Environ. Sci. Technol. 1993, 27, 2502-2510
Sources of Toxic Trace Elements in Urban Air in Illinois Clyde W. Sweet' and Stephen J. Vermette Illinois State Water Survey, 2204 Griffith Drive, Champaign, Illinois 61820
Sheldon Landsberger Department of Nuclear Engineering, University of Illinois, Urbana, Illinois 6 1801 ~~~~~~
Samples of airborne inhalable particulate matter (PM10)were collected in southeast Chicago and East St. Louis and analyzed for 29 trace elements using X-ray fluorescence spectrometry and neutron activation analysis. Wind trajectory analysis, factor analysis, and chemical mass balance modeling were used to determine the most important sources of the toxic trace elements. Industrial complexes were the most significant sources of copper and cadmium in East St. Louis (smelters) and of manganese and chromium in southeast Chicago (steel mills). It is concluded that some sources have a large effect on the concentrations of specific toxic metals in the atmosphere without significantly affecting overall PM-10 levels.
Introduction
areas in Illinois, southeast Chicago, and East St. Louis. Air quality at these locations is among the worst in Illinois in terms of criteria pollutants. One of the areas (southeast Chicago) has been designated by the US. EPA as a nonattainment area. This designation means that the probability of exceeding the PM-10 standard is greater than 95% (11). The two urban study areas are each part of large metropolitan areas that also contain major industrial sources of airborne particles bearing toxic trace metals. In addition, there is a long history of industrial pollution at both sites which has resulted in the contamination of surface soil and urban dust. This material is a potential area source of contaminated airborne particles resuspended by wind or vehicle traffic. Collection of fine and coarse airborne PM-10 particles helps identify the type of source responsible for emitting the particles. Hightemperature sources such as industrial stacks tend to emit fine PM-10 particles (0.5)to known source characteristics. Wind speed and direction were also included to help identify specific sources. For southeast Chicago, the fine component of PM-10 is influenced by the following sources: steel mill emissions and/or oil combustion (Mn, Fe, Cr, Pb, Zn, V, Ni), which account for 39 ?& of the variance; coal combustion/regional sulfate (S, Se, C1, Br, P, and mass), which accounts for 13% of the variance; a crustal source (Al, Si, Ti, and wind speed), which accounts for 8% of the variance; and steel mill fugitive dust (Mn, Cr, Fe, Ca), which accounts for 5% of the variance. The first factor, dominated by metallic
elements and associated with easterlywinds (iron and steel mill fetch) and weak airflow (poor ventilation), suggests stack emissions, particularly iron and steel, as a source. Emissions from oil burning (Ni and V) and other unidentified sources (Pb and Zn) co-vary with the mill emissions. Regional sulfate sources correlate (second factor) with the fine particle mass. The last two factors suggestfugitive emission sources. The distinction between these two is reinforced by the positive loading of wind speed with a soil source, suggesting wind erosion as the suspension mechanism. The lack of a similar relationship with the iron and steel-derived dust supports a mechanical method for dust suspension (e.g., truck traffic). Toxic elements are associated, in part, with each of the identified sources. No auto emissions could be isolated by factor analysis even though they are a source of P b in Chicago. This is probably due to the fact that airborne P b levels have been greatly reduced in recent years (5) and that auto exhaust co-varies with steel emissions and other P b sources. The coarse component of PM-10 in southeast Chicago is influenced by the following sources: a crustal source (Al, Si, Ti, K, Ca and mass), which accounts for 47% of the variance; contaminated urban dust (Fe, Cr, Pb, Br, Cu, Zn), which accounts for 1 2 % of the variance; steel mill dust (Fe, Cr, Mn, Ca), which accounts for 7% of the variance; and coal dust (S, Se and wind speed), which accounts for 6 % of the variance. Fugitive emission sources account for nearly all of the coarse PM-10 variability and are also correlated with total mass in southeast Chicago. It is reasonable to assume that urban activities in proximity to urban ambient air samplers contribute a substantial fraction of resuspended dust. The first factor, identified as soil (crustal), may include as sources local landfills and road dust uncontaminated by industry. Fugitive dusts related to the iron and steel mills are identified as road dust contaminated from iron and steel-related sources (truck spillage, track on, atmospheric fallout), dust entrained from iron and steel industry properties, and coal dust. Factor analysis statistics suggest that for East St. Louis the fine component of PM-10 is influenced by the following sources: a crustal source (Al, Si, Ti, Fe), which accounts for 32 % of the variance; zinc smelter emissions (Zn, Cd, Pb, Sn), which account for 21% of the variance; copper smelter emissions (Cu, Se, P), which account for 13% of the variance; motor vehicle emissions/road dust (Br, Pb, Ca and mass), which account for 9% of the variance; and regional sulfate (S and mass), which accounts for 5% of the variance. The first factor suggests an uncontaminated soil source. The influences of the Zn and Cu smelters within the study area are distinguished from one another. A motor vehicletroad dust source is suggested for the fourth factor. The East St. Louis sampler was located on a roadway median, and it is not surprising that road dust correlates best with the PM-10 fine mass, surpassing that of regional sulfur, which traditionally dominates. For East St. Louis, FA indicates that the coarse component of PM-10 is influenced by the following sources: a soil/road dust source (Al, Si, Ti, K, Fe, MN, Ca and mass), which accounts for 54% of the variance; a steel industry/coal dust source (Fe, Cr, Mn, S, Se), which accounts for 15% of the variance; zinc smelter emissions (Zn, Cd, Pb, Ni), which account for 10% of the variance; and copper smelter emissions (Cu, P), which account for 5% of the variance. The first two factors include enriched
metallic elements. Linkage with soil-related elements in the first factor suggests a road dust source with metallic elements contributed by roadside debris or fallout. Loadings for the second factor do not include soil-related elements (Al, Si,Ti). The association of Se and S with the metallic elements in this factor suggests an industrial source such as steel mills or unidentified, coal-burning industrial sources within the city of St. Louis. As with fine PM-10, both the Zn and Cu smelters are identified from other sources and are distinguished from one another. Separate FA based on specific wind direction sectors were also made for the East St. Louis data. The sources above were re-identified, but several additional sources were also identified including: steel mill emissions (northeast winds, Mn, Cu, Fe, Zn); oil combustion (southwest winds, Ni, V); lead smelter (northeast winds, Pb); incinerator emissions (northwest winds, Zn, K, Pb). At the rural site, a crustal source (Si, Ca, Al, Fe, K, Ti) accounted for 29.4% of the variance of the fine fraction of PM-10. Urban/industrial sources (Mn, Zn, Cu, Cr, Pb, V, C1, P) accounted for 25.7 % of the variance, and regional sulfate (S and mass) accounted for 6.2 % of the variance. For the coarse fraction of PM-10, a crustal dust source (Si, Al, Ca, K, Fe, Ti, Mn, Cr, Sr, Rb and mass) accounts for 41.5% of the variance. This source is positively correlated with wind speed. Urban/industrial sources (Cu, Zn, C1, S, Ni, Se, Br, Pb) account for 30% of the variance. Chemical Mass Balance. The final output in our receptor modeling approach is the source apportionment of the toxic trace elements. Source profiles were developed for the identified sources, and toxic air pollutant apportionment was determined from chemical mass balance (CMB) statistics (22). Often CMB has been carried out on single filters. This approach has the advantage that the number of sources that need to be considered are minimized. If the wind direction is constant during the sampling period, only upwind sources affect the sampling site. The major disadvantage is that a single filter represents only the time period sampled, not average conditions. In order to reflect average conditions, CMB would have to be done on a large number of filters representing all meteorological conditions in proportion to their actual occurrence throughout the year. Even then, some filters representing calm or variable wind conditions with high airborne trace element concentrations would not be represented. In addition, CMB analysis of single filters can easily be distorted by analytical or weighing errors. Our approach was to carry out CMB on the average results reported earlier (Tables 1-111). These data generally reflect average conditions at the sites, and analytical or weighing errors in a few individual filters will not be as significant when averaged into a large data base. Although all sources impacting on the sites need to be considered, we have already narrowed the list considerably using wind trajectory and factor analysis. As a check on the average results, CMB was also run on typical filters representing high impact from major pollution sources at the various sites. To carry out CMB analysis, source profiles were selected from a data base that has been compiled from the literature and from direct measurements (Table V). A "regional background source" was also calculated based on average concentration at the rural site. This source is meant to reflect air coming into the urban study areas that has not received any input from local sources but may be conEnviron. Sci. Technol., Vol. 27, No. 12, 1993 2507
Table V. CMB Source Profiles profile
ref
crustal dust (T41319) steel blast furnace (T28302) coal dust (T21204) coal burning (T11201) coke dust (T21203) copper secondary smelter (T29203) lead slag dust (T201501) incinerator (T17109) oil burning (T11501) motor vehicle exhaust steel composite emissions urban/steel dust (southeast Chicago) steel slag dust zinc smelter coal-fired power plant lead secondary smelter fugitive urban/industrial dust
23 23 23 23 23 23 23 23 23 20 24 12
12 25 25 25 25
taminated by transport from distance sources. Wind trajectory analysis and factor analysis both indicated that trace metals associated with urban/industrial sources are present in the fine and coarse PM-10 particles collected at our rural Illinois site. The selection of which additional sources to use for each urban area was based on emissions inventories, on wind trajectory analysis of ambient measurements, and on factor analysis results. With the exception of the “urban/steel dust” source signatures for southeast Chicago and several fugitive dust profiles from Granite City, the emission source signatures used in the CMB model were not derived from the study areas. Sources are often site specific, and thus source apportionment numbers based on generic source profiles should be viewed with caution. Combinations of these sources are then analyzed until a statistically reasonable fit is obtained that reflects known major sources. The distribution of toxic elements among the source categories can then be calculated. Generally, all elements detectable in the samples and present in the source profiles were used as fitting speciesin CMB analysis. There are many uncertainties in CMB analysis, SO these results should be regarded as only an approximation of the actual distribution of toxic elements among sources. A number of sources identified in the urban areas studied were labeled as fugitive dust. Fugitive source profiles can be very site specific. Dust samples were collected from roads within the Chicago and Granite City areas. Analytical results from these samples were used in this report as an urban fugitive dust profile (urban/steel dust). The concentrations of some toxic elements (Mn, Cr) are up to six times higher in street dust collected in southeast Chicago compared to street dust that is unaffected by steel mills (12). It is apparent that the use of paved and unpaved road dust profiles provided by the United States Environmental Protection Agency 1984 Source Library (23) cannot adequately account for the concentrations of metal in the urban dust in industrial areas. Glover et al. (9) showed that the use of locally generated fugitive dust source signatures can greatly improve the results of CMB analysis. The average percent contribution of identified sources to the toxic elements is presented for southeast Chicago and East St. Louis (Tables VI and VII). The results are reported as a percent of the calculated total source contribution. Model statistics generally show good general agreement between measured and calculated elemental concentrations. Sources identified in the model runs that 2508
Envlron. Sci. Technol., Vol. 27, No. 12, 1993
Table VI. Average Percent Source Apportionment for Southeast Chicago
elementa V (0.42) Cr (0.94) Mn (0.59) Ni (0.96) Cu (0.86) Zn (0.98) As (1.02) Se (0.74) Pb (0.94)
Fine Particles regional coke background combustion steel dust incinerator 28.6 7.1 6.3 70.6 11.4 6.3 46.8 37.3 0.1
elementa
urban dust
V (0.71) Cr (1.03) Mn (0.85) Ni (1.11) Cu (1.04) Zn (1.00) P b (0.79)
63.8 35.7 71.3 55.2 43.6 93.3 59.9
22.1 5.4 0 6.7
26.7 78.8 92.5 10.8 2.9 2.5 0 0 3.8
1.2
0.2 19.0 61.5 0 Coarse Particles blast furnace 30.0 42.9 2.1 38.4 20.5 0 5.4
23.1 5.6 0 0.6 69.8 4.2 36.1 0 6.5
0 5.4 0.8 12.1 14.8 85.9 0 1.7 89.0
motor vehicle
steel 6.1 21.3 26.5 6.6 35.1 6.6 12.9
0 0 0 0.1
0.3 0.1 22.1
a Value in parentheses is the amount of the element calculated by CMB divided by the amount actually measured.
Table VII. Average Percent Source Apportionment for East St. Louis Fine Particles regional zinc cop er vehicle elementa background combustion smelter smeker steel exhaust V (0.46) Cr (0.96) Mn (1.00) Ni (1.15) cu (0.99) Zn (1.03) Cd (0.36) Pb (0.90)
elementa
V (0.56) Cr (0.59) Mn (1.38) Ni (0.66) c u (1.00) Zn (1.06) Cd (0.33) Pb (0.87)
54.8 29.7 31.9 81.7
35.2 18.8 2.8 6.3
2.1
2.1
13.8 38.3 11.8
0.4 1.5 0.1
crustal dust 15.9 9.0 9.1 17.4 0.4 0.2 0 0.1
1.3 1.3
0
2.2
0
0
0.9 85.6 52.5 1.8
5.8 8.4 95.0 0.2 3.5 0.3
Coarse Particles urban zinc smelter dust 81.1 85.9 90.1 72.2 7.5 7.5 5.2 95.4
1.3 5.3 1.0 0 5.1 92.6 93.9 4.6
7.0 43.6 62.5 1.7
0
0.1
0.2 1.5 5.0 84.7
0.7 0
0.8
1.3 1.0 2.0
copper smelter 0 0.2 0 10.2 87.5 0 8.9 0.1
a Value in parentheses is the amount of the element calculated by CMB divided by the amount actually measured.
have no toxic elements (e.g., crustal and sulfate) are not listed. The “goodness of model fit” for each of the toxic elements is shown as a ratio between calculated and measured average ambient concentrations. Generally, good agreement between measured and calculated concentration is shown for the toxic metals. The poorer fit found for some of the elements is probably due to both ambient concentrations and/or source concentrations at or near detection limits and also from inaccuracies in the generic source profiles used. In southeast Chicago, it is apparent that most of the Cr, Mn, Cu, and As on fine particles comes from steel-related stack and fugitive sources. Combustion is important for Se. The attribution of most of the Zn and Pb in fine
particles to an incinerator source is probably inaccurate. Both wind trajectory analysis and CMB for the coarse particles indicate that coarse particle Cr and Mn in southeast Chicago are due primarily to steel-related fugitive emissions. Wind trajectory analysis also shows that steel-related emissions are also an important source of airborne Zn and Pb. This is not reflected in available source profiles for steel-related emissions and leads to unreliable model estimates for these elements. Total mass tends to be underpredicted in fine particles compared to the individual elements. This probably indicates that the source data for carbon are inaccurate. In East St. Louis, toxic elements were also attributed to a wide variety of sources (Table VII). The CMB model indicated that steel-related emissions are responsible for most of the Cr and Mn. Smelters contribute most of the airborne Cu, Zn, and Cd. Finally, the CMB indicates that auto emissions are responsible for most of the Pb. The relatively poor model prediction for Cd could be because a major source of this element has not been identified. However, the close association of Cd and Zn in the rest of this work makes it more likely that the literature value for Cd in Zn smelter emissions is much lower than the true value for East St. Louis. Likewise, P b in fine particles, although well-predicted by CMB, is not attributed to smelter emissions as suggested by wind trajectory analysis. Results from several individual filters from both urban areas were also subjected to CMB analysis. These filters corresponded to episodes when concentrations of PM-10 and many trace metals were relatively high in the study areas. High concentrations of airborne trace metals occur at Bright School in southeast Chicago when winds are from the southeast. Steel and urban dust account for most of the Cr and Mn in Chicago under these conditions. Fugitive emissions of these elements from steel mill areas are the most important sources. As with the average data, lead and zinc are not well-predicted, indicating that the source signature data for these elements are inaccurate. At the East St. Louis site, high trace element levels occur when winds are from the southwest. CMB analyses of individual filters taken during southwest winds are almost identical to averageresults. The similarity between these single filter CMBs and average CMB results indicates that the sources associated with episodes of high trace element pollution play an important role in average airborne trace element concentrations in the study areas. Summary and Conclusions Source identification of toxic trace elements in a complex urban airshed requires the use of a variety of sampling and analysis techniques. The approach taken here has been to compile a data base that includes meteorological information and chemical data for 29 trace elements. Information from wind trajectory and factor analysis and existing emissions inventories is used to identify sources. Once the sources are known, chemical mass balance analysis apportions the airborne pollutants to the various sources. Much of the time, concentrations of the toxic trace elements in urban air approach regional background levels, However, there are periodic episodes during which concentrations of many airborne elements are up to 100times higher than those measured at arural site. These episodes generally occur when the wind blows from the direction of a major point source during the sampling period. Episodes can also occur during periods of low wind speed when mixing is reduced. Since the urban sampling sites
used in this study are often downwind of large point sources, such sources have a significant impact on the average concentrations of many metals in urban air. In contrast, the total mass of inhalable airborne particles is only slightly higher at the urban sites than at the rural site. The average concentrations of most elements measured in this study are similar to those measured in other urban areas around the country. There is considerble variation between urban areas in the concentrations of some toxic elements that appear unrelated to total PM-10 mass (Tables 1-111). Southeast Chicago and East St. Louis have nearly the same average concentration of PM-10. However, from 3-4 times more Mn, Cr, Fe, and Cl are found on airborne particles collected in southeast Chicago as compared to East St. Louis. In turn, concentrations of airborne Cu and Cd in East St. Louis are 5-10 times higher than those in southeast Chicago. Wind trajectory analysis and the composition of airborne particles indicate that steel mills and smelters are major sources of toxic trace elements in Chicago and East St. Louis, respectively. Receptor modeling confirms that in Chicago airborne Mn and Cr come from steel-related industrial process and fugitive sources. The other toxic trace elements that can be apportioned come from avariety of sources including incinerators, oil and coal combustion, resuspended soil, and vehicle emissions. In East St. Louis, the smelters are the primary sources of Zn, Cd, and Cu. Steel emissions are the major source of Mn and Cr. P b comes primarily from auto emissions at this site; however, the smelters are probably also important Pb sources, although this could not be verified by the CMB model due to inadequacies in available source signatures. This variability in the concentrations of toxic elements and their independence of PM-10 mass call into question the strategy of controlling total inhalable particle mass as a surrogate for toxic materials carried on the particles. Some sources may have emissions that are highly enriched in toxic metals but which do not have a significant impact on the total mass of inhalable airborne particles. In these cases, control of traditional PM-10 sources may not do much to improve urban air quality in terms of the toxic trace elements. Literature Cited (1) Nriagu, J. 0. Nature 1989, 338, 47-49. (2) Cole, K. L.; Engstrom, D. R.; Futyma, R. P.; Stottlemyer, R. Environ. Sci. Technol. 1990, 24, 543-549. (3) Galloway, J. N.; Thornton, J. D.; Norton, S. A.; Volchok, H. L.; McLean, R. A. N. Atmos. Environ. 1982,16,1677-1700. (4) Hall, J. V.; Winer, A. M.; Kleinman, M. T.; Lurman, F. W.; Brajer, V.; Colome, S. D. Science 1992, 255, 812-817. (5) Thomson, V. E.; Jones, A.; Haemisegger, E.; Steigerwald, B. J. Air Pollut. Control Assoc. 1985, 35, 535-540. (6) Eisenreich, S. J.; Metzer, N. A.; Urban, N. R.; Robbins, J. A. Environ. Sci. Technol. 1986, 20, 171-174. (7) Saltzman, B. E.; Cholak, J.; Schafer, L. J.; Yeager, D. W.; Meiners, B. G.; Svetlik, J. Enuiron. Sci. Technol. 1985,19, 328-333. (8) Olmez, I.; Sheffield, A. E.; Gordon, G. E.; Houck, J. E.; Pritchett, L. C.; Cooper, J. A.; Dzubay, T. G.; Bennett, R. L. J. Air Pollut. Control Assoc. 1988, 38, 1392-1402. (9) Glover, D. M.; Hopke, P. K.; Vermette, S. J.; Landsberger, S.; D’Auben, D. R. J . Air Waste Manage. Assoc. 1991,41, 294-305. (10) Rahn, K. A.; Lowenthal, D. H. Science 1985,228,275-284. (11) Emison, G. A. In PM-10 Implementation of Standards; C. V. Mathai, Stonefield, D. H., Eds.; Air and Waste Management Association: Pittsburgh, 1988; pp 2-7. Envlron. Scl. Technol., Vol. 27, No. 12, IS93
2509
(12) Sweet, C. W.; Vermette, S. J.; Gatz, D. F. Toric Trace Elements in Urban Air in Illinois; Technical Report No. RR-042; Hazardous Waste Research and Information Center: Champaign, IL, 1990. (13) Sweet, C. W.; Vermette, S. J. Environ. Sci. Technol. 1992, 26, 173-178. (14) Rodes, C. F.; Holland, D. M.; Purdue, L. J.; Rehme, K. A. J.Air Pollut. Control Assoc. 1985, 35, 345-354. (15) Conley, F. W.; Buckman, F. D. Presented a t the 80th Annual Meeting of the Air Pollution Control Association, New York, June 1987; Paper 87-63.1. (16) Illinois Environmental Protection Agency. Illinois 1988 Annual Air Quality Report; IEPA/APC/89-166, Division of Air Pollution Monitoring, IEPA: Springfield, IL, 1989. (17) Kolaz, D. J.; Sweitzer, T. A,; Bomke, D. L. Presented at the 79th Annual Meeting, Air Pollution Control Association, Minneapolis, MN, June 1986; Paper 86-57.4. (18) Dzubay, T. G.; Stevens, R. K.; Gordon, G. E.; Olmez, I.; Sheffield, A. E. Environ. Sci. Technol. 1988, 22, 46-52. (19) Gatz, D. F. Atmos. Environ. 1975, 9, 1-18.
2510
Envlron. Scl. Technol., Vol. 27, No. 12, 1993
(20) Chang, S. N.; Hopke, P. K.; Gordon, G. E.; Rheingrover, S. W. Aerosol Sci. Technol. 1988,8, 63-80. (21) Richman, M. B. J. Climatol. 1986, 6, 293-335. (22) Axetell, K.; Watson, J. G. CMB User's Manual (Version 6.0); Receptor Model Technical Series; EPA-450/4-83-01R. U S . Environmental Protection Agency, Office of Air Quality Planning and Standards, U.S. Government Printing Office: Washington, DC, 1987; Vol. 11. (23) United States Environmental Protection Agency. Receptor Model Source CompositionLibrary;EPA 45014-85-002; U S . E P A Research Triangle Park, NC, 1984. (24) Scheff, P. A.; Wadden, R. A.; Allen, R. J. Environ. Sci. Technol. 1984,18,923-931. (25) Hopke, P. K. Receptor Modeling in Environmental Chemistry; John Wiley & Sons: New York, 1985.
Received f o r review February 25, 1993. Revised manuscript received July 6, 1993. Accepted July 15, 1993." Abstract published in Advance ACS Abstracts, September 1, 1993.
