Validating the Efficiency of the FeS2 Method for Elucidating the Mechanisms of Contaminant Removal Using Fe0/H2O Systems

There is growing interest in using pyrite minerals (FeS2) to enhance the efficiency of metallic iron (Fe0) for water treatment (Fe/H2O systems). This approach contradicts the thermodynamic predicting suppression of FeS2 oxidation by Fe0 addition. Available results are rooted in time series correlations between aqueous and solid phases based on data collected under various operational conditions. Herein, the methylene blue method (MB method) is used to clarify the controversy. The MB method exploits the differential adsorptive affinity of MB onto sand and sand coated with iron corrosion products to assess the extent of Fe0 corrosion in Fe/H2O systems. The effects of the addition of various amounts of FeS2 to a Fe0/sand mixture (FeS2 method) on MB discoloration were characterized in parallel quiescent batch experiments for up to 71 d (pH0 = 6.8). Pristine and aged FeS2 specimens were used. Parallel experiments with methyl orange (MO) and reactive red 120 (RR120) enabled a better discussion of the achieved results. The results clearly showed that FeS2 induces a pH shift and delays Fe precipitation and sand coating. Pristine FeS2 induced a pH shift to values lower than 4.5, but no quantitative MB discoloration occurred after 45 d. Aged FeS2 could not significantly shift the pH value (final pH ≥ 6.4) but improved the MB discoloration. The used systematic sequence of experiments demonstrated that adsorption and coprecipitation are the fundamental mechanisms of contaminant removal in Fe/H2O systems. This research has clarified the reason why a FeS2 addition enhances the efficiency of Fe0 environmental remediation.


Introduction
Metallic iron (Fe 0 ), commonly termed as zero-valent iron (ZVI), is widely used for decentralized safe drinking water provision [1][2][3] and environmental remediation [4][5][6]. Its suitability to remove a large array of biological and chemical pollutants from aqueous solutions has been demonstrated over the past three decades [7][8][9][10][11][12][13][14]. However, Fe 0 is mostly considered as a reducing agent (E 0 = −0.44 V for the redox couple Fe II /Fe 0 ) for reducible species [4,12,15]. This reductionist but wrong view has been motivated by the failure to consider the inherent features of aqueous iron corrosion, which results in the shortcomings of the Fe 0 remediation technology [9]. These features include: (i) the increase of the pH value, coupled with subsequent shielding of the Fe 0 surface by an oxide scale (reactivity loss), rooting the reasoning on galvanic interactions, reference [49] suggested using Fe 0 (E 0 = −0.44 V) to galvanically prevent FeS 2 (E 0 = 0.25 V) oxidation [49]. In this process, Fe 0 is anodically oxidized, and Fe 2+ ions are transported to cathodic sites on the FeS 2 surface, where a protective Fe-oxyhydroxide layer develops. Clearly, Seng et al. [49] demonstrated the suitability of a Fe 0 addition to suppress FeS 2 oxidation. This is exactly the opposite goal of using Fe 0 /FeS 2 systems in water remediation [15]. This controversy calls for a more detailed investigation of the dynamics within the Fe 0 /FeS 2 /H 2 O system.
The suitability of the FeS 2 method for the characterization of the dynamic process of contaminant removal in Fe 0 /H 2 O systems arises from the fact that shifting the pH to lower values increases the solubility of iron [15]. If the pH is decreased to values lower than 4.5, iron precipitation will occur in the bulk solution and not in the vicinity of Fe 0 . In other words, the FeS 2 addition certainly decreases the pH of the system. However, the impact of the pH shift on the decontamination process depends on the extent of the pH shift (minimum pH achieved) and the final pH value. Both the minimum pH and the final pH values depend on several operational factors, including the relative amounts of Fe 0 and FeS 2 and the experimental duration. It is recalled that Fe 0 corrosion increases the pH value while FeS 2 oxidation reduces it [15]. The FeS 2 method implies that the FeS 2 reactivity might be exhausted during the experiment, so that, for longer experimental durations, the final pH should be equal or higher than the initial value.
In the present work, the MB method [41] and the FeS 2 method [44] were combined in order to better characterize the dynamic process of iron corrosion and the interactions within Fe 0 /H 2 O systems, accounting for contaminant removal. Parallel MB discoloration batch experiments in Fe 0 /sand, FeS 2 /sand, and Fe 0 /FeS 2 /sand systems were undertaken for up to 71 d. Under these conditions, various amounts of FeCPs were generated for differential MB discoloration. The FeS 2 mineral was used in its pristine form (pristine FeS 2 ) and after aging for six months (aged FeS 2 ). Additional experiments were performed for the discoloration of methyl orange (MO) and reactive red 120 (RR120); the results are compared.

Dyes
Methylene blue (MB) was used as a tracer of reactivity [41], while methyl orange (MO) and reactive red 120 (RR120) were used as model organic micropollutants [15,51]. The three dyes are widely used to characterize the suitability of various systems for water treatment [51][52][53][54]. The used dyes were of analytical grade. MB was supplied by Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), MO by Tianjin Chemical Reagent Research Institution Co., Ltd. (Tianjin, China), and RR120 by Sigma-Aldrich ® (St. Louis, MO, USA). The dyes were selected due to: (i) similarity in their molecular size (MB and MO) and (ii) differences in their affinity to positively charged iron oxides (Table 1) [45,55]. RR120 was included, because it has a higher molecular size relative to MO. The initial dye concentration used was 10 mg L −1 , equivalent to 31.5 µM for MB, 30.7 µM for MO, and 6.8 µM for RR120. The dye working solutions were prepared by dissolving the corresponding crystals in deionized water. The pH values of the initial solution were 6.5 (MB), 6.6 (MO), and 7.0 (RR120). Other chemicals used in this study included L(+)-ascorbic acid and L-ascorbic acid sodium salt. Ascorbic acid also degrades dyes (in particular, MO) and eliminates interference during iron determination [30,45].

Iron
A standard iron solution (Testing & Certification Co., Ltd., Beijing, China) (1000 mg L −1 ) from the General Research Institute for Nonferrous Metals was used to calibrate the UV/VIS spectrophotometer (Shanghai Jinghua Technology Instrument Co., Ltd, Shanghai, China) used for analysis. All other chemicals used were of analytical grade. In preparation for the spectrophotometric analysis, ascorbic acid was used to reduce Fe III in the solution to Fe II . 1,10 orthophenanthroline was used as reagent for Fe II complexation.

Solid Materials
The used Fe 0 material was purchased from the Shanghai Institute of Fine Technology (Shanghai, China). The Fe 0 used in the current study has been used and described in earlier studies by our research group [45,56]. The material is available as scrap iron with a particle size between 0.05 and 5 mm. Its elemental composition, as specified by the supplier, was: Fe: >99.99%, C: <0.1%, N: <0.1%, and O: <0.1%. Its k Phen value was 13 mg h −1 [56]. The k Phen value is the kinetic parameter of Fe 0 dissolution in a 2-mM 1,10 orthophenanthroline solution, which characterizes the material's intrinsic reactivity [57]. The material was used without any further pretreatment. Fe 0 has been previously proven to be a powerful discoloration agent for MB specifically because the discoloration agents are progressively generated in situ [58,59]. Therefore, the discoloration capacity of used Fe 0 cannot be exhausted within the experimental duration (up to 71 d).

Sand
The used sand was China ISO standard sand, which was used as received without any further pretreatment or characterization. The particle size was between 1.25 and 2.00 mm. Sand was used because of its worldwide availability and its use as an admixing agent to prevent a rapid permeability loss in Fe 0 /H 2 O systems [60,61].

Pyrite (FeS 2 )
The used FeS 2 mineral was from Tongling City, Anhui Province, China. The particle size was between 38 and 48 µm. Its weight composition was 46.0% Fe and 52.2% S; this is equivalent to a purity of 98.2% [29]. FeS 2 was used because of its demonstrated suitability as a pH-shifting agent in Fe 0 /H 2 O systems [62][63][64][65].

Dye Discoloration
Quiescent batch experiments with pristine FeS 2 were conducted in glass test tubes for an experimental duration of 41 d. Dye discoloration was initiated by adding 20.0 mL of the dye solution to a test tube containing 0.0 or 0.1 g of Fe 0 , 0.0 to 0.6 g of FeS 2 , and 0.0 or 0.5 g of sand. Thus, Fe 0 /FeS 2 /sand mixtures contain the same mass of Fe 0 (0.1 g), the same mass of sand (0.5 g), and varying FeS 2 loadings (0.05 to 0.60 g). The corresponding mass loadings of the individual materials varied from 0.0 to 30.0 g L −1 . The blank corresponds to no material addition (0.0 g of Fe 0 , FeS 2 , and sand). Three single aggregate (Fe 0 , FeS 2 , and sand) and two binary (Fe 0 /FeS 2 and Fe 0 /sand) systems were investigated. The discussion is mainly focused on the ternary systems. The seven Fe 0 /FeS 2 /sand systems contained 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6 g of FeS 2 (2.5 to 30.0 g L −1 ) respectively. Besides sustaining the MB method, the sand addition was intended to avoid compaction of the material by gelatinous FeCPs (cementation) [61]. Three sets of experiments were conducted. The first set of experiments using pristine FeS 2 were initiated in April 2019, lasting for 41 d. In October 2019, a second set of experiments using the aged form of the same FeS 2 mineral (aged FeS 2 ) were initiated while extending the dyes to RR120. The second set of experiments lasted for 45 d. The third and last set of experiments entailed following the time-dependent changes in the ternary system for up to 71 d using the following material mixture: Fe 0 (0.1 g or 5.0 g L −1 ) + FeS 2 (0.1 g or 5.0 g L −1 ) + sand (0.5 g or 25.0 g L −1 ). In addition, parallel experiments with each dye were performed.
The efficiency of individual systems for dye discoloration was characterized at laboratory temperatures (about 25 ± 2 • C). The final pH values, the iron concentrations, and the residual dye concentrations were recorded (t = 41 d for pristine FeS 2 and t = 45 d for aged FeS 2 ). All experiments were carried out in triplicates under laboratory (oxic) conditions. The test tubes were protected from direct sunlight, as described in earlier studies [30,45,56].

Analytical Methods
Aqueous dye and iron concentrations were determined by a 752 UV/VIS Spectrophotometer (automatic) (Shanghai Jing Hua Technology Instrument Co., LTD, Shanghai, China). The respective working wavelengths for MB, MO, RR120, and iron were 664.5, 464.0, 515.0, and 510.0 nm ( Table 1). Cuvettes with a 1.0-cm light path were used. The iron determination followed the 1,10 orthophenanthroline method [66]. The spectrophotometer was calibrated for dye concentrations ≤15.0 mg L −1 and iron concentrations ≤10.0 mg L −1 . The pH values were measured by combined glass electrodes (INESA Scientific Instrument Co., Shanghai, China).

Presentation of Experimental Results
In order to characterize the magnitude of the tested systems for dye discoloration, the discoloration efficiency (E) was calculated (Equation (1)): where C 0 was the initial aqueous dye concentration (10.0 mg L −1 ), while C gave the final residual dye concentration. The operational initial concentration (C 0 ) for each case was acquired from a triplicate control experiment without additive materials (so-called blank). This procedure was mainly to account for experimental errors due to dye adsorption onto the walls of the test tubes.

Results and Discussion
Experiments with pristine FeS 2 are considered the reference system for the discussion. April 2019 is operationally considered as t 0 = 0, although the same material was already used in previous works [29,35]. The presentation is mostly focused on the difference between pristine and aged FeS 2 .  The results for MO discoloration showed a similar trend, with a slightly higher iron concentration for aged FeS 2 (up to 6.0 mg L −1 ). For pristine FeS 2 , the pH value decreased substantially from 7.0 to 4.5 (Figure 1a  In both cases, the dye discoloration is maximal in the Fe 0 /sand system ((FeS2) = 0 g L −1 ). This observation is seemingly in contradiction with the evidence that FeS2 improves the efficiency of Fe 0 /H2O systems for contaminant removal [29,62,67]. However, two key features have to be considered: (i) the presence of sand and (ii) the fact that the systems are quiescent, hence no accelerated mass transfer [19,68]. In other words, the progress of the discoloration process is intentionally slowed down in order to better characterize the interactions accounting for contaminant removal in Fe 0 /H2O systems [43,44,69]. Another seemingly controversial observation is the fact that the pristine FeS2 system is the most efficient for both MB and MO.  Figure 2a shows that the E value for MB removal is almost constant at 80% as the pristine FeS2 loadings vary from 0 to 25 g L −1 . The initial E value (75%) for the aged FeS2 system first decreases to  In both cases, the dye discoloration is maximal in the Fe 0 /sand system ((FeS 2 ) = 0 g L −1 ). This observation is seemingly in contradiction with the evidence that FeS 2 improves the efficiency of Fe 0 /H 2 O systems for contaminant removal [29,62,67]. However, two key features have to be considered: (i) the presence of sand and (ii) the fact that the systems are quiescent, hence no accelerated mass transfer [19,68]. In other words, the progress of the discoloration process is intentionally slowed down in order to better characterize the interactions accounting for contaminant removal in Fe 0 /H 2 O systems [43,44,69]. Another seemingly controversial observation is the fact that the pristine FeS 2 system is the most efficient for both MB and MO.  In both cases, the dye discoloration is maximal in the Fe 0 /sand system ((FeS2) = 0 g L −1 ). This observation is seemingly in contradiction with the evidence that FeS2 improves the efficiency of Fe 0 /H2O systems for contaminant removal [29,62,67]. However, two key features have to be considered: (i) the presence of sand and (ii) the fact that the systems are quiescent, hence no accelerated mass transfer [19,68]. In other words, the progress of the discoloration process is intentionally slowed down in order to better characterize the interactions accounting for contaminant removal in Fe 0 /H2O systems [43,44,69]. Another seemingly controversial observation is the fact that the pristine FeS2 system is the most efficient for both MB and MO.  Figure 2a shows that the E value for MB removal is almost constant at 80% as the pristine FeS2 loadings vary from 0 to 25 g L −1 . The initial E value (75%) for the aged FeS2 system first decreases to  Figure 2a shows that the E value for MB removal is almost constant at 80% as the pristine FeS 2 loadings vary from 0 to 25 g L −1 . The initial E value (75%) for the aged FeS 2 system first decreases to about 25% at (FeS 2 ) = 5 g L −1 and then progressively increased to 50% at (FeS 2 ) = 25 g L −1 . This observation suggests that, in the presence of pristine FeS 2 , MB discoloration is achieved by sand and that its surface was not coated to the extent that MB discoloration is affected. Remember that the surface of sand is negatively charged and is an excellent adsorbent for positively charged MB (a cationic dye) [42]. Once the surface of sand is in-situ covered with positively charge FeCPs, it ceases to adsorb MB. This differential adsorptive behavior of MB on sand and iron-coated sand is the cornerstone of the MB method for investigating the reactivity of the Fe 0 /H 2 O system [37,40,41]. Clearly, the pristine FeS 2 had no significant impact on MB adsorptive discoloration under the operational conditions. This demonstrates that iron precipitation was not quantitative to the extent that MB coprecipitation was significant. The situation is different in the presence of aged FeS 2 (Figure 2a). There was practically no pH shift ( Figure 1a) and no significant increase of the iron concentration (Figure 1b), meaning that the in-situ precipitation of FeCPs in the vicinity of Fe 0 significantly impaired the adsorptive MB discoloration by sand. The observed increase of the E value from 25% to 51% (Figure 2a) is attributed to MB coprecipitation with FeCPs. Such a coprecipitation could not occur in the presence of pristine FeS 2 , where MB adsorption on sand was the sole process responsible for discoloration. Figure 2b shows E ≥ 40%, attesting that MO has a higher affinity to in-situ generated FeCPs. The system with pristine FeS 2 achieved much higher MO discoloration than the aged one, indicating that there were FeCPs formations but no coating of the sand surface. This is the reason why MB adsorption by sand was not disturbed in parallel experiments (Figure 2a). The initial E value of 80% ((FeS 2 ) = 0 g L −1 ) decreased to 70% for (FeS 2 ) = 15 g L −1 and remained constant for higher FeS 2 loadings. These results clearly show that FeCPs precipitation was not quantitative and was even less pronounced in systems with low pH values. The lower the pH value, the less the precipitation of FeCPs from the aqueous solution. It is essential to differentiate precipitation in the aqueous phase from precipitation in the vicinity of Fe 0 . The behavior of the aged FeS 2 is related to the second case. Here, precipitation is retarded by sand, which has negative charged surfaces that attract ferrous (Fe 2+ ) and ferric (Fe 3+ ) ions, thereby delaying the availability of free FeCPs for dye coprecipitation. This statement is supported by the increasing trend of MO discoloration as the (FeS 2 ) value increased from 5 to 25 g L −1 (Figure 2b).  . It is seen that the initial pH value of 7.0 decreased after the addition of FeS 2 to values of 6.8 for RR120 and 6.5 for both MB and MO. There was a further pH decrease until day 5, reaching pH 6.4 for both MB and MO. The very close behavior of the MB and MO systems with respect to changes of the pH values was already reported by Gatcha-Bandjun et al. [45]. After day 5, there is a progressive increase of the pH value with the increasing time. This trend is justified by iron corrosion determining the pH value [14]. Note that, in a Fe 0 /FeS 2 /H 2 O system under oxic conditions, the pH value is determined by two concurrent processes: (i) FeS 2 oxidation lowering the pH and (ii) iron corrosion increasing the pH. These two situations are evident in Figure 3b, where FeS 2 oxidation fixed the pH for (FeS 2 ) ≤ 5 g L −1 , while iron corrosion determined the pH for higher FeS 2 loadings. These results were achieved with aged FeS 2 , and the final pH values varied between 6.4 and 7.6 for all investigated systems (variations of only 1.2 pH units). With pristine FeS 2 , the window of pH variation would have been larger and the experimental duration longer for similar observations [14]. Noubactep et al. [43] needed up to 120 d to document this trend in their experiments for U(VI) removal.  Figure 3b summarizes the results of changes of the iron concentrations as a function of the experimental duration. It is seen that more than 3.0 mg L −1 Fe is present in the system with MO, while the two other systems exhibited values lower than 1.0 mg L −1 . Similar results were reported by Phukan et al. [70] for Orange II and Gatcha-Bandjun et al. [45] for MO and were explained by the formation of relative stable complexes between Fe and MO (and Orange II). The fact that more iron was dissolved in the RR120 system than in the MB system can be justified by steric effects by which a larger RR120 disturbs the formation of an oxide scale on Fe 0 , as highlighted by Phukan [71].

MB and MO Discoloration in Fe 0 /FeS2/Sand Systems
The remaining sections will discuss the role of FeCPs on the process of contaminant removal while particularly addressing the role of FeS2 in sustaining the efficiency of Fe 0 /H2O systems. The time-dependent changes of E values for the three used dyes will be comparatively discussed. Figure 4 summarizes the results of changes of the E values as a function of the experimental duration for MB, MO, and RR120. It is seen that there is a general trend of increasing E values with increasing equilibration times. However, some species-dependent observations should be highlighted: (i) E values for RR120 continuously increased from the start of the experiment towards the end from 49% to 85%, (ii) E values for MB also increased from 0 to 60 d and then slightly decreased, and (iii) the initial E value of 15% for MO first increased to 45% and, subsequently, increased towards the end (73%). The differences between MB and MO are explained by the differential affinity of both dyes to FeCPs and sand (Section 3.2). For example, in the initial phase (up to 40 d), the higher values for the MB system are attributed to the availability of the sand surface to adsorb MB, because not enough in-situ generated FeCPs have coated the sand. After day 40, the available sand can be considered completely covered with FeCPs, hence the higher E values for MO and, also, the decrease of the E value for MB at the end of the experiments. The higher E values for RR120 than the other two dyes are justified by steric effects due to the larger molecular size of RR120 (Table 1) and, probably, its lower solubility in water [55].   [70] for Orange II and Gatcha-Bandjun et al. [45] for MO and were explained by the formation of relative stable complexes between Fe and MO (and Orange II). The fact that more iron was dissolved in the RR120 system than in the MB system can be justified by steric effects by which a larger RR120 disturbs the formation of an oxide scale on Fe 0 , as highlighted by Phukan [71].

Dye Discoloration
The remaining sections will discuss the role of FeCPs on the process of contaminant removal while particularly addressing the role of FeS 2 in sustaining the efficiency of Fe 0 /H 2 O systems. The time-dependent changes of E values for the three used dyes will be comparatively discussed. Figure 4 summarizes the results of changes of the E values as a function of the experimental duration for MB, MO, and RR120. It is seen that there is a general trend of increasing E values with increasing equilibration times. However, some species-dependent observations should be highlighted: (i) E values for RR120 continuously increased from the start of the experiment towards the end from 49% to 85%, (ii) E values for MB also increased from 0 to 60 d and then slightly decreased, and (iii) the initial E value of 15% for MO first increased to 45% and, subsequently, increased towards the end (73%). The differences between MB and MO are explained by the differential affinity of both dyes to FeCPs and sand (Section 3.2). For example, in the initial phase (up to 40 d), the higher values for the MB system are attributed to the availability of the sand surface to adsorb MB, because not enough in-situ generated FeCPs have coated the sand. After day 40, the available sand can be considered completely covered with FeCPs, hence the higher E values for MO and, also, the decrease of the E value for MB at the end of the experiments. The higher E values for RR120 than the other two dyes are justified by steric effects due to the larger molecular size of RR120 (Table 1) and, probably, its lower solubility in water [55].

Dye Discoloration
The results in Figure 4 demonstrated the ion-selective nature of the Fe 0 /H 2 O system. In fact, the competitive nature of dye discoloration by sand and iron-coated sand is demonstrated. This result validates the MB method for the characterization of the Fe 0 /H 2 O system, while demonstrating the selectivity of the Fe 0 /H 2 O system for negatively charged species, including bacteria and viruses [72]. In fact, several studies have shown that Fe 0 /H 2 O systems have the capacity to remove both bacteria and viruses in aqueous systems [1,5,9,62]. Given that pathogens are the most widespread contaminant The results in Figure 4 demonstrated the ion-selective nature of the Fe 0 /H2O system. In fact, the competitive nature of dye discoloration by sand and iron-coated sand is demonstrated. This result validates the MB method for the characterization of the Fe 0 /H2O system, while demonstrating the selectivity of the Fe 0 /H2O system for negatively charged species, including bacteria and viruses [72]. In fact, several studies have shown that Fe 0 /H2O systems have the capacity to remove both bacteria and viruses in aqueous systems [1,5,9,62]. Given that pathogens are the most widespread contaminant worldwide, Fe 0 filters for household drinking water treatments are a good candidate to achieve Goal 6 of the UN Sustainable Development Goals [73,74]

Implications on Contaminant Removal Mechanisms
The idea that Fe 0 is an environmental reducing agent prevails in the recent scientific literature [4,6,36,75]. Admixing Fe 0 with reactive FeS2 is accordingly regarded as a tool to enhance the reductive capacity of Fe 0 by two mechanisms: (i) freeing the Fe 0 from in-situ generated FeCPs and delay or even avoid "reactivity loss" and (ii) mediating electron transfer to the contaminants by virtue of the semiconductive nature of FeS2 minerals [15,29,62,76]. This view contradicts the concept of Khudenko [77], demonstrating that Fe 0 can be used to induce the indirect reduction of organic species in water and that copper salts can be used to sustain such indirect reactions. The view of Khudenko [77] was independently introduced by Lipczynska-Kochany et al. [62] using FeS2 to sustain Fe 0 oxidative dissolution and better discussed the observed reductive degradation of CCl4 in the presence of Fe 0 .
Noubactep et al. [43] presented their results on U(VI) removal in Fe 0 /H2O systems in the peerreviewed literature in 2003. These results were available in the grey literature two years earlier [78,79]. These authors demonstrated through a sequence of experiments that no significant U(VI) removal is achieved before quantitative iron precipitation occurs. The same authors then proposed the FeS2 method for the investigation of the mechanism of contaminant removal in Fe 0 /H2O systems [44]. The FeS2 method entails inducing a pH shift to lower values (ideally, lower than 4.5) by pyrite oxidation and monitoring the contaminant removal when the pH subsequently increased as iron corrosion continues. The results demonstrated that electrochemical reduction (electrons from Fe 0 ) plays no significant role in the process of U(VI) removal in the presence of Fe 0 . This observation implies that contaminant removal is a property of corroding iron: the spontaneous precipitation of iron hydroxides. This corresponds to flocculation, which is a conventional chemical process used in water treatments [80]. Thus, immersed Fe 0 in situ generates contaminant scavengers. Moreover, iron hydroxide precipitation occurs in the presence of trace amounts of contaminants, which are simply occluded via coprecipitation [14,18,19,81,82]. Iron hydroxide precipitation on FeS2 corresponds to the

Implications on Contaminant Removal Mechanisms
The idea that Fe 0 is an environmental reducing agent prevails in the recent scientific literature [4,6,36,75]. Admixing Fe 0 with reactive FeS 2 is accordingly regarded as a tool to enhance the reductive capacity of Fe 0 by two mechanisms: (i) freeing the Fe 0 from in-situ generated FeCPs and delay or even avoid "reactivity loss" and (ii) mediating electron transfer to the contaminants by virtue of the semiconductive nature of FeS 2 minerals [15,29,62,76]. This view contradicts the concept of Khudenko [77], demonstrating that Fe 0 can be used to induce the indirect reduction of organic species in water and that copper salts can be used to sustain such indirect reactions. The view of Khudenko [77] was independently introduced by Lipczynska-Kochany et al. [62] using FeS 2 to sustain Fe 0 oxidative dissolution and better discussed the observed reductive degradation of CCl 4 in the presence of Fe 0 .
Noubactep et al. [43] presented their results on U(VI) removal in Fe 0 /H 2 O systems in the peer-reviewed literature in 2003. These results were available in the grey literature two years earlier [78,79]. These authors demonstrated through a sequence of experiments that no significant U(VI) removal is achieved before quantitative iron precipitation occurs. The same authors then proposed the FeS 2 method for the investigation of the mechanism of contaminant removal in Fe 0 /H 2 O systems [44]. The FeS 2 method entails inducing a pH shift to lower values (ideally, lower than 4.5) by pyrite oxidation and monitoring the contaminant removal when the pH subsequently increased as iron corrosion continues. The results demonstrated that electrochemical reduction (electrons from Fe 0 ) plays no significant role in the process of U(VI) removal in the presence of Fe 0 . This observation implies that contaminant removal is a property of corroding iron: the spontaneous precipitation of iron hydroxides. This corresponds to flocculation, which is a conventional chemical process used in water treatments [80]. Thus, immersed Fe 0 in situ generates contaminant scavengers. Moreover, iron hydroxide precipitation occurs in the presence of trace amounts of contaminants, which are simply occluded via coprecipitation [14,18,19,81,82]. Iron hydroxide precipitation on FeS 2 corresponds to the mechanism of acidification suppression, as recently suggested by Seng et al. [49], and is thus in tune with the thermodynamics of the system. Surprisingly, all results published after 2006 and claiming that FeS 2 sustains electrochemical contaminant reductions in Fe 0 /H 2 O systems have not considered the FeS 2 method and the three related references published in Environmental Science & Technology (2003) [43], Environmental Chemistry (2005) [44], and the Journal of Hazardous Materials (2006) [83]. These three journals are from three different publishers (ACS, CSIRO, and Elsevier, respectively) and can be regarded as authoritative. The rationale for this oversight could be the plethora of scientific articles on the broad topic of "remediation using Fe 0 ". However, it is the responsibility of individual researchers and research groups to make sure that our common database reflects the state-of-the-art knowledge. The present work was designed to address this issue while highlighting to all stakeholders, including editorial boards of scientific journals, that "recent is not whole". Specifically, there is a trend to recommend recent references, and the technical management of submitted manuscripts sometimes focus on this questionable aspect. The literature on "remediations using Fe 0 " is an illustration of the fact that bias can be introduced in the scientific literature and maintained for decades. In particular, the mechanism discussed here started in 1994 [16], but Khudenko [77] has been completely ignored since then; yet, this should be the starting point. The concept that Fe 0 is an environmental-reducing agent has never been established, as recently demonstrated by Ebelle et al. [75], but is still prevailing in both the old and recent literature [84][85][86][87].
By using a pyrite mineral in April 2019 (pristine FeS 2 ) and the same six months later (aged FeS 2 ), this work has critically evaluated the state-of-the-art knowledge on the mechanism of contaminant removal in Fe 0 /H 2 O systems in a unique way. An array of experimental conditions could be created that would have been difficult to conciliate if they were not from the same set of experiments. In particular, the fact that decreased pH values (or increased iron concentrations) ( Figure 1) should never be randomly interchanged with increased contaminant removal is elegantly demonstrated ( Figure 5). Figure 5 represents the results of 102 experimental points obtained herein and by Cui [88] in investigating the Fe 0 /FeS 2 /H 2 O system using MB, MO, and RR120. Figure 5a shows clearly that, regardless of the used dye, the dissolved iron level decreased with the increasing pH values. In all the systems, the initial pH value was 6.8 ± 0.2. There is thus no doubt that the FeS 2 addition decreases the pH value and increases the iron concentrations. The extent of the dye discoloration in each system depends on several operation parameters that determine the minimum value (e.g., <4.5 for pristine FeS 2 ) and the final pH value. Sections 3.1-3.3 have presented the results for individual systems. Figure 5b summarizes the pH dependence of the E values. It is seen that the values are scattered without any trend, unlike for the iron level (Figure 5a). mechanism of acidification suppression, as recently suggested by Seng et al. [49], and is thus in tune with the thermodynamics of the system. Surprisingly, all results published after 2006 and claiming that FeS2 sustains electrochemical contaminant reductions in Fe 0 /H2O systems have not considered the FeS2 method and the three related references published in Environmental Science & Technology (2003) [43], Environmental Chemistry (2005) [44], and the Journal of Hazardous Materials (2006) [83]. These three journals are from three different publishers (ACS, CSIRO, and Elsevier, respectively) and can be regarded as authoritative. The rationale for this oversight could be the plethora of scientific articles on the broad topic of "remediation using Fe 0 ". However, it is the responsibility of individual researchers and research groups to make sure that our common database reflects the state-of-the-art knowledge. The present work was designed to address this issue while highlighting to all stakeholders, including editorial boards of scientific journals, that "recent is not whole". Specifically, there is a trend to recommend recent references, and the technical management of submitted manuscripts sometimes focus on this questionable aspect. The literature on "remediations using Fe 0 " is an illustration of the fact that bias can be introduced in the scientific literature and maintained for decades. In particular, the mechanism discussed here started in 1994 [16], but Khudenko [77] has been completely ignored since then; yet, this should be the starting point. The concept that Fe 0 is an environmental-reducing agent has never been established, as recently demonstrated by Ebelle et al. [75], but is still prevailing in both the old and recent literature [84][85][86][87].
By using a pyrite mineral in April 2019 (pristine FeS2) and the same six months later (aged FeS2), this work has critically evaluated the state-of-the-art knowledge on the mechanism of contaminant removal in Fe 0 /H2O systems in a unique way. An array of experimental conditions could be created that would have been difficult to conciliate if they were not from the same set of experiments. In particular, the fact that decreased pH values (or increased iron concentrations) ( Figure 1) should never be randomly interchanged with increased contaminant removal is elegantly demonstrated ( Figure 5). Figure 5 represents the results of 102 experimental points obtained herein and by Cui [88] in investigating the Fe 0 /FeS2/H2O system using MB, MO, and RR120. Figure 5a shows clearly that, regardless of the used dye, the dissolved iron level decreased with the increasing pH values. In all the systems, the initial pH value was 6.8 ± 0.2. There is thus no doubt that the FeS2 addition decreases the pH value and increases the iron concentrations. The extent of the dye discoloration in each system depends on several operation parameters that determine the minimum value (e.g., <4.5 for pristine FeS2) and the final pH value. Sections 3.1-3.3 have presented the results for individual systems. Figure  5b summarizes the pH dependence of the E values. It is seen that the values are scattered without any trend, unlike for the iron level (Figure 5a). Figure 5. Summary of the pH dependence of (a) the iron concentration and (b) the extent of dye discoloration (E value) obtained herein and by Cui [88] while investigating the Fe 0 /FeS2/H2O systems for the discoloration of methylene blue (MB), methyl orange (MO), and reactive red 120. 102 triplicate experiments were performed. The initial pH was 6.8 ± 0.2; data are not sorted. It is hoped that, in establishing that contaminant removals in Fe 0 /H 2 O systems are due to flocculation (adsorption and coprecipitation) ( Figure 6) [89], the results presented herein will redirect the research for the design of the next generation of Fe 0 -based remediation systems, accounting for the ion-selective nature of the systems [14,84,[89][90][91][92]. The view that Fe 0 is a reducing agent [15,93,94] should be immediately abandoned [85]. This view has mediated or supported wrong wordings like zero-valent iron for metallic [14] or reductates for species reducible by Fe 0 according to the relative electrode potentials [94]. In 2013, the notion of electron efficiency was introduced [86,95,96] and has been progressively used by several research groups [86]. The electron efficiency concept strives to reduce Fe 0 wastage by accounting for the redistribution of electrons from Fe 0 to (i) target (reducible) contaminants (e.g., MO), (ii) dissolved oxygen (O 2 ), and (iii) reducible co-contaminants like nitrate (NO 3 − ) [86]. The present work has documented intensive Fe 0 oxidation without MO reduction.
Accordingly, contaminant reductive transformation is the consequence and not the cause or one of the causes of Fe 0 oxidative dissolution.
It is hoped that, in establishing that contaminant removals in Fe 0 /H2O systems are due to flocculation (adsorption and coprecipitation) ( Figure 6) [89], the results presented herein will redirect the research for the design of the next generation of Fe 0 -based remediation systems, accounting for the ion-selective nature of the systems [14,84,[89][90][91][92]. The view that Fe 0 is a reducing agent [15,93,94] should be immediately abandoned [85]. This view has mediated or supported wrong wordings like zero-valent iron for metallic [14] or reductates for species reducible by Fe 0 according to the relative electrode potentials [94]. In 2013, the notion of electron efficiency was introduced [86,95,96] and has been progressively used by several research groups [86]. The electron efficiency concept strives to reduce Fe 0 wastage by accounting for the redistribution of electrons from Fe 0 to (i) target (reducible) contaminants (e.g., MO), (ii) dissolved oxygen (O2), and (iii) reducible co-contaminants like nitrate (NO3 − ) [86]. The present work has documented intensive Fe 0 oxidation without MO reduction. Accordingly, contaminant reductive transformation is the consequence and not the cause or one of the causes of Fe 0 oxidative dissolution. Finally, the very recent results of Seng et al. [49] demonstrate on a purely thermodynamic perspective that Fe 0 should be used to stop FeS2 oxidation. In other words, upon immersion, the FeS2 surface is progressively coated by FeCPs from anodic Fe 0 oxidation. Since the opposite has been favored and largely documented in the Fe 0 remediation literature [15,29,69,76], it is time for more detailed investigations. The present work provides a kinetic-based answer, suggesting that more attention should be paid to the relative dissolution rate of both Fe 0 and FeS2. In particular, the suitability of MB as an operational tracer for the reactivity of the dynamic Fe 0 /H2O system is corroborated. It is very important to state that the MB method is complementing knowledge from the conventional approach while simultaneously providing further insights into the dynamic aspects. Conventionally, dynamic interactions between reactive materials (e.g., Fe 0 and FeS2.) and dissolved species (e.g., contaminants and O2) are investigated by employing a suite of solution chemistry and solid-phase characterization approaches [49,86,[97][98][99][100]. The achieved results are then used to develop Finally, the very recent results of Seng et al. [49] demonstrate on a purely thermodynamic perspective that Fe 0 should be used to stop FeS 2 oxidation. In other words, upon immersion, the FeS 2 surface is progressively coated by FeCPs from anodic Fe 0 oxidation. Since the opposite has been favored and largely documented in the Fe 0 remediation literature [15,29,69,76], it is time for more detailed investigations. The present work provides a kinetic-based answer, suggesting that more attention should be paid to the relative dissolution rate of both Fe 0 and FeS 2 . In particular, the suitability of MB as an operational tracer for the reactivity of the dynamic Fe 0 /H 2 O system is corroborated. It is very important to state that the MB method is complementing knowledge from the conventional approach while simultaneously providing further insights into the dynamic aspects. Conventionally, dynamic interactions between reactive materials (e.g., Fe 0 and FeS 2 .) and dissolved species (e.g., contaminants and O 2 ) are investigated by employing a suite of solution chemistry and solid-phase characterization approaches [49,86,[97][98][99][100]. The achieved results are then used to develop a time series correlation analysis justifying the contaminant removal or transformation in the Fe 0 /H 2 O system [99].

Concluding Remarks
The concept that contaminant removal in Fe 0 /H 2 O systems is caused by the precipitation of solid iron corrosion products (FeCPs) is consistent with many experimental observations. A FeS 2 addition induces a shift of the pH to lower values, but this process is only associated with contaminant removal when quantitative precipitation occurs. Contaminants are removed by adsorption onto FeCPs or occlusion in precipitating FeCPs (coprecipitation). Further, the relative importance of adsorption and coprecipitation can only be speculatively discussed due to the dynamic nature of the system and the difficulty to perform Fe mass balance. The expected ion-selective nature of the Fe 0 /H 2 O system is demonstrated by the differential discoloration of MB and MO. The comparatively higher removal of RR120 is justified by steric effects and its lower solubility. Finally, the differential reactivity of pristine and aged FeS 2 shows that both Fe 0 and reactive additives have to be characterized for their initial intrinsic reactivity and the long-term kinetics of oxidative dissolution. This observation calls for more systematic investigations for the design of the next generation of Fe 0 -based remediation systems.