Metallic Iron for Environmental Remediation: The Fallacy of the Electron Efficiency Concept

The suitability of remediation systems using metallic iron (Fe0) has been extensively discussed during the past 3 decades. It has been established that aqueous Fe0 oxidative dissolution is not caused by the presence of any contaminant. Instead, the reductive transformation of contaminants is a consequence of Fe0 oxidation. Yet researchers are still maintaining that electrons from the metal body are involved in the process of contaminant reduction. According to the electron efficiency concept, electrons from Fe0 should be redistributed to: i) contaminants of concern (COCs), ii) natural reducing agents (e.g., H2O, O2), and/or iii) reducible co-contaminants (e.g. NO3-). The electron efficiency is defined as the fraction of electrons from Fe0 oxidation which is utilized for the reductive transformations of COCs. This concept is in frontal contradiction with the view that Fe0 is not directly involved in the process of contaminant reduction. This communication recalls the universality of the concept that reductive processes observed in remediation Fe0/H2O systems are mediated by primary (e.g., FeII, H/H2) and secondary (e.g., Fe3O4, green rusts) products of aqueous iron corrosion. The critical evaluation of the electron efficiency concept suggests that it should be abandoned. Instead, research efforts should be directed towards tackling the real challenges for the design of sustainable Fe0-based water treatment systems based on fundamental mechanisms of iron corrosion.


INTRODUCTION
Metallic iron (Fe 0 ), also termed as zero-valent iron (ZVI) is widely considered as a cost-effective reducing agent for organic pollutants in groundwater (Henderson and Demond 2007;Guan et al., 2015;Cao et al., 2020;He et al., 2020). The Fe0-based permeable reactive barrier (PRB) technology for groundwater remediation is rooted on this premise (Gillham 2008;Chen et al., 2019;Xiao et al., 2020a;Xiao et al., 2020b;He et al., 2020;Wang et al., 2022). Fe 0 has also been successfully used for the removal of various inorganic contaminants (e.g. As, NO 3− ) and pathogens (e.g., bacteria, viruses) from polluted waters (Richardson and Nicklow 2002;Henderson and Demond 2007;Gheju 2011;Guan et al., 2015;Cao et al., 2020;Kim et al., 2021;Noubactep, 2021). However, these applications are mainly perceived to be derived from the Fe 0 PRB technology for organic pollutants (Obiri-Nyarko et al., 2014;Naseri et al., 2017). The concept that Fe 0 is an electron donor under environmental conditions has never been experimentally established (Warren et al., 1995;Farrell et al., 2001;Lavine et al., 2001;Jiao et al., 2009;Naseri et al., 2017;Cao et al., 2020;Hu et al., 2020;Noubactep, 2022). For example, while investigating the reductive dechlorination of carbon tetrachloride (CT) in Fe 0 /H 2 O systems, Jiao et al. (2009) clearly demonstrated that reducing electrons are not from Fe 0 , although iron corrosion was helpful for CT reductive dechlorination. Their conclusions read as: "The inherent relationship between the dechlorination of CT and the corrosion of iron is attributed to the fact that the adsorbed hydrogen atoms produced during the iron corrosion process are necessary for the dechlorination process of CT." Table 1 summarizes some key arguments presented in the broad scientific literature prior to the advent of the recent Fe 0 remediation technology, and disproving the reductive transformation concept.
The idea that organic pollutants are reductively transformed by Fe 0 was introduced in the scientific literature by Reynolds et al. (1990). Scientists from the University of Waterloo (Canada) were investigating the potential for sampling bias caused by sorption of chlorinated organic species to materials commonly used in groundwater sampling (Lee et al., 2004;Gillham 2008;Cao et al., 2020). Their results revealed losses of chlorinated organic contaminants from water samples in contact with Fe 0based vessels. Hence, reductive dechlorination was proposed as the most likely reaction path (Reynolds et al., 1990;Gillham 2008). This observation coincided with a period when geochemists were looking for suitable materials for the realization of the concept of groundwater remediation using PRBs as introduced in the 1980s by McMurty and Elton (1985). In other words, Fe 0 was considered a reducing agent (or an electron donor) for organic pollutants, because their reductive transformation was observed in its presence, in the Fe 0 /H 2 O system. This coincidence was misinterpreted as a scientific fact and still prevails (Cao et al., 2020;Thakur et al., 2020;Noubactep, 2022). Investigations by Matheson and Tratnyek (1994) and Weber (1996) have been reported to confirm these observations. Moreover, it was claimed that the observation that organic pollutants can be reduced in Fe0/H2O systems was novel (Matheson and Tratnyek, 1994;Gillham 2008). Unfortunately, the then available seminal works of Khudenko (1985), Khudenko (1987), and Khudenko (1991) frontally contradict the claimed novelty (Cao et al., 2020). In particular, in the paper entitled, "Feasibility evaluation of a novel method for destruction of organics" (Khudenko, 1991), Boris Michael Khudenko demonstrated that Cu 2+ cementation by Fe 0 can be used to induce the reductive degradation of organic pollutants ( Table 2). Clearly, Fe 0 is oxidized by Cu 2+ and reaction products (Fe II and H/H 2 species) act as reducing agents for the (organic) contaminants of concern (COCs). In other words, Fe II species resulting from iron corrosion are used for the "destruction of organics." Factually, H 2 and H species also resulting from iron corrosion are reducing agents as demonstrated by Jiao et al. (2009). Table 2 summarizes the experimental conditions and results of Khudenko (1991) with regard to the Fe 0 /H 2 O system. In Khudenko (1991)'s work, an aqueous solution of Direct Yellow 12 was acidified to an initial pH value of 4.5 using H2SO4. Two parallel experiments were performed differing in the addition or non-addition of 100 mg L-1 Cu 2+ (CuSO 4 ). Used Fe 0 was a 1 m long iron wire of 0.2 mm in diameter. The filtrate was stirred by a magnetic stirrer. Results demonstrated that with Cu 2+ addition, complete discoloration was achieved within 2.5 min, while the system without Cu 2+ could not be discolored at all after 60 min of stirring. More Fe 0 was consumed in the absence of Cu 2+ than with Cu 2+ . The second experiment was conducted with a wastewater from finishing operations of a textile mill. The wastewater had a rosy color, and an initial pH value of 8.0. Addition of Cu 2+ alone changed the color to dark blue suggesting formation of Cu 2+ complex. Cementation with Fe 0 and Cu 2+ resulted in 98.5% color removal, while the process without copper resulted in just 10% color removal ( Table 2).
It is surprising that 3 years after the brilliant concept of Khudenko (1991) and its validation using both synthetic and real wastewater, Matheson and Tratnyek (1994) introduced the contradiction without proving Khudenko (1991) wrong. In fact, Matheson and Tratnyek (1994) did not even consider that earlier work by Khudenko (1991). Bigg and Judd, 2000 were the first scientists to cite Khudenko (1991), 2 years after the report that the concept that, Fe 0 is a reducing agent was a "broad consensus" (O'Hannesin and Gillham, 1998). It is also surprising that the view of Matheson and Tratnyek (1994) was claimed to be experimentally validated (Roberts et al., 1996;Weber, 1996) and is still favored by the majority of active researchers on the remediation Fe 0 /H 2 O system (Naseri et al., 2017;Xiao et al., 2020a;Xiao et al., 2020b;Cao et al., 2020;Hu et al., 2020;Hu et al., 2021;Noubactep, 2022). Accordingly, the whole mechanistic discussion is based on the idea that there is some electron transfer from the Fe 0 bulk material to COCs, which are potentially transformed into non-toxic or less toxic species (He et al., 2020). The extent of this reaction was conventionally evaluated using the reaction rate constant k (kobs or kSA) (Johnson et al., 1996;McGeough et al., 2007). However, according to Liu et al. (2013) this is inappropriate because the extent of iron corrosion and the proportion of electrons used for the transformation of COCs should be considered to assess the economics of the system. In other words, the goal was to avoid superfluous Fe 0 dosages which impede the economics of the designed systems (Wu et al., 2014;Shufen et al., 2018, He et al., 2020. During the past 8 years, an important number of papers has been published on the suitability of the electron efficiency concept (EE concept). He et al. (2020) give an excellent overview on the topic, and, interested readers are referred to this very recent review article. To this point, the presentation has highlighted that the EE concept is intrinsically wrong, because Fe 0 does not play any significant role in the process of contaminant reductive transformation in Fe 0 /H 2 O systems (Whitney, 1903;Jiao et al., 2009;Noubactep, 2022). In fact, while Jiao et al. (2009) have proven Matheson and Tratnyek (1994) wrong, Whitney (1903) had already established that under environmental conditions, Fe 0 is oxidized only by protons (H+), even in the presence of dissolved oxygen (O 2 ) and carbonic acid (H 2 CO 3 ) ( Table 1). The present communication aims at demonstrating the fallacy of the EE concept in order to avoid its further propagation. The presentation starts with chemistry of the Fe 0 /H 2 O system, followed by a historical overview on water treatment using Fe 0 , and ends with a critical evaluation of the usefulness of the EE concept.

THE CHEMISTRY OF THE FE0/H2O SYSTEM
Aqueous iron corrosion is an electrochemical process which needs four compartments to occur: an anode, a cathode, a conductor and an electrolyte (Landolt, 2007;Groysman, 2010). In the remediation Fe 0 /H 2 O system, the conductor is the metal body (Fe 0 ), the electrolyte is the polluted water, the anode is an area of the Fe 0 surface where oxidative dissolution occurs (releasing ferrous iron-Fe 2+ ), and the cathode is an area of the Fe 0 surface where electrons left behind by Fe 2+ are transferred to a reducible species.
A key feature of this process is that, the reactions at the anode and the cathode occur simultaneously, and the prerequisite is that the electrolyte must be in contact with both the anode and the cathode (Landolt, 2007;Noubactep, 2014;Noubactep, 2016).
Aqueous iron corrosion proceeds as follows: i) Fe 0 is oxidatively dissolved at the anode to release Fe 2+ , ii) the generated Fe 2+ ions migrate in the polluted water (electrolyte), and iii) then electrons left behind by Fe 2+ are transferred through the metal body (conductor) to a reducible species at the cathode. It is crucial to underline that electrons are transported from the anode to the cathode by Fe 0 . Electron transfer to any adsorbed species only occurs if there is no conduction barrier at the Fe 0 surface (Landolt, 2007;Nesic, 2007;Lazzari, 2008;Groysman, 2010). It is well-known that, at pH > 4.5, an oxide scale forms on the Fe 0 surface and shields it from dissolved species, including  Khudenko (1991 (Stratmann and Mü ller, 1994;Lazzari, 2008). For the EE concept to be applicable, it means that the universal oxide scale on Fe 0 should be electronically conductive, which is not the case in Fe 0 /H 2 O systems (Noubactep 2007;Noubactep, 2008;Noubactep, 2014;Noubactep, 2016). Figure 1 summarized the pathways of contaminant removal and transformations in Fe 0 / H 2 O systems. The electrochemical reaction for aqueous iron corrosion is depicted in Eq. 1: Eq. 1 shows that iron corrosion consumes protons, thereby increasing the pH value. This means that adding protons (acidification) is a powerful tool to intensify iron corrosion where it is needed, for example for H 2 evolution (Ndé -Tchoup é et al., 2020). However, for environmental remediation, a pH shift to lower values is not typically envisaged such that a pH increase occurs as a rule (Lipczynska-Kochany et al., 1994;Matheson and Tratnyek, 1994;Schreier and Reinhard, 1994;Hu et al., 2021). This pH increase favours the formation of ferrous hydroxides [Fe(OH) 2 ], which polymerize and precipitate at the surface of Fe 0 or in its vicinity (Eq. 2). When dissolved oxygen (O 2 ) is present, ferric hydroxides [Fe(OH) 3 ] are formed as well. In the real world, whether the conditions are anoxic or oxic, iron corrosion generates an oxide scale which permanently shields its surface and is made up of several oxides and hydroxides (Odziemkowski and Simpraga, 2004). The process of oxide scale formation and transformation is a dynamic one (Odziemkowski et al., 1998;Sikora and Macdonald, 2000;Nesic, 2007;Lazzari, 2008;Groysman, 2010). It is certain that no electronically conductive oxide scale can be formed at the Fe 0 surface under environmental conditions (Nesic, 2007;Lazzari, 2008). In other words, the oxide scale, acting as diffusion barrier for contaminants and dissolved O 2 , also represents a conductive barrier for electrons from Fe 0 . For this reason, electrons from the metal body cannot (quantitatively) reduce any initially dissolved species (contaminants and O 2 ) ( Figure 1) (Stratmann and Mü ller, 1994;Noubactep 2007;Noubactep, 2008;Jiao et al., 2009;Noubactep, 2013;Noubactep, 2015;Noubactep, 2019;Hu et al., 2020). Clearly, the EE concept is built on a thinking mistake. Accordingly, scientists propagating this concept are justifying their own mistakes by citing past mistakes. The EE concept would have been valid to some extent, if the oxide scale was not present. That is under acidic conditions (pH < 4.5) which is not the pH range of environmental remediation (Gillham 2008;Ghauch, 2015). The barrier nature of the oxide film implies that all reductive transformations are mediated by corrosion products (e.g., Fe II , H/H 2 , Fe 3 O 4 , green rust). However, contaminant removal is mediated by adsorption onto and co-precipitation with solid iron corrosion products (FeCPs). In fixed beds, size-exclusion is the other relevant removal mechanism (Noubactep, 2007, Noubactep, 2008. The Fe 0 research community is reminded of a famous quote by Mahatma Gandhi "An error does not become truth by reason of multiplied propagation, nor does truth become error because nobody sees it." Contextualizing this to the Fe 0 literature, the highlighted mistake (error) has been propagated since the time the work of Khudenko (1991) was ignored. Despite several efforts by our group pointing out this conceptual mistake (Noubactep 2007, Noubactep, 2008Noubactep, 2010a;Noubactep, 2010b;Noubactep, 2014;Noubactep, 2016), the truth has been ignored for 1 decade already (Xiao et al., 2020a;Xiao et al., 2020b;Cao et al., 2020;Cao et al., 2021a;Cao et al., 2021b;Cao et al., 2021c;Cao et al., 2021d;Hu et al., 2021;Noubactep, 2022). However, an accurate fundamental understanding of processes governing contaminant removal is critical in the design and operation of Fe 0 -based systems. This is particularly important given that Fe 0 based remediation systems have wide practical applications. These applications have been discussed in earlier papers (Naseri et al., 2017;Antia, 2020, Huang et al., 2021a, thus, a detailed review is beyond the scope of the present paper. In summary, typical applications of Fe0-based remediation systems documented in literature include: i) decentralized safe drinking water provision in low-income settings (Huang et al., 2021b;Mueller et al., 2021), ii) industrial wastewater treatment systems Kulkarni et al., 2020), iii) recovery of heavy metals from industrial effluents (Vollprecht et al., 2018;Calabrò et al., 2021;Noubactep, 2021), iv) urban stormwater treatment (Rahman et al., 2013;Tian et al., 2019), v) treatment of drainage water from agroecosystems (Das et al., 2017;Lanet et al., 2021), vi) subsurface permeable reactive barriers (PRBs) for remediation of contaminated groundwater (Thakur et al., 2020;Njaramba et al., 2021, Wang et al., 2022, and vii) treatment of domestic wastewater (Wakatsuki et al., 1993;Latrach et al., 2018).
The next section gives some selected examples on how the past decade has ignored available knowledge.
October 2021 | Volume 2 | Article 677813 Bischof (1873) gives an overview of the ancient efforts for the design of household water filters culminating in the design of spongy iron filters. In 1881, spongy iron filters were successfully pilot tested in Antwerp (Belgium), and could enable the supply of drinking water to 200,000 inhabitants for 18 months without any maintenance Antia, 2020;Cao et al., 2020). Spongy iron filters were used to treat yellow-colored water contaminated with pathogens (Devonshire, 1890;Antia, 2020). The yellow coloration reveals that organic substances were present in water. Fe 0 was said to be oxidized by dissolved O 2 , and contaminants were removed by adsorption and occlusion (co-precipitation). In other words, contaminants were removed by mechanisms similar to that in the flocculation process, with the only difference that flocs were not formed in a bulk solution but in the vicinity of individual Fe 0 grains ( Figure 1) (Bojic et al., 2009). The design of Fe 0 filters were not based on the stoichiometry of any chemical reaction, but on the probability to generate enough "flocs" on time to address the extent of water contamination.
Household and community drinking water treatment systems working on the same principles were independently designed later (Antia, 2020). Available designs include; i) the Anderson Process (Devonshire, 1890;van Craenenbroeck, 1998), ii) the Emmons Process (Lauderdale and Emmons, 1951;Lacy, 1952), iii) the SONO arsenic filters (Hussam and Munir, 2007;Neumann et al., 2013) and iv) the IITB (Indian Institute of Technology Bombay, India) arsenic filters (Chaudhari et al., 2014;Banerji and Chaudhari, 2017). It is important to point out that, in all these designs, iron corrosion products (FeCPs) for contaminant scavenging are generated without any addition of chemicals. Designs using common oxidants (e.g. H 2 O 2 , O 3 ) to intensify iron corrosion were also presented (e.g., Gottinger et al., 2013;Kowalski and Sø gaard, 2014;Tepong-Tsindé et al., 2015). Evidently, these findings show that the formation of flocs or FeCPs is critical for contaminant removal, and reaction stoichiometry is not critical in the design and operation of these filters.
The presentation until now shows that Fe 0 based remediation systems have been designed and mostly satisfactorily operated for the past 170 years. During the time before the advent of the PRB technology (1990), efforts to rationalize the efficiency of the systems using reaction stoichiometry were scarce. For example, Gould (1982) found that, in their systems, more Cr VI was reduced than predicted by the cementation process (electrons from Fe 0 ). The same author speculated that some Cr VI was reduced by insitu generated H 2 and pointed out that Fe II species also reduce Cr VI (Gheju, 2018).

Fe 0 for PRBs
Research on Fe 0 PRBs started around 1990 (Gillham 2008). In 1994, the four first peer-reviewed articles were published by four research groups in three different journals (Gillham and O' Hannesin, 1994;Lipczynska-Kochany et al., 1994;Matheson and Tratnyek, 1994;Schreier and Reinhard, 1994) (Supplementary Table S1). Supplementary Table S1 summarizes their current bibliometric evaluation according to SCOPUS Web of knowledge and Table 3 summarizes their experimental conditions. This section demonstrates that the four papers have already revealed all important issues for the discussion of the invalidity of the EE concept. In particular, Schreier and Reinhard (1994)  The common feature of the four papers is that they were investigating the reductive degradation of chlorinatedhydrocarbons in the presence of Fe 0 . It was clear to all of the investigators that FeCPs will form, and influence the reductive process. Among other results, Schreier and Reinhard (1994) observed a lag time in the process of contaminant reduction in Fe 0 /H 2 O systems. In biological systems, for example, a lag time is the time necessary to induce the synthesis of appropriate enzymes (Schreier and Reinhard, 1994). Such lag periods were repeatedly reported in the literature over the years (e.g., Huang et al., 1998;Noubactep et al., 2003;Hao et al., 2005;Cao et al., 2021a;Cao et al., 2021b;Cao et al., 2021c;Cao et al., 2021d). In Fe 0 /H 2 O systems, the lag time is indicative of the time required for the generation of FeCPs following Fe 0 immersion into a polluted water. Proponents of the reductive transformation concept, and the EE concept, often do not provide a rational reason for the lag time.
The currently well-accepted model for the lag time was summarized in a recent review article (Sun et al., 2016). It stipulates that, the lag time corresponds to the time to transform the outer Fe 2 O 3 layer, an air-formed oxide scale into Fe 3 O 4 . In fact, commercial Fe 0 specimens used in water treatment are covered by a pre-existing oxide scale consisting of an inner Fe 3 O 4 layer and an outer Fe 2 O 3 layer. The inner layer is electronically conductive by virtue of the semi-conductive nature of Fe 3 O 4 (band gap: 0.11 eV) (Odziemkowski and Simpraga, 2004;Huang and Zhang, 2005). However, electron transport is hindered by the outer non-conductive Fe 2 O 3 layer. According to this model, contaminant reductive transformation starts when the reduction of the Fe 2 O 3 layer is completed.
The crucial role of in-situ generated FeCPs in adsorbing contaminants (Charlet et al., 1998;Furukawa et al., 2002) and their role in mediating contaminant reduction was demonstrated (White and Peterson, 1996) (Figure 1) (Jiao et al., 2009;Noubactep, 2010b;Noubactep, 2011;Cao et al., 2021a;Cao et al., 2021b). Another important observation of Schreier and Reinhard (1994) was that in experiments conducted at room temperature with Fe 0 , reductive transformations ceased abruptly after some 28 days. The Citation-values in Table 3 suggest that Lipczynska-Kochany et al. (1994) and Schreier and Reinhard (1994) have not continued working on the remediation Fe0/H2O system (Supplementary Table S1). The two other research groups have continuously worked on this field for at least 2 decades (Jeen et al., 2013;Fan et al., 2017). This is a plausible explanation for the fact that the research community could have widely accepted the mistake that contaminant removal is an electrochemical reaction represented by Eq. 3: Table 3 summarized some relevant experimental conditions of the first four peer-reviewed articles on the remediation Fe 0 /H 2 O system. A total of 14 chlorinated hydrocarbons (RCl) were tested. Carbon tetrachloride (CT) was tested in 3 works. The tested RCl initial concentrations varied widely. It is seen that different experimental procedures were employed. While Fe 0 powder (20 and 100 mesh) was used in individual works, there were huge differences in the used mass loading (17-250 g L −1 ), the nature and the volume of the reaction vessels, the mixing intensities (2-175 rpm), the availability of dissolved O 2 , and the experimental duration (1 h-75 days). It has been clearly demonstrated that these differences in the experimental designs are responsible for reported discrepancies (Devlin and Allin, 2005;Henderson and Demond 2007;Gheju 2011;Ghauch, 2015;Guan et al., 2015). Efforts towards more reliable experimental conditions were discussed and constantly actualized for example from 1999 to 2011 by The Interstate Technology and Regulatory Council (www.itrcweb.org) (ITRC, 2011). However, the key factor that the formation of oxide scales in the vicinity of Fe 0 should be favored has received little attention (Devlin and Allin, 2005;Noubactep, 2008;Cao et al., 2021b). This premise implies that only quiescent or very slow-mixed batch experiments would produce results relevant for the design of filters (Noubactep et al., 2009). As concerning column experiments, low-flow-rate conditions should be observed (Colabro et al., 2021). These conditions enable the Fe 2+ concentration to increase in the vicinity of Fe 0 and eventually attain the saturation concentration. In batch experiments, beside quiescent conditions, limited testing volume favor a rapid formation of oxide scales on Fe 0 . The oxide scale ultimately exerts rate control through the control of reactant diffusion transport kinetics (Devlin and Allin, 2005;Noubactep, 2008;Cao et al., 2020;Noubactep 2020;Cao et al., 2021a).
The presentation until now shows that a great deal of work has been done under test conditions that do not permit formation of the universal oxide scale on Fe 0 . Accordingly, reproducible results could be obtained but they are by no means relevant for field conditions. These obvious problems of relevance of results has also favor the introduction of the EE concept which is vehemently refuted herein.

FALSIFYING THE REDUCTIVE TRANSFORMATION CONCEPT
The present work intends to disprove the view that Fe 0 is a reducing agent under environmental conditions. This section revisits Matheson and Tratnyek (1994) in the perspective of assessing whether or not their conclusions were supported by any methodical approach. The clear experimental observation is that there is reductive transformation of chlorinated methanes (RCl) in Fe 0 /H 2 O systems (Reynolds et al., 1990;Gillham and O' Hannesin, 1994;Matheson and Tratnyek, 1994). The question is whether Matheson and Tratnyek (1994) have given a molecular level understanding of this process or established the reaction mechanism. This information is very important for the design of sustainable Fe 0 /H 2 O systems and was even the objective of their investigations (Matheson and Tratnyek, 1994). The history of science teaches that the most powerful advances in surface phenomena (including catalysis) are those that improve the ability to predict the efficiency of engineered systems (Buskirk and Baradaran, 2009). As recently recalled by Scott (2019), the fundamental understanding of the mechanisms of chemical reactions broadens the range of accurate predictions.
Any postulated reaction mechanism is a working hypothesis, whose predictions must be compared with experimental observations (Scott, 2019). Following this principle, Matheson and Tratnyek (1994) considered three possible pathways to justify the RCl 3) which is equivalent to Fe 0 oxidation by RCl (RCl oxidizing agent, electrochemical mechanism) (Pathway 1), ii) reductive dehalogenation by Fe 2+ (Eq. 4) which is equivalent to Fe 2+ oxidation by RCl (chemical mechanism) (Pathway 2), and iii) reductive dehalogenation by H + (Eq. 5) which is equivalent to H + oxidation by RCl (chemical mechanism) (Pathway 3). These three mechanisms are all consistent with the named experimental observation. However, the reaction stoichiometry and the corrosion rate are not known, making the discussion with the spectral signatures of reactants, intermediates and products highly speculative. A profound analysis by Lee et al. (2004) revealed that "no carbon balances between reactants and products have ever been successfully done for many chlorinated hydrocarbons, which indicates that reduction pathways of metalmediated reactions are not fully understood yet." The statement by Lee et al. (2004) is just another hint for the lack of concrete evidence on the reductive transformation concept. The lack of mass-balance disproves the concept in the sense that reduction is not necessarily quantitative as the missing fraction of carbon is rather enmeshed in the matrix of iron corrosion products (co-precipitation) (Eusterhues et al., 2011;Noubactep, 2011). However, in mechanistic discussions, coprecipitation is mostly considered relevant for metallic species (Henderdon and Demond 2007;Colabro et al., 2021). The lack of mass-balance collectively questions the validity of the three postulated mechanisms.
The most important feature from Matheson and Tratnyek (1994) is the approach they used to rule out reduction after Pathway 2 and Pathway 3: Uncatalyzed reduction by dissolved H 2 or Fe 2+ . They performed control experiments with H 2 -saturated water and a 100 mg L −1 FeCl 2 over 15 days in the absence of Fe 0 and could not observe any dehalogenation. They acknowledged that it was "difficult to exclude the possibility that adsorbed Fe 2+ or nascent hydrogen" from Eq. 1 "may be participating in the dehalogenation reaction." Additionally, the amendment of Fe 0 / H 2 O systems with external Fe 2+ or H 2 did not impact RCl reduction.
Finally, the addition of 0.5 mM ethylenediaminetetraacetic acid (EDTA) had no effect on the RCl dehalogenation rate. EDTA was supposed to fix Fe 2+ , avoid hydoxide precipitation and keep the Fe 0 surface free for RCl electrochemical reduction. Matheson and Tratnyek (1994) concluded on the basis of the presented experiments, that "reductive dehalogenation directly coupled with oxidative dissolution of the metal" (Pathway 1) was the "dominant process." It is very important to note that Matheson and Tratnyek (1994) have just initiated a discussion, but just 2 years later, Weber (1996) claimed to have confirmed the electrochemical nature of contaminant reduction in Fe 0 /H 2 O systems. This section seeks to convince the reader that this concept was not established by any scientific approach.
In the absence of mass balance, Matheson and Tratnyek (1994) could not establish the mechanism of RCl reduction. However, there are two more intriguing facts: i) Fe 0 oxidation by water is not discussed, and ii) water is just considered as a proton source for reaction after Eq. 3. The presentation of the authors textually reads, alkyl halides "can also be reduced by iron.
In the presence of a proton donor like water, they typically undergo reductive dehalogenation." This means that the seminal work of Whitney (1903) was ignored as well as thousands of works describing corrosion as resulting from the presence of water, including impurities in natural oil (Brondel et al., 1994) and atmospheric humidity (Stratmann and Müller, 1994). Moreover, the is no iron corrosion in dry (H 2 O free) chlorinated solvents (Rhodes and Carty, 1925;Archer, 1979). Fe 0 corrosion by H 2 O (including moisture) and no Fe 0 corrosion by dry RCl clearly indicates that more attention should have been paid to water as corroding agent in the Fe 0 remediation literature (Ghauch, 2015;Cao et al., 2021b;Noubactep, 2022).
The extent to which it is possible to confirm a reaction mechanism is an issue that has preoccupied researchers in the chemical sciences for many decades (Brenner, 2010;Scott, 2019). In Geology, it is recommended to generate and test multiple working hypotheses in scientific inquiry to guard against drawing premature conclusions (Chamberlin, 1890). The science philosophers Karl Popper and Thomas Kuhn further asserted that scientific hypothesis must be falsifiable, or refutable (Scott, 2019). Following this approach, experiments should be designed to test the viability of multiple proposed reaction mechanisms. A mechanistic hypothesis can be falsified, resulting in its modification or even abandonment (Scott, 2019). Because the reaction mechanism of Matheson and Tratnyek (1994) was falsified by Jiao et al. (2009) it should be abandoned. Moreover, because their concept has not properly considered the redox reactivity of water (E 0 0.00 V), the concept was false at the introduction.
This section calls authors to exercise vigilance to avoid making claims that a proof exists where it does not. To enable progress in the design of sustainable Fe 0 /H 2 O systems, attempts should be made to discredit rather than prove available mechanistic concepts (Scott, 2019). Louis Pasteur once formulated the following advice: "When you believe you have found an important scientific fact, and are feverishly curious to publish it, constrain yourself for days, weeks, years sometimes, fight yourself, try and ruin your own experiments, and only proclaim your discovery after having exhausted all contrary hypotheses" (Scott, 2019).

THE ELECTRON EFFICIENCY CONCEPT: REDOX AND MASS BALANCE PERSPECTIVES
Electron efficiency is defined as the fraction of total electrons from Fe 0 that are used in the reduction of COCs (Eq. 3). The presentation until now has demonstrated that no single electron from Fe 0 can be transferred to COCs because of the presence of the universal oxide scale which is never electronically conductive ( Table 4). If electrons from Fe 0 were transferred to any COC, there would have not been a lag time between the start of the experiment and the start of reductive transformation of COCs (Schreier and Reinhard, 1994;Huang et al., 1998;Hao et al., 2005; Frontiers in Environmental Chemistry | www.frontiersin.org October 2021 | Volume 2 | Article 677813 Cao et al., 2021d). Consequently, COCs, O 2 and co-contaminants are reduced by Fe II , H 2 , Fe 3 O 4 , green rust, and other reducing species generated in the Fe 0 /H 2 O system ( Figure 2). For simplification, it can be assumed that reductive transformations are mainly mediated by Fe 2+ and H 2 from Eq. 1. Eq. 1 implies that the oxidation of 1 mole of Fe 0 produces 1 mole of Fe 2+ and 1 mole of H 2 . Fe 2+ can donate one electron and H 2 two electrons. This means that the electrochemical oxidation of 1 mole of Fe 0 by water (H + or H 2 O-the solvent), indirectly produces 3 moles of electrons for the reduction of O 2 , NO 3 − and all other oxidizing agents. In other words, the EE concept has better considered how the 3 moles of electrons are distributed. In this laborious effort, the mass balance of all involved species must be performed, starting with that of iron, which implies the need for experiments entailing the controlled dissolution of FeCPs. To the best of the authors' knowledge, such a work has not been published. It is very strange, that mechanistic discussions have been performed without complete mass balance for 3 decades (Lavine et al., 2001;Lee et al., 2004;Noubactep, 2011;Cao et al., 2021c). Yet mass balance analysis is critical for proving the validity of the EE concept.
Another important feature about the EE concept is the way it addresses the electrode redox potential. The electrode potential of the Fe II /Fe 0 redox couple (E 0 -0.44 V) is not relevant to explain the reduction potential of the system towards individual contaminants. Instead, the potential of the couples Fe III /Fe II and H + /H 2 , both in adsorbed and dissolved states are considered. For Fe II , it has been demonstrated that, while dissolved Fe II is far less powerful than Fe 0 , adsorbed Fe II , also referred to as structural Fe II is sometimes more powerful than Fe 0 : −0.65 ≤ E 0 (V) ≤ −0.34 (White and Peterson, 1996). This last argument demonstrates that reduction of COCs is favourable, but this reduction of COCs is a chemical process as demonstrated for example by Khudenko (1991) and Jiao et al. (2009).
The last important feature about the EE concept is its usefulness. Reactivity loss and permeability loss are the two main challenges of the Fe 0 remediation technology (Henderson and Demond 2007;Ghauch, 2015;Guan et al., 2015;Cao et al., 2020). Avoiding material wastage is certainly a noble goal, but it should start when the intrinsic reactivity of each relevant material is documented, and its long-term changes characterized (Ndé -Tchoup é et al., 2020). This presents a challenge, because the current situation is that Fe 0 materials are not characterized, and their long term behaviour has not been really investigated Lufingo et al., 2019;Ndé -Tchoup é et al., 2020). For example, steel wool (d ≤ 90 μm), iron wire (d 200 μm) and iron nails (d > 200 μm) are just tested as "Fe 0 materials" in independent researches (Tepong-Tsindé et al., 2019).
In summary, evidence shows that: i) the EE concept fails to account for the lag time widely reported in Fe 0 /H 2 O systems, ii) the application of the EE concept is not based on any iron mass balance, iii) the EE concept disregards the decade old doubts on 4 | Summary of the key factors affecting changes in Fe 0 /H 2 O systems and their implication for the validity of the electron efficiency concept (EE concept).

Process in Fe 0 /H 2 O Remarks Description
Anodic Fe 0 dissolution Driven by the electrode potential of the redox couple Fe II /Fe 0 : E 0 -0.44 V All Fe 0 materials have virtually the same reactivity Fe 2+ migration away from the Fe 0 surface Facilitated by the ionic conduction of the polluted water Saline water is more corrosive than fresh water Transport of electrons to cathodic sites Driven by the electronic conduction of used Fe 0 , and depends on its intrinsic reactivity Fe 0 /S better than Fe 0 Electron transfer at cathodic sites Driven by the difference of the electrode potential relative to E 0 -0.44 V O 2 (E 0 0.84 V) reacts better than H 2 O (E 0 0.00 V) Formation of an iron oxide scale on Fe 0 The oxide scale acts as a barrier for electron transfer. The oxide scale also acts as a diffusion barrier for dissolved species Only water access the Fe 0 surface. The EE concept is not valid. the reducing nature of Fe 0 , and iv) the EE concept does not contribute to solve the two major design issues for Fe 0 /H 2 O systems (i.e., reactive loss and permeability loss).

REVISITING REACTIVITY LOSS AND PERMEABILITY LOSS
The two major problems of Fe 0 -based filters as documented from laboratory experiments, large-scale experiments and field implementations are reactivity loss and permeability loss (Guan et al., 2015;Cao et al., 2020). This corresponds to the state-of-the-art knowledge as summarized 14 years ago by Henderson and Demond (2007). Since then, the research community is divided in two schools. Some few researchers have realized that the dynamic nature of the Fe 0 /H 2 O system implies a different approach of investigation (Ghauch, 2015) while the large majority is still considering that isolated instrumental characterizations would help to understand the system. Clearly, the still praised monitoring of physical changes of Fe 0 features using sophisticated tools including X-ray microcomputer tomography have not mediated the expected understanding of the technology as a whole. In other words, changes of Fe 0 surface, Fe 0 particle size, nature of expanding FeCPs, the reducing porosity can be experimentally documented. However, all these observations are collectively only static snap-shots and their measurements are inaccurate. Therefore, they cannot enable the generation of non-trivial models of the dynamic processes within Fe 0 /H 2 O systems (Santisukkasaem and Das, 2019), especially as these occur over an enormous range of time scales (from few seconds to some 2 decades) (Gillham 2008;Wilkin et al., 2014;Cao et al., 2020).

Reactivity Loss
The present paper posits that the consideration of wrong reactions and improper descriptors has been a major problem the Fe 0 literature. This section clarifies this key issue. A parent compound (e.g., RX) reductively transformed in a Fe 0 /H 2 O system has been assumed to be transformed by electrons from the Fe 0 body according to Eq. 3 (Section Fe 0 for PRBs). A pseudofirst-order model is applied to describe the transformation of RX (Eq. 6). The natural logarithmic transformation of Eq. 6 yields a linear equation with the first-order rate constant kobs as slope (Eq. 7) (Reardon, 1995;Velimirovic et al., 2013). Usually, the mass normalized rate constants (kM, L g−1 h−1) and specific surface area normalized rate constants (kSA, L m−2 h−1) are used to account for the Fe 0 mass and/or the Fe 0 surface area in the systems where C is the concentration at any time and C 0 is the initial concentration of RX, k obs is the pseudo-first-order rate constant (h −1 ) and t (h) is the reaction time. In essence, rooting the reasoning on the available mass loading of Fe 0 particles (g L −1 ) and/or their surface area concentration (m 2 L −1 ) is a bright idea. However, the mistake made in considering Eq. 3 is not corrected.
Eq. 6 contents no single parameter related to water which clearly oxidizes Fe 0 according to the definition of the EE concept (Liu et al., 2013). However, in trying to account for the "fraction of electrons" used to reduce water, the EE concept has replicated the same mistake. Therefore, the EE concept is not able to address the problem of reactivity loss. Reactivity loss has been documented for H 2 generation (Eq. 1) in the absence of any contaminant (Reardon, 1995;Tang et al., 2017;Qin et al., 2018). In fact, reactivity loss is the "natural" time-dependent decrease of the kinetics of iron corrosion (Romanoff, 1957;Melchers and Petersen, 2018). As an intrinsic characteristic, reactivity loss can never be suppressed (Miyajima and Noubactep, 2015). In other words, reactivity loss has occurred in Fe 0 reactive barriers satisfactorily operating for decades (O'Hannesin and Gillham, 1998;Guan et al., 2015;Wilkin et al., 2019). These systems are still efficient because their designs (e.g., Fe 0 characteristics, Fe 0 amount, Fe 0 :aggregate ratio, flow velocity) are somehow favorable. The rationale for their efficiency or the failure of other systems (e.g., Morrison et al., 2006;Ngai et al., 2007) is yet to be investigated.

Permeability Loss
Permeability loss describes the time-dependent decrease of the hydraulic conductivity of Fe 0 -based filter , Domga et al., 2015, Cao et al., 2020, Antia, 2020, Njaramba et al., 2021. While considering Fe 0 as the reducing agent and corrosion by water as a side reaction, permeability loss has been mainly attributed to the precipitation of foreign species evolving for instance CO 3 2− or HCO 3− (Henderson and Demond 2007;Henderson and Demond, 2011;Santisukkasaem and Das, 2019).
However, the fact that contaminants are likely reduced by Fe II species which are transformed to Fe III ones, suggest that (at least) twice more Fe 0 is needed as suggested by the stoichiometry of Eq. 3: 1 mole of Fe 0 exchanges 2 moles of electrons and 1 mole of Fe 2+ only 1 mole of electrons. At pH > 4.5, each oxidized Fe 0 atom is transformed to an iron hydroxide or oxide which volume is at least twice larger than that of the parent metal (Caré et al., 2008;Zhao et al., 2011;Caré et al., 2013;Domga et al., 2015;Yang et al., 2021;Noubactep, 2022). In other words, before considering the contribution of any foreign species (e.g., Ca 2+ , CO 3 2− , HCO 3− ) in the process of permeability loss, the own contribution of iron corrosion should be properly discussed. The contribution of iron corrosion to permeability loss is obvious and has been firstly experimentally documented by Luo et al. (2013) using deionized water (Yang et al., 2021). Properly considering the expansive nature of iron corrosion has prompted Domga et al. (2015) to recommend only hybrid systems (e.g., Fe 0 /sand) for sustainable filters. According to Naseri et al. (2017), in testing a Fe 0 system, the main interest should be to check whether the designed system generated enough contaminant scavengers for efficient water treatment, regardless from the redox properties or even the chemical nature of the pollutants. It is recalled that Fe 0 filters also remove pathogens (You et al., 2005;Bradley et al., 2011;Tepong-Tsindé et al., 2019;Kulkarni et al., 2020) and many other species without redox properties (Richardson and Nicklow 2002;Henderson and Demond 2007;Ullah et al., 2020).

STARTING A NEW AREA IN FE 0 RESEARCH
The core problem of the Fe 0 remediation literature is that tools such as the first-order rate constants (k obs ) or the specific reaction rate first-order constants (k SA ) premised on the misconception that Fe 0 is a reducing agent have had an adverse effect on its development. The EE concept is such a questionable tool. Rather than help the Fe 0 research community to exploit the mainstream corrosion science, these inappropriate tools have isolated the community to a modern knowledge system, reasoning circularly on an avoidable mistake introduced by some few individuals nearly 3 decades ago. Yet despite several efforts highlighting the mistakes, the scientific community continues to propagate the same errors (Xiao et al., 2020a;Xiao et al., 2020b;Cao et al., 2020;Hu et al., 2020;Cao et al., 2021a;Cao et al., 2021b;Hu et al., 2021). Consequently, in terms of progress the Fe 0 remediation technology has not progressed much in the last decade, a scenario referred to as the "lost decade" (Noubactep, 2019). The Fe 0 research community cannot afford to further lose time by further perpetuating the errors of yesteryear, hence the need to root any further research on the correct mechanistic understanding of iron corrosion (Makota et al., 2017;Cao et al., 2020;Hu et al., 2021, Huang et al., 2021. Therefore, to design better systems, there is no other option besides the following: i) to come back to the historical work of Khudenko (1991) as a starting point, and ii) consider progress made by some few research groups during the last decades (Gheju, 2011;Ghauch, 2015;Noubactep, 2015). It is only on the basis of this consideration that research on designing efficient and sustainable Fe 0 will progress in the next coming decades.
Notably, abandoning the view that Fe 0 is an environmental reducing agent will not fix everything at once. It will take time to acquire relevant data pertaining to the specificity and long-term kinetics of iron corrosion as used in environmental remediation and water treatment. Further, it takes time for research funding institutions to adapt their funding practices to support a research in which pilot tests shall last for years exceeding the 2-3 years typical of most funded research project. It is particularly obvious that the common 2 or 3 years grants are not suitable for such research efforts. It is important to bear in mind that, Fe 0 remediation research is going back to its roots: Corrosion Science (Whitney, 1903;Howard, 1910;Dickerson et al., 1979;Nesic, 2007).
The change in strategic direction suggested in this communication is seemingly controversial, because it refutes the consensus of the late 1990s (Gillham and O' Hannesin, 1994). However, the consensual approach has not worked during the past 2 decades, resulting in unsatisfactory technical systems (Morrison et al., 2006;Ngai et al., 2007;Comba et al., 2011;Mueller et al., 2011;Kowalski and Sø gaard, 2014;Bretzler et al., 2020;Ogata et al., 2020;Huang et al., 2021b;Mueller et al., 2021). There is no alternative to abandoning the misleading concept for mere conveniences since the aspirations of younger scientists are yet to be met.
A further argument against the "reducing Fe 0 " is presented by Ito and colleagues (Satur et al., 2007;Tabelin et al., 2017a;Tabelin et al., 2017b;Seng et al., 2019;Parka et al., 2020). It took 1 decade to Ito and his collaborators to establish Fe 0 addition as the most promising tool to suppress pyrite oxidation under environmental conditions (Seng et al., 2019). The idea of the so called carriermicroencapsulation (CME) technique, developed during the second half of the 2000s (Satur et al., 2007), is to coat pyrite with a layer of metal hydroxide or oxide, in order to stop mineral oxidative dissolution which causes acid mine drainage (AMD) (Tabelin et al., 2017a;Tabelin et al., 2017b). Several Al, Fe and Ti complexes with organics were tested over the years but were not really specific to FeS 2 in mine tailings and waste rocks. The bright idea, based on the electrochemistry of Fe 0 (E 0 −0.44 V) and FeS 2 (E 0 0.25 V), was that, in a Fe 0 /FeS 2 mixture, Fe 0 is the anode and dissolved Fe 2+ migrates to the cathode (FeS 2 ) which is passivated upon formation of an oxide scale. Seng et al. (2019) have verified this sound hypothesis using Al 0 and Fe 0 . Interested readers are referred to the related publications. For the presentation herein, its suffices to recall that the semi-conductives properties of FeS 2 have been regarded as the reason for enhanced efficiency of Fe 0 / FeS 2 systems (Xiao et al., 2020a;Xiao et al., 2020b;Hu et al., 2021). It is obvious that thermodynamics is not all but the results of Sheng et al. (2019) are a further call to revisit the operating mode of remediation Fe 0 /H 2 O systems.
The research group of the corresponding author and some few others have been working according to a new paradigm for a decade and have already achieved some good results than can be summarized as: i) Fe 0 is not the source of electrons for the reduction of any contaminant. Tested species include organic (e.g., diclofenac, utriafol) and inorganic (e.g., Cr, Se) compounds (Gheju, 2011;Gheju, 2018;Gheju and Balcu, 2019), ii) the Fe 0 / H 2 O system is ionic selective, negatively charged species are Frontiers in Environmental Chemistry | www.frontiersin.org October 2021 | Volume 2 | Article 677813 favoured (Phukan et al., 2015;Phukan et al., 2016), iii) only hybrid filtration systems (e.g., Fe 0 /pumice, Fe 0 /sand) are sustainable (Ghauch, 2015;Noubactep, 2016), iv) each Fe 0 has a different corrosion rate which is never constant (Btatkeu et al., 2013;Lufingo et al., 2019;Hildebrant et al., 2020), and v) Fe0 filters are a special case of "corrosion in porous media" (Ndé -Tchoup é et al., 2018;Yang et al., 2021). Yang et al. (2021) have recently introduced the most really holistic attempt to predict the service life of Fe 0 filters. They considered for the first time the Faraday's law to tentatively predict the time to clogging. The limitation is that there are no reliable corrosion rates available, this limitation is common to all branches of science investigating corrosion in porous media, including corrosion of reinforcing steel in concrete (Caré et al., 2008;Zhao et al., 2011;Stefanoni et al., 2018;Stefanoni et al., 2019). Therefore, long-term data for iron corrosion are needed (Noubactep, 2022).

CONCLUDING REMARKS
Fe 0 /H 2 O systems have a long history of application in environmental remediation and water treating, and several new applications are emerging. However, the Fe 0 technology suffers from a number of misconceptions and mistakes pertaining to the mechanisms of contaminant removal. In the current communication, evidence was presented highlighting the mistakes of the electron efficiency concept, based on the misunderstanding that contaminant removal occurs according to reaction stoichiometry involving Fe 0 as electron donor (electrochemical mechanism). A number of lines of evidence were presented to demonstrate the invalidity of the EE concept, including: i) the lag time widely reported in Fe 0 /H 2 O systems, ii) the universal formation of a non-conductive oxide layer that acts as a barrier for electron transfer, and iii) lack of mass balance data confirming its validity. Despite these limitations, and several studies demonstrating more plausible mechanisms based on iron corrosion, the bulk of the Fe 0 research community continues to propagate the k SA , k obs and electron efficiency concepts. As a result, Fe 0 research has been nearly stagnant for the past 2 decades, a scenario that has constrained the development of robust and efficient Fe 0 /H 2 O systems. Hence, this communication highlighted the need for a paradigm shift from flawed historical concepts, to a future where the design of the next generation of Fe 0 is rooted on sound fundamental principles based on the science of iron corrosion. The suggested approach would enable an improved exploitation of the capacity of Fe 0 technology beyond it actual stand as summarized by Antia (2020). To achieve this, a critical evidence based on the reactivity and long-term kinetics of Fe 0 is needed, and research to generate such evidence requires longterm funding commitments.

AUTHOR CONTRIBUTIONS
Conception (