Journal of Nanoparticle Research

, Volume 12, Issue 6, pp 2057–2068

Reduction of Se(VI) to Se(-II) by zerovalent iron nanoparticle suspensions

Authors

  • Jovilynn T. Olegario
    • Department of Chemistry and BiochemistrySan Francisco State University
  • Nay Yee
    • Department of Chemistry and BiochemistrySan Francisco State University
  • Marissa Miller
    • Department of Chemistry and BiochemistrySan Francisco State University
  • John Sczepaniak
    • Department of Chemistry and BiochemistrySan Francisco State University
    • Department of Chemistry and BiochemistrySan Francisco State University
Research Paper

DOI: 10.1007/s11051-009-9764-1

Cite this article as:
Olegario, J.T., Yee, N., Miller, M. et al. J Nanopart Res (2010) 12: 2057. doi:10.1007/s11051-009-9764-1

Abstract

The reaction of selenate (Se(VI)) with zerovalent iron nanoparticles (nano Fe0) was studied using both conventional batch equilibrium and X-ray spectroscopic techniques. Nano Fe0 has a high uptake capacity for removal of dissolved Se(VI) reaching concentrations as high as 0.10 Se:Fe molar ratio in the solid product mixture. Kinetic studies of the Se(VI) uptake reaction in batch experiments showed an initial reaction rate (0–30 min) of 0.0364 min−1 which was four times greater than conventional Fe0 powder. Analysis of the oxidation state of Se in the solid products by X-ray absorption near edge structure (XANES) spectroscopy showed evidence for the reduction of Se(VI) to insoluble selenide (Se(-II)) species. Structural analysis of the product by extended X-ray absorption fine structure (EXAFS) spectroscopy suggested that Se(-II) was associated with nano Fe0 oxidation products as a poorly ordered iron selenide (FeSe) compound. The fitted first shell Se–Fe interatomic distance of 2.402 (±0.004) Å matched closely with previous studies of the products of Se(IV)-treated Fe(II)-clays and zero-valent iron/iron carbide (Fe/Fe3C). The poorly ordered FeSe product was associated with Fe0 corrosion product phases such as crystalline magnetite (Fe3O4) and Fe(III) oxyhydroxide. The results of this investigation suggest that nano Fe0 is a strong reducing agent capable of efficient reduction of soluble Se oxyanions to insoluble Se(-II).

Keywords

Fe nanoparticlesReductionSelenium(VI)Selenium(-II)SEMEXAFSSoilsRadioactive waste

Introduction

Selenium (Se) occurs naturally in soil and rocks and is associated with run-off from mining operations (Ryser et al. 2005), irrigation drainage from arid soils (Manning and Burau 1995), and in radioactive wastes (Scheinost and Charlet 2008). The toxicity and mobility of Se in the environment depends on redox conditions (Masscheleyn et al. 1990; Tokunaga et al. 1997) and immobilization reactions such as precipitation of reduced Se and adsorption of Se oxyanions on mineral surfaces (Manceau and Charlet 1994; Goldberg 1985; Peak and Sparks 2002; Peak et al. 2006; Neal and Sposito 1989; Neal et al. 1987). Under aerobic conditions, the predominant forms of Se are the soluble oxyanions selenate (SeO42− or Se(VI)) and selenite (SeO32− or Se(IV)). Under anaerobic reducing conditions insoluble elemental Se (Se(0)) and selenide (Se(-II)) are favored (Scheinost and Charlet 2008; Masscheleyn et al. 1990; Tokunaga et al. 1997). Strategies for removal of soluble Se during water treatment will likely involve manipulating the redox-dependent solubility of Se, especially the formation of insoluble Se(0) and Se(-II).

Abiotic reduction of both Se(IV) and Se(VI) by several iron-containing materials such as Fe(II)/(III) green rust (Myneni et al. 1997), Fe(II)-bearing minerals (Scheinost and Charlet 2008; Charlet et al. 2007; Géhin et al. 2007), FeS2 (Breynaert et al. 2008), iron/iron carbide (Fe/Fe3C) (López de Arroyabe Loyo et al. 2008), zerovalent iron (Fe0) (Qiu et al. 2000; Zhang et al. 2005; Morrison et al. 2002; Scheidegger et al. 2003), and NiFe nanoparticles (Mondal et al. 2004) has been reported. Reduction of Se(VI) by Fe0 foil has been investigated using X-ray spectroscopy and both partial reduction to Se(IV) (Qiu et al. 2000) and total reduction to Se(0) (Scheidegger et al. 2003) have been observed. Corrosion of the Fe0 surface forms Fe(II)/(III) green rust which is capable of both Se(IV) and Se(VI) reduction (Myneni et al. 1997; Scheidegger et al. 2003). A recent study (Scheinost and Charlet 2008) reported Se(IV) reduction by Fe(II)-bearing minerals (mackinawite, magnetite, and siderite) and observed four different Se products depending on Fe(II)-bearing phase and pH (red Se(0), gray Se(0), Fe7Se8, and FeSe). The Se products were a complex mixture of nanoscale clusters of reduced Se(0) and Se(-II) coordinated by Fe(III) (Scheinost and Charlet 2008).

The use of particulate Fe0 powder and nanoparticulate Fe0 as remediation materials and as filter substrates for water purification is increasing (Xiong et al. 2009; Manning et al. 2007; Kanel et al. 2007; Liou et al. 2005; Yang and Lee 2005; Lien and Zhang 2001; Choe et al. 2000; Wang and Zhang 1997). Despite previous studies on the reactions of Se with various Fe0-containing materials, there remains a need for detailed information about the identity and structure of the products of reactions of dissolved Se species (e.g., Se(VI)) with both Fe0 powder (μm-mm scale particles) and Fe0 nanoparticles. The objectives of this study were to (1) investigate the reactivity of unmodified nano Fe0 as a remediation material for dissolved Se(VI), (2) identify the crystal structure and morphology of Se(VI) reduction/nano Fe0 oxidation products using XRD (X-ray diffraction) and scanning electron microscope (SEM), and (3) examine the oxidation state and local atomic environment of Se after reaction of Se(VI) with nano Fe0 by X-ray absorption spectroscopy [X-ray absorption near edge structure (XANES) and X-ray absorption fine structure (EXAFS)].

Experimental

Materials

The reagents used in this study (Na2SeO3, Na2SeO4, FeCl2, HCl, NaOH, NaBH4, and 100% ethanol) were reagent grade and deionized (DI) water as used in all preparations. A 100 mesh electrolytic Fe0 powder (Fisher) was used as a reference material in some experiments without further preparation. Nano Fe0 particles were prepared using a method similar to previous studies (Ponder et al. 2000; Manning et al. 2007) by reacting 500 mL of nitrogen-purged 0.012 M FeCl2 (in 30% ethanol) with 8–10 mL of 0.53 M NaBH4 in a nitrogen-purged flask while stirring. This method created an Fe:B molar ratio = 0.4 and typically yielded 0.334 g nano Fe0 per batch. The synthesized nano Fe0 particles were recovered by rinsing with nitrogen-purged 30% ethanol to remove unreacted reagents followed by rinsing the solid thrice with nitrogen-purged DI water. Samples of nano Fe0 particles were either stored in a nitrogen-purged dessicator or immediately subjected to batch reactions as described below. The nano Fe0 particles used in this investigation have been characterized in a previous investigation using transmission electron microscopy, BET N2 surface area, and XRD (Manning et al. 2007).

Se(VI) adsorption

The kinetics and equilibrium reaction experiment for Se(VI) on nano Fe0 and 100 mesh Fe0 were conducted in 40 mL polycarbonate Oak Ridge centrifuge tubes containing 30.0 mg of Fe0. Equilibrium batch adsorption experiments were conducted using 20.0 mL volumes of varying Se(VI) concentration (0.130–3.00 mM, 0.01 M NaCl) and a reaction time of 48 h. For the kinetics experiments, the starting concentration was 0.130 mM Se in 0.01 M NaCl and samples were taken at 15 min intervals for 2 h. Tubes were shaken on a reciprocating shaker and the solids were separated from the solutions by rapid filtration using 0.2 µm Whatman nitrocellulose membrane filters. Total dissolved Se was analyzed by flame atomic absorption spectrometry (FAAS) with a Varian 220 FS AAS.

XRD and SEM-EDX analyses

The crystallography of the nano Fe0 and the Se–Fe reaction products was investigated by preparing batch reactions of Se(VI)-treated nano Fe0. Glass jars (200 mL) containing 0.500 g of synthesized nano Fe0 were treated with varying Se(VI) concentration (0, 0.50, 1.3, 5.0, and 13 mM Se) in 0.010 M NaCl for 7 days. The equilibrium reaction solutions were analyzed by FAAS for total dissolved Se and the reacted nano Fe0 solids were homogenized and dried under nitrogen atmosphere. Powder XRD was performed on the solids with a Bruker D8 ADVANCE X-ray diffractometer using Cu-Kα source radiation (40 keV), Bragg–Brentano primary beam optics (slit width 1.0 mm), and a scan range from 10–100°2θ at 0.05°2θ increments. An energy-dispersive SolX X-ray detector was used to reject X-ray fluorescence from Fe.

Imaging of the reacted and unreacted nano Fe0 solids was performed with a Zeiss Ultra 55 field emission scanning electron microscope (SEM) equipped with a Gemini FEG column. Samples were mounted on aluminum stubs with carbon paint and heated at 80 °C for 20 min. Images were collected with a secondary electron in-lens detector. Elemental composition was analyzed by energy-dispersive X-ray (EDX) analyses with an Oxford X-Max 80 mm2 Si drift detector at a working distance of 10 mm, accelerating voltage of 13 keV, 35° take-off angle, and a count rate of 10,000 cps. This provided an X-ray resolution of 126 eV (FWHM).

X-ray absorption spectroscopy

Preparation of samples for X-ray absorption spectroscopy (XAS) analyses involved using batches of 0.334 g nano Fe0 (6.0 × 10−3 mol Fe) reacted with 500 mL of N2-purged 0.50 mM Se(IV) or Se(VI) in 0.010 M NaCl for 7 days in 500 mL polycarbonate jars. The pH of the reaction mixture was initially adjusted to 7.0 and became alkaline (pH = 8.2) during the course of the reaction. After 7 days, aliquots of the overlying solution were filtered (0.2 μm) and total dissolved Se was analyzed by FAAS. Sample preparation was performed within 7 days of XAS analysis and samples were rinsed with N2-purged 30% ethanol and stored as hydrated pastes in N2-purged bottles prior to XAS analyses. Reference materials including sodium selenide (Na2Se, Alfa Aesar, 99.8%), synthetic iron selenide (FeSe), elemental Se powder (Se(0), Alfa Aesar, 99.5%), and Se(IV)/Se(V) sodium salts were run by XAS. Synthetic FeSe was prepared on-site at the beam line by mixing stoichiometric equivalents of 0.10 M Na2Se and 0.10 M FeCl2 in a N2 atmosphere and rinsing the dark gray FeSe solid with N2-purged DI water.

All samples of nano Fe0, Se(IV)/Se(VI)-treated nano Fe0, and reference materials were analyzed on GeoSoilEnviroCARS beamline 13 BM-D at Argonne National Laboratory’s Advanced Photon Source (APS). Samples were fixed between kapton tape in slotted 2 mm-thick plexiglass plates and mounted in the X-ray beam at a 45o angle. The Se K-edge absorption spectra were collected from 12,500–13200 eV. Energy step intervals of 0.2 eV from 12,640–12,683 eV (XANES region) and 1.0 eV from 12,683–13,200 eV (EXAFS region) were controlled with a Si(111) monochromator. Depending on signal-to-noise criteria, 6–8 scans were collected per sample. Energy calibration was performed by assigning the K-edge energy of 11,919 eV to a gold foil which resulted in a measured Se K-edge of 12,657.4 (±0.4) eV for a Se(0) standard which was in good agreement with previous work (Ryser et al. 2005). X-ray fluorescence was recorded using a 16 element Ge solid-state array multi-element detector (Canberra) and the incident beam intensity (I0) was recorded using a standard ionization detector.

The EXAFS data analysis was performed using SixPACK (Webb 2005) and IFEFFIT (Newville 2001). Individual K-edge scans were averaged followed by linear pre-edge subtraction, background removal, edge-step normalization, isolation of the χ(k) function with a cubic spline function, followed by k3 weighting. Theoretical EXAFS amplitude and phase functions for Se–Fe, Se–O, and Se–Se single scattering paths were then generated by FEFF 8.0 (Rehr et al. 1992). Fitted structural parameters such as amplitude reduction factor (S02), Fermi shift (E0), inter-atomic distance (R), and Debye–Waller factor (σ2) were fitted in k-space. The S02 value was established as 0.80 by fitting of a 13 mM Na2SeO4 solution EXAFS spectrum with fixed values of N = 4.00 O atoms and RSe–O = 1.644 Å. Within each individual sample fit the E0 shift was constrained to ≤10 eV. Both S02 and E0 shift values were held constant amongst shells. The coordination numbers (N) for shells were held constant at reasonable values appropriate to the shell. Variation in N (when used as an adjustable parameter at the end of fitting) did not significantly improve the fits and had no significant effect on the fitted values of R. Therefore, only integer values of N were used and reported herein. Errors in the k-space fits were determined using the following goodness of fit parameter
$$ {\text{Error}}\;\left( \% \right) = \left( {{{\sum {\left( {\chi_{\exp } - \chi_{\text{fit}} } \right)}^{2} } \mathord{\left/ {\vphantom {{\sum {\left( {\chi_{\exp } - \chi_{\text{fit}} } \right)}^{2} } {\sum {\left( {\chi_{\exp } } \right)}^{2} }}} \right. \kern-\nulldelimiterspace} {\sum {\left( {\chi_{\exp } } \right)}^{2} }}} \right) \times 100 $$
(1)
where χexp and χfit are the experimental and calculated χ(k) function data, respectively. Errors in individual parameters were established using SixPACK output and were ±0.004–0.04 Å for R and ±20% for N.

Results and discussion

Se(VI) batch reactions

The equilibrium adsorption isotherm of Se(VI) on nano Fe0 and 100 mesh Fe0 allowed estimation of the Se uptake capacity (mmol/g) and the Se:Fe molar ratio achieved in the solid product (Fig. 1). In both nano Fe0 and 100 mesh Fe0 reaction suspensions, the final pH was between 8.0–8.2 due to generation of OH from water reduction by Fe0 (Farrell et al. 2001). The uptake of Se(VI) was higher on the nano Fe0 than the 100 mesh Fe0 material due to a higher surface area (Johnson et al. 1996; Manning et al. 2007; Kanel et al. 2007). The speciation of dissolved Se(VI)/Se(IV) determined by hydride generation atomic absorption spectrometry (data not shown) indicated that only Se(VI) was present in the aqueous phase. It is possible that Se(IV) forms as an intermediate reduction product during the Se(VI)–Fe0 reaction; however, it is likely that strong adsorption to Fe oxide/hydroxide surfaces occurs (Manceau and Charlet 1994) thus making Se(IV) undetectable in solution.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig1_HTML.gif
Fig. 1

Equilibrium adsorption isotherms for Se(VI)-treated nano Fe0 and 100 mesh Fe0 powder. Reaction conditions: 30 mg Fe0 solid, 20.0 mL volume of varying Se(VI) concentrations (0.130–3.00 mM) in 0.01 M NaCl, reacted for 48 h. The data points were fit using the Langmuir equation \( X = \left( {{\frac{{Kb{\text{C}}}}{{1 + K{\text{C}}}}}} \right) \)where X is adsorbed Se (mmol/g), K is the Langmuir adsorption constant, b is the maximum adsorption constant, and C is the equilibrium Se concentration (mM). The fitted values K and b included in the equations

The Langmuir adsorption isotherm was used to calculate and compare Se adsorption constants (K) and adsorption maxima (b) for Se(VI) on both Fe0 materials. The Langmuir model assumes adsorption on an intact (stable) surface only and therefore is not a valid mechanistic model given that complex redox reactions cause changes to both adsorbent (Fe0) and adsorbate (Se(VI)) during the reaction. The Langmuir model is used here for comparison and quantitative prediction of Se(VI) uptake only and gives a good overall fit to the uptake curve. The highest Se(VI) treatment achieved a maximum Se uptake of 1.75 mmol Se/g Fe0, a value that corresponds to an Se:Fe molar ratio of 0.10, or approximately one Se atom for every ten Fe atoms. This high Se solid content suggests that the solid product may contain a discrete Se or FeSe phase.

In order to quantify the reaction rate, a pseudo-first-order reaction model was applied to the Se(VI)/Fe0 reaction (Johnson et al. 1996):
$$ {\frac{{{\text{d}}\left[ {{\text{Se}}\left( {\text{VI}} \right)} \right]}}{{{\text{d}}t}}} = - k_{\text{SA}} a_{\text{s}} \rho_{\text{m}} \left[ {{\text{Se}}\left( {\text{VI}} \right)} \right] = - k_{\text{obs}} \left[ {{\text{Se}}\left( {\text{VI}} \right)} \right] $$
(2)
where [Se(VI)] is the Se(VI) concentration (mol/L) in aqueous solution at time t (min), kSA is the specific, surface area-dependent reaction rate constant (L/min m2), as is the specific surface area of the nanoparticles (m2/g), ρm is the nanoparticle suspension density (g/L) and kobs is the observed pseudo-first-order rate constant (min−1). Because of the dynamic nature of the Fe0 suspension, accurate determination of surface area (as) is complicated by Fe0 particle aggregation and rapid corrosion of the Fe0 surface in aqueous solution (Johnson et al. 1996; Nurmi et al. 2005). Thus, we relied on measurement of t and [Se(VI)] to derive kobs and the measured kobs are operationally defined only. The results of deriving kobs for the Se(VI)/Fe0 systems are presented in Fig. 2. It was also observed that evaluation of the initial reaction rate from 0–80 min was most effective in explaining the data. Figure 2a shows the time-dependent removal of Se(VI) from solution by both nano Fe0 and 100 mesh Fe0. The lines in Fig. 2a are the back-calculated results derived from the linearized data in Fig. 2b. The time-dependent removal of dissolved Se(VI) shows the considerably greater rate of Se(VI) removal (~4×) from aqueous solution by nano Fe0 compared with 100 mesh Fe0. The slopes of the best-fit linear equations (R2 ≥ 0.94) yielded kobs values of 0.0364 and 0.0091 min−1 for nano Fe0 and 100 mesh Fe0, respectively. The final Se(VI) removal after 48 h was 0.083 mmol Se g−1 Fe0 which compares with a previously reported Se(VI) removal by nano Fe0 of 0.048 mmol Se g−1 Fe0 (Mondal et al. 2004).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig2_HTML.gif
Fig. 2

Time-dependent uptake of Se(VI) on nano Fe0 and 100 mesh Fe0 powder (a) and linearized data (ln[Se]t/[Se]t=0) including fit of a linear model to the data (b). Reaction conditions: 30 mg Fe0 solid, initial Se concentration ([Se]t=0) = 0.130 mM in 0.01 M NaCl, solution volume = 20 mL. Sampling time intervals = 15 min for 2 h

XRD analysis

The results of XRD analysis of unreacted nano Fe0, 100 mesh Fe0, and the reaction products of Se(VI)-treated nano Fe0 are shown in Fig. 3. Unreacted 100 mesh Fe0 is composed of crystalline bcc α-Fe0 (Fig. 3f), whereas nano Fe0 is X-ray amorphous (Fig. 3a). Previous studies of NaBH4 synthesized nano Fe0 have shown that individual nano Fe0 particles contain a thin corrosion layer of Fe(II)/Fe(III) oxide with a shell of metallic Fe0 visible by TEM analysis and Fe0 grain sizes in particle cores of <1.5 nm (Nurmi et al. 2005; Kanel et al. 2007). Oxidation of nano Fe0 in Se(VI)-free aqueous NaCl solution resulted in diffraction peaks indexed as lepidocrocite (γ-FeOOH) as evidenced by peaks at 27.15, 36.25, and 46.90° 2θ (Fig. 3b). The formation of γ-FeOOH is the result of oxidation of Fe(II) formed on the corroding nano Fe0 surface:
$$ {\text{Fe}}^{0} + 2 {\text{H}}_{ 2} {\text{O}} \to {\text{Fe}}^{ 2+ } + {\text{H}}_{ 2} + 2 {\text{OH}}^{ - } $$
(3)
The Fe(II) formed reacts with OH to form ferrous hydroxide (Fe(OH)2) which then oxidizes to Fe(III) (Farrell et al. 2001; Schwertmann and Taylor 1989; Schwertmann and Cornell 1991). Treatment of nano Fe0 with increasing Se(VI) concentration (Fig. 3c–e) caused a conversion of γ-FeOOH to magnetitie (Fe3O4) when the Se(VI) concentration reached 5.0 mM. No evidence was observed in our XRD data suggesting the formation of a crystalline Se-containing phase. Therefore, it is likely that a poorly crystalline Se-containing phase is present in a mixture with either γ-FeOOH or Fe3O4.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig3_HTML.gif
Fig. 3

XRD patterns of untreated, “pristine” nano Fe0 (a), nano Fe0 reacted 7 days in 0.010 M NaCl (zero Se(VI) control) (b), 0.50 mM Se(VI) (c), 1.3 mM Se(VI) (d), and 5.0 mM Se(VI) (e). Also shown is untreated 100 mesh Fe0 powder (f). Values in parentheses are the corresponding Miller indices for magnetite (pattern e) and crystalline bcc α-Fe0 (pattern f). Data have been vertically offset for clarity

SEM-EDX analysis

Particle shape, size, and morphology are important properties that affect the overall chemical reactivity of nanomaterials. The SEM images of unreacted nano Fe0 showed that the material is composed of individual, spherical particles with size ranges from 30-120 nm that form aggregates and chains (Fig. 4a, b). This particle size range estimate agrees well with prior studies of synthetic nano Fe0 and corresponds to a calculated surface area between 5 and 33 m2/g (Nurmi et al. 2005; Wang and Zhang 1997; Lien and Zhang 2001). Reaction with aqueous 0.5 mM Se(VI) solution resulted in a solid concentration of 0.33 mmol Se/g Fe0 (Se:Fe mol ratio of 0.02) and caused extensive particle corrosion combined with cementation of the oxidized particles into μm-sized aggregates (Fig. 4c). The formation of thin Fe oxide/hydroxide surface features (~10 nm thickness) are γ-FeOOH based on XRD anlyses (Fig. 3c) though considerable amorphous Fe oxide/hydroxide is also present in the solid mixture.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig4_HTML.jpg
Fig. 4

FESEM images of nano Fe0 materials investigated in this study: high resolution image of a cluster of untreated nano Fe0 particles (a), lower resolution image of untreated nano Fe0 showing chains and aggregation of individual particles (b), nano Fe0 corrosion product after 7 days reaction in 0.50 mM Se(VI) (c)

Energy-dispersive X-ray analysis (EDX) combined with 2D mapping (Fig. 5) allowed for determination of the spatial distribution of Se in the solid products. A low magnification reference image was taken (Fig. 5a) in combination with 2D mapping of the Se Kα emission line (11.2224 keV) (Fig. 5b). The average EDX spectrum of the high intensity Se Kα emission regions (boxes, Fig. 5a, b) is shown in Fig. 5c. Bright regions of the reference image in Fig. 5a correspond to higher atomic number Se which correlates well with the high intensity Se Kα emission (red areas in Fig. 5b). The 2D map provides general information suggesting that Se is unevenly distributed throughout the sample and is localized in high concentrations on certain particle surfaces.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig5_HTML.gif
Fig. 5

FESEM-EDX analysis showing a reference SEM image of nano Fe0 reacted with 0.50 mM Se(VI) (a), Se Kα EDX map of the reference image (b), and an EDX spectrum resulting from averaging of the regions of interest (boxes in a and b) (c)

XANES analysis

The XANES data for several Se model compounds including 13.0 mM Se(IV) and Se(VI) solutions, Se(0), Na2Se, and synthetic FeSe were analyzed to establish reference X-ray absorption K-edge energies (E0) (Fig. 6). The measured E0 values for Se(VI), Se(IV), and Se(0) were 12,664.49, 12,661.42, and 12,657.51 eV, respectively (Table 1), which compared favorably with Se(IV), Se(VI), and Se(0) E0 values of 12,664.0, 12,662.0, 12,658.0 eV reported previously (Ryser et al. 2005). The measured E0 for synthetic FeSe and Na2Se were 12,656.25 and 12,656.17 eV, respectively. Recent XAS studies of FeSe reported E0 values of 12,654.4 eV (Scheinost and Charlet 2008) and 12,657.0 eV (Ryser et al. 2005). The XANES edge features for Se(-II)-containing materials are distinctly different from Se(VI), Se(IV), and Se(0) and thus are useful as a fingerprint for the Se(-II) oxidation state. The X-ray absorption energy and intensity of the prominent white line peaks arising from the 1s → 4p transition (Fig. 6) varies with both Se valence and bonding environment (Pickering et al. 1995). In particular, as shown in Fig. 6, the intensity of the white line peak decreases in regular series from Se(VI) (E = 12667.0) to Se(-II) (E = 12659.2) due to increased electron population of the valence 4p orbitals.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig6_HTML.gif
Fig. 6

Normalized Se K-edge XANES spectra of Se-containing model compounds

Table 1

Se K-edge EXAFS fitting results for model systems (Na2SeO4 and Na2SeO3, Se(0), Na2Se and FeSe), and Se(IV)- and Se(VI)-treated nano Fe0

Sample

[Se]

E0 (eV)

Shell

(S02)b

N

ΔE0 (eV)

R (Å)c

σ22)

Goodness of fit (%)

Na2SeO4

13.0 (mM)

12664.49

Se–O

0.80

4.00

8.3

1.643 (0.004)

0.0014

6.3

Na2SeO3

13.0 (mM)

12661.42

Se–O

0.80

3.00

12.4

1.699 (0.003)

0.00115

5.3

Se(0)

99.9 (%)

12657.51

Se–Se

0.82

2.00

6.6

2.356 (0.002)

0.0055

4.3

Se–Se

0.82

8.00

6.6

3.51 (0.02)

0.0308

Se–Se

0.82

4.00

6.6

4.01 (0.02)

0.0170

Na2Se

63.2 (%)

12656.17

Se–Se

0.80

4.00

12.2

2.351 (0.006)

0.0084

6.0

FeSe

7.40a

12656.25

Se–Fe

0.84

4.00

8.0

2.390 (0.004)

0.011

5.5

Se–Se

0.84

8.00

8.0

3.98 (0.04)

0.027

Se(IV)–nano Fe0

0.577

12656.18

Se–Fe

0.80

4.00

8.0

2.399 (0.004)

0.008

6.3

Se–Se

0.80

8.00

8.0

4.03 (0.04)

0.018

Se(VI)–nano Fe0

0.529

12657.99

Se–Fe

0.80

4.00

8.0

2.402 (0.004)

0.007

5.3

Se–Se

0.80

8.00

8.0

4.02 (0.04)

0.021

aSe concentration in solids are mmol/g unless specified in parentheses

bThe amplitude reduction factor (S02) was set at 0.840 during initial fitting and then allowed to vary as a final fitting step. N = coordination number (fixed during the fitting procedure), E0 = Fermi shift, R = interatomic distance, σ2 = Debye–Waller factor; goodness of fit parameter = \( \left( {{{\sum {\left( {\varvec{\chi}_{{\user2{\rm exp}}} -\varvec{\chi}_{{\user2{\rm fit}}} } \right)}^{\user2{2}} } \mathord{\left/ {\vphantom {{\sum {\left( {\varvec{\chi}_{{\user2{\rm exp}}} -\varvec{\chi}_{{\user2{\rm fit}}} } \right)}^{\user2{2}} } {\sum {\left( {\varvec{\chi}_{{\user2{\rm exp}}} } \right)}^{\user2{2}} }}} \right. \kern-\nulldelimiterspace} {\sum {\left( {\varvec{\chi}_{{\user2{\rm exp}}} } \right)}^{\user2{2}} }}} \right)\user2{ \times }100 \) where χexp and χfit are the experimental and calculated χ(k) function data, respectively

cInteratomic distance (R) error term in parentheses

Quantitative XANES analysis was performed on the synthetic FeSe sample by linear combination of normalized model spectra (Fig. 7a). Partial oxidation of Se(-II) to Se(0) was detected in the synthetic FeSe sample as indicated by a reduction in the intensity of the XANES peak at 12,665.6 eV and an increase in the white line feature (12,659.2 eV). The FeSe experimental spectrum was well described by a linear combination model of 74% Se(-II) and 26% Se(0) (Fig. 7a) though the post-edge region of the FeSe spectrum was slightly under-predicted. Despite atmospheric control of the FeSe sample it is apparent that partial oxidation of Se(-II) to Se(0) had occurred. Nonetheless, the FeSe, Se(-II), and Se(0) XANES spectra in the this study are nearly identical to those reported in recent investigations (López de Arroyabe Loyo et al. 2008).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig7_HTML.gif
Fig. 7

Least squares fitting of sodium selenide (Na2Se) and elemental Se (Se0) model compound spectra to synthetic iron selenide (FeSe) (a) and fitting of Na2Se and synthetic FeSe to Se(VI)-treated nano Fe0 data (b). Experimental data (points) and model data (lines) are overlaid including the individual contributions of the model compound spectra with percent contributions in parentheses. Also shown below each plot are the residuals (fit-data) for the fits

The XANES spectrum of Se(VI)-treated nano Fe0 suggests that Se(VI) had been fully reduced to Se(-II) as determined by quantitative XANES reconstruction (Fig. 7b). The Se(VI)-treated nano Fe0 was described by 91% Se(-II) (as Na2Se) plus 9% synthetic FeSe. The differences in the XANES spectra of synthetic FeSe and Se(VI)-treated nano Fe0 may be due to the unique chemical environment of Se(-II) formed from reduction of Se(VI) on nano Fe0 surfaces. Formation of extremely small, nanoscale FeSe particles with structure and coordination numbers different from bulk FeSe were detected during the reduction of Se(IV) by Fe/Fe3C nanoparticles (López de Arroyabe Loyo et al. 2008). Also, whereas FeSe was synthesized from homogeneous mixing of Fe2+ and Se2−, Se(VI)-treated nano Fe0 underwent a more complex chemical pathway to form FeSe. The final products include FeSe closely associated with γ-FeOOH, Fe3O4, plus amorphous Fe2+ and Fe3+ hydroxides.

EXAFS analysis

The local atomic structure of Se(VI)-treated nano Fe0 was investigated by EXAFS spectroscopy. An additional sample of Se(IV)-treated nano Fe0 was also investigated for comparison. The results of fitting the EXAFS data are given in Table 1 and shown in Fig. 8. The Se(0) spectrum showed a pronounced first Se–Se backscatterer shell at R = 2.356 (±0.002) Å from N = 2 Se atoms. This structural description agrees favorably with previous EXAFS work on gray, monoclinic Se(0) where R = 2.39 (±0.01) Å (N = 2) (Scheinost and Charlet 2008) and R = 2.35 (±0.01) Å (N = 2.2) (Ryser et al. 2005) were reported. A second, low intensity Se–Se shell at R = 3.51 (±0.02) Å (N = 8) (Table 1) was used in the fit with a small statistical improvement. This second Se–Se shell was justified on the basis of the α-monoclinic Se(0) (Se8 ring structure) (Cherin and Unger 1972). The Na2Se sample was fit with one Se–Se shell of N = 4 Se atoms at 2.351 (±0.006) Å. The Na2Se crystal structure is iso-structural with Na2S (antifluorite structure) and contains Se atoms occupying the face-centered lattice positions and Na atoms occupying the tetrahedral interstitial sites (Sangster and Pelton 1997). Attempts to include Na in the first shell fitting were unsuccessful. Thus, the pronounced peak at 2.351 Å in the Fourier transform (Fig. 8) was determined to be the result of Se nearest neighbors. A similar Se–Se shell at 2.30 Å was detected by EXAFS in CdSe nanoclusters (Marcus et al. 1991).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9764-1/MediaObjects/11051_2009_9764_Fig8_HTML.gif
Fig. 8

Se K-edge EXAFS spectra (k3-weighted χ(k) functions) and Fourier transformed radial structure functions of elemental Se0, Na2Se, synthetic FeSe, Se(IV)-treated nano Fe0, and Se(VI)-treated nano Fe0. Points are experimental data and lines are the fits using SIXPACK software (see numerical results in Table 1). The EXAFS data are uncorrected for phase-shift

The FeSe compound EXAFS data (Fig. 8) was fit with N = 4.00 Fe atoms at an Se−Fe interatomic distance of R = 2.390 (±0.004) Å and a second shell of N = 8 Se atoms at R = 3.98 (±0.04) Å (Table 1). This Se–Fe shell was similar to previous EXAFS work on FeSe (Ryser et al. 2005; Scheinost and Charlet 2008) and the Se product of Se(IV)-treated FeS (Breynaert et al. 2008). The crystal structure of α-FeSe is composed of a stack of edge-sharing FeSe4-tetrahedral layers with a tetragonal unit cell with cell dimensions of a = 3.77 Å and c = 5.52 Å (Hsu et al. 2008). Though the synthetic FeSe produced and analyzed by EXAFS in this study was hydrous and poorly ordered, the FeSe4-tetrahedral arrangement provided a reasonable model for the atomic environment around Se.

The experimental Se(VI)- and Se(IV)-treated nano Fe0 EXAFS data were fit with two backscatter shells and compared with Se(0), Na2Se, and synthetic FeSe model compound data. The Se(VI) and Se(IV) treatments yielded nearly identical EXAFS data and had structural similarities with FeSe (Table 1). The Se(IV)-treated nano Fe0 sample exhibited a slightly higher intensity of the second Se–Se shell at 4.03 (±0.04) Å than the Se(VI)-treated sample. The first Se–Fe shells at 2.399 (±0.004) Å (Se(IV)) and 2.402 (±0.004) Å (Se(VI)) were in good agreement with synthetic FeSe (R = 2.390 Å) but were in poor agreement with Se(0) (R = 2.356 Å) and Na2Se (R = 2.351 Å). In addition, the intensity of the first Se–Se shell of the Se(0) sample was considerably lower than Se-treated nano Fe0 samples.

The Se(VI)-treated nano Fe0 product most closely matched the structural environment (i.e., interatomic distances and coordination numbers) of synthetic FeSe. This finding agrees with previous work on Se(IV)-treated Fe/Fe3C nanoparticles where FeSe was the proposed product (López de Arroyabe Loyo et al. 2008). The low intensity Se–Se shell at R = 4.02 (±0.04) Å was consistent with the weak second shell EXAFS data of FeSe at 3.98 (±0.04) Å. Previous EXAFS investigations of tetragonal FeSe (Scheinost and Charlet 2008; Breynaert et al. 2008) proposed Se–Se shells at R = 3.67–3.71 and 3.99 Å to describe the second and third backscatterer shell environment. Fitting an intermediate Se–Se shell at R = 3.67–3.71 Å to our EXAFS data did not produce a significant improvement in the overall fit in either k or R space. It is noteworthy that the intensity of the Se–Fe first shell peak for the Se(VI)- and Se(IV)-treated nano Fe0 is greater than FeSe but is similar to the Se–Se intensity of Na2Se (Fig. 8). This is consistent with the XANES analysis that suggested that the electronic environment of Se in Se-treated nano Fe0 was similar to that of Na2Se. Though the EXAFS data from the Na2Se model compound did not compare favorably with the interatomic distances of the Se(IV)- and Se(VI)-treated nano Fe0 samples, it is possible that ultra-small, nanoscale clusters of FeSe cause an increase in contributions from Se–Se bonding in the first shell which increases the backscattering intensity of EXAFS.

Conclusions

When comparing XANES, EXAFS, and XRD data we conclude that the Se-containing product of the Se(VI)-treated nano Fe0 is short-range ordered FeSe. No crystalline FeSe or FeSe2 phases were detected by XRD. We found no evidence for Se(IV) formation in the final Se(VI) reduction product. The oxidation state of Se is clearly Se(-II), though it is possible that poorly ordered domains of Se(0) could be present as part of the Fe corrosion layer, especially when environmental conditions are more oxidizing or if partial reduction of Se(VI) occurs.

The Se-containing product is part of a mixture of Fe(II)/(III) and Fe(III) oxides/hydroxides present in the corrosion product of oxidized nano Fe0. This chemical/structural environment is undoubtedly unique and proved challenging to reproduce from comparison with model compound XANES spectra. The XANES results suggested the Se product was nearly identical with Na2Se whereas the EXAFS spectra contained features similar to both FeSe and Na2Se.

The Se(-II)-containing product is highly insoluble and becomes occluded with multiple layers of Fe(II)/(III) and Fe(III) oxides/hydroxides. Reoxidation of Se(-II) will first yield Se(0), which is insoluble and thus immobile, followed by further oxidation to Se(IV). The Se(IV) anion is strongly bound to Fe(III) oxides/hydroxides by inner-sphere adsorption and thus will likely not leach out of the Fe corrosion product layers. Though this Se bonding environment in oxidized nano Fe0 is favorable for immobilizing Se, the long-term stability of the product will require additional study and is an ongoing project in our laboratory.

Acknowledgments

We thank Matt Newville (APS-GSECARS) for assistance with XAS data collection and Sam Webb (SSRL) for assistance with SIXpack software work. The XAS work was performed at GSECARS (Sector 13, APS), Argonne National Laboratory, which is supported by the NSF (EAR-0217473), the U.S. DOE (DE-FG02-94ER14466), and the State of Illinois. This work was supported by the Research Corporation Cottrell College Science Program (CC6462 and CC5444) and the NSF-MRI program (0421285).

Copyright information

© Springer Science+Business Media B.V. 2009