Environ. Sci. Technol. 2002, 36, 677-682
Significance of Semivolatile Diesel Exhaust Organics for Secondary HONO Formation LUKAS GUTZWILLER,† FRANK ARENS,† URS BALTENSPERGER,† H E I N Z W . G A¨ G G E L E R , † , ‡ A N D M A R K U S A M M A N N * ,† Paul Scherrer Institute, CH-5232 Villigen PSI, Switzerland, and Department of Chemistry and Biochemistry, University of Bern, CH-3012 Bern, Switzerland
The atmospheric origin of nitrous acid (HONO) is largely unknown despite its estimated importance as an OH source during daytime due to its rapid photolysis. Recently, primary HONO contained in automobile exhaust as well as secondary HONO formation on soot particles have been invoked as possible HONO sources, but none of them is able to account for the observed HONO to NOx ratios of up to 0.04 in the atmosphere. In this paper, we show that semivolatile and/or water-soluble species contained in diesel exhaust are significantly involved in secondary HONO formation. These species are not associated with soot when the exhaust exits the tailpipe. To quantify these species and to assess the reaction kinetics leading to HONO, experiments were performed in which filtered but hot diesel exhaust gas interacted with a glass surface as well as a water film mimicking dry and wet surfaces to which exhaust might be exposed. A fraction of 0.023 of the NOx emitted was heterogeneously converted to HONO, which is at least three times more than the primary HONO emissions by diesel engines and a fraction of 50 larger than HONO formed on diesel soot particles that do not contain the semivolatile organics.
Introduction Nitrous acid (HONO) seems to be an important precursor of OH radicals in the troposphere due to its photolysis by sunlight (1). Field studies indicate that HONO is mainly formed heterogeneously from NO2 on ground or airborne surfaces such as aerosol particles and cloud droplets. Overnight, HONO can make up 0.01-0.04 of the nitrogen oxides in urban areas (1). The two main reactions currently considered as HONO sources are reactions 1 and 2, which are both aqueous-phase mechanisms but could also occur on surfaces within adsorbed water:
2NO2 + H2O w HONO + HNO3
3.0 × 107 M-1 s-1 (2) (1)
NO2 + HSO3- w NO2- + HSO3 (radical)
3.0 × 105 M-1 s-1 (3) (2a)
2HSO3 + H2O w H2SO3 + H2SO4
5.0 × 105 M-1 s-1 (4) (2b)
* Corresponding author phone: +41/56/310-4049; fax: +41/56/ 310-4435; e-mail:
[email protected]. † Paul Scherrer Institute. ‡ University of Bern. 10.1021/es015673b CCC: $22.00 Published on Web 01/12/2002
2002 American Chemical Society
Reports on the kinetics of reaction 1 on the surfaces of various glass types, Teflon, and synthetic fiber carpet differ by orders of magnitude (5-11), but it is unlikely that they may sustain a significant HONO source in the troposphere. A further reaction of NO and NO2 with water (12) has not been considered to be an important process under atmospheric conditions (13). Since both the HONO concentrations and the HONO to NOx ratio generally scale with the degree of pollution, combustion processes have been suspected to be involved in HONO formation. Direct emissions of HONO from automobile exhaust are usually below 1% of nitrogen oxides (14-16) and are thus unlikely the major HONO source. Therefore, considerable research effort was directed toward investigating the reaction of NO2 with soot, another exhaust gas component (17-22). The main conclusion from these studies was that the soot surface contains functionalities substituted to the large carbonaceous structures or individual condensed organic species, which are able to reduce NO2 to HONO with yields up to 100% in a reaction of the type:
NO2 + {C-H}red w HONO + {C}ox
(3)
where {C-H}red stands for a surface-bound species or class of compounds. The presence of a large variety of functionalities associated with soot has been demonstrated by various authors (23-25). Stadler and Rossi (22) were able to extract {C-H}red from flame soot in a polar solvent, but a quantitative analysis showed that soot from all sources investigated contains by far not enough {C-H}red-type species to convert significant amounts of NO2 in the troposphere (19, 20, 26). The starting point of the present investigation was that if soot contains low vapor pressure organic compounds able to reduce NO2 to HONO, it will also be likely that exhaust gas contains more volatile organic species with the same ability. Whether a semivolatile organic compound within a complex gas mixture remains in the gas phase or partitions to the condensed phase also depends strongly on its water solubility or its affinity to condensed organic species. Although several reactions of type 3 are known to occur in aqueous solution (27-29), their importance in atmospheric chemistry has not yet been assessed, nor is it known which of the tremendous number of species in combustion exhaust (30) could be candidates for such a reaction. Therefore, the secondary HONO formation by reaction 3 with volatile reactants provided by diesel exhaust (as an important source of combustion gases) was assessed in three types of experiments: first, hot diesel exhaust was condensed on tube sections at various temperatures; off-line, these tube sections were exposed to NO2 to test for HONO formation and hence characterize the volatility of {C-H}red. Second, in an on-line experiment, these species were continuously absorbed into a flowing water film and reacted with varying concentrations of NO2 in order to determine the kinetics of reaction 3. In a further on-line experiment, a large glass surface was continuously exposed to the exhaust gas (including {C-H}red and NO2), and the extent of NO2 to HONO conversion was monitored downstream of the glass surface.
Experimental Section Diesel exhaust was generated using an electrical power generator (Kubota GL-4500S, maximum load 4.5 kW) without exhaust treatment device. The engine emissions have been well-characterized by Arens et al. (19) using standard diesel fuel. The on-line experiments (interaction with water and glass surface) were performed using so-called biodiesel with VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
677
FIGURE 1. Schematic design of the exhaust transfer line from the 4.5-kW diesel engine to the WWFT. For exhaust condensate sampling, the wetted wall flow tube was replaced by condensation columns (without adding synthetic air). The insert shows the heated injection capillary reaching into the WWFT (aqueous phase experiment). For the third experiment, the capillary was slightly pulled back, and a glass fiber disk was placed between the capillary and the WWFT. a roughly 10 times lower sulfur content (50 ppm) than standard diesel fuel in order to minimize the pathways of sulfur oxidation of which reaction 2 is only a minor one (3). Since the NOx emission indices are not affected by the sulfur content, the dilution ratios of the present experiments were calculated using the emission indices of NOx determined by Arens et al. using standard diesel (19). This seems to be an adequate procedure since the NOx concentrations varied by less than 15% with respect to fuel consumption at a given engine load. The aim of the first experiment was to identify the temperature range at which {C-H}red undergoes phase transfer. A fraction of 5 L min-1 of the undiluted and unfiltered exhaust passed through a noninsulated aluminum tube of 330 cm length and 8 mm i.d., passively cooled by ambient air. A stable temperature gradient along the tube was established due to heat transfer. Transfer of {C-H}red to the condensed phase occurs depending on the temperature of a given tube section. Although the relative humidity was kept below 100% up to about 310 cm tube length, solvation of polar {C-H}red in water adsorbed on the tube walls close to the tube end might also occur. The Reynolds number at the tube exit was estimated to be 880 (5 L min-1 flow, dynamic viscosity of 0.152 cm2 s-1 for air at 20 °C). Because of the higher temperature at the tube inlet, the flow regime along the tube was probably not perfectly laminar. Moreover, turbulences may have been induced by the pressure variations typically occurring in tail pipes. Therefore, some particles may have been deposited inside the tube. The tube was exposed to exhaust during 2 h at 3 kW engine load. After sampling, the tube was cut into sections of different lengths. In the laboratory, each section was then exposed to a flow of 13N-labeled NO2 at 95 ppb and 30% RH (31). The total NO2 concentration was monitored using a chemiluminescence detector (Monitorlabs model 8841). A trap coated with sodium carbonate was placed downstream of the exposed tube section, and the decay of the radioactive HO13NO (half-life 9.96 min) was monitored using a γ-detector placed next to the trap. This setup was identical to that used by Arens et al. (19), except that the soot samples were replaced by the tube sections. The radioactive tracer technique is very well-suited for monitoring HONO concentrations at low ppb levels (19). Figure 1 shows the scheme for the on-line experiments in which the exhaust gas was transferred to a wetted wall flow tube (WWFT) in order to determine the aqueous phase 678
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
kinetics of reaction 3 as well as the yield of HONO from the interaction of diesel exhaust with a glass surface. The exhaust temperature was 180 and 230 °C at 1 and 3 kW, respectively, when about 100th of the exhaust (5 L min-1) was diverted toward the transfer line and diluted by about 1 L min-1 preheated synthetic air to decrease the dew point and to keep the overall residence time below 2 s. A first T-connector open to the atmosphere before the admission valve (both stainless steel) allowed the reduction of the pressure inside the transfer line and hence the adjustment of the dilution ratio. From the admission valve, the gas passed a particle filter (glass fibers in a stainless steel housing, Stroehlein, at 200 °C) connected to a 4 mm i.d. PFA tube of 4 m length and heated to 100 °C. At the end of the transfer line, the diluted exhaust contained less than 2 × 104 particles per cm3 as confirmed by a differential mobility analyzer coupled to a particle counter. Before injection into the WWFT via a heated capillary (at 100 °C, see insert of Figure 1), the backing pressure was further reduced by a second T-connector open to the atmosphere. The bottom part of the capillary had an i.d. of 0.2 mm and allowed for a throughput in the range of 2-3 mL min-1, which after dilution with synthetic air yielded an optimum dilution ratio of about 200 as determined based on the NOx concentration of 500 and 1000 ppb at 1 and 3 kW load, respectively (see below). In the aqueous-phase kinetic experiments, the NO2 concentration was varied by adding up to 2000 ppb NO2 in the dilution air before the whole gas mixture was fed into the WWFT for interaction with its water surface. The WWFT has been described in detail by Zellweger et al. (32). In short, the gas sample is drawn at 0.48 L min-1 through a vertically mounted glass tube of rectangular shape and fed into a NOx analyzer (Monitorlabs model 8841). A flow of 0.82 mL min-1 ultrapure water (18 MΩ) is injected at the top of the glass tube and forms a homogeneous water film flowing down the rough inside glass surface of 240 cm2 surface area. At the tube bottom, the water film (effluent) is pumped off and preconcentrated on two alternating ion traps (TAC-LP1, Dionex) using a peristaltic pump. By switching the 10-port valve (Supelco) on which the two preconcentrator columns were mounted, the anions were eluted alternatively using a carbonate eluent and analyzed for nitrite, nitrate, sulfite, and sulfate by ion chromatography (AS4A-SC, Dionex; 2 mL min-1) with a time resolution of 5.5 min. Because of the high exhaust dilution ratio of about 200, the effluent had a pH close to 6 so that efficient HONO sampling was ascertained without adding carbonate buffer (32). The system was calibrated using appropriate dilutions of 1000 ppb by weight (Merck) standard solutions. The residence time of 25 ( 5 s of the water film in the flow tube was measured by injecting a burst of ions into the water supply and measuring the breakthrough time by conductivity detection. The feasibility of this apparatus for kinetic measurements was checked by confirming the rate constant of reaction 1 (2) within the uncertainty of the residence time. In the third experiment, the WWFT was not used for kinetic measurements but to monitor the HONO yield from the interaction of exhaust with a glass surface. In this case, a disk of compressed glass fibers (GF6, Schleicher & Schuell, 25 mm diameter, ca. 480 cm2 BET surface area) was placed between the injection capillary and the WWFT. Moreover, the relative humidity of the dilution air was varied instead of the NO2 concentration as in the second experiment. Fourth, in an attempt to combine experiments one and three, unfiltered hot exhaust was first sampled in a condensation column at 4 °C rather than the temperature resolving tube described above, resulting in a clear aqueous solution of pH 2.4. Then, 50 µL of this solution () condensate) was given on a glass fiber disk and exposed to a humidified NO2/air flow. The resulting HONO was monitored like in experiment three using the WWFT but this time off-line.
FIGURE 2. Rate of HONO formation determined off-line for the condensation tube sections (right axis, black bars) at 95 ppb NO2 and 30% RH. The tube sections correspond to decreasing temperature (curve T, gray solid line, left axis) and increasing relative humidity (curve RH, black solid line, left axis) during sampling.
Results Temperature-Resolved HONO Formation in an Exhaust Condensation Tube. Figure 2 shows the HONO formation profile as a function of temperature obtained by exposing the different tube sections to NO2. The HONO formation is increasing as a function of decreasing sampling temperature and is peaking around 60 °C near the end of the tube. This HONO formation was attributed to reaction 3. Although this experiment does not allow us to assess unambiguously which of the vapor pressure of {C-H}red or its polar nature affects more its partitioning to the condensed phase, we conclude that the most important species involved in reaction 3 are depositing at a temperature below 100 °C and that these species are at least stable enough to partially survive the exposure to large NO2 concentrations as well as the period of time (a few hours) between sampling and analysis. Interaction of Diesel Exhaust with a Water Surface. In this on-line experiment, the pseudo-first-order kinetics of reaction 3 was assessed by keeping the supply of (watersoluble) {C-H}red from the exhaust constant while the NO2 concentration was varied. The critical task consisted in differentiating the contributions to nitrite due to reactions 1-3 and those due to dissolution of primary HONO emitted by the engine. Figure 3 shows the measured concentrations of nitrite as a function of NO2 added. As expected, the nitrite concentration increases nonlinearly as a function of NO2 due to the second-order component contributed by reaction 1. Measured nitrate is exclusively associated with reaction 1 and not affected by primary nitric acid emissions from the engine since the nitrate signal did not change when the exhaust gas was switched off and the NO2 concentration kept constant. Therefore, the contribution of reaction 1 to measured nitrite can be accounted for by subtracting nitrate from the measured nitrite. Oxidation of dissolved sulfur(IV) to sulfate is known to be affected by various oxidants such as O2, O3, and H2O2 as well as metal ions (3) besides NO2. As even without adding NO2 no HSO3- was observed, the main fate of sulfite seemed to be one of these additional oxidation mechanisms. Moreover, given the kinetics of reaction 2a (3), at most 10% of the observed 70 nmol L-1 sulfate could originate from sulfite oxidation by NO2 during the short residence time in the WWFT. Therefore, a conservative estimate for the contribution of reaction 2 is obtained by subtracting the total sulfate signal. The remainder of the nitrite signal (nitrite - nitrate - sulfate) in Figure 3 shows an intercept with the y-axis at 180 nmol L-1, which is the contribution of the primary HONO emission. (Since reaction 3 is expected to be linear in NO2, its contribution at 0 ppb NO2 is assumed to be null.) Primary HONO amounts to a HONO:NOx ratio of 0.0067 or, relating the NOx emission
FIGURE 3. Aqueous-phase kinetics of diesel exhaust with dissolved NO2. The ion concentrations are plotted against the gas-phase NO2 concentration (NO2 from engine plus added NO2) detected at the exit of the WWFT. The triangles correspond to the total nitrite signal, and the dots are obtained by subtraction of the nitrate and sulfate signal from the nitrite signal. Their slope of 0.13 nmol/L ppb yields the pseudo-first-order rate constant of reaction 3. The extrapolated intercept of the dots with the y-axis corresponds to the primary HONO represented as a dotted line. The hatched surface is attributed to nitrite formation due to reaction 3 and correspondingly to secondary HONO. indices to the fuel consumption via exhaust flow (see Table 1 in Arens et al.; 19), 71 mg of HONO/kg of diesel fuel. Although some uncertainty is induced through the use of these emission relations, this number is in good agreement with the primary HONO emission indices obtained by Kurtenbach et al. (15) (0.0065 and 88 mg of HONO/kg) in a road traffic tunnel. From the linear slope of the remaining nitrite, we obtain the pseudo-first-order rate constant for reaction 3 equivalent to k3‚{C-H}red ) 0.65 s-1 assuming a Henry constant for NO2 of 0.014 M bar-1 (2) and a fast gasliquid transfer of NO2 (phase mixed regime). As the slope does not level off up to 2000 ppb NO2 or 250 nmol of nitrite/L (primary HONO subtracted), we conclude that {C-H}red is in excess as compared to dissolved NO2 and must correspond to at least the same amount as primary HONO (180 nmol L-1) equivalent to a fraction of at least 0.009 of total NOx or 98 mg/kg of burnt fuel. The NO2 concentration was not further increased because the uncertainty with subtracting the contribution by the second-order reaction 1 would become too large. The experiment was also carried out at engine loads other than 3 kW; no load dependence was observed. Interaction of Diesel Exhaust with a Glass Surface. Before the diesel exhaust was admitted, several blank experiments were performed with the glass fiber disks. When the disks as obtained from the manufacturer (no special cleaning was applied) were mounted at the inlet of the WWFT and exposed to NO2 (50 and 400 ppb, 62% RH), a small change in nitrite (less than 10%) was observed, which was near the detection limit indicating that the formation of gas-phase HONO due to reaction 1 was very small. Nevertheless, the extent of reaction 1 could be estimated after prolonged exposure to NO2 during 14 h. Ion chromatographic analysis then showed the presence of nitrate, as most of the HNO3 formed due to reaction 1 remained on the glass surface. This allowed us to get an estimate for the fraction of NO2 surface collisions resulting in reaction, γrxn, amounting to 10-8. This value is 2 orders of magnitude lower than the value observed by Kleffmann et al. (5) for initial uptake on a clean quartz surface but in agreement with other measurements on borosilicate glass (6-9), on nylon (10), and on vinyl-coated wallpaper (11). The same low reactive uptake was observed when 30 µL of 0.2 N sulfuric acid solution was given on the glass fibers and exposed to 104 ppb NO2 at 89% RH. In the last blank experiment, the glass surface was concurrently exposed to about 10 ppb SO2 and 400 ppb NO2 (at low and high RH), VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
679
less than 10% when storing the condensate for 1 week. We assume that the above value is an order of magnitude lower as compared to the one obtained in the on-line experiment due to the highly oxidative conditions during sampling since the condensate was sampled at 46 ppm NO2. Nevertheless, this indicates that the HONO-forming species may be preserved for several hours, particularly in acidic media.
Discussion
FIGURE 4. Glass fiber disk exposed to hot diesel exhaust and the HONO formation relative to NO2 was monitored using the WWFT and a NOx analyzer. The plotted HONO corresponds to nitrite after subtraction of nitrate and primary HONO of which the latter one was determined in the previous experiment. The average NOx mixing ratio was 569 ( 143 ppb corresponding to a dilution ratio of 220 ( 55. The average value of 0.023 ( 0.006 is represented by the line. The different symbols correspond to different days. which again did not lead to a significant increase of the HONO signal. After these blank experiments, the diesel exhaust (1-kW engine load) was passed over a glass fiber disk, and the formation of gas-phase HONO was monitored on-line as nitrite in the effluent of the WWFT. As in the previous experiment, reactions 1-3 may also take place here. Therefore, the nitrate signal due to reaction 1 was subtracted likewise. However, no sulfite or sulfate was observed so that in this case reaction 2 does not contribute to the nitrite signal in the WWFT. Primary HONO contained in the exhaust before reaction on the glass surface is also detected as nitrite, as shown in the previous experiments, and its fraction of 0.0067 relative to NOx is also subtracted. The remaining nitrite signal corresponds to secondary HONO formed on the glass fiber surface and in the water film according to reaction 3 (reactions 1 and 2 on the glass surface can be excluded, see blank experiment). In Figure 4, the secondary HONO is expressed as a fraction of NOx emitted by the engine. This secondary HONO to primary NOx ratio is independent of the relative humidity and amounts to 0.023 ( 0.006 corresponding to 244 mg of secondary HONO/kg of fuel burnt. The only NO2 available for reaction was supplied directly from the exhaust, i.e., no NO2 was added separately, and amounted to 265 ( 83 ppb. In an experiment not shown here, additional NO2 was admitted but did not affect the absolute HONO yield. This means that the extent of HONO formation was controlled by the on-line supply of new {C-H}red reactants with the exhaust gas and not by the NO2 concentration. The scatter of data plotted in Figure 4 seems to be associated with the day-to-day variation of the ambient engine operating conditions. Combining the pseudo-first-order rate constant k3‚[{CH}red] ) 0.65 s-1 determined in the aqueous-phase experiments with the minimum amount of {C-H}red emission given by the HONO:NOx ratio obtained on the glass surface yields an upper limit for the apparent second-order rate constant k3 of 106 M-1 s-1 for reaction 3 in the aqueous phase at neutral pH conditions. This number corresponds to a rough estimate since the exact nature of the reaction is still not clear. Finally, when 50 µL of freshly sampled diesel condensate (clear solution, i.e., without soot particles) collected from the exhaust at 4 °C at 1-kW engine load were given on the glass fibers and exposed during 3 h to 110 ppb NO2 at high relative humidity (more than 90%), an equivalent yield of 21 mg of HONO/kg of burnt fuel was formed. However, the HONO formation capacity of the condensate decreased to 680
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
Both off-line and on-line experiments presented above clearly show that diesel exhaust contains a water-soluble class of compounds that undergoes transfer to the condensed phase below 100 °C. This class of compounds is able to reduce NO2 to nitrite in the aqueous phase or HONO on a surface, similar to {C-H}red in reaction 3. Moreover, its HONO formation capacity of 244 mg of HONO/kg of burnt fuel is roughly 50 times more important than the one of its low-volatility counterparts associated with soot (4.7 mg/kg of fuel) determined by Arens et al. (19). The two off-line experiments also show that {C-H}red is sufficiently stable to remain reactive on time scales of hours to days. As tail pipe emissions generally occur at temperatures above 100 °C, it seems clear that the most significant part of the {C-H}red type species is emitted as gas-phase species. Whether under realistic conditions, these species partition to an aerosol, fog droplets, or ground surface sensitively depends on ambient conditions such as temperature and humidity as well as the initial dilution ratio (from a few hundred up to several thousands). It is therefore reasonable that in the study by Arens et al. (19) the semivolatile members of {C-H}red were not condensed on the soot particles since the particles contained in the exhaust were diluted into dry nitrogen and rapidly passed through denuders to remove all reactive gas-phase species including NOx and volatile organic compounds. The same is probably true for flame soot investigations (18, 21, 22) in which soot was sampled in the hot part of a flame and then placed into a vacuum reactor for analysis. If {C-H}red ever had a chance to condense on the sample surface, it has most probably evaporated when the sample was placed under vacuum. The desorption processes of semivolatile organic compounds from diesel soot have recently been reviewed by Strommen and Kamens (33). Both on-line experiments presented here indicate that when diesel exhaust interacts either with a solid or with a liquid surface, secondary HONO or its precursor nitrite can be formed through reaction 3 leading to HONO:NOx ratios up to 0.023. This HONO:NOx ratio is at least a factor of 3 larger than the primary HONO emission from diesel engines as recently summarized by Kurtenbach et al. (15) and confirmed in the present experiments. In the road tunnel study by Kurtenbach, however, significant HONO formation was observed when NO2 interacted with material scratched from the tunnel wall. The HONO formation on the porous tunnel residue decreased rapidly and was interpreted by the authors as surface deactivation byproducts. In fact, reactivation of this material containing 13.5 wt % carbon could be achieved by flushing synthetic air over it. The authors attributed the HONO formation to reaction 1 rather than a noncatalytic reaction similar to reaction 3. They calculated an initial reaction probability of several 10-6, based on the surface area of the tunnel wall from which the material was scratched. Given the porous nature of the tunnel residue, this surface area corresponds to a lower limit; therefore, the above γrxn seems very large to be attributed to reaction 1, especially when compared to our findings using borosilicate fibers as substrate. An alternative interpretation of the rapidly decreasing reaction rate would be by invoking reaction 3
and reactivation through diffusion from the porous material similarly to observations made on bulk soot (18). Similarly, Kleffmann et al. (16) observed an increase of about 1.5 ppb HONO/h at about 10 ppb NO2 after transferring diesel exhaust from a commercial engine into a smog chamber at 50% RH (primary HONO subtracted). This HONO formation was only observed under humid conditions (50% RH) and was therefore attributed to reaction 1 occurring on the chamber walls. It remains however questionable whether such a significant increase can be explained by reaction 1 alone without invoking a mechanism such as reaction 3 due to {C-H}red adsorbed on the chamber walls. It is beyond the scope of this work to identify the {CH}red-type species or class of compounds. The relevant literature reveals substances such as OH- and OCH3substituted aromatics (28) as well as amines (27) undergoing electron transfer reactions with NO2 yielding nitrite in the aqueous phase. The analysis of the exhaust condensate by liquid and gas chromatography coupled to mass spectrometry (34) revealed about 0.54 mg of phenols (phenol as well as several alkylphenols and their nitro derivatives)/kg of fuel burnt, which is at least 2 orders of magnitude lower than necessary to explain our results with this class of compounds. We therefore suggest that other substances such as complex phenols (dihydroxybenzenes, guaiacol, syringol, and corresponding derivatives from polyaromatic hydrocarbons) as well as amines might be involved. Assessing the amount of total hydrocarbon (THC) in the exhaust using a flame ionization detector (J. U. M. Engineering, Munich model VE5) yields THC:NOx ratios of 36 and 7% by mass at 1- and 3-kW engine load, respectively, of which only the value at high load is in agreement with emissions from high-duty vehicles (35). This means, on one hand, that our power generator engine is not necessarily representative of current in-use diesel engines. On the other hand, either many hydrocarbons species are involved in heterogeneous HONO formation according to reaction 3 or the HONO formation mechanism is still more complex than expressed by the simple scheme depicted by reaction 3. However, the main conclusion of this paper that semivolatile organic compounds contained in combustion exhaust may constitute a significant source of HONO via reaction 3 is not questioned, since in the liquidphase experiment the HONO formation due to reaction 3 was at least as important as the primary HONO. Since the identity of the range of NO2 reducing species is unknown, the determined second-order rate constant k3 ) 106 M-1 s-1 corresponds to the overall rate constant when all the {C-H}red species are lumped together into one class. If OH-substituted aromatic compounds are involved, the deprotonated form is the most reactive partner, which renders these reactions strongly pH dependent. The pKa values of such species are mostly above 8, so that it would seem reasonable that k3 is at least 2 orders of magnitude lower in a neutral solution (29). On the other hand, aromatic amines might also efficiently react with NO2 under acidic conditions (27). What might be the atmospheric implications of the present results? For an air mass with freshly emitted diesel exhaust containing 40 ppb NOx (20 ppb NO2) corresponding to 0.92 ppb {C-H}red and with a sufficient water content to sustain the kinetics of reaction 3, we expect an initial rate of nitrite formation of 0.64 ppb/h or a half-life for {C-H}red of about 1 h at 20 ppb NO2. This result means that all the {C-H}red contained in diesel exhaust might have reacted with NO2 to nitrite within a few hours following emission. The partitioning of nitrite into the gas phase as HONO depends on the acidity of droplets or aerosol particles. The pH of droplets correlates with the content of liquid water and neutralizing species of the air parcel and varies strongly as the parcel moves and is transformed over time. Thus, it may well be that reaction 3
takes place in a neutral droplet, while it releases HONO into the gas phase as soon as the pH drops below about 3.5 upon drying.
Acknowledgments We thank Norbert Heeb and Martin Kohler of EMPA (The Swiss Federal Institute for Materials Testing and Research) for the analysis of the diesel condensate by liquid and gas chromatography coupled to mass spectrometry. We also thank the Federal Office for Education and Science for their funding within DIFUSO (ENV4-CT97-0390) and NITROCAT (EVK2-CT-1999-00025), two EC projects under the fourth and fifth Framework Program, respectively.
Literature Cited (1) Lammel, G.; Cape, J. N. Chem. Soc. Rev. 1996, 361-369. (2) Cheung, J. L.; Li, Y. Q.; Boniface, J.; Shi, Q.; Davidovits, P.; Worsnop, D. R.; Jayne, J. T.; Kolb, C. E. J. Phys. Chem. A 2000, 104, 2655-2662. (3) Lee, Y.-N., Schwartz, S. E. In Precipitation Scavenging, Dry Deposition and Resuspension; Pruppacher, H. R., Semonin, R. G., Linn, W. G. N., Eds.; Elsevier: New York, 1983; Vol. 1, pp 453-470. (4) Lyon, R. K.; Cole, J. A. Combust. Flame 1989, 77, 139-143. (5) Kleffmann, J.; Becker, K. H.; Wiesen, P. Atmos. Environ. 1998, 32, 2721-2729. (6) Hatakeyama, S.; Akimoto, H.; Sakamaki, F. Int. J. Chem. Kinet. 1983, 15, 1013-1029. (7) Pitts, J. N.; Sanhueza, E.; Atkinson, R.; Carter, W. P.; Winer, A. M.; Harris, G. W.; Plum, C. N. Int. J. Chem. Kinet. 1984, 16, 919-939. (8) Jenkin, M. E.; Cox, R. A.; Williams, D. J. Atmos. Environ. 1988, 22, 487-498. (9) Svensson, R.; Ljungstro¨m, E.; Lindqvist, O. Atmos. Environ. 1987, 21, 1529-1539. (10) Perrino, C.; De Santis, F.; Febo, A. Atmos. Environ. 1988, 22, 1925-1930. (11) Wainman, T.; Weschler, C. J.; Lioy, P. J.; Zhang, J. F. Environ. Sci. Technol. 2001, 35, 2200-2206. (12) Calvert, J. G.; Yarwod, G.; Dunker, A. M. Res. Chem. Intermed. 1994, 20, 463-502. (13) Andres-Hernandez, M. D.; Notholt, J.; Hjorth, J.; Schrems, O. Atmos. Environ. 1996, 30, 175-180. (14) Kirchstetter, T. W.; Harley, R. W.; Littlejohn, D. Environ. Sci. Technol. 1996, 30, 2843-2849. (15) Kurtenbach, R.; Becker, K. H.; Gomes, J. A. G.; Kleffmann, J.; Lo¨rzer, J. C.; Spittler, M.; Wiesen, P.; Ackermann, R.; Geyer, A.; Platt, U. Atmos. Environ. 2001, 35, 3385-3394. (16) Kleffmann, J.; Heland, J.; Kurtenbach, R.; Lo¨rzer, J. C.; Wiesen, P.; Ammann, M.; Gutzwiller, L.; Rodenas Garcia, M.; Pons, M.; Wirtz, K.; Scheer, V.; Vogt, R. Submitted for publication in Environ. Sci. Technol. (17) Ammann, M.; Kalberer, K.; Jost, D. T.; Tobler, L.; Ro¨ssler, E.; Piguet, D.; Ga¨ggeler, H. W.; Baltensperger, U. Nature 1998, 395, 157-160. (18) Gerecke, A.; Thielmann, A.; Gutzwiller, L.; Rossi, M. J. Geophys. Res. Lett. 1998, 25, 2453-2456. (19) Arens, F.; Gutzwiller, L.; Baltensperger, U.; Ga¨ggeler, H. W.; Ammann, M. Environ. Sci. Technol. 2001, 35, 2191-2199. (20) Kleffmann, J.; Becker, K. H.; Lackhoff, M.; Wiesen, P. Phys. Chem. Chem. Phys. 1999, 1, 5443-5450. (21) Longfellow, C. A.; Ravishankara, A. R.; Hanson, D. R. J. Geophys. Res. 1999, 104, 13833-13840. (22) Stadler, D.; Rossi, M. J. Phys. Chem. Chem. Phys. 2000, 2, 54205429. (23) Chughtai, A. R., Jr.; Welch, W. F.; Akhter, M. S.; Smith, D. M. Appl. Spectrosc. 1990, 44, 294-298. (24) Kirchner, U.; Scheer, V.; Vogt, R. J. Phys. Chem. A 2000, 104, 8908-8915. (25) Al Abadleh, H. A.; Grassian, V. H. J. Phys. Chem. A 2000, 104, 11926-11933. (26) Aumont, B.; Madronich, S.; Ammann, M.; Kalberer, M.; Baltensperger, U.; Hauglustaine, D.; Brocheton, F. J. Geophys. Res. 1999, 104, 1729-1736. (27) Saltzman, B. E. Anal. Chem. 1954, 26, 1948-1955. (28) Forni, L. G.; Mora-Arellano, V. O.; Packer, J. E.; Willson, R. L. J. Chem. Soc., Perkin Trans. 2 1986, 1-6. (29) Alfassi, Z. B.; Huie, R. E.; Neta, P. J. Phys. Chem. 1986, 90, 41564158. VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
681
(30) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 1999, 33, 1578-1587. (31) Kalberer, M.; Tabor, K.; Ammann, M.; Parrat, Y.; Weingartner, E.; Piguet, D.; Ro¨ssler, E.; Jost, D. T.; Tu ¨ rler, A.; Ga¨ggeler, H. W.; Baltensperger, U. J. Phys. Chem. 1996, 100, 15487-15493. (32) Zellweger, C.; Ammann, M.; Hofer, P.; Baltensperger, U. Atmos. Environ. 1999, 33, 1131-1140. (33) Strommen, M. R.; Kamens, R. M. Environ. Sci. Technol. 1999, 33, 1738-1746.
682
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
(34) Heeb, N. Swiss Federal Institute for Materials Testing and Research (EMPA), Du ¨ bendorf, Personal communication. (35) Yanowitz, J.; McCormick, R. L.; Graboski, M. S. Environ. Sci. Technol. 2000, 34, 729-740.
Received for review August 31, 2001. Revised manuscript received November 26, 2001. Accepted November 29, 2001. ES015673B
