Survey of volatile organic compounds associated with automotive ...

Atmospheric Environment 35 (2001) 4017–4031

Survey of volatile organic compounds associated with automotive emissions in the urban airshed of Sa* o Paulo, Brazil ! a, Joachim D. Pleila,*, Thomas A. Hartlagea, Maribel Colon M. Lucia Guardanib, M. Helena Martinsb a

National Exposure Research Laboratory, US Environmental Protection Agency, MD-44, Research Triangle Park, NC 27711, USA b * Paulo, SP, Brazil Companhia de Tecnologia de Saneamento Ambiental (CETESB), Sao Received 1 September 2000; accepted 11 March 2001

Abstract The Metropolitan Region of S*ao Paulo (MRSP), Brazil, is one of the largest metropolitan areas in the world (population 17 million, approx.) and relies heavily on alcohol-based fuels for automobiles. It is estimated that about 40% of the total volume of fuel is ethanol with some vehicles using pure ethanol and others a gasoline/ethanol blend. As such, S*ao Paulo is an excellent example of an oxygenates-dominated airshed of mobile sources and is most likely indicative of the future in heavily populated areas in the US such as Los Angeles where ‘‘oxy-fuels’’ are becoming an important replacement for the conventional pure petroleum-based fuels. In this work, we surveyed the ambient air to identify and quantify the organic compounds associated with the evaporative and exhaust emissions of these fuels and to begin to understand the potential for human exposure. Because this was an initial test without detailed prior knowledge of the airshed of the area, we applied two different air sampling methods for various time periods to assess the ambient concentrations of a variety of polar and nonpolar volatile organic compounds (VOCs). For quality assurance (QA), we collected all the samples in duplicate (whole-air samples in Summa canisters and adsorbent-based samples on Perkin-Elmer Air Toxics tubes) at various flow rates to test performance. All samples were collected over identical time frames, typically for 1-, 2-, and 4-h periods per day at six different locations over a period of 1 week. Overall S*ao Paulo results demonstrate that mean concentrations of single-ring aromatics are 2–3 times higher, volatile aldehydes are 5–10 times higher, and simple alcohols 10–100 times higher as compared to results of a recent study performed by EPA in the Los Angeles basin. C4–C11 n-alkanes were only slightly elevated in S*ao Paulo. Published by Elsevier Science Ltd. Keywords: Aldehydes; VOCs; Aromatics; Ethanol fuel; Automobile emissions; Summa canisters; Adsorbent tubes

1. Introduction The US Environmental Protection Agency’s (EPA) Office of Research and Development (ORD), EPA’s Office of International Activities (OIA), and the Companhia de Tecnologia de Saneamento Ambiental (CETESB) of S*ao Paulo, SP, Brazil, joined efforts to study volatile organic compounds (VOCs) in urban air *Corresponding author. Tel.: +1-919-541-4680; fax: +1-919541-3527. E-mail addresses: [email protected] (J.D. Pleil). 1352-2310/01/$ - see front matter Published by Elsevier Science Ltd. PII: S 1 3 5 2 - 2 3 1 0 ( 0 1 ) 0 0 1 7 8 - 9

of the Metropolitan Region of S*ao Paulo (MRSP), Brazil. In addition to fostering international collaboration, this work had the dual purpose of estimating typical ambient VOC levels in a large city dominated by alcohol-fueled vehicles and determining what type of ambient measurement methodology is most applicable to the S*ao Paulo ambient air mixture. The air quality of the MRSP is affected by a complex system of mobile and stationary sources mixing in a topographically diverse area. The urban area comprises 8000 km2 ranging in elevation from 650 to 1200 m. The metropolitan area is traversed by a river just to the north

4018

M. Colo!n et al. / Atmospheric Environment 35 (2001) 4017–4031

of the downtown area and another to the west. The metropolitan area sits in a plain that is bounded to the north and west by higher elevations, to the south by two reservoirs, and is about 70 km northwest of the Atlantic Ocean. The estimate for the population of the city is 9.9 million and for the metropolitan region approximately 17 million (CETESB, 1998). As such, S*ao Paulo is one of the biggest cities in the world. The Metropolitan Region of S*ao Paulo has about 60,000 industries. In Brazil, the use of ethanol as a fuel began as a consequence of the National Alcohol Program, a strategic program launched in 1975 as an attempt to mitigate oil import dependence. Neat ethanol and gasohol (a mixture of gasoline with anhydrous ethanol) were introduced as automotive fuels. When Brazil had to decrease ethanol production, an alternative fuel blend was developed using a mixture of 60% ethanol, 33% methanol and 7% gasoline. This blend was the bestknown alternative in terms of air pollution control, fuel supply, and quality; it did not compromise the National Motor Vehicle Pollution Control Program (Szwarc et al., 1991). During our study, however, only a few gas stations in the MRSP had this alternative blend to supply consumers. The fleet was about 6 million registered vehicles. Approximately 71% of all vehicles used a ‘‘gasohol’’ blend containing 24% ethanol, 22% of vehicles used pure ethanol, and the remaining 7% (primarily trucks and buses) used diesel fuel. It is estimated that ethanol constitutes about 40% of all vehicular fuel used by volume. Because traffic congestion is also a major problem in the city, the local government has imposed a driving ban one day during the week based on arbitrarily assigned (Monday, Tuesday, Wednesday, etc.) license plate designators. S*ao Paulo is an excellent example of an oxygenatedcompounds-dominated airshed as caused by mobile sources. This could be an indicator of the future in heavily populated areas in the US such as Los Angeles where ‘‘oxy-fuels’’ are also becoming an important replacement for conventional pure petroleum-based fuels. Though there have been many studies performed with respect to ethanol and other alternate fuels in the US, the huge scale of ethanol use in the S*ao Paulo region is unique. Little is known about the types and mixtures of organic compounds associated with such great amounts of potential evaporative and exhaust emissions, nor is it known how these chemicals interact at higher concentrations in a complex urban airshed. Two different styles of air sampling methods were applied to assess the ambient concentrations of a variety of polar and nonpolar VOCs. We used whole-air sample collection with evacuated stainless-steel canisters and battery-operated samplers, as well as adsorbent tubes with a variety of different battery-operated pumping systems. Each type of sample was collected in duplicate,

and all samples for each method were collected over identical time frames of 1, 2, and 4 h to demonstrate their use in different scenarios. Six varied sites throughout the city were chosen, including an urban scale station in Ibirapuera Park near the center of the city, inside a major tunnel, a site near the domestic airport, and other downtown sites impacted heavily by major highways and stagnant traffic. All sites (except for the tunnel site) were standard CETESB air pollution monitoring stations so we also had access to simultaneous meteorological and criteria pollutant data. Fig. 1 is a map that shows the overall S*ao Paulo region and the individual sampling sites within the city. This project gave EPA an opportunity to test existing methods in a complex scenario and provided CETESB with new (albeit limited) information on the trace organic composition of the airshed in S*ao Paulo. In the long run, this work benefits both the US and Brazil in that this basic information can be used to design and implement better sampling and measurement strategies in both countries.

2. Experimental methods Overall coordination of the sampling sites, transportation, equipment staging, and on-site laboratory support was provided by CETESB personnel. Sampling locations for this study were chosen to coincide with existing CETESB environmental monitoring stations so that we had access to meteorological and criteria pollutant information as well as security for the equipment. The specific canister and adsorbent tube sampling plan was developed by EPA personnel to simultaneously provide information about the organic composition of the airshed and about the performance of the various sampling methods. Field work was conducted as a group effort by the CETESB and EPA collaborators. 2.1. Sampling equipment The whole-air samples were collected with portable, battery-operated personal whole air sampler (PWAS) units that use mass flow control to collect a constant flow of air into an evacuated sampling container. PWAS prototypes (Whitaker et al., 1995) were originally developed by EPA under a research contract with the Research Triangle Institute (Research Triangle Park, NC; contract 68-02-4544) and have since been redesigned as a commercially available package under a Cooperative Research and Development Agreement (CRADA file 0121-95) between EPA’s National Exposure Research Laboratory (Research Triangle Park, NC) and Environmental Supply Corporation (Durham, NC). Sample-collection canisters were stainless steel


4019

Fig. 1. Map of S*ao Paulo city with individual sampling sites.

with an interior surface deactivation based either on the Summa electropolish technique as supplied by SIS, Inc. (Moscow, ID), and Biospherics, Inc. (Hillsboro, OR), or on the Silcosteel fused-silica vapor deposition method as supplied by Restek Corporation (Bellefonte, PA). We used 1- or 1.8-l canisters depending on sampling time with stable 12-ml/min flow. The adsorbent tube samples were collected using two SKC AirChek 2000 pumps, two SKC Pocket pumps (SKC Inc., Eighty Four, PA), and two Bios AirPro Surveyor 2 pumps (Bios International Company, Butler, NJ). The Bios pumps were used at two different flow rates (17.0 and 67.0 ml/min); the SKC pumps used a flow rate of 67.0 ml/min. Air Toxics tubes (Perkin-Elmer, Norwalk, CT, USA) were used during this study. 2.2. Sampling schedule and sites The general plan employed three distinct sampling periods each day starting with 1, then 2, and finally 4 h, with short breaks (approx. 20 min) between them for sample changes. Different sampling times were used to demonstrate and evaluate sampler performance and to cover most of the work day; no attempt was made to interpret the results from individual times. Each day we used a different site. For each sampling period we collected a set of canisters and two to four sets of solid adsorbent tubes depending on availability. Due to logistics, we could collect only 1 and 2 h samples at Dom Pedro Park in the morning, and one set of samples (for 1 h) inside the Tunel 9 de Julho in the afternoon.

The Congonhas samples were collected on a Saturday, all others were collected on weekdays, though traffic was similarly busy regardless of day. A total of 90 samples were collected by the two methods during this study. All samples were transported back to EPA laboratories in Research Triangle Park (RTP), NC, for analysis. Six monitoring sites from different parts of the city of S*ao Paulo were chosen for sampling: Cerqueira Cesar, Ibirapuera Park, Lapa, Congonhas, Dom Pedro Park, and Tunel 9 de Julho (see Fig. 1). The sites and sampling conditions are briefly described below. Cerqueira CesarFa CETESB monitoring site at the ! campus of the Faculdade da Saude Publica. This site was located along a major avenue (Av. Doutor Arnaldo) and across from a cemetery. Traffic at this site was heavy in the morning and late afternoon. Weather was overcast with occasional rain and temperatures in the low 20s1C. Ibirapuera ParkFa CETESB monitoring site at a city park in the middle of the metropolitan area. The weather was sunny with scattered clouds and temperatures in the mid-20s1C. There was no nearby traffic (within 0.3 km). LapaFa CETESB monitoring site at the grounds of a city disposal yard and garden located along a major avenue (Marginal Tiet#e) and the Tiet#e river. Across the river was a large printing company. Traffic at this location was constantly busy with eight lanes, with the strong presence of trucks. The weather was warm and sunny with temperatures in the high 20s1C.

4020


CongonhasFa CETESB monitoring site at a public school yard located in a residential/commercial area along Av. dos Bandeirantes. Traffic was always busy at this location, with the strong presence of trucks. Across from the monitoring site was a gas station, and the Congonhas Airport was within 0.3 km; there was frequent commercial airline traffic over the site. The weather was warm and sunny with temperatures in the high 20s1C and low 30s1C. Dom Pedro ParkFa CETESB monitoring site located in downtown S*ao Paulo. Dom Pedro Park is surrounded by three major avenues; the Tamanduate!ı River runs through the park. The monitoring site is located along do Estado Av. The weather was sunny and cool with temperatures in the mid-20s1C. Tunel 9 de JulhoFa very busy automobile and bus tunnel (no trucks allowed) running under a series of roadways. The sampling location was about 150 m inside the tunnel from the entrance; temperatures were in the low 20s1C. Though wind speed and direction (as well as a variety of other weather and criteria pollutants data) were simultaneously collected by CETESB, these were not used as parameters in our data analyses. Since all sites were within 10 m of very busy roads (except for the park which was within 300 m of a highway), and because the traffic is the dominant VOCs source, long range transport was not considered an important variable. 2.3. Sample preparation and analysis 2.3.1. Canisters Sampling canisters were prepared at EPA laboratories with a cleanup procedure that involved repeated flushing and evacuation at 1001C and a final sealed vacuum of 2.5 10@4 Torr (or better) using an automated Model 960 canister cleaning system (XonTech, Inc., Van Nuys, CA). Once in the field at the CETESB laboratory, each canister was checked prior to use for vacuum integrity using a digital pressure gauge (constructed by ManTech, Inc., RTP, NC). Sampler flow rates were set and checked daily before and after sampling using a digital electronic flow calibrator #21606 (Restek Corporation, Bellefonte, PA). Each ambient sample was transported to the EPA laboratories where it was pressurized with a neutral gas (Scientific Grade Zero Air, National Specialty Gases, Durham, NC), and a dilution factor was calculated based on pre- and post-pressurization absolute pressure. Analyses of canisters were performed via gas chromatography–mass spectrometry (GC–MS) using protocols derived from EPA method TO-14 (Winberry et al., 1989). The analytical instrumentation was fully automated to extract an aliquot (100 ml) from the canister and then cryogenically concentrate the extract at

@1501C and thermally desorb/inject it at 1501C onto a capillary column. The oven temperature program started with a 2-min hold at @501C and then ramped to 2001C at 81C/min for analysis at full scan (33–350 amu) with a mass spectrometer. All analyses were performed with a Graseby-Nutech 3550A cryoconcentrator (Graseby-Nutech, Smyrna, GA) with a 16-canister autosampler interfaced to an ITS40 (Magnum) GC–MS ion trap instrument (Finnigan MAT, San Jose, CA). The analytical column consisted of sequentially joined SPB-1 precolumn (6 m 0.53 mm i.d. 1.0 mm stationary phase) and Rtx-Wax Crossbond-PEG analytical column (60 m 0.25 mm i.d. 0.50 mm stationary phase), both from Restek Corporation. QA standards and calibration standards were prepared by ManTech, Inc., EPA’s onsite support services research contractor under Contract 68-D5-0049. For the canister samples, C4–C11 n-alkanes, C4–C9 n-aldehydes, seven single-ring aromatics, four volatile alcohols, and three common chlorinated hydrocarbons were quantified. 2.3.2. Adsorbent tubes Before shipping tubes to S*ao Paulo, each was preconditioned for 15 min at 3501C in a pure helium stream to remove any impurities and then sealed with brass Swagelok fittings and Teflon ferrules. Processed tubes were individually wrapped in aluminum foil and sealed in individual glass culture tubes; sets of 20 tubes were then stored in clean aluminum cans filled with cushioning material to prevent breakage during transportation. The same packing method was applied after sample collection for transporting samples back to the EPA laboratory. Prior to each sampling event in S*ao Paulo, all flow rates for the tube pump were set and checked using a DC-1 Dry Cal Meter (Bios International Corporation, Butler, NJ). Tube samples were analyzed with a PerkinElmer ATD-400 automatic thermal desorption system (Perkin-Elmer Corp., Norwalk, CT) interfaced to a Varian Star 3400CX gas chromatograph and a Varian Saturn 2000 ion trap mass spectrometer (Varian, Walnut Creek, CA). The analytical column was a Rtx-1 (60 m 0.32 mm i.d. 1.0 mm stationary phase) from Restek Corporation. Tubes were desorbed at 2001C inside the ATD-400, and analytes were preconcentrated onto an internal Air Monitoring Trap (Perkin-Elmer Corp., Norwalk, CT) held at 271C; the subsequent thermal desorption of the trap for GC injection was made at 2801C. The GC temperature program consisted of an initial temperature of 351C held for 5 min followed by a temperature increase at a rate of 61C/min up to the final temperature of 2101C which was held for 0.83 min. For the adsorbent samples, C4–C9 n-aldehydes, 10 single-ring aromatics, four volatile alcohols, and two common chlorinated hydrocarbons were quantified.


2.4. Data reduction Upon chemical analysis, the data were organized by site, by sampling system, and by duplicates in spreadsheets (Lotus 1-2-3, R4, Cambridge, MA). Statistical analyses and graphs were produced with GraphPad Prism, version 2 (GraphPad Software, Inc., San Diego, CA). We also composited all canister data and all tube data as overall indicators of the S*ao Paulo airshed and to serve as a methods comparison. To put these data in perspective, we gathered similar information from recent US studies performed by EPA and tabulated overall results.

3. Results and discussion The overall field study was a success; all equipment operated properly despite a day of near-record heat, constant high humidity, and occasional rain. The CETESB monitoring stations were excellent field study sites in that we could deploy samplers on the roof (about 8 ft off the ground) to avoid low obstacles to airflow. We had access within 5–15 mm to all of the areas impacted by heavy traffic; only the Ibirapuera Park site had a traffic buffer zone of about 300–1500 mm in any direction. 3.1. General results The overall results of this study are presented in Table 1 along with some comparison data from recent work performed in the US. The data are presented as overall means and standard errors of the means (SEMs) for the S*ao Paulo canister and adsorbent tubes data. Similar information is presented from a study performed in RTP, NC, and in Nashville, TN (McClenny et al., 1998); in the Los Angeles basin in Azusa, CA (Daughtrey et al., 1998); and from other published studies in Southern California (Zielinska et al., 1997 for alcohols; Grosjean et al., 1996 for aldehydes; Fraser et al., 1997 for alkanes; and Lonneman, 1998 for data from 1997 only). A cursory inspection of the canister and adsorbent tubes data in Table 1 shows that the SEMs for canister data are appreciably smaller than for the tube data, indicating the probability of better precision for the canisters. If we allow for SEMs, the data means are essentially equivalent for methylene chloride, benzene, trichloroethene, toluene, butanal, and nonanal. There appears to be a technical problem with the remaining aldehydes measured by the adsorbent tubes as they do not match well with the canisters or with the patterns from the other studies. These issues will be addressed in more detail later. From a more general perspective, both canister and adsorbent tube data show that levels in S*ao Paulo

4021

exceed those found in the US studies, including Azusa and Los Angeles, CA. This is especially obvious for ethanol where the canisters show a mean of 414 ppbv, which greatly exceeds the 7.5-, 5.3-, and 17.7-ppbv means from other California studies (Daughtrey et al., 1998; Zielinska et al., 1997; Lonneman, 1998, respectively). This is expected because ethanol comprises approximately 40% of all fuel in S*ao Paulo, whereas the Southern California estimate is only about 3% for both methanol and ethanol combined (Zielinska et al., 1997). Additionally, our S*ao Paulo data match well with measurements made by Gee and Sollars (1998) in Caracas, Quito, Santiago, and S*ao Paulo during 1995 and 1996. Table 2 shows where their data overlap with ours (hexane, heptane, and six single-ring aromatic compounds). Gee and Sollars also include two Asian cities, Bangkok and Manila, where many of the ambient levels are appreciably elevated over their western hemisphere counterparts, especially for toluene, ethylbenzene, and the xylene isomers. Though we found benzene levels to be somewhat lower in our measurements in this comparison, we believe these numbers accurately reflect the airshed during our sampling period in 1998. 3.2. Site comparisons The six monitoring sites were expected to demonstrate similar VOC signatures as they were all heavily impacted by nearby traffic. Table 3a lists our canister concentration results distributed in columns by sampling site (the adsorbent tube results are discussed later). As expected, the Tunel 9 de Julho site has the highest levels of compounds associated with automotive sources, in particular, the ethanol level (1017 ppbv), which is well over twice the study mean (414 ppbv). Additionally, benzene is at 11.0 versus its mean study value of 2.6 ppbv, toluene is at 13.7 versus 9.0 ppbv, and m, pxylene is at 10.8 versus 4.6 ppbv; all the n-alkanes and most of the n-aldehydes from the Tunel samples also exceed their mean study values. Although Ibirapuera Park’s concentration values for most analytes are lower than the overall study mean, the values are definitely not uniformly the absolute lowest of the sites. We attribute this to the fact that the park is centrally located within the metropolitan area and that these data were collected on hot, sunny days with little wind. The data from the Dom Pedro site are appreciably lower as a whole than the other sites despite the fact that it is a downtown site; we attribute this to the cooler windy weather that day and to the fact that all the samples were collected in the morning. All individual sites demonstrate the same general VOC pattern as the overall mean values. Table 3b lists the tube data for each site; the patterns are similar to those of the canisters. However, in contrast to the canisters data, the more volatile aromatics are higher in the Congonhas and Lapa sites’

4022

Table 1 Overall comparisons (ppbv) n ¼ 78

n ¼ 280

n ¼ 325

n ¼ 4082425

S*ao Paulo Cans

S*ao Paulo Tubes

RTP

Nashville

Azusa

Mean

SEM

Mean

SEM

Mean

Mean

Mean

10.6 414.0 69.1 1.3

1.4 54.0 12.2 0.3

1.9 90.0 2.5 0.2

0.6 25.0 1.4 0.1

Methylene chloride Trichloroethene Tetrachloroethene p-Dichlorobenzene

0.5 0.2 0.6 0.0

0.0 0.0 0.1 0.0

0.6 0.2 1.3 0.1

0.2 0.0 0.5 0.0

0.2 0.0

0.0 0.0

0.1 0.0

0.0

0.0

Benzene Toluene Ethylbenzene m; p-Xylene Styrene o-Xylene 4-Ethyltoluene 1,3,5-Trimethylbenzene 1,2,4-Trimethylbenzene

2.6 9.0 2.0 4.6 0.7 1.5

0.4 0.7 0.2 0.5 0.1 0.2

2.5 15.1 3.5 7.8 1.6 2.4 1.8 0.7 2.3

1.0 5.2 0.9 1.8 0.5 0.5 0.4 0.1 0.1

0.4 0.4 0.1 0.2 0.1 0.1

0.0 0.0 0.0 0.0 0.0 0.0

0.0 1.6

0.0 0.1

3.6 0.0 0.4 0.0 0.2 10.2

2.4 0.0 0.1 0.0 0.0 7.1

0.9 0.4 1.4 0.4 0.7 1.1

0.3 0.2 0.1 0.0 0.1 0.1 0.0

Methanol Ethanol 2-Propanol 1-Propanol

Butanal Pentanal Hexanal Heptanal Octanal Nonanal

6.1 13.0 5.5 4.3 4.4 4.6

0.3 0.9 0.3 0.4 0.5 0.6

Butane Pentane Hexane Heptane Octane Nonane Decane Undecane

6.9 7.2 2.0 1.2 1.0 0.5 0.3 0.2

1.6 1.8 0.3 0.2 0.1 0.1 0.0 0.0

n ¼ 30 LA Study-Various

LA Study-Lonneman

SEM

Mean

SEM

Mean

SEM

7.5

0.36

15.2 5.3

na na

16.7 17.7 4.4

1.2 5.8 5.4

0.0 0.0

0.1 0.0

0.1 0.0

1.3 4.4

0.2 2.6

0.0

0.0

0.1

0.0

0.5 0.7 0.1 0.3 0.0 0.1 0.0 0.1 0.2

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

1.3 4.0 0.5 1.7 0.2 0.7 0.2 0.2 0.6

0.0 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0

1.7 5.1 0.7 2.6 0.5 0.9

1.3 4.2 0.7 2.8 0.4 1.0

0.4 0.9

0.5 1.1

0.1 0.0 0.0 0.0 0.0 0.0

1.3 0.1 0.3 0.3 0.5

0.0 0.0 0.0 0.0 0.0

0.6 0.1 1.1 0.3 0.5

0.1 0.0 0.0 0.0 0.0

0.7 0.4 0.4 0.6 0.5 1.1

1.11 0.95 0.61

0.2 1.1 0.3

0.46 0.33

0.3 0.3

0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.6 0.2 0.0 0.0 0.0 0.0 0.0

0.0 0.0 0.0 0.0 0.0 0.0 0.0

2.1 4.0 1.5 0.7 0.2 0.1 0.3 0.2

0.1 0.2 0.1 0.0 0.0 0.0 0.0 0.0

6.9 6.3 1.9 0.9 0.4 0.3 0.4 0.2

3.1 2.9 1.2 0.7 0.3 0.3 0.3 0.2

1.2 1.4 0.8 0.5 0.3 0.3 0.3 0.2

SEM

SEM

0.1 0.1 0.1 0.1 0.2 0.4 na na na na na na na na


n ¼ 34

Compound

4023

M. Colo!n et al. / Atmospheric Environment 35 (2001) 4017–4031 Table 2 Levels of VOCs in Asian and Latin American cities (ppbv) Pollutant

City Caracasa, Venezuela

Hexane

5.3

Heptane

2.5

Benzene

Quitoa, Ecuador

Santiagoa, Chile

Bangkoka, Thailand

Manilaa, Philippines

S*ao Pauloa, Brazil

S*ao Paulo, Brazil Canisterb

S*ao Paulo, Brazil Tubesb

6.5

8.3

2.7

0.8

3.2

8.4

2.0

2.7

1.2

4.4

1.6

4.6

5.6

3.9

5.2

2.6

1.7

Toluene

7.6

4.0

7.8

48.9

44.2

7.4

9.0

9.2

Ethylbenzene

1.1

0.5

1.5

8.3

5.0

1.4

2.0

2.3

m; p-xylene

3.7

1.5

5.7

18.5

12.7

4.2

4.6

5.8

o-xylene

1.3

0.5

2.0

6.6

3.8

1.4

1.5

1.9

1,3,5-TMB

0.7

0.9

1.3

1.8

0.6

a b

2.0

0.6

Reference data from Gee and Sollars (1998). Results from our November 1998 study in S*ao Paulo, Brazil.

tubes data than in the Tunel tubes data. The Dom Pedro tubes data also demonstrate lower overall levels like their canister counterparts.

combined, we have generated a comparison statistic difference parameter, di , defined as ‘‘percent difference’’, calculated by di ¼ ðS1i @S2i Þ=ðS1i þ S2i Þ * 200%;

3.2.1. Sampling method comparisonsFduplicates precision Because all samples were collected in duplicate, we could generate statistics to demonstrate a measure of analytical precision as well as comparisons among methods. As a visual demonstration, we chose the aromatic hydrocarbons, benzene, toluene, ethylbenzene, m; p-xylene, o-xylene, and styrene, as a class. Fig. 2a is a scatter plot of paired canister samples, and Fig. 2b is a similar plot of paired adsorbent tube samples. The canister duplicates exhibit a much tighter pattern around the theoretical ‘‘y ¼ x’’ line of perfect agreement than do the tubes duplicates. Though not statistically rigorous, we can make a simple calculation for comparison purposes that shows that the linear ‘‘goodness of fit’’ for all aromatic hydrocarbons has correlation values of r2 ¼ 0:9876, sy; x ¼ 0:3658 ppbv for canister duplicates and r2 ¼ 0:8547, sy; x ¼ 1:222 ppbv for tubes duplicates data (r2 is the square of the correlation coefficient and sy; x is the standard error of the mean of the vertical deviation from perfect agreement). To put the examples from Figs. 2a and b for intramethod comparison (i.e., comparing canister versus canister or tube versus tube data) into a mathematical framework for all of the data (and all compounds)

where i refers to the ith data set in time, and S1 and S2 are the duplicate measures. The sets of di ’s are then an indicator of the relative precision for that particular data set. Because the order of the samples (S1i or S2i ) is arbitrary, the sign is irrelevant and the average of di ’s would actually underestimate the scatter statistics. Therefore, we chose to generate the root mean square (RMS) for each set of di values, qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi Sðdi Þ2 ; RMSðdi set Þ ¼ to achieve a relative percent value for the expected positive difference among paired duplicate samples. The results from the above calculations are given in Table 4a and b (canisters and tubes, respectively), for each compound at each site, plus an average value pffiffiffi for the whole data set. Statistically, RMSðdi set Þ= 2 is an estimator of CV, the mean coefficient of variation; this is presented as the final column in Table 4a and b for the combined sample set. For all sites combined, the canisters (Table 4a) demonstrate mean duplicate precision expressed as CV: 18–37% for the alcohols, 3–10% for the aromatics, 3–22% for the chlorinated compounds, and 12–40% for the aldehydes. In comparison, the tubes data (Table 4b) include a number of entries where there are insufficient

4024

Table 3(a) Compound

Cerqueira Cesar

Ibirapuera Park

Lapa

Congonhas

Dom Pedro

Tunel 9 de Julho

n¼6

n¼6

n¼6

n¼6

n¼4

n¼2

Mean

SEM

Mean

SEM

Mean

SEM

Mean

SEM

Mean

Methylene chloride Chloroform Trichloroethene Tetrachloroethene p-Dichlorobenzene

Mean

SEM

2.1 28.8 8.6 0.6

12.6 1017.5 29.8 0.2

1.1 17.9 5.9 0.2

0.8 0.5 0.2 0.6 0.0

0.1 0.2 0.0 0.1 0.0

0.5 0.1 0.1 0.5 0.0

0.1 0.0 0.0 0.1 0.0

0.6 0.0 0.4 1.2 0.0

0.1 0.0 0.0 0.1 0.0

0.3 0.0 0.0 0.2 0.0

0.0 0.0 0.0 0.1 0.0

0.4 0.0 0.1 0.3 0.0

0.1 0.0 0.0 0.0 0.0

0.3 0.0 0.1 0.3 0.0

0.0 0.0 0.0 0.0 0.0

Benzene Toluene Ethylbenzene m; p-Xylene o-Xylene Styrene

3.5 11.9 2.3 5.3 2.0 1.0

0.5 0.7 0.1 0.5 0.2 0.2

1.4 7.4 1.4 2.5 0.9 0.5

0.3 1.5 0.3 0.6 0.2 0.2

1.4 11.7 3.1 6.0 1.6 0.4

0.1 1.2 0.3 0.5 0.1 0.1

3.5 6.5 1.3 4.6 1.6 0.5

0.6 0.8 0.2 0.9 0.3 0.0

0.9 4.4 0.8 1.6 0.5 0.4

0.1 0.3 0.1 0.1 0.0 0.0

11.0 13.7 3.0 10.8 4.1 2.7

0.4 0.3 0.1 0.2 0.1 0.1


5.4 15.9 5.3 4.7 3.4 2.4

1.2 0.9 0.9 0.3 0.7 0.8

6.0 11.0 5.9 4.2 5.3 5.3

0.5 1.0 0.6 0.9 1.3 1.3

7.0 10.7 5.7 3.8 4.1 5.4

0.7 1.5 0.6 0.7 0.8 1.4

6.4 13.3 4.8 4.7 5.2 5.4

0.5 1.7 0.9 1.0 1.7 2.2

5.0 10.2 4.3 3.3 3.2 3.8

0.7 1.5 0.7 0.7 0.5 0.9

6.3 29.1 7.2 6.8 4.1 3.0

1.8 1.3 1.1 1.8 1.7 0.8


5.8 5.9 2.6 1.6 1.5 0.8 0.5 0.3

0.3 0.5 0.2 0.2 0.1 0.1 0.1 0.1

3.7 2.9 1.1 0.6 0.5 0.3 0.3 0.1

1.0 0.7 0.3 0.1 0.2 0.1 0.0 0.0

2.9 2.5 1.0 0.7 0.7 0.3 0.3 0.2

0.2 0.1 0.1 0.0 0.2 0.0 0.0 0.0

19.8 21.9 4.2 1.9 1.3 0.5 0.3 0.1

6.8 7.4 1.1 0.3 0.2 0.0 0.0 0.0

1.9 1.7 0.7 0.5 0.3 0.2 0.1 0.1

0.1 0.1 0.0 0.1 0.1 0.0 0.0 0.0

11.1 14.7 5.1 3.7 3.0 1.5 0.9 0.5

0.1 0.4 0.1 0.1 0.0 0.0 0.0 0.0


(a) Average (ppbv) and standard error of the sample mean for the samples collected with canisters at different sites in S*ao Paulo, Brazil Methanol 9.2 2.4 12.9 4.6 13.4 2.5 6.4 1.7 6.9 Ethanol 467.8 79.1 397.1 125.7 313.1 97.6 525.9 108.7 138.4 2-Propanol 51.8 2.4 104.9 36.7 95.2 27.8 35.1 4.7 30.7 1-Propanol 1.9 1.2 1.1 0.5 1.7 0.5 1.2 0.6 1.0

SEM

Table 3(b) Compound

Cerqueira Cesar

Ibirapuera Park

Lapa

Congonhas

Dom Pedro

Tunel 9 de Julho

n ¼ 16

n ¼ 16

n ¼ 12

n¼6

n¼4

n¼2

Average

SEM

Average

SEM

Average

SEM

Average

SEM

Average

SEM

Average

SEM

0.0 15.8 0.1 0.0

0.0 1.0 0.0 0.0

0.0 104.8 0.0 0.1

0.0 2.5 0.0 0.0


0.1 0.0 0.2 0.1

0.0 0.0 0.0 0.0

0.3 0.0 0.5 0.0

0.0 0.0 0.1 0.0

0.7 0.5 1.6 0.0

0.1 0.1 0.1 0.0

0.0 0.0 0.2 0.0

0.0 0.0 0.1 0.0

0.0 0.0 0.1 0.0

0.0 0.0 0.0 0.0

0.1 0.0 0.1 0.0

0.0 0.0 0.0 0.0

Benzene Toluene Ethylbenzene m; p-Xylene Styrene o-Xylene 4-Ethyltoluene 1,3,5-Trimethylbenzene 1,2,4-Trimethylbenzene

0.8 6.4 2.0 5.8 1.3 2.0 2.0 0.7 2.6

0.2 0.2 0.1 0.4 0.1 0.1 0.2 0.1 0.2

0.8 6.3 1.4 2.9 0.6 1.0 0.6 0.2 0.8

0.2 1.0 0.2 0.4 0.1 0.1 0.1 0.0 0.1

1.5 12.2 3.6 7.1 0.7 1.8 1.0 0.4 1.2

0.1 1.0 0.4 0.8 0.1 0.2 0.1 0.0 0.1

2.9 5.8 1.1 4.0 0.7 1.4 1.2 0.5 1.7

0.6 0.9 0.1 0.7 0.0 0.2 0.2 0.1 0.3

0.0 1.5 0.5 1.4 0.4 0.4 0.3 0.1 0.4

0.0 0.2 0.1 0.1 0.1 0.0 0.0 0.0 0.0

0.4 5.2 1.7 7.3 2.5 2.7 2.8 1.3 4.4

0.0 0.1 0.0 0.1 0.1 0.1 0.1 0.0 0.0


0.0 0.0 0.9 0.0 0.3 0.2

0.0 0.0 0.2 0.0 0.1 0.0

0.1 0.0 0.0 0.0 0.1 26.2

0.0 0.0 0.0 0.0 0.0 26.1

0.2 0.0 0.1 0.0 0.1 0.1

0.0 0.0 0.1 0.0 0.0 0.0

0.1 0.0 0.6 0.0 0.3 0.3

0.0 0.0 0.4 0.0 0.1 0.1

0.0 0.0 0.0 0.0 0.1 0.1

0.0 0.0 0.0 0.0 0.0 0.0

0.0 0.0 1.9 0.0 0.0 0.0

0.0 0.0 0.1 0.0 0.0 0.0


(b) Average (ppbv) VOCs and standard error of the sample mean for solid sorbent for each sampling site in S*ao Paulo, Brazil Methanol 0.0 0.0 5.9 1.9 0.0 0.0 0.0 0.0 Ethanol 11.0 6.1 59.8 9.4 102.9 13.6 42.6 13.4 2-Propanol 0.1 0.1 0.8 0.2 1.1 0.2 0.0 0.0 1-Propanol 0.0 0.0 0.0 0.0 0.2 0.0 0.2 0.0

4025

4026


Fig. 2. (a) Scatter plot of paired canister samples. (b) Scatter plot of paired adsorbent tube samples.

data to perform meaningful calculations; these are designated as ‘‘NA’’ and were excluded from the overall statistics. The overall precision of the tubes aromatic compounds and chlorinated compounds measurements is similar to that of the canisters (albeit slightly worse). There are not enough data, however, to generate any statistics for methanol, pentanal, hexanal, and heptanal; also a number of additional ‘‘NA’’ entries scattered among the data decrease the value of some of the overall statistics. As such, the tube methodology does not provide the same level of performance as the canister method and cannot be recommended for measuring most of the oxygenated species. 3.2.2. Sampling method comparisonsFprecision of tubes versus canisters Precision calculations become a little more complicated for this data set when performing an inter-method comparison because each duplicate pair of canister data has one or more simultaneous duplicate pairs of adsorbent tube data. To generate these statistics and still use all available data, we chose to generate multiple di ’s as follows: di ðmÞ ¼ STðmÞi @SC1i = STðmÞi þSC1i * 200 and di ðnÞ ¼ STðnÞi @SC2i = STðnÞi þSC2i * 200;

where m and n refer to odd and even tube numbers, respectively; SC1i and SC2i are the first and second canisters of any duplicate pair; and ST(m)i and ST(n)i are the first and second tubes of any of the multiple corresponding duplicate pairs. This way, the first canisters and first tubes of any given ith data set in time are compared, and the second canisters are compared to the second tubes of a set. Because it is arbitrary which sample of any given set is chosen as the first, we assure ourselves that the statistical comparison is fair and complete. Because an intra-method bias has scientific meaning for these data, we chose to generate mean and SEM values of comparison (rather than RMS values) for the di (m or n) for all individual compounds across the combined sites data. Table 5 displays the mean percent difference as composites for each site, and the mean and SEM values for the composited data. For this comparison, we left the tube ‘‘nondetects’’ as part of the data set (interpreted as zeros); therefore, the statistics reflect these often major associated differences. For all sites combined, there are no valuable comparisons for the aldehydes, and the alcohols statistics are weak except for 1-propanol. The comparisons among the aromatic compounds indicate an overall negative bias for the tubes versus the canisters (except for styrene), with the most volatile aromatic compound, benzene, having the largest negative bias. The set of chlorinated compounds also demonstrates a mostly negative bias (except for p-dichlorobenzene). Coupled with the intra-method variability data, especially with regard to the erraticity of nondetects, these results indicate that the tube data are most likely the cause of the large differences in these comparisons.

4. Conclusions The VOC profile of the airshed in S*ao Paulo is dominated by ethanol with elevated methanol and 1- and 2-propanol at levels well above measurements available for US cities (see Table 1). The overall data trend also shows levels of C4–C9 n-aldehydes to be approximately 10 times higher than measurements made by others for Los Angeles or Azusa, CA. We conclude that the use of alcohol-based fuels is the primary source for these differences because alcohol comprises about 40% of all mobile fuel by volume in S*ao Paulo as compared to 3% in Los Angeles. As discussed in the results section above, we found that the overall patterns of VOCs at different sites in the city are very similar (see Fig. 3). Because this includes the Tunel 9 de Julho site deep within an automotive tunnel, we conclude thatthe overall driver for the VOCs profile in the cityis the vehicular traffic rather than industrial sources. This reasoning further confirms the conclusion thatthe relative overabundance of ethanol, alcohols, and aldehydic species (as compared

Table 4 Intra-method precisiona Compound

Cerqueira Cesar

Ibirapuera Park

Lapa

Congonhas

Dom Pedro

Tunel 9 de Julho

All sites combined n ¼ 15

RMS % diff

RMS % diff

RMS % diff

RMS % diff

RMS % diff

RMS % diff

Average RMS % diff

CV (%) of diff

26.3 48.4 33.5 14.7

67.8 3.6 27.7 67.8

33.5 37.1 57.7 18.9

17.5 3.5 39.5 199.5

37.0 26.1 37.4 52.7

26.2 18.5 26.4 37.2


8.8 10.3 3.5 6.5

18.5 29.9 0.9 77.4

20.0 9.1 15.0 41.5

6.9 7.8 1.8 43.9

8.0 18.7 1.9 23.5

25.8 18.0 3.3 0.0

14.7 15.7 4.4 32.1

10.4 11.1 3.1 22.7

2.8 3.5 1.9 4.3 6.9 4.1

0.9 4.1 3.7 3.2 13.8 2.7

7.6 14.8 13.8 13.7 31.2 13.2

1.9 1.6 2.7 6.9 11.1 1.6

2.1 2.1 3.1 3.8 9.0 1.6

7.3 4.5 3.5 4.5 10.3 4.3

3.8 5.1 4.8 6.1 13.7 4.6

2.7 3.6 3.4 4.3 9.7 3.2


25.0 6.0 20.4 20.3 62.3 64.6

10.2 11.5 19.3 24.5 33.7 63.4

13.3 23.4 21.2 40.1 33.8 30.2

14.7 31.1 17.3 35.6 33.8 73.1

23.3 20.5 31.9 50.7 37.2 51.0

58.0 8.8 31.6 52.4 83.7 53.0

24.1 16.9 23.6 37.3 47.4 55.9

17.0 11.9 16.7 26.3 33.5 39.5


4.2 3.8 3.6 6.5 20.6 3.0 7.2 69.3

2.2 12.7 4.3 10.9 36.6 1.9 14.5 50.2

16.5 11.1 16.1 16.9 38.4 21.5 17.7 60.3

3.6 3.3 3.3 13.9 15.4 7.6 9.4 66.9

4.6 6.7 5.8 16.0 92.1 1.3 1.1 21.2

2.2 4.8 3.6 3.3 0.8 0.9 3.0 11.0

5.5 7.1 6.1 11.3 34.0 6.0 8.8 46.5

3.9 5.0 4.3 8.0 24.0 4.3 6.2 32.9

42.9 19.6 32.3 40.7

NA 22.3 34.5 34.4

NA 37.1 NA 26.0

NA 15.8 6.0 18.4

NA 4.7 NA 3.9

42.9 19.9 24.3 24.7

NA 14.1 17.2 17.4

Benzene Toluene Ethylbenzene m; p-Xylene o-Xylene Styrene

(b) Tubes duplicate samples Methanol NA Ethanol NA 2-Propanol NA 1-Propanol NA

4027

37.2 59.3 55.7 14.0


(a) Canister duplicate samples Methanol 39.9 Ethanol 5.0 2-Propanol 10.0 1-Propanol 1.1

4028

Table 4 (continued ) Compound

Benzene Toluene Ethylbenzene m; p-Xylene Styrene o-Xylene 4-Ethyltoluene 1,3,5-Trimethylbenzene 1,2,4-Trimethylbenzene Butanal Pentanal Hexanal Heptanal Octanal Nonanal a

Ibirapuera Park

Lapa

Congonhas

Dom Pedro

Tunel 9 de Julho


RMS % diff

RMS % diff

RMS % diff

RMS % diff

RMS % diff

RMS % diff

Average RMS % diff

CV (%) of diff

NA 25.7 10.6 5.2

14.6 35.4 21.6 10.1

15.9 13.8 16.0 46.4

NA 36.9 16.5 60.7

22.6 NA 25.4 12.8

37.8 NA 8.9 NA

22.7 28.0 16.5 27.0

16.1 19.8 11.7 19.1

39.7 5.6 7.8 8.2 5.8 8.3 6.1 5.0 4.7

32.7 20.7 14.1 11.3 11.9 10.8 8.2 16.6 6.9

10.9 15.6 16.3 16.3 14.1 16.6 17.2 16.4 17.9

34.2 12.3 7.5 3.1 6.9 4.1 2.1 1.4 1.9

27.9 30.1 26.6 20.0 17.0 18.6 11.8 7.5 8.0

14.4 2.2 4.9 3.8 6.0 3.9 5.5 2.9 1.3

26.6 14.4 12.9 10.5 10.3 10.4 8.5 8.3 6.8

18.8 10.2 9.1 7.4 7.3 7.3 6.0 5.9 4.8

NA NA 35.1 44.7 15.0 26.8

11.4 NA NA NA 32.5 29.7

14.8 NA NA NA 42.4 34.6

NA NA 71.1 1.2 18.1 20.4

97.3 NA NA NA 0.7 14.7

41.2 NA 53.1 22.9 21.7 25.2

29.1 NA NA NA 15.4 17.9

NA NA NA NA NA NA

NA indicates insufficient non-zero values for calculation. CV is the coefficient of variation of the overall % difference between duplicates.



Cerqueira Cesar

Table 5 Inter-method precisionFtubes vs. canisters simultaneous samples Ibirapuera Park

Lapa

Congonhas

Dom Pedro

Tunel 9 de Julho


Mean %diff

Mean %diff

Mean %diff

Mean %diff

Mean %diff

Mean %diff

Average mean %diff

SEM of mean %diff

Methanol Ethanol 2-Propanol 1-Propanol

@200.0 @193.0 @199.5 @12.6

@82.8 @119.1 @194.4 2.5

@200.0 @68.6 @189.4 54.3

@166.7 @164.1 @200.0 18.4

@199.9 @154.7 @198.5 1.0

@200.0 @162.6 @200.0 33.7

@174.9 @143.7 @196.9 16.2

16.2 15.1 1.5 8.5


@178.3 @159.5 @92.7 185.5

@69.9 @109.6 @30.2 138.2

@17.7 19.1 19.5 100.7

@165.7 @126.8 @2.5 70.4

@152.5 @199.5 @129.6 30.8

@83.5 @199.7 @106.6 0.0

@111.3 @129.3 @57.0 87.6

22.0 28.1 21.0 23.7

Benzene Toluene Ethylbenzene m; p-Xylene Styrene o-Xylene

@133.8 @62.4 @14.5 1.2 17.0 @6.9

@100.0 @33.8 @16.0 @2.8 7.6 @7.9

0.6 @1.0 @3.8 @2.4 28.0 @10.9

@29.8 @13.0 @13.2 @11.9 18.4 @18.0

@192.2 @100.0 @39.9 @15.7 @1.8 @22.4

@184.7 @90.0 @55.3 @38.5 @6.7 @43.2

@106.7 @50.0 @23.8 @11.7 10.4 @18.2

27.4 14.0 6.7 5.0 4.5 4.7


@196.5 @200.0 @136.5 @198.4 @157.4 @115.3

@189.8 @199.9 @198.3 @195.7 @187.9 @164.2

@186.4 @199.5 @192.5 @197.8 @189.8 @189.6

@196.2 @200.0 @156.8 @196.4 @172.9 @161.8

@198.0 @199.7 @199.7 @199.7 @192.9 @189.8

@200.0 @200.0 @116.1 @200.0 @200.0 @200.0

@194.5 @199.9 @166.6 @198.0 @183.5 @170.1

1.8 0.1 12.3 0.6 5.4 10.7


Cerqueira Cesar

Compound

4029

4030


Fig. 3. Comparison of VOCs concentration (ppbv) by site.

to US cities) is most likely from the evaporation and combustion of alcohol-based fuels. The single-ring aromatic compounds and the C4–C11, n-alkanes are similar in concentration or slightly elevated compared to Los Angeles and appreciably higher than the smaller US cities of Research Triangle Park, NC, and Nashville, TN. We attribute this to a combination of big city industrial activity and the fraction of mobile source emissions from petroleumbased fuels. The levels of the few common chlorinated compounds measured are unremarkable. In a broader worldwide comparison for a subset of hydrocarbon compounds (see Table 2), we found that our data match well with other researchers’ data from large South American cities, including S*ao Paulo, and that there is a definite difference in alkyl aromatics for the Pacific rim cities of Bangkok and Manila where they are 4–6 times higher. Because the benzene values are roughly similar for all of the studies and thus indicative of vehicular traffic, we conclude that the exaggerated alkyl aromatic concentrations are most likely due to differences in industrial emissions. Based on the consistency of data as presented in both Tables 1 and 2 for overall typical averages, we conclude that our measurements are truly representative of the VOCs in the S*ao Paulo airshed. The statistical inter- and intra-method comparisons demonstrate an overall better performance for the canister method. We conclude that this may be due in part to the rigors of shipping the tube samples via air,

but most likely reflects the overall differences between these methods as experienced in our laboratories. The tube data give reasonable overall estimates for the straight-chain and aromatic hydrocarbons; thus, this method is deemed useful if logistics (size and handling) are primary issues. If more precise data are required, the canisters would be a better choice despite their greater size, cost, and weight and overall greater ‘‘hands-on’’ requirements in the field and laboratory.

Acknowledgements The authors thank the CETESB staff members who assisted in field sampling and made us feel welcome in S*ao Paulo, the CETESB management for providing travel funding for this work, EPA’s Office of International Activities for arranging this collaborative effort, and members of NERL/EPA and ManTech laboratory staff for advice and assistance in analyzing and interpreting the data. Special thanks to Karen Oliver and E. Hunter Daughtrey of ManTech Environmental for providing calibration standards and data from Azusa, CA, and Research Triangle Park, NC, field work; to Jack Suggs of US EPA for statistical advice and William A. Lonneman from the US EPA/SEE program for also providing field work data. The scientific portion of this work was funded by EPA, including infrastructure support from ManTech Environmental Technology, Inc., under EPA Contract


68-D5-0049 for calibration and quality assurance standards. This work has been subjected to EPA review and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. References ! CETESB, 1998. Relatorio Anual de Qualidade do Ar no Estado de S*ao Paulo 1997. CETESB-Companhia de Tecnologia de Saneamento Ambiental, S*ao Paulo, Brazil, August 1998. Daughtrey, E.H., Adams, J.R., Kronmiller, K.G., Oliver, K.D., Yoong, M.J., 1998. Speciated and total non-methane organic carbon hourly measurements at the Azusa monitoring site during the 1997 Southern California Ozone study (SCOS97). Proceedings of the EPA/AWMA Symposium on Measurement of Toxic and Related Air Pollutants, Cary, NC. Air a Waste Management Association, Sewickley, PA, September 1998, pp. 851–862. Fraser, M.P., Cass, G.R., Simoneit, B.R.T., Rasmussen, R.A., 1997. Air quality model evaluation data for organics. 4. C2– C36 non-aromatic hydrocarbons. Environmental Science and Technology 31 (8), 2356–2367. Gee, I.L., Sollars, C.J., 1998. Ambient air levels of volatile organic compounds in Latin American and Asian cities. Chemosphere 36 (11), 2497–2506. Grosjean, E., Grosjean, D., Fraser, M.P., Cass, G.R., 1996. Air quality model evaluation data for organics. 2. C1–C14 carbonyls in Los Angeles air. Environmental Science and Technology 30 (9), 2687–2703. Lonneman, W.A., 1998. Comparison of hydrocarbon composition in Los Angeles for the year 1968 and 1997. Proceedings

4031

of the EPA/AWMA Symposium on Measurement of Toxic and Related Air Pollutants, Cary, NC. Air a Waste Management Association, Sewickley, PA, September 1998, pp. 356–365. (NoteFused 1997 data only, also included some unpublished data from personal communication). McClenny, W.A., Daughtrey, E.H., Adams, J.R., Oliver, K.D., Kronmiller, K.G., 1998. Volatile organic compound concentration patterns at the New Hendersonville monitoring site in the 1995 Southern Oxidants Study in the Nashville, Tennessee area. Journal of Geophysical Research 103 (17), 22509–22518. Szwarc, A., Branco, G.M., Farah, E.L., Costa, W., 1991. Alcohol crisis in Brazil: the search for alternatives. Ninth International Symposium on Alcohol Fuels, Florence, Italy, 12–15 November 1991. Whitaker, A.D., Fortmann, R.C., Lindstrom, A.B., 1995. Development and testing of a whole-air sampler for measurement of personal exposure to volatile organic compounds. Journal of Exposure Analysis and Environmental Epidemiology 5 (1), 89–100. Winberry, W.T., Murphy, N.T., Riggan, R.M., 1989. Method TO-14 in compendium of methods for the determination of toxic organic compounds in ambient air. Environmental Protection Agency, Research Triangle Park, NC, EPA 600/ 4-89-017, US. Zielinska, B., Shire, J., Harshfield, G., Pasek, R., 1997. The concentration of oxygenated compounds in the Los Angeles, CA, area following the introduction of reformulated gasoline. Air a Waste Management Association 90th Annual Meeting and Exhibition, Toronto Ontario, Canada, June, 1997. Air a Waste Management Association, Sewickley, PA, Paper No. 97-RP139.06.