Article pubs.acs.org/IECR
Characterization of Arsenic Contamination on Rust from Ton Containers Gary S. Groenewold,*,† Recep Avci,‡ Robert V. Fox,† Muhammedin Deliorman,‡ Zhiyong Suo,‡ and Laura Kellerman‡ †
Idaho National Laboratory, 2351 North Boulevard, Idaho Falls, Idaho 83415, United States Image and Chemical Analysis Laboratory, Montana State University, Bozeman, Montana 59717, United States
‡
S Supporting Information *
ABSTRACT: The speciation and spatial distribution of arsenic on rusted steel surfaces affect both measurement and removal approaches. The chemistry of arsenic residing in the rust of ton containers that held the chemical warfare agents bis(2chloroethyl)sulfide (sulfur mustard) and 2-chlorovinyldichloroarsine (Lewisite) is of particular interest, because while the agents have been decontaminated, residual arsenic could pose a health or environmental risk. The chemistry and distribution of arsenic in rust samples were probed using imaging secondary ion mass spectrometry (SIMS), X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy, and scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/EDX). Arsenic in the III and or V oxidation state is homogeneously distributed at the very topmost layer of the rust samples and is intimately associated with iron. Sputter depth profiling followed by SIMS and XPS shows As at a depth of several nanometers, in some cases in a reduced form. The SEM/EDX experiments show that As is present at a depth of several micrometers but is inhomogeneously distributed; most locations contained oxidized As at concentrations of a few percent; however, several locations showed very high concentrations of As in the zerovalent form. These results indicate that the rust material must be removed if the steel containers are to be cleared of arsenic.
■
INTRODUCTION
The As(III) oxides will hydrolyze to form polymeric species. When this array of possible species interacts with environmental surfaces having heterogeneous surface chemistry, the surface speciation can be very diverse. Arsenic speciation on surfaces can include inner-sphere and outer-sphere complexes and can include many different forms.5 In particular, iron oxides, including hydrous ferric oxide, ferrihydrite, goethite, and hematite, will form complexes with arsenate and arsenite and have been used for removal of arsenic from water.6−10 Hydrous ferric oxide has been used to capture arsenic by adsorption, in which arsenate replaces hydroxide in forming inner-sphere coordination complexes.7,11 Zerovalent iron has been used to remove arsenic from water and for stabilizing arsenic in contaminated soil,12 although mechanistic investigations indicate that iron oxide on the surface of Fe(0) is responsible for adsorption.13 However, Fe(0) itself will reduce As(V) to As(III) and is probably capable of reducing As to As(0).13 Prior studies point to the fact that As can be better manipulated if speciation is understood, and so, a variety of approaches have been employed to characterize As speciation, including sequential aqueous extraction, X-ray fluorescence, Xray photoelectron, and X-ray absorption spectroscopies and vibrational spectroscopies.5 In general, these studies have shown that arsenate tends to be bidentate, inner-sphere coordinated to the iron oxide minerals,14−19 which is consistent
Arsenic is a contaminant of concern in areas where there are residual materials from chemical weapons activities. These activities principally occurred in the last century, but there is ongoing concern related to contamination in industrial settings where chemical warfare material demilitarization is occurring. For example, there is significant arsenic contamination (existing as both organic and inorganic species) left over from chemical warfare research conducted in Germany,1 the former Soviet Union,2 and in the United States.3 Incineration plants have been used to destroy legacy chemical weapons, including arsenical agents; however, removal of arsenic entails conversion to the oxide species, which can then be precipitated from aqueous solution as ferric arsenate. In the United States, bis(2chloroethyl)sulfide (mustard) was stored in ton containers, and some batches also contained 2-chlorovinyldichloroarsine (also known as Lewisite).3 Industrial equipment such as the ton containers and detonation chambers that have been in contact with the agents can develop iron oxide deposits that hold arsenic, which can display diverse speciation. The variable As speciation complicates removal strategies, because it affects how strongly the metalloid is bound to the matrix. As oxides in the V oxidation state such as As2O5, As4O10, and H3AsO4 tend to be very soluble in water, but removal using a water wash is not usually effective because arsenic acid is moderately strong (pK1 = 2.2) and its conjugate base (arsenate) forms insoluble precipitates with transition metal oxides.4 The As(V) species are also strong oxidants, forming reduced As(III) species in oxidation reactions. H3AsO3 (arsenous acid) is very weak (pKa = 9.2), but it still displays good solubility and also greater toxicity compared to arsenate. © 2013 American Chemical Society
Received: Revised: Accepted: Published: 1396
July 20, 2012 January 2, 2013 January 2, 2013 January 2, 2013 dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
used was substantially less than 1 mg and visually consisted of several grains and/or clumps that could be held between the tines of sharp forceps. Once in place, the particle sample was pressed into the foil22 using a glass microscope slide. This spread the particle aggregates over an area that was a small fraction of a millimeter across. The indium-mounted samples were then placed on the sample holder. Typically, a 140 μm × 140 μm area was scatter rastered using the primary ion beam in a pulsed fashion. The temporal width of a single primary ion pulse was 14 ns, and the repetition rate was 10 kHz. Every fifth pulse was used to fire a pulsed electron gun, which compensated for electrical charging of the sample. The primary ion source was a microfocused Ga+ gun that operated at 1.3 nA DC, at +15 kV relative to ground. The target stage was biased nominally at +3.0 keV for analysis, optimized (± ∼100 V) for emission of secondary ions. The nominal primary ion impact energy was 12 keV. The flux density was calculated at 2.2 × 1010 ions/(s cm2). Particles were typically analyzed for 3 min with a total dose imparted to the samples of 3.9 × 1012 ions/ cm2. X-ray Photoelectron Spectroscopy (XPS). The analysis was conducted on a Physical Electronics 5600ci XPS system equipped with monochromatized Al Kα X-rays. The analysis area of the sample was ∼0.8 mm in diameter. Electron emissions were collected at 45° to the normal of the surface, and the spherical-sector-analyzer pass energy was selected as 23.5 eV for high-resolution scanning and as 46.95 eV for a survey to achieve optimum energy resolution and count rate. The data acquisition and data analysis were performed using RBD AugerScan software. Scanning Auger Electron Spectroscopy (AES) and Imaging. Auger electron spectra and the AES elemental maps of selected species were obtained using a Physical Electronics 660 scanning Auger microprobe (PHI 660) system (Physical Electronics Inc., Eden Prairie, MN). A single-pass cylindrical mirror analyzer (CMA) working at ∼1% of its pass energy was used. AES spectra consisting of E*N(E) versus E were obtained in the energy region of interest where N(E) represents the number of electrons collected by the CMA at kinetic energy E, where all the electrons confined within the cones defined by 42 ± 6° are collected as a function of E. Typical target currents were about 50 nA for a primary beam energy of 10 keV focused to a diameter of about 0.4 μm. AES spectra were collected from selected points of interest, while elemental AES maps were collected from areas typically varying in size from of 100 × 100 μm2 to 200 × 200 μm2 areas at 128 × 128 pixel resolution. The Auger electrons originate from the top several nanometers of the sample. Scanning Electron Microscopy/Field Emission Energy Dispersive X-ray Spectroscopy (SEM/EDX). Micrographs were taken using a field emission scanning electron microscope (Supra 55VP, Zeiss, Thornwood, NY) in the Image and Chemical Analysis Laboratory at Montana State University. Samples were mounted to a sample holder using carbon tape and imaged at 1 kV. The elemental data was collected by energy dispersive X-ray spectroscopy in spot mode using a Si(Li) detector (Princeton Gamma-Tech Instruments, Inc., Princeton, NJ) at 20 kV. Analysis of the elemental data was performed with Spirit software (Princeton Gamma-Tech Instruments, Inc., Princeton, NJ).
with the strong binding seen in precipitation and sorption studies. However, in some instances, more soluble As species can be formed, particularly at high surface concentrations. In the present study, there is a similar desire to understand As speciation as it exists on rust flake material that formed inside ton containers that were used to hold sulfur mustard and Lewisite. These containers have been emptied and decontaminated by heating to 1000 °F for greater than 15 min, which is a process that destroys any residual agent but of course cannot destroy arsenic. Reuse or recycling of the equipment may require removing arsenic, and a target level for arsenic surface contamination of 1 μg/cm2 was stipulated by Army Regulation 385-10 in August 2012. If removal is undertaken, a detailed knowledge of the speciation and spatial distribution of arsenic will be highly beneficial for guiding technology selection and application. In this report, arsenic has been studied in the rust formed on the surface using a variety of surface techniques, which showed As present at a variety of depths and oxidation states.
■
EXPERIMENTAL SECTION Sample Origin. Samples were harvested from eductor tubes that were part of ton containers used to store Lewisite and/or mustard. The ton containers were decontaminated by heating them to 1000 °F for greater than 15 min in an autoclave. The surface of the eductor tubes consisted of a coating of rust, which could be readily flaked off, providing samples for the present study (Figure 1). The outermost layer of the samples
Figure 1. Photographs of rust flake samples from eductor tubes.
tended to have a bright orange coloration, consistent with Fe(III) oxide. Inner layers tended to have a more metallic gray color, suggesting the presence of Fe in more reduced oxidation states. In addition to coloration differences, there were also geometric differences, namely, samples were curved and contained both convex and concave sides. The different geometries do not necessarily correlate to inner and outer rust layers, since the curved rust flake originated from both the outside and inside of the eductor tube. In addition to the orange and metallic gray flakes, black particulate rust samples were also collected and analyzed. The black particulate tended to originate from between the orange flake and gray metallic flake. Imaging Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS). Analyses were performed using a triple focusing time-of-flight secondary ion mass spectrometer (TRIFT ToF-SIMS, Physical Electronics, Eden Prairie, MN),20,21 located in the Image and Chemical Analysis Laboratory at Montana State University. Particulate samples were placed on squares of indium foil (∼1 cm2). The amount 1397
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
■
Article
RESULTS AND DISCUSSION Imaging ToF-SIMS. SIMS provides information from the top nanometer or less of the sample and hence provides a view of the composition and chemistry within the first couple of atomic/molecular layers of the rust flake. Analyses were conducted using a ToF-SIMS instrument, operated in both the positive (cation) and negative (anion) modes. The two modes are complementary in that metal species and organics tend to be readily observed in the cation mode, while electronegative elements and oxo species tend to be seen in the anion mode. In addition, variations in elements and molecular species as a function of sample depth can be measured. This experiment is conducted by alternately acquiring a ToF-SIMS analysis and then removing a fraction of a nanometer of the surface by sputtering it using a high intensity Ga+ beam. A ToF-SIMS analysis is subsequently acquired. SIMS, Cation Analyses. The analyses of orange flake samples were dominated by [Fe]+ and its corresponding isotopes (Figure 2). Other notable inorganic ions are [FeOH]+
where n = 0, 1, and 2, respectively (Table S1, Supporting Information). The lower abundance m/z 235 is assigned [AsFe2O3]+. The fact that As-containing ions are seen in both the presputtered and postsputtered analyses suggests that the As contamination exists to some depth within the sample, a conclusion consistent with the negative ion SIMS and XPS analyses. The presence of molecular ions containing both As and Fe indicates that the two elements are proximally located on the sample surface and suggests the presence of covalently bond moieties such as As−O−Fe on the surface, consistent with the inner-sphere complexes identified in X-ray absorption fine structure spectroscopy studies of arsenate bound to iron oxide surfaces.5,14−19 The heterometallic ions are accompanied by several ions containing two Fe atoms, namely, [Fe2]+, [Fe2O]+, [Fe2O2H]+, [Fe2O2H]+ (m/z 112, 128, 144, and 145, respectively), and [Fe3O3]+ at m/z 216. Another interesting ion at m/z 141 is probably [FeGaO]+, which grows in relative abundance with increasing primary ion dose (consistent with the behavior of [Ga]+ at m/z 69). The appearance of Fe−O containing molecular ions is consistent with the dominant composition of the rust samples. A small group of ions at m/z 206−208 is indicative of Pb contamination on the surface. In Figure 2, these ions are very small; however, their relative abundances are close to those expected for the [Pb]+ isotopic pattern, and the ions are significantly more abundant in the spectra acquired for the gray metallic flake and the black particulate (Figures S1 and S2, Supporting Information); in these experiments, the Pb isotopic pattern is unmistakable. Another cluster of ions in the spectrum of the orange flake from m/z 198−204 corresponds to Hg. This was not seen in the analyses of the gray metallic flake or the black particulate but was observed strongly in the XPS analyses and is a significant contaminant as shown by the inductively coupled plasma−mass spectrometry analyses.23 Normally, Hg responds strongly in a SIMS analyses, and so, the lack of strong response here suggests that the Hg exists mainly at some depth but is not a significant surface contaminant. The origin of the Pb is not known; however, Hg is known to have been present as a contaminant in some lots of sulfur mustard.3 The surfaces of the rust samples were relatively homogeneous from a microscopic spatial perspective. The ToF-SIMS instrument generates chemical maps of the analyzed regions, which showed spatial variability that was attributable to uneven surface morphology, which results in variations in secondary ion yield. There was no variation in ion ratios, which can be seen by comparing the total ion image to the [Fe]+ image before sputtering (Figure S3, images a1 and a2, Supporting Information). Areas of high total ion intensity correlated with high [Fe]+ intensity. The [AsO]+ image (Figure S1, a3, Supporting Information) was much lower in intensity (consistent with low intensity in the spectrum) but showed that the arsenic was distributed across the rust particle in much the same manner as was iron. The situation did not notably change after sputtering; homogeneous distribution of iron and arsenic was observed across the sample, indicating that, within the top nanometer, there is no dramatic localization of the As. Analyses of the two other types of samples, that is, the gray metallic flake and the black particulate, produced positive SIMS spectra that were very similar to that of the orange flake, in terms of species produced and fractional abundances. This indicates that the chemical composition of the topmost layer of the surfaces of the three types of rust samples are relatively
Figure 2. Cation SIMS spectra of orange flake: (a) unmodified sample, prior to sputtering; (b) analysis subsequent to 60 s sputtering, which removed ∼0.3 nm from the surface.
and [Ga]+ at m/z 73 and 69; the latter originates from the primary ion beam and is implanted during the analysis and during surface sputtering. Other lower abundance ions are principally derived from organic surface contamination. All surfaces contain adsorbed organic molecules that originate from the environment to which the sample has been exposed, and these show up in the SIMS analyses, which are highly sensitive for their detection. The organics in this analysis do not contain information pertinent to the As contamination. Sputtering the sample for 60 s results in removal of approximately the top 0.3 nm of material from the surface. This estimate is based on a prior calibration of a crater sputtered into a boron-doped silicon that was subsequently measured using atomic force microscopy. Removal of 0.3 nm eliminates a large fraction of the adsorbed organics and significantly improves observation of ions derived from the inorganic material. A very notable ion at m/z 91 corresponds to [AsO]+ and is seen in both the presputtered and sputtered analyses. Close examination of the slightly higher m/z range reveals strong evidence for close association between As and Fe: ions at m/z 131, 147, and 163 correspond to [AsFeOn]+, 1398
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
oxidation states.24 Ion beam induced reduction of arsenate likely accounts for the majority of the [AsO2]− intensity, as indicated by ion beam sputtering experiments and analyses of spiked samples described below. Spectra acquired after the surface of the sample was sputtered are depleted in organic species; however, [AsO2]− and the halide-containing species remain relatively constant or are even enhanced in their intensity. This indicates that arsenic is distributed throughout the near-surface bulk, encompassing perhaps the first several nanometers of the rust. Sputtering practically eliminates the [H]− and [OH]− peaks and the zinc and nitrate surface species, which shows that these are surface species. Sputtering also resulted in a large decrease in the intensity of [AsO3]− in the analyses of the orange flake rust, which probably indicates that As is reduced by the sputtering process. The decrease is quantified in terms of the As-fractional abundance of [AsO3]−, which decreased from 0.31 to 0.12 in this sample (Figure 4), and similar decreases were seen in the sputtered
similar, which is consistent with the fact that the samples had been in contact with similar atmospheric environments. The spectra of the metallic gray flake and black particulate (Figures S1 and S2, respectively, Supporting Information) both had higher intensities for [Pb]+ and [Na]+ compared to the orange flake and did not contain notable [Hg]+. This suggests that Pb and Na are located within in the rust coating, nearer to the base metal, while Hg is located near the surface. One rust sample was equilibrated with a 1 mM solution of sodium arsenate, dried, and then, analyzed (Figure S4, Supporting Information). The resulting positive ion mass spectrum showed the same ions as did the spectra of the three different rust samples, with the salient difference being the intensities of the heteronuclear As−Fe containing ions [AsFe]+, [AsFeO]+, [AsFeO2]+, and [AsFe2O3]+ were all more abundant in the arsenate-equilibrated sample. SIMS, Anion Analyses. Anion analysis (before sputtering, Figure 3a) of the oxidized, orange colored rust flake revealed
Figure 3. Anion SIMS spectra of orange rust fleck: (top) before sputtering; (bottom) after sputtering. −
Figure 4. Plot of As fractional abundances for the three rust samples and the As-sorbed rust.
−
significant [O] and [OH] , typical of an oxide surface, and abundant [Cl]− and [F]− as well. An abundant [H]− was also observed and can arise from surface organics or hydroxyl moieties. These peaks account for better than 90% of the ions sputtered from the surface. Chlorine also shows up as m/z 51 and 53, which are attributed to [ClO]−, and m/z 70, 72 and 74, which is [Cl2]−. [ClO]− suggests that the rust material may have been treated with bleach. A significant [NO3]− peak is observed at m/z 62, and a cluster of isotopic ions from m/z 169 to 175 is [ZnCl3]−. Zn is a component of the metal and upon oxidation forms Zn2+ leading to the chloride cluster (Figure 3a). A lower abundant isotopic cluster is seen from m/z 161 to 165 and corresponds to [FeCl3]−. Of highest interest are the arsenic-containing ions at m/z 107 and 123 that correspond to [AsO2]− (As(III)) and [AsO3]− (As(V)), respectively. The appearance of the ion [AsO3]− indicates As(V) on the surface of the rust particles. On the surface, the predominant As(V) species will be present as the AsO43− anion; however, direct evidence for this is not contained in the SIMS spectrum. Instead [AsO3]− is likely formed by decomposition of AsO43−-containing species. [AsO2]− suggests the presence of As(III) but does not prove it because of the reductive nature of the primary ion bombardment process, which converts oxy species to lower
analyses of the gray flake and black particulate rust. Sputterinduced As reduction is supported by analysis of a control sample that was a rust flake that was equilibrated with 1 mM arsenate solution for 2 h, decanted, dried, and analyzed. The objective of this experiment was to overwhelm the As signal with ions derived from As(V) on the surface and to see whether this biased the intensities of the oxyanions toward a more intense As(V) species [AsO3]−. The anion spectrum contained [AsO3]− and [AsO2]− that were almost 10 times more abundant than the corresponding peaks in the spectra of the other rust samples (compare Figure S5 with S6 and S7, Supporting Information and Figure 3). However, the Asfractional abundance of [AsO3]− was ∼0.26 (Figure 4), which is a value not appreciably different from that of the three rust samples. This suggests that As-fractional abundance for [AsO3]− ranging from 0.23 to 0.31 represents As in the V oxidation state on the topmost layer of the rust samples. When the As-equilibrated sample was sputtered, the As-fractional abundance of [AsO3]− dropped from 0.26 to 0.09, indicating that sputtering is reducing As on the surface that probably started out with the majority of the As present in the V oxidation state. 1399
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
Anion analyses of the gray metallic flake and the black particulate rust samples produced spectra (Figures S6 and S7, Supporting Information) that were similar to those of the orange flake samples in almost all respects. One difference was that both of these samples had a more intense [F]− signal, indicating more fluoride compared to the orange flake. The origin of the fluoride is not known. The anion SIMS images showed some apparently inhomogeneous ion distributions across the surfaces, for example, the total ion image shows good signal intensity across the area sampled, with the exception of a dark “canyon” running across the middle of the image (Figure S8, Supporting Information). However, when one compares the relative ion intensities from high intensity and low intensity regions, there is very little variation in the mass spectra generated. Thus the “canyon” and other features are due to irregular surface morphology, the consequence of which is that some surface locations are out of focus relative to the ion microscope that is the secondary ion source of the ToF-SIMS. The most abundant secondary anion is [Cl]−, and it furnishes an image that is qualitatively identical to the total ion image. Images for [F]− and [AsO2]− are much less intense but again display the same areal features as did the total ion image. XPS. XPS measures electrons emitted from the first nanometer of the sample,5 which is a depth regime similar to that of the SIMS analyses. XPS can generate a wide variety of peaks that can overlap and confuse assignment, and so, a gray rust flake sample was equilibrated with 1 mM arsenate, in order to produce spectra in which the As XPS signature was unequivocal (Figure 5). The most intense As XPS peak is the
from As(0), which has a 3d energy around 41 eV. In addition to the As signature, significant peaks corresponding to O 1s, Fe 2p3, Fe 2p1, C 1s, and O 1s were also observed. The sample was sputtered for 4 min using a 2.5 kV Ar+ beam, which removed approximately 4 nm from the top of the sample. Reanalysis showed that the carbon atom percentage decreased by more than 10%, while both Fe and oxygen increased. The arsenic atom percentage remained about the same. These observations indicate that the carbon in this sample is located principally on the surface, while the arsenic concentration is fairly constant down to at least 4 nm. The rust was dislodged from the surface of the eductor tube upon sampling, and three visually distinct types of materials were collected that originated from different spatial realms relative to the base metal. Material on the surface and furthest from the base metal had an orange appearance indicative of highly oxidized iron. Analyses of these orange flake samples revealed a spectrum (Figure S9, Supporting Information) similar to that of the As-dosed gray flake described above. In the orange flake, the As atom percentage was lower compared to the black particulate or gray flake described later on, but values were still in the percent range (Table S3, Supporting Information). Sputtering this sample removed the As, which indicated that it was localized in the first 4 nm (the depth that is reached during a 4 min sputtering event). Keeping in mind that the orange flake is the outermost material in the overall rust flake structure, this also means that the majority of the As is localized at some depth and is not present on the surface. This conclusion is supported by the XPS analyses of the black particulate, a material that is localized between the outer orange flake, and the inner gray metallic particulate. The full spectrum contains the same major peaks corresponding to As, Fe, O, C, and Cl (Figure S10, Supporting Information) and indicates a significantly higher carbon concentration on the surface (Table S3, Supporting Information). As concentration (3.4%) is also higher than on the orange flake. Both C and As decrease upon sputtering to 4 nm but are nowhere near depleted. A closer examination of the lower energy range (Figure S11, Supporting Information) shows a couple of low-intensity peaks that indicate the presence of Hg but appears not to be a major contaminant in the surface analyses. Sputtering largely removes Hg, indicating that it is localized on the surface of the black particulate. The As 3d peak is located near 44 eV in both the pre- and postsputter analyses, again indicating As in the III and or V oxidation states and also indicating that the 2.5 kV argon sputter beam is not causing reduction of As in these samples (see below). The gray metallic flake was that portion of the rust structure that was closest to the basis metal. XPS analysis indicated 3.8% As and 25.7% carbon, with the remainder of the sample consisting of Fe and O (Figure S12, Supporting Information). Subsequent to sputtering, the carbon decreased to around 10%, but interestingly, the As percentage increased, nearly doubling, which indicates that the As is not predominantly on the surface but resides in a near-surface realm greater than 4 nm deep. A plot of the low electronvolt portion of the spectrum (Figure S13, Supporting Information) with an expanded y axis shows that some of the As is probably not in the III or V oxidation states but instead is elemental. This conclusion is derived from the observation of a small peak at 40.4 eV, which is interpreted in terms of the As 3d signal shifted to lower energy as a consequence of a fraction of the As existing in the zerovalent state.13,25 The As 2p3 line also displays a low-energy peak at
Figure 5. XPS spectrum of gray rust flake, equilibrated with 1 mM AsO43−: (a) prior to sputtering; (b) after a 4 min sputter (removal of top 4 nm).
As 2p3 line at 1326.8 eV (Table S2, Supporting Information), which was used to quantitate the As atom percentage at 13.8% (Table S3, Supporting Information). The assignment was confirmed by the presence of the As 2p1 peak at 1362 eV, As 2p3 and As 2p1 lines at 143 and 148 eV, and the As 3d line at 44 eV. The As 3d energy is indicative of oxidation state,5 with a value of 44 eV being indicative of As in the III state (see below).13,25 This value can be differentiated from the As(V) value at 45 eV, although differentiation can be challenging as a result of the line width and the presence of multiple species on the surface. The oxidized As species are easily distinguished 1400
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
particularly the conclusion that peaks emerging at slightly lower energy after sputtering are derived from As(0). Since this phenomenon was not observed after sputtering other high-As samples, it is not likely that As(0) is produced by reduction induced by the Ar+ beam. Interestingly, Su and Puls also concluded that As migrated to the interior of zerovalent iron, which was actually coated with iron oxide.27 As will be discussed later, the existence of pockets of highly reduced As is consistent with the EDX analyses. The spectrum in Figure 6 also shows a remarkable quantity of Hg (2.1% as indicated by the intensity of the Hg 4f7 peak at 101 eV). The fact that it is very nearly completely sputtered away indicates that it is localized in the top 10 nm. The data suggest that there may be a correlation between high Ascontaining regions and Hg, although more measurements are needed to evaluate the potential relationship. Scanning AES and Imaging. An imaging AES analysis was conducted for one of the gray flake surfaces in order to provide a cross check of the observations and conclusions derived from the XPS and ToF-SIMS analyses. The broad area Auger spectrum was consistent with an iron oxide sample contaminated with a significant fraction of arsenic (Figure 7).
1323.2 eV (reduced As) in addition to the main peak at 1326 eV (oxidized As). Multiple rust samples were analyzed, with similar results. In several of the metallic gray samples, As atom percentages exceeded 10%, although these values varied significantly, which reflects the highly heterogeneous nature of the eductor tube rust. The variable As percentages are consistent with the EDX results, which showed regions of high As concentration, where the As was in a reduced form. One metallic gray flake did reveal strong evidence for reduced arsenic, in the form of a shift in the XPS peaks to lower energy once the topmost layers were removed by sputtering. An initial XPS analysis prior to sputtering showed significant As (7.5%) as measured using the As 2p3 line at 1326 eV (Figure 6, dashed trace). Peaks at
Figure 6. X-ray photoelectron spectrum of gray metallic flake, both before and after sputtering for 10 min (depth equivalent 10 nm): (a) high-energy regime, As 2p lines; (b) low-energy region, As Auger, 3p and 3d lines.
Figure 7. Auger spectrum of rust flake, concave side.
268, 232, 208, 149, 145, and 45 eV correspond to As Auger and XPS lines as indicated, and in addition, the spectrum showed notable C 1s and Hg 4f peaks. The sample was then sputtered using the Ar+ 2.5 kV ion beam, and the XPS spectrum was remeasured. For each of the As XPS and Auger peaks, a second peak emerged at slightly lower energy, which likely indicates the presence of a significant fraction of zerovalent As at depth. Literature values for the As 3p line indicate that As(III) and As(V) are found at 44 and 45 eV, in good agreement with the presputter measurement here (although the peak width is too wide to permit distinction between these two oxidation states). The position of the As 3p peak emanating from As(0) has been measured at 41 eV, which is in excellent agreement with the measurement made after the sputtering event. The interpretations are based on prior research; arsenite and arsenate removal with zerovalent iron was demonstrated by Bang and coworkers, and the oxidation state of sorbed As was evaluated using XPS.13 The spectrum of a coupon treated with As(V) contained an As 3d peak at 45.3 eV,13 consistent with the value of 45.5 eV reported by Wagner et al.26 The spectrum acquired for the coupon treated with As(III) showed two peaks at 44.9 and 41.4 eV,13 consistent with the values of 44.2 and 41.5 eV reported by Wagner et al.26 Adsorption of arsenate and arsenite on zerovalent iron was also examined by Su and Puls, who measured As(III) at 44.1 eV and As(V) at 45.2 eV.27 These prior studies support the current oxidation state assignments,
The small peak at 1218 eV was correlated with the As 1 line at 1228 eV, and other salient peaks in the spectrum correspond to Fe and O, with minor peaks attributable to C and Cl. When spectra from specific regions were compared, the As1 line was at times absent, which is consistent with the inhomogeneity of the sample surface. The chemical surface inhomogeneity was highlighted by scanning Auger images. In the example in Figure 8, As is present across most of a 100 × 100 μm2 area, with the exception of a roughly circular region seen on the lower righthand side of image d. This feature is also low in Fe, but high in oxygen and chlorine, suggesting that As is not as strongly correlated with highly oxidized regions of the surface. This conclusion is consistent with ToF-SIMS and XPS measurements of the highly oxidized orange flake samples, which showed lower As signals compared to other more reduced rust samples. SEM/EDX. Scanning electron micrographs of the convex surface of a rust flake sample showed irregular morphology characterized by cracks in the material (Figure 9). In this particular sample, the convex surface was proximate to the bulk metal, having originated from the interior of the eductor tube. EDX was used to measure the elemental composition of the samples. X-rays used for identification and quantification of the elements are generated from depths of 1 μm or more, and so, the EDX measurements provide an analysis of this depth 1401
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
Figure 8. Scanning Auger images of rust flake, convex side. Brighter regions indicate higher concentrations. Images b−f are at the same magnification as image a but are plotted at half size.
Figure 10. Atom percentages plotted versus oxygen percentage, convex side of rust flake (proximate to metal bulk). Values measured using EDX.
and the fact that these are located close to the values measured for the oxidized points indicates that the majority of the sample is in fact oxidized. However, spectra were acquired from several points for which oxygen was very depleted, indicating the presence of metallic iron and arsenic, consistent with prior reports of arsenic reduction on zerovalent iron.13 These reduced pockets do not reflect the majority of the sample; however, it is conceivable that they could represent 5−15% of the total material. The micrographs of the concave sample were also complex, with multiple geometric forms present on the surface (Figure S14, Supporting Information). However EDX analyses of various points did not support significant chemical variability on this side of the rust flake. Fe accounted for ∼18−24 atom %, with As ranging from ∼1−5% (Figure 11). The broad area values were slightly less oxidized compared to the individual points measured, suggesting that there are in fact phases in the Figure 9. Scanning electron micrograph of a rust flake, convex side. Elemental atom percentages measured via EDX. (top) Magnification 400×; (bottom) magnification 3000×.
regime, as opposed to surface analyses that are provided by the SIMS, XPS, and AES experiments described above. The broad area composition shown in Figure 9 can be distilled to an elemental composition of approximately As2Fe2O4, which cannot be correlated with oxide forms possessing Fe and or As in their normal oxidation states and hence instead suggests a reduced environment. The different geometric forms shown in the micrographs could suggest varying composition, and this expectation was supported by localized analyses performed by focusing the electron beam on a very small region and measuring the resultant X-rays. These experiments showed wide variability in elemental percentages (Figure 9), indicating the presence of local phases ranging from very oxidized to very reduced. The spot analyses can be grouped on the basis of atomic composition (Figure 10), which shows that the majority of the sample bulk is oxidized as expected but that there are notable phases that are significantly reduced. The percentages for the broad area analyses represent an average measurement,
Figure 11. Atom percentages plotted versus oxygen percentage, concave side of rust flake (surface of tube material). Values measured using EDX. 1402
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
Article
Both XPS and scanning Auger spectroscopy provide data from the first several nanometers and hence probe a depth regime slightly deeper than that sampled by the ToF-SIMS. The depth regimes probed by ToF-SIMS and XPS were extended by removing surface layers with sputtering and periodically alternating analyses. As was found in nearly every sample examined. The precise energy of the As 3d line indicated either the III or V oxidation state; however, these could not be distinguished. When some of the gray metallic flake samples were sputtered with Ar+, the As XPS lines were shifted to slightly lower energy, signaling the presence of As(0) at a few nanometers below the surface. In most cases, this did not occur, which showed that Ar+ sputtering did not induce reduction. Sputtering to 3 nm did not eliminate the As lines. A complementing scanning Auger analysis added to the consistency of this conclusion, showing that As was not colocated with regions high in oxygen but instead localized with regions of high Fe. The near-surface bulk was analyzed using EDX, which measures atomic composition down to a depth of several micrometers. The EDX analyses generated atomic percentages for the orange flake samples that featured high oxygen percentages and As values that were only a few percent. In contrast, the gray flake samples showed a lower oxygen content, indicative of reduced regions. EDX scans of microscopic features revealed the presence of locales where As and Fe were much reduced, most likely in the elemental oxidation state, consisting of 70% As in some cases. Neighboring regions showed a more normal oxygen percentage. These observations were interpreted in terms of significant spatial heterogeneity and the existence of significant pockets of reduced As and Fe, extending to a depth of several micrometers. The presence of arsenic at several depth regimes existing as arsenate or arsenite species suggests that it is bound to iron oxides via inner-sphere coordination, which implies that separation of the arsenic from the rust is probably impracticable. Leaching of arsenic from the rust will probably not occur to any great extent, which means that arsenic in these mineral forms will probably not be a source of ground- or surface water contamination. Removal of arsenic contamination will require an approach that loosens the rust particulate from the metal surface of the steel. Finally, field measurement of arsenic will require an approach capable of measurement in a complex iron oxide matrix and capable of identifying arsenic at a depth of several micrometers, particularly if speciation information is needed.
broad area that are reduced. However, these can only account for a small percentage of the total material. The fact that arsenic is detected with significant abundance by EDX raises the question of how it migrated to a depth of one to several micrometers into the iron oxide. A ton container made of low carbon steel and holding Lewisite can undergo oxidation, with entrainment of arsenic decomposition products into the oxide layer. This process may be accelerated when the container is heated to greater than 1000 °F in order to achieve agent decontamination; the iron oxide becomes highly porous, which may facilitate permeation of reduced arsenic through the iron oxide. This explanation would be consistent with the spatial correlation of reduced arsenic pockets with reduced iron oxide species. These possibilities cannot be distinguished on the basis of the present experiments, but should be investigated further, since arsenic at these depths will likely be more difficult to remove.
■
CONCLUSIONS A detailed study of the chemistry of As-contaminated rust was conducted in order to have a thorough understanding of the challenges associated with mitigating hazards and to guide removal of the contamination. The samples were derived from eductor tubes that had once been part of ton containers that were used to store mustard and also Lewisite. The containers had been drained and decontaminated, which left behind significant rust material on the surface of the eductor tube. The rust was liberated from the surface of the tubes in flakes and displayed coloration that suggested that the outside of the rust (in contact with the chemical agents and decontamination solutions) was highly oxidized. Underneath that was a black amorphous layer, and finally, there was a gray metallic flake that was in closest proximity to the steel. The results of investigations using four different analytical techniques that probe two different depth regimes show that As is concentrated in the gray metallic flake that is closest to the base metal and that there are pockets where As is reduced to the elemental state. As is also present in significant quantities in the black particulate rust samples and in the orange flake. Sputtering experiments to a depth of 3 nm did not eliminate the As signatures, and the fact that strong As lines were seen in the EDX analyses indicates that As has permeated the material to a depth of several micrometers. Both the SIMS and XPS analyses indicate that As is oxidized on the surface of the rust samples. As(V) is clearly present on the topmost layer of the samples but is probably accompanied by As(III) at least to a depth of several nanometers. As(0) can also be seen at this depth in some of the gray flake samples. ToF-SIMS provides the shallowest interrogation of the surfaces of the samples, probing the top nanometer. The cation SIMS spectra displayed molecular ions containing As, Fe, and O that strongly indicate the existence of As and Fe in close proximity on the sample surface, most probably bound by a linking O atom via inner-sphere coordination.5,18 When the samples were sputtered to a depth of 0.3 nm, the As signature was largely unchanged. The anion SIMS spectra are dominated by abundant oxygen, fluoride, and chloride. The molecular ion signatures for As in the form of AsO3− and AsO2− are also observed in significant abundance, and unequivocally indicated that a significant fraction of the As on the surface exists in the V oxidation state. Upon sputtering, the ratio of AsO3− and AsO2− changed in favor of AsO2−, suggesting that As in the nearsubsurface exists more prevalently in the III oxidation state.
■
ASSOCIATED CONTENT
S Supporting Information *
Peak identifications for SIMS spectra, XPS spectra, and XPS atom percentages; additional positive and negative ion SIMS spectra and images with supporting XPS spectra and EDX images. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest. 1403
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404
Industrial & Engineering Chemistry Research
■
Article
coprecipitated and adsorbed arsenate. Geochim. Cosmochim. Acta 1993, 57, 2251−2269. (20) Schueler, B. W. Microscope imaging by time-of-flight secondary ion mass spectrometry. Microsc., Microanal., Microstruct. 1992, 3 (2− 3), 119−139. (21) Schueler, B.; Sander, P.; Reed, D. A. A time-of-flight secondary ion microscope. Vacuum 1990, 41 (7−9), 1661−1664. (22) Reich, D. F. Sample Handling. In ToF-SIMS Surface Analysis by Mass Spectrometry; Vickerman, J. C., Briggs, D., Eds.; IM Publications and SurfaceSpectra Limited: Chichester, U.K., 2001; p 123. (23) Fox, R. V.; Groenewold, G. S.; Avci, R.; Deliorman, M.; Suo, J.; Kellerman, L. Evaluation of selected inorganic CWM byproducts on ton container eductor tubes and steel coupons, and implications for laser cleaning. In INL/LTD-12-24808; Idaho National Laboratory: Idaho Falls, ID, 2012. (24) Groenewold, G. S.; Delmore, J. E.; Olson, J. E.; Appelhans, A. D.; Ingram, J. C.; Dahl, D. A. Secondary ion mass spectrometry of sodium nitrate: comparison of ReO4− and Cs+ primary ions. Int. J. Mass Spectrom. Ion Processes 1997, 163 (3), 185−195. (25) Bang, S.; Korfiatis, G. P.; Meng, X. Removal of arsenic from water by zero-valent iron. J. Hazard. Mater. 2005, 121 (1−3), 61−67. (26) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F.; Muilenberg, G. E. Handbook of X-ray Photoelectron Spectroscopy; Perkin−Elmer Corporation, Physical Electronics Division: Eden Prairie, MN, 1979. (27) Su, C.; Puls, R. W. Arsenate and arsenite removal by zerovalent iron: kinetics, redox transformation, and implications for in situ groundwater remediation. Environ. Sci. Technol. 2001, 35 (7), 1487− 1492.
ACKNOWLEDGMENTS This research was funded by the United States Army, Program Manager for Non-Stockpile Chemical Materiel.
■
REFERENCES
(1) Martens, H. The old chemical warfare (OCW) incineration plants in Munster, Germany. In Applied Science and Analysis Newsletter; Applied Science and Analysis, Inc.: Kane’ohe, HI, 1997; http://www. asanltr.com/ASANews-97/chem_demil_plant.html. (2) Ionov, L.; Zubtsovsky, N.; Makarova, L.; Reshetnikov, S. Chemical and biological-ecological aspects of risk assessment for Lewisite destruction. In Ecological Risks Associated with the Destruction of Chemical Weapons; Kolodkin, V. M., Ruck, W., Eds.; Springer: Dordrecht, the Netherlands, 2006; pp 217−221. (3) Council, N. R. Disposal of Chemical Munitions and Agents; The National Academies of Science and Engineering: Washington, DC, 1984. (4) Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements, 2nd ed.; Butterworth Heinemann: Oxford, U.K., 1997; p 1250. (5) Wang, S. L.; Mulligan, C. N. Speciation and surface structure of inorganic arsenic in solid phases: a review. Environ. Int. 2008, 34 (6), 867−879. (6) Mishra, S. P.; Das, M.; Dash, U. N. Review on adverse effects of water contaminants like arsenic, fluoride and phosphate and their remediation. J. Sci. Ind. Res. 2010, 69 (4), 249−253. (7) Mohan, D.; Pittman, C. U. Arsenic removal from water/ wastewater using adsorbentsa critical review. J. Hazard. Mater. 2007, 142 (1−2), 1−53. (8) Cheng, H. F.; Hu, Y. N.; Luo, J.; Xu, B.; Zhao, J. F. Geochemical processes controlling fate and transport of arsenic in acid mine drainage (AMD) and natural systems. J. Hazard. Mater. 2009, 165 (1− 3), 13−26. (9) Harris, G. B.; Krause, E. The disposal of arsenic from metallurgical processes: its status regarding ferric arsenate. In Extractive Metallurgy of Copper, Nickel and Cobalt. Fundamental Aspects; Reddy, R. G., Weizenbach, R. N., Eds.; The Minerals, Metals and Materials Society: Warrendale, PA, 1993; Vol. 1, pp 1221−1237. (10) Jiang, J. Q. Removing arsenic from groundwater for the developing worlda review. Water Sci. Technol. 2001, 44 (6), 89−98. (11) Höll, W. Mechanisms of arsenic removal from water. Environ. Geochem. Health 2010, 32 (4), 287−290. (12) Kumpiene, J.; Lagerkvist, A.; Maurice, C. Stabilization of As, Cr, Cu, Pb and Zn in soil using amendmentsa review. Waste Manage. 2008, 28 (1), 215−225. (13) Bang, S.; Johnson, M. D.; Korfiatis, G. P.; Meng, X. Chemical reactions between arsenic and zero-valent iron in water. Water Res. 2005, 39 (5), 763−770. (14) Farquhar, M. L.; Charnock, J. M.; Livens, F. R.; Vaughan, D. J. Mechanisms of arsenic uptake from aqueous solution by interaction with goethite, lepidocrocite, mackinawite and pyrite: an X-ray absorption spectroscopy study. Environ. Sci. Technol. 2002, 36, 1757−1762. (15) Fendorf, S.; Eick, M. J.; Grossl, P.; Sparks, D. L. Arsenate and chromate retention mechanisms of goethite 1. Surface structure. Environ. Sci. Technol. 1997, 31, 315−320. (16) Grossl, P.; Eick, M. J.; Sparks, D. L.; Goldberg, S.; Ainsworth, C. C. Arsenate and chromate retention mechanisms on goethite 2. Kinetic evaluation using a pressure-jump relaxation technique. Environ. Sci. Technol. 1997, 31, 321−326. (17) Manceau, A. The mechanism of anion adsorption on iron oxides. Geochim. Cosmochim. Acta 1995, 59, 3647−3653. (18) Waychunas, G. A.; Kim, C. S.; Banfield, J. F. Nanoparticulate iron oxide minerals in soils and sediments: unique properties and contaminant scavenging mechanisms. J. Nanopart. Res. 2005, 7 (4−5), 409−433. (19) Waychunas, G. A.; Rea, B. A.; Fuller, C. C.; Davis, J. A. Surface chemistry of ferrihydrite: Part 1. EXAFS studies of the geometry of 1404
dx.doi.org/10.1021/ie301937j | Ind. Eng. Chem. Res. 2013, 52, 1396−1404