The key role of contact time in elucidating the mechanisms of enhanced decontamination by Fe0/MnO2/sand systems

Metallic iron (Fe0) has shown outstanding performances for water decontamination and its efficiency has been improved by the presence of sand (Fe0/sand) and manganese oxide (Fe0/MnOx). In this study, a ternary Fe0/MnOx/sand system is characterized for its discoloration efficiency of methylene blue (MB) in quiescent batch studies for 7, 18, 25 and 47 days. The objective was to understand the fundamental mechanisms of water treatment in Fe0/H2O systems using MB as an operational tracer of reactivity. The premise was that, in the short term, both MnO2 and sand delay MB discoloration by avoiding the availability of free iron corrosion products (FeCPs). Results clearly demonstrate no monotonous increase in MB discoloration with increasing contact time. As a rule, the extent of MB discoloration is influenced by the diffusive transport of MB from the solution to the aggregates at the bottom of the vessels (test-tubes). The presence of MnOx and sand enabled the long-term generation of iron hydroxides for MB discoloration by adsorption and co-precipitation. Results clearly reveal the complexity of the Fe0/MnOx/sand system, while establishing that both MnOx and sand improve the efficiency of Fe0/H2O systems in the long-term. This study establishes the mechanisms of the promotion of water decontamination by amending Fe0-based systems with reactive MnOx.

www.nature.com/scientificreports/ Fe 0 corrosion rate as well-documented in the corrosion literature and referred to as 'passivation' [25][26][27][28] . However, "reactivity loss" has been introduced in the post-1990 literature to characterize the limited electron transfer from the metal body to some dissolved contaminants 16,19 . Given that under natural conditions Fe 0 is corroded only by protons from water dissociation (Eq. 1) 29 , Miyajima and Noubactep 30 argued that reactivity loss is a mirage. In fact, "reactivity loss" has also occurred in Fe 0 -based permeable reactive barriers successfully working for up to two decades [31][32][33][34] . On the other hand, Roh et al. 35 reported on Fe 0 specimens from World War I still corroding in soils. Clearly, it can be argued that the old motto "rust never rests" is valid for Fe 0 filters where corrosion additionally occurs under immersed conditions. The question is, how to ensure that Fe 0 oxidation with changing corrosion rates still secures clean water in the long-term?
During the past decade, substantial experiences have been accumulated on increasing the efficiency of Fe 0 / H 2 O systems by admixing Fe 0 with other materials (Table 1) 15,36 . However, these efforts were mostly misled by the misconception that Fe 0 is a reducing agent 36 . Fortunately, available data can be re-interpreted based on the chemistry of the system. It suffices to consider that reduction is not a relevant contaminant removal mechanism, and that contaminant reduction is never mediated by electrons from the metal body 11 . For example, MnO 2 is not reduced by Fe 0 (Eq. 2), but rather by Fe 2+ (Eq. 3a) (Fig. 1). Equation 3b depicts that MnO 2 reductive dissolution by Fe 2+ induces acidification of the system (releases protons). O 2 and other dissolved species are equally reduced by Fe 2+ and other reductive species present in the Fe 0 /H 2 O system (e.g., H 2 , Fe 3 O 4 , green rust) [37][38][39][40] . Thus, it is established that contaminants are reduced by an indirect mechanism (Fig. 1), and that this process continues even after virtual surface passivation (which is thus not a "loss of their reactivity"). Successful efforts to overcome Fe 0 passivation include the addition of gravel 41,42 , magnetite 15,43 , MnO x 44,45 , pyrite 36,46 , and sand 47,48 . The presence of inert sand improves the efficiency of even batch Fe 0 /H 2 O systems for water treatment 49 . However, the studies testing other reactive materials have not considered inert systems (e.g., sand) as operational references. Only Ndé-Tchoupé et al. 50 did such a comparison. However, the objective was to test pozzolan as an alternative filling material to sand for Fe 0 filters. In other words, while testing magnetite (Fe 3 O 4 ) as admixing agent for the reductive transformation of contaminants 43 , a reference Fe 0 /sand should have been considered in parallel experiments. The inclusion of an operational reference enables a better understanding of the specific action of the reactive additive (here Fe 3 O 4 ). Following the science of aqueous iron corrosion under environmental conditions 25,28 , this study premises that iron passivation is delayed by avoiding the precipitation of iron corrosion products in the vicinity of the metal. Thus, Fe 2+ and Fe 3+ ions are consumed instead of coating sand in Fe 0 /sand systems, and Fe 2+ ions are additionally consumed in the reductive dissolution of MnO x (Eq. 2) in the Fe 0 /MnO x / sand systems (Fig. 1). Note that all other aggregates including Fe 3 O 4 and granular activated carbon are in-situ coated by FeCPs like sand and the postulated effects are not realizable in the long-term (Table 1).
Fe 0 is used to efficiently remove various contaminants such as turbidity, pathogens, and dissolved species from aqueous solutions [51][52][53][54][55][56][57][58][59][60] . Chemical pollutants in the aqueous phase can be ions, molecules, and colloids. For reducible dissolved species, there is a trend to consider Fe 0 -based materials (E 0 = − 0.44 V) as (strong) reducing agent 51,56,60 , and contaminant reductive transformation as an electrochemical process 58 . It is evident that colloids,  59 . In particular, there is need to elucidate how insoluble Fe(OH) 3 contribute to the co-precipitation of pollutants from the aqueous phase. The objective of this study is to investigate the impact of MnO x addition on the efficiency of Fe 0 /H 2 O systems for MB discoloration as a function of the experimental duration (contact time). The specific objective is to confirm the suitability of 'MB discoloration' as powerful tool for the characterization of decontaminantion processes in Fe 0 /H 2 O systems while using MnO x and sand to control the availability of 'free' FeCPs. The extent of MB discoloration is investigated in five different systems: (i) Fe 0 alone, (ii) sand alone, (iii) Fe 0 /sand, (iv) Fe 0 / MnO x , and (v) Fe 0 /MnO x /sand for 7, 18, 25 and 47 days. A comparison of the results from the five systems provides critical information on the contaminant removal mechanisms and the role of MnO x .

Materials and methods
The theory of iron and manganese cycle in a Fe 0 /MnO x /sand system. Initially (t 0 = 0), when Fe 0 , MnO x and sand are put into the solution, there is no dissolved iron and no dissolved manganese in the system ( Table 2). At t > t 0 , Fe 0 is dissolved by protons (water) to generate H 2 and Fe 2+ (Eq. 1). Fe 2+ induces the reductive dissolution of MnO x (Eq. 3) [61][62][63][64][65] . At t > t 0 , the Fe 0 /MnO x /sand system hosts dynamic processes which might continue after Fe 0 depletion. In fact, the mixture of Fe and MnO x minerals is a very complex reactive system that has been investigated for more that a century [66][67][68] . The uniqueness of the Fe 0 /MnO x /sand system is that Fe minerals are generated in-situ and are comparatively more reactive than aged minerals like goethite or hematite. Because the pH of the system is larger than 5.0, in the absence of ligands, Fe and Mn hydroxides have very low solubility and precipitate not far away from their points of nucleation 65,69 . The dynamics within the Fe 0 /MnO x /sand entail a series of interchanges of iron and manganese from older to younger forms as follows: (i) dissolution of Fe 0 and MnO x , (ii) migration of Fe 2+ , Fe 3+ and Mn 2+ from the areas of their generation to areas where precipitation will occur, and (iii) precipitation in one or more forms of iron and manganese hydroxide.
In the Fe 0 /MnO x /sand system, iron and manganese chemically precipitate at the surface of MnO x , sand or in the bulk solution. Due to the good adsorptive affinities of Fe 2+ and Fe 3+ for sand surface, it is assumed that Only water has access to the metal surface. Fe 2+ and H 2 are stand-alone reducing agents. MnO 2 and other relevant dissolved species (e.g. RX) are reduced by Fe 2+ and H 2 . Upon the oxidation of Fe 2+ , various solid iron hydroxides/oxides (e.g. Fe(OH) 3 ) precipitate and act as contaminant scavengers. RX stands for an halogenated hydrocarbon. The key information is that MnO 2 is not reduced by Fe 0 . Table 2. Time-dependent inventory of reactive species in the four investigated systems. t 0 corresponds to the start of the experiment, while t ∞ corresponds to the time required for Fe 0 depletion. It is assumed that MnO 2 is quantitatively converted to MnOOH without impact on MB discoloration. FeCPs: Fe corrosion products. FeCPs can be free or coated on sand. (Adapted from ref. 70  www.nature.com/scientificreports/ the deposition of Fe hydroxides at its surface (coating) will compete with Fe 2+ consumption by the reductive dissolution of MnO x until sand coating is completed. Thereafter, the "free" precipitation of iron and manganese occurs and the final products are deposits of more or less pure iron and manganese ores 66,71 . In other words, the investigated Fe 0 /MnO x /sand system is a ternary system only at the start of the experiment. It then turns to a mixture of Fe 0 , iron oxide-coated sand, iron oxide-coated MnO x , Fe/Mn shales, etc. Even after Fe 0 depletion, the Fe/Mn mineral mixture will still be a reactive one, with a great potential for water treatment by both abiotic and biotic processes 67,68,72 . Experimental details. This experimental section is adapted from Cao et al. 70 using the same experimental design and two more MnO 2 minerals.
Solutions. The used methylene blue (MB-Basic Blue 9 from Merck) was of analytical grade. The working solution was 10.0 mg L -1 prepared by diluting a 1000 mg L -1 stock solution. The stock solution was prepared by dissolving accurately weighted MB in tap water. The use of tap water rather than deionised water was motivated by the fact that tap water is closer to natural water in its chemical composition. The MB molecular formula is C 16 H 18 N 3 SCl corresponding to a molecular weight of 319.85 g. MB was chosen in this study because of its wellknown strong adsorption onto solids 70 .
Solid materials. Metallic iron (Fe 0 ). The used Fe 0 material was purchased from iPutech (Rheinfelden, Germany). The material is available as filings with a particle size between 0.3 and 2.0 mm. Its elemental composition as specified by the supplier was: C: 3.52%; Si: 2.12%; Mn: 0.93%; Cr: 0.66% while the balance was Fe. The material was used without any further pre-treatment. Fe 0 was proven as a powerful discoloration agent for MB given that discoloration agents in the form of FeCPs are progressively generated in-situ 70 .
Manganese dioxide (MnO 2 ). The tested natural MnO 2 -bearing minerals was Manganit from Ilfeld/Harz, Thüringen (Germany). The mineral was crushed and fractionated by sieving. The fraction 0.5-1.0 mm was used without any further pre-treatment. No chemical, mineralogical nor structural characterizations were performed. MnO 2 is a reactive mineral [73][74][75] and is used to delay the availability of 'free' iron corrosion products (FeCPs) in the system. This results in a delay of quantitative MB discoloration 30 .
Sand. The used sand was a commercial material for aviculture ("Papagaiensand" from RUT-Lehrte/Germany). The sand was used as received without any further pre-treatment. The particle size was between 2.0 and 4.0 mm. Sand was used as an adsorbent because of its worldwide availability and its use as admixing agent in Fe 0 barriers 50,76 . The adsorption capacity of sand for MB has been systematically documented as early as in 1955 by Mitchell et al. 77 .  where, C 0 is the initial aqueous MB concentration (ideally 10.0 mg L -1 ), while C gives the MB concentration after the experiment. The operational initial concentration (C 0 ) for each case was acquired from a triplicate control experiment without additive material (so-called blank). This procedure was to account for experimental errors during dilution of the stock solution, MB adsorption onto the walls of the reaction vessels, and all other possible side reactions during the experiments.

Results and discussion
Evidence for the complexity of the Fe 0 /MnO 2 /sand systems. Figure 2 compares the extent of MB discoloration in the six investigated systems for 7 and 47 days. Figure 2a clearly shows that, after 7 d, only MnO 2 had not significantly discolored MB (4%) while Fe 0 alone depicts the best discoloration efficiency (62%). The E values for the other systems varied between 31 and 40%. The increasing order of efficiency was: MnO 2 < sand < Fe 0 /MnO 2 < Fe 0 /MnO 2 /sand < Fe 0 /sand < Fe 0 . These results can be regarded as counter-intuitive since binary (Fe 0 /MnO 2 , Fe 0 /sand) and ternary (Fe 0 /MnO 2 /sand) performed less than Fe 0 alone. In conventional shaken or stirred batch experiments, involved processes are accelerated to the extent that achieved results are the intuitive ones observed after 47 days (Fig. 2b) 78 . Figure 2b compares the extent of MB discoloration in the six systems after 47 days. Compared to the results after 7 days, the extent of MB discoloration has increased to more than 50% in all systems, except MnO 2 alone. Based on the absolute E values, the increasing order of efficiency was: MnO 2 (4%) < sand (51%) < Fe 0 /sand It is interesting to note that Fe 0 /sand performed less than Fe 0 alone and the two MnO 2 -bearing systems. This observation alone confirms that MnO 2 -amendment enhances the efficiency of Fe 0 /H 2 O systems by "reinforcing" corrosion (Eq. 3), but only in the long-term. Thus, the complexity of the ternary system as well as the need to understand its operation model is apparent. This is achieved herein by investigating the systems for 7, 18, 25 and 47 days. This corresponds to following the fate of aqueous MB (discoloration) as the contact time increases from 7 to 47 days 59,70,71 . In particular the variation of the pH value in the systems will be discussed in detail.
Effect of the contact time on the Fe 0 /MnO 2 /sand system. Figure 3a compares the extent of MB discoloration in Fe 0 /MnO 2 /sand systems for the four tested contact times (7, 18, 25 and 47 d) and Fig. 3b depicts the corresponding changes in pH values. It is seen that the lowest extent of MB discoloration corresponds to 18 d contact time. This means that after 7 days the system performed better than after 18 d. The observation can be regarded as counter-intuitive, while the monotonous increase of the pH value (Fig. 3b) is intuitive. The investigated systems were 0 ≤ [Fe 0 ] (g L -1 ) ≤ 45, with [MnO 2 ] = 2.3 g L -1 and [sand] = 45 g L -1 . This means that [Fe 0 ] = 0.0 g L -1 corresponds to a MnO 2 /sand system or simplified to the sand system as MnO 2 has no adsorptive affinities for MB (Fig. 2). In other words, the counter-intuitive observation corresponds to the effect of MnO 2 on the Fe 0 /H 2 O system. A key feature from Fig. 3a is that there is an intuitive monotonous increase of the E value with increasing Fe 0 loading for all four tested contact times. This suggests that if the experiments were performed by different investigators, the given interpretations would have been conclusive and even convincing. The tested experimental conditions were selected based on past works 30,79 to achieved such results. In addition, most of the observations www.nature.com/scientificreports/ made by researchers in Fe 0 /H 2 O systems are just static snap-shots (mostly inaccurately measured) of processes occurring over an enormous range of time scales 80 . Following this premise, it was necessary to further vary the experimental conditions to maximize the chance to make more relevant observation 81 . One really intriguing observation is that the MB previously removed (t < 7 days) was released back to the solution at day 18 such that MB discoloration was lower even in the sand system (Fig. 3a). This MB desorption is rationalized by the pH decrease accompanying MnO 2 reductive precipitation as given in Eq. (3b). By decreasing the pH value, further adsorption onto sand is inhibited and the previously adsorbed MB is desorbed and released into solution (Fig. 3a). On the other hand, the process of Fe 0 dissolution by MnO 2 implies intensified interactions at the bottom of the assay tubes which slowed down the diffusion of MB from the bulk solution. Iron corrosion determined the extent of MB discoloration and the intuitive increase of MB discoloration with increasing Fe 0 loading is observed in all systems only after a pseudo-steady state is established in the systems or the capacity of MnO 2 is exhausted. Alyoussef 59,70,71 tested a parallel system with 4.5 g L -1 of MnO 2 and observed a larger decrease of MB discoloration for 18 days. Similar observations were made by Noubactep et al. 82 in their experiments for uranium removal. Figure 3b shows that for [Fe 0 ] > 7.5 g L -1 , the pH value monotonously increases with increasing Fe 0 loading. For [Fe 0 ] < 7.5 g L -1 , there were some fluctuations justified by the co-occurrence of Fe 0 corrosion (consuming protons-Eq. 1) and MnO 2 reductive dissolution (producing protons-Eq. 3b) to fix the pH of the systems. Again, once the oxidation capacity of MnO 2 is exhausted, iron corrosion controls the pH of the system.
The comparative evaluation of the time-dependent changes of E and pH values has clarified the operating mode of MnO 2 in enhancing the efficiency of the Fe 0 /H 2 O system without any solid phase characterization. This discussion has equally not considered the redox reactivity of MnO 2 for Fe 0 (and MB). Only the availability of "free" FeCPs was considered in the investigated single, binary and ternary aggregate systems. Achieved results corroborate the usefulness of varying several operational parameters to better understand complex dynamic systems 81,83,84 . Significance of the findings. Operating mode of remediation Fe 0 /H 2 O systems. This study has confirmed that Fe 0 in engineered filtration systems is oxidized by protons to ferrous ion (Fe 2+ ) (Eq. 1). Fe 2+ ions are partly transferred to the surface of available aggregates (e.g. MnO 2 and sand) and is oxidized further to ferric ion (Fe 3+ ) and deposited on the aggregates as hydroxides (in-situ coating) (Fig. 1). Iron oxide-coated sand is a good adsorbent for several contaminants including chromium 39,85 , pathogens 86,87 and phosphates 88,89 . Fe 0 oxidation also contributes to produce anoxic conditions which are favorable for the abiotic reductive transformation of several dissolved species including chlorinated compounds 38,90 . Unlike sand and other inert aggregates, MnO 2 is reactive and uses Fe 2+ for its reductive dissolution (Eq. 3). Because the reaction occurs at the surface of MnO 2 (Fig. 1), Fe 0 passivation is delayed until the oxidative capacity of MnO 2 is exhausted. Results presented herein have demonstrated these mechanisms excellently, while benefiting from the tracer nature of methylene blue (MB method) 30,79 . In fact, mechanistic discussions are often complicated by the need to consider the redox reactivity of both Fe 0 and MnO 2 with the contaminant of concern 72 . In other words, one major output of this research is that the popular hypothesis to rationalize reductive transformations in Fe 0 /H 2 O systems is faulty 91 . The hypothesis that Fe 0 is an electron donor for dissolved has been seriously challenged during the past 15 years, however, the questioned view is still prevailing 11,36,59 .
The stoichiometry of electrochemical reactions (similar to Eq. 2) has been routinely used to design Fe 0 remediation systems 92,93 . The evidence that twice more Fe 0 is needed to exchange the same number of electrons when reduction is induced by Fe 2+ implies that the service life of Fe 0 -based systems has been wrongly estimated 57 . The statement is valid regardless of the approach used to estimate the efficiency of the system. However, the main problem has been the failure to properly consider the expansive nature of iron corrosion, which makes only hybrid systems viable in the long term 14 .
The importance of hybrid Fe 0 /H 2 O systems. The long history of Fe 0 filtration systems teaches that only hybrid systems are sustainable. The Bischof filters, applied both for household and large-scale uses, contained a reactive zone made up of 25% sponge iron (vol/vol) mixed with gravel 41,94 . The Multi-Soil-Layering of Wakatsuki et al. 89 contained only 15% Fe 0 (w/w) (iron fillings) mixed with 15% Fe 0 (w/w) pelletized jute and balanced with zeolite (60%). The phosphate filters of Erickson et al. 95 contained only up to 5% steel wool balanced with sand. All these systems operated for more that 1 year without clogging. In the framework of subsurface permeable reactive barriers, O'Hannesin and Gillham 31 tested a reactive wall containing 22% Fe 0 balanced with gravel and reported on good hydraulic properties in the long term. Other systems with 100% Fe 0 have failed because of loss of porosity coupled with the early development of preferential flow paths in the Fe 0 permeable reactive barrier 96 . However, the availability of preferential flow paths was globally attributed to mineral precipitation (e.g. calcium carbonate, iron oxides, sulfide minerals). The key point is that iron oxides resulting from corrosion products are more abundant and universally present, and their generation should be reduced by "diluting" Fe 0 with non-expansive aggregates like gravel or sand.
All systems containing a pure Fe 0 layer (100%) were reported to be efficient but not sustainable [97][98][99] . The most prominent example is probably the use of iron filings for selenium removal from agricultural drainage water by the Harza Process 98,100 . In 1985, Harza Engineering Co. tested a pilot-scale process using iron filings in flow-through beds. The testing was discontinued because the beds quickly cemented with precipitates 100 . The study concluded that the advantage of Fe 0 filtration is to decrease Se concentration to very low concentrations. The mechanism of Se removal was further investigated and it was established that Se is not reduced by an electrochemical mechanism 98 . Furthermore, Fe 0 filters were suggested as a polishing step following microbial treatments 100 3 -and Se VI , and recently for the mitigation of pathogens (bacteria) from dairy manure. The fact that a hybrid system, initially developed for chemically reducible micro-pollutants is performing well for pathogens corroborate the idea that it suffices to sustain iron corrosion to achieve water treatment 37,[52][53][54]106 . As discussed in the "Introduction", Huang et al. 43 have not convincingly demonstrated the specificity of their hybrid system (Fe 0 /Fe 3 O 4 ). This is particularly the case in a context where Fe 0 /sand systems are already essentially more sustainable than pure Fe 0 (100%) 14,49 . The present work also confirms previous results that any additive to Fe 0 basically delay the availability of corrosion products under typical field conditions. The observed enhanced performance results from sustained iron corrosion in the whole system. The question then arises, what makes MnO 2 a specific admixing aggregate for Fe 0 filters?
The suitability of hybrid Fe 0 /MnO 2 systems. The presentation until now has demonstrated that applying Fe 0 for water treatment is promising as mixing Fe 0 with other aggregates delays passivation or sustain treatment efficiency. Moreover, substantial experiences have been accumulated on the functionality of hybrid systems for water treatment ("The importance of hybrid Fe 0 /H 2 O systems"). The knowledge that Fe 0 acts as generator of contaminant scavengers (and never as reducing agent) implies that adsorption and co-precipitation are the fundamental mechanisms of contaminant removal in Fe 0 /H 2 O systems. Hybrid systems tested as means to prevent iron passivation include amendment with granular activated carbon (GAC), magnetite (Fe 3 O 4 ), manganese oxides (MnO x ), pyrite (FeS 2 ), and sand 15 . Among these aggregates, MnO x and FeS 2 are the most chemically reactive 36,71 . Both aggregates induce a pH shift to more acidic values. However, because iron corrosion increases the pH, it is possible to find the optimal Fe 0 /FeS 2 and/or Fe 0 /MnO 2 ratio for case-specific water treatment. Therefore, long-term systematic testing with well-characterized materials is necessary.
Note that Fe 0 is a generator of iron oxides, and adding Mn oxides (MnO x ) to the system creates a very complex system, which is not new to geochemists, but which is yet to be investigated in the context of water treatment 68,72 . In fact, taken individually, the redox reactivity of these minerals plays important roles in the fate and transformation of many contaminants in natural environments 59,[61][62][63][64][70][71][72] . Available works mostly investigate simple model systems with few contaminants 68,72 . To bridge the gap between simple model systems and complex environmental systems, a profound understanding of the redox reactivity of Mn-and Fe-oxides in complex model systems toward water decontamination is urgently needed. The effects of natural ligands (Cl -, HCO 3 -, PO 4 3-, SO 4 2-) and natural organic matter (NOM) on the redox reactivity of Fe 0 /MnO 2 systems need to be investigated as well. Moreover, there is need to investigate the following: (i) fate of contaminants in Fe 0 systems, and (ii) the safe disposal of spent Fe 0 materials, including their use as filler material in novel construction materials, and the behavior of contaminants in such materials.

Concluding remarks
This study clearly delineates the important role of reactive MnO x minerals on the process of water treatment using Fe 0 -based systems. The presence of MnO x induces Fe 2+ oxidation at the mineral surface, resulting in a significant delay of Fe 0 passivation compared to that attained in Fe 0 and Fe 0 /sand systems. Being a natural mineral or a soil resource, its incorporation in Fe 0 filters reinforces the frugality of this already demonstrated affordable system. It is expected that adding MnO x to Fe 0 /H 2 O will create geochemical dynamics in the system which would sustain iron corrosion and maintain the efficiency of system for water decontamination for the long term. This would make Fe 0 filters a sustainable solution for decentralized safe drinking water provision and enable the realization of universal access to safe drinking water and even on a self-reliant manner. To bridge the existing knowledge gaps, the need for further research entailing long-term testing of Fe 0 systems was highlighted.