Environ. Sci. Technol. 1999, 33, 1970-1978

Toxic Air Contaminants in Porto Alegre, Brazil ERIC GROSJEAN,† REINHOLD A. RASMUSSEN,‡ AND D A N I E L G R O S J E A N * ,† DGA, Inc., 4526 Telephone Road, Suite 205, Ventura, California 93003, and Department of Environmental Science and Engineering, Oregon Graduate Institute of Science and Technology, P.O. Box 9100, Portland, Oregon 97291-1000

Information on ambient levels of toxic air contaminants is a critical component of programs aimed at regulating air emissions to protect public health. Yet, this information is not available in many countries. Toxic air contaminants have been measured near a busy highway in downtown Porto Alegre, Brazil, from 3/20/96 to 4/16/97. Of the 42 compounds listed in U.S. EPA Method TO-14, 21 were not detected (concentrations < 0.1 ppb), 3 were detected in only a few samples (CHCl3, HClCdCCl2, and p-dichlorobenzene), and 3 were present at concentrations typical of background locations (CH3Cl, CH3CCl3, and CFC-113). Comparisons of urban/background location concentration ratios and of urban concentrations vs those of CO indicate no local emissions for CCl4, small and variable sources for CFC-11 and CFC-12, local sources other than vehicles for CH2Cl2 and Cl2CdCCl2, and vehicles as the major source of 1,3-butadiene, styrene, and aromatic hydrocarbons. Ambient concentrations of 1,3-butadiene and aromatic hydrocarbons were well correlated to those of benzene (R ) 0.92-0.98) and those of CO (R ) 0.88-0.97). These correlations are used to estimate mid-1996 vehicle emission rates, e.g., (197 ( 50) × 103 kg/year for 1,3-butadiene. A comprehensive comparison is made of our results (halogenated compounds measured in Porto Alegre, halogenated compounds not detected in Porto Alegre, and aromatic hydrocarbons with focus on the toluene/ benzene, (m + p)-xylene/benzene, and (m + p)-xylene/ toluene ratios) with literature data for background and urban locations.

Introduction Regulatory agencies in many countries continue to explore means to reduce public exposure to airborne toxic chemicals (1, 2). The identification of toxic air contaminants and the measurement of their ambient concentrations are critically important components of control programs aimed at regulating air emissions of toxic chemicals in order to protect public health. For several countries, including the United States, a large body of information is available regarding the nature and ambient concentrations of toxic air contaminants (3-9). This is particularly true for the aromatic hydrocarbons (1, 5, 8-25), which have received attention as toxic air contaminants and as important precursors to the formation * Corresponding author phone: (805)644-0125; fax: (805)644-0142; e-mail: [email protected]. † DGA, Inc. ‡ Oregon Graduate Institute of Science and Technology. 1970

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

of ozone (1, 26, 27) and of secondary organic aerosols (1, 8, 27-30). However, information on ambient levels of toxic air contaminants is either very limited or not available for many other countries. We report ambient concentrations of toxic air contaminants in Porto Alegre, Brazil, for the ca. 1 year period March 20, 1996 to April 16, 1997. To our knowledge, no comprehensive survey of toxic air contaminants has been carried out in South America prior to this work. Porto Alegre is of particular interest, due to a combination of vehicle fuels that is unique in the world. Of the ca. 600 000 vehicles registered in the Porto Alegre metropolitan area, none is running on gasoline alone. The three vehicle fuels used in Porto Alegre are ethanol (17% of total vehicles; the ethanol fuel is hydrated ethanol (5% water) and contains e5% gasoline), diesel (9% of total vehicles, mostly buses and trucks), and a mixture of 85 vol % gasoline and 15 vol % methyl-tert-butyl-ether, hereafter referred to as MTBE (74% of total vehicles). The diesel fuel has a low sulfur content (0.5%) and contains no ethanol and no MTBE. The light-duty vehicles that run on a mixture of 85% gasoline and 15% MTBE in Porto Alegre are an exception in Brazil, where ethanol (22 vol %) is used instead of MTBE as the oxygenated additive to gasoline. Vehicle emissions have a significant impact on air quality in the city of Porto Alegre, where mobile sources account for 98.9% of total CO emissions (31). In this study, we have analyzed ambient air samples collected in Porto Alegre for the 42 toxic air contaminants listed in the U.S. Environmental Protection Agency Compendium Method TO-14 (32): 1,3-butadiene, benzene, C7C9 aromatic hydrocarbons, styrene, the halocarbons CFC11, CFC-12, CFC-113, and CFC-114, and the saturated aliphatic, unsaturated aliphatic, and aromatic halogenated hydrocarbons that are listed in the Results and Discussion. The objectives of this article are to report ambient concentrations of toxic air contaminants in Porto Alegre from March 1996 to April 1997, to describe temporal variations in ambient concentrations, to investigate major sources, i.e., mobile and/ or stationary, to estimate emission rates, and to compare the results to literature data for other urban and nonurban locations. Most samples were collected near a major highway at a downtown location, during the morning period of high vehicle traffic. Thus, our results should strongly reflect vehicle emissions, and it was hoped that in this way emission rates could be estimated, relative to those of CO, for those toxic contaminants that are generally emitted by vehicles, e.g., 1,3-butadiene and benzene. It was also hoped that under these conditions those toxic air contaminants that do not correlate with CO could be more convincingly shown to originate from sources other than vehicles, i.e., stationary sources, for which no information is available in Porto Alegre. Ambient concentrations and emission rates of formaldehyde and acetaldehyde, which have received much attention as toxic air contaminants (7, 33), have also been measured in Porto Alegre and will be discussed in future work. The results presented here have been obtained as part of a larger project that combines field measurements and laboratory and computer kinetic modeling studies of the impact of vehicles on air quality in Porto Alegre (34-37).

Experimental Methods Samples of ambient air were collected in Porto Alegre (latitude 30° 02′ S, longitude 51° 14′ W, elevation 10 m, population ) 1.5 million for the city and 3.4 million for the metropolitan area) using 850 mL SUMMA electropolished stainless steel 10.1021/es980578x CCC: $18.00

 1999 American Chemical Society Published on Web 04/30/1999

canisters. Of the 48 samples collected during the ca. 1 year period March 20, 1996-April 16, 1997, 46 samples were collected at downtown locations and 2 samples were collected at a “background” location ca. 20 km southeast of Porto Alegre and upwind of the city on the 2 days of sample collection. Of the 46 urban samples, 44 were collected at the same location on a 30 m wide divider strip in the middle of a major highway that brings a large portion of the vehicle trafficcars, buses, and trucks-to and from the center of Porto Alegre. This sampling location was a few blocks from the city center and near a major bus terminal (Rodoviaria). The two other urban samples were also collected near busy streets in the downtown area, one at the main market in the city center and the other at a bus station. The sampling duration was 1 min for all samples. At the Rodoviaria, samples were collected ca. once a week or once every two weeks. In addition, two sets of samples were collected on 3 consecutive days, April 8-10 and April 14-16, 1997. All urban samples were collected during the morning period of high vehicle traffic as indicated by continuous measurements of CO. Samples were analyzed by gas chromatography with flame ionization detection (GC-FID) and by gas chromatographymass spectrometry (GC-MS). The two analytical methods have been described elsewhere (34-36). Comprehensive validation studies of the canister sampling, GC-FID, and GCMS protocols have been reported previously (38-43). All samples were analyzed by GC-FID, yielding ambient concentration data for ca. 66 compounds including those we discuss in this article: 1,3-butadiene, styrene, benzene, toluene, (m + p)-xylene (the two isomers could not be resolved), o-xylene, ethylbenzene, p-ethyltoluene, 1,3,5trimethylbenzene, and 1,2,4-trimethylbenzene + sec-butylbenzene (the two compounds could not be resolved; GC-MS analysis of two samples indicated that s-butylbenzene accounted for e1% of the unresolved peak). Of the 48 samples analyzed by GC-FID, 23 samples (all from urban locations) were analyzed by GC-MS for the 42 compounds listed in U.S. EPA Compendium Method TO-14 (32) and 17 samples (including the two samples collected at the background location) were analyzed by GC-MS for five compounds: CCl3F (CFC-11), CCl2F2 (CFC-12), CH2Cl2 (methylene chloride), CCl4 (carbon tetrachloride), and Cl2CdCCl2 (tetrachlorethylene, PCE). The GC-MS detection limit for all halogenated compounds was 0.1 ppbv. Wind direction, wind speed, and other meteorological data were recorded throughout the study. On the days and at the times of sample collection, wind directions (percent of total) were 24% SE, 18% E, 16% S, 16% N, 10.5 NW, 8% W, 5% NE, and 2.5% SW. Wind speeds (percentage of total) were 16, 8, 24, 16, 13, 5, 5, 2.5, 10.5, and 0% for 0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, and >4 m/s. These percentages were very similar to those for the entire 1 year period studied.

Results and Discussion Ambient concentrations of toxic air contaminants have been measured in Porto Alegre from March 20, 1996 to April 16, 1997. Individual results for the 48 samples (2 at a background location and 46 at downtown locations during the morning period of high vehicle traffic) are not listed due to space limitations and are available from the authors and as part of a report (44). Individual data are given in Figure 1 and Figure 2 using six compounds as examples. A summary of the data is given in Table 1, which includes for each compound the number of measurements, the number of samples in which the compound was detected, the range of concentrations measured during the 1 year period studied, the average concentration and the corresponding standard deviation, and the ratio of the average concentration measured at the downtown locations to that measured at the background location.

FIGURE 1. Ambient concentrations of methylene chloride (top), CFC-11 (middle), and CFC-12 (bottom) in downtown Porto Alegre, March 20, 1996 to April 16, 1997. Compounds Not Detected and Compounds Detected in a Few Samples. With a detection limit of 0.1 ppbv, 21 of the 42 compounds listed in Compendium Method TO-14 were not detected: ClF2C-CF2Cl (CFC-114), vinyl chloride, methyl bromide, ethyl chloride, vinylidene chloride, allyl chloride, 1,1-dichloroethane, cis-1,2-dichloroethylene, 1,2-dichloroethane, 1,2-dichloropropane, cis-1,3-dichloropropene, trans1,3-dichloropropene, 1,1,2-trichloroethane, 1,2-dibromoethane, 1,1,2,2-tetrachloroethane, chlorobenzene, benzyl chloride, o-dichlorobenzene, m-dichlorobenzene, 1,2,4trichlorobenzene, and hexachloro-1,3-butadiene. Thus, there appears to be no large mobile and/or stationary source for these 21 halogenated compounds in Porto Alegre. For the most reactive of these compounds, i.e., the six chlorinated alkenes, emissions from sources, if any, may be offset in part by removal from the atmosphere due to reactions with the hydroxyl radical and with ozone (7). Three compounds were detected in only a few samples: p-dichlorobenzene (0.3 ppb in one sample), chloroform (0.1 ppb in four samples), and trichloroethylene (0.1, 0.6 and 1.2 ppb in four, one, and one samples, respectively). For these three compounds, mobile and/or stationary sources may exist in Porto Alegre, but our results are too limited to draw conclusions. Literature data for ambient concentrations of VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1971

FIGURE 2. Scatterplots of ambient concentrations of (m + p)-xylene (top), 1,3-butadiene (middle), and tetrachloroethylene (bottom) vs those of CO. these three halogenated compounds at other locations are discussed later in this section. Comparison of Data for Urban and Background Locations. If ambient levels of the compounds studied reflect emissions from the metropolitan Porto Alegre area, concentrations measured at the downtown locations should be higher than those measured at the background location. As the results in Table 1 indicate, this was the case for methylene chloride, tetrachloroethylene, 1,3-butadiene, styrene, and all aromatic hydrocarbons. For example, the average benzene concentration in Porto Alegre was ca. 20 times higher than that measured at the background location. This was also the case for carbon monoxide, acetylene, MTBE, and ethanol, which are indicators of vehicle emissions and for which the urban/background concentration ratios given in Table 1 were comparable in magnitude to those of the toxic air contaminants. Urban concentrations of carbon tetrachloride were identical to those measured at the background location. This observation suggests that there is no major source of CCl4 in Porto Alegre. Ambient levels of CCl4 measured in this study, i.e., 0.1 ppbv, are consistent with literature data for CCl4 at background locations (3-9, 43, 45, 46). Urban concentrations of CFC-11 and CFC-12 were similar to those measured at the 1972

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

background location, and the range of concentrations measured in the urban samples suggests that there may be small and variable sources of CFC-11 and CFC-12 in Porto Alegre. Inferences regarding urban sources of chloroform, trichloroethylene, p-dichlorobenzene, methylchloroform, methylene chloride, and CFC-113 in Porto Alegre cannot be made from examination of the urban/background concentration ratio (these six compounds were not measured at the background location). As noted in the preceding section, chloroform, trichloroethylene, and p-dichlorobenzene were detected in only a few urban samples and therefore do not have significant urban sources. For methylchloroform, methyl chloride, and CFC-113, sources can be estimated by comparison with literature data for other background and urban locations; see discussion later in this section. Temporal Variations in Ambient Concentrations. Scatterplots of ambient concentrations vs sampling date revealed three types of patterns which are illustrated in Figure 1 using CH2Cl2, CFC-11, and CFC-12 as examples. Ambient levels of CCl4, CFC-11, CFC-113, and CH3CCl3 were essentially constant from March 1996 to April 1997. Ambient levels of CFC-12 were constant from March 1996 to November 1996 and then decreased from December 1996 to March 1997. Concentrations of 1,3-butadiene, styrene, CH2Cl2, CH3Cl, PCE, and aromatic hydrocarbons did not exhibit seasonal variations. For these compounds the amplitudes of seasonal variations, if any, were smaller than day-to-day differences in concentrations, which were recorded during two periods of 3 consecutive days each. Day-to-day variations in ambient concentrations reflect day-to-day changes in meteorology and in vehicle traffic density. Diurnal variations, although not investigated for the toxic air contaminants, were well documented from continuous measurements of CO during the 1 year period studied (34, 35). Temporal variations in ambient concentrations of the aromatic hydrocarbons followed the same pattern, thus indicating common sources for these compounds. This was also the case for 1,3-butadiene, and, with more scatter, for styrene. Although scatterplots contained a few outliers, as would be expected from changes in vehicle fleet composition, etc., from one sample to the next, they indicated that the aromatic hydrocarbon/benzene and 1,3-butadiene/benzene ambient concentration ratios were essentially constant during the 1 year period studied. Linear regression of ambient concentrations of 1,3-butadiene and the alkylbenzenes vs those of benzene yielded correlation coefficients ranging from 0.92 to 0.98. Sources of Toxic Air Contaminants. For the toxic air contaminants whose concentrations in Porto Alegre are significantly higher than those measured outside the city, ambient levels in Porto Alegre may reflect vehicle emissions, stationary source emissions, or both. There is no information on stationary source emissions in Porto Alegre for the compounds of interest. For vehicle exhaust emissions, several compounds are available that can be used as indicators. These indicators include CO, for which vehicles account for 98.9% of the total emissions in the city of Porto Alegre (31, 34), and acetylene, whose ambient levels in Porto Alegre are closely correlated to those of CO (34, 35). To examine the contribution of vehicle emissions to ambient levels of toxic air contaminants, we constructed scatterplots of ambient concentrations of these compounds vs those of CO (see examples in Figure 2). These scatterplots indicated good correlation with CO for all aromatics and for 1,3-butadiene, no correlation with CO for CCl4, CH3CCl3, CFC11, CFC-113, and CH3Cl, and poor correlation with CO for CH2Cl2, PCE, and CFC-12. Scatterplots such as that shown in Figure 2 for (m + p)-xylene indicate that vehicles are a major source of 1,3-butadiene and aromatic hydrocarbons. Vehicles may also be a major source for styrene, although

TABLE 1. Data Summary urban locations (b) concentration, ppb averagee

compounda

Nd

Dd

range

1,3-butadiene styrene benzene toluene o-xylene (m + p)-xylene ethylbenzene p-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene carbon tetrachloride CFC-11 CFC-12 methylene chloride tetrachloroethylene (PCE) methyl chloroform methyl chloride CFC-113 chloroform trichlorethylene p-dichlorobenzene carbon monoxide acetylene ethanol MTBE

46 46 46 46 46 46 46 46 46 46 38 38 38 38 38 23 23 23 23 23 23 46 46 46 46

46 19 46 46 46 46 46 46 46 46 38 38 38 24 29 15 12 16 4 6 1 46 46 44 46

0.2-3.6 0.1-1.5 0.8-9.5 1.1-26.7 0.3-3.2 0.7-8.3 0.3-4.6 0.4-2.9 0.2-1.8 0.4-4.0 0.1-0.2 0.3-3.30 0.1-1.00 0.1-2.40 0.1-4.6 0.1-0.3 0.1-0.3 0.1-1.30 0.1 0.1-1.2 0.3 504-9186 5.5-109.4 0.4-68.2 0.2-17.1

1.21 ( 0.76 0.58 ( 0.48 3.71 ( 2.17 5.53 ( 4.47 1.27 ( 0.76 3.26 ( 1.96 1.33 ( 0.94 1.23 ( 0.70 0.65 ( 0.36 1.60 ( 0.90 0.103 ( 0.016 0.492 ( 0.482 0.411 ( 0.214 0.513 ( 0.586 1.06 ( 1.21 0.120 ( 0.056 0.150 ( 0.067 0.194 ( 0.298 0.367 ( 0.455 3240 ( 1980 37.7 ( 23.4 11.9 ( 13.2 6.4 ( 4.3

concn at background location, ppbc,e

concn ratio, urban/background

0.045 ( 0.000 64

a Not included are the 21 halogenated compounds listed in the text (see Results and Discussion) that were not detected (detection limit ) 0.1 ppb) in any of the 23 urban samples analyzed by GC-MS for Compendium Method TO-14 toxic contaminants. b Three downtown locations, most samples collected during morning period of high vehicle traffic. c Two samples. d N ) number of measurements, D ) number of observations with a detection limit of 0.1 ppb. e Uncertainty of (1 standard deviation. f NA ) not analyzed.

TABLE 2. Linear Least Squares Regression of Ambient Concentrations of Toxic Air Contaminants vs Those of CO and Estimated Toxic Air Contaminant Emission Rates compound

103 × slopea,b

interceptb,c

R

emission rate, metric tonnes per yeard

1,3-butadiene benzene toluene ethylbenzene (m + p)-xylene o-xylene p-ethyltoluene 1,3,5-trimethylbenzene 1,2,4-trimethylbenzene styrene

0.333 ( 0.015 1.061 ( 0.041 1.480 ( 0.106 0.397 ( 0.033 0.912 ( 0.045 0.359 ( 0.018 0.337 ( 0.024 0.153 ( 0.011 0.415 ( 0.021 0.147 ( 0.056

0.047 ( 0.054 0.135 ( 0.146 0.369 ( 0.392 0.021 ( 0.117 0.254 ( 0.157 0.096 ( 0.063 0.149 ( 0.086 0.142 ( 0.037 0.235 ( 0.074 0.065 ( 0.220

0.964 0.971 0.905 0.878 0.953 0.951 0.907 0.912 0.950 0.535

197 ( 50 907 ( 229 1491 ( 388 461 ( 121 1059 ( 270 417 ( 106 443 ( 115 201 ( 52 545 ( 139 (167 ( 76)e

a Forty-six samples collected in downtown Porto Alegre from March 20, 1996 to April 16, 1997, with a few outliers omitted from regression analysis after examination of scatterplots for entire data sets. Concentration units are ppb for CO and for all other compounds. b Uncertainty of (1 standard deviation. c Units: ppb. d From CO emission rate and regression parameters converted to mass basis; see text for stated uncertainty. e Estimate only, large uncertainty due to scatter; see low value of R.

the scatterplot of ambient styrene vs ambient CO exhibits substantial scatter. Methylene chloride, CFC-12, and PCE show weak correlations with CO (R ) 0.51, 0.49, and 0.45, respectively), indicating that they may be emitted in part by vehicles (CFC-12 is commonly used in automobile air conditioning units). However, the data scatter precludes any estimate of vehicle-related emissions. The scatterplots for PCE (Figure 2) and CH2Cl2 (not shown) indicate that these two compounds have sources other than vehicle emissions. These sources in Porto Alegre are not known at the present time. Estimated Vehicle Emissions of 1,3-Butadiene and Aromatic Hydrocarbons. For toxic air contaminants that are emitted by vehicles, it is possible to estimate emission rates using the emission rate of CO and the compound/CO ambient concentration ratio. This approach involves a

number of assumptions which have been discussed elsewhere (34). To estimate toxic air contaminant emission rates, we examined scatterplots of ambient concentrations of the compound of interest vs those of CO. We excluded statistical outliers (from none to four for each compound), and we calculated linear least squares regression parameters. The results are listed in Table 2 and, except for styrene, indicate high correlation with CO (R ) 0.88-0.97) with near-zero intercepts that are much smaller than the average hydrocarbon concentrations. We then multiplied the slopes given in Table 2 by the mid-1996 Porto Alegre CO emission rate, i.e., 306 700 metric tonnes per year (34). This yielded the mid-1996 estimated emission rates that are listed in Table 2, e.g. 197 ( 50 metric tonnes per year for 1,3-butadiene. The related uncertainties are those associated with the slopes of VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1973

TABLE 3. Literature Data for Ambient Concentrations of the 21 Halogenated Compounds That Were Not Detected in Porto Alegre concentration, ppta background locations compound CFC-114 vinyl chloride methyl bromide ethyl chloride vinylidene chloride cis-1,2-dichloroethylene allyl chloride 1,1-dichloroethane 1,2-dichloroethane 1,2-dichloropropane cis-1,3-dichloropropene trans-1,3-dichloropropene 1,2-dibromoethane 1,1,2-trichlorethane

SHb 10

NHb 12 20 9.9

40