FeS Nanoparticles: Electrolyte Composition Effects on

Oct 18, 2012 - Division of Environmental and Biomolecular Systems, Oregon Health ... that the higher reactivity of Fe/FeS is due to greater abundance ...
0 downloads 0 Views 559KB Size
Article pubs.acs.org/est

Reactivity of Fe/FeS Nanoparticles: Electrolyte Composition Effects on Corrosion Electrochemistry David Turcio-Ortega,† Dimin Fan,† Paul G. Tratnyek,†,* Eun-Ju Kim,‡ and Yoon-Seok Chang‡ †

Division of Environmental and Biomolecular Systems, Oregon Health & Science University, 20000 NW Walker Road, Portland, Oregon 97006, United States ‡ School of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang, 790-784, Republic of Korea S Supporting Information *

ABSTRACT: Zerovalent iron nanoparticles (Fe0 NPs or nZVI) synthesized by reductive precipitation in aqueous solution (Fe/FeO) differ in composition and reactivity from the NPs obtained by reductive precipitation in the presence of a Ssource such as dithionite (Fe/FeS). To compare the redox properties of these types of NPs under a range of environmentally relevant solution conditions, stationary powder disk electrodes (PDEs) made from Fe/FeO and Fe/FeS were characterized using a series of complementary electrochemical techniques: opencircuit chronopotentiometry (CP), linear polarization resistance (LPR), electrochemical impedance spectroscopy (EIS), and linear sweep voltammetry (LSV). The passive films on these materials equilibrate within minutes of first immersion and do not show further breakdown until >1 day of exposure. During this period, the potentials and currents measured by LPR and LSV suggest that Fe/FeS undergoes more rapid corrosion and is more strongly influence by solution chemical conditions than Fe/FeO. Chloride containing media were strongly activating and natural organic matter (NOM) was mildly passivating for both materials. These effects were also seen in the impedance data obtained by EIS, and equivalent circuit modeling of the electrodes composed of these powders suggested that the higher reactivity of Fe/FeS is due to greater abundance of defects in its passive film.



material of the latter type is Fe0 coated with iron sulfides (Fe/ FeS), which can be prepared efficiently using a single-step synthesis involving dithionite.16,17 We recently described optimization of this synthesis method, the composition and structure of the resulting NPs, and the reactivity of this material with TCE in solutions of deionized water.16 To reliably assess the potential advantages of Fe/FeS in field applications, however, it is necessary to obtain a more in depth understanding of the role that passive film composition and solution conditions play in determining the reactivity of this system. Based on prior work, several aspects are of particular interest. The first is whether the role of iron sulfide is as a mediator of reaction between Fe0 and contaminants from solution, since both similarities and differences have been described regarding the reduction of chlorinated solvents by iron sulfides (e.g., mackinawite) vs iron oxides (e.g., magnetite).18−20 The second aspect is to identify the effect of major organic absorbates such as surfactants or natural organic matter (NOM), since these can have large impacts on the mobility and reactivity of Fe0 NPs.5,21,22 A third aspect is to understand the

INTRODUCTION Among the many ways to enhance or expand the applicability of zerovalent metals to remediation of contaminated materials, formulation as nanoparticles (NPs) has received the most attention.1−3 Most of these formulations consist of NPs with an Fe0 core and (mainly) iron oxide shell (Fe/FeO) in aqueous suspensions stabilized by various surfactants.4,5 These materials have potential advantages including high reactivity and substantial mobility in porous media, but there are uncertainties such as whether adequate quantities of Fe0 can be delivered sufficient distances in the subsurface to provide a sustainable solution to contamination at real field sites.6 To maximize the benefits and avoid the limitations of conventional Fe0 NP formulations, many alternatives have been developed. These included addition of catalytic metals such as Pd or Ni (e.g., for degradation of polychlorinated dibenzodioxins, PCDDs7); encapsulation in colloidal matrices (e.g., silica8,9 and carbon10,11); and immobilization in organic membranes.12,13 A more subtle approach is controlled modification of the NP shell to impart more selective reactivity with contaminants, less superfluous reaction with water or aquifer matrix materials, and/or colloidal properties that favor transport. These modifications can involve minor constituents of the NP oxide shell (i.e., doping14,15) or replacement of oxide shell with a passive film of significantly different composition. A © 2012 American Chemical Society

Received: Revised: Accepted: Published: 12484

August 23, 2012 October 16, 2012 October 18, 2012 October 18, 2012 dx.doi.org/10.1021/es303422w | Environ. Sci. Technol. 2012, 46, 12484−12492

Environmental Science & Technology

Article

effects of inorganic cations and anions commonly found in groundwater and that are known to be activating or passivating vis-à-vis corrosion, including Ca2+, Mg2+, Cl−, carbonate, silicate, etc.23−25 In addition to the above concerns, there is the cross-cutting issue of how the core/shell structure, composition, and reactivity of NPs change over time. Some studies have focused on “aging” of Fe/FeO NPs over the time period from first immersion of the particles up to about 1 month of exposure to solution.26−32 However, other studies have chosen to bypass the early stages of aging by “pre-exposure” to solution conditions (e.g., refs 5 and 33), based on the assumptions that aged NPs would give more stable and reproducible contaminant degradation kinetics and that the results would be more relevant to full-scale environmental applications. The latter assumption was true when most groundwater remediation was done with Fe0 NPs that were manufactured before delivery to a site, but current practice has changed to favor onsite synthesis and prompt injection of the resulting NP suspension.34 Under this scenario, the early stages of NP aging could be the most important in controlling the outcome of a remediation application. This period is also likely to be the most dynamic with respect to NP properties, so characterization of these changes is best done with methods that are in situ and real time, such as the electrochemical techniques used in this study. The electrochemical approach taken here is intended to complement other material and reactivity characterizations of Fe/FeS presented elsewhere. In prior work,16 an optimized method of preparing Fe/FeS was developed, based on the structure and composition of the resulting material and its reactivity with a model contaminant (trichloroethylene) in pure water solutions. The goal of this study was to better understand the fundamental factors that control the reactivity of Fe/FeS over a range of solution conditions, and the effect of solution conditions on TCE degradation kinetics will be addressed in a separate manuscript.35 While the practical motivation for this work was to assess whether Fe/FeS offers substantial advantages over Fe/FeO for treatment of contaminated groundwater, the results also address a number of issues with more fundamental or general significance. These issues include (i) the differences in reactivity of zerovalent metal nanoparticles with a shell composed of metal sulfides rather than oxides, (ii) the dynamics of change in NP properties during the time period after first exposure to new solution conditions, and (iii) the distinctly different types of effects that inorganic and organic adsorbants have on the electronic properties on Fe0 NP coatings under solutions conditions.

material by reverse osmosis of brown water from a wetland pond in Georgetown, SCand is similar in composition and properties to Suwannee River Humic Acid37 The concentration of NOM was expressed as dissolved organic carbon (mg/L as DOC) assuming the organic carbon content of NOM-GT was 48.3%.38 The full procedure for synthesis of Fe/FeS NPs has been described elsewhere.16 Briefly, Na2S2O4 (3.0 g) was dissolved in 1 L of 0.8 M NaBH4, and the resulting solution was then added dropwise (40−50 drops min−1) to 0.5 M FeCl3 until a 3:1 volume ratio of the two solutions was reached. The resulting Fe/FeS suspension was transferred to an anaerobic chamber (95% N2 with 5% H2), and the solids were recovered by flash drying (as described previously39). The dry powder recovered was stored in the anaerobic chamber. Electrochemical Methods. Electrochemical characterizations were performed with an Autolab PGSTAT30 (EcoChemie, Utrecht, The Netherlands) and a 3-electrode cell containing a powder disk electrode (PDE) as the working electrode, a Pt wire counter electrode, and a Ag/AgCl reference electrode. All potentials are reported relative to the Ag/AgCl reference. PDEs were prepared with roughly 0.02 g of the dry Fe/FeO or Fe/FeS, using the procedure that we developed previously40 and have since used with several electrochemical techniques on a variety of systems.5,22,25,26,36,41 In this study, we applied a complementary combination of electrochemical techniques in a sequence that efficiently proceeds from less-destructive to more-destructive: first open-circuit chronopotentiometry (CP) with brief interruptions for linear polarization resistance (LPR) measurements, followed by electrochemical impedance spectroscopy (EIS), and finally linear sweep voltammetry (LSV). For each experiment, the PDE was freshly packed with Fe/ FeO or Fe/FeS and then immersed into test media consisting of either 0.1 M NaCl; 0.01 M NaCl; 0.01 M CaCl2; 0.01 M MgCl2; 0.1 M TRIS at pH values of 2.1, 7.5, and 10.7; or 21 mg/L NOM (all prepared in deoxygenated, deionized (DI) water). CPs were collected immediately after immersion, for approximately 1 h. The first LPR measurement was made when the CP was interrupted at 30 min and a second LPR was obtained when the CP was completed (at 60 min). For LPR measurements, ± 10 mV with respect to the open circuit potential (Eoc) was applied to the PDE at a scan rate of 2 mV/s. EIS was performed over the full range of frequencies recommended for our potentiostat: 10 kHz to 10 mHz. Seven data points were measured at an amplitude of 10 mV. The LSVs were obtained at the end of each batch of experiments by polarizing the PDE at a scan rate of 2 mV/s from −250 mV to +1100 mV with respect to Eoc.

EXPERIMENTAL SECTION Chemicals and Materials. All chemical solutions were prepared by dissolving reagent-grade chemicals (FeCl3·6H2O, NaBH4, NaCl, CaCl2·2H2O, MgCl2, NaOH, and HCl) into deionized water that was degassed with high-purity N2 for 2 h. Tris buffer was prepared from 0.1 M tris(hydroxymethyl)aminomethane with pH adjustment using 0.1 M HCl or 0.1 M NaOH. NOM solutions were prepared with the same standard material that we have referred to as NOM-GT in previous studies of NOM effects on the reactivity of Fe0.5,22,36 This material was originally provided by Baohua Gu (Oak Ridge National Laboratory, Oak Ridge, TN)who obtained the raw

RESULTS AND DISCUSSION In our original report on synthesis and characterization of Fe/ FeS with dithionite,16 we varied its concentration and found that material obtained using ∼2.0 mg/L dithionite gave the highest rate of trichloroethylene (TCE) reduction (before and after normalization to surface area) in batch experiments containing pure water. For this study, we used only Fe/FeS prepared with 3.0 mg/L dithionite (to be consistent with other follow-up studies, such as ref 35), but varied the solution chemistry by adding HCl, NaOH, NaCl, MgCl2, CaCl2, TRIS, and NOM. The reactivity of Fe/FeS in these solutions without the presence of any contaminantwas characterized by electrochemical methods, and these data are presented in





12485

dx.doi.org/10.1021/es303422w | Environ. Sci. Technol. 2012, 46, 12484−12492

Environmental Science & Technology

Article

Figure 1. Nyquist (left) and Bode (right) plots of EIS data for Fe/FeO (blue) and Fe/FeS (red) in (A) 0.1 and 0.01 M NaCl, (B) 0.01 M CaCl2 and 0.01 M MgCl2, (C) 0.01 M TRIS at pH 2.1, 7.5, and 10.7, (D) deoxygenated DI water and 21 mg/L NOM.

Linear Polarization Resistance. After 30 and 60 min of immersion in the various test solutions, the potential of the PDE was ramped ± 10 mV around Eoc at a scan rate of 2 mV/s. The resulting current vs potential data were used to determine the polarization resistance at Eoc as defined by eq 1.43

the order that the methods were applied. A companion study using Fe/FeS made the same way and a similar range of solution conditionsbut focused the effects of solution conditions on TCE reduction rateswill be published separately.35 Chronopotentiometry. The results from chronopotentiometry (CP) are presented and discussed in Section 1 of the Supporting Information (SI). The data show that changes in open-circuit potential (Eoc) during the first hour after immersion of the PDE are typical of what we have seen in prior work with various types of Fe/FeO.5,26,42 The CPs for PDEs composed of Fe/FeS are more dynamic in several ways (i.e., change in Eoc with time or with solution conditions), suggesting the material is generally more reactive. Further insight into the enhanced reactivity Fe/FeS is provided by the additional electrochemical characterizations performed on each experiment.

⎛ ΔE ⎞ B ⎜ ⎟ = Rp = ⎝ Δi ⎠ E → 0 icorr

(1)

where Rp is the polarization resistance, icorr the corrosion current, and B is a constant that depends on the anodic and the cathodic slopes (βa, βc) of the polarization curves:

B=

βa βc 2.303(βa + βc)

(2)

To evaluate eq 2, the constants βa and βc were obtained by Tafel Analysis of the corresponding segments of the polar12486

dx.doi.org/10.1021/es303422w | Environ. Sci. Technol. 2012, 46, 12484−12492

Environmental Science & Technology

Article

electrolytes and at high f (>102 Hz) in lower pH TRIS buffered electrolytes. The former effect may reflect the high capacitance of Fe/FeS (see below) and the latter may reflect ionic associations involving the protonated TRIS, but neither effect can be fully interpreted without further study. Equivalent Circuit Modeling. For a more systematic and quantitative evaluation of the EIS results, the impedance data were fit to an equivalent circuit (EC) model using the method developed by Boukamp48,49 and implemented in the Autolab FRA software (Ver. 4.9.007). A common, minimal EC model was selected based on analogy between the experimental conditions used in this study and other systems where a conductive metal surface is coated with an imperfectly conductive porous film (e.g., steel in concrete,50 degraded polymer coated metals,51 or nanoparticle reinforced polymer protected metals52). The model assumes the overall impedance response is determined by the three types of interfaces represented in Figure 2: (i) the interface were the polished

ization curves obtained by LSV (described later in the section on Linear Sweep Voltammetry). All of the values of Rp and icorr obtained by LPR are given in Table S1 of the SI. The results obtained at 30 and 60 min are similar, consistent with the lack of significant changes in the CP data during this time interval (SI Figure S1). Other trends in the LPR datadue to differences between Fe/FeO and Fe/FeS under the various solution conditions testedare discussed the Summary Analysis section below. Electrochemical Impedance Spectroscopy. The multiple aspects of EIS data are usually presented in several complementary formats, including Nyquist (complex plane) and Bode plots. Nyquist plots represent the imaginary (Z″) vs real (Z′) parts of the impedance measured over the frequency range of the perturbation (in this case, a sinusoidal applied potential). Both types of plots are interpreted as the result of an equivalent circuit consisting of a resistance in series with a combination of a resistor and a capacitor, which represent the polarization resistance and double layer capacitance, respectively. In a typical Nyquist plot, the left (high frequency) end of the semicircular response represents impedance from ohmic resistance and the right (low frequency) end of the semicircle is the sum of the ohmic resistance and the polarization resistance.44 The Nyquist plots obtained in Farrell’s studies of contaminant removal from solution by zerovalent iron45,46 were fairly typical of the expected results for a bulk Fe0 electrode undergoing a simple corrosion process. However, as shown in Figure 1, the diverse range of conditions used in this study resulted in Nyquist plots that reflect two capacitive loops at very high frequencies for Fe/FeO and Fe/FeS. For each solution condition, Fe/FeS gave relatively lower Z′ (i.e., less impedance) than Fe/FeO, suggesting that the surface coating on Fe/FeS might support greater overall rates of charge transfer. The greatest effect of solution conditions was with NOM, which gave markedly greater impedance than any of the inorganic salts media (Figure 1-D1). Bode plots of EIS data show the log of impedance modulus (|Z|) or phase angle (Phase) vs log frequency (f). In Bode modulus plots, the impedance at the high frequency limit is the ohmic resistance (Rs) and the difference between impedance at the low and high frequency limits is the polarization resistance (or charge transfer resistance, Rct). The Bode modulus plot of the EIS data from this study is shown at the right side of its corresponding Nyquist plot in Figure 1, whereas expected both Fe/FeO and Fe/FeS gave values of Rs that are relatively low for solutions of inorganic salts and high for solutions of organic polyelectrolytes. For each solution condition, the values of |Z| are greater for Fe/FeO than Fe/FeS, suggesting less impedance and more reactivity for Fe/FeS, which is consistent with the trend in Z′ noted from the Nyquist plots. Also evident in both the Nyquist and Bode modulus plots is the large effect of NOM, which shows higher impedance, further indicating lower reactivity (for both Fe/FeO and Fe/FeS) in this medium. Bode phase plots for actively corroding Fe0 typically show a peak of about 40 deg at 10 Hz due to the reaction Fe0 + 2 H+ → H2 + Fe2+ (e.g., as seen in the electrochemical studies of contaminant degradation on an Fe0 wire electrode by Farrell45,47). However, EIS data from this study gave Bode phase plots that are comparatively featureless, suggesting that the overall degree of passivation is high for both Fe/FeO and Fe/FeS. Two subtle but distinctive features of the data for Fe/ FeS are the phase increase at low f (