Environ. Sci. Technol. 2002, 36, 2484-2490
Resuspension of Soil as a Source of Airborne Lead near Industrial Facilities and Highways T H O M A S M Y O U N G , * ,† D E O A . H E E R A M A N , ‡,§ G O R K E M S I R I N , †,⊥ A N D LOWELL L. ASHBAUGH‡ Department of Civil and Environmental Engineering, Air Quality Group, Crocker Nuclear Laboratory, University of California, Davis, One Shields Avenue, Davis, California 95616
Geologic materials are an important source of airborne particulate matter less than 10 µm aerodynamic diameter (PM10), but the contribution of contaminated soil to concentrations of Pb and other trace elements in air has not been documented. To examine the potential significance of this mechanism, surface soil samples with a range of bulk soil Pb concentrations were obtained near five industrial facilities and along roadsides and were resuspended in a specially designed laboratory chamber. The concentration of Pb and other trace elements was measured in the bulk soil, in soil size fractions, and in PM10 generated during resuspension of soils and fractions. Average yields of PM10 from dry soils ranged from 0.169 to 0.869 mg of PM10/g of soil. Yields declined approximately linearly with increasing geometric mean particle size of the bulk soil. The resulting PM10 had average Pb concentrations as high as 2283 mg/kg for samples from a secondary Pb smelter. Pb was enriched in PM10 by 5.36-88.7 times as compared with uncontaminated California soils. Total production of PM10 bound Pb from the soil samples varied between 0.012 and 1.2 mg of Pb/kg of bulk soil. During a relatively large erosion event, a contaminated site might contribute approximately 300 ng/m3 of PM10-bound Pb to air. Contribution of soil from contaminated sites to airborne element balances thus deserves consideration when constructing receptor models for source apportionment or attempting to control airborne Pb emissions.
Introduction Historically, most airborne lead (Pb) was emitted by industries handling Pb-bearing materials and by vehicles (e.g., refs 1 and 2). As these sources have declined in significance in the United States because of stricter emissions standards and the phase-out of leaded gasoline, ambient Pb concentrations have fallen significantly. In the Los Angeles, CA, area, for example, typical Pb levels have declined from above 3000 * Corresponding author phone: 530-754-9399; fax: 530-752-7872; e-mail:
[email protected]. † Department of Civil and Environmental Engineering. ‡ Air Quality Group, Crocker Nuclear Laboratory. § Present address: Beak International Inc., 14 Abacus Rd., Brampton, Ontario, Canada L6T 5B7. ⊥ Present address: Dokuz Eylu ¨l U ¨ niversitesi, Mu ¨ hendislik Faku ¨ ltesi, C¸ evre Mu ¨ h. Bo¨lu ¨ mu ¨ Kaynaklar Kampu ¨ s, Buca, 35160, Izmir, Turkey. 2484
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002
ng/m3 in the early 1970s (3) to around 20 ng/m3 in the late 1990s (4). However, spikes in airborne Pb concentration above 300 ng/m3 are still occasionally observed at daily monitoring stations in California (4). Fugitive emissions have emerged as a leading potential cause of such spikes, particularly in semiarid or arid regions where soils may serve as an important source of particulate matter to the atmosphere (5, 6). Over 40% of the airborne Pb in the South Coast Air Basin was estimated to result from resuspension of Pb contaminated soils in 1989 based on a mass balance study, but the amount of resuspended material and the concentration of Pb within it were average values obtained from literature sources (7). Particulate matter less than 10 µm in aerodynamic diameter (PM10) generated from soils has been implicated in the longrange transport of Pb and may be an important source of Pb in house dust resulting in elevated blood Pb levels in children (5, 8). Receptor modeling is commonly employed to estimate the contribution of various sources to ambient concentrations of particulate-bound elements (9). Geologic source material compositions in chemical mass balance receptor models are typically determined from samples at a limited number of sites collected near the receptor or are derived from tabulated average soil compositions. However, trace element concentrations in soils may greatly exceed average levels in areas impacted by anthropogenic emissions such as industrial facilities and highways, and differences between contaminated and uncontaminated soils may be even greater in the fine particle fraction that is subject to long-range transport and deposition within the respiratory tract. There has been little systematic investigation of the contribution of contaminated soils to local element balances in ambient air. The goal of this study was to determine source profiles of elements within PM10 generated from soil samples collected near industrial facilities and highways and to estimate their potential impact on local airborne Pb concentrations. Fine soil particles may be suspended in the atmosphere by the action of wind or by mechanical disturbances (e.g., tilling, mowing). Movement of aerosol-sized soil particles is strongly related to the occurrence of saltation as sand-sized particles are lifted and redeposited, disrupting aggregate structure and ejecting smaller particles ( B > 150 µm, 150 > C >75 µm, 75 > D > 45 µm, 45 > E > 38 µm, and F < 38 µm) to determine the particle size distribution and to prepare size-fractionated material for subsequent study. Approximately 4 g of fraction A of each sample were powdered for chemical analysis; similar preparation of the other fractions was not necessary because of their smaller particle sizes. An unpowdered sample of each fraction A was reserved for measurement of PM10 formation potential. Analysis of Bulk Soils and Fractions. Pb and potential co-contaminants Ni, Cu, and Zn were analyzed in bulk samples and their size fractions using energy-dispersive X-ray fluorescence spectrometry (XRF; Kevex 0700 XES Control with 5230 Energy/Digital Converter and 4460 Pulse Converter). Elements other than Pb were chosen for analysis because they are chemicals of environmental concern, they are likely to be found in some soils impacted by industrial activity, and they could be analyzed conveniently and accurately by XRF in conjunction with Pb. Zr was the target element for Pb analyses (30 kV, 0.2 mA, 350 counts), and Ge was the target element for Ni, Cu, and Zn analyses (20 kV, 0.5 mA, 350 counts). For each batch, either the Zr-targeted element (Pb) or Ge-targeted elements (Ni, Cu, and Zn) were analyzed along with relevant standard reference materials obtained from the National Institute of Standards and Technology (NIST 2709, 2710, 2711) or the United States Geological Survey (USGS AEG GXR4, GXR5). After analysis of a sample batch, the concentrations of Pb and co-contaminants were calculated by an automated calibration procedure. Measurement of PM10 Formation Potential. PM10 content was measured by resuspending bulk soils or fractions in a dust resuspension test chamber (22). The four basic components of the resuspension apparatus include (i) a dust resuspension chamber that separates PM10 from the soil sample, (ii) a collection chamber that collects suspended airborne dust, (iii) a Sierra Andersen PM10 inlet that separates PM10 in the resuspended dust from larger resuspended material, and (iv) an IMPROVE (Interagency Monitoring of Protected Visual Environments) sampler that collects PM10 onto Teflon filter cartridges. Each experiment was initiated by placing a quantity of the sample (0.5-1.0 g of bulk soil or a particular size fraction) into the dust resuspension chamber, sealing it, and forcing air (3.5 L/min) through the soil sample for 15 s. The 3.5 L/min flow rate corresponds to a velocity of 15.4 cm/s at the bottom of the resuspension chamber and 6.1 cm/s in the expanded cross section at the middle of the chamber. Particles with an aerodynamic diameter less than about 40 µm are carried out of the resuspension chamber and into the collection chamber via a 1-cm diameter aluminum tube (22). The dust is mixed VOL. 36, NO. 11, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2485
TABLE 2. Bulk Soil Characteristics and Corresponding PM10 Data from Each Sitea Pb enrichment site GM PM BP LS SB RS
dg (µm)
Cu
344 (14) 41.2 (27.7) 89.4 (38.7) 197 (13) 81.1 (54.9) 285.1 (151.7)
5.33 (0.06) 5.98 (1.62) 4.89 (0.98) 8.95 (0.46) 7.59 (2.72) 4.85 (0.84)
CA avg
soil Pb (mg/kg)
soil Ni (mg/kg)
soil Cu (mg/kg)
soil Zn (mg/kg)
PM10 yield (mg/g)
PM10 Pb (mg/kg PM10)
PM10 Pb yield (mg/kg soil)
fine soil/ coarse soil
PM10/ bulk soil
65.2 (3.9) 99.4 (5.9) 66.3 (4.6) 636 (522) 132 (79.9) 135 (162) 23.9 (13.8)
42.3 (7.8) 3.13 (0.70) 6.09 (3.47) 23.7 (3.9) 12.1 (10.3) 110 (170) 57.0 (80.0)
25.7 (1.3) 12.3 (4.9) 11.8 (6.0) 39.0 (12.7) 18.9 (7.7) 30.0 (10.6) 28.7 (19.3)
86.8 (10.8) 62.9 (11.4) 49.5 (11.9) 106 (27.5) 620 (614) 97.3 (49.6) 149 (32)
0.198 (0.120) 0.697 (0.161) 0.560 (0.166) 0.578 (0.484) 0.869 (0.455) 0.169 (0.081)
544 (177) 110 (61) 118 (68) 2283 (2147) 184 (153) 79.5 (6.7)
0.104 (0.073) 0.0713 (0.0235) 0.0628 (0.0297) 1.23 (1.55) 0.147 (0.099) 0.012 (0.005)
2.43 (0.41) 1.28 (0.14) 1.42 (0.06) 1.12 (0.25) 1.55 (0.70) 7.83 (8.34)
8.31 (2.54) 1.10 (0.57) 1.74 (0.96) 3.03 (1.41) 1.34 (0.43) 1.31 (0.91)
a Average concentrations of Pb, Ni, Cu, and Zn for 50 benchmark California Soils (CA avg) are included for comparison (24). Numbers in parentheses are standard deviations; dg represents the geometric mean particle diameter Cu is the uniformity coefficient defined in the text.
throughout the chamber and is collected through a Sierra Andersen PM10 inlet connected to an IMPROVE sampler containing preweighed 25-mm Teflon filters. An experiment consists of 11 air pulses (15 s), each followed by a 5-min period for collection of dust on the filter cartridges. At the end of a run, a delay of 2.5 min is provided before turning off the vacuum to allow all suspended particles to be removed from the collection chamber. The process of soil resuspension and dust collection is automated with a timed relay device. The Teflon filter is then reweighed and the amount of PM10 removed from the sample is calculated as the difference between the initial and final weight of the filter. The PM10 yield is calculated as the weight of PM10 collected on the filter per gram of soil initially charged to the system. Most of the resuspension experiments were replicated (65% duplicate, 20% triplicate), but 15% of the results were based on a single trial. Analysis of Filters. Lead concentrations on the filters from the resuspension experiments were determined by energydispersive XRF at the Crocker Nuclear Laboratory using previously reported methods (23). Concentrations were also measured for 21 other elements (S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Ga, As, Se, Br, Rb, Sr, Y, Zr, and Hg) routinely determined in filter analyses. The baseline calibration of the system consists of analyzing a series of 30 commercial and NIST elemental standards. At the beginning of every analytical session, the calibration was checked with a series of 15 standards and by reanalyzing several samples with concentrations known from previous analysis sessions. The spectra were analyzed by an automated code developed especially for particulate samples. The program removes the background, searches for all peaks, and fits each peak to a Gaussian distribution. Overlapping emission line interferences were eliminated using a series of algorithms. Analyses were performed only after the calibration and reanalysis samples were within the required 4% precision limits. The standard and reanalysis trays were reanalyzed at the end of each session. None of the elements measured, including Pb, were found at concentrations above method detection limits in any of the blank filters analyzed. Elemental concentrations on the particulate matter (ng/g) are determined by multiplying the areal density of each element in ng/cm2 by the available filter area (2.2 cm2) and dividing by the particulate mass recovered on the filter.
Results and Discussion Soil Lead Concentrations. Average Pb concentrations at the six types of sites studied ranged from 65.2 mg/kg at site GM 2486
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 11, 2002
to 636 mg/kg at site LS (Table 2). Coefficients of variation for the Pb concentrations were below 7% for the sites with average Pb concentrations below 100 mg/kg (sites GM, PM, and BP) while the remaining three sites (LS, SB, and RS) had coefficients of variation greater than 60%. The small coefficients of variation at the less contaminated sites suggest that the emission source was not controlling soil Pb levels at these sites because samples taken at various distances and directions from the presumed emission source had similar Pb concentrations. Pb concentrations at these sites therefore appear to be controlled by geologic background or regional-scale Pb deposition rather than emissions from the particular facility. The coefficients of variation for the roadside samples are not directly comparable to the others because samples came from three different locations around the state, rather than from multiple locations at a single facility. Average Pb levels at all of the sites exceed the average of 23.9 mg/kg (p e 0.05) measured for 50 benchmark California soils (24). For site SB, average downwind Pb concentrations were significantly higher (p e 0.05) than upwind concentrations. At site LS, Pb concentrations in samples collected within 200 m of the source were significantly higher (p e 0.01) than those in samples collected more than 200 m from the source. The Pb emission source therefore appears to control the surface soil Pb concentrations at these two sites. Sites GM, PM, and BP may be viewed as having “urban/industrial background” levels of Pb contamination in the surface soils because concentrations are elevated as compared with statewide averages for uncontaminated soils but the emission source does not appear to control soil Pb concentrations. In some cases, the relatively low soil Pb levels measured may be a result of the large distance between sampling locations and the source zone (site GM), but in other cases, they seem to reflect a true absence of significant Pb contamination (site PM). In contrast to the results for Pb, none of the sites had average levels of Zn, Ni, or Cu that were significantly above (p e 0.05) statewide California soil averages. Correlations among Pb, Ni, Cu, and Zn concentrations in the soil samples were examined to provide clues about the elemental contamination signature that might be expected in PM10 generated from each type of site. For the roadside samples, a statistically significant correlation (p e 0.05) was identified between Pb and Ni concentrations, suggesting a highway source for both elements. Ni emissions from catalyst equipped automobiles have been reported (9) and, although they would not have chronologically coincided with signifi-
cant automotive Pb emissions, both would be related to traffic levels at the roadside sites. Correlation of Pb with Zn and Cu was also expected for the roadside soils, but the observed correlation coefficients (0.911 and 0.986) were not statistically significant (p e 0.05) because of the small sample size (N ) 3) for the roadside soils. Soil Pb levels were also significantly correlated (p e 0.05) with Ni at site GM. Particulate matter emissions of up to 1800 µg/g of Ni have been reported from glass making operations (9). Ni and Cu were correlated with soil Pb at site LS consistent with particulate composition data from Pb furnaces of >10 000 µg/g Ni and >2000 µg/g Cu (9). A significant correlation (p e 0.05) observed between soil Pb and Cu concentrations at site SB would be expected based on the 6900 µg/g of Cu content in sandblasting emissions (9). As expected, no significant interelement correlations were observed at the facilities with Pb levels not controlled by the emission source (sites GM, PM, and BP) with the exception of Ni at site GM. As observed in previous studies, Pb has a tendency to be associated with smaller soil size fractions. These fractions have higher specific surface area and generally have higher contents of organic matter or Fe/Al oxides that may serve as Pb binding sites. This tendency to accumulate in smaller size classes was assessed by calculating the ratio of the Pb concentration in the
