Fenton pre-oxidation of natural organic matter in drinking water treatment through the application of iron nails

ABSTRACT This study investigated for the first time the efficiency of an advanced oxidation process (AOP) zero valent iron/hydrogen peroxide (ZVI/H2O2) employing iron nails for the removal of Natural Organic Matter (NOM) from natural water of Regent’s Park lake, London, UK. The low cost of nails and their easy separation from the water after the treatment make this AOP attractive for water utilities in low- and middle-income countries. The process was investigated as a pre-oxidation step for drinking water treatment. Results showed that UV254 removal in the natural water was lower than that of simulated water containing commercial humic acid (HA), indicating a matrix effect. Statistical analysis confirmed the maximum removal of dissolved organic carbon (DOC) in natural water depends on the initial pH (best at 4.5) and H2O2 dosage (best at 100% excess of stoichiometric dosage). DOC and UV254 removals under this operational condition were 51% and 89%, respectively. Molecular weight (MW) and specific UV absorbance (SUVA254) were significantly reduced to 74% and 78%, respectively. Formation of Chloroform THM in natural water sample after the ZVI/H2O2 process (initial pH 4.5) was below the limit for drinking water, and 48% less than the THM formation in the same water not subjected to pre-oxidation. Characterization of oxidation products on the iron-nail-ZVI surface after the ZVI/H2O2 treatment by SEM, XRD, and XPS identified the formation of magnetite and lepidocrocite. Results suggest that the investigated ZVI/H2O2 process is a promising technology for removing NOM and reducing THM formation during drinking water treatment.


Introduction
Surface waters in various regions contain natural organic matter (NOM) with a significant percentage of humic acids (HA) [1]. These are known precursors in the formation of disinfection by products (DBPs), in particular toxic chlorinated organics produced by the reaction with chlorine during drinking water treatment [2,3].
Amongst the halogenated DBPs, trihalomethanes (THMs) are the ones found in higher concentrations in drinking water [4,5]. These are associated with increased risk of cancer, especially in the bladder, and adverse reproductive outcomes [6]. Making the situation worse in WTPs, the application of pre-chlorination for taste/ odour or iron/manganese control prior to coagulation can increase the generation of a variety of DBPs [7]. In addition, in conventional water treatment trains, the steps of coagulation and flocculation followed by sedimentation or flotation, and filtration are only partially effective at removing NOM [8,9].
AOPs are the WTP technologies that have attracted most interest in the last two decades because of their potential to reduce the formation of DBPs [3,16]. AOPs are generally applied for commercial drinking water under low to moderate oxidation conditions. Even in these operating conditions, NOM with higher molecular weight compounds is partially oxidized and transformed into smaller ones, such as aldehydes and carboxylic acids, which are more biodegradable molecules [17,18]. These chemical modifications can result in the overall decrease of Total Organic Carbon (TOC) and Disinfection Byproduct Formation Potentials (DBPFP) in the treated water [17,19]. And among the AOPs, the Fenton process has already been proposed to optimize the removal of several organic contaminants from waters [8,[20][21][22][23]. This process utilizes hydrogen peroxide (H 2 O 2 ) and soluble iron (II) salts to produce a highly reactive radical (HO•) capable of oxidizing organic compounds in solution. Previous Fenton studies (photo-Fenton and Electro-Fenton) were reported to potentially remove organics with a wide molecular weight range from the humic acid structure, achieving the fragmentation of hydrophobic fractions into hydrophilic fractions [24,25]. In recent years, alternatives to soluble iron salts in the Fenton process have been investigated, mainly aimed at the use of solid elemental iron (Fe 0 /ZVI) as this is an efficient electron donor [26,27]. However, the disadvantages of removing metalic iron in the form of particulate powder during and at the end of the treatment have driven interest to a search for other forms of ZVI [27][28][29][30][31][32][33][34][35][36] which may allow easier separation from the water. Among these studies, similar rates of TOC removal have been reported when using ZVI in less conventional forms (such as metal shavings) compared to commercial particulate ZVI powder [36].
This strategy of employing low-cost commercially available ZVI materials can be attractive for the application of the Fenton (ZVI/H 2 O 2 ) process [27], mainly for municipalities that may lack financial resources to afford advanced water treatment processes with complex equipment such as ozone generators or UV photo-reactors. For example, the cost of ZVI powder with 97%-99% of purity range from 60 to 150.0 $ USD/ Kg [37], while the cost of commercial iron-nails range from 1.0 to 3.0 $ USD/Kg. Considering this, some researchers have applied commercial iron-nails in different water/ wastewater treatments e.g. in sand filters during the filtration process for arsenic removal [38][39][40][41], in Electrocoagulation process for removal of mercury (II) [42], and in ultrasound-assisted Fenton process for wastewater treatment [43]. However, the application of commercial iron-nails for NOM removal in the drinking water treatment plants has not been reported. Therefore, the main purpose of the present work was to investigate the removal of NOM in natural surface water using commercial iron-nails in the ZVI/H 2 O 2 process as a pre-oxidation step to conventional WTPs. To the authors' knowledge, this is the first study applying ZVI in the form of ironnails to the ZVI/H 2 O 2 process to remove NOM and reduce THM formation during drinking water treatment.

Materials and chemicals
Commercial HA (Sigma-Aldrich), hydrogen peroxide (50% w/w) (Solvay Group), THM standards (Supelco; purity of 99.9%), sodium polystyrene sulfonates (PSS, Polymer Standards Service GmbH, Germany), methyl-tertiary-butyl-ether (MtBE) (Sigma-Aldrich), sodium sulfite (Merck;purity of 99%), and all other chemical reagents were of analytical grade and the solutions were prepared with ultrapure water (Millipore Milli-Q, resistivity > 18.2 MΩ.cm). The ZVI nail was obtained in the form of small low-carbon-steel nails (AISI 1010) with 0.43 cm² g −1 (surface area per gram of each nail). The chemical composition (mass concentration -%) performed by optical emission spectroscopy -OES (Spectromax Instrument) confirmed that the total trace elements content in the nail is less than 1%, as shown in Table A Natural water samples were collected from the Regent's Park lake (London, UK -51°31'28" N 0°09'15" W) (Figure A.1) in 1-L bottles for each sample, during August and September 2018.

Preparation of samples
Simulated water was prepared by dissolving 0.01 g of commercial HA in 15 mL of 0.1 M NaOH solution [20] and was homogenized on a shaker (175 rpm) for ± 14 h to fully dissolve. Hydrogen peroxide was used to prepare a 1.0 g L −1 stock solution.
The stoichiometric H 2 O 2 dosage applied to the simulated water containing HA and natural water was 8.6 mg L −1 per mg L −1 of dissolved organic carbon (DOC) [44,45]. Stoichiometric dosages of sodium sulfite [46] were added to the samples at the end of ZVI/H 2 O 2 experiments to extinguish the remaining concentrations of H 2 O 2 and to prevent the influence of sodium sulfite excesses on the chlorine extinction at the chlorination step. The pH of the samples subjected to the ZVI/H 2 O 2 process was adjusted to pH 7.0 at the end of each run, before the chlorination step.
Preliminary chlorine demand tests on water samples collected from Regent's Park indicated that a chlorine dosage of 4.0 mg L −1 was required for providing a residual concentration of free chlorine of 0.5 ± 0.15 mg L −1 [47]. After the ZVI/H 2 O 2 process, samples were subjected to 30 min and 24 h of contact time (Ct) with chlorine (4.0 mg L −1 ) [48,49]. The sodium sulfite was added at the end of each contact time to extinguish the remaining concentrations of chlorine. After the chlorination step, the samples were stored according to USEPA Method 551.1 [50] for later analysis of THM formation. THM standards were prepared by dissolving in methanol, at 2000 μg mL −1 each.

Analytical procedures
The UV 254 (Shimadzu UV 1800 spectrophotometer), DOC (TOC-L CPH model TOC analyser) and specific UV absorbance [SUVA (L mg −1 m −1 ) = UV 254 /DOC * 100] determinations were obtained according to USEPA Method 415.3 [51] and Standard Methods 5910B [52]. For molecular weight (MW) analysis, a calibration was performed using a set of four sodium polystyrene sulfonates (PSS, Polymer Standards Service GmbH, Germany) together with Acetone (R² = 0.998). Of the various standards available for MW measurement, PSS have been widely applied [53][54][55][56][57][58][59][60] as they are more similar to NOM with regards to charge density [53,54]. After the calibration step, a semi-logarithmic linear calibration curve was obtained and used to determine the MW. These calibration curves were constructed by determining the peak retention times of the patterns with narrow MW distribution, following methodologies and equations by Liu and Fitzpatrick [53] and Zhou et al. [54]. The MW values of NOM before and after the ZVI/ H 2 O 2 reaction were determined by high pressure size exclusion chromatography (HPSEC) on the HPLC system (Perkin Elmer, UK) equipped with a 200 series pump, autosampler and 254 nm UV detector, following Liu and Fitzpatrick's methodology [53]. The analyses of total THM were performed by gas chromatography (GC) (Claurus 500 model) with column (Restek Rxi-5ms, 30 m × 0.25 mm ID) coupled with mass spectrometry (MS) by liquid/liquid extraction with methyl-tertiarybutyl-ether (MtBE), following USEPA Method 551.1 [50] and Rasheed et al. [61]. GC analytical conditions are given in Table A.2. The H 2 O 2 concentration was monitored according to Solvay-Peroxidos Brasil's method [62] on a HACH 890 colorimeter at 420 nm (detection interval 0.1-10.0 mg L −1 ). The determination is based on the reaction between H 2 O 2 and a titanium (IV) salt in an aqueous acidic solution to produce a yellow complex of perititanic acid.
The surface morphology of the ZVI nail was analysed using scanning electron microscope (SEM) (FEG, 'Field Emission Gun') (Zeiss Merlin, and Zeiss Ultra 55) in both secondary electrons (SE) and backscattered electrons (BSE) imaging modes. The operating voltage for imaging ranged from 1.5 to 5 kV. X-ray energy dispersive spectroscopy (XEDS) provided qualitative compositional information about the surface deposits on the ZVI nail. XEDS elemental maps were acquired using the Zeiss Merlin FEG-SEM equipped with an Oxford Instruments low voltage X-Max Extreme 100 windowless silicon drift detector (SDD). All XEDS data were collected with incident electron beam energy in the range of 1.5-3 kV, which significantly reduces the volume of the specimen that generates the XED spectra; thus, it was possible to obtain qualitative compositional data from a surface/near surface region of ∼ 20-50 nm in depth. The oxides species on the ZVI nail surface after ZVI/ H 2 O 2 reaction were identified by X-ray diffraction (XRD) using Proto AXRD system diffractometer θ-2θ diffractometer (284 mm diameter) and X-ray photoelectron spectroscopy (XPS) using spectrometer Thermo K-alpha (E = 1486.6 eV). The dissolved Fe 2+ formed during the experiments was measured at the time of aliquoting after filtering the samples in a 0.45 μm membrane filter. The analyses were performed on a HACH 890 colorimeter at 520 nm (10-phenanthroline method adapted from Standard Methods for the Examination of Water and Wastewater) [63][64][65].

Experimental procedure
Prior to the start of the experiments, the ZVI nail was subjected to a chemical cleaning/pickling step (by immersion in 1% HCl solution for 10 min) and was washed three times with distilled water after the pickling to remove any surface dirt and remaining acid. ZVI/H 2 O 2 runs were conducted following a statistical factorial design -2³. All experiments were performed in 1-L beakers under agitation at 250 rpm in Jar-test equipment (temperature 23 ± 1°C). The pH of the solution was initially adjusted to the set initial value with H 2 SO 4 (0.01 M) and the required H 2 O 2 dosage was injected into the solution. The pH stability was verified for the possible need for re-adjustment, and then the ZVI was added triggering the start of the reaction. The aliquots were previously filtered through a 0.45 μm membrane filter (Millipore) and analyses of Fe 2+ , residual H 2 O 2 , and pH were performed immediately. In the respective times of taking the aliquots, the oxidation reaction by H 2 O 2 was extinguished with sodium sulfite (except for H 2 O 2 measuring). The remaining samples were stored under refrigeration (4°C ) for further analysis.

Statistical analysis
Statistical analysis was performed by Statistica 8.0, Stat-Soft (Inc., USA) software to obtain the best operational conditions for the oxidation of NOM. A factorial design in two levels and three variables (2³) was used to evaluate the effect of the independent variables (initial pH, H 2 O 2 dosage, and ZVI nail amount) and their levels on the ZVI/H 2 O 2 process (Table A.3). Among such experiments, three were performed at the central point conditions for the estimation of the experimental error. ANOVA test was carried out to assess the difference in significance between data, and a p-value < 0.05 was considered statistically significant.

Characteristics of the simulated water containing HA and natural water
Although commercial HA has been used as the research matrix in some studies [66,67], it is important to consider the composition differences between commercial HA solutions and natural waters. Such differences result in variation of the efficiency of chemical water treatments applied in these diverse matrices [68]. In this way, the performance of ZVI/H 2 O 2 was evaluated to treat simulated water containing commercial HA and real natural water samples collected from the Regent's Park lake. Table 1 shows the characterization of the simulated water ([HA] = 5 mg L −1 ) and surface water from Regent's Park lake, London. SUVA 254 has been used as one of the parameters of better correlation with the organic matter aromaticity and aquatic humic composition [5,66]. The high SUVA 254 value obtained from the simulated water containing HA (6.2 L mg −1 m −1 ) indicates the presence of higher aromatic carbon content and hydrophobic compounds than in the natural water (2.21 L mg −1 m −1 ). DOC concentration was significantly higher in the natural water, highlighting the variety of possible organic substances present in natural waters, not necessarily with higher hydrophobic compounds.
Consequently, differences in the efficiency of the ZVI/ H 2 O 2 process were observed. Greater UV 254 removal in the simulated water containing HA (93.3% ± 0.95) compared to that in natural water (69.9% ± 2.94) can be seen in Figure (Table  1), the higher UV 254 removal obtained in this sample occurred probably due to the larger number of active sites in its aromatic structure available for reacting with the HO• [21,[69][70][71]. In addition, carbonate and bicarbonate ions are typically present in natural water [44] (From Table 1 conductivity of natural and simulated water are = 1007 ± 39.6 and 37.6 ± 0.1 μs.m −1 , respectively). Some HO• radical reacts with carbonate/bicarbonate ions, therefore, the HA removal in natural water may be lower than that in simulated water.
During the first 15 min of ZVI/H 2 O 2 reaction in the simulated water containing HA, no changes in UV 254 were observed, as shown in Figure A.2. This may have been due to the interference of iron ions dissolved in solution, which can absorb ultraviolet light at 253.7 nm [52].

Effect of initial pH
The influence of initial pH on NOM degradation by the ZVI/H 2 O 2 processes applied to Regent's Park water was investigated over the pH range of 4.5-6.5. Figure 1(a) confirms the reduction of DOC when the initial pH was elevated, with 51%, 30% and 8% of removal, corresponding to initial pH 4.5, 5.5 and 6.5, respectively. The same trend was observed for UV 254 removal ( Figure A.3). As shown in Figure 1(b), the ZVI/H 2 O 2 applied in a more acidic pH range allowed greater leaching of ferrous ions from the nails. This increased leaching promotes the formation of radicals with sufficient oxidative potential to degrade the organic contaminants [72]. However, when the aqueous solution reaches neutral pH values, the oxidation rate of the process is delayed by the formation of solid iron oxyhydroxides (Equations (1)-(4)), which are weaker activating agents than Fe 2+ (aq), resulting in decreased rate of production of HO• radicals [22,26,56]. The initial raise followed by the decay of Fe 2+ concentrations in aqueous medium after 5 min of treatment (at pH 4.5 and 5.5) shown in Figure 1(b) may indicate that there is a fast initial step of Fe dissolution caused by H 2 O 2 oxidation according to Equation (1). This would be followed by its (Fe 2+ ) consumption in reaction with H 2 O 2 to form radicals with oxidative potential (Equation (2)). In addition to this reaction step, the reduction in the concentration of Fe 2+ in the solution can occur due to the formation of solid iron oxyhydroxides (Equation (4)). Previous studies using micro powder ZVI in the ZVI/H 2 O 2 process reported attained concentrations of Fe 2+ (aq) similar to those obtained in this study [32,73]. As the AOP investigated is a step prior to conventional water treatment steps, the remaining iron ions from ZVI/ H 2 O 2 process can be used as part of the coagulant in the following step of drinking water treatment (coagulation/ flocculation). In the present work, it was not possible to measure the total iron concentration and compare the results with the standards for drinking water. However, the total concentration of iron in the final treated water should not be higher than that required by legislation in the final treated water (maximum limit allowed for total [Fe] by Brazilian standard: 0.3 mg L −1 [47] and British standard: 0.2 mg L −1 [74]). Figure 2 shows that the pH of the natural water and simulated water containing HA increased during the reaction. This may indicate the kinetic prevalence of H + consuming reaction (Equation (1)) together with OHgeneration reaction (Equation (2)) which is known to be much faster than Equation (3). Thus, in the course of the overall reaction, the OHconsumed in Equation (4) is not enough to compensate and neutralize the H + consumption and OHgeneration by the sum of reaction steps (Equation (1)) and (Equation (2)) [26,36,75]. One of the main factors that hinders the applicability of the conventional Fenton process is the need for a subsequent pH adjustment to a higher value (close to neutrality) for the subsequent coagulation/flocculation steps in the WTPs, increasing the process costs [76]. Thus, the proposed ZVI/H 2 O 2 process shows the additional advantage of spontaneously increase the pH to a convenient range (6-7.5 in natural water), along the course of the reaction. Consequently, a pH adjustment step using a base for the subsequent coagulation/flocculation steps may not be necessary, reducing treatment costs and process complexity.   [78]. In this study, after the H 2 O 2 concentration reached <0.5 mg L −1 , the possibility of NOM degradation by oxidation was considered to have ceased.

Effect of H 2 O 2 and ZVI nail dosages
The run dosed with 100% excess of H 2 O 2 (85 mg L −1 ) at initial pH 4.5 showed a fast oxidant consumption in the first 15 min of reaction, being about 66% of H 2 O 2 consumed (k = 0.0725 min −1 (Equation (5))). This behaviour may be a consequence of the reaction between the H 2 O 2 and H + ions with the ZVI nail surface, simultaneously releasing ferrous ions into the solution [79]. Results of Figure 3 shows higher concentration of ferrous ions in aqueous medium at the beginning of the ZVI/H 2 O 2 reaction, which provided a higher catalytic effect to the H 2 O 2 consumption [44,80]. Consequently, greater removal of DOC and UV 254 was observed in this time interval. After this time, the rate of HO• generation may have been reduced due to the lower concentration of ferrous ions available to react with H 2 O 2 , which justifies the lower H 2 O 2 consumption (k = 0.0642 and 0.0585 min −1 for 30 and 45 min, respectively) and reduced DOC removal. The removal of UV 254 after 15 min (shown in Figure 3) probably occurred from reactions of other iron species with the remaining H 2 O 2 , which can produce reactive oxygen species with lower oxidative potential, but still sufficient to reduce the humic structures of NOM [44].
The calculations of the pseudo-first order constants of the H 2 O 2 concentration decay were performed according to Equation (5)   Considering that only the dissolved ferrous iron concentration was measured in this study, it is suggested for future studies to determine also the dissolved ferric iron concentration, as both species would be present during the process [73].

Effects of ZVI/H 2 O 2 on DOC, UV 254 , SUVA 254 and MW
The effects of ZVI/H 2 O 2 advanced oxidation on DOC, UV 254 , SUVA 254 , and MW parameters in the natural water were evaluated from the best experimental condition obtained in this study (pH = 4. 5   The partial mineralization of DOC achieved from lower pH level followed by higher H 2 O 2 dosage contributed to the expressive reductions of the humic fractions, hydrophobic content, and consequently, molecular weight (indicated by parameters UV 254 , SUVA 254 , and MW, respectively), as shown in Table A.5. The results show lower DOC mineralization efficiency compared to UV 254 removal (DOC = 51% and UV 254 = 89%), corroborating previous AOPs studies for HA removal [5,44,48,69,70]. However, the significant removal of SUVA 254 (78%) indicates a reduction in hydrophobic carbon content as a loss of aromatic structures and conjugated double bonds presents in NOM [5,44,48,69,70]. It is worth mentioning that because of the reduction of SUVA 254 , it was verified that ZVI/H 2 O 2 treatment also favoured the reduction of MW of NOM contained in natural water in 74% (after 60 min of ZVI/H 2 O 2 treatment) (Table A.5). The reduction of these medium/ micro molecules indicates the conversion of large organic structure into simpler molecules [70]. The initial MW value obtained in this study is similar to what has already been detected in some surface waters for potabilization [82]. These results reinforce that those possible reactions between HO• and NOM (such as HO• radical addition reactions to aromatic sites) may not be sufficient for the full range of mineralization [83], but favour the reduction of the structural complexity of the aromatic fractions of NOM [44].
Additional experimental results are presented in Table A.6, confirming the greater DOC and UV 254 removals as the pH range is reduced and the H 2 O 2 dosages are increased on the ZVI/H 2 O 2 process.

THM formation
In this study, Chloroform (CHCl 3 ) was the only THM species detected before and after the ZVI/H 2 O 2 reaction (ZVI nail = 50 g L −1 ; [H 2 O 2 ] = 100% of excess and initial pH = 4.5 and 6.5) in the chlorinated natural water. The possibility that bromide was converted to bromate in the treated water [84,85], and it was no longer available to be incorporated into the NOM, should be considered with caution. Complementary results (at acidic pH ranges) indicated a low oxidation rate of bromide during the ZVI/H 2 O 2 treatment (bromide before ZVI/ H 2 O 2 = 0.24 mg L −1 and bromide after ZVI/H 2 O 2 = 0.21 mg L −1 ). Previous studies have also identified CHCl 3 in surface water and wastewater in higher concentrations compared to brominated species [45,48,86,87].
The CHCl 3 formation before and after the ZVI/H 2 O 2 process are shown in Figure 4, confirming a significant reduction in CHCl 3 formation in samples treated at initial pH 4.5. The efficient removal of humic/aromatic structures of the NOM observed within 60 min of the ZVI/H 2 O 2 process at pH 4.5 (UV 254 removal: 89% and 53% to initial pH 4.5 and 6.5, respectively) may explain this reduction of the CHCl 3 formation. Previous studies showed that the structure of NOM and of its by-products could determine the concentration of THM formed during the chlorination step [3,87]. After 24 h of Ct, the sample treated with ZVI/H 2 O 2 at pH 4.5 showed a 48% reduction in CHCl 3 formation compared to the untreated water.
In contrast to the pH 4.5 results, the high concentration of CHCl 3 observed after the ZVI/H 2 O 2 treatment at initial pH 6.5 (Ct 24h = 156.4 ± 20.48 μg L −1 ) exceeded the maximum values for total THMs allowed for drinking water according to the Brazilian and British standards (< 100 μg L −1 ) [47,74], as shown in Figure 4 (dotted line). Golea et al. [88] showed a significant correlation (R² = 0.82) between the formation of THMs and values of the parameter UV 254 in surface waters used for drinking water. Other researchers also confirmed that the high CHCl 3 formation may be a consequence of the lower efficiency to remove HA [44,89,90]. This high CHCl 3 value in samples treated at pH 6.5 in this study may be due to the formation of more chlorine reactive byproducts in addition to the low reduction of the aromatic organic structure observed after AOP ZVI/H 2 O 2. However, to prove the influence of these byproducts on the formation of CHCl 3 , more in-depth study and

Characterization of ZVI nail and its corrosion products
The oxidation of ZVI nail used in the ZVI/H 2 O 2 process (pH 0 = 4.5; H 2 O 2 = 100% of excess and ZVI nail = 50 g L −1 ) was confirmed by electron microscopy analyses. In addition to chemical results, these analyses provided images of the oxidized surface of the metal and the morphology of its oxides, helping understand the oxidative process. SEM results showed less ZVI nail corrosion on its surface prior to the reaction with H 2 O 2 , which was supported by the XEDS elemental maps ( Figure A.5  (a)). Different characteristics of the ZVI nail surface were observed after the reaction with 100% excess of H 2 O 2 , confirming the intense surface oxidation of ZVI nail after its reaction with H 2 O 2 (and H + ions of the solution), producing an uneven rough surface ( Figure A.5  (b)). The formation of ferrous ions as a product of ZVI oxidation favour the degradation of organic matter, as these ions are responsible for reacting with H 2 O 2 to produce HO• in the Fenton process [31,91,92].
The presence of the oxide species on the surface of the reacted ZVI nail was confirmed by XRD analysis. The identified species were: (1) Iron (Fe) ( (Figure A.6). Previous studies based on water/ wastewater treatment applying commercial iron powders also reported the formation and deposition of such species on the surface of the reacted ZVI [32,[91][92][93][94], as shown in this investigation.
SEM characterization confirmed the complex oxide morphologies on the ZVI nail surface after the oxidative process ( Figure 5 (a-d)). The secondary electron image of the oxide in Figure 5(b) exhibited a plate-like or flakelike structure, generally related to the presence of lepidocrocite hydroxide [93,94]. The fine spherical oxides (less than 200 nm) shown in Figure 5(c) are consistent with the morphology of magnetite [95,96]. The oxide morphology shown in Figure 5(d) may be lepidocrocite in a different formation stage when compared to the structure stage shown in Figure 5 Figure A.7 (b)). The surface of the reacted ZVI nail peaked at 531.42 eV, which indicated typical OHhydroxide-binding O 1s binding energy [95,96], and energy values at 529.91 eV indicating binding between O and Fe [96]. The formation of these oxides and hydroxides on the surface of the ZVI plays a critical role in removing contaminants, either by adsorption, precipitation, co-precipitation [99] or by favouring the formation of a passive layer on the metal [94], which limits the mass transport of ferrous ions from ZVI to the aqueous solution [36,100]. The presence of this multi-oxide layer transition-zone was confirmed by SEM/BSE analysis from the ZVI nail cross-section specimen after the ZVI/ H 2 O 2 reaction ( Figure 6). The different contrast in this image is due to the atomic number (compositional) difference between the regions, with the brighter contrast associated with the highest atomic weight (Fe in this figure) [101] and the grey contrast indicative of the oxide layer on the ZVI nail surface.
The presence of these complex oxides on the surface of the ZVI nail formed during the ZVI/H 2 O 2 process (verified by SEM, XRD, and XPS analyses) forms the passive layer (identified by SEM/BSE and cross-section analyses), which can limit the mass transport of iron ions from the ZVI nail to the aqueous solution. Thus, it can be speculated that the NOM removal after certain ZVI/H 2 O 2 reaction time (as discussed in Section 3.7) might have been influenced by this oxide layer formed.

Overall remarks about the iron-nails ZVI/ H 2 O 2 process
The removal of DOC (51%) and UV 254 (89%) attained from the application of the ZVI/H 2 O 2 process in natural water shown in this investigation indicated an important contribution of this system as a NOM pre-oxidation strategy in water treatment processes.
Experimental results revealed that NOM removal was predominant at the initial time period of the ZVI/H 2 O 2 reaction, especially over DOC removal. The AOP applied in this investigation at initial pH 4.5 showed that in 15 min of reaction, the DOC mineralization reached 46% with 66% H 2 O 2 consumption (k = 0.0725 min −1 ). In 30 min of reaction, DOC removal reached 51% with lower H 2 O 2 consumption velocity (k = 0.064 min −1 ) and lower concentration of dissolved Fe 2+ in the solution (Figure 3). These results suggest that other radicals with lower oxidative redox potential radicals [78,[102][103][104] and other iron species formed in this time interval are factors potentially responsible for the reduction in mineralization rates (Equations (6)- (8)).
The greatest DOC removal observed in the initial 15 min of the reaction was referred to as Stage I, as shown in Figure 7 (Fe 0 /H 2 O 2 and Fe 2+ /H 2 O 2 ). After this reaction time, they may have overlapped another stage, which is the controlling step of the overall reaction of this process, referred to as Stage II (Figure 7). Then, it could be hypothesized that the results of the ZVI/H 2 O 2 experiments (at initial pH 4.5 and 5.5) applied  to Regent's Park lake's water show the presence of these two stages (Figure 7).
The slowing down of the mineralization of the organic content observed in this investigation may be due to the influence of several factors. For example, the pH of the aqueous medium may have also influenced the oxidative process kinetics (as shown in Figure 1). This is because different radical species with lower oxidative potential could have been generated in the system, especially when the pH of the aqueous medium is above 5.0 [92]. This observation explains the lower NOM removal and, consequently, higher Chloroform THM formation, as identified in the ZVI/ H 2 O 2 experiment at the initial pH 6.5 ( Figure 4). Also, the formation and deposition of the oxide/hydroxide layer on the surface of the ZVI nail oxidized identified in this paper (shown in Figures 5 and 6 and Figures A.6-A.7) needs to be investigated, as it may slow down the metal dissolution step [105,106]. The effect and practical implications of this passive layer formation are being investigated and will be discussed in future publications.
It is highlighted that previous investigations with other AOPs, such as O 3 /UV, O 3 /H 2 O 2 , UV/H 2 O 2 , photo-Fenton, and Electrical discharge plasma for NOM and HA removal (as pre-oxidation step or after filtration step) [44,71,87,107,108], showed similar DOC removal in comparison to this paper (applied as a pre-oxidation step, without previous physico-chemical treatment). However, the ZVI/H 2 O 2 process investigated here has the advantage of making use of an easily accessible and low-cost reagent (i.e. commercial iron-nails) for the preliminary removal of NOM in drinking water treatment.
In addition to monitoring the NOM, a set of contaminanting elements that may be present in natural and treated waters such as Al, Cd, Cr, and Cu were analysed before and during the ZVI/H 2 O 2 process (initial pH 4.5 and 6.5, ZVI = 50 g L −1 and H 2 O 2 100% excess of stoichiometric dosage). From the experimental conditions applied in this study, the removal of these elements was not observed as shown in Table A.7. At the end of the experiments, Cr, one of the elements present in the composition of the ZVI-nail (Table A.1) presented concentrations close to that found in natural water before ZVI/H 2 O 2 treatment. Thus, it is not possible from these results to affirm a Cr leaching effect from the ZVI-nail to the treated water. Also, element concentrations were well bellow drinking water standards (in mg L −1 : Al: 0.1; Cd: 0.005; and Cu: 2.0) [47].
The authors propose further studies with nail-ZVI/H 2 O 2 in continuous mode to evaluate the iron nails performance. Also, the additional reduction of NOM which is expected to occur in the steps of the conventional treatment train after the ZVI/H 2 O 2 pre-oxidation discussed in this paper is an interesting proposal for future investigations on the efficiency of coagulation and quantification of THM formation.

Conclusions
. Two variables had a significant effect (p < 0.05) on the mineralization of NOM present in natural water: H 2 O 2 concentration (best at 100% of excess) and initial pH (best at 4.5). The NOM removal evaluated by DOC, UV 254 , SUVA 254 , and MW after ZVI/H 2 O 2 treatment in these operating conditions led to lower concentration and lower reactivity of its remaining structure with chlorine, resulting in the best condition, with 48% less formation of THM Chloroform in comparison with no pre-oxidation. The ZVI nails employed in the present study were significantly oxidized by H 2 O 2 in the presence of H + ions in the solution, confirming their reactivity and capacity of producing aqueous Fe ions with catalytic activity for the Fenton reaction. The formation of different oxide structures on their surface was verified as resulting from oxidative reactions. . The observed spontaneous increase in pH along the course of the reaction from initial pH 4.5 to the range pH 5.5-6 favours the insertion of this pre-oxidation step in conventional water treatment trains, as this range of final pH after pre-oxidation is typical of surface waters. Fe hydroxide precipitates formed in the water as part of the reaction mechanism can aid the coagulation-flocculation process and can be subsequently incorporated in the sludge removed in the sedimentation stage. . In addition to these results, the low-cost of commercial iron-nails, their wide accessibility, and the fact that they can be easily separated from the water after the pre-oxidation reaction step or when necessary, make our proposed AOP very attractive as simple pre-treatment for NOM removal in WTPs. This process may be of interest to water utilities with limited financial resources and infrastructure to implement a more sophisticated and costly water advanced pre-oxidation process. data analysis. Special thanks to Dr R. G. Palgrave at the Department of Chemistry, University College London, for XPS analyses and technical support. The Royal Parks are acknowledged for authorising the water sampling at the Regent's Park lake.

Disclosure statement
No potential conflict of interest was reported by the author(s).