Benzene: A Secondary Pollutant Formed in the Three-Way Catalyst

Nov 25, 2004 - ANNA-MARIA FORSS, ‡. DOMINIK STEFFEN, ‡,§. AND. NORBERT V. HEEB †. Laboratory of Organic Chemistry and Laboratory of Internal...
0 downloads 0 Views 196KB Size
Environ. Sci. Technol. 2005, 39, 331-338

Benzene: A Secondary Pollutant Formed in the Three-Way Catalyst S T E F A N B R U E H L M A N N , * ,† ANNA-MARIA FORSS,‡ D O M I N I K S T E F F E N , ‡,§ A N D NORBERT V. HEEB† Laboratory of Organic Chemistry and Laboratory of Internal Combustion Engines, Swiss Federal Laboratories for Materials Testing and Research (EMPA), U ¨ berlandstrasse 129, 8600 Du ¨ bendorf, Switzerland

Benzene emissions from a relevant proportion of today’s gasoline-driven passenger cars and light-duty vehicles can increase by up to 2 orders of magnitude when driving at high engine load (e.g., on highways). Under such conditions, post-catalyst benzene levels exceeded those found precatalyst. As a consequence, formation of benzene in the catalyst was postulated. To further reduce ambient air concentrations of benzene, these critical operating conditions must be carefully avoided. Here, we report in detail to what extent and at what operating conditions catalystinduced benzene and toluene formation can occur. For that purpose, a EURO-1 passenger car (1.8 L, model year 1995) fulfilling the valid regulations, equipped with a new, twolayered, Pd-CeO2-Al2O3/Rh-ZrO2-Al2O3 three-way catalyst was operated at steady state on a chassis dynamometer at 100, 125, and 150 km/h at variable air to fuel ratios. Pre- and post-catalyst exhaust gas concentrations of benzene, toluene, C2-, and C3-benzenes were monitored at a time resolution of 0.5 Hz by means of chemical ionization mass spectrometry. A net benzene formation window, ranging from pre-catalyst exhaust gas temperatures of 600-730 °C and λ-values of 0.83-0.95, with a pronounced minimum at 0.87, was observed. Dealkylation reactions of aromatic hydrocarbons are assumed to be the major pathway leading to benzene.

Introduction Aromatic hydrocarbon emissions from gasoline-driven vehicles have a major impact on the ambient air quality (1). In Switzerland, the contribution of mobile sources to ambient air benzene concentrations was estimated to be 75% in 2000 (2). Alkylated benzenes are of concern due to their high photochemical ozone-formation potential (3-9); however, benzene is an even higher risk to human health. It is classified by the WHO and the U.S. EPA as a human carcinogen causing leukemia (10, 11). For 2001, the U.S. EPA (12) reported a mean average benzene concentration, in urban ambient air, of 2.4 µg/m3 (0.70 ppbv). For 2000, a weighted, ambient air benzene concentration of 1.9 µg/m3 (0.55 ppbv) was reported for the Swiss population (2). The risk for leukemia at a lifetime * Corresponding author phone: +41 1 823 4604; fax: +41 1 823 4041; e-mail: [email protected]. † Laboratory of Organic Chemistry. ‡ Laboratory of Internal Combustion Engines. § Present address: DaimlerChrysler Switzerland, Bernstrasse 55, 8952 Schlieren, Switzerland. 10.1021/es049755m CCC: $30.25 Published on Web 11/25/2004

 2005 American Chemical Society

exposure of 1 µg/m3 is quantified to be 6 × 10-6 by the WHO (13). A risk of one in a million is accepted for carcinogenic compounds such as benzene. The risk to the Swiss population based on the current lifetime exposure to benzene is an order of magnitude higher than the accepted risk of one in a million. Therefore, a significant reduction in benzene emissions originating predominantly from incomplete gasoline combustion and evaporation losses is necessary (2). The implementation of the three-way catalyst (TWC) technology to gasoline-driven vehicles was a major step toward reducing traffic-related emissions. For a complete conversion of all pollutants, it is important to operate the catalyst with a stoichiometric air to fuel ratio. λ is the quotient between the actual air to fuel ratio and the stoichiometric one. At optimal conditions, the current exhaust gas aftertreatment technology is able to reduce benzene emissions by up to 98% (14). Elevated emissions can be observed during cold-start and at under-stoichiometric conditions (λ < 1). This is the case when the closed loop control is switched off under full load. While increased benzene emissions in the cold-start phase are related to incomplete conversion in the catalyst, elevated benzene concentrations under hot conditions are caused by catalyst-induced formation. Up to 50 times higher benzene than toluene concentrations were determined at the end of the urban part of the U.S. Federal Test Procedure 75 (15). Benzene formation in the catalyst from static measurements on engine test benches was reported (16, 17), and a catalyst-induced benzene formation at a λ of 0.95 was postulated from chassis dynamometer measurements (18). Dealkylation of alkylated benzenes is a prominent mechanism for the formation of benzene on Pt, Ni, and Rh catalysts in steam reforming model studies (19, 20), and demethylation of toluene over a rhodium-based catalyst was identified as a benzene formation pathway (21). At transient driving, we observed negative conversion efficiencies indicating a net formation of benzene, and to a minor degree, of toluene when operating a EURO-1 TWC vehicle at high engine load (22). Investigations of the benzene emission characteristics of 20 EURO-1 and 6 EURO-2 vehicles studied on the chassis dynamometer at EMPA revealed that, on average, benzene emission factors (g/km) for both fleets significantly increased by 2 orders of magnitude at highway driving conditions (23, 14). Due to its negative impact on human health, current benzene levels in urban air and emissions from roads have to be reduced by about 1 order of magnitude. It is, therefore, of importance to precisely identify the critical vehicle and catalyst conditions that induce benzene formation. The studies performed in the 1980s and early 1990s at static engine modes, as well as the more recent findings from representative vehicle fleets operated at transient driving (real-world cycles), do not provide sufficient insight into what conditions and to what extent benzene is formed in the catalyst. We, therefore, investigated in more detail those operating conditions that are relevant for a catalyst-induced benzene formation. Since an investigation of these short time events in a transient driving cycle is not possible, we operated the test vehicle at steady-state conditions and at several working points. Chemical ionization mass spectrometry (CI-MS) was applied to determine pre- and post-catalyst concentrations of aromatic hydrocarbons, and the conversion efficiencies were derived therefrom.

Experimental Section Vehicle. All tests were carried out on a compact passenger car (BMW, 318is E36, model year 1995) with an engine VOL. 39, NO. 1, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

331

displacement of 1.8 L and a manually switched five-speed gearbox. The car had a mileage of 63 000 km and was equipped with a new, original spare part, TWC for this study. The new catalyst was preconditioned under real-world driving conditions for 3000 km in order to properly activate the catalyst and to strip off production residues. In this configuration, the test vehicle (EURO-1) met the Swiss FAV-1 regulations, which are based on the U.S. Federal Test Procedure 75 driving cycle with limitations for carbon monoxide (CO), total hydrocarbons (THC), and nitrogen oxides (NOx) of 2.1, 0.25, and 0.62 g/km, respectively. The test vehicle was subjected to transient measurements that were not part of this study. The same fuel and catalyst was used, but regulated emissions were measured in diluted exhaust. When driven in phase I of the New European Driving Cycle (NEDC), which includes a preconditioning and coldstart phase, the test vehicle emitted 9.38, 1.20, 0.30, and 0.070 g/km of CO, THC, NOx, and benzene, respectively. A fleet representative of Switzerland of 20 passenger cars (model year 1991-1996), emitted on average 4.24 ( 2.16, 0.56 ( 0.31, 0.49 ( 0.25, and 0.029 ( 0.014 g/km of CO, THC, NOx, and benzene, respectively, in the same test cycle (23). However, these values were detected with fuel containing only 2% benzene. At NEDC phase II, which includes transient extra-urban driving and highway driving, emissions of 0.07, 0.01, 0.01, and 0.001 g/km of CO, THC, NOx, and benzene were found for this vehicle. Corresponding values for the entire fleet are 0.48 ( 0.66, 0.04 ( 0.03, 0.21 ( 0.14, and 0.002 ( 0.003 g/km. The emission behavior of the test vehicle can be judged as working slightly below average at low temperatures (cold-start) and excellent at elevated temperatures. Its benzene emissions are also comparable. Fuel. The gasoline exhibited a total content of monoaromatic hydrocarbons of 37.4 vol %, including a benzene content of 3.0 vol %. Its detailed composition was determined by means of gas chromatography. Mole fractions of 0.11, 0.37, 0.33, and 0.20 for benzene, toluene, and the group of C2- and C3-benzenes, respectively, were determined. With the exception of the benzene content (3% vs 2.5%) and a slightly too high vapor pressure (66.2 vs 45-60 kPa), the gasoline fulfilled the specifications for a category II gasoline according to the World-Wide Fuel Charter (24). This fuel category is used in markets where U.S. Tier 0 or Tier 1, EURO-1 and EURO-2, or equivalent emission standards are required. Catalyst. Two identical catalysts were purchased as original spare parts. One was operated on the vehicle after conditioning for 3000 km. To determine its structure and chemical composition, the other was freed from its metal mantle and subjected to physical and chemical measurements. The washcoat structure was elucidated using a scanning electron microscope (SEM, 1455, LEO, Germany), equipped with an energy-dispersive X-ray detector (EDX, 7353, Oxford Instruments, U.K.). The chemical composition of a powder sample was specified by wavelength dispersive X-ray fluorescence spectroscopy (WD-XRF). For the semiquantitative determination, a WD-XRF spectrometer equipped with an Rh X-ray source and the corresponding filters (PW 2400, Phillips, The Netherlands) was used. The presence of Rh was verified using a WD-WRF spectrometer fitted with a Cr X-ray source (PW 1404, Phillips, The Netherlands). The BET surface areas were determined using a surface area analyzer (Coulter SA 3100, Beckman, USA). The catalyst consisted of two ceramic bodies (cordierite) of identical geometric dimensions with a honeycomb-like structure and a density of 62 cells per square centimeter, corresponding to 400 cells per square inch. The cordierite bodies (855 g) had a length of 11.5 cm and a face surface of 119 cm2. The washcoat had a twolayered structure. In the bottom layer, Pd was supported on 332

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005

a mixture of CeO2 and Al2O3. The top layer consisted of Rh supported on a mixture of ZrO2 and Al2O3. A BET surface area of 23 m2/g was measured. Considering a density of 0.43 g/cm3 (25) for the uncoated support, a BET surface area of 74 m2/g was calculated. Based on its chemical composition and design, the catalyst can be assigned to the second generation of TWCs, which were introduced in 1989 and since then widely used throughout the world (26-28). Exhaust Gas Analysis. Measurements of all pollutants were performed on undiluted exhaust gas held at 180 °C during sampling. To calculate the flow rate of the raw exhaust gas, the exhaust gas stream was diluted using a constant volume sampling system (CVS, Horiba, Japan) after passing the sampling sites. Since the fraction of the raw exhaust gas needed for the analysis was smaller than 1% under all applied conditions, the calculated raw exhaust gas flow rates were not corrected for that supplied to the analyzers. Concentrations of regulated compounds such as carbon monoxide (CO), total hydrocarbons (THC), and nitrogen oxides (NOx) as well as those of the aromatic hydrocarbons, benzene (C6H6), toluene (C7H8), and the C2- (C8H10) and C3-benzenes (C9H12) were measured on a chassis dynamometer at the Swiss Federal Laboratories for Materials Testing and Research (EMPA, Du ¨ bendorf, Switzerland). NOx was determined as NO but quantified as NO2 equivalents using a chemiluminescence detector (CLA-150, Horiba, Japan). CO was measured with a nondispersive infrared detector (AIA-110S, Horiba, Japan), and THC emissions were quantified by applying a flame ionization detector (FMA-126, Horiba, Japan) calibrated with propane using a mean molecular THC mass of 13.85 g/mol. The corresponding calibration gases (Sauerstoffwerke Lenzburg, Switzerland) consisted of NO (1000 ppm), CO (5 vol. %), and propane (2000 ppm). Calibration of these detectors was performed at the beginning of each measuring day. The pre-catalyst λ signal was observed using a λ meter (LA3, Etas, Germany) in combination with a linear λ sensor (Bosch, Germany). Measurements of aromatic hydrocarbons were carried out by means of chemical ionization mass spectrometry (Airsense 2000 Multichannel Analyzer, V&F Analysentechnik GmbH, Austria), observing the corresponding positively charged molecular ions using mercury as the ionization gas (Hg+, 10.4 eV). Calibration of the mass spectrometer was carried out prior to every test run using two calibration gases: one consisting of benzene, toluene, and p-xylene in synthetic air (10 ppm each, Carbagas, Switzerland), and the other containing isopropylbenzene in synthetic air (1 ppm, Praxair, Belgium). The class of C2-benzenes consists of o-, m-, and p-xylenes and ethylbenzene, whereas the class of C3-benzenes includes eight isomers (namely, three trimethyl-, three methylethyland two propylbenzene isomers). A weighted response factor for the C3-benzenes was derived from a gas chromatographic analysis of the applied gasoline and the CI-MS response factors for each isomer. With less than 5% deviation, this response factor corresponded to that of 3-ethyl methylbenzene. Therefore, C3-benzene concentrations are reported here as 3-ethyl methylbenzene equivalents. Pre- and post-catalyst exhaust gas temperatures as well as the oil temperature of the engine were monitored using NiCr/NiAl thermocouples and were kept in acceptable ranges with the aid of an air conditioning system and a fan capable of simulating a wind speed of up to 160 km/h. Experimental Procedure. All tests were performed at constant velocity and engine full load to obtain stabilized steady-state conditions. We found that transient vehicle operation was not applicable for a detailed study of the catalyst-induced benzene formation. Each individual test run was effected according to the following procedure: (i) the car was preconditioned to reach a minimum oil temperature of 80°C, (ii) the velocity of the vehicle was set by the chassis

FIGURE 1. λ-dependent conversion efficiencies of regulated pollutants and corresponding pre-catalyst temperatures at 100 (+), 125 (9), and 150 (4) km/h. Only the measurements in a comparable λ-window of 0.88 ( 0.02 were considered for calculation of mean values given in Table 1. dynamometer driving in fifth gear, and (iii) engine full load and λ-value was adjusted by the aid of the accelerator pedal. After 12 min of driving, both the oil and the pre-catalyst exhaust gas temperature were constant. The pre-catalyst emissions were monitored for a total of 16 min, the sampling site was switched while the car was kept running under constant conditions, and the post-catalyst emission data was collected for another 14 min. Applying this sampling protocol, steady-state conditions were obtained, and the pre- and postcatalyst exhaust concentrations could be derived from the CI-MS response of the individual pollutants.

Results and Discussion So far, catalyst-induced benzene formation was observed only during fuel-rich combustion and at elevated exhaust gas temperatures, which typically occur during highway driving or acceleration. The catalyst conversion efficiencies (η) for regulated and unregulated pollutants are therefore discussed for such conditions. According to the formula η ) 1 - (post-catalyst concentration/pre-catalyst concentration), λ-dependent conversion efficiencies were calculated. Emissions of Regulated Pollutants. The λ-dependent conversion efficiencies of total hydrocarbons (THC), carbon monoxide (CO), and nitrogen oxides (NOx) are shown in Figure 1, and the corresponding operating conditions and mean emission factors for regulated pollutants as well as for aromatic hydrocarbons are given in Table 1. No dependence on λ could be observed for NOx conversion, which can be expected as only reducing conditions were investigated. For the same reason, direct CO oxidation is hindered, and its conversion is low but slightly increasing at λ above 0.95. Conversion of THC proceeds differently for a given λ depending on the exhaust gas temperature. At a comparable λ of about 0.9, mean THC conversion efficiencies of 0.08,

0.39, and 0.76 were achieved at 100, 125, and 150 km/h, respectively, with mean pre-catalyst exhaust gas temperatures of about 620, 700, and 780 °C (Table 1). Under these conditions no relevant CO conversion was detected with efficiencies of 0.05, 0.004, and 0.11. Under the applied reducing conditions, direct oxidation is hindered, and the partial CO conversion is believed to proceed via a water gas shift reaction that is catalyzed by ceria (29, 30). Both CO and hydrogen (H2) are known to be present under reducing conditions. Therefore, the abatement of NOx is strongly enhanced by the reducing agents present, leading to excellent conversions of 0.98, 0.99, and 1.00. Summing up the behavior of the catalyst for the emissions of regulated pollutants, we noticed that the catalyst is working as expected for fuel-rich combustion conditions, namely, successfully reducing NOx, moderately converting THC, but only poorly oxidizing CO. This performance is not a sign of a malfunctioning catalyst but has to be anticipated in situations where the oxygen storage component is not able to bridge the lack of oxygen. We assume that the presence of rhodium and to a lesser extent palladium, which are known to be active in steam reforming (31), enable a moderate conversion of THC mainly at higher temperatures. Emissions of Aromatic Hydrocarbons. Figure 2 displays the conversion efficiencies of aromatic hydrocarbons at 100, 125, and 150 km/h versus λ (please note the different scale for benzene). At 100 km/h, a pre-catalyst temperature of 617 ( 22°C, and a mean λ of 0.86 ( 0.02 (Table 1), conversion of benzene and toluene proceeds with nearly constant but negative mean efficiencies of -0.38 and -0.09, respectively. These negative conversion efficiencies indicate a net formation of both benzene and toluene. Mean post-catalyst emission factors of 206 and 416 mg/km were obtained for benzene and toluene, respectively. Conversion efficiencies VOL. 39, NO. 1, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

333

TABLE 1. Operating Conditions and Mean Emission Factors of Regulated and Unregulated Pollutantsa velocity

λ temperature

100 km/h (n ) 7b/8) pre-catalyst pre-catalyst post-catalystc

exhaust gas flowd rotational speed residence timee performance THC η CO η NOx

pre-catalyst post-catalyst pre-catalyst post-catalyst pre-catalyst post-catalyst

η benzene η toluene η C2-benzene η C3-benzene η

pre-catalyst post-catalyst pre-catalyst post-catalyst pre-catalyst post-catalyst pre-catalyst post-catalyst

Operating Conditions 0.86 ( 0.02 °C 617 ( 22 °C 373 ( 11 L/s 39.8 ( 1.3 rpm 3157 ( 26 s 0.015 ( 0.001 kW 41.0 ( 1.4 Emission Factors of Regulated Pollutants mg/km 2734 ( 147 mg/km 2511 ( 153 0.08 mg/km 89500 ( 11500 mg/km 85400 ( 12700 0.05 mg/km 2997 ( 471 mg/km 68 ( 43 0.98 Emission Factors of Aromatic Hydrocarbons mg/km 150 ( 18 mg/km 206 ( 22 -0.38 mg/km 380 ( 22 mg/km 416 ( 25 -0.09 mg/km 489 ( 49 mg/km 450 ( 32 0.08 mg/km 516 ( 43 mg/km 417 ( 54 0.19

125 km/h (n ) 8)

150 km/h (n ) 6)

0.88 ( 0.01 693 ( 6 453 ( 5 51.3 ( 0.5 3840 ( 6 0.011 ( 0.001 54.2 ( 0.7

0.88 ( 0.02 773 ( 10 542 ( 7 65.2 ( 0.9 4599 ( 4 0.009 ( 0.001 68.2 ( 0.9

2533 ( 72 1553 ( 154 0.39 78400 ( 7500 78100 ( 8100 0.004 4460 ( 580 25 ( 5 0.99

2794 ( 170 680 ( 166 0.76 79700 ( 15200 70600 ( 15200 0.11 5034 ( 882 20 ( 4 1.00

167 ( 12 625 ( 61 -2.74 363 ( 20 246 ( 50 0.32 408 ( 18 95 ( 23 0.77 387 ( 34 44 ( 12 0.89

213 ( 19 186 ( 116 0.12 418 ( 30 59 ( 19 0.86 415 ( 28 32 ( 8 0.92 373 ( 47 23 ( 5 0.94

Only data points in a comparable λ-window of 0.88 ( 0.02 were considered. b One run not measured with CI-MS. c Measured at the end of pipe. d Conditions: T ) 273.15 K, p ) 101325 Pa, Constant: R ) 8.314 J/(mol‚K). e The residence time was approximated dividing the catalyst volume (L) by the exhaust gas flow rate (L/s). a

of C2- and C3-benzenes increase slightly with λ at 100 km/h. A net reduction with positive mean conversion efficiencies of 0.08 and 0.19 was observed for C2- and C3-benzenes, respectively, corresponding to post-catalyst emission factors of 450 and 417 mg/km. At 125 km/h, a pre-catalyst temperature of 693 ( 6 °C and an enlarged λ-window from 0.83 to 0.98, clear changes appear for the conversion efficiencies of all aromatic hydrocarbons. Conversion efficiencies of C2- and C3-benzenes exhibit a nonlinear increase with rising λ. Abatement of these two substance classes proceeds with positive conversion efficiencies over the entire λ-range. Mean values of 0.77 and 0.89 are reported for the C2- and C3-benzenes, respectively. Please notice that only those data points laying in a λ-window of 0.88 ( 0.02 were considered for the calculation of mean values (Tables 1 and 2). This allows a comparison of the three data sets at a similar air to fuel ratio. Thus, under these conditions the post-catalyst emissions were diminished to 95 and 44 mg/km for C2- and C3-benzenes, respectively. Toluene shows a slightly different behavior at 125 km/h. Below a λ of 0.84, we noticed slightly negative conversion efficiencies as already observed at 100 km/h. Exceeding a λ of 0.84, toluene conversion becomes positive and shows a nearly linear dependence on λ, comparable to that observed for THC (Figure 1). For the data subset at 125 km/h, a mean toluene conversion efficiency of 0.32 and a post-catalyst emission factor of 246 mg/km was found at λ ) 0.88 ( 0.01. However, the conversion of benzene proceeds very differently to those of the alkyl benzenes. A positive conversion, indicating a net degradation, is observed only at λ > 0.95. Even under these conditions, the benzene oxidation is always less efficient than that of all other alkyl benzenes as 334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005

well as the entire class of hydrocarbons. In contrast to these compounds, benzene is formed very efficiently over the catalyst when driving at 125 km/h with pre-catalyst exhaust gas temperatures exceeding 670 °C. If the velocity is further increased to 150 km/h, corresponding to an additional rise of the pre-catalyst exhaust gas temperature to about 780 °C, good to excellent conversion efficiencies of 0.86, 0.92, and 0.94 for toluene, C2-, and C3benzenes, respectively, were found. The corresponding postcatalyst emission factors are further reduced to 59, 32, and 23 mg/km, respectively. In this high-temperature regime, benzene is converted with either negative or slightly positive efficiencies from -0.74 to 0.47, corresponding to a mean benzene emission factor of 186 mg/km. At elevated temperatures, the benzene and toluene conversion still depend on λ to some degree, whereas the catalytic degradation of C2- and C3-benzenes is not affected by λ. Molar Distribution and Degree of Alkylation. Table 2 displays the mass proportion of individual aromatic hydrocarbons with respect to the THC group of compounds, as well as their molar distribution. The aromatic hydrocarbons accounted for 51-56 and 44-65 mass % in the pre- and post-catalyst exhaust gas, respectively. A moderate increase of the benzene/THC ratio was observed in the pre-catalyst exhaust gas with values of 5.5, 6.6, and 7.6 mass % at 100, 125, and 150 km/h, respectively. Post-catalyst, a substantial increase was noticed with a maximum benzene/THC ratio of about 40 mass % at 125 km/h, where the catalyst-induced benzene formation is strongly favored and the benzene degradation is still suppressed. We assume that under these conditions benzene has become the most prominent individual hydrocarbon present in the post-catalyst exhaust gas.

FIGURE 2. λ-dependent conversion efficiencies of aromatic hydrocarbons at 100 (+), 125 (9), and 150 (4) km/h. Only the measurements in a comparable λ-window of 0.88 ( 0.02 were considered for calculation of mean values given in Table 1. Negative conversion efficiencies indicate a net formation over the catalyst.

TABLE 2. Variations in Mean Exhaust Gas Composition across the Catalysta velocity

100 km/h 125 km/h 150 km/h (n ) 7b/8) (n ) 8) (n ) 6)

Aromatic Hydrocarbon/THC Mass Proportion benzene/THC

pre-catalyst post-catalyst toluene/THC pre-catalyst post-catalyst C2-benzene/THC pre-catalyst post-catalyst C3-benzene/THC pre-catalyst post-catalyst aromatic HC/THC pre-catalyst post-catalyst

% % % % % % % % % %

5.5 8.2 13.9 16.6 17.9 17.9 18.9 16.6 56.2 59.3

6.6 40.3 14.3 15.9 16.1 6.1 15.3 2.9 52.3 65.1

7.6 7.4 15.0 8.7 14.8 4.8 13.4 3.3 50.8 44.1

Mole Fraction (xi)c benzene toluene C2-benzenes C3-benzenes

pre-catalyst post-catalyst pre-catalyst post-catalyst pre-catalyst post-catalyst pre-catalyst post-catalyst

0.13 0.18 0.28 0.30 0.31 0.29 0.29 0.23

0.16 0.67 0.30 0.22 0.29 0.08 0.25 0.03

0.19 0.68 0.32 0.18 0.27 0.09 0.22 0.05

1.62 0.47 1.15

1.52 0.52 1.00

Degree of Alkylation (z) z z ∆z

pre-catalyst post-catalyst

1.75 1.57 0.18

a Only data points in a comparable λ-window of 0.88 ( 0.02 were considered. b One run not measured with CI-MS. c Mole fraction of aromatic hydrocarbons based on CI-MS response.

From the CI-MS measurements, we calculated the mole fraction (xi) of an individual aromatic hydrocarbon (i) within the group of aromatic hydrocarbons according to formula xi

) ni/Σni where ni corresponds to the exhaust concentration of compound (i). The chemical composition of the fuel, determined with gas chromatography, is shown in Figure 3. Mole fractions of 0.11, 0.37, 0.33, and 0.20 were found for benzene, toluene, and the group of C2- and C3-benzenes, respectively. The corresponding mole fractions of the aromatic hydrocarbons in the pre- and post-catalyst exhaust gas are given in Table 2 and are also shown in Figure 3. A rather wide and stable molar distribution was observed for the three pre-catalyst data sets. Increasing the velocity from 100 to 150 km/h, and with it the pre-catalyst exhaust gas temperature from about 620 to 780 °C, resulted in a small increase of the benzene mole fraction from 0.13 to 0.19, whereas that of the C3-benzenes decreased from 0.29 to 0.22. The observed pre-catalyst molar distributions are similar to those of the applied fuel. Again, the exhaust gas composition changed significantly when introducing a catalyst (e.g., when driving at 125 km/h); a narrow molar distribution, rich in benzene but poor in alkyl benzenes, was obtained after the catalyst. The mole fraction of benzene increased up to 0.67 in the post-catalyst exhaust gas, whereas that of the C3benzenes decreased to 0.03. All alkyl benzenes obey the empirical formula C6+yH6+2y, where y is the degree of alkylation of an individual aromatic hydrocarbon (i). Since four classes of aromatic hydrocarbons with y ) 0, 1, 2, and 3 were subjected to our measurements, a mean degree of alkylation (z) can be calculated using the formula z ) (Σxiyi). Table 2 reports the degree of alkylation of the pre- and post-catalyst exhaust gas as well as the difference (∆z). As mentioned, the aromatic hydrocarbon composition of the pre-catalyst exhaust gas changed only slightly with decreasing z-values of 1.75, 1.62, and 1.52 at increasing speeds of 100, 125, and 150 km/h, respectively. From the gaschromatographic analysis of the fuel, a comVOL. 39, NO. 1, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

335

FIGURE 3. Molar distributions of benzene, toluene, C2-, and C3-benzenes in the fuel as well as in the pre- (9) and post-catalyst (0) exhaust gas in a λ-window of 0.88 ( 0.02. parable z-value of 1.61 was obtained. Minor changes are observed after the catalyst at 100 km/h with z ) 1.57, but the degree of alkylation is clearly affected at 125 and 150 km/h, with reduced post-catalyst values of 0.47 and 0.52. Thus, one can notice that the degree of alkylation within the investigated class of aromatic hydrocarbons is considerably diminished after the catalyst. This reduction in the degree of alkylation may be explained by a more efficient conversion of alkyl benzenes as compared to benzene. Nevertheless, since we know that more benzene leaves the catalyst than has entered, we assume that an efficient benzene formation via dealkylation of alkyl benzenes is much more probable than an inhibited degradation of benzene. This hypothesis is supported by results from others (19-21). Considering that benzene is a carcinogen with a negative impact on human health and also regarding the intensive benzene formation observed, the critical conditions that allow a catalyst-induced benzene formation are discussed in more detail below. Catalyst-Induced Benzene Formation. A strong influence of the temperature on the catalyst performance was recognized. While driving at 100 km/h with a pre-catalyst exhaust gas temperature of about 620 °C, a moderate net formation of benzene (η ) -0.38) is observed. A highly efficient benzene formation occurred over the catalyst when driving at 125 km/h and a pre-catalyst exhaust gas temperature of about 700°C (Figure 2). The benzene conversion efficiencies are negative in a broad λ-range of 0.83 up to 0.95. At a λ-value of 0.83, a negative conversion efficiency of -1.42 was observed. While increasing λ, the corresponding conversion efficiency value further decreases until a minimum is reached at λ ) 0.87. In this status, with an air to fuel ratio of 12.83 (stoichiometric ) 14.7), a strongly negative benzene conversion efficiency of -3.42 is obtained, corresponding to preand post-catalyst emission factors of 168 and 740 mg/km, respectively. At this point 4.4 times more benzene leaves the catalyst than enters it. A further increase in λ results in a linear increase of the conversion efficiency with, for the first time, positive conversion at λ ) 0.96. The measurement point 336

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005

obtained with the highest λ of 0.98 resulted in a positive conversion efficiency of 0.82. At 150 km/h, resulting in a temperature of about 780 °C, a slightly positive mean benzene conversion efficiency of 0.12 was obtained. From these data, we deduce that the intrinsic formation via dealkylation becomes relevant at about 600 °C, whereas the direct oxidation of benzene remains ineffective until a pre-catalyst temperature of around 700°C is reached. Primarily with higher temperature but also with increasing λ, an alternative degradation pathway becomes dominating. This results in positive benzene conversion efficiencies at λ-values above 0.96 and pre-catalyst temperatures above 730 °C. From these findings, we identify a net benzene formation window between λ-values of 0.83 up to 0.95 and pre-catalyst exhaust gas temperatures from 600 up to 730 °C. Our findings are in good agreement with results from others. Grimm (16) reported an intensive catalyst-induced benzene formation when operating a 2.0 L engine (Mercedes Benz M102 E20) at maximum torque and rotational speed with different catalytic systems. Conversion efficiencies of η ) -1.6 and -1.2 were determined across an Fe-ZrO2-Ce and a Pt-Al2O3+ Rh-Al2O3-Ce catalyst, respectively, at a λ-value of 0.86 and a pre-catalyst exhaust gas temperature of 722 °C. Pelz et al. (17) observed, on a 2.3 L engine (Mercedes Benz M102 E) at full load and a pre-catalyst exhaust gas temperature of 830 °C under λ < 1 conditions, a moderate benzene formation with η ) -0.28 across a catalytic system for which no chemical specification was given. Summers and Silver (18) detected η of -0.61 and -0.55, on a 2.5 L passenger car (Chevrolet Citation, 1982) equipped with a new platinum/rhodium 5:1 catalyst at λ ) 0.95 and precatalyst exhaust gas temperatures of 450 and 478 °C, respectively. This, with respect to the low temperatures, relatively strong benzene formation may be explained with the high loading of rhodium, which is known to be very active in steam reforming (29, 31). Since the reported experiments, including ours, were all carried out under steady-state conditions, it is of interest to

compare these findings with previous results obtained from transient driving (real-world cycles). A fleet consisting of three EURO-2 passenger cars and three EURO-2 light-duty vehicles, on average, emitted only about 1.0-3.5 mg/km benzene under optimal conversion conditions (η ) 0.98) but their emissions increased by 2 orders of magnitude, up to 80 and 110 mg/km, at increased velocity with negative conversion efficiencies of -0.38 and -0.49 at 135 and 145 km/h, respectively (14). In an earlier study on a EURO-1 passenger car (Renault Express 1.4 L, 1995) equipped with a Pd/Rh TWC, a net benzene formation indicated by conversion efficiencies of -0.50 and -0.92 was observed at 135 and 145 km/h, respectively. Under these conditions, the engine was operated in a fuel-rich regime with a reported mean λ-value of 0.93 ( 0.06 (22). These findings also agree with those reported here, namely, that driving under steady-state conditions (125 km/h) with a λ of 0.93 results in a benzene conversion efficiency of -1.06. Net benzene formation at exhaust gas temperatures between 600 and 730 °C with λ below 1 is typically found at full load enrichment, which is motivated 2-fold. First, maximum engine power is achievable around a λ of 0.85 for combustion kinetic reasons. Second, lower exhaust gas temperatures reduce thermal stress to both engine and catalyst, preventing noble metal sintering as well as evaporation losses. Considering the reported disadvantages, this practice should, in our opinion, be carefully avoided. Any additional measure that prevents catalyst-induced benzene formation (e.g., the development of heat-resistant catalyst material) should be implemented in future vehicle technology. The benzene formation potential of individual alkylbenzenes present in the pre-catalyst exhaust gas is currently under investigation. Further research on other potential benzene precursors should be carried out, as there is no certainty yet whether dealkylation of alkylbenzenes is the only relevant pathway towards benzene formation. Fuelrich operation is considered essential during the reduction as well as the desulfation step of NOx storage-reduction catalysts (32). With respect to the findings reported here, a secondary benzene formation in this operating phase may also be possible and should be investigated prior to wider application of this exhaust gas aftertreatment technology.

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

Acknowledgments The authors thank R. Luescher and C. Ruedy, EMPA Laboratory for Internal Combustion Engines, for providing technical assistance on the chassis dynamometer. C. J. Saxer, EMPA Laboratory of Organic Chemistry, is acknowledged for his support in computer-based data evaluation and his practical contribution during the measuring campaign. We thank P. Lienemann and M. Trottmann, EMPA Laboratory of Inorganic Analytical Chemistry, and B. Bommer, EMPA Laboratory for High Performance Ceramics, for the characterization of the catalyst.

Literature Cited (1) Grosjean, E.; Rasmussen, R. A.; Grosjean, D. Toxic air contaminants in Porto Allegre, Brazil. Environ. Sci. Technol. 1999, 33 (12), 1970-1978. (2) Swiss Agency for the Environment, Forests and Landscape. Benzol in der Schweiz. Schriftenreihe Umwelt 2003, No. 350. (3) Atkinson, R. Gas-phase tropospheric chemistry of organiccompoundssa review. Atmos. Environ. 1990, 24A (1), 1-41. (4) Carter, W. P. L.; Atkinson, R. Computer modeling study of incremental hydrocarbon reactivity. Environ. Sci. Technol. 1989, 23 (7), 864-880. (5) Carter, W. P. L. Development of ozone reactivity scales for volatile organic-compounds. J. Air Waste Manage. Assoc. 1994, 44 (7), 881-899. (6) Derwent, R. G.; Jenkin, M. E.; Saunders, S. M. Photochemical ozone creation potentials for a large number of reactive

(21)

(22)

(23)

(24) (25)

(26) (27) (28) (29) (30)

hydrocarbons under European conditions. Atmos. Environ. 1996, 30 (2), 181-199. Derwent, R. G.; Hov, O. Computer modeling studies of the impact of vehicle exhaust emission controls on photochemical airpollution formation in the United-Kingdom. Environ. Sci. Technol. 1980, 14 (11), 1360-1366. Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Atmospheric chemistry of tropospheric ozone formationsscientific and regulatory implications. J. Air Waste Manage. Assoc. 1993, 43 (8), 1091-1100. Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Tropospheric air pollution: ozone, airborne toxics, polycyclic aromatic hydrocarbons, and particles. Science 1997, 276 (5315), 1045-1052. WHO (IARC), International Agency for Research on Cancer. IARC Monographs on the Evaluation of Carcinogenic Risk to Humans; IARC: Lyon, 1987; Suppl. 7, p 120. U.S. EPA. Carcinogenic Effects of Benzene: An Update; EPA/ 600/P-97/001-F; U.S. Environmental Protection Agency: Washington, DC, 1998. U.S. EPA. 2001 Urban Air Toxics Monitoring Program; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2001; 140 pp. WHO. Air Quality Guidelines, 2nd ed.; Regional Office for Europe: Copenhagen, Denmark, 2000; http://www.who.dk/air/ Activities/20020620_1. Heeb, N. V.; Forss, A.-M.; Weilenmann, M. Pre- and postcatalyst-, fuel-, velocity- and acceleration-dependent benzene emission data of gasoline-driven EURO-2 passenger cars and light duty vehicles. Atmos. Environ. 2002, 36 (30), 4745-4756. Dearth, M. A.; Gierczak, C. A.; Siegl, W. O. Online measurement of benzene and toluene in dilute vehicle exhaust by massspectrometry. Environ. Sci. Technol. 1992, 26 (8), 1573-1580. Grimm, H. Die Emission von Benzol und anderen Kohlenwasserstoffen im Abgas von Ottomotoren und ihre Beeinflussung durch katalytische Abgasnachbehandlung. Ph.D. Thesis, University of Technology Clausthal, Germany, 1988, pp 1-166. Pelz, N.; Dempster, N. M.; Hundleby, G. E.; Shore, P. R. The composition of gasoline engine hydrocarbon emissions-an evaluation of catalyst and fuel effects. SAE Tech. Pap. Ser. 1990, No. 902074. Summers, J. C.; Sliver, R. G. Catalytic control issues associated with the use of reformulated gasolines. SAE Tech. Pap. Ser. 1990, No. 902072. Duprez, D.; Miloudi, A.; Delahay, G.; Maurel, R. Selective steam reforming of aromatic-hydrocarbons. 4. Steam conversion and hydroconversion of selected monoalkyl-benzenes and dialkylbenzenes on Rh catalysts. J. Catal. 1984, 90 (2), 292-304. Duprez, D.; Miloudi, A.; Delahay, G.; Maurel, R. Selective steam reforming of aromatic-hydrocarbons. 5. Steam conversion and hydroconversion of selected monoalklbenzenes and dialkylbenzenes on Pt and Ni catalysts. J. Catal. 1986, 101 (1), 56-66. Patterson, M. J.; Angove, D. E.; Cant, N. W.; Nelson, P. F. The formation of benzene and chlorobenzene during the oxidation of toluene over rhodium-based catalysts. Appl. Catal. B 1999, 20 (2), 123-131. Heeb, N. V.; Forss, A.-M.; Bach, C.; Mattrel, P. Velocity-dependent emission factors of benzene, toluene and C2-benzenes of a passenger car equipped with and without a regulated 3-way catalyst. Atmos. Environ. 2000, 34 (7), 1123-1137. Heeb, N. V.; Forss, A.-M.; Bach, C.; Schweizer, T. Velocitydependent benzene emission factors of three-way catalyst passenger cars and light duty vehicles from time- resolved exhaust gas analysis. Int. J. Vehicle Des. 2001, 27 (1-4), 296305. ACEA; Alliance; EMA; JAMA: World-Wide Fuel Charter; 2002; http://www.acea.be. Siemund, S.; Stiebels, S. Abgasemissionen. In Handbuch Verbrennungsmotor; van Basshuyssen, R., Scha¨fer, F., Eds.; Verlag Vieweg: Braunschweig/Wiesbaden, 2002. Heck, R. M.; Farrauto, R. J. Automobile exhaust catalysts. Appl. Catal. A 2001, 221 (1-2), 443-457. Gandhi, H. S.; Graham, G. W.; McCabe, R. W. Automotive exhaust catalysis. J. Catal. 2003, 216 (1-2), 433-442. Farrauto, R. J.; Heck, R. M. Catalytic converters: state of the art and perspectives. Catal. Today 1999, 51 (3-4), 351-360. Barbier, J., Jr.; Duprez, D. Steam effects in 3-way catalysis. Appl. Catal. B 1994, 4 (2-3), 105-140. Bunluesin, T.; Gorte, R. J.; Graham, G. W. Studies of the watergas-shift reaction on ceria-supported Pt, Pd, and Rh: implications for oxygen-storage properties. Appl. Catal. B 1998, 15 (12), 107-114.

VOL. 39, NO. 1, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

337

(31) Whittington, B. I.; Jiang, C. J.; Trimm, D. L. Vehicle exhaust catalysis. 1. The relative importance of catalytic-oxidation, steam reforming and water-gas shift reactions. Catal. Today 1995, 26 (1), 41-45. (32) Sassi, A.; Noirot, R.; Rigaudeau, C.; Belot, G. Influence of catalyst support geometry and wall thickness on NOx trap desulfation: rich/lean wobbling strategies and hexagonal cell supports for

338

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005

high SO2 selectivity during desulfation. Top. Catal. 2004, 30-31 (1-4), 267-272.

Received for review February 17, 2004. Revised manuscript received July 27, 2004. Accepted October 4, 2004. ES049755M