH) on the surface.

ZVI oxidation during the column operation
The minerals formed by oxidation of ZVI during the column operation were identified by XRD analysis. After operation and reactivation of the ZVI-packed bed reactors, about half of the remaining iron in the column after the operation and reactivation was transformed to iron (oxyhydr)oxide mineral species: magnetite (Fe 3 O 4 ), lepidocrocite (γ-FeOOH), and goethite (α -FeOOH) (Fig. 7, Table 3). These results indicated that ZVI was oxidized to iron (oxyhydr)oxides by the continuous NTO reduction under anoxic conditions.

Reductive transformation of NTO by ZVI
ZVI reductively transformed NTO into its amine daughter product, ATO, indicating that the mechanism of NTO removal by ZVI involves reduction of nitro groups. It is well established that nitro-containing compounds such as nitroguanidine and nitrobenzene are reductively transformed by ZVI into their amine daughter product, with occasional detection of nitroso or hydroxylamine intermittent products (Agrawal and Tratnyek 1995;Rios-Valenciana et al. 2022). In our experiments, nitroso or hydroxylamine products were not detected during the reductive transformation of (1) NTO by ZVI. Formation of an hydroxylamine intermediate (3-hydroxylamino-1,2,4-triazol-5-one) was previously observed by our research group during the anaerobic biotransformation of NTO due to the slow rate of the reaction (Krzmarzick et al. 2015).

Effect of initial pH influent on the longevity of ZVI-packed column
This study investigated the effect of initial medium pH on NTO reduction in batch experiments and continuous column experiments. Batch experiments confirmed that NTO reduction by ZVI was faster and more prolonged in assays with acidic pH compared to circumneutral to moderately alkaline pH (Figs. 2 and 3). The results are consistent with the findings from previous studies reporting higher reduction rates for other nitro-aromatic compounds such as atrazine, nitrobenzene, and TNT with ZVI at acidic pH conditions than weak acidic to alkaline conditions (Dombek et al. 2001;Dong et al. 2010;Yin et al. 2012;Zhang et al. 2010). We propose that the faster reaction of ZVI is due to the acidic conditions preventing the precipitation of iron corrosion products (Dong et al. 2010;Yin et al. 2012). High pH promotes the formation of a passive layer consisting of iron corrosion products on the ZVI surface, with the associated decrease in the rate of electron transfer (Song and Carraway 2005). Therefore, low pH conditions seem to allow a better electron transfer during NTO reduction by ZVI by eschewing the formation of a thick passivating iron (oxyhydr)oxide layer.

Loss of NTO reducing capacity by iron corrosion products
Reducing the formation of an iron (oxyhydr)oxide passivation layer on the ZVI surface may be the key factor to extend Table 2 Average values of operational parameters in columns fed with acidic influent and circumneutral influent † The effluent pH during the column experiments is shown in Fig. S1 † † We assume that Fe 0 is oxidized to Fe II by NTO reduction. Thus, 1 mol of NTO is reduced by 3 mol of Fe 0 † † † Released iron in the effluent during the operation is expressed as total iron due to the possible conversion of Fe II to Fe III in basic conditions

Fig. 5
Iron mass percentage of remaining iron and dissolved iron released during NTO reduction operation and reactivation process in flow-through columns packed with ZVI with acidic NTO influent (pH 3.0) and circumneutral NTO influent (pH 6.0). Iron mass values are provided in Table 2 the service life of ZVI-packed columns treating NTO. In the PBRs operated in this research, ZVI was oxidized to iron (oxyhydr)oxide minerals such as magnetite (Fe 3 O 4 ), lepidocrocite (γ-FeOOH), and goethite (α -FeOOH) by the continuous NTO reduction and anoxic iron corrosion, which were unable to reduce NTO. We verified in a batch experiment that synthesized magnetite does not reduce NTO and showed no adsorption after 90 min (Fig. S2). In addition, iron (oxyhydr)oxides such as goethite and ferrihydrites (Fe III 2 O 3 •0.5H 2 O) did not reduce NTO, but could adsorb some NTO, and to a lesser extend ATO, at positively charged surface sites at pH 7.0 (Khatiwada et al. 2018a;Linker et al. Fig. 6 SEM images of unreacted ZVI (a), ZVI pretreated with 1 M HCl (b), reacted ZVI collected from the NTO reducing column treating pH-3.0 influent for ca. 12,500 PVs (c), and reacted ZVI collected from the NTO reducing column treating pH-6.0 influent for ca. 11,000 PVs (d) Fig. 7 Normalized diffractograms of unreacted ZVI (a), reacted ZVI collected from the NTO reducing column treating pH-3.0 influent for ca. 12,500 PVs (b), reacted ZVI collected from the NTO reducing column treating pH-6.0 influent for ca. 11,000 PVs (c), and reference minerals (panel d-g) 2015). However, significant adsorption at ferric surface sites in the passivation layer was not observed during our column experiments as confirmed by the high recovery of NTO as ATO plus unreacted NTO (92.0-103.4%, Table 2).
The decreased NTO reducing capacity of the columns was effectively recovered by dissolving the iron corrosion products using 1 M HCl for 100 PVs. Dissolution of Fe III during the regeneration process (Fig. 5, Table 2) provided evidence of iron (oxyhydr)oxide precipitates on the surface of ZVI in the columns. Other studies have also confirmed that HCl was effective in dissolving the iron corrosion products formed on the surface of ZVI particles (Cornell et al. 1976;Hemmelmann et al. 2013;Sultana et al. 2014). These results confirm that regeneration with HCl is a simple and effective approach to regenerate ZVI and extend the service life of PBRs treating NTO.

Decreased hydraulic conductivity by iron corrosion products
The current study observed a moderate decrease in the hydraulic conductivity of the column fed with the pH-3.0 influent after extended operation (29% loss after 10,900 PVs or 180 days) ( Table 2). Previous studies have reported that iron corrosion products caused a decrease of the hydraulic conductivity of permeable reactive barriers (PRB) packed with ZVI (Henderson and Demond 2007;Phillips et al. 2010;Santisukkasaem and Das 2019;Xin et al. 2018;Yang et al. 2021). Corrosion of ZVI to maghemite and magnetite led to a 50% decrease in hydraulic conductivity of a ZVI column fed with tap water for 1 month (Santisukkasaem and Das 2019). In a ZVI-filled PRB operated for 10 years, iron oxide minerals, mainly magnetite, formed a concreted zone restricting the hydraulic conductivity (Phillips et al. 2010). As a result of oxidative transformation, the pores of the ZVI bed can become restricted by the increased tortuosity and volume of iron corrosion products (Moraci et al. 2016;Yang et al. 2021). The expected volumetric expansion for magnetite, goethite, and ferrihydrite was calculated to 2 to 4.5 based on Faraday's law (Yang et al. 2021). The reduced pore volume resulting from the volumetric expansion of iron corrosion products could decrease hydraulic conductivity over time and require replacement of bed materials.

Comprehensive evaluation of ZVI-packed bed reactor
ZVI-packed bed reactors were shown to be a promising technology to remediate NTO-contaminated water as an effective and prolonged treatment system in this study. The performance of the laboratory-scale ZVI reactors was tested under accelerated conditions with a considerably shorter HRT (1 h) than the typical residence time of ZVI-based PRB (i.e., 216 h (9 days) by Wilkin and Puls (2004); 394-535 h (16.4-22.3 days) by Bartlett and Morrison (2009)). Despite the short HRT, the ZVI column treating pH-3.0 influent displayed a long service life, transforming effectively NTO to ATO for 2930 PVs (C/C 0 = 0.15) (Fig. 4). Although the ZVI column operated with pH-6.0 influent had a relatively short service life (ca. 250 PVs until C/C 0 = 0.15), a much longer service time would be expected in practice due to the longer residence time typically applied in groundwater remediation systems. In addition to this, water containing NTO could be naturally acidic because of the low pKa of the chemical (3.76) (Taylor et al. 2015).

Conclusions
ZVI-packed bed reactors can effectively reduce NTO over long-term periods of operation to its amine daughter product, ATO, under acidic to circumneutral pH conditions. ZVI is concomitantly oxidized to iron (oxyhydr)oxide minerals such as magnetite, lepidocrocite, and goethite by continuous NTO reduction under anoxic conditions. Precipitation of iron (oxyhydr)oxide minerals on the ZVI surface can hinder NTO reduction in the ZVI-packed remediation technology; therefore, acidic conditions are preferred to alleviate iron oxide precipitation over circumneutral or alkaline conditions. In addition, the acid treatment can regenerate