Nanoscale Metallic Iron for Environmental Remediation: Prospects and Limitations

Abstract

The amendment of the subsurface with nanoscale metallic iron particles (nano-Fe⁰) has been discussed in the literature as an efficient in situ technology for groundwater remediation. However, the introduction of this technology was controversial and its efficiency has never been univocally established. This unsatisfying situation has motivated this communication whose objective was a comprehensive discussion of the intrinsic reactivity of nano-Fe⁰ based on the contemporary knowledge on the mechanism of contaminant removal by Fe⁰ and a mathematical model. It is showed that due to limitations of the mass transfer of nano-Fe⁰ to contaminants, available concepts cannot explain the success of nano-Fe⁰ injection for in situ groundwater remediation. It is recommended to test the possibility of introducing nano-Fe⁰ to initiate the formation of roll-fronts which propagation would induce the reductive transformation of both dissolved and adsorbed contaminants. Within a roll-front, Fe^II from nano-Fe⁰ is the reducing agent for contaminants. Fe^II is recycled by biotic or abiotic Fe^III reduction. While the roll-front concept could explain the success of already implemented reaction zones, more research is needed for a science-based recommendation of nano-Fe⁰ for subsurface treatment by roll-fronts.

Introduction

The development of new methods and materials for environmental remediation is a real challenge for the scientific community. Such technologies will only be adopted by industry if they can exhibit marked improvements in efficiency, affordability or eco-compatibility compared with conventional techniques. The use of metallic iron (Fe⁰) in subsurface reactive permeable barriers has been identified as such a technology (Gillham and O’Hannesin 1991, 1994; Matheson and Tratnyek 1994; Blowes et al. 1997; O’Hannesin and Gillham 1998). Since this discovery almost 20 years ago, extensive research of the Fe⁰/H₂O system has been performed in an attempt to understand the controlling mechanisms behind the remediation of redox-amenable contaminant species using Fe⁰-based materials (Scherer et al. 2000; Henderson and Demond 2007; Bartzas and Komnitsas 2010; Gillham 2010; Li and Benson 2010; Noubactep 2010a, b; Noubactep and Caré 2010a; Kim et al. 2010; Comba et al. 2011; Courcelles et al. 2011; Gheju 2011; Jeen et al. 2011; Noubactep 2011a). Two different tools are commonly used to optimise the efficiency of Fe⁰ for aqueous contaminant removal: (1) reducing the particle size of Fe⁰ down to the nanoscale (nano-Fe⁰) (Wang and Zhang 1997; Ponder et al. 2000) and (2) using bimetallic systems (Muftikian et al. 1995; Korte et al. 2000). In recent years, there has been considerable interest into combining the two methods (Schrick et al. 2002; Karn et al. 2009; Nagpal et al. 2010; Chen et al. 2011; Mossa et al. 2011).

Since the original proof of concept study into the application of nano-Fe⁰ for water treatment at Lehigh University, USA (Wang and Zhang 1997), research within this field has boomed. On August 20th 2011, a search at “ACS publications”, “Science Direct” (Elsevier journals), and “Springer journals” using key words “nanoscale” and “zerovalent iron” yielded 740 peer-reviewed articles (Table 1). As shown in Table 1, 234 articles have already been published in the first 7 months of 2011 by the three selected publishers. This clearly demonstrates the interest within academia for this technology.

Table 1.

Results of a web-search for three relevant publishers demonstrating the current interest within academia for the nano-Fe⁰ technology (search: 20 August 2011)

Publisher	Search’s results
	Period	Total	2011
ACS publications	2004 to 2011	263	96
Elsevier journals	1993 to 2011	392	112
Springer journals	2003 to 2011	85	26
	Total	740	234

In recent years, several review articles and critical views on nano-Fe⁰ for environmental remediation have been published (Zhang et al. 1998; Gillham 2003; Zhang 2003; Li et al. 2006; Macé 2006; Macé et al. 2006; Tratnyek and Johnson 2006; Karn et al. 2009; Pradeep and Anshup 2009; Agarwal and Joshi 2010; Gillham 2010; Müller and Nowack 2010; Noubactep and Caré 2010b; Cullen et al. 2011; Noubactep 2011b; Peralta-Videa et al. 2011; Shi et al. 2011; Truex et al. 2011a, b). However, the original discussion on the suitability of nano-Fe⁰ for in situ field applications (Gillham 2003, 2010) has not been satisfactorily addressed (Macé 2006; Noubactep 2011b). Moreover, a recent comparison between field applications of Fe⁰ of different particle sizes (nanometres, micrometres, and millimetres) for field applications has clearly demonstrated the superiority of mm-Fe⁰ (average efficacy 97%) (Comba et al. 2011). The decreasing order of reactivity was mm-Fe⁰ (97%) > μm-Fe⁰ (91%) > nano-Fe⁰ (65%). Expectably, the lower efficiency of nano-Fe⁰ is due to their high reactivity (Gillham 2003, 2010; Noubactep and Caré 2010b). Therefore, the question arises on the fundamental necessity to further increase the reactivity of nano-Fe⁰ by using a noble metal combination.

The Problem

Nano-Fe⁰ technology for environmental remediation was introduced as an alternative to the conventional Fe⁰ walls mostly for inaccessible aquifers (Masciangioli and Zhang 2003; Li et al. 2006). The very small particle size of nano-Fe⁰ (1–100 nm) would allow the material to penetrate deep into soil networks (Ghauch et al. 2009; Karn et al. 2009; Comba et al. 2011; Gheju 2011; Truex et al. 2011a, b).

Due to the exponential relationship between specific surface area (SSA) and radius (R = d/2) of a perfectly spherical object (SSA = 4πR ²), as a rule, a decrease in Fe⁰ particle size increases the surface area per gram by up to three orders of magnitude (Macé et al. 2006; Chen et al. 2011). In other words, the inverse relationship between Fe⁰ particle size and reactivity is due to a greater density of reactive sites on the particle surface at smaller scale. The following three claims have been made with regard to the use of nano-Fe⁰ for aqueous contaminant removal (Comba et al. 2011): (1) some aqueous contaminant species that have been proven as unsuccessful for remediation using μm-Fe⁰ and mm-Fe⁰ can be effectively removed using nano-Fe⁰, (2) nano-Fe⁰ can be used for more rapid degradation of contaminants, and (3) the formation of some undesirable by-products during remediation using μm- and mm-Fe⁰ can be avoided by using nano-Fe⁰. Such processes whilst correct are all linked to the greater reactivity that nano-Fe⁰ possesses due to its size (reactive surface area). When performed in conditions without a large nano-Fe⁰ stoichiometric excess, e.g. a system analogous to the environment, it may prove that such claims will be unfounded (Nagpal et al. 2010; Noubactep and Caré 2010b; Noubactep 2011b; Sakulchaicharoen et al. 2010; Nagpal et al. 2011).

An undisputed drawback with regards to the use of nano-Fe⁰ for environmental applications is their strong tendency to aggregate and adhere to solid surfaces (Li et al. 2006; Tratnyek and Johnson 2006; Karn et al. 2009; Sakulchaicharoen et al. 2010; Comba et al. 2011; Gheju 2011; Truex et al. 2011a, b). Karn et al. (2009) listed some parameters that influence nano-Fe⁰ adsorption onto soil and aquifer materials: (1) the surface chemistry of soil and Fe⁰ particles, (2) the groundwater chemistry (e.g. ionic strength, pH and presence of natural organic matter), and (3) the hydrodynamic conditions (pore size, porosity, flow velocity and degree of mixing or turbulence). Several methods have been developed for the stabilization of nano-Fe⁰ particles over the past decade and proven efficient to sustain the reactivity of nano-Fe⁰ (Karn et al. 2009; Sakulchaicharoen et al. 2010; Comba et al. 2011; Gheju 2011). One factor that has been overlooked, however, is the impact volumetric expansion has on the mobility of (1) residual Fe⁰, (2) primary corrosion products (Fe^II and H₂) and contaminants. The volume of any corrosion product (Fe hydroxide or oxide) is higher than that of the original metal (Fe⁰). The ratio between the volume of expansive corrosion product and the volume of iron consumed in the corrosion process is called “rust expansion coefficient” (η) (Anstice et al. 1993; Caré et al. 2008; Zhao et al. 2011). Volumetric corrosion products are likely to: (1) contribute to porosity loss, (2) impact the retention of contaminants and transformation products, and (3) increase the particle agglomeration.

Another area of heightened research is with regard to the determining the toxicity of nano-Fe⁰, with mixed results reported (Grieger et al. 2010; Gheju 2011). For example, Barnes et al. (2010) reported minimal change to the structure of a river water community due to the addition of nano-Fe⁰ while Diao and Yao (2009) reported nano-Fe⁰ particles as highly cytotoxic towards both gram-positive and gram-negative bacteria species.

While taking into account all known influencing parameters, the following seven features have to be systematically studied in order to optimise the general applicability of this technique (Karn et al. 2009; Tervonen et al. 2009; Gheju 2011): (1) mobility changes due to nano-Fe⁰ volumetric expansion during corrosion, (2) the bioavailability of Fe⁰ and corrosion products (Fe^II/Fe^III species, H/H₂), (3) the ecotoxicity of Fe⁰ and its corrosion products, (4) the bioaccumulation of Fe⁰ and its corrosion products, (5) the translocation potential of nano-Fe⁰, (6) the long-term reactivity of nano-Fe⁰ particles and (7) the speciation, persistence and fate of contaminants and their transformation products. A major contributing factor to the latter point is that little is known (compared to permeable reactive barrier technologies) about the extent contaminants are removed via size exclusion using nano-Fe⁰.

Only when all seven “operational drivers” have been determined can the global community have full faith in the technology.

Objectives of the Study

The present communication is focused on the “field persistence” or reactive “life span” of nano-Fe⁰ particles. For in situ applications, a keen understanding of nano-Fe⁰ reactive fate is essential for effective and prudent site clean-up. The knowledge of which is likely to largely underpin decisions as to (1) the choice of material selected, (2) the mechanism of application and (3) the strategy (if any) for repeated treatments.

In the current work, a multidisciplinary approach is used to analyse the relationship between nano-Fe⁰ reactivity and its performance for in situ field applications. The discussion is based on the contemporary knowledge of the mechanism of aqueous contaminant removal by Fe⁰ (Noubactep 2007, 2008, 2010c, 2011d, e). Much of the impetus for this work has come from the work of Noubactep and Caré (2010b) who have challenged the concept that nano-Fe⁰ is a strong reducing agent for contaminant reductive transformation.

Nanoscale Metallic Iron for Environmental Remediation

To date, nano-Fe⁰ particles have been reported as largely successful for water and soil treatment (Pradeep and Anshup 2009; Agarwal and Joshi 2010; Litter et al. 2010; Comba et al. 2011; Crane et al. 2011). A wide variety of redox-amenable organic and inorganic species and non-reducible species (e.g. Cd and Zn) have been efficiently treated. Similar to μm- and mm-Fe⁰, adsorption is considered important only for non-reducible species (Celebi et al. 2007; Li and Zhang 2007; Noubactep 2007, 2008, 2010c; Boparai et al. 2011; Noubactep 2011c, d; Xiao et al. 2011). For example, Boparai et al. (2011) reported that heavy metals are either reduced (e.g. Cu²⁺ and Ag²⁺) at, or directly adsorbed (e.g. Zn²⁺ and Cd²⁺) onto the Fe⁰ surface. They further argued that “the controlling mechanism is a function of the standard redox potential of the contaminant”. Recent work has however challenged this concept (Noubactep 2010c, 2011b, d, e), which is explained below.

Contaminant Reduction by Nano-Fe⁰

The chemical reaction between Fe⁰ and redox-amenable aqueous species is considered to involve three steps: (1) direct electron transfer from Fe⁰ at the metal surface or through a conductive oxide film on Fe⁰ (direct reduction), (2) catalyzed hydrogenolysis by the H/H₂ (indirect reduction mechanism 1) and (3) reduction by Fe^II species resulting from Fe⁰ corrosion (indirect reduction mechanism 2). In this constellation, H₂ is supposed to result from H₂O reduction during anoxic iron corrosion (Reardon et al. 2008; Chen et al. 2011). However, evidence exists in the literature, e.g. Stratmann and Müller (1994), that even under external oxic conditions, Fe⁰ is oxidized by H₂O (or more precisely by H⁺) and O₂ by Fe^II (Table 2). Despite the significant reaction rate exhibited by nano-Fe⁰ due to its high surface area, such processes are considered to occur (discounting any quantum size effects) independent of particle size.

Table 2.

Relevant redox couples for the process of aqueous Fe⁰ dissolution and oxide scale formation in a passive remediation Fe⁰/H₂O system. These processes are thermodynamically the same for all Fe⁰ particle sizes. Observed differences are due to kinetics aspects

Electrode reactions			Eq.
Fe0	⟺	Fe2++2e−	(I)
Oxic conditions
O2 + 2H2O + 4e−	⟺	4OH−	(IIa)
2H2O + 2e−	⟺	H2 + 2OH−	(IIb)
Anoxic conditions
O2 + 4H+ + 4e−	⟺	2H2O	(IIIa)
2H+ + 2e−	⟺	H2	(IIIb)

These processes are thermodynamically the same for all Fe⁰ particle sizes. Observed differences are due to kinetics aspects

Table 2 summarizes the half reactions for the aqueous oxidation of Fe⁰ under both anoxic and oxic conditions. Thermodynamically, the major cathodic reaction depends on the availability of molecular O₂ (E ⁰ = 0.81 V). In the absence of O₂, Fe⁰ is oxidized by H⁺ (E ⁰ = 0.00 V). It can therefore be stated that the rate of Fe⁰ oxidation is dictated by the concentration of dissolved O₂, H⁺ and H₂O in proximity to Fe⁰ surfaces. Le Chatelier’s principle also states that the consumption of Fe^II (via oxidation to Fe^III) will also result in an increase in Fe⁰ oxidation.

The electrode potential of the redox couple Fe^II/Fe⁰ is −0.44 V, a value which is independent of the particle size (nanometre, micrometre or millimetre). The value −0.44 V is considered largely unchanged due to the presence of alloying materials (e.g. low alloy steel and bimetallic systems).

As a consequence, statements including “nano-Fe ⁰ are more reactive than μm-Fe ⁰ and mm-Fe ⁰” are misleading; as the reactivity of Fe⁰ (discounting quantum size effects), is independent of the particle size. Any enhanced reactivity reported is likely to be due to the significantly high surface area of nano-Fe⁰ compared with other forms. A second statement “bimetallic nano-Fe ⁰ is more reactive that monometallic nano-Fe ⁰” is also a qualitative statement, as the reactivity of the materials depends on numerous factors associated with the materials synthesis route and varies depending on the chemistry of the chosen alloying metal. Ideally, comparisons should be made versus standard reference materials using established standard experimental protocols (Noubactep et al. 2009), which once established, will significantly improve the design of future field applications.

Limitations of the Nano-Fe⁰ Technology

The efficiency of nano-Fe⁰ for aqueous contaminant reduction faces some key issues for in situ applications in porous media. These challenges include: (1) the strong tendency of aggregation/agglomeration, (2) the rapid settlement on subsurface solid phases and (3) the porosity and permeability loss of porous media (Behrens et al. 2000; Dickinson and Scott 2010; Cullen et al. 2011; Mossa et al. 2011). Aggregation and settlement limit nano-Fe⁰ transport through porous media. Porosity and permeability loss limit nano-Fe⁰ transport to target contaminants. It was demonstrated that nano-Fe⁰ may travel only a few centimetres in porous media from the injection position under typical groundwater conditions (Tratnyek and Johnson 2006; Johnson et al. 2008; Comba et al. 2011; Gheju 2011). Accordingly, recent efforts have been made to (1) increase the porosity of porous media, (2) mechanically increase the distribution of nano-Fe⁰ and/or (3) chemically modify nano-Fe⁰ for improved aqueous mobility in porous networks.

Improving the Efficiency of the Nano-Fe⁰ Systems

Dispersion Agents

Methods to improve the aqueous mobility of nano-Fe⁰ have received the greatest research interest. It has been determined that the key to improving particle mobility is found in modifying their surface properties such that the nano-Fe⁰ have significantly improved colloidal stability and a commensurate reduction in the likelihood of adherence to mineral surfaces. Several synthetic methods are now available to produce more mobile nano-Fe⁰. Efficiently tested dispersants include anionic surface chargers (e.g. polyacrylic acid), non-ionic surfactants, starch and oil (Wu et al. 2009; Fang et al. 2011; Jiang et al. 2011; Mossa et al. 2011; Tong et al. 2011).

Bimetallic Combinations

In recent years, noble metals have been used to increase the reactivity of monometallic nano-Fe⁰ (Nagpal et al. 2010; Yuan et al. 2010; Zhu et al. 2010; Kadu et al. 2011; Mossa et al. 2011). As mentioned above, this appears counterintuitive as nano-Fe⁰ is already highly reactive due to its size (Wang and Zhang 1997; Zhang 2003; Macé et al. 2006; Ghauch et al. 2009) and is unstable during synthesis, storage and application (Jiang et al. 2011). This chemical instability has been documented as a key reason for the observed lower efficiency exhibited by nano-Fe⁰ systems compared with μm- and mm-Fe⁰ (Comba et al. 2011). Accordingly, it is questionable whether further enhancing the reactivity of nano-Fe⁰, e.g. by plating with more noble elements, may be of any benefit. The reactivity of nano-Fe⁰ will be discussed in the next section on the basis of mathematical modelling.

Significance of Increased Reactivity

The Problem

The increased Fe⁰ reactivity from mm to nm size should be better characterized. The relative reactivity of four different materials is discussed on the basis of 1 kg Fe⁰: 1 nm-Fe⁰ (d ₀ = 25 nm), 1 μm-Fe⁰ (d ₀ = 25 μm) and 2 mm-Fe⁰ (d ₀ = 250 and 1,000 μm). Calculations for the number particles (N) in 1 kg of each material and the number of layers (N′) in each particle are made after the Eqs. 1 and 2 presented in details elsewhere (Noubactep and Caré 2010a; Noubactep et al. 2010).

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ N = \frac{M}{{{\rho_{\text{Fe}}} \cdot \left[ 4/3\left( { \pi \cdot R_0^3} \right) \right] }} $$\end{document} 1

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ N\prime = {2} \cdot \left[ {{4}/{3}\left( {\pi R_0^3} \right)} \right]/{a^{{3}}} $$\end{document} 2

where M is the mass of Fe⁰ (here 1 kg), ρ _Fe is the specific weight of Fe (7,800 kg/m³), R ₀ is the initial radius of the Fe particle (d = 2*R ₀) and a the lattice parameter (a = 2.866 Å).

The results are summarized in Table 3. It can be seen that the number of layers of Fe⁰ in individual particles varies from 87.2 for nano-Fe⁰ to more than 3*10⁶ for mm-Fe⁰ (d = 1 mm). In the meantime, the number of particles in 1 kg decreased from 1.96*10¹⁸ for nano-Fe⁰ to only 3.1*10⁴ for mm-Fe⁰. The ratio of the number of Fe⁰ layers in each particle to the number of Fe⁰ layers in nano-Fe⁰ varies from 1 to 4*10⁴. This ratio corresponds to the relative time (τ) as defined later (section 3.2). On the other hand, the ratio of the number of particles in 1 kg of nano-Fe⁰ to the number of particles in the same mass of each other materials varies from 1 to 6.4*10¹³. These results are summarized in Fig. 1. Instead of the mass of Fe⁰, the number of electrons released by the conversion of Fe⁰ to Fe^II is used to assess the kinetics of Fe⁰ consumption. This is discussed in the next section.

Table 3.

Summary of the values of the number of particles contained in 1 kg of each material, the number of layer making up each particle and estimation of the relative time (τ)

Size	d (μm)	n particles (−)	n layers (−)	n layers/n nano (−)	n nano/n particles (−)	τ (−)
Nano-Fe0	25*10−3	1.96*1018	87.2	1.0	1.0	1.0
μm-Fe0	25	1.96*109	87.2*103	103	109	103
mm-Fe0	250	1.96*106	87.2*104	104	1012	104
mm-Fe0	1,000	3.06*104	3.49*105	4*104	6.4*1013	4*104

The life span of nano-Fe⁰ is operationally considered as the unit of time while assuming uniform corrosion. τ coincides with the ratio of the number of layers of Fe in each particle to that of nano-Fe⁰. The ratio of the number of particles in individual systems is also given

Fig. 1.

Comparison of the evolution of the kinetics of electron release and the number of layers in each Fe⁰ particle as a function of the particle size. It is shown that smaller particles release huge amounts of electrons within a very short time. Calculations are made for 1 kg of Fe⁰ material

Relative Corrosion Kinetics of Fe⁰ Materials

For the discussion in this section, uniform corrosion for spherical particles is assumed. Individual particles corrode independently until material depletion. It is further assumed for simplicity that individual layers corrode with the same kinetics independent of particle size. The latter assumption is conservative as larger particles react slower than smaller (Macé et al. 2006; Reardon et al. 2008; Nagpal et al. 2010). With these assumptions, a relative time (τ) can be defined while taking the time for the corrosion of the smallest particle (here 87.2 layers of nano-Fe⁰) or t _{∞, nano} as unit.

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \tau = t/{t_{{\infty, {\text{ nano}}}}} $$\end{document} 3

Accordingly, one unit of time corresponds to the time to nano-Fe⁰ depletion. Remember that all 1.96*10¹⁸ particles in the 1 kg of nano-Fe⁰ simultaneously corrode with the same kinetics. The results of the calculations are presented in Fig. 2. From Fig. 2a and Table 3, it can be seen that after nano-Fe⁰ depletion, the material with 1,000-μm (or 1 mm) diameter will still react for more than 3*10⁴ times longer than the time necessary for nano-Fe⁰ depletion (τ = 4*10⁴, see Table 3). Figure 2b shows that the mm-Fe⁰ with 250-μm diameter is depleted after about 10⁴*t _{∞, nano}.

Fig. 2.

Kinetics of the process of Fe⁰ exhaustion at nano-, micro- and millimetre scale as for: a d ≤ 1,000 μm and b d ≤ 200 μm

Based on the assumptions above, the service life of a nano-Fe⁰ particle can be estimated. Table 4 summarizes the results of such estimations while varying the service life of a 1 mm Fe⁰ particle from 5 to 40 years. This assumption is based on the fact that conventional Fe⁰ walls are supposed to function for several decades (here up to 4 decades). Results show (Table 4) that the maximum life span of a nano-Fe⁰ is about 8.8 h (less than 1 day). In other words, following approximately 9 h from subsurface deployment it is suggested that all nano-Fe⁰ would be reactively exhausted. The success of this is dependent on three key factors: (1) the hydrodynamic conditions: pore size, porosity, flow velocity and degree of mixing or turbulence, (2) the water chemistry and the affinity of nano-Fe⁰ and its transformation products to the soil materials and (3) the reactivity of Fe⁰.

Table 4.

Estimation of the value of the life span (t _∞) of a nano-Fe⁰ particle with 25 nm diameter for barrier life spans (t) from 10 to 40 years

t (years)	5	10	15	20	25	30	35	40
t μm (days)	45.7	91.3	137.0	182.6	228.3	273.9	319.6	365.3
t μm (years)	0.2	0.3	0.5	0.7	0.9	1.0	1.2	1.4
t ∞ (hours)	1.1	2.2	3.3	4.4	5.5	6.6	7.7	8.8

The considered conventional reactive wall contains granular Fe⁰ with a diameter of 1 mm. For comparison, the relative life span (in days and years) of the micrometric particles is given

It is certain that the dynamic process of transformation of concentric layers of Fe⁰ atoms to concentric layers of iron (hydr)oxides cannot be linear (Noubactep and Caré 2010b). In fact, effects similar to “case hardening” for food and wood drying will lead to “surface hardened layers” (Tarvainen et al. 2006; Fernando et al. 2008) leading to differential kinetics/extents of Fe⁰ passivation for different particle size ranges. In other words, the extent of restricted corrosion rates through resulting surface hardened layers will be different for nm-, μm- and mm-Fe⁰. Bearing this in mind, the very short relative life span of a nano-Fe⁰ estimated above will be used for the discussion in this work. It is certain that “case-hardening”-like effects will prolong this hypothetical life span to some days or weeks.

Extent of Iron Corrosion from Fe⁰ Materials

A discussion as to the extent of Fe⁰ consumption is limited in the present section to τ = 1 or t _{∞, nano}. It is considered for simplification that the sole iron corrosion product is Fe₃O₄. The corresponding coefficient of volumetric expansion is η _Fe3O4 = 2.08 (Eq. 4) (Caré et al. 2008). Using ρ = M/V, the volume of Fe corresponding to 1 kg Fe⁰ is calculated as 127.0 mL (V ₀). This is the initial volume of Fe⁰ (V ₀). Following corrosion, this volume is partly or totally consumed. The volume (ΔV) corresponding to the volume of pores occupied by the volumetric expansion of corrosion products can be estimated.

Assuming that the coefficient of volumetric expansion (η) (“rust expansion coefficient” or “specific volume”) (Anstice et al. 1993; Caré et al. 2008; Zhao et al. 2011) of the reaction products is:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \eta = {V_{\text{oxide}}}/{V_{\text{Fe}}} $$\end{document} 4

where V _oxide is the volume of the reaction product and V _Fe the volume of the parent Fe⁰.

The volume ΔV characterizing the extent of porosity loss due to volumetric expansion is given by Eq. 5:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta V = \left( {\eta - {1}} \right)*{V_{{{\text{consumed}}\,{\text{Fe}}}}} $$\end{document} 5

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta V = {V_{\infty }} - {V_0} = \nu *\left( {{\eta_{\text{Fe3O4}}}-{1}} \right)*{V_0} $$\end{document} 5a

Where V _{consumed Fe} is the volume of consumed Fe⁰ at time t _∞, V ₀ is the volume occupied by the initial Fe⁰ particles and ν (ν ≤ 1) is the fraction of the initial amount of Fe⁰ (1 kg) which has reacted at t _{∞, nano}. V _∞ is the total volume occupied by residual Fe⁰ and in situ formed corrosion products. t _∞ corresponds to nano-Fe⁰ depletion (25 nm in this section). In the discussion on the reactivity at nanoscale, t _∞ corresponds to the depletion of the material with 10-nm diameter (Section 4).

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {V_{\infty }} = \eta *{V_{{{\text{consumed}}\,{\text{Fe}}}}} + \left( {{V_0} - {V_{{{\text{consumed}}\,{\text{Fe}}}}}} \right) $$\end{document} 6

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {V_{\infty }} = {V_0 * }\left[ {{1 } + \nu *\left( {{\eta_{\text{Fe3O4}}} - {1}} \right)} \right] $$\end{document} 6a

The volumetric expansion (ΔV, Eq. 5) can be characterized as percent of the initial volume (V ₀) using Eq. 7:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \Delta V\left( \% \right) = {1}00*\nu *\left( {{\eta_{\text{Fe3O4}}}-{1}} \right) $$\end{document} 7

Table 5 summarizes the results. It is shown that at τ = 1 (nano-Fe⁰ depletion), a volume augmentation of 108% has occurred in the nano-Fe⁰ system, with volume augmentations in all other systems lower than 0.5%. This clearly shows that the porosity of the subsurface will be significantly influenced by nano-Fe⁰ at t _∞. Remember that 100% reactive exhaustion of nano-Fe⁰ is predicted to occur by approximately 9 h time. During this same period the porosity loss due to expansive iron corrosion is likely to be negligible for all other Fe⁰ particle size fractions. Calculations for Akageneite β-FeOOH (η _FeOOH = 3.48) as sole corrosion products shows that V _∞,nano = 448.7 mL, ΔV = 320.5 mL or 250.4%. The examples of Fe₃O₄ (anoxic) and FeOOH (oxic) demonstrate the crucial importance of the nature of formed corrosion products for the discussion of the extent of porosity loss.

Table 5.

Estimation of the extent of porosity loss (ΔV) due to the volumetric expansion of iron corrosion for Fe⁰ particles of different sizes

Size	m consumed (kg)	ν (%)	V ∞ (mL)	ΔV (mL)	ΔV (%)	n Fe(II) (moles)	n Fe3O4 (moles)	n electrons (moles)	n nano/n electrons (−)
nm-Fe0	100	100.00	264.16	137.16	108.00	17.857	5.9524	35.714	1
μm-Fe0	3*10−3	0.30	127.41	0.41	0.32	0.053	0.0178	0.107	335
mm-Fe0	3*10−4	0.03	127.04	0.04	0.03	0.005	0.0018	0.011	3,342
mm-Fe0	7.5*10−5	0.01	127.01	0.01	0.01	0.001	0.0004	0.003	13,369

The operational unit of time is arbitrarily the time to nano-Fe⁰ depletion (t _∞)

V ₀ volume occupied by the initial Fe⁰ particles, V _∞ volume occupied by residual Fe⁰ and in situ formed corrosion products. ΔV volume of pore occupied by the volumetric expansion of corrosion products, m _consumed mass of Fe⁰ consumed, ν percent of Fe⁰ depletion, n _Fe(II) number of moles of corroded Fe⁰, n _Fe3O4 number of moles of generated iron corrosion products, n _electrons  = 2*n _Fe(II) number of electrons released by corroded iron, n _nano/n _electrons the ratio of the number of electrons produced in by nano-Fe⁰ to n _electrons

Another important aspect of Fe⁰ consumption is given by the number of moles of Fe⁰ that have been oxidized (Table 5). Assuming contaminant reduction, Table 5 shows that after τ = 1, 35.71 moles of electrons have been released in the nano-Fe⁰ system but less than 0.11 moles in all other systems. In other words, up to 35.71 moles of electrons are available for contaminant reduction per kg nano-Fe⁰ within a few hours of reaction (<9 h). But what proportion of the electrons produced would reach the contaminant within this period? That is the major question to be answered for the further development of the nano-Fe⁰ technology for in situ applications.

Reactivity of Nano-Fe⁰ Materials

The presentation until now has discussed the reactivity of nano-Fe⁰ in comparison to larger scale Fe⁰. Section 4 will focus only on the nanoscale size fraction (d ≤ 100 nm). Equations 1–7 will be used and the particle size will vary from 10 to 100 nm. As stated above t _∞ is for a nano-Fe⁰ of 10 nm in diameter and the reaction proceeding until 100% reactive exhaustion has been achieved.

Fe⁰ Reactivity at Nanoscale

Table 6 summarizes the results of calculations for the number of Fe⁰ particles and number of layers of Fe⁰ in each nano-Fe⁰. It is shown that 1 kg of the material with d = 10 nm contains 1,000 times more particles than a material of d = 100 nm.

Table 6.

Summary of the values of the number of particles contained in 1 kg of each nano-Fe⁰, the number of layer making up each particle and estimation of the relative time (τ)

d (nm)	n particles (−)	n layers (−)	n layers/n 10 (−)	n 10/n particles (−)	τ (−)
10	3.06*1019	34.9	1.0	1.0	1.0
25	1.96*1018	87.2	2.5	16	2.5
50	2.45*1011	174.5	5.0	125	5.0
75	7.25*1016	261.7	7.5	422	7.5
100	3.06*1016	348.9	10.0	1,000	10.0

The life span of the material with the smallest particle site (d = 10 nm) is operationally considered as the unit of time while assuming uniform corrosion. The ratio of the number of particles in 1 kg of d = 10 nm to that of other d values is also given

Table 6 also shows that the maximum value of the relative time (τ) is 10 (or 10¹). This is more practical for graphical representations than situations where nano-Fe⁰ are compared with larger particles (t ≤ 10⁴). The physical significance of τ is more important, it means that if a nano-Fe⁰ with a diameter of 10 nm depletes after 2 days, the material with a diameter 100 nm will deplete after 20 days. For field applications, the selection of the particle size to be used should be dictated by site specific characteristics. Which diameter could quantitatively reach the contaminants before depletion? And what fraction of the material will have already oxidized on the path? What is the impact of this oxidation on the transport of nano-Fe⁰ in the porous aquifer? These are some key questions to be answered in order to give this possibly very efficient technology a scientific basis.

Figure 3 summarizes the evolution of the volumetric expansion in all five nano-Fe⁰ systems. It can be seen from Fig. 3a that the smallest material (d = 10 nm) experiences the 108% volumetric expansion within a short time (τ = 1) while the larger materials (d = 100 nm) needs ten more time for the same expansion. Accordingly, beside the question whether the material will reach the contaminant under site specific conditions, the question has to be answered how the volumetric expansion will impact the aquifer porosity (and permeability).

Fig. 3.

Kinetics of the process of porosity loss at nanoscale as characterized by: a the percent volumetric expansion for the five considered particle sizes, and b the absolute value of ΔV (mL) at τ = 1 for two different iron corrosion products (Fe₃O₄ and FeOOH)

Figure 3b compares the variation of the volumetric expansion for two different iron corrosion products, Fe₃O₄ and FeOOH, which are considered the most likely products in anoxic and oxic aquifers respectively. It also shows that when designing a nano-Fe⁰ injection strategy, however, the availability of oxidizing species (e.g. MnO₂ and O₂) must also to be taken into account. Figure 3a shows that under both conditions the trend of porosity loss is similar but the extent is proportional to the coefficient of volumetric expansion (η). In particular, at τ = 1, the system with the material d = 10 nm experiences 250% volumetric expansion under oxic conditions and only 110% under anoxic conditions. As a result the kinetics of more rapid Fe⁰ corrosion in an oxygen-rich environment must also be considered for an effective treatment strategy.

Fe⁰ Reactivity of Nano-bimetallics

The reactivity of monometallic nano-Fe⁰ can be improved by combining it with a noble metal. Assuming α (α > 1) the coefficient of reactivity enhancement, the relation between the relative time of a bimetallic system (τ _Fe/M) and that of a non-plated metal (τ _Fe) is given by Eq. 6:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {\tau_{\text{Fe}}} = \alpha *{\tau_{{{\text{Fe}}/M}}} $$\end{document} 8

To characterize the impact of plating on nano-Fe⁰, the material with the largest size (d = 100 nm) will be plated by three hypothetical metals (M ⁰₁, M ⁰₂ and M ⁰₃) to yield a reactivity factor of 2.5 (for Fe⁰/M ⁰₁), 5 (for Fe⁰/M ⁰₂) and 10 (for Fe⁰/M ⁰₃). The considered α values of (α ≤ 10) are realistic and even conservative. In fact, reported reactivity enhancement is essentially larger (Chen et al. 2011; Zhuang et al. 2011). For example, Zhuang et al. (2011) reported that palladized nano-Fe⁰ promoted the dehalogenation kinetics for polybrominated diphenyl ethers by orders of magnitude equal to 2, 3 and 4 (α ≥ 100). The results of the calculations for the four systems (d = 10 nm) are summarized in Fig. 4. The system with d = 10 nm is represented for comparison. It can be seen that system Fe⁰/M ⁰₃ (d = 100 nm) is as reactive as Fe⁰ (d = 10 nm). Given that the reactivity of nano-Fe⁰ (d = 100 nm) could already significantly been too high in some situations, the results from Fig. 4 strongly question the suitability of plating at nanoscale. Accordingly, while the application of bimetallic Fe⁰ is definitively useful at μm and mm scale, it usefulness at nanoscale is likely inappropriate. It can also be noted that by increasing the reactivity of the material the rate at which volumetric pore clogging also increases. As a consequence it should be acknowledged that there exists a conceptual play-off between increased reaction rate and increased porosity loss, the impact of which will vary depending on the physiochemical conditions of each contaminated site.

Fig. 4.

Calculated extent of Fe⁰ exhaustion as a function of the relative time (τ) for three ideal bimetallic systems based on the material with 100 nm diameter. The material with 10 nm diameter is represented for comparison

Characterizing the Process of Reactivity Loss

To better characterize the process of porosity loss due to the volumetric expansion of nano-Fe⁰, the evolution of the porosity of a sand column filled with nano-Fe⁰ will be discussed as volumetric expansion proceeds. A laboratory column with a height h (h = 75.0 cm) and diameter D (D = 5.0 cm) is composed of spherical sand particles (d = 5.0 mm). The compactness of the column is ideally C = 0.64 (Noubactep and Caré 2010a, Noubactep et al. 2010). The pore volume is given by Eq. 9:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {V_{\text{pore}}} = V*\left( {{1} - C} \right) $$\end{document} 9

where V is the apparent volume of the sand column (V = h*π*D ²/4).

It is supposed that the nano-Fe⁰ particles fill the inter-granular porosity of the sand column V _pore without modifying the compactness C and the apparent volume V of the sand column. The residual porosity of the sand column (V′_pore) is given by Eq. 10:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ V{\prime_{\text{pore}}} = V*\left( {{1} - C} \right) - {V_0} $$\end{document} 10

where V ₀ is the volume of the Fe particles.

The evolution of the residual porosity (V′_pore) as nano-Fe⁰ particles undergo volumetric expansive corrosion is considered by introducing the specific volume (η) of the reaction products according to Eq. 11:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ V_{\text{pore}}^{\prime } = V*\left( {{1} - C} \right) - \left( {{V_0} - {V_{{{\text{consumed}}\,{\text{Fe}}}}}} \right) - \eta *{V_{{{\text{consumed}}\,{\text{Fe}}}}} $$\end{document} 11

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ V_{\text{pore}}^{\prime } = V*\left( {{1} - C} \right) - {V_0} - \left( {\eta - {1}} \right)*{V_{{{\text{consumed}}\,{\text{Fe}}}}} $$\end{document} 11a

where V _{consumed Fe} is the volume of nano-Fe⁰ particles which is consumed at a given time.

Equations 9 through 11 are very useful to design reactive zone. However, they are limited to describe the initial (V _pore) and a final conditions (V′_pore) regardless on the nature of iron corrosion products and the kinetics of the process.

Using a sand column comparable to one of those used by Moraci and Calabrò (2010) and 1 kg of nano-Fe⁰, the process of pore filling (porosity loss) can be better characterized. For simplicity, nano-Fe⁰ considered as transported by a biodegradable dispersant which does not significantly contribute to porosity loss. As shown in Section 3.3, 1 kg of nano-Fe⁰ occupies a volume of 127 mL. The initial pore volume of the sand column calculated after Eq. 9 is 530.36 mL (100% porosity), i.e. a capacity for approximately 4.17 kg of nano-Fe⁰. Filling the initial pore volume of the sand column (530.36 mL) with 1 kg of nano-Fe⁰ (127.00 mL) yields a 23.9% porosity loss (Table 7). This, however, does not take into account the expansive nature of iron during oxidative corrosion.

Table 7.

Summary of the extent of porosity loss (ΔV _pore in %) as 1 kg of nano-Fe⁰ (V ₀ = 127 mL) is corroded to various iron oxides

Fe species	η (−)	V ∞ (mL)	ΔV (mL)	V′pore (mL)	ΔV pore (%)
Fe0	1	127	0	403.4	23.9
Fe3O4	2.08	264.2	137.2	266.2	49.8
Fe2O3	2.12	269.2	142.3	261.1	50.8
α-FeOOH	2.91	369.6	242.6	160.8	69.7
γ-FeOOH	3.03	384.8	258.0	145.6	72.6
β-FeOOH	3.48	442.0	315.0	88.4	83.3
Fe(OH)2	3.75	476.3	349.3	54.1	89.8
Fe(OH)3	4.2	533.4	406.4	−3.0	100.0
Fe(OH)3∙3H2O	6.4	812.8	685.8	−282.4	100.0

V _∞ is the volume of iron oxide at Fe⁰ exhaustion. The initial volume of pore (V _pore) is 530.4 mL and V′_pore is the residual pore volume at Fe⁰ exhaustion. The absolute value of negative values for V′_pore corresponds to the mass of nano-Fe⁰ which cannot oxidize because of lack of space for volumetric expansion

Using the eight possible iron corrosion products documented by Caré et al. (2008) and their respective coefficient of volumetric expansion (2.08 ≤ η ≤ 6.40), the extent of porosity loss is calculated and summarized in Table 7. The results show that the residual volume of pores (V′_pore) decreases with increasing η values and is zero for Fe(OH)₃ and Fe(OH)₃∙3H₂O (100% porosity loss). Ferrihydrite (Fe(OH)₃∙3H₂O) is the largest known iron corrosion products. In other words, depending on environmental conditions as little as 1 kg of nano-Fe⁰ could clog the tested column. Although this discussion considers the nature of the corrosion products, there are other important factors which must be considered. The negative values (−3.04 and −282.4 mL) corresponds to the mass of Fe⁰ which will not oxidized because of lack of space for expansion (Noubactep and Caré 2010a; Noubactep et al. 2010).

The extent of porosity loss (ΔV in %) given in Table 7 assumes uniform distribution of nano-Fe⁰ in the whole column. This is, however, not a very good field representation. For example, if 1 kg of nano-Fe⁰ (V ₀ = 127 mL) is uniformly distributed only in the first third of the column (V ^1/3_pore = 176.8 mL), with Fe₃O₄ as the primary corrosion product (ΔV = 137.16 mL) a 78% porosity loss can be expected. For all other oxide phases it is calculated that complete porosity loss (100%) will precede nano-Fe⁰ reactive exhaustion. However, in the practice a system with 78% porosity loss is considered as clogged. One possibility to avoid the clogging of the entrance zone of a porous system is to intermittently inject calculated amounts of nano-Fe⁰. The volume to be injected at each event and the time scale between two injections are necessarily determined by site specific characteristics (e.g. porosity of the aquifer and water flow rate).

Discussion

A primary reason behind the interest into the use of nano-Fe⁰ particles over μm-Fe⁰ and mm-Fe⁰ particles for water treatment is ascribed to a significant increase the materials efficiency (Comba et al. 2011; Gheju 2011). For example, as reported by Vodyanitskii (2010), Kanel et al. (2006) reported near-total remediation of a 1 mg L⁻¹ As^V solution within only 10 min by a nano-Fe⁰ with a specific surface of 24 m²/g, whereas the same goal was achieved by mm-Fe⁰ (1–2 m²/g) after only 4 days or 5,760 min (ratio of time, 570; average ratio of surface, 16). However, this experimental evidence is highly qualitative as neither the number of atoms directly accessible at the surface nor the intrinsic reactivity of individual materials are considered in both cases (Noubactep and Caré 2010b). For a better comparative result, the following three key conditions must be considered: (1) the intrinsic Fe⁰ reactivity should be characterized, (2) the amount of used materials should be based on the reaction stoichiometry and (3) the experimental conditions should be relevant for field applications. In particular, the driving force for the transport of contaminants and Fe species should be relevant for field situations: (1) mixing operation (type and intensity) in batch studies and (2) flow rate and column dimensions in column studies (Noubactep et al. 2009).

Transport of Nano-Fe⁰ to the Contaminants

The efficiency of nano-Fe⁰ for the in situ treatment of a contaminated aquifer body is intrinsically linked to the extent of physical contact between Fe⁰ and any aqueous contaminant species present. In some circumstances, contaminants could diffuse to the suspended Fe⁰ particles and be degraded in the aqueous phase. However, typically the suspended Fe⁰ particles must migrate to the contaminants. As Fe⁰ particles are transported from the injection zone to the target contaminant plume by natural groundwater, diffusion experiments under relevant groundwater velocity, using site specific aquifer materials are essential in order to effectively assess the suitability of nano-Fe⁰ for in situ applications (Antia 2011).

Contaminants are typically partitioned between sediment and water phases in a “pseudo-equilibrium” state. Therefore, it is likely that Fe⁰ particles whilst acting to reduce any soluble contaminants are also likely to promote the dissolution of a range of adsorbed chemical species (Le Chatelier’s principle). However, as water is also a redox-amenable species the specific mechanism for nano-Fe⁰ reactivity in a range of conditions is difficult to resolve (Tratnyek and Johnson 2006). In other words, nano-Fe⁰ is readily oxidized by H₂O during subsurface migration to the target contaminant plume and also competes with any other redox-amenable (including contaminants) present in the groundwater. Additionally, expansive iron corrosion will yield voluminous iron (hydr)oxides (Table 7) with limited aqueous mobility due to (1) an increased size and weight and (2) a possible increased affinity to aquifer material.

This discussion has intentionally neglected the segregation between parent compounds, the reaction products and their relative affinity to Fe⁰ and Fe (hydro)oxides. The fact that the core Fe⁰ is always covered by oxide layers is also neglected for simplification. The process of preferential flow which is crucial in predicting mass transfer in the subsurface is also not considered (Flury and Flühler 1994; Fryar and Schwartz 1998; Simunek et al. 2003; Clothier et al. 2008; Allaire et al. 2009). However, it is clearly shown, that due the acute redox sensitivity of nano-Fe⁰ and the subsequent significant formation of highly voluminous oxidative corrosion products it is likely that for environmentally relevant distances (metre), a significant proportion of the originally injected nano-Fe⁰ will remain “clogged” in pore spaces.

Beside the transport of nano-Fe⁰ to contaminants, the possibility of quantitative contaminant desorption and their subsequent transformation by suspended Fe⁰ could be considered. However, it is not likely that concentration-gradient-driven mass transfer could be quantitative at considered distances (metre). It should be recalled that the slow kinetics of contaminant desorption form aquifer materials is the major cause of the ineffectiveness of the pump-and-treat technology for groundwater remediation (McMurty and Elton 1985; Mackay and Cherry 1989; Starr and Cherry 1994).

This section has shown that it is likely that the success of nano-Fe⁰ for in situ remediation is seriously limited by the intrinsic formation of voluminous iron corrosion products (Tratnyek and Johnson 2006; Comba et al. 2011; Gheju 2011; Truex et al. 2011a, b). Bearing this in mind, the next section suggests an alternative nano-Fe⁰ subsurface deployment mechanism that more effectively takes into account the aforementioned nano-Fe⁰ hydraulic mobility issues than conventional injection processes: the formation of a nano-Fe⁰ “redox front” injection array system for progressive contaminant reduction. The geochemical process of redox front migration is a well-documented one (Fryar and Schwartz 1998; Min et al. 2005; Sidborn and Neretnieks 2007).

Nano-Fe⁰ as Source of Fe^II for a Redox Front?

The Concept

The progressive consumption of mm-Fe⁰ (Fig. 1; Table 3) is the guarantee for the long-term efficiency of reactive barriers (Comba et al. 2011). In fact, continuously generated small amount of high reactive iron minerals (Gu et al. 1999; Su and Puls 2001; Furukawa et al. 2002; Kohn et al. 2005; Noubactep 2007, 2008, 2010c, 2011b, c, d) are sufficient for the removal of contaminants which are present in trace amounts (Palmer and Wittbrodt 1991). As discussed above, for nano-Fe⁰ however, (1) Fe⁰ reactive exhaustion typically occurs in a relatively short time scale (<9 h) and (2) it is likely that nano-Fe⁰ subsurface mobility is significantly retarded or even prevented due to the volumetric expansive nature of iron corrosion (Zhao et al. 2011). As a consequence an alternative method of subsurface deployment is suggested in the current work: the deployment of a linear nano-Fe⁰ injection array orientated perpendicular to the flow direction of the contaminant plume. The injected nano-Fe⁰ can effectively form a redox front (roll-front) which migrates through the contaminated zone and transforms the contaminants during its migration as illustrated in Fig. 5. The Fe^II/Fe^III roll-front travels across the contaminated zone with all possible mechanisms (e.g. diffusion, dispersion, convection and preferential flow) and the contaminants are transformed and immobilized during the cycle Fe^II⇔Fe^III. In other words, it is a plume of Fe^II/Fe^III formed from injected nano-Fe⁰ which migrates through the contaminated zone and “sweeps” the contaminants. As a consequence this method considers all nano-Fe⁰ mobility issues.

Fig. 5.

Schematic diagram of the flow process of the U-shaped redox front through a contaminated zone. Despite the relative importance of preferential flow paths, contaminants are “swept” by the roll-front. The red circle is the contaminated zone. This zone is transformed to blue the circle upon treatment by the Fe^II/Fe^III front

Nano-Fe⁰ as Fe^II Generator

Nano-Fe⁰ in the aqueous phase is certainly a Fe^II/Fe^III producer. Fe^II species are the main reducing agents for contaminants under both anoxic and oxic conditions (Stratmann and Müller 1994; Nesic 2007; Kiser and Manning 2010; Noubactep 2010c, 2011c, d; Zhuang et al. 2011). Microbial activity could regenerate Fe^II (bio-corrosion) for more contaminant reduction (Vodyanitskii 2010). In this case, more contaminant is reduced than can be predicted from the reaction stoichiometry. In order words, the operating mode of nano-Fe⁰ for contaminant reduction can be summarized as follows: (1) Fe⁰ is oxidized to produce Fe^II, (2) Fe^II reduces the contaminant and is oxidized to Fe^III and (3) a proportion of Fe^II is regenerated by the biological reduction of Fe^III. Accordingly, before Fe⁰ depletion, there are three sources of Fe^II: (1) the Fe⁰-mediated abiotic oxidation by H₂O, (2) the Fe⁰-mediated abiotic oxidation by Fe^III and (3) the biological reduction of Fe^III. After Fe⁰ depletion, the only remaining source of Fe^II is the biological reduction of Fe^III. Provided that the appropriate micro-organism species are present in the subsurface, this process, however, could conceptually proceed for a significantly long-time period (Cullen et al. 2011).

Evidence suggests that such micro-organism colonies can be sustained by a consistent supply of Fe^II, Fe^III and molecular hydrogen (H/H₂). Another further process that is worth noting is the generation of atomic or molecular hydrogen (H/H₂) by Fe⁰-mediated hydrolysis reactions, which is likely to aid and the aforementioned biotic processes (Cullen et al. 2011).

The abiotic conversion of Fe^III to Fe^II has been successfully utilised in the hydrometallurgy industry, for example Lottering et al. (2008) reported on the sustainable use of MnO₂ for the abiotic regeneration of Fe^III for U^IV oxidation.

The fate of contaminant reduction products is discussed in the next section.

Mechanism of Contaminant Removal by Injected Nano-Fe⁰

The successful application of nano-Fe⁰ injection technology for in situ remediation is highly dependent on a comprehensive understanding of the fundamental processes governing the processes of contaminant removal. The hitherto discussion has focused on reductive transformations by nano-Fe⁰. However, contaminant reductive transformation is not a guarantee for contaminant removal (Noubactep 2010c, 2011c). Additionally, certain reaction products are more toxic than their parent compounds (Jiao et al. 2009). Accordingly, efforts have to be focused on the specific mechanism of aqueous contaminant removal. Relevant removal processes include: (1) adsorption, (2) chemical precipitation, (3) co-precipitation, (4) size exclusion or straining and (5) volatilization (Crawford et al. 1993a, b; Noubactep 2007, 2008, 2010c; Vodyanitskii 2010; Eusterhues et al. 2011; Johnson et al. 2011; Noubactep 2011d, e). Chemical precipitation is a characteristic of inorganic compounds when the solubility limit is exceeded (Crawford et al. 1993a, b; Kalin et al. 2005). Volatilization is subsequent to chemical transformation yielding gaseous species like AsH₃, CH₄, CO₂, H₂ and N₂. In Fe⁰ reactive barrier systems, contaminants are efficiently removed by the combination of adsorption, co-precipitation and size exclusion within the engineered barrier (Noubactep 2010d; Noubactep and Schöner 2010; Noubactep 2011a). As a result the current discussion concentrates on such processes.

With Fe⁰ (<1 m²/g) first transformed to voluminous hydroxides species (>100 m²/g) and subsequently transformed to oxides (<40 m²/g), contaminant size exclusion (straining) is driven by the dynamic cycle of expansion/compression accompanying the corrosion process (Pokhrel and Viraraghavan 2008; Vodyanitskii 2010; Noubactep 2011a). During these cycles contaminants are enmeshed and sequestrated in a “matrix” of iron corrosion products.

For conventional nano-Fe⁰ injection arrays, size exclusion may play an important role (1) in proximity for Fe⁰ particles and (2) by reducing the pore space during expansive corrosion of the materials. However, if roll-fronts are formed as discussed above, the extent of permeability loss in aquifer will be limited. The roll-front could act as a colloidal reactive barrier for the removal of parent contaminants and reaction products. Species are removed or immobilized by colloids and not because they are reduced. More research is needed to test this hypothesis.

Fe^II/Fe^III Redox Front as a Colloidal Reactive Barrier

Aqueous contaminants have been reported to be quantitatively removed both during abiotic and biotic (1) oxidation of Fe^II and (2) reduction of Fe^III (Pokhrel and Viraraghavan 2008; 2009; Pokhrel et al. 2009; Vodyanitskii 2010). On the other hand, injection of Fe^III salts for adsorptive contaminant removal has been reported (Morrison and Sprangler 1993; Morrison et al. 1996). Accordingly, the migration of the Fe^II/Fe^III redox front may be coupled to quantitative contaminant removal by adsorption and co-precipitation.

The primary reason for contaminant removal during these redox reactions is the colloidal nature of in situ generated Fe species (Fe(OH)₂ and Fe(OH)₃) (Hanna and Boily 2010), which necessarily experience volumetric contraction to form oxides (of Fe^II or Fe^III). Contaminants are first adsorbed by highly reactive colloids and are co-precipitated during transformation to amorphous and crystalline oxides (Ghauch et al. 2010; Noubactep 2011c, d).

Concluding Remarks

Constructed geochemical barriers of metallic iron (Fe⁰) have been used for groundwater remediation since 1996 (O’Hannesin and Gillham 1998; Scherer et al. 2000; Henderson and Demond 2007; Gillham 2010, Comba et al. 2011; Gheju 2011). In recent years, however, nano-Fe⁰ has received proclaim as a new tool for water treatment due to (1) improvements in reactivity and associated aqueous contaminant removal performance compared with conventional materials and (2) the option of subsurface deployment via injection for targeted in situ treatment of contaminant plumes (Comba et al. 2011; Gheju 2011).

Considering reactivity first, the current work has highlighted the need for prudent use of terminology. Discounting any quantum size effects, which are only prevalent for Fe⁰ less than approximately 10 nm in diameter, the reactivity of nano-Fe⁰ as a function of surface area is no more reactive than larger forms. Nano-Fe⁰ only exhibits such high reactivity due to it significantly high surface area as a function of mass/volume. Despite this, a recent trend in research has been the development of bimetallic nano-Fe⁰ wherein the combination of a noble metal acts to further increase the reactivity of nano-Fe⁰. It is argued in the current work that as reactive exhaustion is already achieved by monometallic nano-Fe⁰ in the order of minutes this seems counterintuitive for the majority of environmental applications.

Considering the nano-Fe⁰ subsurface injection procedure, in the current work it has been highlighted that the hydraulic mobility of the particles is likely to be significantly retarded by voluminous expansion due to particle corrosion. An alternative nano-Fe⁰ injection procedure has been suggested herein. The injected nano-Fe⁰ effectively forms an in situ migrating front which possibly reductively transforms contaminant and removes reduced species by adsorption and co-precipitation.

It is also outlined in the current work that a number of studies with experiments “proclaimed” as analogous to environmental systems are largely overlooked a range of operational drivers including changes in nano-Fe⁰ (1) reactivity and (2) voluminous as a function of time. It is hoped that the huge literature on redox front migration (Reynolds and Goldhaber 1978; Posey-Dowty et al. 1987; Romero et al. 1992; Read et al. 1993) and the cycle of iron in the hydrosphere (Vodyanitskii 2010) will now be used for the further development of nano-Fe⁰ injection technology.