Even today the remediation of organic contaminant source zones poses significant technical and economic challenges. Nanoscale zero-valent iron (NZVI) injections have proved to be a promising approach especially for source zone treatment. We present the development and the characterization of a new kind of NZVI with several advantages on the basis of laboratory experiments, model simulations and a field test. The developed NZVI particles are manufactured by milling, consist of 85 % Fe(0) and exhibit a flake-like shape with a thickness of <100 nm. The mass normalized perchloroethylene (PCE) dechlorination rate constant was 4.1 × 10−3 L/g h compared to 4.0 × 10−4 L/g h for a commercially available reference product. A transport distance of at least 190 cm in quartz sand with a grain size of 0.2–0.8 mm and Fe(0) concentrations between 6 and 160 g/kg (sand) were achieved without significant indications of clogging. The particles showed only a low acute toxicity and had no longterm inhibitory effects on dechlorinating microorganisms. During a field test 280 kg of the iron flakes was injected to a depth of 10–12 m into quaternary sand layers with hydraulic conductivities ranging between 10−4 and 10−5 m/s. Fe(0) concentrations of 1 g/kg (sand) or more [up to 100 g/kg (sand)] were achieved in 80 % of the targeted area. The iron flakes have so far remained reactive for more than 1 year and caused a PCE concentration decrease from 20.000–30.000 to 100–200 µg/L. Integration of particle transport processes into the OpenGeoSys model code proved suitable for site-specific 3D prediction and optimization of iron flake injections.
During the last 10–15 years there has been an increasing number of scientific publications investigating different characteristics and capabilities of nanoscale zero-valent iron (NZVI) particles with most of them showing NZVI to be a promising substance for groundwater and in particular source remediation (e.g. Henn and Waddill 2006; Phenrat et al. 2011; Kuiken 2010). Major advantages in this technology are the injectability of the particles in immediate vicinity of the contaminant source, high reaction rates due to the large surface area and a large number of relevant contaminants which are treatable, especially chlorinated hydrocarbons (CHCs) as, e.g. per- or trichloroethylene. Despite these advantages and promising results of laboratory and field studies, NZVI still do not represent standard technology for source treatment of CHCs or other contaminants. This is mainly due to the low availability in Europe and to the high costs. Today, NZVI is typically produced by reductive precipitation of Fe(0) from solutions containing Fe2+ using strong reductants like borohydride or hydrogen. Further developmental opportunities of this production technology are so far limited due to the relatively high costs of the required chemicals.
With this contribution, we want to introduce a new kind of NZVI particles produced for source remediation by grinding of macroscopic raw materials of elementary iron to the micrometer and nanometer scale without the need of expensive chemicals. Besides particle production, the properties and functionality of the new particles as well as their field applicability are described. The development was an iterative process between particle production on the one hand and verification of reactivity, mobility and ecological sustainability on the other. In addition to the testing of different raw materials, grinding additives and grinding procedures, the development of coatings to prevent early sedimentation caused by aggregation (Phenrat et al. 2007, 2008) was investigated and the size, shape, surface area, structure and Fe(0) content of the particles were characterized.
Reactivity characteristics of NZVI must fulfill two requirements. First, immediate reactivity has to be high to achieve short treatment times and second, longterm reactivity has likewise to be sufficiently high to provide reactivity over several weeks or even months. Such periods are essential since not all of the contaminant mass will be affected by NZVI particles directly after injection, due to the heterogeneous distribution of aquifer permeability, contaminant mass and NZVI transport.
For a successful source treatment with NZVI, which means removal of the bulk contaminant mass, a sufficiently high Fe(0) concentration and a particle distribution as homogeneous as possible are important. The Fe(0) demand would be 28 g/kg (sand) for a PCE contamination with a residual saturation of 5 %. However, so far lab or field experiments including a quantification of Fe(0) mass could verify appropriate Fe(0) concentrations only up to a distance of a few decimeters from the injection point [e.g. Phenrat et al. (2010), Taghavy et al. (2010)]. Even if relatively short distances of, for example, 2 m between injection points can be covered using advanced injection technology like direct-push, it has to be ensured that the injected NZVI can be transported such distances at sufficient concentrations.
Besides the improvement of application-oriented aspects, a further goal was to promote confidence in the transferability from lab to field scale and in the ecotoxicological and microbial innocuousness of the new particles. A systematic and quantitative evaluation of NZVI applicability as a remedial approach for CHC contaminants in soil and groundwater under field-relevant conditions has been reported only recently (Phenrat et al. 2011). Henn and Waddill (2006) supplied evidence that NZVI effectively degraded contamination and reduced the mass flux from the source as a critical metric identified for source treatment. Many early studies were reported as successes with limited discussion addressing challenges associated with NZVI application at a field scale (Elliott and Zhang 2001; Zhang 2003). Zhao (2013) provides a good review of NZVI application for CHC remediation, underlining the lack of systematically built up experience under field conditions. O’Carroll et al. (2012) point out a significant need for field studies that demonstrate rigorous site characterization, optimization of on-site injection infrastructure, design of injection fluid properties and reporting methods to reduce the unreliability associated with NZVI delivery. Our field-scale pilot study aimed to provide answers to these questions.
To optimize NZVI injection and distribution for field applications, the particle development was accompanied by the adaption of a numerical model. While the classical filtration theory (Yao et al. 1971) was shown to be inappropriate to explain NZVI transport in porous media in many cases, studies considering more complex approaches like blocking, ripening and straining processes have proven to be able to describe transport behavior under different hydraulic and geochemical boundary conditions (Bradford et al. 2002; 2003, 2009; 2011; Johnson and Elimelech 1995; Johnson et al. 1996; Ko and Elimelech 2000; Loveland et al. 2003; McDowell-Boyer et al. 1986; Sun et al. 2001b; Tiraferri et al. 2011; Tosco and Sethi 2009, 2010). However, more-dimensional model applications considering transport behavior of NZVIs for field conditions are quite rare (Cullen et al. 2010; Kanel et al. 2008; Sun et al. 2001a). Therefore, the objectives of the reactive transport modeling part were (1) to identify and parameterize governing transport processes under different injection conditions and (2) to transfer and apply these findings to the field scale using more-dimensional simulations including model verification by field investigations.
Although there are already various studies showing only low acute toxicity for NZVI materials (Marsalek et al. 2012; Phenrat et al. 2009; El-Temsah and Joner 2012; Wang et al. 2012), these findings cannot directly be transferred to newly developed particles, since different chemical composition, coatings or shape of NZVI can have an influence on the toxicity. Since microbial degradation can additionally contribute to the removal of main target contaminants like chlorinated ethylenes (Tiehm and Schmidt 2011; Aktas et al. 2012), inhibition of the microbial degradation by NZVI should be excluded. Moreover, reductively dechlorinating and also other hydrogenotrophic bacteria can act as a sink for the hydrogen produced during anaerobic NZVI corrosion and thus mitigate the potential for clogging. The suspending agents and stabilization materials of NZVI formulations can function as an additional source of electron donors for biological dechlorination. Thus, the combination of abiotic dechlorination with NZVI and biological dechlorination can have synergetic effects.
All these topics are needed for the throughout evaluation of the characteristics and capabilities of this new and promising type of NZVI. This paper will therefore summarize all of these aspects for the developed particles to give a brief overall view. More detailed and extended presentations of the individual topics will be part of subsequent publications.
To develop a method to produce nanoparticles of elementary iron in a two-stage top-down process, various commercial iron powders (carbonyl iron, sponge iron, cast iron and iron powder) were tested. Within the project various grinding equipments, grinding additives and milling times were investigated (Table 1). The focus was set on grinding tests with different mills, such as vibrating mills, ball mills and stirred ball mills in a laboratory as well as on a semi-industrial scale. The selection of the grinding media was a challenge because of the formation of hydrogen in an aqueous agent (tap water, deionized water; pH value controlled). Different organic media were tested; in the end mono ethylene glycol (MEG) was selected. Consequently, the iron particles were produced in a two-stage procedure. The first step involved dry grinding with inhibitors as corrosion protection up to a particle size <40 µm. Wet fine grinding, the second process step, was realized with MEG as the grinding liquid and addition of a surfactant. Sponge iron (Höganäs, NC100.24) was used as the raw material for production batches A1–A4 and Atomet 57 or 58 (Rio Tinto, Quebec Metal Powders Ltd) for batches B1–B5 which also included the particles for field applications selected by reactivity and mobility investigations.
To study the effect of coatings on the size and agglomeration of iron particles at first, a series of experiments were performed using freshly prepared nanoparticles, as long as no grinded particles were available. Afterwards the results were transferred to grinded particles. According to the procedure described by Bonder et al. (2007) iron nanoparticles are formed by chemical reduction of iron(II) salts using boron hydrate as a reducing agent. Iron(II) sulfate FeCl2 × 4H2O (Fluka), sodium borohydride NaBH4 (Sigma Aldrich), polyethylene glycol PEG 400 (Fluka, linear molecule; molecular weight 380–420 g/mol) and sodium dodecyl sulfate SDS (Merck) used for the preparation of preliminary test particles were of analytical grade and were used as received. Iron nanoparticles were synthesized by reduction of FeCl2 in ethanol (96 %). 25 ml of 6 mmol NaBH4 was drop-by-drop added to a stirred solution of 1.8 mmol FeCl2 and a PEG concentration between 0.5 and 2 mmol in 250 mL cooled (0 °C) ethanol. Because of the air-sensitivity of the iron particles, the solution was kept in an inert gas atmosphere (N2). After 2 h stirring the particles precipitated. The supernatant solvent was poured off and the particles suspended in de-ionised water (18 MΩcm). The reduction was accomplished in the presence of different polyethylene glycols (PEGs), which enclose the freshly formed iron particles. This procedure leads to small particles with narrow size distribution. Particle size distribution used for the coating development was determined with a Zetasizer Nano ZS (Malvern).
The Fe(0) content of the milled particles was determined by measuring the hydrogen production in relation to the total amount of dissolved iron in acidic conditions (sulfuric acid). Dissolved iron concentrations were determined by Atomic Absorption Spectroscopy (AAS). Depending on the iron concentration of the sample either graphite furnace—AAS (GF-AAS, SpectrAA-400 Varian) or flame—AAS (F-AAS, GBC 906AA) was used. Furthermore, the concentration of dissolved iron was determined by the photometric method according to DIN 38401-E1.
To analyze the size distribution of particle batches a Mastersizer 2000 system (Malvern) was used. This technique is based on laser diffraction and measures the intensity of light scattered as a laser beam passes through a dispersed particulate sample. Calculations of the size distribution use the created scattering pattern and assume a spherical shape of the particles.
The specific surface area (BET) is an important parameter for the characterization of NZVI particles. The determination is based on the measurement of the adsorption of nitrogen on the surface according to DIN EN ISO 9277: 1995. The measurement was performed using the areameter Area-Max 1 by CIS Seifert. Since the particles must be in a dry state, the suspension was first removed and the particles were coated with methanol to prevent oxidation. Before each BET measurement, the surface of the samples had to be de-gased to avoid contamination with oxygen. The degassing took place in a separate thermostat-controlled heating device at a temperature of 105 °C until methanol evaporated. For the analysis of the particle size, shape, surface and structure, a Phenom electron microscope was used.
Reactivity comparison of different particle batches
To compare the particles produced under different milling conditions, reactivity tests were conducted in degassed and demineralized water in a 0.5-L gas-proof reactor. The solution was continuously stripped with nitrogen to achieve anaerobic conditions. The aquatic NZVI suspension was stirred at 200 rpm to minimize mass transport effects of contaminants on the reactivity. During the reaction the pH was controlled by titro-processors which kept the pH constant at 7 by adding hydrochloric acid. Dehalogenation rates were estimated by measuring the contaminant concentration and the dissolved iodide as a function of experimental time during the deiodination of iopromide (iodinated X-ray contrast media). Iopromide was quantified by LC–ESI-MS/MS (HP 1100, Agilent, Waldbronn, Germany; Quattro-LC, Micromass, Manchester, UK)
NZVI from NanoIron s.r.o. (N25) was used as a reference material. N25 particles have an average BET surface area of 20–25 m2/g and an average particle size of 50 nm (Zhuang et al. 2012). However, in aqueous solution the particles aggregate to clusters of >1 µm (determined by dynamic light scattering, DLS). The particle shells consist of a thin iron oxide layer (Fe3O4, FeO) which encloses the Fe(0) core (core–shell-structure). Before each batch test the Fe(0) content of the nanoparticles was determined by measuring the hydrogen production under acidic conditions.
The central part of the stand-alone column setup system was a syringe pump which is controlled by a Siemens Logo, controlling the switches of the solenoid valves to change between refilling and pump action of the syringes. Moreover it operated a membrane pump which supplies degassed water to a storage container. The water quality was monitored while the degassed water was pumped into a PCE mixing container to supply the syringe pump with a PCE solution with concentrations between 80 and 120 mg/L, comparable to the concentrations in a source zone.
The syringe pump assured a continuous and almost pulse-free discharge of the PCE solution through the columns. It was possible to take water samples from sampling ports before and after each column to measure PCE concentration, the degradation products chloride and metabolites, and pH. Outflow boundaries were controlled via a constant-head tank. The overflow rate of the constant-head tank was measured to confirm the flow rates from the syringes. The columns were made of glass, had a length of 200 cm and an inner diameter of 3.6 cm. Quartz sand was used as a porous medium (grain size 0.3–0.8 mm) with a bulk density of 1.67 g/cm3, a corresponding porosity of approx. 0.36 and a pore volume (PV) of approx. 700 cm3. Each column was controlled by an allocated syringe pump with flow rates about 175 cm3/days resulting in seepage velocities of 0.5 m/days. To insert the NZVI particles (B2) in the column, 700 mL of a suspension with an Fe(0) concentration of 10.4 g/L (1 PV) was injected from the top of the vertically positioned column, which resulted in a total ZVI mass of 7.3 g within the first 0.5 m of the column and a maximum Fe(0)-concentration of 16 g/kg sand. To compare the newly developed with other NZVI particles, N25S was used in a separate column as a reference material. The Fe(0) concentration of this suspension during the injection was 20 g/L. The total mass of ZVI inside the column was 14 g (within 700 mL suspension), distributed along the 2 m column. The maximum ZVI concentration reached was 17 g/kg. First-order reaction rate constants, calculated from the PCE inflow and outflow concentrations of each column, were normalized to the respectively injected ZVI mass. The normalized rate by the mass of ZVI per unit aqueous phase volume (k m) was calculated according to Taghavy et al. (2010).
Transport investigations were performed in horizontal column experiments (L 2 m, ID 8 cm, quartz sand 0.2–0.8 mm, n approx. 0.35). To determine the best particle production batch, the different particle batches were injected at similar injection conditions with Fe(0) concentrations of about 10 g/L and filter velocities (v f) of 0.5 m/h. In this study, only the results of the two best batches with the best transport characteristics (B2 and B4) were represented. To compare the developed particles with commercial particles, an experiment with same injection conditions and N25S particles was performed additionally.
To investigate if a particle distribution range of approximately 2 m or more and a sufficient Fe(0) concentration in the sand can be achieved, an experiment with the batch A4 was performed at similar injection conditions (filter velocity: 0.5 m/h, Fe(0): 9 g/L, quartz sand 0.25–1.0 mm, porosity: 0.36) but with a considerably greater injected Fe(0) mass.
During injection the magnetic susceptibility was measured using a core sensor (Co. Bartington, Core Scanning Sensor Type MS2C) to quantify the concentration of total Fe(0) (Fe 0tot ) which indicates the sum of mobile and immobile Fe(0). The mass of retained Fe(0) in the sand was the best variable for comparing the experiments because it could be determined exactly from Fe 0tot .
Particle transport modeling
As an extensive basis for the present study, column tests conducted with commercial NZVI particles N25S under variable injection velocities (0.5–4.5 m/h), particle concentrations (5–17 g/L), and initial permeabilities (K = 6E−09–5E−11 m2) were simulated using the matlab-based code E-MNM1D (Tosco and Sethi 2010), considering blocking and straining as governing deposition processes. Because of a similar transport behavior of N25S and the developed particles (B4), both deposition processes were also considered and parameterized by simulations of a column test with the developed particles in preparation for the preliminary injection test at the field site and for simulations at field scale. For these field-scale simulations, the multi-component, multi-dimensional flow and transport code OpenGeoSys (OGS) (Kolditz et al. 2012) was used and modified (Hornbruch et al. in prep.). The mass transport is described in the following Eq. 1, where C i and S i are the concentration of dissolved and attached species i, respectively, t is the time, v f the Darcy velocity, n the water filled porosity, and D the diffusion–dispersion tensor.
where S is the attached concentration in the solid phase for each interaction site i, ρ b the bulk density, k a,i and k d,i the attachment and detachment rate constants, respectively, and A the excluded area parameter according to Johnson et al. (1996) as the negative inverse value of a maximum sediment retention capacity S max, x the travel distance, d 50 the mean soil diameter and β i an exponent controlling attachment dynamics.
For field simulations, a homogeneous radial model was used considering a decreasing flow velocity with increasing distance from injection. The permeability was chosen to be 1E−12 m2 based on direct-push investigations, while the effective porosity was assumed to be 20 %. The smallest grid spacing at the injection point was 0.10 m due to high flow velocities in the vicinity of the injection well using an injection rate of 13 L/min and an injection concentration of 2.5 g/L.
A field-scale pilot test was vital for assessing and validating the applicability of the newly developed NZVI particles, concepts and methodologies. NZVI field application was implemented at the demonstration site Breite St. in Braunschweig, Germany. The site was formerly used as a dry cleaning facility and is highly contaminated with PCE between 10 and 14 m below ground. The rather moderate hydraulic conductivity fine sands (1–5 × 10−5 m/s) exhibit PCE aqueous concentrations between 20 and 50 mg/L.
A proper dimensioning of the field-scale pilot-test in terms of spatial configuration, injection strategy and setup of the monitoring system cannot be achieved without a preliminary field-scale injection experiment. Injection relevant parameters were investigated during direct-push (DP) injection of 2 m3 NZVI slurry (2.5 g/L) through pressure-activated injection probes at about 13 L/min under a pressure of 5–6 bar. These preliminary test injections were performed at two different injection points depth oriented into horizons with similar characteristics to the pilot-test target area. Monitoring of the spatial NZVI distribution around injection points was carried out through DP-Liner sampling at the same depths followed by lab analytics through calibrated magnetic susceptibility measurements on the liner-based soil samples.
Based on these findings, a full field-scale pilot application was designed (spacing of injection wells, injection rates, injection pressure, depth and type of injection, NZVI concentration) and implemented in summer 2012. Five newly developed NZVI in situ monitoring systems were installed in July 2012 prior to the injection. The multilevel ports for monitoring the NZVI spatial distribution through online metering of the magnetic susceptibility were equipped with additional multilevel groundwater sampling devices. To monitor the reaction zone in terms of upgradient background, respectively downgradient effects, multilevel monitoring well groups (1″ inner diameter monitoring wells installed in a nested setup with depth-oriented screened intervals) were installed prior to the NZVI injection. 280 kg of iron was injected as slurry (10 g/L) at 14 injection points during the first 2 weeks of August 2012 at 4 depths between 10 and 12 m below ground. For each injection target horizon, 500 L slurry was injected in an omni-directional setup. Following the injection, the NZVI radius of influence was investigated by liner-based soil sampling. Spatial distributions of contaminants of concern and milieu parameters within the reaction zone and neighboring up- and downgradient areas were monitored over 1 year through a series of groundwater sampling events and analysis.
Investigation of the biological activity during field application
For the investigation of biological activities during the field application of NZVI, dechlorinating organisms were assessed by nested PCR. Investigated organisms were Dehalococcoides sp. (Smits et al. 2004), Desulfitobacterium sp. (Smits et al. 2004), Desulfomonile tiedjei (El Fantroussi et al. 1997a, b), Dehalobacter sp. (Smits et al. 2004) and Desulfuromonas sp. (Löffler et al. 2000). The occurrence of these halorespiring microorganisms at a given site can be taken as an indicator for the degree of dechlorination occurring at this site (Schmidt et al. 2006). For the assessment of acute toxicity, the inhibition of luminescent bacteria was measured according to DIN EN ISO 11348-1.
Aged NZVI were evaluated with standardized acute aquatic and mechanism-specific ecotoxicological tests. The nanomaterials were investigated in a worst-case scenario as ultrasonicated suspensions. The particle suspensions were dried in an oven over night at 80 °C under air circulation. The powder weighed in with a maximal variance of 1 % and introduced into the corresponding aquatic media. The concentration range for of the aquatic test was up to 1,000 mg/L.
As acute toxicological tests, the Daphnia magna acute immobilization test according to DIN EN ISO 6341:2010, the prolonged fish embryo toxicity test (96 h) according to DIN EN ISO15088:2009 with Danio rerio (Braunbeck et al. 2005), the algae growth inhibition test DIN 8692:2010 with Desmodesmus subspicatus and as a mechanism-specific test the Ames fluctuation test according to ISO 11350:2011 with Salmonella typhimurium (Reifferscheid et al. 2012) were applied.
Results and discussion
After milling, the particles were stored in ethylene glycol to avoid oxidation. On average fresh particles in suspensions consisted of 85 % Fe(0) (±2; n = 3). Investigations of the Fe(0) stability showed a reduction of about 10 % during 4-month storage. Figure 1 displays the measured particles size distribution of the final particle suspension after the two-step milling process [particle size: d 10/d 50/d 90 = 1,63/4,53/16,21 µm]. The BET analysis of the particles suspension shows a specific surface area of 18 m2/g (raw material: 0.1 m2/g; intermediate product: 0.7 m2/g).
Due to the milling process the particle suspension contained different agglomerates, probably causing a wide particle distribution. The largest measured agglomerates were 45 μm in size, the smallest 200 nm.
This was confirmed by a scanning electron microscopy (SEM) image of dried particles (Fig. 2). The SEM image shows that larger particles can consist of several layers of small particles.
Furthermore, the final iron particles can be described as a flake shape resulting from the milling process. While the lateral size of the flake is approximately several micrometers, the particle thickness is less than 100 nm. This stands in contrast to the mathematical model of the mastersizer 2000 system and has to be considered. The mathematical model of the mastersizer assumes a spherical shape of a particle for its calculation. Furthermore, it predicts a larger particle size when the laser beam hits a particle in front instead of side.
Utilizing PEG 400, a size distribution between 24 and 164 nm was found 6 h after synthesis. 18 h later the size distribution had increased (33–255 nm). Particles stored in nitrogen atmosphere for many days exhibit no significant precipitation. Synthesis of NZVI also was made in aqueous solution (pH between 9 and 10) without addition of PEG. In this case, iron particles precipitated within a few hours. Hence, PEG is able to diminish agglomeration if it is added during synthesis of nanoparticles.
Similar practice is advantageous if iron particles are made by milling. Ethylene glycol is an alternative to PEG 400. In the presence of ethylene glycol, suspensions of milled iron are stable for many days even if diluted with deionized water (1:100). However, addition of oxygen results in agglomeration and complete precipitation within 18 h. Ethylene glycol coatings are hydrophilic and permeable for oxygen (Bonder et al. 2007). Fe(II) ions are formed during the oxygen-induced corrosion process of iron particles. We assume, that Fe(II) ions play a vital role in agglomeration. The addition of small amounts of sodium dodecyl sulfate (SDS, 10 mM) inhibits precipitation in the presence of oxygen appreciably. A possible explanation is the formation of Fe(II)–SDS complexes, which form a protective layer on the iron particles inhibiting the corrosion process (Rajendra et al. 2002). Agglomeration induced by ions also is observable while adding drinking water. Two samples of ethylene glycol-coated milled iron particles were diluted with potable water under inert gas atmosphere in a glove box. Without SDS a spontaneous precipitation occurred. The sample containing 10 mM SDS also showed initiation of precipitation. However, after 4 h the supernatant solution was still a dark suspension. In this experiment, calcium and magnesium ions in potable water might be responsible for agglomeration.
Reactivity comparison of different particle batches
The present study investigates the reactivity of NZVI in batch experiments depending on varying parameters of the milling process. The production batches of NZVI differ with regard to basic material, milling duration, and type of dispersant chemicals used: in all degradation experiments, iopromide is subject to a pseudo-first-order rate kinetics. The concentration of iopromide decreases exponentially over time with a very good correlation (Fig. 3; Table 2). Degradation of iopromide is verified by an increase of the iodide concentration. Furthermore, a constant DOC concentration throughout experimental time excludes a decrease in iopromide concentration caused by adsorption onto the reactor surface in addition to dehalogenation. This is in accordance with results of Stieber et al. (2011). Apparent dehalogenation rates during experiments with tested milled particles are comparable (except A1) to the rate which is observed during experiments with the reference nanoparticles N25.
The particle batch A1 shows a significantly lower degradation rate of iopromide because of a lower surface area available for the reaction. The shorter milling duration of A1 (only one milling step) probably results in larger particles and a lower surface area of the particle suspension. In experiments with fine-milled Fe(0), Matheson and Tratnyek (1994) report that the dehalogenation performance is a linear function of the iron surface area concentration. In the present study, higher Fe(0) amounts correspond to higher iron surfaces available in the batch reactor which might be the reason for the slightly higher degradation rate of the particle batch A2.
Within the present batch system, the organic contaminant (iopromide) is effectively de-halogenated by the milled iron particles. Further experiments will focus on the reactivity of NZVI particles in conditions comparable to the aquifer (sand columns).
Column experiments have been performed to simulate remediation of PCE solution in a source zone with NZVI particles. Within the first 12 days (3 pore volumes), the particles showed the highest reactivity resulting in a decrease of the dissolved PCE concentrations of up to 90 % (95–9 mg/L). Thereafter, the reaction decelerated and leveled off at a constant PCE reduction of approx. 50 % (Fig. 4). This change in reactivity is probably the result of mineralogical transformations on particle surfaces. Stoichiometric chloride formation showed an almost complete degradation of the reduced PCE mass. Furthermore, throughout the experiments, only a small amount of 2.8 mg TCE total mass and no other metabolites were detected. The particles showed a persistent and stable reaction for more than 40 days (10 pore volumes) without any significant changes of pH due to dechlorination or anaerobic corrosion.
Within the first PV, a PCE reduction rate constant (k obs) of 1.1 × 10−1 L/h was observed, after 10 PV the k obs had an average of 4.3 × 10−2 L/h. Under the same experimental conditions, N25S and B2 particles were directly compared (Fig. 5). When normalized by the mass of ZVI per unit aqueous phase volume (k m), the B2 particles showed approximately one magnitude larger rate constants than the reference particles N25S (Fig. 5). The average value of this rate constant was 4.1 × 10−3 L/g h for B2 particles and 4.0 × 10−4 L/g h for N25S. The better performance of the newly developed particles is probably caused by the mechanical activation of the surface during the milling process resulting in a flake shape with a highly reactive non-stabilized surface. N25S particles are deactivated by part-oxidation of the surface to enhance their stability against reactions like anaerobic corrosion.
The ZVI mass normalized rate constant of the B2 particles is consistent with the magnitude of the normalized rate constant for RNIP particles (NZVI by TODA) as reported by Liu et al. (2005). In comparison to the mass normalized rate constant of micron-sized iron powder of 4.8 × 10−4 L/g h measured by Doong and Lai (2006), B2 shows a one-order magnitude larger rate constant (average 4.1 × 10−3 L/g h). For field application, a longer contact time between the reactive surface and the dissolved contaminants is needed to improve the efficiency of the degradation.
At comparable retained Fe(0) masses of 33–40 g, B3 particles had the greatest distribution range of 90 cm if Fe(0)tot >1 g/kg (sand) is taken as a reference compared to 35 cm for N25S and 24 cm for B1 (Fig. 6). Besides the longer transportation length of batch B3, these particles additionally showed the lowest Fe(0) concentration gradients and the most homogenous distribution. The lack of additives during the pre-milling process for batch B1 is assumed to be responsible for its relatively low transportability. The used grinding additives as well as the Fe(0) content and iron impurities seemed to be the controlling factors for the distribution range. The flake-like shape resulting from the grinding process is assumed to be the reason for slower sedimentation and better transport characteristics compared to the chemically precipitated N25S particles with a more spherical shape (Komar and Reimers 1978).
NZVI particles in the experiment with the higher injected volume of A4 suspension could be transported throughout the whole column producing Fe(0)tot concentration of at least 27.5 g/kg (sand) up to 1.7 m flow path (Fig. 6). Fe(0)tot concentrations near the injection point accounted for 100 g/kg (sand) and more. To our knowledge, the results presented here are the first proof that NZVI particles can be spread over 2 m with resulting Fe(0) concentrations sufficiently high for source treatment.
During the preliminary field injection tests, the proof of NZVI presence and persistence achieved through liner-soil sampling indicated a radius of influence (ROI defined by the 0.5 g/L NZVI iso-concentration) of about 1 m. Following the field-scale pilot-test injection, liner-based soil sampling validated the previous findings in terms of NZVI ROI and confirmed DP direct injection as a NZVI subsurface delivery method being able to achieve a quasi-homogeneous coverage of the reaction zone (Fig. 7).
Groundwater monitoring for 1 year following the field-scale pilot-test injection in August 2012 revealed a drastic decrease of the PCE concentrations from 20–30 mg/L down to 100–200 µg/L both in the source area and downgradient plume (Fig. 8). The increase of the PCE concentrations immediately after the NZVI injection is probably due to mechanical mobilization from neighboring NAPL bodies. A typical initial increase of degradation products (TCE, cis-DCE) followed by a decrease was observed. VC was detected only in concentrations below 4 µg/L. Ethene was produced in high concentrations both in the source area (up to 21 mg/L) and in the downgradient plume (up to 3.5 mg/L). Such ethene concentrations suggest initial PCE concentrations of about 120 mg/L. The increase of chloride from 60 to 200 mg/L further supports the conclusions about high initial PCE concentrations indicating an additional input from the PCE-NAPL. Despite the temporary accumulation of cis-DCE pointing to additionally occurring biological degradation processes, the fast production of high amounts of ethene indicates iron-mediated abiotic reductive dechlorination.
Further groundwater monitoring is planned and an additional liner-soil sampling campaign should reveal the present NZVI spatial distribution as well as NZVI consumption during 1 year, as important steps towards mass balance and efficiency estimations.
Biological activity during field application
Dechlorinating microorganisms were detected by PCR in the source area and in the downgradient plume at all sampling times. Furthermore, formation and accumulation of cis-DCE point to the occurrence of biological dechlorination. Thus, there was no sustained inhibition of the naturally occurring dechlorinating microorganisms by injection of NZVI. Rather biological dechlorination was stimulated by the organics introduced during NZVI application, despite transient unfavorable pH values due to fermentation. Decreasing sulfate concentrations indicated a microbial sulfate reduction. Both dechlorination as well as sulfate reduction consume hydrogen, thus avoiding gas clogging. At some measurement points, a transient inhibitory effect on luminescent bacteria up to a maximum LID (Lowest ineffective dilution) of 16 was measured, which returned to no effective toxicity in the course of monitoring.
Comparative studies in the laboratory using soil and groundwater from the site confirmed the findings of the field application and showed a stimulation of biological dechlorination by injection of NZVI suspension.
Using the fitted parameterization reported in Table 3, the measurement data of the particle concentration profile in the column test and its simulation with E-MNM1D as well as with the extended OGS are in satisfactory agreement (Fig. 9), showing mainly a typical shape of strained particles (Tosco and Sethi 2010; Bradford et al. 2002).
Even though the variances of the measurements of the sediment cores, probably caused by unknown inhomogeneous permeability distributions, are partially severe (Fig. 9), the simulation is quite similar to the measured mean particle distribution indicating the suitability of the developed model and the parameterization for supporting application design. For 0.5 g/kg (sand) as a concentration used to define the transport length of Fe(0)tot, the predicted particle spreading is about 0.7 m. To consider pore clogging effects due to particle deposition, the OGS model approach is actually applied with a permeability–porosity relationship.
Since the aquatic tests use aerobic organisms, only oxidized NZVI were tested. In the environment, NZVI are transformed into oxides thus being a more realistic object of investigation. The particles agglomerated within minutes and were deposited at the bottom of the test vessels in the presence of oxygen.
In the Daphnia magna immobilization test, the particles caused immobility at concentrations below 1,000 mg/L. An attachment of the particles to the animals’ body was observed which was counteracted by additional movement. After molting, these particles were shed and only minor consequent attachment was observed. The samples were not mutagenic in the Ames fluctuation test. The fish embryo toxicity test was the most sensitive assay in the applied biotest battery. The embryos of Danio rerio showed both lethal and sublethal (delayed hatching) effects at concentrations below 100 mg/L. The nanomaterials associated with the fish eggs’ chorion and impacted the evaluation. Any unhatched fish eggs were opened at the end of the test.
The algae growth inhibition test was not applicable as the algae clogged between the particles, thus resulting in shading and a growth inhibition by physical means. Within the Ames fluctuation test, the bacterial density measurement was affected by deposited nanomaterials in the exposure vessels. As a solution, a multi-point measurement was suggested to overcome this problem.
We compared our results with studies on iron oxide nanomaterials and aged zero-valent nanomaterials. The low toxicity is in agreement with results of Filser et al. (2013) and Sun et al. (2011) for iron oxide nanoparticles and of Marsalek et al. (2012) and Phenrat et al. (2009) for aged zero-valent iron nanomaterials. However, a study by Li et al. (2009) on adult medaka fish using iron oxide nanoparticles reported deleterious changes on gills and intestine filament at concentrations of 5 mg/L and suggested oxidative stress as the mode of action. Moreover El-Temsah and Joner (2012) reported an effect on earthworms‘ reproduction in concentrations <100 mg/kg for aged NZVI.
Based on these results, aged NZVI can be classified in aquatic acute category 3 following the globally agreed system of classification and labeling of chemicals (European Community 2008). Consequently, with regard to ecotoxicity, the use of NZVI can be recommended for remediation purposes. However, both the effects of a prolonged or chronic exposure and interactions with other chemicals are unknown.
As a result of the research activities presented, a new type of NZVI particles with advantages over preexisting NZVI particles concerning reactivity, transportability and economy (<50 €/kg) was developed for groundwater remediation. The relatively high reactivity of the iron flakes may originate from the specific shape of the particles exhibiting higher specific surface area compared to more spherical shapes. The manufacturing process and the resulting shape may also enable the development of pits and edges favoring pitting corrosion and thus enhancing electron transfer. Furthermore, corrosion and electron transfer are increased by impurities in the ZVI, the amount of which is higher in the iron flakes due to the raw materials used than in relatively pure precipitated nanoiron particles.
The newly developed iron flakes are the only NZVI material for which transportability in relevant concentrations for up to 1 m and more could be confirmed by solid-phase analysis in the lab and the field. These comparatively good transport characteristics with a relatively homogeneous particle distribution and low-permeability decreases are probably also the result of the flake-like shape decreasing the particle sedimentation rate and pore plugging since thin particles retained in pore throats allow higher water flow compared to spherical particles.
These practical important advantages together with the measured low acute toxicity of the particles, the observed synergies with microbial dechlorination and the developed possibility for numerical optimization of field applications illustrate that the iron flakes have a significant potential for source treatment of CHCs and other contaminants. The iron flakes are at present produced for research purposes and on demand.
Aktaş Ö, Schmidt KR, Mungenast S, Stoll C, Tiehm A (2012) Effect of chloroethene concentrations and granular activated carbon on reductive dechlorination kinetics and growth of Dehalococcoides spp. Bioresour Technol 103:286–292
Bonder MJ, Zhang Y, Kiick KL, Papaefthymiou V, Hadjipanayis GC (2007) Controlling synthesis of Fe nanoparticles with polyethylene glycol. J Magn Magn Mater 311:658–664
Bradford SA, Yates SR, Bettahar M, Simunek J (2002) Physical factors affecting the transport and fate of colloids in saturated porous media. Water Resour Res 38:1327
Bradford SA, Simunek J, Bettahar M, van Genuchten MT, Yates SR (2003) Modeling colloid attachment, straining, and exclusion in saturated porous media. Environ Sci Technol 37:2242–2250
Bradford SA, Simunek J, Bettahar M, van Genuchten MT, Yates SR (2006) Significance of straining in colloid deposition: evidence and implications. Water Resour Res 42:W12S15
Bradford SA, Torkzaban S, Leij F, Simunek J, van Genuchten MT (2009) Modeling the coupled effects of pore space geometry and velocity on colloid transport and retention. Water Resour Res 45:W02414
Bradford SA, Torkzaban S, Simunek J (2011) Modeling colloid transport and retention in saturated porous media under unfavorable attachment conditions. Water Resour Res 47:W10503
Braunbeck T, Boettcher M, Hollert H, Kosmehl T, Lammer E, Leist E, Rudolf M, Seitz N (2005) Towards an alternative for the acute fish LC (50) test in chemical assessment: the fish embryo toxicity test goes multi-species—an update. Altex 22:87
Cullen E, O’Carroll DM, Yanful EK, Sleep B (2010) Simulation of the subsurface mobility of carbon nanoparticles at the field scale. Adv Water Resour 33:361–371
DIN EN ISO 9277:1995 Determination of the specific surface area of solids by gas adsorption using the BET method; German version DIN ISO 9277:1995; DIN German Institute for Standardization, p 12
DIN EN ISO 15088:2009 Water quality—determination of the acute toxicity of waste water to zebrafish eggs (Danio rerio) (ISO 15088:2007); German version EN ISO 15088:2008. DIN EN ISO 15088. DIN German Institute for Standardization, p 20
DIN EN ISO 6341:2010 Water quality—determination of the inhibition of the mobility of Daphnia magna Straus (Cladocera, Crustacea)—acute toxicity test (ISO/DIS 6341:2010); German version prEN ISO 6341:2010. DIN EN ISO 6341. DIN German Institute for Standardization, p 33
DIN EN ISO 8692:2010 Water quality—fresh water algal growth inhibition test with unicellular green algae (ISO 8692:2010); German version EN ISO 8692:2010. DIN EN ISO 8692. DIN German Institute for Standardization, p 30
Doong R, Lai Y (2006) Effect of metal ions and humic acid on the dechlorination of tetrachloroethylene by zerovalent iron. Chemosphere 64:371–378
El Fantroussi S, Mahillon J, Naveau H, Agathos SN (1997a) Introduction and PCR detection of Desulfomonile tiedjei in soil slurry microcosms. Biodegradation 8:125–133
El Fantroussi S, Mahillon J, Naveau H, Agathos SN (1997b) Introduction of anaerobic dechlorinating bacteria into soil slurry microcosms and nested-PCR monitoring. Appl Environ Microbiol 63(2):806–811
Elliott DW, Zhang WX (2001) Field assessment of nanoscale bimetallic particles for groundwater treatment. Environ Sci Technol 35:4922–4926
El-Temsah YS, Joner EJ (2012) Ecotoxicological effects on earthworms of fresh and aged nano-sized zero-valent iron (nZVI) in soil. Chemosphere 89:76–82. doi:10.1016/j.chemosphere.2012.04.020
European Community (2008) Regulation (EC) No 1272/2008 of the European Parliament and of the Council of 16 December 2008 on classification, labelling and packaging of substances and mixtures, amending and repealing Directives 67/548/EEC and 1999/45/EC, and amending Regulation (EC) No 1907/2006 (Text with EEA relevance). In: Union OJotE (ed), 1272/2008
Filser J, Arndt D, Baumann J, Geppert M, Hackmann S, Luther EM, Pade C, Prenzel K, Wigger H, Arning J, Hohnholt MC, Koser J, Kuck A, Lesnikov E, Neumann J, Schutrumpf S, Warrelmann J, Baumer M, Dringen R, von Gleich A, Swiderek P, Thoming J (2013) Intrinsically green iron oxide nanoparticles? From synthesis via (eco-)toxicology to scenario modelling. Nanoscale 5:1034–1046. doi:10.1039/c2nr31652h
Henn KW, Waddill DW (2006) Utilization of nanoscale zero-valent iron for source remediation—a case study. Remediation 16:57–77. doi:10.1002/rem.20081
Hornbruch G, Strutz T, Dahmke A, Köber R. (in prep.) Simulation of NZVI transport in porous media under variable injection conditions
ISO 11350:2011 Water quality—determination of the genotoxicity of water and waste water—Salmonella/microsome fluctuation test (Ames fluctuation test). ISO 11350. ISO International Organization for Standardization, p 42
Johnson PR, Elimelech M (1995) Dynamics of colloid deposition in porous media—blocking based on random sequential adsorption. Langmuir 11:801–812
Johnson PR, Sun N, Elimelech M (1996) Colloid transport in geochemically heterogeneous porous media: modeling and measurements. Environ Sci Technol 30:3284–3293
Kanel SR, Goswami RR, Clement TP, Barnett MO, Zhao D (2008) Two dimensional transport characteristics of surface stabilized zero-valent iron nanoparticles in porous media. Environ Sci Technol 42:896–900
Ko CH, Elimelech M (2000) The “shadow effect” in colloid transport and deposition dynamics in granular porous media: measurements and mechanisms. Environ Sci Technol 34:3681–3689. doi:10.1021/es0009323
Kolditz O, Bauer S, Bilke L, Böttcher N, Delfs JO, Fischer T, Görke UJ, Kalbacher T, Kosakowski G, McDermott CI, Park CH, Radu F, Rink K, Shao H, Shao HB, Sun F, Sun YY, Singh AK, Taron J, Walther M, Wang W, Watanabe N, Wu Y, Xie M, Xu W, Zehner B (2012) OpenGeoSys: an open-source initiative for numerical simulation of thermo-hydro-mechanical/chemical (THM/C) processes in porous media. Environ Earth Sci 67:589–599. doi:10.1007/s12665-012-1546-x
Komar PD, Reimers CE (1978) Grain shape effects on settling rates. J Geol 86(2):193–209
Kuiken T (2010) The project on emerging nanotechnologies and nanoremediation. Environ Earth Sci 60(4):903–907
Li H, Zhou Q, Wu Y, Fu J, Wang T, Jiang G (2009) Effects of waterborne nano-iron on medaka (Oryzias latipes): antioxidant enzymatic activity, lipid peroxidation and histopathology. Ecotoxicol and Environ Saf 72:684–692. doi:10.1016/j.ecoenv.2008.09.027
Liu Y, Choi H, Dionysiou D, Lowry GV (2005) Trichloroethene hydrodechlorination in water by highly disordered monometallic nanoiron. Chem Mater 17:5315–5322
Löffler FE, Sun Q, Li J, Tiedje JM (2000) 16S rRNA gene-based detection of tetrachloroethene dechlorinating Desulforomonas and Dehalococcoides species. Appl Environ Microbiol 66(4):1369–1374
Loveland JP, Bhattacharjee S, Ryan JN, Elimelech M (2003) Colloid transport in a geochemically heterogeneous porous medium: aquifer tank experiment and modeling. J Contam Hydrol 65:161–182
Marsalek B, Jancula D, Marsalkova E, Mashlan M, Safarova K, Tucek J, Zboril R (2012) Multimodal action and selective toxicity of zerovalent iron nanoparticles against cyanobacteria. Environ Sci Technol 46:2316–2323. doi:10.1021/es2031483
Matheson LJ, Tratnyek P (1994) Reductive dehalogenation of chlorinated methanes by iron metal. Environ Sci Technol 28:2045–2053
McDowell-Boyer LM, Hunt JR, Sitar N (1986) Particle transport through porous media. Water Resour Res 22:1901–1921
O’Carroll D, Sleep B, Krol M, Boparai H, Kocur C (2012) Nanoscale zero valent iron and bimetallic particles for contaminated site remediation. Adv Water Resour 51:104–122
Phenrat T, Saleh N, Sirk K, Tilton RD, Lowry GV (2007) Aggregation and sedimentation of aqueous nanoscale zerovalent iron dispersions. Environ Sci Technol 41:284–290
Phenrat T, Saleh N, Sirk K, Kim H-J (2008) Stabilization of aqueous zerovalent iron dispersions by anionic polyelectrolytes: adsorbed anionic polyelectrolyte layer properties and their effect on aggregation and sedimentation. J Nanoparticle Res 10:795–814
Phenrat T, Long TC, Lowry GV, Veronesi B (2009) Partial oxidation (“aging”) and surface modification decrease the toxicity of nanosized zerovalent iron. Environ Sci Technol 43:195–200
Phenrat T, Cihan A, Kim HJ, Mital M, Illangasekare T, Lowry GV (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 44:9086–9093
Phenrat T, Fagerlund F, Illangasekare T, Lowry GV, Tilton RD (2011) Polymer-modified Fe0 nanoparticles target entrapped NAPL in two dimensional porous media: effect of particle concentration, NAPL saturation, and injection strategy. Environ Sci Technol 45:6102–6109. doi:10.1021/es200577n
Rajendra S, Reenkala SM, Anthony N, Ramaraj R (2002) Synergistic corrosion inhibition by the sodium dodecyl sulphate–Zn2+ system. Corros Sci 44:2243–2252
Reifferscheid G, Maes H, Allner B, Badurova J, Belkin S, Bluhm K, Brauer F, Bressling J, Domeneghetti S, Elad T (2012) International round-robin study on the Ames fluctuation test. Environ Mol Mutagen 53:185–197
Schmidt K, Stoll C, Tiehm A (2006) Evaluation of 16S-PCR detection of Dehalococcoides at two chloroethene-contaminated sites. Water Sci Technol: Water Supply 6(3):129–136
Smits THM, Devenoges C, Szynalski K, Maillard J, Holliger C (2004) Development of a real-time PCR method for quantification of the three genera Dehalobacter, Dehalococcoides, and Desulfitobacterium in microbial communities. J Microbiol Methods 57:369–378
Stieber M, Putschew A, Jekel M (2011) Treatment of pharmaceuticals and diagnostic agents using zero-valent iron—kinetic studies and assessment of transformation products assay. Environ Sci Technol 45(11):4944–4950
Sun N, Elimelech M, Sun NZ, Ryan JN (2001a) A novel two-dimensional model for colloid transport in physically and geochemically heterogeneous porous media. J Contam Hydrol 49:173–199
Sun N, Sun NZ, Elimelech M, Ryan JN (2001b) Sensitivity analysis and parameter identifiability for colloid transport in geochemically heterogeneous porous media. Water Resour Res 37:209–222
Sun J, Wang S, Zhao D, Hun FH, Weng L, Liu H (2011) Cytotoxicity, permeability, and inflammation of metal oxide nanoparticles in human cardiac microvascular endothelial cells. Cell Biol Toxicol 27:333–342. doi:10.1007/s10565-011-9191-9
Taghavy A, Costanza J, Pennell KD, Abrioala LM (2010) Effectiveness of nanoscale zero-valent iron for treatment of PCE-DNAPL source zone. J Contam Hydrol 118:128–142
Tiehm A, Schmidt KR (2011) Sequential anaerobic/aerobic biodegradation of chloroethenes—aspects of field application. Curr Opin Biotechnol 22(3):415–421
Tiraferri A, Tosco T, Sethi R (2011) Transport and retention of microparticles in packed sand columns at low and intermediate ionic strengths: experiments and mathematical modeling. Environ Earth Sci 63(4):847–859
Tosco T, Sethi R (2009) MNM1D: a numerical code for colloid transport in porous media: implementation and validation. Am J Environ Sci 5:517–525
Tosco T, Sethi R (2010) Transport of non-newtonian suspensions of highly concentrated micro- and nanoscale iron particles in porous media: a modeling approach. Environ Sci Technol 44:9062–9068. doi:10.1021/es100868n0013
Wang C, Wang L, Wang Y, Liang Y, Zhang J (2012) Toxicity effects of four typical nanomaterials on the growth of Escherichia coli, Bacillus subtilis and Agrobacterium tumefaciens. Environ Earth Sci 65(6):1643–1649
Yao K-M, Habibian MT, O’Melia CR (1971) Water and waste water filtration. Concepts and applications. Environ Sci Technol 5:1105–1112. doi:10.1021/es60058a005
Zhang WX (2003) Nanoscale iron particles for environmental remediation: an overview. J Nanopart Res 5:323–332
Zhao MZ (2013) In situ dechlorination in soil and groundwater using stabilized zero-valent iron nanoparticles: some field experience on effectiveness and limitations. In: Novel solutions to water pollution, Chapter 6, pp 79–96, Chapter doi:10.1021/bk-2013-1123.ch006, ACS Symposium Series, vol 1123, ISBN13:9780841227545e, ISBN:9780841227552
Zhuang Y, Jin LT, Luthy RG (2012) Kinetics and pathways for the debromination of polybrominated diphenyl ethers by bimetallic and nanoscale zerovalent iron: effects of particle properties and catalyst. Chemosphere 89(4):426–432
This work is part of the joint project NAPASAN (Nanoparticles for ground water remediation) which was funded by the German Federal Ministry for Education and Research (BMBF) under the Grant Number 03X0097 within the research program NanoNature (Nanotechnologies for Environmental Protection—Value and Impact) which is part of the framework program WING (Material Innovations for Industry and Society).
About this article
Cite this article
Köber, R., Hollert, H., Hornbruch, G. et al. Nanoscale zero-valent iron flakes for groundwater treatment. Environ Earth Sci 72, 1 (2014). https://doi.org/10.1007/s12665-014-3239-0
- Nanoscale zero-valent iron
- Field test
- Numerical model