Customer exposure to MTBE, TAME, C6 alkyl methyl ethers, and benzene during gasoline refueling.

We studied customer exposure during refueling by collecting air samples from customers' breathing zone. The measurements were carried out during 4 days in summer 1996 at two Finnish self-service gasoline stations with "stage I" vapor recovery systems. The 95-RON (research octane number) gasoline contained approximately 2.7% methyl tert-butyl ether (MTBE), approximately 8.5% tert-amyl methyl ether (TAME), approximately 3.2% C6 alkyl methyl ethers (C6 AMEs), and 0.75% benzene. The individual exposure concentrations showed a wide log-normal distribution, with low exposures being the most frequent. In over 90% of the samples, the concentration of MTBE was higher (range <0.02-51 mg/m3) than that of TAME. The MTBE values were well below the short-term (15 min) threshold limits set for occupational exposure (250-360 mg/m3). At station A, the geometric mean concentrations in individual samples were 3.9 mg/m3 MTBE and 2. 2 mg/m3 TAME. The corresponding values at station B were 2.4 and 1.7 mg/m3, respectively. The average refueling (sampling) time was 63 sec at station A and 74 sec at station B. No statistically significant difference was observed in customer exposures between the two service stations. The overall geometric means (n = 167) for an adjusted 1-min refueling time were 3.3 mg/m3 MTBE and 1.9 mg/m3 TAME. Each day an integrated breathing zone sample was also collected, corresponding to an arithmetic mean of 20-21 refuelings. The overall arithmetic mean concentrations in the integrated samples (n = 8) were 0.90 mg/m3 for benzene and 0.56 mg/m3 for C6 AMEs calculated as a group. Mean MTBE concentrations in ambient air (a stationary point in the middle of the pump island) were 0.16 mg/m3 for station A and 0.07 mg/m3 for station B. The mean ambient concentrations of TAME, C6 AMEs, and benzene were 0.031 mg/m3, approximately 0.005 mg/m3, and approximately 0.01 mg/m3, respectively, at both stations. The mean wind speed was 1.4 m/sec and mean air temperature was 21 degreesC. Of the gasoline refueled during the study, 75% was 95 grade and 25% was 98/99 grade, with an oxygenate (MTBE) content of 12.2%. ImagesFigure 1Figure 2Figure 3Figure 4

<0.02-51 mg/m3) than that of TAME The MTBE vlues were well below the short-term (15 min) tehold limits set for ocupational exposure (250-360 mg/m3). At station A, the geometric mean concentrains in individual samples were 3.9 mg/m3 MTBE and 2.2 mg/m3 TAME. The corsponding values at station B were 2.4 and 1.7 mg/m3, respecly. The average refueling (sampling) time was 63 sec at station A and 74 sec at sation B. No statisticaily significant difference was observed in customer eTosures bewen the two service stations. The overall geometrc means (n = 167) for an adjusted 1-min refueling time were 3.3 mg/im3 MTBE and 1.9 mg/m3 TAME. Each day an ntegrate d breathing zone sample was also collected, corresponding to an arithmetic mean of20-21 r lings. The overall arithmetic mean concentrtions in the integrated samples (n -8) were 0.90 mg/mn3 for benzene and 0.56 mg/m3 for C6 AMEs calculated as a group. Mean MTBE concentrations in ambient air (a stationary point in the middle of the pump island) were 0.16 mg/m3 for station A and 0.07 mg/m3 for station B. The mean ambient concentrations ofTAME, C6 AMEs, and benzene were 0.031 mg/m3, -0.005 mg/nm3, and -0.01 mg/m3, respctiey, at both stations. The mean wind speed was 1.4 m/sec and mean air temperature was 21°C. Of the gasoline refueled during the study, 75% was 95 grade and 25% was 98/99 grade, with an oxygenate (MTBE) content of 12.2%. Key words bernene, gasoline, methyl tbutyl. ether, oxyfuel, oxygenate, pump island, ing sece station, eamyl methyl ether. Environ Healb Penpeetl07:133-140 (1999). [Online 13January 19991 bap://e/epnal.niehbs.nib.gv/doesl1999/107p133-14Ovainiotal/abstra tml Methyl tert-butyl ether (MTBE) and tertamyl methyl ether (TAME) were introduced in the mid-1970s as gasoline additives. MTBE was originally used as an octane enhancer. In the 1990s, however, the reduction of motor vehicle emissions of CO and hydrocarbons has become an important reason for its use and, indeed, for the use of oxygenates in general. Unleaded gasoline that contains oxygenates such as MTBE, TAME, ethanol, methanol, or ethyl tertbutyl ether (ETBE) is commonly used in several western European countries and North America. At present, all of the gasoline used in Austria, Sweden, and Finland is unleaded (1). In Europe, the oxygenates are generally ethers and, unlike in the United States, little ethanol is currendy used (4. In Finland, unleaded gasoline with 11% MTBE has been used since 1991. Reformulated gasoline with a typical MTBE content of 11-12% was put on the market in 1994. In 1995, a new reformulated product was introduced, containing an oxygenate mixture of MTBE, TAME, and other alkyl methyl ethers (AMEs). The new product is based on etherification of the reactive C5-C6 isoolefins present in fluid catalytic cracking light gasoline fractions (3). The etherification is carried out with methanol and isoamylenes to obtain TAME, and with methanol and tertiary C6 olefins to obtain heavier ethers. The result is a mixture of hydrocarbons, TAME, and heavier ethers, such as tertiary hexyl methyl ethers. This mixture is then added to gasoline. MTBE can also be added to the same gasoline pool.
Because MTBE is now the most common oxygenate, its toxicologic effects and, to some extent, human exposure levels have been investigated (mainly in the 1990s) (4)(5)(6). The exposure of customers during refueling with MTBE-oxygenated gasoline was assessed in a study carried out at two Finnish service stations (7). No studies have been reported on driver exposure to TAME or other AMEs present in the oxygenate mixture. The aim of this study was to assess the exposure of customers to MTBE, TAME, and C6 AMEs during refueling. We also measured the exposure to benzene. The field measurements, also including a stationary sampling point at the pump island, were carried out at two Finnish service stations.
There are no set threshold concentrations for such short-term exposures as customer exposure during refueling. For occupational MTBE exposures, there are limit values for an 8-hr working day in some European countries and in the United States, ranging from 90 to 180 mg/m3 (8)(9)(10)(11). In Sweden and in The Netherlands, short-term (15 min) occupational exposure limit values of 250 and 360 mg/m3, respectively, have also been set. Futhermore, a reference concentration of 3 mg/m3 has been set by the U.S. EPA as a guideline for longterm exposures via environmental air (14. No threshold limit values (TLVs) exist for TAME and C6 AMEs.
The United States currently has a shortterm standard of 8 mg/m3 and an 8-hr TLV of 1.6 mg/m3 (11) for benzene. In Europe, national standards exist, but the European Union has also set a common limit value of 3.25 mg/m3 (8 hr), which will come into effect with a 3-year transition period from the year 2000; during the transition period, the limit is 9.75 mg/m3 (13).

Materials and Methods
Service stations andgasolines. The study was carried out in summer 1996 at two self-service stations located in the Helsinki, Finland, metropolitan area, representing the typical Nordic station in size and design. Service station A was an urban station with two adjacent main roads with high traffic densities. Service station B was a roadside station. The measurements were carried out on 19-22 August at station A and on 26-29 August at station B.
Both stations had three pump islands in front of the service station building, with a total of 12 pumps at station A and eight pumps at station B for gasoline and diesel oil. At station A, six of the pumps were for dispensing 95-and 99-research octane number (RON) gasoline and six for 98-RON gasoline/diesel oil. At station B, seven of the pumps were for dispensing 95-RON gasoline, three for 98-RON gasoline, six for 99-RON gasoline, and three for diesel oil. The gasoline dis-All gasoline grades sold during the study Samples were taken of the 95-RON penser pistols were equipped with rubber (unleaded 95-, 98-, and 99-RON) were of the gasoline to determine its contents of oxy-"splash collars," which are in general use reformulated type and of summer grade, with genates and benzene. The samples were in Finland. Both stations had "stage I" the following maximum contents: lead 0.003 taken at the beginning of the exposure vapor recovery systems that collect vapors g/l, benzene 1% (v/v), and sulfur 0.01% measurements and after deliveries of gasoduring gasoline unloading (but not with (w/w). The oxygen content was specified as line. The samples were collected in glass stage II recovery, which collects vapors 2.0-2.7% (w/w) and the Reid vapor pressure bottles, which were then tightly sealed and released from a vehicle's gasoline tank was 67-68 kPa (at 37.8°C). The density of transferred to a test laboratory of the gasoduring refueling).
The values obtained are shown in Table 1    pIes were collected at each station.
Pump island measurements. The ambient air measurements at the pump islands were carried out on the same days as the breathing zone measurements. The sampling point was in the center of the forecourt between the two pumps on the middle pump island. Two parallel samples were collected in charcoal tubes on each measurement day from 0800 to 1600 hr. This stationary sampling point was 1.5 m above the ground at station A and 1.8 m above the ground at station B.
Meteorologic conditions. Air temperature and relative humidity were measured hourly in the center of the forecourt (pump island sampling point) with a Vaisala HMI 31 Humidity and Temperature Indicator (Vaisala Oy, Vantaa, Finland). Wind speed and direction were measured continuously with a wind gauge (Typ 1482, Wilh. Lambrect KG, Gottingen, Germany) located on the roof of the service station building about 5 m above the ground.
Method ofsampling. The sampling was performed using Du Pont S 2500 pumps (E.I. du Pont de Nemours & Co., Wilmington, DE), which pulled air through the charcoal tubes (SKC 226-01; 150 mg of adsorbent) at 1 1/min (breathing zone samples) or 0.2 l/min (pump island samples). At the service stations, the charcoal tubes were stored in a cooled box (+2°C) after sampling. Each day after sampling, the tubes were stored at -20°C until analysis.
Method ofanalysis. The analytes were desorbed from the charcoal with 1 ml of benzene-free CS2 (Rathburn Chemicals, Walkerburn, UK) in a glass vial. After 1 hr at room temperature with occasional vibration, the vials were mounted on a GC autosampler (HP 7673; Hewlett-Packard, Palo Alto, CA) and a 2-pl subsample was injected into the GC. The GC instrument was a Hewlett-Packard 5890 equipped with a 5790 A quadrupole mass-selective detector (electron ionization mode). The temperature of the injector was 250°C, and a slow injection was used.
Calibration standards were made by the phase equilibrium method by adding 1 ml of a standard mixture in CS2 to a glass vial containing charcoal, the amount and batch of the charcoal being the same as that used for samples. The mixture was allowed to stand at room temperature for 1 hr before analysis. The stock standard solution was prepared weekly, and the calibration standards were prepared daily. The stock standard solutions of MTBE, TAME, and benzene were prepared from pure solvents (>97%), whereas the calibration standards of C6 AMEs were prepared from an etherified standard mixture containing TAME, C6 AMEs, and hydrocarbons. The mixture was obtained from the oil company (Neste Oy), which also supplied the information on the density and percent composition of the standard mixture.
The detection limits were 0.04 pg/in1 for MTBE and 0.03 pg/ml for TAME, which correspond to atmospheric concentrations of 0.04 and 0.03 mg/m3, respectively, for a 1-liter air sample and 0.0004 mg/m3 and 0.0003 mg/im3, respectively, for a 90-liter air sample.
The detection limits for C6 AMEs and benzene were in the range of0.02-0.05 pg/m. Results Breathing zone measurements/individual samples. The overall average concentrations of MTBE and TAME in individual breathing zone samples for the two service stations are shown in Table 4. The data were log-normally distributed; therefore, geometric means (GM) were calculated as the central value. In these calculations, each value below the detection limit was assigned a value L/2 (half of the detection limit). At station A, 2/83 MTBE and 2/83 TAME results were below detection limits. At station B, 4/84 MTBE and 3/84 TAME results were below detection limits.
Because the refueling times varied (23-189 sec), average concentrations for 1 min (C1 min) were calculated for all individual samples (concentration measured x refueling time in seconds/60 sec). GMs were also provided for these values ( Table 4). The time-adjusted values allowed the exposures of individual customers and the average exposures of the two service stations to be compared. ing Cl min val-values were tested using the Student's t-test. :station A) and There were no significant differences (95% 3). The differ-confidence level) in MTBE or TAME expoof the Cl min sures between the two service stations (p = 0.13, MTBE; p = 0.38, TAME).
)nes and overall The Cl min values of MTBE and TAME were added, and these sum values are shown in Figures 1 and 2 variables. The difference in customer exposure between vehicles with and without a converter was not statistically significant (pvalue for MTBE was 0.17 and for TAME 0.08).
The sampling site was recorded as the location of the pump island and as a compass point indicating on which side of the pump island the vehicle was parked during refueling. There were 9 sampling sites at station A and 10 at station B. The number of samples taken at each sampling site ranged from 1 to 17. The mean concentrations of more than 12 refuelings (data not shown) were tested 140. ._e E using analysis ofvariance (ANOVA). No statistically significant differences were obtained among the sampling sites.
Twenty-nine different makes of cars refueled at both stations, the most abundant being Toyota (n = 26), Opel (n = 16), Nissan (n = 15), and Volvo (n = 13). There were 1-6 different models within one make, for a total of 68 models. The data did not allow correlation of the make and model of the vehide with the time spent in refueling or with the exposure levels. The mean refueling time was 63 sec at station A and 74 sec at station B (overall mean 68 sec).
Breathing zone measurements/integrated samples. The results of the breathing zone measurements by integrated sampling are shown in Table 5. Each sample consisted of 20-21 refueling operations, thus corresponding to daily arithmetic means of individual samples. The variation among the 4 sampling days at each station was small; therefore, the arithmetic mean was chosen as the central value for the integrated samples.
The arithmetic mean concentrations of the four integrated samples were 9.7 mg/m3 MTBE and 5.3 mg/m3 TAME at  Sum concentrations calculated for the group of C6 AMEs were 12% (station A) and 11% (station B) of the corresponding mean concentrations of TAME, and 4.2% (station A) and 3.9% (station B) of the total amounts of C4-C6 AMEs measured. The mean benzene concentration was 6.4% (station A) and 6.3% (station B) of the total amount of C4-C6 AMEs measured. In all, the mean proportions of each individual component measured in the breathing zone were very similar at the two stations.
Pump island measurements. The daily and the mean concentrations of C4-C6 alkyl methyl ethers and benzene in ambient air measured in the middle of the pump island are presented in Tables 2 and 3. Each day, the highest concentrations were measured for MTBE. The daily levels of C6 AMEs were low but mostly detectable; only four of the values were below the detection limit. The arithmetic mean concentration of MTBE was 0.16 mg/m3 at station A and 0.069 mg/m3 at station B. The mean concentration of TAME was the same at both service stations (0.031 mg/m3). The mean concentrations of C6 AMEs were also very similar at the two stations, ranging from 0.0005 mg/m3 (2,3-dimethyl-2-methoxybutane) to 0.0027 mg/m3 (2-methyl-2-methoxypentane). The benzene level was slightly higher at station A (0.012 mg/m3). For daily TAME concentrations and wind speed, the Pearson's coefficient of correlation (r) was 0.95 (p = 0.05) at station A and 0.99 (p = 0.01) at station B. At station B, MTBE (r = 0.96; p = 0.04) and benzene (r = 0.99; p = 0.006) were also highly correlated with the daily wind speed. At station A, the r values for MTBE and benzene were also rather high (0.71 and 0.88), but because of the small number of samples (n = 4), they did not reach statistical significance.

Discussion
Breathing zone measurements. The concentrations of MTBE in customers' breathing zones were higher than those of TAME in over 90% of the samples, although the amount of TAME in 95-RON gasoline was at least three times the MTBE concentration. This is explained by the clearly higher vapor pressure of MTBE (55 versus 17 kPa at 37.8°C) (4). Nevertheless, there was great variation in the TAME/MTBE ratio among the samples. The concentration of TAME was on average 60-70% of the MTBE concentration but, as can be seen in Figures 1 and 2, the TAME/MTBE ratio ranged from 0.1 to 1.0 in individual samples. This variation may have been partly due to gasoline fumes in the car tank from a previous fill-up. The oxygenate composition of these fumes may have been different from that of the gasoline added.
The levels of both MTBE and TAME in the individual samples varied widely, from below detection limits to concentrations at least 1,000 times higher, resulting in a GSD of up to 7 for the overall mean concentrations. Because of the high sensitivity of the method, only six samples remained below detection limits. Low exposures were the most frequent. At least 50% of C1 mi values were below 5 mg/m3 for MTBE or 3 mg/m3 for TAME, although the highest exposure levels (Cl min) measured were around 100 mg/m3 for MTBE and 70 mg/m3 for TAME.
High variations in the data contributed to the fact that a Student's t-test showed no statistically significant difference between the two service stations, despite the higher overall GMs for MTBE and TAME at station A. At station A, the air temperature was higher (40C), as was the wind speed, which should in turn reduce the exposures. In all, the weather conditions at both stations were representative of north or central European summer conditions. The mean volume of gasoline dispensed was almost the same at both stations, but the mean refueling time was 11 sec shorter at station A. The difference in mean refueling time may be explained by different adjustments of gasoline flow in the pump pistols.  (Fig. 3). At station A, however, the integrated samples yielded higher arithmetic mean concentrations than did the individual samples. As can be expected for a highly skewed log-normal distribution, the GM values (representing the proper central values for the exposure concentrations) are much lower than the arithmetic means.
The integrated samples and the arithmetic means overestimate the average exposure during refueling.
Despite the above, the integrated samples were useful in clearly showing the differences in the volatilities of the components measured. Figure 4 presents the average concentrations in liquid gasoline and the corresponding average airborne concentrations measured by integrated sampling. For example, the proportions of MTBE and C6 AMEs were rather similar in gasoline, but the amount of MTBE found in air samples was 15 times higher. The Reid vapor pressure of benzene (22 kPa) is only slightly higher than that of TAME (17 kPa), but according to our results, benzene is much more easily volatilized than TAME. This may be due to a positive discrepancy of Rauolt's law, which has been observed for benzene in hydrocarbon mixtures (14).
In the Finnish service station study conducted in 1995 (7), an overall GM value of 6 mg/m3 was obtained for MTBE during a 1-min refueling time. The range for four measurement periods was 4.6-8.2 mg/m3, and the MTBE content of the gasoline was 1 1%. In the present study, the  gasoline used for refueling contained on average 2.7% of MTBE, and the corresponding overall average concentration of MTBE in air was 3.3 mg/m3. The latter figure seems to be higher than expected on the basis of the much lower MTBE content of the gasoline (averaging 25% of the MTBE content in the aforementioned study). One explanation may be gasoline vapors from the car tanks, originating from a previous refueling, as explained above. Nevertheless, the highest atmospheric concentrations measured in the 1995 study were about 200 mg/m3, as compared with 50 mg/m3 measured in the present study.
The arithmetic mean concentration of benzene calculated for the eight integrated samples was 0.9 mg/m3. This equals the value obtained in an earlier study in which 228 refuelings were measured in Cincinnati, Ohio (15). The benzene concentration of the regular unleaded gasoline was 0.92% (v/v) in this study, whereas in our study (7) it was slightly lower: 0.75% (w/w) corresponding to 0.64% (v/v). When the results of the two studies are calculated for a 1-min refueling time, 1.4 mg/m3 and 1.0 mg/m3 mean concentrations are obtained for the Cincinnati study and our study, respectively. These values are consistent with the slightly different benzene contents of the gasolines in these studies. In a previous study from 1984-1985 (16), 21 self-fill operations were measured, and an average benzene concentration of 4.3 mg/m3 was obtained for a 2min refueling time, which corresponds to 8.6 mg/m3 calculated over a 1-min period. The average benzene content of the gasoline was reported to be 4% (w/w). Again, this result is consistent with ours.
When the individual exposures found in the present study are compared with the occupational short-term limits (15 min), it can be concluded that even the highest MTBE and benzene concentrations at the customer breathing zone are well below these limits. The highest MTBE exposure (C1 mi) was 106 mg/m3, and the highest benzene exposure was estimated (based on integrated samples) to be 1/10 of the MTBE C1 I value, 10 mg/m3. The upper 95% confidence limit (UCL) for a 1-min refueling time was calculated to be 114 mg/m3 for MTBE and 56 mg/m3 for TAME. For the sum of MTBE and TAME, the 95% UCL was 169 mg/m3.
Pump island measurements. The daily concentrations of the components measured correlated well with the wind speed especially at station B. No such correlation was found in our 1995 study (17). This may be due to the shifting wind during that study, whereas the wind was rather stable throughout the measurements for the present study.
The average pump island concentrations were either similar (TAME, C6 AMEs, benzene) at the two stations or higher (MTBE) at station A, although the wind speed was higher and the volume of gasoline sold was lower at that station. These factors should contribute to lower air concentrations. In this case, the lack of contribution from these factors might be explained by the sampling, sites being doser to the pumps at station A. Emissions from passing traffic are unlikely to have markedly affected the impurity levels at the pump island, as evidenced by our previous study carried out in the vicinity of two service stations (17).
The mean pump island concentrations of MTBE measured in the summer in a previous study (17) were on average 0.9 mg/im3. The MTBE content of the gasoline was 1 1%  and the volume of gasoline dispensed was 21,000 1/24 hr. In the present study, the corresponding mean concentration was slightly lower (0.11 mg/m3) when considering the average MTBE content of the gasolines, which was estimated to be around 5%. The volume of gasoline dispensed was 12,000 1/8 hr. It should be noted, however, that the sampling points are not fully comparable because the service stations were not the same in the two studies. However, the air temperature and wind speed were rather similar in both studies. In two Italian service station studies (18), mean MTBE concentrations of 0.11 mg/m3 (2.1% MTBE) and 0.25 mg/m3 (2.7% MTBE) in gasoline were measured at the pump island. In a U.S. service station study (19), an overall GM of 0.39 mg/m3 was obtained for MTBE at five non-stage II self-service stations when the MTBE content of the gasoline was around 15%. The overall means (0.11 mag/m3; 5% MTBE in gasoline) of our study are not inconsistent with the above studies.
For benzene, the overall GM concentration at the pump island in the U.S. study was 0.032 mg/m3 (range 0.003-0.17 mg/m3) (19). The benzene content of the gasoline averaged 1.3% (v/v); the benzene content in our study was about 0.64% (v/v). Again, there is no discrepancy between the U.S. study and ours: the overall benzene concentration at the pump island was 0.01 mg/m3 in our study.

Conclusions
The individual exposure measurements in customers' breathing zones yielded widely distributed concentrations, with low concentrations being the most frequent. Due to the higher volatility of MTBE, its concentration was higher in most samples than the TAME concentration, despite the threefold higher content of TAME in the gasoline. On average, a customer was exposed to an MTBE concentration of 3.3 mg/m3 and to a TAME concentration of 1.9 mg/m3 for 1 min. No statistically significant difference was observed in customer exposures to MTBE or TAME between the two service stations or between vehides with and without catalytic converters. The average concentrations of benzene and C6 AMEs in customers' breathing zones were about 1/10 and 1/15, respectively, of the average MTBE concentration.
The pump island measurements showed the highest overall mean concentration for MTBE (0.11 mg/m3) and second highest for TAME. The concentrations of benzene and C6 AMEs were low but measurable. A correlation was found between wind speed and daily pump island concentration.