Scholarly article on topic 'Methods for characterizing the fate and effects of nano zerovalent iron during groundwater remediation'

Methods for characterizing the fate and effects of nano zerovalent iron during groundwater remediation Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Keywords
{Groundwater / Remediation / "Zerovalent iron" / Fate / Effects / Characterization}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Zhenqing Shi, Dimin Fan, Richard L. Johnson, Paul G. Tratnyek, James T. Nurmi, et al.

Abstract The emplacement of nano zerovalent iron (nZVI) for groundwater remediation is usually monitored by common measurements such as pH, total iron content, and oxidation–reduction potential (ORP) by potentiometry. However, the interpretation of such measurements can be misleading because of the complex interactions between the target materials (e.g., suspensions of highly reactive and variably aggregated nanoparticles) and aquifer materials (sediments and groundwater), and multiple complications related to sampling and detection methods. This paper reviews current practice for both direct and indirect characterizations of nZVI during groundwater remediation and explores prospects for improving these methods and/or refining the interpretation of these measurements. To support our recommendations, results are presented based on laboratory batch and column studies of nZVI detection using chemical, electrochemical, and geophysical methods. Chemical redox probes appear to be a promising new method for specifically detecting nZVI, based on laboratory tests. The potentiometric and voltammetric detections of iron nanoparticles, using traditional stationary disc electrodes, rotating disc electrodes, and flow-through cell disc electrodes, provide insight for interpreting ORP measurements, which are affected by solution chemistry conditions and the interactions between iron nanoparticles and the electrode surface. The geophysical methods used for characterizing ZVI during groundwater remediation are reviewed and its application for nZVI detection is assessed with results of laboratory column experiments.

Academic research paper on topic "Methods for characterizing the fate and effects of nano zerovalent iron during groundwater remediation"

ARTICLE IN PRESS

CONHYD-03100; No of Pages 19

Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Review article

Methods for characterizing the fate and effects of nano zerovalent iron during groundwater remediation

Zhenqing Shi3*, Dimin Fanb, Richard L. Johnson b, Paul G. Tratnyekb'**, James T. Nurmic, Yuxin Wud, Kenneth H. Williams d

a School of Environment and Energy, South China University of Technology, Guangzhou, Guangdong 510006, PR China b Institute of Environmental Health, Oregon Health & Science University, 3181 SWSam Jackson Park Road, Portland, OR 97239, United States c Engineering Science Department, Clackamas Community College, 19600 MolallaAve., Oregon City, OR 97045, United States d Earth Sciences Division, Lawrence Berkeley National Laboratory, #1 Cyclotron Road, MS 74R0316C, Berkeley, CA 94720, United States

ARTICLE INFO ABSTRACT

The emplacement of nano zerovalent iron (nZVI) for groundwater remediation is usually monitored by common measurements such as pH, total iron content, and oxidation-reduction potential (ORP) by potentiometry. However, the interpretation of such measurements can be misleading because of the complex interactions between the target materials (e.g., suspensions of highly reactive and variably aggregated nanoparticles) and aquifer materials (sediments and groundwater), and multiple complications related to sampling and detection methods. This paper reviews current practice for both direct and indirect characterizations of nZVI during groundwater remediation and explores prospects for improving these methods and/or refining the interpretation of these measurements. To support our recommendations, results are presented based on laboratory batch and column studies of nZVI detection using chemical, electrochemical, and geophysical methods. Chemical redox probes appear to be a promising new method for specifically detecting nZVI, based on laboratory tests. The potentiometric and voltammetric detections of iron nanoparticles, using traditional stationary disc electrodes, rotating disc electrodes, and flow-through cell disc electrodes, provide insight for interpreting ORP measurements, which are affected by solution chemistry conditions and the interactions between iron nanoparticles and the electrode surface. The geophysical methods used for characterizing ZVI during groundwater remediation are reviewed and its application for nZVI detection is assessed with results of laboratory column experiments.

© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Article history:

Received 17 December 2014

Received in revised form 26 February 2015

Accepted 9 March 2015

Available online xxxx

Keywords:

Groundwater

Remediation

Zerovalent iron

Effects

Characterization

Contents

1. Introduction................................................................................................................0

2. Preparation and formulation of nZVI..........................................................................................0

3. Essential chemistry of nZVI..................................................................................................0

4. Distribution of nZVI in the subsurface........................................................................................0

5. Classification of characterization methods....................................................................................0

* Corresponding author. Tel./fax: +86 20 39380508. ** Corresponding author. Tel.: +1 503 346 3431; fax: +1 503 346 3427.

E-mail addresses: zqshi@scut.edu.cn (Z. Shi), tratnyek@ohsu.edu (P.G. Tratnyek).

http://dx.doi.org/10.1016/jjconhyd.2015.03.004

0169-7722/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

ARTICLE IN PRESS

2 Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

6. Direct methods for characterization of the fate and effects of nZVI................................0

6.1. Microscopic and spectroscopic techniques.......................................0

6.2. Color and absorptivity................................................0

6.3. Total iron......................................................0

7. Methods for indirect characterization of the fate and effects of nZVI...............................0

7.1. Total dissolved iron..................................................0

7.2. pH.........................................................0

7.3. Dissolved oxygen...................................................0

7.4. Concentration of contaminants............................................0

7.5. Conservative tracers.................................................0

7.6. Dissolved hydrogen..................................................0

7.7. Redox active probes.................................................0

7.8. Potentiometry and voltammetry: ORP measurements..................................0

7.8.1. ORP: conventional stationary electrodes....................................0

7.8.2. ORP: rotating disc electrodes.........................................0

7.8.3. ORP: flow-through cell electrodes.......................................0

7.9. Geophysical methods.................................................0

8. Implications and recommendations..............................................0

Acknowledgments........................................................0

References...........................................................0

1. Introduction

Nano zerovalent iron (nZVI) is one of the most extensively studied types of environmental nanoparticles because of its use in groundwater remediation. Many laboratory studies have been conducted to investigate multiple aspects related to nZVI application, including synthesis methods (He and Zhao, 2007; Wang and Zhang, 1997), material characterization of nZVI freshly prepared by various methods (Baer et al., 2010a, 2012; He and Zhao, 2005; Nurmi et al., 2005, 2011; Sarathy et al., 2008; Yan et al., 2010b), mobility in porous media (He et al., 2007, 2009; Johnson et al., 2009; Phenrat et al., 2009; Phenrat et al., 2010), and interactions/reactions with a wide range of contaminants (Fan et al., 2013,2014; Ling and Zhang, 2014a,b; Liu and Lowry, 2006; Liu et al., 2007; Wang and Zhang, 1997). This laboratory-based work is sufficiently well-developed and abundant that it has been the basis for multiple reviews of the process-level aspects of potential applications of nZVI in remediation (O'Carroll et al., 2013; Tosco et al., 2014; Tratnyek and Johnson, 2006; Yan et al., 2013).

Compared with the advanced state of literature from process-level laboratory-scale studies with nZVI, relatively few detailed studies have been reported on the application of nZVI for groundwater remediation at the field scale. In the few studies that have been well documented (Bennett et al., 2010; Elliott and Zhang, 2001; He et al., 2010; Henn and Waddill, 2006; Johnson et al., 2013; Kocur et al., 2014; Quinn et al., 2005 ; Wei et al., 2010), the effectiveness of nZVI injection is usually assessed mainly by the changes in the concentrations of target contaminants of concern (CoCs), even though there are multiple processes other than degradation by nZVI that might contribute to CoC attenuation (displacement, dilution, stripping into off-gassing H2, etc.). More comprehensive methods for characterizing the fate and effects of nZVI — including nZVI delivery, reactivity, transformation, effects on contaminants, and impact on the surrounding environment — during ground-water remediation are needed to elucidate the potential of this technology. The only study to date that emphasizes the methodological aspects of assessing nZVI transport and fate is

described in two of our previous publications (Johnson et al., 2013; Shi et al., 2011). This review is the culmination of that effort, including a synthesis of results and insights from prior work, supplemented with complementary data that we had not previously reported. It begins with a brief review of essential precursory topics (composition of nZVI formulations and reactivity of nZVI with media), followed by classification and discussion of ex situ characterization methods (performed on water samples from monitoring wells), and then in situ characterization methods (mainly geophysical). The overall goal is to advance the quality of site characterization during and after subsurface emplacements of nZVI and provide a more reliable basis for interpreting and optimizing the performance of this technology.

2. Preparation and formulation of nZVI

Most nZVI is produced with chemical reduction methods, either by H2 under high temperature (e.g., the commercial product known as RNIP-10DP and RNIP-M2 from Toda Corp.), or by borohydride reduction of aqueous Fe(II) under ambient conditions (FeBH) (which is the most common on-site synthesis method for laboratory and field studies). Bare nZVI particles usually have a core-shell structure composed of an Fe0 core covered by a thin layer of iron (hydr)oxides (Li and Zhang, 2006; Martin et al., 2008; Nurmi et al., 2005). Due to electrostatic and magnetic attractions, bare nZVI particles in aqueous suspension quickly form aggregates that are large enough to have poor mobility in porous media (Phenrat et al., 2007). To minimize aggregation and maximize mobility, nZVI is typically prepared and/or deployed in the presence of additives such as surfactants or other polyelectrolytes. For example, RNIP-M2 from Toda Corp. was nZVI with surface modification by aspartate. Among numerous methods for surface modification of nZVI (reviewed in Tratnyek et al. (2011) and Yan et al. (2013)), the use of carboxymethyl cellulose (CMC) stabilizers for nZVI surface modification (CMC-nZVI) has attracted the most interest for field application (He and Zhao, 2005,2007; He et al., 2007,2009,2010; Johnson et al., 2013; Kocur et al., 2014).

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Unlike many stabilizers that are added after particle synthesis, CMC is usually added prior to synthesis to allow formation of CMC-Fe(II) complexes, and this results in the formation of stable nZVI suspension with very small particle size. Another nZVI formulation, NANOFER by NANO IRON s.r.o. (Rajhrad, Czech Republic), is a commercial product and has been used relatively recently in multiple laboratory studies and field demonstrations (Eglal and Ramamurthy, 2014; Kadar et al., 2011; Klimkova et al., 2011; Laumann et al., 2013,2014; Lerner et al., 2012; Zhuang et al., 2012). NANOFER 25S consists of nZVI particles with the average primary particle diameter about 50 nm, and surface modification with iron oxides and polyacrylic acid.

The varying chemical and physical properties of nZVI particles from different synthesis methods directly affect the fate and transport of nZVI in subsurface environments, and thus their detection. Rapid changes in water chemistry and electrometric properties of impacted areas are expected upon nZVI injection, and the extent of these impacted areas will be greater for nZVI with small size and good physical stability. The oxidation of nZVI particles (to nZVIox) complicates the characterization of nZVI transport because of differences in the physical, spectroscopic, and electrometric properties of nZVIox vs. the original reduced nZVI (Johnson et al., 2013; Liu et al., 2014; Shi et al., 2011). Also, the small size of some nZVI particles requires special attention to distinguish nZVI particles from dissolved iron species in water samples. In some respects, the polyelectrolyte stabilizers and other ingredients used in synthesis of nZVI can be helpful for mapping nZVI impacted regions of the subsurface, but these ingredients may also complicate the use of other characterization methods, such as geophysics.

3. Essential chemistry of nZVI

The utility of nZVI derives from its relatively high reactivity, but this reactivity (and that of some components used to prepare nZVI, such as borohydride) also impacts the overall fate and effects of the injected material in ways that strongly influence options for their characterization. To a large extent, the reactivity of nZVI is determined by the corrosion of Fe0, which consumes dissolved oxygen, other oxidizing species (including contaminants), and protons; resulting in anoxia, elevated H2 and pH, and formation of various Fe(II) or Fe(III) species. The relationships between the reactions that give rise to these effects are summarized in Fig. 1.

The reactions shown in Fig. 1 apply to any type of ZVI, but the small particle size and high specific surface area of nZVI mean that the extent of corrosion (Rxn 1) can result in substantial or even complete depletion of the Fe0. This introduces the prospect that in situ reactions initiated by nZVI may become limited by the quantity of nZVI delivered. Conversely, the combination of reaction between nZVI and natural in situ oxidants (O2, H2O, natural organic matter, some minerals, etc.) comprises a natural reductant demand (NRD), which competes with the preferred reaction of nZVI with contaminants.

The reactions between nZVI and NRD result in the formation of oxidized iron species, including Fe(II) in anoxic environment or Fe(III) in oxic environment (e.g., Fig. 1, Rxns 2 and 3). Formation of the mixed-valent iron oxide magnetite

Fig. 1. The major chemical reactions (Rxns) contributing to subsurface conditions after the addition of nZVI (shown generically as Fe0). (1) Corrosion of Fe0; (2) and (3) precipitation of Fe(II) and Fe(III) oxyhydroxides; (4) transformation of Fe oxides. Solid phases are shown in bold.

from corrosion of Fe0 under anoxic conditions — as suggested by Rxn 2 in Fig. 1 — occurs by a disproportionation known as the Schikorr reaction (Filip et al., 2014). In typical field nZVI injection trials (where initial conditions are aerobic), Fe(III) oxides are expected to form first because nZVI rapidly reacts with dissolved oxygen (DO). After DO is consumed, H2O becomes the major contributing NRD of nZVI, yielding dissolved Fe(II), H2, and OH- (Fig. 1, Rxn 1), which react further to form Fe(II/III) oxides (Fig. 1, Rxn 2).

Additional possible reactions that are characteristic of nZVI in groundwater include abiotic reduction of nitrate to ammonia; stimulation of biotic sulfidigenesis; precipitation of iron (oxyhydr)oxides, sulfides (mackinawite, FeS), carbonates (siderite, FeCO3), and phosphates (vivianite, (Fe3(PO4)2-8H2O)); and sequestration of metals by reduction, adsorption, and precipitation/co-precipitation (ranging from alkali metal cations to transition metal oxyanions). For all of these reactions, their roles in contaminant removal during remediation applications of nZVI have been studied extensively (as noted above), but less attention has been given to their use as indicators of the fate and effects of nZVI when it is deployed in situ.

Extrapolation from general consideration of chemical processes, as described above, to field-scale, in situ applications of nZVI must consider additional factors that often are not well represented in laboratory studies. One such factor is aging of the nZVI, which is negligible (usually avoided) in most laboratory studies but substantial (and unavoidable) in most field studies. For example, batch experiments performed with commercial RNIP nZVI have given evidence of significant passivation after one month of exposure to common ground-water anions (Reinsch et al., 2010; Sarathy et al., 2008). Furthermore, we showed in a more recent study that the presence of sulfide caused conversion of up to 10% of the original nZVI (FeBH) into FeS within 24 h (Fan et al., 2013). However, it should be noted that the kinetics of passive film formation or transformation are not only dependent on the type of anions, but also dependent on the properties of nZVI. For CMC-nZVI, due to its smaller particle size and higher specific surface area than most of other nZVI formulations (He and Zhao, 2007), aging and related transformation reactions could occur at rates fast enough to impact nZVI transport during the early stages of emplacement.

ARTICLE IN PRESS

4 Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

4. Distribution of nZVI in the subsurface

Successful remediation of contaminated source zones using nZVI necessitates that the nZVI be sufficiently mobile in the subsurface to reach the target treatment zone. To achieve this, a variety of nZVI emplacement strategies have been developed. Nearly all strategies involve injection of the nZVI in conjunction with some organic amendment (e.g., biopolymer, surfactant, shear-thinning fluid). These provide a number of potential benefits, including: (1) decreased filtration losses (i.e., decreased particle sticking coefficient), (2) decreased aggregation, (3) enhanced sweep efficiencies, and (4) an external carbon source for enhanced biodegradation. Applications in which nZVI/organic suspensions are injected into either conventional wells or drive points using either pressure injection or gravity feed have been demonstrated (Johnson et al., 2013). For most field demonstrations, delivery of nZVI into target treatment zones has been limited due to filtration of the nZVI by the porous media, and/or preferential flow (which results in bypassing significant portions of the treatment zone). Regions that are not directly impacted by nZVI may still be impacted indirectly as reactions of the nZVI alter the in situ biogeochemistry and these changes spread with natural transport of the groundwater.

Most scenarios for nZVI emplacement by injection into wells are expected to create a distribution of nZVI and nZVI-impacted fluids in the subsurface similar to that shown conceptually in Fig. 2. Briefly, nZVI will propagate from the injection well into the aquifer with a distribution pattern that is largely determined by the heterogeneous structure of the aquifer and the filtration properties of the nZVI (zone A in Fig. 2). Transport distances of a meter or less seem to be typical (Johnson et al., 2013), but a portion of nZVI may be transported up to a few meters along preferential pathways with high permeability. The distance and distribution of nZVI emplacement are strongly influenced by the injection velocity (e.g., higher velocities near the injection well minimize deposition, while slower velocities away from the well favor deposition) (Krol et al., 2013).

Depending on pre-existing redox conditions in the aquifer, a portion of the nZVI may be oxidized during transport (to nZVIOx, zone B in Fig. 2). As suggested in the figure, the oxidized material can travel farther than the unoxidized material (nZVI) because oxidation takes place primarily at the leading edge of the nZVI front. In general, nZVI movement will stop when the high velocities that are applied during injection end. Beyond the distance to which the nanoparticles are transported, dissolved species will continue to move under the influence of groundwater flow. These species may include Fe2+, dissolved hydrogen, the dissolved organic phase (e.g., CMC) as well as ions associated with synthesis of the nZVI (e.g., borate) (zone C in Fig. 2). Since many of these species are biogeochemically active, their concentrations down-gradient from the injection well will likely persist for a few meters at most. Beyond that point, non-reactive species will continue to move with the groundwater (zone D in Fig. 2). In general, there will be little contaminant attenuation in this last zone (D).

The conceptual model represented by Fig. 2 provides a rubric for classifying the methods and criteria used to characterize the in situ fate and effects of nZVI emplacement. In general, each of the zones in Fig. 2 can be characterized by groundwater sampling during the injection process. As discussed below, nZVI is readily visible in groundwater down to concentrations of about 10 mg L-1. Thus, groundwater samples collected during injection (from zone A, Fig. 2) will become black wherever nZVI concentrations are sufficiently high to provide contaminant degradation by direct reaction with ZVIred. Similarly, the oxidized material (ZVIox) is visible by eye and/or spectrophotometry when it is present (Johnson et al., 2013) and dissolved reactive species can be measured by conventional means, thereby delimitating zone B. After injection has ended, concentrations of iron nanoparticles in groundwater decrease rapidly due to aggregation, deposition, and filtration. As a consequence, characterization of the nZVI distribution after injection usually depends on indirect measurements, collection of solid-phase samples, and/or using geophysical methods.

Fig. 2. Expected distribution of nZVI and nZVI impacted fluids resulting from injection of nZVI into a moderately permeable aquifer. Features are not drawn to scale. NPs: nanoparticles.

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx 5

5. Classification of characterization methods

Depending on the method of sample collection and preservation, a wide variety of sample characterization methods can be useful for determining the fate and effects of nZVI. For direct detection of any type of anthropogenic nanoparticles in environmental media, there are few sensitive and specific methods and there is no proven and widely-used protocol for direct detection of nZVI in the field. Electron microscopy and spectroscopy, color and absorptivity, and total iron (by standard colorimetric procedures) are among the most used direct detection methods.

Instead, the field studies reported so far (as well as many laboratory column studies) have relied mainly on monitoring methods that are indirect in that their response is not necessarily due to the nZVI per se but rather is mediated by products of reactions between nZVI and the medium. These reactions consume dissolved O2 and reducible contaminants (sometimes generating diagnostic products), generate Fen/Fem species and H2, increase the pH, and lower the oxidation-reduction potential (ORP). Thus, the nZVI impacted zone may be delineated by (1) changes in water chemistry parameters such as pH and dissolved O2, H2, and Fe2+ or (2) changes of electrometric properties such as ORP and complex conductivity. In the following two sections, we discuss both direct and indirect methods for characterization of the fate and effects of nZVI.

6. Direct methods for characterization of the fate and effects of nZVI

To fully characterize the field-scale transport of injected nZVI, some methods must be employed that can directly and unequivocally detect the presence of Fe0 (i.e., nZVI that is at most partially oxidized). For this, the first step involves sampling and sample handling in such a way as to preserve even small quantities of Fe0 in small particles. Then, Fe0 can be identified using a combination of direct microscopic and spectroscopic techniques, including scanning electron microscopy (SEM), transmission electron microscopy (TEM), Mossbauer spectroscopy, X-ray photoelectron spectroscopy (XPS), etc. These methods can provide definitive evidence for the presence of nZVI, but they are not practical for routine application to non-research projects in the field. In addition, there are other techniques with possible, but not yet demonstrated, applications to nZVI detection, such as field-flow fractionation interfaced to inductively coupled plasma mass spectrometry (FFF/ICP-MS) and flow cytometry.

For field work, the methods often interpreted as evidence for nZVI are less direct than the electron microscopic and spectroscopic methods. These include total iron, absorptivity at wavelengths indicative of iron species (see below), and visual inspection for coloration that can be routine analyses in the field for nZVI detection.

6.1. Microscopic and spectroscopic techniques

Electron microscopy has been employed to provide direct measurement of the properties of nZVI particles and their interactions with environmental matrix and contaminants (e.g. Baer et al., 2008, 2010a,b, 2012; Kanel et al., 2006; Kocur et al.,

2014; Ling and Zhang, 2014a,b; Nurmi et al., 2005; Sarathy et al., 2008). Scanning electron microscopy coupled with X-ray fluorescence analysis (SEM/XRF) is routinely used to distinguish sub-micron nZVI particles from aquifer and organic materials, and the chemical imaging by energy-dispersive spectroscopy (EDS) coupled to SEM can provide elemental compositions of selected nZVI particles. Transmission electron microscopy has been widely used to measure the size and morphology of nZVI particles and the aggregates. The development of TEM has provided the capability of detecting nZVI particles from nanoscales to angstrom scales. Most recently, by using spherical aberration-corrected scanning transmission electron microscopy (STEM), the core-shell structure of a single nZVI particle and its interaction with heavy metals can be visualized at the angstrom scales (Ling and Zhang, 2014a,b).

Advanced spectroscopic techniques can be also used to characterize the chemical compositions of nZVI particles, especially under the impact of contaminants and environmental media. Mossbauer spectroscopy has unique capabilities for identifying iron mineral phases, valence distribution, and structural environment of iron in various environmental media, and has been used to characterize the dynamic properties of nZVI particles (Filip et al., 2014; Kanel et al., 2006; Petala et al.,

2013). X-ray photoelectron spectroscopy is another surface sensitive technique for nZVI surface characterization and its interactions with heavy metals (Baer et al., 2008,2010b; Li and Zhang, 2007; Martin et al., 2008; Nurmi et al., 2005, 2011 ; Yan etal., 2010a,b, 2012).

The above electron microscopic and spectroscopic techniques, especially when used in combination, provide the most direct and definitive characterization of nZVI derived particles. However, because they are ex situ analyses requiring specialized instrumentation, they do not provide the kind of rapid, routine, and efficient detection methods needed for monitoring nZVI deployments in the field. Furthermore, most of these methods require considerable sample preparation, which is difficult to do without significant sample alteration (Baer et al., 2012; Nurmi et al., 2011). New electron microscopic techniques suitable for in situ detection of engineered nanoparticles are being developed (Benn and Westerhoff, 2008; Kent et al.,

2014), but sample preparation is still required and a potential source of misleading results.

Other techniques based on ICP-MS have been developed for nanoparticle detections in environmental samples. The FFF/ICP-MS has been used for detecting a variety of nanoparticles including Ag, iron (hydr)oxides, TiO2, etc. (Plathe etal., 2013; Poda et al., 2011), making it a promising technique for direct measurement of nZVI particles. Another technique, single particle ICP-MS (spICP-MS), is able to quantify the size and concentration of nanoparticles simultaneously, even at very low concentrations (Lee et al., 2014). Although, no application of these methods to nZVI detection has been reported, to our knowledge, it is possible that these ICP-MS based techniques may be suitable for nZVI detection, at least for laboratory-based research. Similarly, although not yet having been used for nZVI detection, advanced flow cytometry methods coupled with fluorescence and electrochemical detection may provide an efficient, realtime means of identifying the presence of nZVI in studies of groundwater applications.

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

6.2. Color and absorptivity

The presence of nZVI, especially at relatively high concentrations, can be judged simply based on the color of water samples (Johnson et al., 2013). Black or dark water samples from monitoring wells indicate the presence of nZVI, while lightly colored water indicates the absence of nZVI (if colorless) or presence of nZVIOX (if yellow) only. For freshly-prepared CMC-nZVI in laboratory, nZVI is visually evident at concentrations as low as ~10 mg L-1. In field samples, this number is expected to be higher due to the background color from natural dissolved or colloidal materials. Furthermore, when significant nZVI is apparent from inspection of water sample color (i.e., suspensions are black), it is possible that there has been significant oxidation of nZVI to nZVIOX (due to NRD) and visual inspection of sample color is not a sensitive, or quantitative, way to detect this change (Johnson et al., 2013).

Absorptivity measured by UV-Vis spectrometry can be used to quantify concentrations of nZVI in water samples, after appropriate calibration with nZVI standards (Johnson et al., 2013; Phenrat et al., 2007). Absorbance at 508 nm has been used for this purpose in a number of studies (Basnet et al., 2013; Kocur etal., 2013,2014; Phenrat et al., 2007; Saleh et al., 2008). However, this absorbance is fairly close to the isosbestic point of absorbance curves for nZVI oxidation to nZVIox (Fig. 3). Based on the results shown in Fig. 3, and additional considerations discussed in Johnson et al. (2013), absorbance at 800 nm can be attributed entirely to nZVI and we chose to use absorbance at 800 nm to characterize breakthrough of CMC-nZVI during a pilot scale, field test (Johnson et al., 2013). We found that this measurement — with calibration on appropriate standard solutions — compared to independent measurements of total iron by other colorimetric methods (see below) gave a sensitive and efficient protocol for characterizing the quantity and condition of transported nZVI under field conditions.

6.3. Total iron

Total iron measurement by conventional colorimetric methods is another routinely used method for monitoring

nZVI deployment in field. Given that total iron measurement can be done easily in both the laboratory and field, this approach has become one of the most commonly used methods to evaluate nZVI transport. This method can only be applied if the background concentration of dissolved (or colloidal) iron is low compared to nZVI dose, but this condition should be satisfied if the emplacement is at all successful. The major shortcoming of total iron measurement is that it does not differentiate nZVI from other oxidized iron species, such as dissolved Fe(II), Fe(II) or Fe(III) oxides. This issue has been mostly ignored in many column (and some field) studies of nZVI transport, some of which rely on total iron as the primary parameter to assess nZVI transport (Esfahani et al., 2014; Jiemvarangkul et al., 2011; Laumann et al., 2014; Raychoudhury et al., 2014; Wei and Li, 2013). Operationally, this might be justified because well-prepared and stabilized nZVI suspensions, particularly those of CMC-nZVI, cannot be separated into dissolved and particulate phases by centrifuga-tion or filtration. However, the small size and high reactivity of CMC-nZVI can magnify this problem because faster nZVI dissolution in water, as discussed earlier, might occur even within the time frame of typical laboratory column experiments. At the pilot and field scale, it has already been shown that the NRD of the subsurface porous media will oxidize nZVI all the way to Fe(III). Therefore, total iron measurement alone might result in significant overestimation of quantity of nZVI being delivered, and thus must be coupled with other techniques, such as UV-Vis or chemical redox probes (discussed later), in order to give a complete and unbiased characterization of nZVI transport.

7. Methods for indirect characterization of the fate and effects of nZVI

Most of the methods used for nZVI characterization are indirect, in which the presence or properties of nZVI was characterized based on the changes of water chemistry parameters (e.g. Fe2+, pH, DO, H2, CoCs) or electrometric properties of impacted environments (e.g. ORP, complex conductivity) due to reactions driven by nZVI. Individually,

Fig. 3. Absorbance of nZVI suspensions with various degrees of oxidation by titration with air (volumes of air are indicated in the legend). (a) Absorbance spectra with evaluated wavelengths marked with dashed lines and (b) response curves fitted to sigmoid calibration model (assuming the plateau absorbances reflect purely nZVI and nZVIox — left and right, respectively — and the oxidation reaction is simply nZVI ^ nZVIox).

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx 7

each of these parameters is evidence only for "impacts" of nZVI emplacement, not the actual presence of nZVI, or even nZVIox. However, when considered in combinations, as complementary indicators, the parameters can provide evidence for both the occurrence (delivery) and effects (reactivity) of nZVI in the injection zone and nearby areas of the subsurface.

7.1. Total dissolved iron

The dissolved product of nZVI oxidation is mainly Fe(II) species because Fe(III) is negligibly soluble in water. Increased concentrations of dissolved Fe(II) are usually observed during field deployments of nZVI, and dissolved Fe(II) can account for the majority of measured total iron in field water samples after nZVI injection (Elliott and Zhang, 2001). Upon further oxidation (e.g., due to NRD), Fe(II) is oxidized to various Fe(III) (hydr)oxides, depending on the reaction conditions (Fig. 1). These products have been characterized extensively in the laboratory, but much less data is available from field studies. The stability of dissolved Fe(II) and Fe(III) may be increased by ligands present in the groundwater, including dissolved organic matter.

To analyze total dissolved iron, water samples are filtered through membrane filters, typically with 0.2 or 0.45 |jm pore size. However, some nZVI particles stabilized by organic polymers, such as CMC-nZVI, are smaller than the pore size of most membrane filters used in the field, even after exposure to subsurface media (He et al., 2007, 2010; Kocur et al., 2014). Therefore, under these conditions, it is not feasible to separate truly dissolved iron from total iron. In addition, considering the abundance of iron minerals in natural environments and complex iron biogeochemical cycling in the subsurface, the total dissolved iron is likely affected by other background reactions/processes, especially at sample locations relatively distant from injection wells. In practice, dissolved iron analysis is often coupled with total iron measurements (as described above), with the difference providing one preliminary indicator of iron speciation.

7.2. pH

Corrosion of nZVI consumes H+ and/or releases OH~ (Fig. 1), usually resulting in elevated pH, with pH values of 9-11 being common in laboratory studies with unbuffered suspensions of nZVI (Zhang, 2003). The effect of nZVI on groundwater pH is moderated by the buffer capacity of the subsurface media as well as the distance of monitoring wells to the nZVI injection wells (Elliott and Zhang, 2001; Kocur et al., 2013; Wei et al., 2010). Therefore, while it is typical for pH to increase 2-3 units in the vicinity of injection wells, water samples from monitoring wells frequently show little or no change in pH (Wei et al., 2010). Where nZVI becomes completely oxidized, the groundwater pH could eventually recover to the pre-injection values. Taken together, these factors make pH changes an aggregate indicator of impacts from nZVI, but not one that is diagnostic of any particular effect.

7.3. Dissolved oxygen

The reaction between nZVI and DO is the fastest among all the reactions that typically contribute to NRD in groundwater,

and complete depletion of DO is often observed upon nZVI injection near the injection well (He et al., 2010; Johnson et al., 2013; Kocur et al., 2014). A rapid decrease of DO in ground-water is a good indicator of successful emplacement of nZVI, but not necessarily the breakthrough of nZVI, as the arrival of nZVI usually occurs after complete consumption of DO. Additionally, low values of DO do not necessarily indicate the persistence of nZVI within the impacted zones if there is significant NRD due to natural processes that may continue to react with nZVI. Dissolved oxygen at the monitoring wells may recover to background values with time, due to loss of reactivity for the injected nZVI and/or mixing with oxic groundwater.

7.4. Concentration of contaminants

Ultimately, the most important indicator of the success of nZVI emplacement is degradation of contaminants, the first indicator of which is usually lowering contaminant concentrations in monitoring well water samples. nZVI causes removal of most organic contaminants by reduction and inorganic contaminants by a combination of reduction and sequestration processes. The reductive dehalogenation of chlorinated organic compounds by Fe0 has been studied in detail since the early work by Matheson and Tratnyek (1994), and these reactions have been demonstrated using nZVI under field conditions (Elliott and Zhang, 2001; Glazier et al., 2003; He et al., 2010; Wei et al., 2010). The sequestration of heavy metals by nZVI is also very well characterized in laboratory studies (Li and Zhang, 2007; Ling and Zhang, 2014a,b; O'Carroll et al., 2013; Yan et al., 2012), and field applications are being explored.

In addition to degradation or sequestration by direct contact with Fe0 associated with nZVI, there are a variety ways that the effects of nZVI emplacement could indirectly cause contaminant removal. These include contaminant reaction with nZVI that is depleted of Fe0 but still composed of reactive and reducing solid phases, Fe(II) from Fe0 corrosion that has migrated to and adsorbed onto other aquifer solid surfaces, etc. These removal mechanisms still contribute to overall remediation of the site contamination, but complicate the interpretation of contaminant data as evidence for nZVI transport. Furthermore, nZVI emplacement could lower apparent contaminant concentrations simply by displacement or dilution of contaminated water, and these generally are not evidence of successful remediation.

7.5. Conservative tracers

Tracers that are non-reactive (i.e., not subject to significant sorption or transformation) usually have been used prior to and/or during nZVI injections to help determine the distribution of the mass of fluid injected during nZVI emplacement. This information is usually interpreted in the conventional fashion: as an indication of the hydrological response of the system to perturbation by fluid injection. The traditional conservative tracers, such as bromide ions and fluorescein, have been applied in this way (He et al., 2010; Johnson et al., 2013; Kocur et al., 2014), although it must be recognized that the high reactivity of nZVI could cause transformation of some tracers (e.g., degradation and quenching of fluorescein fluorescence). Additionally, some species that arise from the preparation

of nZVI, such as sulfate, borate, and chloride ions could serve as "semi-conservative" tracers of nZVI impacted fluids in situ. In some cases, when nZVI is formulated with a distinctive carrier material, the carrier can be used as a tracer, such as has been done with nZVI supported on carbon colloids (Busch et al., 2014a,b; Mackenzie et al., 2012).

7.6. Dissolved hydrogen

The concentration of dissolved H2 in groundwater is usually low, because it is limited by various microbiological processes including denitrification, iron reduction, sulfate reduction, and methanogenesis (Blodau, 2011). However, the reaction between nZVI and water can produce significant quantities of H2, so elevated H2 concentrations in groundwater might be observed as a result of nZVI injection. With time, H2 concentrations are expected to be lowered back to background levels due to microbial utilization of H2 as an electron donor. While measuring H2 concentration is not often done during nZVI emplacement, redox potentials are often measured and they are strongly affected by H2. This was demonstrated in our previous study (Shi et al., 2011) and is discussed below, in Section 7.8.

Another use of measured H2 concentrations is worth mentioning here because it is complementary. Production of H2 by reacting nZVI with acid has been used as an assay for the Fe0 content of nZVI (Kocur et al., 2014; Phenrat et al., 2009), based on the assumption that none of the iron (hydr)oxides that result from Fe0 are capable of further reduction of water. This assay has been implemented in a commercial device for verifying the integrity of nZVI preparations in the field before deployment (http://www.nanoiron.cz/en/nzvi-tester).

77. Redox active probes

Redox active chemical probes are compounds that undergo fast, usually reversible, and diagnostic reactions in response to redox conditions. In most cases, the oxidized and reduced forms of a probe redox couple have distinct UV-Vis absorbance or fluorescence spectra, which allows direct and easy monitoring of redox conditions by spectrophotometry. Redox active chemical probes have long been used to characterize the redox conditions of biogeochemical systems, including sediments (Tratnyek et al., 2001), groundwater (Jones and Ingle, 2001, 2005) and, more recently, Fe(II) bearing pure mineral phases (Orsetti et al., 2013). However, applications of this approach have mostly been limited to laboratory studies and its potential application to detection and characterization of nZVI in field applications has never been explored.

A major advantage that redox probes have over colorimetric iron assays and electrode measurements of redox potential (discussed above and below, respectively) is the possibility that chemical redox probes can be designed to differentiate among various redox active species in multi-component media, based on the reactivity between chemical probes and various environmental reductants. For example, it is shown later that both nZVI and H2 give strongly negative ORP when measured with a platinum (Pt) electrode. But while nZVI rapidly reduces chemical redox probes, the redox probes do not react with dissolved H2 (alone, in the absence of activation, such as with a catalytic noble metal). Fig. 4 illustrates another example, where

S 1.5 @

0 1 2 3 4 5 6 7 time (min)

Fig. 4. Reduction kinetics of indigo disulfonate (I2S) by mixtures composed of different amounts of CMC-nZVI and aqueous Fe(II) with 10 mg L—1 total iron concentration monitored by UV-Vis spectrophotometry (pH = 7.2 buffered by 10 mM HEPES; initial [I2S] = 140 |M).

indigo disulfonate (I2S) (E7 = — 0.129 V) is used as a chemical redox probe to differentiate between reduction by aqueous Fe(II) and by Fe0 from CMC-nZVI. In these batch kinetic experiments, I2S was added to the mixtures composed of different amounts of CMC-nZVI and aqueous Fe(II) with the same total iron concentration equal to 10 mg L—1. Over time, the change in absorbance was monitored at 610 nm, which is the maximum absorbance wavelength for oxidized I2S (Tratnyek et al., 2001). It can be seen that I2S was not reduced at all by aqueous Fe(II), but increasing the concentration of CMC-nZVI reduced the plateau absorbance, suggesting more total reduction of I2S. Thus, these results allow the kinetic (reduction rate constant) and capacity (plateau concentration) aspects to be extracted from the data to quantify the amount of Fe0. The selectivity of I2S to Fe0 may prove to be a useful way to distinguish Fe0 and aqueous Fe(II), which is difficult to achieve by solution iron or ORP measurements. Because the concentration of this probe can easily be measured with a field portable UV-Vis spectrophotometer, this technique can potentially be deployed in the field to monitor delivery and transformation of CMC-nZVI. However, achieving the full potential of chemical redox probes for characterizing in situ redox conditions will require further development and validation.

7.8. Potentiometry and voltammetry: ORP measurements

Potentiometry — especially for measurement of ORP — is one of the most widely used indirect methods for nZVI detection in the field. Despite the complexity of interpreting ORP results (O'Carroll et al., 2013; Shi et al., 2011), complementary potentiometric and voltammetric methods, coupled with iron analysis, are useful because they provide insights into the reactivity and transport of nZVI in porous media.

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

The introduction of nZVI into the subsurface should create strongly reducing conditions in the impacted region, and this region is often demarcated by highly negative ORP values obtained in monitoring well water samples. These ORP measurements have almost always been made with conventional combination ORP electrodes, which contain a Pt working electrode and a reference cell. ORP values measured in this way usually are very low (typically — 400 to — 600 mV vs. SHE) near the injection well and increase with distance from the injection point, eventually up to levels typical of unimpacted groundwater at the site (Elliott and Zhang, 2001; Glazier et al., 2003; He et al., 2010; Henn and Waddill, 2006; Johnson et al., 2013; Kocur et al., 2014). Despite the apparent simplicity of this method, overall conformity of the results with expectations, and frequent application of this approach in laboratory and field studies, there are a multitude of potential pitfalls that must be avoided if ORP is to be a useful method for characterizing nZVI emplacement.

Many of the fundamental issues with interpretation of measured ORPs stem from the fact that they are mixed potentials (Emix) that include the contributions of all the redox couples in the system weighted by their exchange current densities (i0), which effectively quantifies the sensitivity of the electrode to each redox couple. In the standard definition of Emix (Eq. (1)), it is also assumed that each individual redox couple is described by the Nernst equation, although this is not usually true for some relevant redox couples (nitrate/ammonia, sulfate/sulfide, etc.).

E -VJ!-^-RT. {Redi} 1 m

Emix X|i?|ri niF {°xi}{H+}^ 1

For each half-reaction between reduced species (Redi) and oxidized species (Ox) Eq. (1) includes the corresponding values of exchange current density (i,0), and the stoichiometric coefficients for electrons (ni) and protons (a). The implications of Eq. (1) for the interpretation of ORP measurements in nZVI impacted waters were developed thoroughly in Shi et al. (2011) and these results have been applied in several more recent studies (Adeleye et al., 2013; Su et al., 2014). In general, the major half reactions contributing to Emix in the presence of

nZVI include dissolved Fen, H2, and, to lesser degree, reduced S or B species, if they are present.

In systems containing suspended nZVI, we showed that an additional factor plays a very large role in affecting ORP measurement: the deposition of nanoparticulate iron onto the electrode surface, resulting in a surface film that can transform the response of the ORP electrode (Shi et al., 2011). The range of possible interactions that the electrode responds to is summarized in Fig. 5. To resolve the contributions ofthese interactions, we have made ORP measurements using a variety of electrode configurations: including stationary electrodes (STEs), rotating disc electrodes (RDEs) (Shi et al., 2011), and flow-through cell electrodes (FCEs) (Johnson et al., 2013). A cross-section of new and old results obtained by these methods is discussed in the following three sections.

7.8.1. ORP: conventional stationary electrodes

For conventional ORP measurements done with a Pt electrode in the field (e.g., by immersing the Pt electrode into a freshly collected sample of groundwater), the effects of nZVI particle settling and the deposition of nZVI on electrode surfaces are usually overlooked. It is likely, however, that these effects are not negligible, based on the general considerations summarized in Fig. 5. To evaluate the significance of these effects, and develop a verified basis for interpreting ORP measurements in nZVI containing systems, data from controlled experimental measurements are essential. We started an extensive study of ORP measurements on nZVI suspensions in Shi et al. (2011) and additional results are reported below.

Fig. 6 demonstrates the effect of settling on ORP measured in suspensions of nZVI. In this experiment, two Pt STEs were positioned at different depths in an unstirred nZVI suspension and ORP was recorded at both electrodes over time. The reactor headspace was purged with Ar to avoid O2 diffusion into the reactor. During the measurement period, settling of the nZVI was visually evident and the ORPs rebounded after the front of particles settled below each Pt electrode surface (Fig. 6). Clearly this result demonstrates that ORP measurements are very sensitive to the proximity of nZVI to the electrode, although the detailed interpretation of this change in electrode response must take into account the whole range

Fig. 5. Interactions/reactions between an electrode, aqueous solution, and nanoparticles of zerovalent iron. Reprinted with permission from Shi et al. (2011 ). Copyright 2011 American Chemical Society.

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Fig. 6. ORP measured by two Pt electrodes located at different depths in an unstirred nZVI suspension. The suspension was fully mixed when the electrodes were first immersed (t = 0) and the front of settling nZVI passed the top and bottom electrodes at -325 and 850 min, respectively. The suspension was 50 mg L—1 RNIP-10DP (same as in Shi et al. (2011)) in 3.33 mM bicarbonate buffered DO/DI water.

of effects shown in Fig. 5 (Shi et al., 2011). The practical implication of this result is that ORP measurements on nZVI impacted water samples must be done in a way that consistently controls the effects of particle settling (and probably also aggregation and agglomeration).

7.8.2. ORP: rotating disc electrodes

An RDE provides controlled mixing, and we found that well mixed suspensions of nZVI can be obtained at rotation rates greater than 4000 RPM with a 3 mm RDE (using 200 mg L-1 RNIP-10DP) and other conditions detailed in Shi et al. (2011). Under these conditions, the effect of nZVI settling is minimized, and interfacial mass transfer at the electrode is maximized, which allows controlled and reproducible ORP measurements (Shi et al., 2011), thereby providing the best opportunity for rigorous interpretation of the results of ORP measurements of nZVI suspensions.

During measurements with an RDE, the ORP of the nZVI suspension decreased quickly with time and then approached the lowest value (ORPmin) under anaerobic conditions (Shi et al., 2011 ). The relationship between ORPmin values and nZVI concentrations was established based on nZVI coverage on the electrode surface (Shi et al., 2011),

ORPmin =

ORPlim K'djnZVI] 1 + K 'd[nZVI\

in which ORPmin is the lowest ORP value recorded by the electrode at one specific nZVI concentration. The ORP]im and Kd' are model parameters that depend on the properties of nZVI, the electrode, and the operational conditions.

As indicated by Eq. (2), the ORPmin values gave a Langmuir type dependence on the nZVI concentrations: ORPmin values quickly decrease with the increase of nZVI concentrations

when nZVI concentrations are low and the further increase in nZVI concentrations does not reduce ORPmin values proportionally. While Eq. (2) establishes the direct relationship between ORP measurements and nZVI concentrations, the oxidation of nZVI likely affects the ORP measurements with time. Even at very low nZVI concentrations (e.g. < 50 mg L—1), our RDE measurements showed lower ORP values (Shi et al., 2011) than those observed in field monitoring wells close to the injection wells with significant nZVI breakthrough (e.g. Johnson et al. (2013)). This suggests that the deposition of oxidized nZVI particles on the electrode surface is important in interpreting ORP measurements under field conditions, and this important conclusion is discussed further below. Building on this result, time-dependent ORP measurements have recently been used to explore the reactions among Pb2+, NO—, and nZVI particles, since ORP was affected by the transformation of nZVI particles deposited on the electrode surfaces and the different redox reactions involving Pb2+, NO——, and iron species in the measured systems (Su et al., 2014).

To further investigate how ORP measurements are influenced by transformation of nZVI particles deposited on the electrode surface, we conducted linear sweep voltammetry (LSV) experiments in the same RDE reactor that we reported in our previous study (Shi et al., 2011). In these experiments, the reactor headspace was purged with Ar, and then the purging was stopped to allow air to slowly reenter the cell and oxidize the nZVI. At selected times, LSVs were measured, by scanning from —1.0 to 0 V (vs. Ag/AgCl) at 10 mV s—1. That LSV data are presented below as a Tafel plot (log |i| vs. E), where the corrosion potentials (Ecorr) are located at the singularities (where i = 0). Initially, the ORP recorded by RDE decreased quickly and approached a minimum value around — 645 mV, consistent with ORP measurements made by other methods (e.g., as shown in Fig. 6). The small peak at 7 min is due to the LSV experiment performed at that time. The ORP only stayed at this low value briefly, and then rebounded (Fig. 7a), in this case due to the oxidation of nZVI. At selected times along the curve in Fig. 7a, Ecorr was recorded, and it shifted toward more positive values with time (Fig. 7b). The shift in Ecorr suggests progressive oxidation of nZVI particles that deposit onto the electrode surface, which in turn suggests that some of the shift in ORP shown in Fig. 7a is due to changes in the composition of the nZVI in the reactor. These two sets of potentials — ORP obtained with the RDE and Ecorr obtained by LSV — form a nice linear correlation (Fig. 7c) over the range of the data collected. Deviations from this linear correlation are expected, however, over a broader range of conditions, where there is more opportunity for differential contributions of the various effects represented in Fig. 5, such as the kinetic effect of electrode responses in nZVI suspensions during the first few minutes (Shi et al., 2011).

7.8.3. ORP: flow-through cell electrodes

To extend the results we obtained with an RDE (Shi et al., 2011) to flow conditions (to simulate a scenario where groundwater from the monitoring well is analyzed continuously), we used an FCE system to measure the ORP of sample well waters pumped during a pilot scale nZVI injection test (Johnson et al., 2013). The FCE system was set up with two 3 mm working electrodes, one Pt and one glassy carbon (GC), which allowed simultaneous monitoring of ORP continuously

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Fig. 7. Comparison of (a) ORP obtained with a Pt RDE, (b) Ecorr obtained by LSV, and (c) the correlation between the two. Time in this experiment represents the nZVI oxidation due to oxygen intrusion. Performed with 200 mg L-1 RNIP-10DP in 3 mM bicarbonate media.

throughout the experiment with both electrode materials. In other aspects, the FCE is similar to the RDE in that it should minimize the effects of the mass transfer processes and nZVI settling. Since the exchange current densities (i0) for the various redox couples are different between Pt and GC, the FCE system configured with both working electrodes has the potential to help characterize a number of effects of nZVI that are not well understood, including those involving dissolved H2, NRD, and nZVI deposition onto ORP probes.

Combining ORP measurements using an FCE and total iron analysis, we first investigated the transport of two representative nZVI suspensions, RNIP-M2 and CMC-nZVI, in a clear-glass laboratory column (2.54 cm i.d. glass column, 15.2 cm length) packed with 0.5 mm glass beads. Before each experiment, the column was flushed with 3.33 mM deoxygenated NaHCO3 to remove any background turbidity and dissolved O2. The FCE (with freshly polished working electrodes) was connected to the column effluent and ORP was monitored continuously. The FCE effluent was collected with a fraction collector, with 5 min time steps, and analyzed for total iron. Each experiment was initiated by injecting the nZVI suspension into the column using a ceramic pistol pump at the fixed flow rate (1 mL min-1) from a reservoir of nZVI suspension. The nZVI suspension was continuously sparged with Ar and mixed by an RDE, which also served to monitor the ORP of the nZVI suspension (Shi et al., 2011).

The results of these column tests (Fig. 8) show sharp changes in ORP and total iron concentrations approximately when a front of weakly retarded species would be expected to breakthrough (in this case, 0.64 pore volumes « 20 min). Note that the ORP dropped and the total iron concentration increased at approximately the same time. Using CMC-nZVI, both GC and Pt electrodes give ORP values that agree closely. The low plateau values, around - 600 mV vs. Ag/AgCl, are typical for ORP of nZVI suspensions under anoxic conditions with negligible NRD. In contrast, RNIP-M2 was less mobile in the column, with most of the nZVI depositing in the first 4 cm of the column (based on visible color). The ORP of this column's effluent only decreased to about - 300 mV vs. Ag/AgCl, suggesting that the dominant redox couple under these conditions was Fe2+/Fe3+. As with CMC-nZVI, there was little difference between ORP values of GC and Pt electrodes with RNIP-M2.

Since RNIP-M2 was less mobile, we selected CMC-nZVI for use in most subsequent column studies, following protocols similar to those used for Fig. 8, except using the columns (2.22 cm i.d. plastic column, 17.8 cm length) packed with sand from the large experimental aquifer program (LEAP) used in our previous field-scale tests (grain size = 0.3 mm; hydraulic conductivity = 0.02 cm s-1) (Johnson et al., 2013). Generally, the results show patterns of ORP changes (Fig. 9a) and total iron breakthrough (Fig. 9b) with time that are similar to what we observed in the glass bead column experiments with CMC-nZVI injection. The breakthrough of total iron, as expected, was proportional to the initial nZVI concentrations injected into the column (Fig. 9b). The ORP values stayed low even after total iron breakthrough dropped, mainly due to the coating of iron nanoparticles (nZVI and nZVIox) deposited on the electrode surface. However, the results also show that varying the influent concentration of CMC-nZVI had little effect on the ORP of the GC or Pt FCE (Fig. 9a), probably because these experiments were all done at relatively high nZVI doses, which corresponds to the ORP]im case described by Eq. (2). The small

Fig. 8. Comparison between ORP (obtained with an FCE containing Pt and GC working electrodes) and total iron after the addition of nZVI (CMC-nZVI and RNIP-M2) to a glass column (2.54 cm i.d., 15.2 cm length) packed with 0.5 mm glass beads. Flow rate was 1 mL min-1.

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Fig. 9. Breakthrough curves obtained by introducing CMC-nZVI in columns (2.22 cm i.d. and 17.8 cm length) packed with LEAP sand, monitored by (a) ORP with an FCE and (b) total iron in the effluent. nZVI concentrations are shown in the legend. Flow rate = 0.84 mL min-1.

difference in ORPmin with nZVI concentration (Fig. 9a) may also be because CMC-nZVI was rapidly oxidized by the sand (due to the NRD of this material), resulting in a thin layer of nZVIox coating on the electrode surface, which created a constant environment at the electrode interface affecting the ORP measurement regardless of the total quantity of iron in the columns.

Another major feature of the data in Fig. 9a is that the Pt FCE gave significantly lower ORP values compared with that of the GC FCE. This difference can be attributed to the effects of both dissolved H2 as discussed in our previous study (Shi et al., 2011) and NRD of the sand, which results in the formation of iron oxide coating on the electrode surface. Under the conditions of these experiments, dissolved H2 dominated Pt FCE response and resulted in low ORP values with the Pt FCE, irrespective the iron oxide coating. Since the GC FCE is insensitive to dissolved H2, the iron oxide coating dominates the response, resulted in significant higher ORP. Additional evidence for oxidation of nZVI by the sand was presented in our previous study on nZVI transport at the field scale (Johnson et al., 2013). As a further check on our interpretations of the FCE electrode response in terms of iron oxide coatings and dissolved H2, we conducted experiments in which we polished the FCE at intermediate times during the column experiments. As shown in Fig. 9a for 756 mg L—1 CMC-nZVI, polishing the FCE resulted in a significant drop in ORP measured with the GC FCE (due to removal the iron oxides coating), and little change in the ORP of the Pt FCE (because dissolved H2 is unaffected by the surface coating).

Our interpretation of the results obtained with the FCE and sand columns (Fig. 9a) also provides additional insight into the results obtained in the glass bead column experiments (Fig. 8) where NRD should be negligible. For the glass-bead column experiments with CMC-nZVI, the measured ORP values are relatively negative and consistent with the range of expected ORPs calculated for the dissolved H2 concentrations in waters impacted by ZVI corrosion (as described in Shi et al. (2011) and elaborated further below). However, in contrast to the results obtained with the sand column, the glass bed column gave little difference in the ORP measured with Pt and GC (cf. Figs. 8 and 9). This result is consistent with the lack of NRD in the glass bead column, so the nZVI was sufficient to produce the very

low ORP of fully reduced nZVI, thereby masking the different contributions of H2 to the mixed potentials at Pt vs. GC.

To further investigate the effect of NRD in the LEAP sand on nZVI transport, we conducted column experiments similar to those shown in Fig. 9, but with nZVI introduced as a series of four 10 min CMC-nZVI pulses (8.4 mLof 70 mg L—1 CMC-nZVI for each pulse). The pulses were applied consecutively without disturbing the column, but the FCE electrodes were polished between each pulse (the FCE cell is external to the column). We hypothesized that the NRD of the LEAP sand would be consumed by the sequential nZVI injections, thereby allowing more reduced nZVI to reach the end of the column and decreasing the amount of nZVIox in the column effluent. Based on the results of experiments described above, we expect that the decrease in nZVI to nZVIox conversion in the column would result in more reduced (or less oxidized) material depositing on the electrodes in the FCE, and that this would cause the ORP measured by GC FCE to become more negative with each pulse/ cleaning cycle (because it is dominated by the iron phases); whereas there would be little change in ORP from Pt FCE (which is already low due to the contribution of H2).

The data obtained from the pulsed column experiments (Fig. 10) support the hypothesized behavior. There was little difference between ORP measured by the Pt FCE for different nZVI injections (Fig. 10a), consistent with Emix at the Pt FCE being dominated by dissolved H2 and relatively unaffected by the extent of iron oxide formation on the electrode surface. For the GC FCE, which is insensitive to the H2/H+ redox couple, the ORP kinetic curves shifted down with more nZVI pulse injections (Fig. 10b) because there is more breakthrough of nZVI from the columns after previous injections have consumed the NRD of the column packing material. The lowest ORP observed during each nZVI pulse injection were close for the first two injections, about — 200 mV, indicating little breakthrough of nZVI and that the GC FCE response was dominated by Fe2+/Fe3+ redox couples. After two nZVI injections, the minimum ORP gradually decreased to about — 500 mV, indicating an increasing contribution of nZVI to GC FCE response.

Overall, the results of monitoring column effluent ORP using an FCE (Figs. 8-10) suggest that this may be a worthwhile way

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Fig. 10. Continuous monitoring of ORP of CMC-nZVI during column transport with pulse nZVI injection: (a) PtFCEand (b) GCFCE. For each injection, 70 mgL—1 CMC-nZVI was injected for 10 min and then followed by 110 min 3.33 mM NaHCO3. The same injection was repeated four times and the FCE was polished between injections. Flow rate is 0.84 mL min-1.

to obtain real-time evidence on breakthrough of nZVI impacted fluids. Comparison of ORP measured with Pt and GC in an FCE allows characterization of the extent of iron oxide deposition on the electrode surface and this can serve as an indicator of the aquifer NRD and the degree of nZVI oxidation by the NRD. However, correct interpretation of these data requires careful consideration of the complicating factors discussed above: including the selective response of different electrode materials to various redox couples, the unknown and possibly variable contributions of some redox couples to the electrode mixed potential, variable sensitivities of electrodes to particles with different size distributions, and the deposition of nZVI or its oxidation products onto the electrode surface resulting in a surface with altered properties. The detailed interpretation of

ORP response provided here for this (relatively controlled) experiment was only possible with the complementary information provided by simultaneous measurement of ORP with Pt and GC electrodes.

One aspect of this problem that we have investigated in even further detail is the contribution of dissolved H2 (from corrosion of Fe0) to the measured ORP (Shi et al., 2011). The theoretical Eh of the H+/H2 couple was calculated as a function of pH2 and pH (Fig. 11), and we annotated this result with approximate regions corresponding to relevant environmental scenarios. When controlled by natural microbiological processes, the calculated Eh based for the H+/H2 couple is between — 0.3 V and — 0.6 V (vs. Ag/AgCl) for typical groundwater pHs; whereas for groundwater impacted by Fe0 corrosion, both pH

Fig. 11. The effect of dissolved [H2] on ORP of Pt electrode calculated with Nernst equation at various pH values (1 atm and 22 °C). Modified from Shi et al. (2011).

Fig. 12. Impact of varying dissolved [H2] on ORP measured with a Pt flow-through cell electrode (3 mM NaHCO3 at pH = 8.4 and flow rate = 2 mLmin-1).

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

and H2 concentrations will be higher, which corresponds to calculated Eh values between — 0.6 V and — 0.8 V (vs. Ag/AgCl) (Fig. 11).

To test the theoretical model represented by Fig. 11, we measured ORP as a function of H2 concentration using a Pt FCE, and the results are shown in Fig. 12. For these experiments, H2 gas was dissolved in the 3.33 mM NaHCO3 solution at pH 8.4 and then the solution was continuously pumped into the FCE at a fixed flow rate (2 mL min-1). The ORP measured with the FCE was continuously recorded over time. As shown in Fig. 12, the ORP measurements were affected by both the H2 concentration, as expected based on the Nernst equation, and the kinetics, presumably due to the mass transfer process. At relatively high H2 concentrations, the ORP of the Pt electrode can approach a minimal value within a few minutes. Therefore, it is expected that, in the nZVl suspensions, dissolved H2 will play a major role in determining the response of the Pt electrode (Shi et al., 2011). However, in the field, for monitoring wells in which no significant nZVl was observed, the dissolved H2 concentrations are likely low and the impact on ORP measurements is expected to be minimal because of slow kinetics at low H2 concentrations (Fig. 12).

Overall, our studies on ORP measurements using STE, RDE, and FCE highlight the precautions that must be taken when interpreting ORP measurements of water samples containing nZVl in both laboratory and field studies. ln addition to the mixed potentials arising from multiple redox couples, the deposition of nZVl particles on the electrode surface and the settling of nZVl particles should be carefully considered when conducting ORP measurements and interpreting the results. Dissolved H2 appears to play an important role in affecting Pt electrode response, but its impact is dependent on the H2 concentrations (and therefore the concentrations of nZVl) and kinetics. Based on our results, electrode cleaning prior to and during the measurements is crucial in order to evaluate the reactivity of nZVl in water samples.

7.9. Geophysical methods

The technologies discussed above for studying the fate and effects of nZVl during its transport and interaction with the surrounding environment all require direct access to samples from the subsurface. Therefore, the results from these methods represent discrete points in space and (usually) time, which means they provide relatively low resolution with respect to the spatial and temporal dynamics that are expected during, and soon after, the emplacement of nZVl into the subsurface.

Geophysical methods are different in that they provide in situ characterization of subsurface conditions with continuous resolution (in space and/or time). Although most geophysical methods are indirect — in that they usually require petro-physical models in order to infer physicochemical properties from the measured data — these methods can provide characterization of large study domains efficiently and remotely (i.e., without direct contact with the imaging target).

Geophysical methods have been used to study a variety of hydrogeological and geomicrobiological processes in shallow subsurface (see reviews by Atekwana and Slater (2009), Rubin and Hubbard (2005) and Vereecken et al. (2004)). More recently, geophysical methods have also been used to describe installations of in situ permeable reactive barriers (PRBs) and

even to provide evidence for changes in conditions within the PRB (Slater et al., 2005; Wu et al., 2005, 2006, 2008, 2009). Additionally, there is some evidence from laboratory studies that geophysical methods can detect nanoparticles in porous media (Joyce et al., 2012), but the application of these methods to deployments of nZVl in groundwater remediation applications has received little consideration to date.

Because the intrinsic electrical and magnetic properties of ZVl materials are significantly different than that of the natural soils, these properties are promising targets for geophysical methods that track nZVl distribution. Magnetic susceptibility measurements have been used in single field borehole and laboratory experiments to detect the presence of nZVl (Kober et al., 2014; Vecchia et al., 2009). Few studies have attempted to apply this method at larger scale (Buchau et al., 2010), in part because these applications require the detector be placed close to the vicinity of the nZVl. Comparing with magnetic susceptibility measurements, electrical geophysical methods, especially complex conductivity (or resistivity), have been used more frequently for the study of ZVl materials. Complex conductivity (o*) measures the complex electrical conduction behavior of the study material upon excitation by an external current source. lt can be expressed in terms of a magnitude (|o|) and a phase shift or real (o') and imaginary (o") components:

o* =1 = ias* = o '( ( + io "( ( pt

= r^ = Vo '• lp*l

VP'2 + P"2

^ = tan

= — tan

o " ' o '

where p* is the complex resistivity, p' and p" are the real and imaginary resistivities, e* is the permittivity, a is the angular frequency and i = 1.

The conductivity magnitude (|o|), or the real conductivity (o'), measures the ability of the material to conduct charges, and the phase shift ($), or imaginary component (o"), measures the magnitude of charge polarization that occurs at mineral grain/electrolyte interfaces mainly at low frequencies. While values at single frequencies (e.g. at 0.1,1, or 10 Hz) are often used to describe the electrical properties of a material, these parameters are frequency dependent, so the choice of the frequency to use can impact the magnitude of the signals significantly. An alternative method is to acquire the signals across a broad frequency spectrum (e.g., from 0.1 to 1000 Hz) and then invert the spectral data for global parameters that are frequency independent. A single Cole-Cole type dispersion model (Cole and Cole, 1941) is often used to invert for a global conduction (often denoted as o0) and polarization (mn) parameters (Lesmes and Frye, 2001) based on the following equation (Jones, 2002),

o *(a) = o 0

(iar)c

1 + (iar)c(1—m)

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

where o0 is the conductivity at DC frequency, t is the mean relaxation time (a parameter relevant to the characteristic particle or pore size), c is a shape exponent (typically 0.1-0.6) and m is the chargeability, a measure of the polarization magnitude (mn = m x o0).

The conductivity and polarization responses of a material are functions of many parameters that include soil matrix characteristics (e.g. permeability, saturation level, pore fluid salinity) as well as the properties of the electrical double layer (EDL) at mineral/water interfaces. For ZVI, these properties are also impacted by the charge transfer processes across the mineral/water interface during redox reactions, which is illustrated in Fig. 13 (Slater et al., 2005).

While it is typical for most natural materials to have a phase response in the range of a few to 10-20 milli-radians (mrad), the phase response from ZVI can easily exceed 100 mrad depending on its concentration, thus providing enough contrast with its surrounding environments for electrical imaging (Slater and Binley, 2003; Slater et al., 2005; Wu et al., 2005, 2006). Extending this to mixtures of (micron and millimeter sized) ZVI and Ottawa sand, Slater et al. (2005) found close correlations between electrical properties and the quantity of ZVI (Fig. 14). The use of surface area of ZVI per pore volume (SFe(pore)) in Fig. 14 as the independent variable is based on the fact that electrical charge polarization occurs on mineral surface, thus mineral surface area is one of the major factors controlling the magnitude of the polarization signal. Fig. 14 shows a linear correlation between SFe(pore) and mn (or o") (R2 > 0.97) indicating the dominant control of ZVI on the electrical signals of the mixture with negligible

Fig. 14. Conductivity (o0), induced polarization (mn and oi Hz") and the relaxation time constant (Tn) changes as a function of ZVI surface area per unit pore volume (SFe(pore)) for Ottawa sand-ZVI mixtures that vary from 0.1% to 100% ZVI volume concentration. Figure adapted from Slater et al. (2005).

contributions from Ottawa sand. Note that the conductivity (a0) of the mixture only starts to increase significantly when the ZVI volumetric fraction is above 20%, an indication of possible bridging between the ZVI particles above a critical threshold to form continuous electronic pathways. A concurrent and abrupt increase of the conductivity normalized

Fig. 13. Conceptual model for possible polarization mechanisms in metal-particle-containing soils such as sand-Fe0 mixtures. (a) Without applied electric field (E = 0); (b) with applied electric field (E ^ 0). IP1 represents the diffusive polarization mechanism along metal surface; IP2 represents the redox reaction related polarization mechanism. Symbols are r = metallic particle radius; 1/k = EDL thickness; and refr = r + f (1/k) is the postulated polarization sphere radius sensed with IP. The electric double layer thickness is exaggerated for illustration purposes. Figure adapted from Slater et al. (2005).

ARTICLE IN PRESS

Z. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

relaxation time constant (Tn = t x o0) as shown in Fig. 14 is consistent with this hypothesis. This study also investigated the effects of the electrolyte concentration on the polarization magnitude and time constant. They found a linear correlation between electrolyte activity and the polarization magnitude (mn) for a few types of electrolytes and also observed a directional (from decrease to increase) change in the relaxation time constant (t) with increasing electrolyte activity above a certain threshold, which are attributed to possible changes in charge mobility (Lesmes and Frye, 2001).

The properties determined by complex resistivity (or conductivity) methods may provide information not only on the quantity of ZVI present, but also on the condition and reactivity of the ZVI. The latter is expected because all the electrical parameters (|o|, o', o", mn, and t) are partly determined by electrical properties of the particle-water interface (Wong, 1979) that can be altered by the ambient condition and its impacts on the reactivity of the ZVI. To explore this possibility, electrical measurements were performed with laboratory columns packed with ZVI/sand mixtures and cores extracted from a PRB in Kansas City, Missouri (Wu et al., 2005, 2006, 2008, 2009). During these experiments, mineral precipitation was induced by controlling pore fluid chemistry (e.g., manipulating concentrations of carbonate vs. sulfate). The results showed that mineral precipitation affected the apparent electrical properties of these materials by (i) changing the surface area of ZVI due to the precipitation of a variety of iron oxides on ZVI surfaces and (ii) altering the mineralogy of precipitated phases on the ZVI surface. Specifically, the precipitation of calcite tends to reduce the polarization responses of ZVI while magnetite/maghemite precipitation can enhance such a signal due to the differences ofthe electrical properties between calcite and magnetite and how they compare to the electrical properties of ZVI (Wu et al., 2006).

Field applications of electrical methods to characterize ZVI emplacement for subsurface remediation have also been carried out (Slater and Binley, 2003). In this field study, crosshole electrical tomographic imaging was conducted on an eight-year old ZVI PRB, and the results demonstrated that both electrical resistivity and induced polarization could be used to delineate the boundary of the PRB. The high conductivity values closely tracked the geometry of the ZVI according to the engineering construction plan and the phase shift signals provided a good representation of the PRB geometry as well.

The geophysical studies discussed above were focused on conventional PRBs where the granular ZVI has average particle size of 0.4 to 1 mm. Emplacement of nZVI is sometimes done as a gallery of sequential injections that aims to produce a treatment zone that is similar to a PRB, so geophysical methods may apply to nZVI emplacements in a manner similar to what has been demonstrated for courser-grained ZVI. This possibility has received very little investigation, however, and there is only one published study to date (Joyce et al., 2012). In that study, induced polarization measurements were made on nZVI/sand mixtures (and a few other types of relevant nanoparticles) in laboratory column experiments. They observed a significant phase response in the range of tens of mrad with a 20% nZVI content by weight with nZVI that was fresh, without additives, and mechanically mixed with clean sand. In field applications of nZVI, however, the material is coated with surfactants or

other polyelectrolytes to improve stability and mobility of the suspended solids. These additives are likely to alter the material's electrical properties significantly because these organic surfactants are normally nonconductive, thus shielding the electrical effects of nZVI itself.

To investigate the sensitivity of geophysical methods to nZVI under more realistic conditions, we conducted preliminary experiments using nZVI coated with insulative mineral oil. The coated nZVI slurry was injected into a sand packed, water saturated column until a total of 1.5% by weight nZVI was injected. After electrical measurements were collected on day 0, a 1% by weight H2O2 solution was injected started from day 1 and continued until day 13 (at 0.5 mL h-1) with an intention to stimulate Fenton's reactions in order to oxidize the surface mineral oil coating and thereby reexposing the nZVI. A phase response below 0.5 mrad (similar to pre-injection condition) was observed after the initial injection of the nZVI (day 0 in Fig. 15). Subsequent H2O2 fluid injection for the next 13 days was accompanied with increasing phase shift responses over time indicating possible exposure (and oxidation) of the nZVI. These preliminary results indicate that although induced polarization signals might be used to track nZVI after its surface has been exposed, its usage for tracking the initial injection of surfactant coated nZVI could be limited. However, it is worth pointing out that our study was preliminary and we did not characterize the electrical conductivity signals from surfactant coated nZVI. The use of nZVI and the addition of surfactant could potentially change electrical conductivity of the nZVI slurry significantly, making it a more suitable target for resistivity imaging during the initial injection phase.

In conclusion, available studies in the lab and field have illustrated the sensitivity of electrical geophysical signals not only to the existence and placement of ZVI but also to the corrosion and precipitation processes that significantly impacts its reactivity. Such results demonstrated the potential of

Fig. 15. Induced polarization phase shift from nZVI slurry injected into a sand packed, water saturated column. A total of 1.5% by weight nZVI was injected. The surface of nZVI was coated with mineral oil and a 1% hydrogen peroxide solution was injected at 0.5 mL h-1, from day 1 to 13.

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

electrical geophysical methods for monitoring field scale in-situ (n)ZVl installment and its fate and transport over time. However, field testing of these methods has been limited so far, particularly for nZVl applications. Further investigation should focus on coupling these geophysical methods with other direct measurement technologies discussed in previous sections to test their capabilities for tracking the installation, fate, and transport of nZVl at both lab and field scales.

8. Implications and recommendations

The complexity of nZVl properties and reactions in subsurface environments necessitate a suite of complementary detection methods in order to fully characterize the fate and effects of nZVl during groundwater remediation. Caution must be taken when assessing the performance of nZVl injection based on only few measured parameters. Simple direct methods including total Fe, color and absorptivity, and some conventional indirect methods based on water chemistry and electrometric properties are most applicable in field applications. Some new methods, although still in laboratory experimental stages, such as chemical redox probes and geophysical methods, may provide new options for nZVl detection in the field in the near future.

Currently, most efforts in both laboratory and field studies emphasize the characterization of nZVl after field injection, which helps to assess the effectiveness of nZVl for the reduction of CoC in the field, the ultimate goal of nZVl injection. One aspect that often is overlooked, however, concerns the long-term effects of nZVl on subsurface biogeochemistry, and therefore on the sustainability of remediation efforts based on nZVl. For example, increased dissolved H2 and Fe2+ not only provide evidence for nZVl emplacement but may also create favorable environments for microbes to facilitate long-term bioremediation. The organic amendments added during nZVl synthesis or injection to facilitate transport may also provide carbon and energy for microbial metabolism, thereby enhancing microbial activities and thus degradation and sequestration of contaminants in groundwater. Further research to characterize these effects is needed to completely understand the effects of nZVl injection during groundwater remediation.

Acknowledgments

A portion of this work was funded by the Strategic Environmental Research and Development Program (SERDP) as part of ER-1485 (Fundamental Study of the Delivery of Nanoiron to DNAPL Source Zones in Naturally Heterogeneous Field Systems). This report has not been subject to review by SERDP and therefore does not necessarily reflect their views and no official endorsement should be inferred.

References

Adeleye, A.S., Keller, AA, Miller, R.J., Lenihan, H.S., 2013. Persistence of

commercial nanoscaled zero-valent iron (nZVl) and by-products.

J. Nanoparticle Res. 15(1), 1-18. Atekwana, E.A., Slater, L, 2009. Biogeophysics: a new frontier in earth science

research. Rev. Geophys. 47 (4), RG4004. Baer, D.R., Amonette, J.E., Engelhard, M.H., Gaspar, D.J., Karakoti, A.S.,

Kuchibhatla, S., Nachimuthu, P., Nurmi, J.T., Qiang, Y., Sarathy, V., Seal, S.,

Sharma, A., Tratnyek P.G., Wang, C.M., 2008. Characterization challenges for nanomaterials. Surf. lnterface Anal. 40 (3-4), 529-537.

Baer, D.R., Gaspar, D.J., Nachimuthu, P., Techane, S.D., Castner, D.G., 2010a. Application of surface chemical analysis tools for characterization of nanoparticles. Anal. Bioanal. Chem. 396 (3), 983-1002.

Baer, D.R., Amonette, J.E., Dohnalkova, A., Engelhard, M.H., Kuchibhatla, S., Liu, J., Nachimuthu, P., Nurmi, J.T., Tratnyek P.G., Wang, C.M., 2010b. The importance of complementary information provided by surface analysis, electron microscopy and in situ characterization of nanoparticles. Microsc. Microanal. 16 (Supplement S2), 408-409.

Baer, D.R., Tratnyek, P.G., Qiang, Y., Amonette, J.E., Linehan, J., Sarathy, V., Nurmi, J.T., Wang, C.M., Antony, J., 2012. Synthesis, characterization, and properties of zero-valent iron nanoparticles. Environmental Applications of Nanomaterials: Synthesis, Sorbents and Sensors 2nd edition. pp. 49-86.

Basnet, M., Ghoshal, S., Tufenkji, N., 2013. Rhamnolipid biosurfactant and soy protein act as effective stabilizers in the aggregation and transport of palladium-doped zerovalent iron nanoparticles in saturated porous media. Environ. Sci.Technol.47 (23), 13355-13364.

Benn, T.M., Westerhoff, P., 2008. Nanoparticle silver released into water from commercially available sock fabrics. Environ. Sci. Technol. 42 (11), 4133-4139.

Bennett, P., He, F., Zhao, D., Aiken, B., Feldman, L., 2010. ln situ testing of metallic iron nanoparticle mobility and reactivity in a shallow granular aquifer. J. Contam. Hydrol. 116(1-4), 35-46.

Blodau, C., 2011. Thermodynamic control on terminal electron transfer and methanogenesis. ln: Tratnyek, P.G., Grundl, T.J., Haderlein, S.B. (Eds.), Aquatic Redox Chemistry. ACS Symposium Series. American Chemical Society, Washington, DC, pp. 65-83.

Buchau, A., Rucker, W.M., de Boer, C.V., Klaas, N., 2010. lnductive detection and concentration measurement of nano sized zero valent iron in the subsurface. Sci. Meas. Technol. 4 (6), 289-297.

Busch, J., Meissner, T., Potthoff, A., Oswald, S.E., 2014a. lnvestigations on mobility of carbon colloid supported nanoscale zero-valent iron (nZVl) in a column experiment and a laboratory 2D-aquifer test system. Environ. Sci. Pollut Res. 21 (18), 10908-10916.

Busch, J., Meissner, T., Potthoff, A., Oswald, S.E., 2014b. Transport of carbon colloid supported nanoscale zero-valent iron in saturated porous media. J. Contam. Hydrol. 164,25-34.

Cole, K.S., Cole, R.H., 1941. Dispersion and absorption in dielectrics, vol. l: alternating current field. J. Chem. Phys. 9,341-351.

Eglal, M.M., Ramamurthy, A.S., 2014. Nanofer ZVl: morphology, particle characteristics, kinetics, and applications. J. Nanomater. 1-11.

Elliott, D.W., Zhang, W., 2001. Field assessment of nanoscale bimetallic particles for groundwater treatment. Environ. Sci. Technol. 35 (24), 4922-4926.

Esfahani, A.R., Firouzi, A.F., Sayyad, G., Kiasat, A.R., 2014. Transport and retention of polymer-stabilized zero-valent iron nanoparticles in saturated porous media: effects ofinitial particle concentration and ionic strength. J. lnd. Eng. Chem. 20 (5), 2671-2679.

Fan, D., Anitori, R.P., Tebo, B.M., Tratnyek, P.G., Lezama Pacheco, J.S., Kukkadapu, R.K., Engelhard, M.H., Bowden, M.E., Kovarik, L., Arey, B.W., 2013. Reductive sequestration of pertechnetate (99TcO—) by nano zerovalent lron (nZVl) transformed by abiotic sulfide. Environ. Sci. Technol. 47 (10), 5302-5310.

Fan, D., Anitori, R.P., Tebo, B.M., Tratnyek, P.G., Lezama Pacheco, J.S., Kukkadapu, R.K., Kovarik, L., Engelhard, M.H., Bowden, M.E., 2014. Oxidative remobi-lization of technetium sequestered by sulfide-transformed nano zerovalent lron. Environ. Sci. Technol. 48 (13), 7409-7417.

Filip, J., Karlicky, F., Marusak Z., Lazar, P., Cernlk, M., Otyepka, M., Zboril, R., 2014. Anaerobic reaction of nanoscale zerovalent iron with water: mechanism and kinetics. J. Phys. Chem. C118 (25), 13817-13825.

Glazier, R., Venkatakrishnan, R., Gheorghiu, F., Walata, L., Nash, R., Zhang, W., 2003. Nanotechnology takes root. Civ. Eng. 73 (5), 64-69.

He, F., Zhao, D.Y., 2005. Preparation and characterization of a new class of starch-stabilized bimetallic nanoparticles for degradation of chlorinated hydrocarbons in water. Environ. Sci. Technol. 39 (9), 3314-3320.

He, F., Zhao, D., 2007. Manipulating the size and dispersibility of zerovalent iron nanoparticles by use of carboxymethyl cellulose stabilizers. Environ. Sci. Technol. 41 (17), 6216-6221.

He, F., Zhao, D., Liu, J., Roberts, C.B., 2007. Stabilization of Fe-Pd nanoparticles with sodium carboxymethyl cellulose for enhanced transport and de-chlorination oftrichloroethylene in soil and groundwater. lnd. Eng. Chem. Res. 46 (1), 29-34.

He, F., Zhang, M., Qian, T., Zhao, D., 2009. Transport of carboxymethyl cellulose stabilized iron nanoparticles in porous media: column experiments and modeling. J. Colloid lnterface Sci. 334 (1), 96-102.

He, F., Zhao, D., Paul, C., 2010. Field assessment of carboxymethyl cellulose stabilized iron nanoparticles for in situ destruction of chlorinated solvents in source zones. Water Res. 44 (7), 2360-2370.

ARTICLE IN PRESS

18 Z. Shi et al / Journal of Contaminant Hydrology xxx (2015) xxx-xxx

Henn, K.W., Waddill, D.W., 2006. Utilization of nanoscale zero-valent iron for source remediation—a case study. Remediat. J. 16 (2), 57-77.

Jiemvarangkul, P., Zhang, W., Lien, H.L., 2011. Enhanced transport of poly-electrolyte stabilized nanoscale zero-valent iron (nZVI) in porous media. Chem. Eng. J. 170 (2-3), 482-491.

Johnson, R.L., Johnson, G.O., Nurmi, J.T., Tratnyek P.G., 2009. Natural organic matter enhanced mobility of nano zerovalent iron. Environ. Sci. Technol. 43 (14), 5455-5460.

Johnson, R.L., Nurmi, J.T., Johnson, G.S.O., Fan, D.M., Johnson, R.L.O., Shi, Z.Q., Salter-Blanc, A.J., Tratnyek, P.G., Lowry, G.V., 2013. Field-scale transport and transformation of carboxymethylcellulose-stabilized nano zero-valent iron. Environ. Sci. Technol. 47 (3), 1573-1580.

Jones, D.P., 2002. Investigation of Clay-Organic Reactions Using Complex Resistivity. Colorado School of Mines (403 pp.).

Jones, B.D., Ingle Jr., J.D., 2001. Evaluation of immobilized redox indicators as reversible, in situ redox sensors for determining Fe(III)-reducing conditions in environmental samples. Talanta 55 (4), 699-714.

Jones, B.D., Ingle Jr., J.D., 2005. Evaluation of redox indicators for determining sulfate-reducing and dechlorinating conditions. Water Res. 39 (18), 4343-4354.

Joyce, R.A., Glaser, D.R., Werkema, D.D., Atekwana, E.A., 2012. Spectral induced polarization response to nanoparticles in a saturated sand matrix. J. Appl. Geophys. 77,63-71.

Kadar, E., Tarran, G.A., Jha, A.N., Al-Subiai, S.N., 2011. Stabilization of engineered zero-valent nanoiron with Na-acrylic copolymer enhances spermiotoxicity. Environ. Sci. Technol. 45 (8), 3245-3251.

Kanel, S.R., Greneche, J.M., Choi, H., 2006. Arsenic(V) removal from groundwa-ter using nano scale zero-valent iron as a colloidal reactive barrier material. Environ. Sci. Technol. 40 (6), 2045-2050.

Kent, R.D., Oser, J.G., Vikesland, P.J., 2014. Controlled evaluation of silver nanoparticle sulfidation in a full-scale wastewater treatment plant. Environ. Sci. Technol. 48 (15), 8564-8572.

Klimkova, S., Cernik M., Lacinova, L., Filip, J., Jancik D., Zboril, R., 2011. Zero-valent iron nanoparticles in treatment of acid mine water from in situ uranium leaching. Chemosphere 82 (8), 1178-1184.

Kober, R., Hollert, H., Hornbruch, G., Jekel, M., Kamptner, A., Klaas, N., Maes, H., Mangold, K.M., Martac, E., Matheis, A., Paar, H., Schaffer, A., Schell, H., Schiwy, A., Schmidt, ICR, Strutz, T.J., Thummler, S., Tiehm, A., Braun, J., 2014. Nanoscale zero-valent iron flakes for groundwater treatment. Environ. Earth Sci. 72 (9), 3339-3352.

Kocur, C.M., O'Carroll, D.M., Sleep, B.E., 2013. Impact of nZVI stability on mobility in porous media. J. Contam. Hydrol. 145,17-25.

Kocur, C.M., Chowdhury, A.I., Sakulchaicharoen, N., Boparai, H.K., Weber, K.P., Sharma, P., Krol, M.M., Austrins, L., Peace, C., Sleep, B.E., O'Carroll, D.M., 2014. Characterization of nZVI mobility in a field scale test. Environ. Sci. Technol. 48 (5), 2862-2869.

Krol, M.M., Oleniuk, A.J., Kocur, C.M., Sleep, B.E., Bennett, P., Xiong, Z., O'Carroll, D.M., 2013. A field-validated model for in situ transport of polymer-stabilized nZVI and implications for subsurface injection. Environ. Sci. Technol. 47 (13), 7332-7340.

Laumann, S., Micic, V., Lowry, G.V., Hofmann, T., 2013. Carbonate minerals in porous media decrease mobility of polyacrylic acid modified zero-valent iron nanoparticles used for groundwater remediation. Environ. Pollut 179,53-60.

Laumann, S., Micic, V., Hofmann, T., 2014. Mobility enhancement of nanoscale zero-valent iron in carbonate porous media through co-injection of polyelectrolytes. Water Res. 50, 70-79.

Lee, S., Bi, X., Reed, R.B., Ranville, J.F., Herckes, P., Westerhoff, P., 2014. Nanoparticle size detection limits by dingle particle ICP-MS for 40 elements. Environ. Sci. Technol. 48 (17), 10291-10300.

Lerner, R.N., Lu, Q., Zeng, H., Liu, Y., 2012. The effects ofbiofilmon the transport of stabilized zerovalent iron nanoparticles in saturated porous media. Water Res. 46 (4), 975-985.

Lesmes, D.P., Frye, K.M., 2001. The influence of pore fluid chemistry on the complex conductivity and induced-polarization responses of Berea Sandstone. J. Geophys. Res. 106,4079-4090.

Li, X., Zhang, W., 2006. Iron nanoparticles: the core-shell structure and unique properties for Ni(II) sequestration. Langmuir 22 (10), 4638-4642.

Li, X., Zhang, W., 2007. Sequestration of metal cations with zerovalent iron nanoparticles — a study with high resolution X-ray photoelectron spectroscopy (HR-XPS). J. Phys. Chem. C 111 (19), 6939-6946.

Ling, L., Zhang, W., 2014a. Reactions of nanoscale zero-valent iron with Ni(II): three-dimensional tomography of the "hollow out" effect in a single nanoparticle. Environ. Sci. Technol. Lett. 1 (3), 209-213.

Ling, L., Zhang, W., 2014b. Sequestration of arsenate in zero-valent iron nanoparticles: visualization of intraparticle reactions at angstrom resolution. Environ. Sci. Technol. Lett. 1 (7), 305-309.

Liu, Y., Lowry, G.V., 2006. Effect of particle age (Fe0 content) and solution pH on NZVI reactivity: H2 evolution and TCE dechlorination. Environ. Sci. Technol. 40 (19), 6085-6090.

Liu, Y., Phenrat, T., Lowry, G.V., 2007. Effect of TCE concentration and dissolved groundwater solutes on nZVI-promoted TCE dechlorination and H2 evolution. Environ. Sci. Technol. 41 (22), 7881-7887.

Liu, A., Liu, J., Pan, B., Zhang, W., 2014. Formation of lepidocrocite ([gamma]-FeOOH) from oxidation of nanoscale zero-valent iron (nZVI) in oxygenated water. RSC Adv.4 (101), 57377-57382.

Mackenzie, K., Bleyl, S., Georgi, A., Kopinke, F.D., 2012. Carbo-iron — an Fe/AC composite — as alternative to nano-iron for groundwater treatment. Water Res. 46 (12), 3817-3826.

Martin, J.E., Herzing, A.A., Yan, W., Li, X.Q., Koel, B.E., Kiely, C.J., Zhang, W., 2008. Determination of the oxide layer thickness in core-shell zerovalent iron nanoparticles. Langmuir 24 (8), 4329-4334.

Matheson, L.J., Tratnyek P.G., 1994. Reductive dehalogenation of chlorinated methanes by iron metal. Environ. Sci. Technol. 28 (12), 2045-2053.

Nurmi, J.T., Tratnyek P.G., Sarathy, V., Baer, D.R., Amonette, J.E., Pecher, K., Wang, C.M., Linehan, J.C., Matson, D.W., Penn, R.L., Driessen, M.D., 2005. Characterization and properties of metallic iron nanoparticles: spectroscopy, electrochemistry, and kinetics. Environ. Sci.Technol. 39 (5), 1221-1230.

Nurmi, J.T., Sarathy, V., Tratnyek, P.G., Baer, D.R., Amonette, J.E., Karkamkar, A., 2011. Recovery of iron/iron oxide nanoparticles from solution: comparison of methods and their effects. J. Nanoparticle Res. 13 (5), 1937-1952.

O'Carroll, D., Sleep, B., Krol, M., Boparai, H., Kocur, C., 2013. Nanoscale zero valent iron and bimetallic particles for contaminated site remediation. Adv. Water Resour. 51,104-122.

Orsetti, S., Laskov, C., Haderlein, S.B., 2013. Electron transfer between iron minerals and quinones: estimating the reduction potential of the Fe(II)-goethite surface from AQDS speciation. Environ. Sci. Technol. 47 (24), 14161-14168.

Petala, E., Dimos, K., Douvalis, A., Bakas, T., Tucek J., Zboril, R., Karakassides, M.A., 2013. Nanoscale zero-valent iron supported on mesoporous silica: characterization and reactivity for Cr(VI) removal from aqueous solution. J. Hazard. Mater. 261,295-306.

Phenrat, T., Saleh, N., Sirk K., Tilton, R.D., Lowry, G.V., 2007. Aggregation and sedimentation of aqueous nanoscale zerovalent iron dispersions. Environ. Sci. Technol. 41 (1), 284-290.

Phenrat, T., Kim, H.J., Fagerlund, F., Illangasekare, T., Tilton, R.D., Lowry, G.V., 2009. Particle size distribution, concentration, and magnetic attraction affect transport of polymer-modified Fe0 nanoparticles in sand columns. Environ. Sci. Technol. 43 (13), 5079-5085.

Phenrat, T., Cihan, A., Kim, H.J., Mital, M., Illangasekare, T., Lowry, G.V., 2010. Transport and deposition of polymer-modified Fe0 nanoparticles in 2-D heterogeneous porous media: effects of particle concentration, Fe0 content, and coatings. Environ. Sci. Technol. 44 (23), 9086-9093.

Plathe, K.L., von der Kammer, F., Hassellov, M., Moore, J.N., Murayama, M., Hofmann, T., Hochella, M.F., 2013. The role of nanominerals and mineral nanoparticles in the transport of toxic trace metals: field-flow fractionation and analytical TEM analyses after nanoparticle isolation and density separation. Geochim. Cosmochim. Acta 102,213-225.

Poda, A.R., Bednar, A.J., Kennedy, A.J., Harmon, A., Hull, M., Mitrano, D.M., Ranville, J.F., Steevens, J., 2011. Characterization of silver nanoparticles using flow-field flow fractionation interfaced to inductively coupled plasma mass spectrometry. J. Chromatogr. A1218 (27), 4219-4225.

Quinn, J., Geiger, C., Clausen, C., Brooks, K., Coon, C., O'Hara, S., Krug, T., Major, D., Yoon, W.S., Gavaskar, A., Holdsworth, T., 2005. Field demonstration of DNAPL dehalogenation using emulsified zero-valent iron. Environ. Sci. Technol. 39 (5), 1309-1318.

Raychoudhury, T., Tufenkji, N., Ghoshal, S., 2014. Straining of polyelectrolyte-stabilized nanoscale zero valent iron particles during transport through granular porous media. Water Res. 50, 80-89.

Reinsch, B.C., Forsberg, B., Penn, R.L., Kim, C.S., Lowry, G.V., 2010. Chemical transformations during aging of zerovalent iron nanoparticles in the presence of common groundwater dissolved constituents. Environ. Sci. Technol. 44 (9), 3455-3461.

Rubin, Y., Hubbard, S.S., 2005. Hydrogeophysics. Springer.

Saleh, N., Kim, H.J., Phenrat, T., Matyjaszewski, K., Tilton, R.D., Lowry, G.V., 2008. Ionic strength and composition affect the mobility of surface-modified Fe0 nanoparticles in water-saturated sand columns. Environ. Sci. Technol. 42 (9), 3349-3355.

Sarathy, V., Tratnyek P.G., Nurmi, J.T., Baer, D.R., Amonette, J.E., Chun, C.L., Penn, R.L., Reardon, E.J., 2008. Aging of iron nanoparticles in aqueous solution: effects on structure and reactivity. J. Phys. Chem. C 112 (7), 2286-2293.

Shi, Z., Nurmi, J.T., Tratnyek, P.G., 2011. Effects of nano zero-valent iron on oxidation-reduction potential. Environ. Sci. Technol. 45 (4), 1586-1592.

Slater, L., Binley, A., 2003. Evaluation of permeable reactive barrier (PRB) integrity using electrical imaging methods. Geophysics 68 (3), 911-921.

Slater, L., Choi, J., Wu, Y., 2005. Electrical properties of iron-sand columns: implications for induced polarization investigation and performance monitoring of iron-wall barriers. Geophysics 70 (4), G87-G94.

ARTICLE IN PRESS

2. Shi et al. / Journal of Contaminant Hydrology xxx (2015) xxx-xxx 19

Su, Y., Adeleye, A.S., Zhou, X., Dai, C., Zhang, W., Keller, AA, Zhang, Y., 2014. Effects of nitrate on the treatment of lead contaminated groundwater by nanoscale zerovalent iron. J. Hazard. Mater. 280,504-513.

Tosco, T., Petrangeli Papini, M., Cruz Viggi, C., Sethi, R., 2014. Nanoscale zerovalent iron particles for groundwater remediation: a review. J. Clean. Prod. 77,10-21.

Tratnyek P.G., Johnson, R.L., 2006. Nanotechnologies for environmental cleanup. Nano Today 1 (2), 44-48.

Tratnyek, P.G., Reilkoff, T.E., Lemon, A.W., Scherer, M.M., Balko, B.A., Feik L.M., Henegar, B.D., 2001. Visualizing redox chemistry: probing environmental oxidation-reduction reactions with indicator dyes. Chem. Educ. 6 (3), 172-179.

Tratnyek P.G., Salter-Blanc, A.J., Nurmi, J.T., Amonette, J.E., Liu, J., Wang, C., Dohnalkova, A., Baer, D.R., 2011. Reactivity of zerovalent metals in aquatic media: effects of organic surface coatings. ln: Tratnyek P.G., Grundl, T.J., Haderlein, S.B. (Eds.), Aquatic Redox Chemistry. ACS Symposium Series. American Chemical Society, Washington, DC, pp. 381-406.

Vecchia, E.D., Luna, M., Sethi, R., 2009. Transport in porous media of highly concentrated iron micro- and nanoparticles in the presence of xanthan gum. Environ. Sci.Technol. 43 (23), 8942-8947.

Vereecken, H., Binley, A., Revil, A., Titov, K., 2004. Applied Hydrogeophysics 71. Springer.

Wang, C., Zhang, W., 1997. Synthesizing nanoscale iron particles for rapid and complete dechlorination of TCE and PCBs. Environ. Sci. Technol. 31 (7), 2154-2156.

Wei, C.J., Li, X.Y., 2013. Surface coating with Ca(OH)2 for improvement of the transport of nanoscale zero-valent iron (nZVl) in porous media. Water Sci. Technol. 68 (10), 2287-2293.

Wei, Y.T., Wu, S.C., Chou, C.M., Che, C.H., Tsai, S.M., Lien, H.L., 2010. lnfluence of nanoscale zero-valent iron on geochemical properties of groundwater and vinyl chloride degradation: a field case study. Water Res. 44 (1), 131-140.

Wong, J., 1979. An electrochemical model of the induced-polarization phenomenon in disseminated sulfide ores. Geophysics 44,1245-1265.

Wu, Y., Slater, L., Korte, N., 2005. Effect of precipitation on low frequency electrical properties of zerovalent iron columns. Environ. Sci. Technol. 39 (23), 9197-9204.

Wu, Y., Slater, L., Korte, N., 2006. Low frequency electrical properties of corroded iron barrier cores. Environ. Sci. Technol. 40 (7), 2254-2261.

Wu, Y., Slater, L., Versteeg, R., LaBrecque, D., 2008. A comparison of the low frequency electrical signatures of iron oxide versus calcite precipitation in granular zero valent iron columns. J. Contam. Hydrol. 95 (3-4), 154-167.

Wu, Y., Versteeg, R., Slater, L., LaBrecque, D., 2009. Calcite precipitation dominates the electrical signatures of zero valent iron columns under simulated field conditions. J. Contam. Hydrol. 106,131-143.

Yan, W., Herzing, AA, Kiely, C.J., Zhang, W., 2010a. Nanoscale zero-valent iron (nZVl): aspects of the core-shell structure and reactions with inorganic species in water. J. Contam. Hydrol. 118 (3-4), 96-104.

Yan, W., Herzing, AA., Li, X., Kiely, C.J., Zhang, W., 2010b. Structural evolution of Pd-doped nanoscale zero-valent iron (nZVl) in aqueous media and implications for particle aging and reactivity. Environ. Sci. Technol. 44 (11), 4288-4294.

Yan, W., Ramos, M.A.V., Koel, B.E., Zhang, W., 2012. As(lll) sequestration by iron nanoparticles: study of solid-phase redox transformations with X-ray photoelectron spectroscopy. J. Phys. Chem. C 116 (9), 5303-5311.

Yan, W., Lien, H.L., Koel, B.E., Zhang, W., 2013. lron nanoparticles for environmental clean-up: recent developments and future outlook Environ. Sci. Processes lmpacts 15 (1), 63-77.

Zhang, W., 2003. Nanoscale iron particles for environmental remediation: an overview. J. Nanoparticle Res. 5 (3-4), 323-332.

Zhuang, Y., Jin, L., Luthy, R.G., 2012. Kinetics and pathways for the denomination of polybrominated diphenyl ethers by bimetallic and nanoscale zerovalent iron: effects of particle properties and catalyst. Chemosphere 89 (4), 426-432.