Investigation of the Microcharacteristics of PM2.5 in Residual Oil Fly

The particle size and morphology were characterized from bright field images recorded ..... Clearing House for Inventories and Emission Factors (CHIEF...
0 downloads 0 Views 1MB Size
Environ. Sci. Technol. 2004, 38, 6553-6560

Investigation of the Microcharacteristics of PM2.5 in Residual Oil Fly Ash by Analytical Transmission Electron Microscopy YUANZHI CHEN, NARESH SHAH,* FRANK E. HUGGINS, AND GERALD P. HUFFMAN University of Kentucky, 533 South Limestone Street, Lexington, Kentucky 40508-4005

Atmospheric emissions from combustion of residual oils often consist of carbonaceous material and metal compounds, both of which are of concern for health and environmental issues. In this study, particulate matter fractions with aerodynamic diameters nominally less than 2.5 µm (PM2.5) in two residual oil fly ash (ROFA) samples generated from combustion experiments were investigated by analytical transmission electron microscopy (TEM) techniques, including energy-dispersive X-ray spectroscopy, selected area electron diffraction (SAED), high-resolution TEM, and electron energy loss spectroscopy (EELS). Carbonaceous particles, which dominate both samples, exist in two distinctive forms: as soot aggregates with spherical primary particles of size 10-80 nm that exhibit a concentric arrangement of graphitic layers around the particle center and as larger spherical or irregular-shaped porous residual char particles of size 1-20 µm that usually have anisotropic microtextures and contain organic sulfur species. Such carbon-rich particles were often observed to be coated with inorganic species, notably transition metals (V, Ni, Fe, Zn) in the form of sulfates, oxides, vanadates, and phosphates. In this respect, they therefore differ from similar carbonaceous particles generated in combustion of diesel fuels that lack significant inorganic species. Crystalline phases of vanadium, nickel, and iron oxides and multi-element oxides were identified by the SAED technique. The valence state of V in some V-rich oxide particles probed by EELS was found to vary from +2 to +5. Individual transition metal sulfate, oxide, and phosphate particles are typically compositionally complex, containing multiple metallic elements. These microcharacteristics of individual PM2.5 particles revealed by electron microscopy techniques should be important parameters to include in future toxicological investigations of ROFA PM.

Introduction Available ambient measurements of PM2.5 suggest that anthropogenic combustion sources, fires, and other emitters of condensable and secondary origins contribute a large percentage of the overall ambient PM2.5 mass in most areas (1). The U.S. EPA estimates that in the year 2000, out of 756 000 ton of primary PM2.5 emissions due to all fuel combustion * Corresponding author phone: (859)257-5119; fax: (859)257-7215; e-mail: [email protected]. 10.1021/es049872h CCC: $27.50 Published on Web 11/13/2004

 2004 American Chemical Society

in the United States, 22 000 ton was due to residual oil combustion. Additionally, 21 000 ton of primary PM2.5 was emitted by residual oil combustion in marine vessels (out of 608 000 ton for all transportation applications). The U.S. EPA estimate of all anthropogenic PM2.5 emissions in the United States is 21 926 000 ton (2). Residual oil fly ash (ROFA) is a byproduct generated from combustion of residual fuel oil. A number of metallic elements (e.g., Mg, V, Fe, Ni, Cu, Zn, and Pb), which are originally present in the crude oil or derive from contamination of product streams by refinery catalysts and equipment, constitute a significant fraction of ROFA (3-5). Toxicological studies have shown that the pulmonary injury induced by ROFA exposure is associated with its high contents of transition metals, especially those in the form of species that are readily soluble in aqueous fluids (6-8). In addition, many of the transition metal compounds in ROFA have well-known catalytic abilities to activate important chemical reactions in the atmosphere and biochemical processes within the human body. These health and environmental effects have a close relationship with the physical and chemical characteristics of individual particles, such as size, morphology, solubility, composition, and oxidation state. It is therefore essential to use appropriate analytical methods to obtain such information. Several bulk analysis techniques, such as chemical composition analysis (4), Fourier transform infrared spectroscopy (FTIR) (4), X-ray diffraction (XRD) (4, 5, 9), and X-ray absorption fine structure (XAFS) spectroscopy (3, 10, 11) have been applied to identify the chemical forms of ROFA. However, these bulk techniques are unable to give detailed information about individual particles and studies on the microcharacteristics of individual ROFA particles are limited. A recent electron microscopic study on the aerosol in a nickel refinery showed that not a single particle out of 1170 examined consisted of a well-defined stoichiometric substance (12). As a result, the authors questioned the appropriateness of regulatory decisions based on results of toxicological studies generated from exposures of animals to specific individual nickel compounds. As will be presented herein, individual ROFA particles are also frequently compositionally complex. Since toxicological behavior is believed to be directly related to individual PM particle characteristics, information on the chemical and physical complexity of ROFA PM2.5 particles should lead to improved understanding of the appropriate toxicological mechanisms. In our previous studies, we have used several analytical techniques, including XAFS, XRD, scanning electron microscopy (SEM), nuclear magnetic resonance (NMR), X-ray fluorescence (XRF), and inductively coupled plasma/mass spectrometry (ICP/MS) to characterize the molecular structure and microstructures of several ROFA samples (3, 5). Here, we present the results of an in-depth study using various analytical transmission electron microscopy (TEM) techniques, including energy dispersive X-ray spectroscopy (EDS), high-resolution TEM (HRTEM), selected area electron diffraction (SAED), and electron energy loss spectroscopy (EELS) to characterize the morphology, composition, crystalline phases, and elemental valence states of two typical ROFA PM2.5 samples that are rich in metallic compounds. The results provide detailed information about the characteristics of single ROFA PM2.5 particles and complement the previous results from bulk analysis methods. VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6553

FIGURE 1. Typical carbonaceous particles in ROFA PM2.5. (a) Soot aggregate with spherical primary particles. (b) An irregular shaped char particle. Insets are SAED patterns and 002 lattice fringe images. The 002 arcs in the SAED pattern of the char particle reflect an anisotropic microtexture.

Experimental Section Samples. The fuel oils were combusted at the U.S. EPA’s National Risk Management Research Laboratory (NRMRL) using a North American three-pass fire-tube package boiler. A detailed description of this boiler is given elsewhere (13). The combustion product gases laden with fly ash (ROFA) passed through a cyclone where the fly ash (ROFA) was separated aerodynamically into fine (PM2.5) and coarse (PM2.5+) particle-laden streams. Teflon-coated glass fiber filters were used to strip fine ROFA (PM2.5) solids from the ambient air-cooled, post-cyclone stream and stored in glass bottles for further analysis. Several grams of PM2.5 samples derived from baseline No. 5 (BL#5) and high sulfur No. 6 (HS#6) oils, which were previously found (3, 5) to be richest in metal content, were brought to the University of Kentucky for further analyses. Several drops of homogenized suspensions prepared by ultrasonicating a few milligrams of the ROFA PM2.5 sample in acetone were deposited onto TEM grids coated with a lacey carbon film. Using acetone to disperse inorganic powders is a commonly used technique for TEM sample preparation. Because of the high temperatures of combustion, fly ash usually does not contain much acetone-soluble material. It is possible to condense moisture and some VOCs on the fly ash during cooling. Acetone suspension dissolves this “binder” material and breaks up aggregates to provide physically separated individual particles amenable for TEM analysis. Acetone may dissolve some light organics in the sample but does not interact with the inorganic particles. Cross-sectional samples for TEM analysis were also made by embedding the ROFA PM2.5 samples in EMBed-812 resin and then cutting ultrathin sections with a diamond knife in a Reichert Ultracut E ultramicrotome. Instrumentation. The two ROFA PM2.5 samples were studied with a JEOL JEM-2010F field emission analytical transmission electron microscope equipped with an Oxford EDS detector, a STEM (scanning TEM) unit and a GIF/PEELS (Gatan Imaging Filter/Parallel EELS) system. The particle size and morphology were characterized from bright field images recorded by a Gatan 794 slow-scan CCD (charge coupled device) camera. The elemental compositions were determined by analyzing the EDS spectra. SAED was used to identify the crystalline phases. EELS spectra were recorded in diffraction mode (image-coupled mode) with an energy resolution of 1 eV (full-width at half-maximum of zero-loss peak) and a dispersion rate of 0.2 eV/channel and analyzed by using Digital Micrograph software. Built-in macros of Digital Micrograph software were also used to process raw 6554

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

FIGURE 2. Comparison of carbon K-edge EELS spectra from (a) char particles with graphitic layers more parallel to the electron beam, (b) soot aggregate particles with randomly orientated graphitic layers, (c) char particles with graphitic layers more normal to the electron beam, and (d) TEM grid amorphous carbon films. (Intensities are approximately normalized at σ* peak at 292.4 eV.) EELS data by subtracting background and by correcting for the contribution of plural inelastic scattering in some thick samples.

Results and Discussion Since both ROFA PM2.5 samples effectively showed similar characteristics with regard to chemical compositions and speciation, no effort has been made to differentiate them in this study. The following discussion can be considered a summary of the general characteristics of high-sulfur ROFA PM2.5 samples examined in this study by the different TEM techniques. Carbonaceous Particles. Carbon is the dominant element in these ROFA samples. Based on the loss of ignition (LOI) values, the carbon contents for the BL#5 and HS#6 samples are approximately 64 and 87 wt %, respectively. Two kinds of distinctive carbonaceous particles were found in these samples (Figure 1). The first kind consists of residual fuel char or coke particles that are derived from the incomplete combustion of the less volatile fraction of the residual fuel oil and subsequent carbonization. These rather large carbonaceous particles typically have a porous spherical morphology (5) with a typical size range of 1-100 µm. Usually these particles are separated by cyclone into larger size

FIGURE 3. Typical EDS spectra of char (top) and soot (bottom) particles. EDS spectra are approximately normalized to the height of the carbon peak for comparison. (PM2.5+) fraction. However, occasionally, in turbulent combustion atmosphere and in the exhaust pathways, these large particles shed small pieces as irregularly shaped flakes as shown in Figure 1. The other kind of carbonaceous particle consists of residual fuel soot particles exhibiting fractal-like chain structures similar to those of diesel soot PM. The spherical primary particles have a size varying from 10 to 80 nm (most at 30-40 nm), and the whole aggregate size can be as large as 1 µm. The formation mechanism is probably similar to that of diesel soot and involves rearrangement and nucleation of fragment fuel molecules, surface growth by condenseation, and agglomeration due to Brownian motion (14, 15). The microstructural differences between soot and char particle types are reflected in their SAED patterns. The small soot particles show broad and diffuse 002 and hk0 rings, which indicate a turbostratic structure since hkl reflections that characterize and define a graphitic structure are absent. The larger char particles also show a turbostratic structure, but the diffraction rings are sharper and more intense, which indicates an increase in the dimensions of the basic structural unit (BSU). Furthermore, the char particles often show anisotropic microtextures. The 002 arcs in the SAED pattern (Figure 1b, inset) indicate a preferential orientation of BSUs or pore walls. This preferential structure is absent in the soot particles, since their SAED patterns show complete rings. The high resolution images (002 lattice fringe images) were used to give direct visualization of the microstructures (Figure 1, insets). The soot primary particles exhibit a microtexture based on a concentric arrangement of graphitic layers around the particle center, similar to the layered structure of an onion. Multiple spherical nuclei are frequently found in the primary particles. These nuclei or “growth centers” have also been observed in diesel soot (16), supporting the speculation that fuel soot derives from nucleation of hydrocarbon molecular droplets followed by collision and surface growth by condensation (14). The char particles show the order-deficient arrangement of clusters consisting of several subparallel graphitic layers. Most of the fringe lengths are less than 5 nm, and the stacks typically vary from three to seven layers. Carbon K-edge EELS spectra of some typical carbonaceous particles are shown in Figure 2. The peaks at 285.4 and 292.4 eV correspond to electron transitions from the 1s level to antibonding π* and σ* states, respectively. Trends toward ordered or amorphous structures can be derived from these peak features. The σ* peak at 292.4 eV is prominent in the carbonaceous particles examined in this study but is smeared out in the amorphous carbon films of the TEM grid. The loss of structural order results in a relaxation of the selection

FIGURE 4. Micrograph and EDS spectrum of a typical composite inorganic particle with multiple elements.

FIGURE 5. Comparison of vanadium L-edge EELS spectra of various vanadium oxide standards. The feature at 530-535 eV is due to the oxygen K-edge. rules for the 1s to σ* transition (17) and produces a broad featureless peak (290-310 eV) in the EELS spectra of the TEM amorphous carbon films. This trend is also reflected in the broadening of the π* peak. Therefore the carbonaceous particles in ROFA possess more ordered microstructures than do amorphous carbon films. Structural anisotropy of fuel char particles is demonstrated by the variation of the intensity of the π* peak (Figure 2, spectra a and c) (i.e., particles with graphitic layers more closely parallel to the electron beam will produce higher intensity π* peaks), whereas particles with graphitic layers more normal to the electron beam produce lower intensity π* peaks. This reasoning is also valid for EELS spectra collected at normal large collection angles. VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6555

FIGURE 6. (a and b) Micrograph and SAED pattern of rodlike crystalline and amorphous vanadium oxide particles, respectively. (c) Micrograph of a vanadium oxide nanocrystal. The inset shows its lattice fringe image and the FFT of the image. (d) EELS spectra recorded from above particles (spectra a-c correspond to particles in panels a-c, respectively). On the other hand, such anisotropy was not observed for the fuel soot particles as the variation of π* peak intensity is relatively small among different soot aggregates, most probably due to the relatively random distribution of graphitic layers. Typical EDS spectra of char and soot particles exhibit a dominant C peak and much weaker but discernible O and S peaks (Figure 3). Usually the S content of soot particles is much less than that of char particles. This compositional difference is consistent with the formation mechanisms of the soot and char particles. Since the char particles do not undergo a vaporization-condensation process, inorganic elements from the parent oil fuel (e.g., V and P) are more readily preserved. XAFS analysis (5) has shown an increase in the sulfate form at the expense of the thiophenic form with decrease in the carbon content (higher burn-off) of ROFA. Both types of carbonaceous particles, especially the porous char particles, are often found mixed/coated with exterior transition metal compounds. Such particles could effectively act as a catalyst “support” for transition metal compounds that catalyze chemical reactions in the atmosphere and in biochemical processes in the body. 6556

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

Inorganic Particles. Previously (5), the concentrations of inorganic constituents of BL#5 and HS#6 ROFA PM2.5 samples were measured by acid digestion followed by ICP/MS analyses. S, V, Zn, Ni, Mg, and Cu were observed in significant concentrations with Pb, As, Cr, Mn, Sb, and Cd at lower concentrations. XAFS analyses indicated that the major chemical species present are transition metal sulfates, sulfides, and oxides (5). Such species are also observed with the TEM techniques used in this study. In addition, other chemical species that could be present based on TEM studies include vanadates, phosphates, and aluminosilicates. These chemical species either exist separately or fuse together forming multi-element composite particles. A submicron spherical particle containing Al, Si, P, S, Ca, Ti, V, Fe, and Ni (Cu is due to the TEM grid) is shown in Figure 4. These composite particles typically have a relatively larger size (>0.5 µm) and appear to consist mainly of sulfates and silicates. The following section will discuss these species in more detail. The focus will be on the major transition metals (V, Ni, and Fe). Vanadium Species. Vanadium sulfates are the major chemical form of vanadium in ROFA PM2.5 samples. These sulfate particles often incorporate other metallic elements,

FIGURE 7. Micrograph and SAED pattern of an ammonium vanadate particle. such as Ni, Zn, Fe, Ca, and Na. However, it is difficult to get more detailed information on the chemical form from the EELS and SAED analysis because these particles are sensitive to the electron beam (possibly due to being hydrated). Our earlier study (5) indicated that VOSO4‚xH2O is the dominant form of vanadium. However, it is very likely that these hydrated vanadium sulfates are mixed with other metal sulfates or incorporate other metal atoms resulting in chemically complex sulfates. Vanadium-rich oxide particles were also frequently found in the samples. Since the vanadium atom has an outer electron shell arrangement of 3d34s2, it can form several oxides with different valence states depending on the number of d and s electrons lost. Previously, XRD indicated that V2O5 was present in the samples (5). It should be noted that the XRD technique works well for crystalline compounds with concentrations typically >1 wt %. For amorphous or crystalline compounds with low concentrations and small dimensions (nanometer size), other techniques are needed. EELS combined with SAED was used to probe the valence states of vanadium oxides in this study. The EELS spectra of several model vanadium oxide compounds examined under the same experimental condition as the ROFA PM samples are given in Figure 5. Using the spectrum of V2O5 as an example, the two peaks at 519 and 525.4 eV correspond to vanadium L3-edge (2p3/2) and L2-edge (2p1/2), respectively. The peaks above 530 eV correspond to oxygen K-edge features that are related to the excitations of oxygen 1s electrons to empty hybridization orbitals of vanadium and oxygen. Detailed analyses of oxygen K-edge EELS spectra of transition metal oxides using molecular orbital theory and ligand field theory have been reported (18, 19). As the valence state of vanadium in its oxides increases from +2 (in VO) to +5 (in V2O5), there is a chemical shift for the peak positions of the L3- and L2-edges

from 516.8 to 519 eV and from 523.4 to 525.4 eV, respectively. In addition, the L2/L3 intensity ratio and the intensity of the peak at 530 eV assigned to the electron transition from the O 1s level to 2t2g(V3dπ + O2pπ) orbitals increase (18, 20). The observed trend agrees well with the near edge X-ray absorption fine structure (NEXAFS) studies of vanadium oxides (20, 21). Therefore these spectral features can be used to probe the valence states of vanadium oxides in ROFA, especially for those amorphous vanadium species whose microstructures are not revealed by SAED. Several vanadium oxides identified by using EELS, SAED, and HRTEM are shown in Figure 6. Figure 6a shows a rodlike vanadium oxide (VO2) that has an orthorhombic crystalline structure (PDF 42-0876). The EELS spectrum (Figure 6d) indicates that the valence state is close to +4. Another rodlike amorphous vanadium oxide particle is shown in Figure 6b, and its EELS spectrum indicates that the valence state is between +3 and +4. Vanadium oxide particles with different morphologies such as angular and irregular shapes were also found in the samples. In addition, a small amount of vanadium oxide nanoparticles are present in the samples. The HRTEM image and FFT (fast Fourier transform) of a vanadium oxide nanocrystal with a tetragonal structure (PDF 15-0629) are shown in Figure 6c. Its EELS spectrum suggests a valence state between +2 and +3. By using EELS, the valence states of vanadium oxides in the ROFA sample have been found to vary from +2 to +5. Vanadium also exists in the form of vanadate compounds. Figure 7 shows the micrograph of an ammonium vanadate particle that has a monoclinic crystalline structure (PDF 310075). The SAED pattern in the [100] zone axis direction gives direct information on the b axis (b ) 3.66 Å). Other crystalline vanadates identified are CaV3O7 (PDF 26-0337) and Ni3(VO4)2 (PDF 37-0353). EDS analyses also suggest that there are sodium vanadates in the samples. Crystalline sodium vanadates, such as NaVO3‚3.5H2O and NaVO3‚1.9H2O, have been previously identified in oil fly ash by XRD (22). The relative occurrence of vanadates is much less than that of the vanadium sulfates but appears to approximate that of vanadium oxides. In addition, some vanadium phosphates were also found in the ROFA samples. Figure 8 shows the EDS and EELS spectra of an amorphous vanadium phosphate particle. The near-edge structures of V and O closely resemble those of (VO)2P2O7 (23). (Coincidentally, it is an active catalyst used in commercial oxidation of butane to maleic anhydride.) Nickel Species. Nickel sulfate is the dominant form of nickel in ROFA. As expected, nickel sulfate particles frequently incorporate other metallic elements, typically Mg and Zn. Since Mg and Zn both exist in the 2+ oxidation state, substitution with Ni2+ is common. An example of these complex nickel sulfate particles is shown in Figure 9. The EDS spectrum shows three major peaks for S, Ni, and O, suggesting a sulfate form. However, from the EDS analysis it is impossible to determine whether it is hydrated or not. Since the particle was sensitive to the electron beam, it is very likely that it exists in a hydrated form. The selected area diffraction pattern showed an amorphous microstructure.

FIGURE 8. EDS spectrum, micrograph (left), and EELS spectrum (right) of a vanadium phosphate particle. VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6557

FIGURE 9. Micrograph and EDS spectrum of a complex nickel sulfate particle incorporating other metallic elements (Mg, V, Zn, and perhaps Cu).

FIGURE 10. Micrograph of a nickel oxide (NiO) particle. Insets are its high-resolution image and SAED pattern. The high-resolution image gives direct visualization of the two-dimensional lattice arrangement with a lattice constant a ) 0.418 nm in the [001] direction. Nickel oxides were not found in previous XRD studies, but have been identified by SAED and HRTEM in this study. A crystalline NiO particle with a size about 60 nm is shown in Figure 10. The SAED pattern corresponds to a cubic structure (PDF 04-0835), and the high resolution image gives a direct visualization of the two-dimensional lattice arrangement with a lattice constant a ) 0.418 nm in the [001] direction. Nickel has been found to readily form a spinel

crystalline structure with other transition metal elements. Figure 11 shows a nickel aluminum oxide particle that has a crystalline structure corresponding to NiAl2O4 (PDF 10-0339). The EDS spectrum recorded from this particle also shows some Zn and Fe. Therefore, the general formula should be written as XY2O4, where X ) Ni2+, Zn2+, and Fe2+ and Y ) Al3+ and Fe3+. Similar Ni spinel compounds in ROFA PM have been identified by Ni XAFS spectroscopy (24). As discussed above, there appear to be some vanadates in the samples. Ni is one element capable of forming vanadates. A micrograph of a crystalline Ni3(VO4)2 particle that has an orthorhombic structure is shown in Figure 12. The EDS spectrum recorded from this particle showed a Ni:V atomic ratio to be approximately 3:2, supporting the chemical form (Ni3(VO4)2) identified by SAED. Iron Species. Iron was found in the forms of sulfates, sulfides, phosphates, and oxides. Iron sulfate and phosphate particles typically incorporate other elements such as Na, Mg, V, Ni, and Zn. Based on the EDS analyses, composite compounds such as Na3Fe(SO4) and (Zn, Fe)3(PO4)2 could be present. Some of these composite compounds are sensitive to the electron beam (possibly because of the hydrated forms) making accurate analysis difficult. Iron oxides typically exist in the crystalline forms. Hematite (Fe2O3) and magnetite (Fe3O4) have both been identified. Fe may incorporate other atoms forming oxides with a spinel crystalline structure. Figure 13 shows an iron,vanadium oxide ((Fe,V)3O4) that has a cubic structure. The lattice constant derived from the SAED pattern is very close to that of magnetite. Several iron, vanadium oxide particles with amorphous microstructures were also identified in the samples. Other Species. Some aluminosilicate particles with spherical or irregular shaped morphologies, and typical sizes larger than 0.5 µm have been found in the samples. Other transition metals such as Ti, Cr, Cu, and Zn were less frequently observed. They mainly exist in the form of sulfates, oxides, and less frequently, phosphates. Small amounts of lead compounds are present in the sample, typically in the form of sulfates. The EDS analyses indicate that Ba, Ca, and Cu are frequently associated with Pb compounds. The crystalline phase Cu6PbO8 with a cubic structure (PDF 07-0028) has been identified by SAED. A small amount of sulfides is also present in the samples. Nanometer-sized crystalline CaS (PDF 08-0464) particles were identified by SAED and EDS. Other sulfides, such as nickel sulfides, have been identified in a recent leaching-XAFS study (24). Some rarely reported boron species in ROFA have been identified by EELS and SAED. A micrograph of a boron phosphate particle is shown in Figure 14. The EDS spectrum only shows C, O, and P peaks, and the minor B peak is barely seen due to the detector efficiency and resolution. Therefore,

FIGURE 11. Micrograph and EDS spectrum of a crystalline oxide particle (indicated by the arrow) that contains mainly Ni and Al and lesser amounts of Zn and Fe. The SAED pattern is indexed as NiAl2O4 that has a spinel crystalline structure. The generic spinel formula can be written as XY2O4, where X ) Ni2+, Zn2+, and Fe2+ and Y ) Al3+and Fe3+. 6558

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

FIGURE 12. Micrograph (inset is the SAED pattern) and EDS spectrum of a nickel vanadate (Ni3(VO4)2) particle.

FIGURE 13. Micrograph and SAED pattern of an iron vanadium oxide ((Fe,V)3O4) particle. it is hard to surmise the real chemical form based purely on the EDS analysis. By carefully examining its EELS spectrum, the boron K-edge has been identified and found compatible with that of BPO4. The SAED pattern further reveals that it has a tetragonal crystalline structure with lattice constants a ) 4.34 Å and c ) 6.64 Å (PDF 34-0132).

Implications of Results to Health Studies. The application of analytical electron microscopy techniques for the characterization of ROFA PM2.5 provides much information about the size, morphology, composition, crystallinity, and elemental oxidation states of individual particles, which cannot be obtained by conventional bulk analysis methods. The most important finding in the current study is that most individual inorganic particles in ROFA PM2.5 exhibit a much more complex chemistry than has been previously realized. These composite particles typically exist in the form of multi-metal sulfates that are readily soluble; they are therefore potentially bioavailable and a concern for human health. Other species with multiple elements identified in this study include oxides (spinels), phosphates, and vanadates. Such speciation information should be used to augment current epidemiological and toxicological studies based on pure compound data alone. Another important finding is that considerable numbers of particles are in the nanometer size range, although they are relatively insignificant by weight. These crystalline or amorphous nanoparticles possess larger surface area and are likely to exhibit enhanced solubility and reactivity compared to larger particles of similar compositions. Therefore, even relatively insoluble species such as vanadium and nickel oxides or sulfides may exhibit increased bioavailability in such small sizes. Differences between crystalline and amorphous substances of similar chemical species should also be taken into account in toxicological studies. The last implication of the results of this study is that multiple techniques are necessary to characterize ROFA and

FIGURE 14. Rare boron phosphate particle (BPO4) was identified by EDS, EELS, and SAED. The boron K peak cannot be discerned in the EDS spectrum (right top) but was clearly revealed in the EELS spectrum (right bottom). The SAED pattern (inset in the left) corresponds to a tetragonal crystalline structure. VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6559

other PM2.5 samples comprehensively. Minor species such as vanadates and phosphates can be identified by microanalysis techniques but were not found in previous bulk analyses. In contrast, nickel sulfides were rarely observed in this study but were identified in a complementary XAFS investigation (24), most likely because the sulfides only exist in the interiors of carbonaceous particles and are not easily accessible to TEM methods. The complementary nature of bulk and microanalysis techniques is clearly necessary for in-depth chemical speciation of ROFA PM2.5. Analytical electron microscopy combined with bulk techniques such as XAFS spectroscopy should find significant future application in identification of toxic phases in aerosols and other environmental studies.

Acknowledgments The authors are grateful to Dr. William P. Linak and Dr. C. Andrew Miller of U.S. Environmental Protection Agency (Research Triangle Park, NC) for generating the ROFA samples and to Dr. Alan Dozier for his assistance in the use of the TEM. This work was supported by the National Science Foundation under CRAEMS Grant CHE-0089133.

Literature Cited (1) U.S. EPA. Clearing House for Inventories and Emission Factors (CHIEF). http://www.epa.gov/ttn/chief/eiip/pm25inventory/ index.html. (2) National Air Quality and Emissions Trends Reports2003 special studies edition; U.S. EPA Report 454/R-03-005. Also available at http://www.epa.gov/airtrends/. (3) Huggins, F. E.; Shah, N.; Huffman, G. P.; Robertson, J. D. Fuel Process. Technol. 2000, 65-66, 203-218. (4) Henry, W. M.; Knapp, K. T. Environ. Sci. Technol. 1980, 14, 450-456. (5) Huffman, G. P.; Huggins, F. E.; Shah, N.; Huggins, R.; Linak, W. P.; Miller, C. A.; Pugmire, R. J.; Meuzelaar, H. L. C.; Seehra, M. S.; Mannivannan, A. J. Air Waste Manage. Assoc. 2000, 50, 11061114. (6) Dreher, K. L.; Jaskot, R. H.; Lehmann, J. R.; Richards, J. H.; McGee, J. K.; Ghio, A. J.; Costa, D. L.; J. Toxicol. Environ. Health 1997, 50, 285-305.

6560

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

(7) Dye, J. A.; Adler, K. B.; Richards, J. H.; Dreher, K. L. Am. J. Physiol. 1999, 277, L498-L510. (8) Lambert, A. L.; Dong, W.; Selgrade, M. K.; Gilmour, M. I. Toxicol. Appl. Pharmacol. 2000, 165, 84-93. (9) Mannivannan, A.; Seehra, M. S. ACS Fuel Div. Preprints 2000, 45, 446-450. (10) Galbreath, K. C.; Zygarlicke, C. J.; Toman, D. L.; Huggins, F. E.; Huffman, G. P. Combust. Sci. Technol. 1998, 134, 243-262. (11) Galbreath, K. C.; Zygarlicke, C. J.; Huggins, F. E.; Huffman, G. P., Wong, J. L. Energy Fuels 1998, 12, 818-822. (12) Weinbruch, S.; Aken, P.; Ebert, M.; Thomassen, Y.; Skogstad, A.; Chashchin, V. P.; Nikonov, A. J. Environ. Monit. 2002, 4, 344350. (13) Miller, C. A.; Linak, W. P.; King, C.; Wendt, J. O. L. Combust. Sci. Technol. 1998, 134, 477-502. (14) Kittelson, D. B. J. Aerosol Sci. 1998, 29, 575-588. (15) Kim, D.; Gautam, M.; Gera, D. J. Aerosol Sci., 2002, 33, 16091621. (16) Ishiguro, T.; Takatori, Y.; Akihama, K. Combust. Flame 1997, 108, 231-234. (17) Berger, S. D.; Mckenzie, D. R.; Martin P. J. Philos. Mag. Lett. 1988, 57, 285-290. (18) Grunes, L. A.; Leapman, R. D.; Wilker, C. N.; Hoffmann, R.; Kunz, A. B. Phys. Rev. B 1982, 25, 7157-7173. (19) De Groot, F. M. F.; Grioni, M.; Fuggle, J. C. Phys. Rev. B 1989, 40, 5715-5723. (20) Chen, J. G.; Kim, C. M.; Fru ¨ hberger, B.; DeVries, B. D.; Touvelle, M. S. Surf. Sci. 1994, 321, 145-155. (21) Chen, J. G.; Fru ¨ hberger, B.; Colaianni, M. L. J. Vac. Sci. Technol. 1996, 14, 1668-1673. (22) Bacci, P.; Del Monte, M.; Longhetto, A.; Piano, A.; Prodi, F.; Redaelli, P.; Sabbioni, C.; Ventura, A. J. Aerosol Sci. 1983, 14, 557-572. (23) Lin, X. W.; Wang, Y. Y.; Dravid, V. P.; Michalakos P. M.; Kung, M. C. Phys. Rev. B 1993, 47, 3477-3481. (24) Huggins, F. E.; Huffman, G. P.; Linak, W. P.; Miller, C. A. Environ. Sci. Technol. 2004, 38, 1836-1842.

Received for review January 23, 2004. Revised manuscript received August 11, 2004. Accepted August 18, 2004. ES049872H