Environ. Sci. Technol. 2005, 39, 7609-7615
A Novel Method for Determination of Size-Resolved, Submicrometer Particle Traffic Emission Factors S A R A J A N H A¨ L L * A N D MATTIAS HALLQUIST Department of Chemistry, Atmospheric Science, S-412 96 Go¨teborg, Sweden
A novel approach to determine size-segregated particle number emission factors for traffic is presented. It was proven that using limited data sets (800-2000 samples) statistically significant emission factors from road traffic can be extracted. In this study data from four sites were used for calculating emission factors (rural and urban roadside, urban rooftop, and urban background). The measurements were performed using SMPS/DMPS (scanning or differential particle sizers) from TSI and commercial gas analyzers. Describing the particle concentration as a ratio to an exhaust trace gas, e.g. NOx, the dilution effect will be minimized. This ratio is easily compared among different studies. By knowledge of the emission factor of the chosen trace gas the emission ratio can be converted to an emission factor for particle numbers of defined particle sizes. For the presented method only one measurement site is needed, where the difference between high and low (background) traffic exposure is used. To define high and low traffic exposure, the best result was obtained using high ratio of [NO] to [NO2] and low [NOx], respectively. Emission ratios for 10-100-nm particles at two road sites, one high-speed 90-kmph rural case and one urban, slower, and more congested situation, were determined to (35(15) × 1014 and (24(8) × 1014 particles per mole NOx, respectively.
Introduction In recent years the health problem of atmospheric particles has been earning increasing attention from the public and policy makers. Regulations are focusing on inhalable particles, i.e., PM10, the mass of particles smaller than 10 µm in diameter. Still, the research has shown that a large part of the health problems are in fact associated with ultrafine particles, as the size of the particles decides how far into the lung they will penetrate (1). Most of the ultrafine particles, i.e., particles below 100 nm in diameter, are deposited in the deepest end of the lung, or the alveolar region. Those particles are more numerous, but do not contribute significantly to the total mass of particles, e.g., PM10. If the focus of the health concern is on the smaller particles, the physical measure of particles should be number density. How the concentration of particles in the surrounding air is affected by changes in emissions is usually calculated by models, and the result is thus directly affected by the emission data used in the model. These emissions are given as emission factors, i.e., emitted amounts related to traffic work, fuel sold, or other variables. The access to emission factors for * Corresponding author e-mail:
[email protected]. 10.1021/es048208y CCC: $30.25 Published on Web 08/25/2005
2005 American Chemical Society
size-segregated particles, or even total number of particles, is limited (2, 3). Ultrafine particles in urban areas are mostly emitted by traffic. Traffic can be regarded as either a large number of individual point sources or as an average over the traffic fleet, i.e., a line or area source. The point sources are affected by fuel, vehicle type, individual driving patterns, etc. The emissions from specific vehicles are frequently measured in laboratories, as dynamometer studies, where specified driving patterns are used to mimic real-world driving conditions (4). For a typical traffic source (e.g., a road segment) the vehicle to vehicle variation cannot be distinguished. The average line source can be estimated using data from individual vehicles if the traffic composition and driving pattern are known. The line source can also be measured directly, either onroad, e.g., following the traffic with a measuring van (5, 6), or stationary at the road side (7-10). Normally the concentrations of pollutants are measured, and in order to get the emissions the dilution process has to be established. One way to know the dilution is to measure inside tunnels, where the dilution rate can be determined easily using a tracer. The tracer is either emitted by the same source as the exhaust, e.g. CO2, or emitted close to the source, which is often done using SF6 (11-14). The tracer should be a stable compound that does not transform quickly enough to change its concentration by any process other than dilution. The dilution of the tracer is then assumed to equal the dilution of the exhaust. At roadside sites meteorological models are used to calculate the dilution ratio (14, 15). A large number of emission factors for gaseous species are obtained by the described processes (3, 16). When using these methods on number of particles emitted, some additional difficulties arise for the dynamic and numerous particles below 50 nm in diameter. At low dilution ratios, the number of particles decreases by coagulation and the surfaces available for gas-to-particle conversion decrease. Limited dilution gives limited number of particles and larger particles, which has been observed both in low-diluted dynamometer studies and in tunnels (3). The aerosol can also be affected by other atmospheric processes such as deposition and evaporation. The process being the most important is the one with the shortest time scale. At road side the dilution time scale is often the shortest (17-19). In this study ultrafine particles (ufp) were size-selected and counted, using scanning or differential mobility particle sizers (SMPS or DMPS) from TSI at four different sites in the Go¨teborg area. The sites were located both close to heavy traffic and in background areas. The concentrations of particles and gases, meteorological parameters, and at one site traffic variables were measured. These data were used to find a relationship between emissions of submicrometer particles, by number, and emissions of a gaseous compound, e.g. nitrogen oxides, nitric oxide, or carbon monoxide. In this way a size-distributed emission ratio for number of submicrometer particles can be achieved, relating the particle emissions directly to gas emissions. In addition to being used to normalize the particle emission, the concentrations of the trace gaseous species were used to distinguish between high and low traffic exposure of the site. The various options in the use of the trace gases are evaluated and a final standard method is recommended. This method can be applied directly to other data sets and the use of ratios facilitates a direct comparison between studies. The emission factors obtained in this study are compared to existing particle emission factors. VOL. 39, NO. 19, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7609
TABLE 1. Average Concentrations and Meteorological Parameters during the Campaigns (With the Standard Deviation of the Data)
total N ( × 103 cm-3) ufp (% of total N) NO (ppb) NO2 (ppb) NO/NO2 CO (ppb) temp (°C) wind speed (m s-1) a
Measured at Tagene.
b
Surte May 2001
GU May-June 2003
Femman February-March 2004
Garda May 2004
5.1 ( 4.4 90 ( 8 8.7 ( 14 9.0 ( 7.3 0.81 ( 0.68 n.a. 16 ( 8a 1.8 ( 1.4a
5.5 ( 2.8 68 ( 13 2.6 ( 3.6 5.3 ( 2.9 0.47 ( 0.34 n.a. 13 ( 3b 3.5 ( 1.5b
4.6 ( 3.7 90 ( 6 16 ( 18 20 ( 8 0.58 ( 0.88 189 ( 162 0.6 ( 0.9 3.4 ( 1.9
7.0 ( 5.7 91 ( 6 23 ( 38 16 ( 9 1.06 ( 1.10 121 ( 174 11 ( 3b 2.1 ( 1.0b
Measured at Femman.
FIGURE 1. Diurnal variation of ultrafine particle number concentration and nitric oxide at the four sites. The arrows indicate to which axis the graph belongs.
Experimental Sites and Instrumentation. Emission factors of submicrometer particles were extracted from measurement data collected at four sites in the Go¨teborg region (58 °N and 12 °E), which were selected to illustrate different types of locations. The sites were one rural roadside site with commuting traffic (Surte), one urban background site (Go¨teborg University, GU), one urban rooftop site (Femman), and one densly trafficked urban roadside site (Garda). Nitrogen oxides were measured by chemiluminiscence (Eco Physics CLD700AL), while CO was measured by infrared absorption (Environnement CO12Module or Unor 610). Particle number size distributions were measured by either a differential mobility particle sizer, DMPS TSI-3932 (Surte, Femman, and Garda) or a scanning mobility particle sizer, SMPS TSI-3936 (GU and Garda). The DMPS and SMPS were intercalibrated. The characteristics of the four sites are summarized in Table 1. The rural road site (Surte) has been described previously (2, 7). It is located at the curbside of a straight, four-lane road with few emitters other than the traffic on the road itself. The traffic, 18,000 vehicles per day, is running steadily at approximately 90 kmph, and the speed limit on the road is shifting between 70 and 90 kmph close to the measurement site. Generally, this site is not heavily polluted but has a clear diurnal variation. The urban road site (Garda) is in general a more polluted site, where, in addition to the traffic on the road itself, nearby roads and other urban sources contribute to the measured 7610
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 19, 2005
concentrations. The traffic intensity is larger, i.e., 90,000 vehicles per day, and congestions are frequent, especially during rush hours. The speed limit is 70 kmph, but the actual speed is often limited by the congestions. The measurements took place 10 m from the curbside of the road. The urban rooftop (Femman) at 30 m height above the ground level is affected by several roads in the center of Go¨teborg. However, the traffic-related particles observed at Femman have spent more time in the atmosphere than the recently emitted particles observed at the road sites. This was shown by, e.g., lower NO to NO2 ratio. Noteworthy is that the measurements at Femman took place during February and March with low morning temperatures and frequent temperature inversions, causing enhanced concentration of pollutants. The urban background site (GU) is situated at the Chemistry Department of Go¨teborg University. It is located approximately 2 km from the city center (i.e., Femman) and 15 m above and 20 m away from the closest road, with a traffic intensity of 8000 vehicles per day. The GU site is thus characterized by very low NO and NO2 concentrations and the [NO] to [NO2] ratio indicates that the site is mostly exposed to aged traffic emissions. However, some occasions with local traffic exposure can be extracted from the measurements. Figure 1 presents the average diurnal variation of NO and numbers of particles for all sites. As has been shown previously the morning rush hours are clearly shown for the road sites, while the afternoon rush hours are diminished, as observed in most urban areas (7, 8, 20). Regarding NO, the
TABLE 2. Correlation Coefficients for Number of ufp, i.e., 10-100-nm Particles, and for the Gases CO, NO, NO2, NOx, and the Ratio of NO to NO2 ufp ufp NO/NO2 NO NO2 NOx CO ufp NO/NO2 NO NO2 NOx ufp NO/NO2 NO NO2 NOx CO ufp NO/NO2 NO NO2 NOx
NO/NO2
NO
NO2
Femman, urban rooftop, 2045 samples 1.00 0.52 0.59 0.71 1.00 0.89 0.67 1.00 0.74 1.00
GU, urban background, 5149 samples 1.00 -0.06 0.14 0.46 1.00 0.78 0.07 1.00 0.52 1.00 Garda, urban roadside, 1542 samples 1.00 0.72 0.78 0.66 1.00 0.93 0.58 1.00 0.69 1.00
Surte, rural roadside, 868 samples 1.00 0.43 0.72 0.65 1.00 0.75 0.30 1.00 0.69 1.00
NOx
CO
0.65 0.89 0.99 0.83 1.00
0.48 0.78 0.82 0.61 0.81 1.00
0.38 0.41 0.81 0.92 1.00
n.a. n.a. n.a. n.a. n.a.
0.79 0.91 0.99 0.79 1.00
0.41 0.52 0.45 0.28 0.44 1.00
0.75 0.64 0.96 0.86 1.00
n.a. n.a. n.a. n.a. n.a.
road site at Garda exhibits the largest peak in the concentration, while the road site at Surte has lower concentration during all hours. Table 2 shows the correlation between selected quantities describing the four sites. The trends depictured in Figure 1 are also found in the correlation matrix. The sites Surte, Garda, and Femman had good correlation between nitrogen oxides and numbers of particles, while the correlation was not as good for the background site. The correlations of ultrafine particles and NO was higher than with NO2 at the road sites, as compared to the rooftop and background sites, where ultrafine particles correlated better with NO2 than NO. This can be explained by slightly more aged air masses observed at these sites, including a larger fraction of transported particles in the concept of ultrafine particles. Emission Factors. The data collected during the four different campaigns have been used to calculate sizedistributed particle emission ratios and emission factors for on-road traffic, using a novel simplified approach. The calculation of the emission factor is done by determination of an emission ratio and then using an existing emission factor for gaseous species. The emission ratio, ER, is defined as the ratio between the averaged particle concentrations and a traffic-related gas concentration average, both measured in the traffic-emitted aerosol, see eq 1.
ER )
([particle]emission) ([related gas]emission)
)
([particle]high-traffic) -([particle]background) ([related gas]high-traffic) - ([related gas]background)
1
In equation 1 an overline denotes the average over a number of samples. The emissions, i.e., [particle]emission and [related gas]emission, are defined by the difference in average
concentrations measured during high-traffic conditions and averages under background conditions. The two data sets “high-traffic” and “background” must be carefully defined. Ideally the background should be measured during the same conditions, e.g. location and time, as the high-traffic situation but excluding the traffic source. This is impossible and the measurement has to be conducted either at a different time or at a different location. Measuring at the same location but separated in time requires consideration of the background diurnal variation. Unfortunately, the urban background often has a diurnal pattern similar to that of the traffic exposure at the site, and is thus eventually underpredicted. However, if one use a separate site for monitoring of the background condition one has to consider the risk that the two sites occasionally are exposed to other local sources. Obvious the two-site method also needs two identical experimental setups. For this study only one site was used, and accordingly the background was defined as the average of the low trace gas concentration condition. It was noted that both the high- and low-traffic conditions were spread over the day, i.e., no preferable time of the day. The high-traffic conditions are selected from the data set, using high concentration of a gas as a tracer or “selecting variable”, and not by traffic intensity counts. The reason for not using traffic counts is that the meteorological effects dominate and make the correlation between traffic and traffic-related particles lower than the correlation between traffic-related gases, such as NO, and particle concentration (7). If the background is defined due to high dilution decreasing the concentrations instead of only reduced traffic influence, the amount of traffic exhaust included will affect the particle and related gas concentrations equally. When calculating the averages needed for eq 1, numbers of data points used for high-traffic and background conditions have to be decided. As a defined base case, 20% of the data were used for the background and the high-traffic situation averages, respectively. The impact of number of data points used when calculating the emission ratios was evaluated and described later. The calculated emission ratio is given in units of particles per mole of a gas compound, and can be normalized to the appropriate gas emission factor to give a particle emission factor, e.g., particles per consumed amount of fuel or per vehicle and kilometer. The relating gas compound is not necessarily the same compound as the one selecting the highor low-traffic-influenced situation. In the base case NOx was used, but also normalizing to NO or CO concentration is reported for comparison. The most important issue when selecting the relating gas species is that it must be emitted from traffic, in known amounts, and should be commonly and easily measured. The method described above relies on the ability to find a good tracer for high-traffic and background conditions, respectively. Nitrogen oxides (NOx) are well-known emitted species from combustion sources such as traffic. Most of the nitrogen oxides from combustion sources is emitted as NO and then converted into NO2 in the atmosphere by reaction with ozone. Additionally, NO2 may be photochemically transformed back to NO. Further away from sources these two conversion routs between NOx species will be balanced and crucially dependent on photochemical activity, as described by Leighton’s relationship. The chemistry of NOx in the atmosphere can thus be used to estimate the age of emissions. As a first approximation, fresh emissions contain high [NO] and low [NO2], while aged air masses contain low [NO] and high [NO2]. However, polluted air contains high NOx concentration (high [NO] and [NO2]) while clean air has low [NOx] levels (low [NO] and [NO2]). This may influence the results if [NO] or [NO2] are used directly as indicators. Thus, if the site is exposed to both polluted air masses and VOL. 39, NO. 19, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7611
TABLE 3. Average Concentrations and Wind Speed for High-Traffic and Background Cases, as Selected by Different Variables at the Four Sitesa
FIGURE 2. Resulting emission ratios calculated according to the base case for the four sites. For the base case the background was defined by low NOx concentrations, and high-traffic was defined by high NO to NO2 ratio, and both cases used 20% of the data. clean air masses the ratio of [NO] to [NO2] can be used to indicate fresh combustion emissions. To find the background (clean) conditions, low concentrations of total NOx are ideal as an indicator. The atmospheric fate of particles and the tracers should be as similar as possible. The fate of NO will be removal by conversion into NO2 by oxidation, while the time scales of particles are dependent on their size, where smaller particles (below 20 nm) are removed more quickly than the accumulation mode particles (around 100 nm). Thus, it would be beneficial to measure close to the sources.
Results The emission ratios, shown in Figure 2, were calculated at the four sites using a “base case”, where high-traffic conditions were selected by high [NO] to [NO2] ratio, the background by low NOx concentrations, and 20% of the data were used to describe each of the two conditions. The major features from the calculated ratios are that the rural road site has increased numbers of particles between 10 and 30 nm and that the rooftop site lacks them. The small particles at the rural site can originate from higher traffic speed. A large number of selecting variables were used for comparison and the obtained concentrations for high-traffic and background conditions are shown in Table 3. Upon changing tracers and the amount of data, the calculated emission ratios at the four sites reveal some important characteristics of the particle emissions and its aging in the atmosphere. The effect of shifting the lifetime of the selecting gas, for the high-traffic situation, was studied, using the road sites as examples, see Figure 3. The emission ratios give two peaks in the size distribution; one at 20 nm and one at 70 nm. Changing the selecting variable between NO to NO2 ratio, NO, NOx, and NO2 shows that the 20-nm peak is more related to short lifetimes than the 70-nm peak. At the urban rooftop site, Femman, the distance to the traffic prevents the smallest of the particles from reaching the site, i.e., if this site is used to calculate emission ratios, the size distribution lacks the smallest particles in the spectra. The spectra of the emission ratio for Femman reveal a broad stable 30-nm mode which is more or less independent of trace gas used. These measurements occurred during wintertime when the site exhibited stable wintertime inversion condition with low photochemical activities, i.e., low concentration of O3. The size distribution is then reflecting a combined inversion aerosol where the smallest particles are removed faster than, e.g., NO. Using different trace gases for determining high-traffic condition showed the largest effect for the urban background site, GU. The effect of which trace 7612
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 19, 2005
selecting variable
NO/NO2
NO ppb
NO2 ppb
NOx ppb
CO ppb
ufp × 103 cm-3
WS m s-1
NO/NO2 NO NOx NO2 CO ufp
2.6 2.6 2.5 2.0 2.0 2.2
73 75 75 65 53 65
Garda 26 99 28 103 29 104 31 95 22 75 27 91
256 222 209 170 371 219
11.5 12.3 12.4 11.8 9.8 14.2
1.7 1.4 1.4 1.3 1.7 1.4
NO/NO2 NO NOx NO2 ufp
1.9 1.6 1.4 1.1 1.2
22 28 28 24 22
Surte 11 33 18 47 20 48 21 46 17 40
n.a. n.a. n.a. n.a. n.a.
6.9 8.8 9.0 9.0 10.9
2.3 1.3 1.1 0.8 1.4
NO/NO2 NO NOx NO2 ufp
0.9 0.9 0.6 0.5 0.4
7 7 7 6 3
16 21 23 23 17
n.a. n.a. n.a. n.a. n.a.
3.9 4.7 5.7 6.0 7.7
4.2 3.9 3.1 2.8 3.0
NO/NO2 NO NOx NO2 CO ufp
1.8 1.8 1.8 1.5 1.5 1.3
79 80 80 73 66 62
Femman 37 117 40 121 41 121 42 116 33 99 36 98
346 350 350 326 408 295
6.9 7.4 7.5 7.6 6.3 9.3
2.1 2.0 2.0 2.0 2.1 2.9
GU 6 8 10 10 8
Background Site WS NO/NO2 NO NO2 NOx CO ufp (NOx selected) ppb ppb ppb ppb × 103 cm-3 m s-1 Femman GU Surte Garda
0.1 0.5 0.3 0.3
1 1 1 2
7 3 2 6
8 4 2 8
123 n.a. n.a. 35
1.9 2.8 2.4 2.8
4.4 4.5 2.4 2.6
a Background conditions are always selected by low [NO ] due to x the limited effect of using alternating selecting variables.
FIGURE 3. Emission ratios calculated at the four sites, using NO2, NOx, NO, or NO to NO2 ratio as high-traffic selector, and NOx as background selector. gas to use is thus more profound for sites with low traffic flow where the difference between high and low traffic exposure is smaller. Carbon monoxide (CO) was also tested as an alternative to using NOx related variables as a tracer of traffic. The
FIGURE 4. Emission ratios calculated at the road sites, using CO as the selecting variable and either CO or NOx as the relating gas. behavior of CO in the atmosphere is considerably different compared to NOx, since CO has a much longer lifetime. Figure 4 shows the calculated emission ratios using the CO concentration in comparison to the preferred [NO] to [NO2] ratio for Garda, the urban road site. There was no significant difference between using either of the two tracer methods, if the emission ratios were calculated normalized to NOx concentration. However, if normalized to [CO], there was a significant difference, and by using CO as a traffic indicator the emission ratio is underestimated by almost a factor of 2. The ratio between CO and NOx varies among different combustion processes. Vehicles using gasoline generally emit more CO than diesel-fueled vehicles, and the difference is even larger for noncatalyst gasoline vehicles. In this study, the applied method was used on limited data sets and it was not possible to extract vehicle fleet composition using different tracer compounds. With larger data sets, e.g., higher time resolution, it should be possible to perform such evaluations. Particle number was also evaluated as a selecting variable. Particles are emitted from different sources, and different sizes of particles have different lifetimes. When the total number of particles (10-370 nm) is used as a selecting variable, also accumulation mode particles, i.e., transported air masses, can be included in the definition of high-traffic, causing an overestimation of particles in relation to the normalizing gas concentration. Hence, an emission ratio calculated in this way can only be used as an upper limit of the true emission ratio of the traffic studied. To avoid including accumulation mode particles, a limited particle size interval for defining high-traffic conditions can be selected. However, defining a part of the size spectra as a selecting variable introduces a distorted size distribution, depending heavily on the size limit used in the definition, and is therefore not recommended. In the selection process of background condition (low traffic exposure) the choice of trace gas has little effect on the data from road sites while there was a small but noticeable effect for the rooftop and background sites. Still, all backgrounds have rather low concentrations, especially in the smallest size range, and only the 70-nm peak is affected. The results from GU, the University background site, having the hardest requirements for extracting an accurate traffic aerosol, shows that using low NOx was favorable to define the background conditions. In the above emission ratio determinations, 20% of the data were used to calculate the average for high-traffic and background conditions, respectively. Figure 5 shows, in 5% intervals, the effect of using a larger or smaller fraction of the data set to calculate the emission ratios. The effect on calculated emission ratios was small and only the University site was very sensitive to the choice of numbers of data points, due to the limited traffic exposure at this site. It was obvious that data from road sites with high traffic exposure are better suited for emission ratio calculations. The fact that number of data points used does not affect the outcome much indicates that the calculated emission ratio is a good measure
FIGURE 5. Different parts of the data are defined as high- and low-traffic situations, respectively. Each line shows one of the 5% intervals between 5 and 95%. The base case, i.e., 20% is indicated by stars. The ratio of [NO] to [NO2] is used as selecting variable for high-traffic and [NOx] is used as background for all four sites. of the actual emission ratio, i.e., a quality assurance. Noteworthy is that the chosen “base case”, i.e., averaging over 20% of the data to describe traffic influenced or background situations, respectively, was arbitrary and might not be optimal for other conditions/sites. It would be preferable to use the same classification, if possible, to facilitate a comparison. The applied method uses the ratio of the numbers of particles to the relating gas species and corrects for background concentrations and therefore it is remarkably robust to the choice of numbers of data points. The total number emission ratios, shown in Table 4, were calculated as the sum of the particles in the size distribution. Indicated in bold are the emission ratios preferred for emission factor calculations, based of the above discussion. A comparison between the different ways of sorting and normalizing data gives information about the sites and the reliability of the data. From the data in Table 4 it is striking that a selecting procedure using ultrafine particles, ufp, will overestimate the emission factors, while the other methods are similar in magnitude. For the emission factors normalized by [NO] the difference between selecting variables is larger than when normalizing to [NOx], which is a more stable variable. When normalizing to [NOx], the emission factor determined at the road sites can be extracted with all presented methods with the exception of ufp. For the sites further away from traffic, especially the GU site, the influence of other sources becomes more important and a tracer method with a short lifetime, i.e., [NO] to [NO2] ratio, must be used. Using [CO] to normalize the emission factor gives results similar to those of normalizing to [NOx]. However, there was a significant effect when using the [CO] selection method that shows the effect of sorting and normalizing to the same parameter. Normalization by [NO] or [NOx] of the road sites emission ratios reflects the fraction of NO emitted from traffic. The number of particles emitted per mole of NOx is 70-80% of the number of particles emitted per mole of NO, at the road sites. The range 70-80% NO of total NOx is slightly lower than what is expected directly from the exhaust, probably due to initial oxidation in the initial mixing process. The errors given in Table 4 are the statistical errors calculated at the 95% confidence level. From our intracalibration of the particle instruments we estimate a maximum instrumental error of 20% in particle number concentrations. For the gases the calibration of the instruments give an uncertainty of less than 2 ppb. The resulting emission ratios VOL. 39, NO. 19, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7613
TABLE 4. Emission Ratios for Ultrafine Particles Calculated Based on Low NOx Defining the Background, and Ratio of NO to NO2, NO, NOx, NO2, CO, or Number of Ultrafine Particles (ufp) Defining the High-Traffic Exhausta selecting variable site
unit NO)-1
NO/NO2
NO
NOx
NO2
CO
ufp
30 ( 3 50 ( 11 51 ( 13 16 ( 2
32 ( 3 73 ( 10 57 ( 10 17 ( 2
32 ( 3 133 ( 12 60 ( 10 17 ( 2
35 ( 4 170 ( 15 67 ( 11 19 ( 2
34 ( 5 n.a. n.a. 17 ( 2
45 ( 4 528 ( 29 95 ( 12 29 ( 2
Garda GU Surte Femman
1014
Garda GU Surte Femman
1014 (mole NOx)-1
24 ( 3 22 ( 5 35 ( 9 11 ( 1
25 ( 3 27 ( 4 35 ( 6 12 ( 1
25 ( 3 37 ( 3 35 ( 6 12 ( 1
25 ( 3 41 ( 4 37 ( 6 13 ( 1
26 ( 4 n.a. n.a. 12 ( 1
33 ( 3 94 ( 5 55 ( 7 20 ( 1
Garda Femman
1014 (mole CO)-1
10 ( 1 6(1
13 ( 1 6(1
14 ( 1 6(1
16 ( 2 7(1
5(1 4(1
15 ( 1 11 ( 1
(mole
a Averages over 20% of the samples are used in each of the selections, and the standard error is given. Ratios used in emission factor calculations are shown in bold.
TABLE 5. On-Road Emission Factors EF km-1
EF (NOx) g km-1
ER (g NOx)-1
ER (mole NOx)-1
LDVe HDVf
4.6 × 1014 63 × 1014
9.0c 42c
0.5 × 1014 1.5 × 1014
23.5 × 1014 69 × 1014
fleet
4.5 × 1014b
1.4d
2.0 × 1014
92 × 1014
9
fleet
5 × 1014
1.4d
4 × 1014
160 × 1014
21
fleet
4.6 × 1014
1.4c
3.3 × 1014
151 × 1014
3
LDVe, diesel LDVe, gas
1 × 1014 0.1 × 1014
1.4d 1.4d
1 × 1014 0.1 × 1014
40 × 1014 4 × 1014
22 22
fleet
1.75 × 1014
1.4d
1.25 × 1014
60 × 1014
15
fleet
n.a.
n.a.
1.0 × 1014
35 × 1014
this study
n.a.
0.7 ×
24 ×
site description California, tunnel >10 nm >10 nm Australia, highway 15-700 nm Minnesota, highway >3 nm Stockholm, tunnel 3-900 nm Sweden, lab 10 nm to 10 µm 10 nm to 10 µm Australia, road side 5-900 nm Sweden, rural road side 10-100 nm Sweden, urban roadside 10-100 nm
fleet
n.a.
a
1014
1014
In intalics are given related to kg fuel (not km). b Recalculated from per mile. c NOx emission factors from the same study. from literature. e Light duty vehicles. f Heavy duty vehicles.
determined at the two road sites, including statistical and instrumental errors, were (24 ( 8) × 1014 and (35 ( 15) × 1014 particles per mole NOx for Garda and Surte, respectively. These values can directly be used for comparisons among measurement campaigns and conditions.
Discussion To convert the emission ratios into particle number emission factors an emission factor of, e.g., NOx must be used. For our data an emission factor for Swedish conditions (1.4 g NOx per km) was used for normalization (3). Table 5 summarizes the obtained emission factors together with emission factors calculated from reported data sets, either measured at road sites or in laboratories. The 1.4 g NOx per km has been used for studies when no NOx emissions have been presented. From the summarized data in Table 5 it was noted that the emission factors from the present study generally are lower than the previously reported emission factors. The data in the present study includes 10-100-nm particles, while some of the other studies have used a lower cut off (e.g., 3 nm). Numbers of particles between 3 and 10 nm can occasionally be very high and will influence the comparison (23). Furthermore, there are some uncertainties in the NOx emissions at individual sites, which directly influence the comparison. Regarding the size distribution of emitted particles from traffic it is known that vehicles emit two distinct modes of 7614
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 19, 2005
d
ref
16 16
this study NOx emission factor
particles: one nucleation mode around 20 nm and one mode around 70 nm (3, 6, 24). It has been shown previously that the nucleation peak consists of volatile material, due to different concentrations when the exhaust is diluted by air of different temperatures (24) and as the peak evaporates if a thermal denuder is used in front of the SMPS system (25). In a chase experiment the nucleation mode was shown to consist of organic components (6). At the sites used in this study, the road sites clearly show a distinct 20-30-nm mode. The 70-nm mode is present in the Garda emission, while for Surte it is visible only as a shoulder. By comparing the magnitude of the two peaks between the two road sites the vehicles at the rural site are shown to emit twice the amount of 20-30-nm particles, while the mode at 70 nm is ca. 30% higher at the urban site compared to the rural site. Vehicles at the rural site run at higher engine temperatures, due to high speed, producing less soot but consuming more lubrication oil, giving higher concentrations of nucleation mode particles (20-30 nm) and less of the soot aggregates (70 nm). The larger total numbers of particles emitted at the high-speed rural case is in accordance with earlier studies by Kittelson et al (21). In summary, the selection, using high [NO] to [NO2] ratio for the high-traffic aerosol, and low [NOx] for background, makes the extraction of data, when only traffic sources affect the roadside measurement station, possible. By using the normalizing procedure to a compound emitted from the same
source, favorably a stable variable like NOx, the determination becomes independent of calculations of dilution/meteorology. The novel method gives particle emission factors comparable to reported data while being easily applicable and robust.
Acknowledgments MISTRA, the Swedish Foundation for Strategic Environmental Research, is acknowledged for financial support and J.Gust.Richerts research funding is acknowledged for a new COanalyzer.
Literature Cited (1) Voutilainen, A.; Kaipio, J.; Pekkanen, J.; Timonen, K.; Ruuskanen, J. Theoretical analysis of the influence of aerosol size distribution and physical activity on particle deposition pattern in human lungs, Scand. J. Work Environ. Health 2004, 30 (Suppl.2), 7379. (2) Janha¨ll, S.; Jonson, A° . M.; Molnar, P.; Svensson, E.; Hallquist, M. Size resolved traffic emission factors of submicrometer particles. Atmos. Environ. 2004, 38 (26), 4331-4340. (3) Kristensson, A.; Johansson, C.; Westerholm, R.; Swietlicki, E.; Gidhagen, L.; Wideqvist, U.; Vesely, V. Real-world traffic emission factors of gases and particles measured in a road tunnel in Stockholm, Sweden. Atmos. Environ. 2004, 38, 657-673. (4) Holmen, B. A.; Ayala, A. Ultrafine PM emissions from natural gas, oxidation-catalyst diesel, and particle-trap diesel heavyduty transit buses. Environ. Sci. Technol.2002, 36 (23), 50415050. (5) Gouriou, F.; Morin, J. P.; Weill, M. E. On-road measurements of particle number concentrations and size distributions in urban and tunnel environments. Atmos. Environ. 2004, 38 (18), 2831-2840. (6) Canagaratna, M. R.; Jayne, J. T.; Ghertner, D. A.; Herndon, S.; Shi, Q.; Jimenez, J. L.; Silva, P. J.; Williams, P.; Lanni, T.; Drewnick, F.; Demerjian, K. L.; Kolb, C. E.; Worsnop, D. R. Chase studies of particulate emissions from in-use New York city vehicles. Aerosol Sci. Technol. 2004, 38, 555-573. (7) Molna´r, P.; Janha¨ll, S.; Hallquist, M. Roadside measurements of fine and ultrafine particles at a major road north of Gothenburg. Atmos. Environ. 2002, 36 (25), 4115-4123. (8) Wehner, B.; Birmili, W.; Gnauk, T.; Wiedensohler, A. Particle number size distributions in a street canyon and their transformation into the urban-air background: measurements and a simple model study. Atmos. Environ. 2002, 36, 2215-2223. (9) Gramotnev, G.; Brown, R.; Ristovski, Z.; Hitchins, J.; Morawska, L. Determination of average emission factors for vehicles on a busy road. Atmos. Environ. 2003, 37 (4), 465-474. (10) Nanzetta, M.; Holmen, B. Roadside particle number distributions and relationships between number concentrations, meteorology, and traffic along a northern California freeway. J. Air Waste Manage. Assoc. 2004, 54, 540-554.
(11) Pokharel, S. S.; Bishop, G. A.; Stedman, D. H. An on-road motor vehicle emissions inventory for Denver: an efficient alternative to modeling. Atmos. Environ. 2002, 36, 5177-5184. (12) Hwa, M.-Y.; Hsieh, C.-C.; Wu, T.-C.; Chang, L.-F. Real-world vehicle emissions and VOCs profile in the Taipei tunnel located at Taiwan Taipei area. Atmos. Environ. 2002, 36, 1993-2002. (13) Wingfors, H.; Sjo¨din, A° .; Haglund, P.; Brorstro¨m-Lunde´n, E. Characterisation and determination of profiles of polycyclic aromatic hydrocarbons in a traffic tunnel in Gothenburg, Sweden. Atmos. Environ. 2001, 35, 6361-6369. (14) Gidhagen, L.; Johansson, C.; Sto ¨ m, J.; Christenssom, A.; Swietlicki, E.; Pirjola, L.; Hansson, H.-C. Model simulation of ultrafine particles inside a road tunnel. Atmos. Environ. 2003, 37, 20232036. (15) Jamriska, M.; Morawska, L. A model for determination of motor vehicle emission factors from on-road measurements with a focus on submicrometer particles. Sci. Total Environ. 2001, 264, 241-255. (16) Kirchstetter, T. W.; Harley, R. A.; Kreisberg, N. M.; Stolzenburg, M. R.; Hering, S. V. On-road measurement of fine particle and nitrogen oxide emissions from light- and heavy-duty motor vehicles. Atmos. Environ. 1999, 33 (18), 2955-2968. (17) Ketzel, M.; Berkowicz, R. Modelling the fate of ultrafine particles from exhaust pipe to rural background: an analysis of time scales for dilution, coagulation and deposition. Atmos. Environ. 2004, 38, 2639-2652. (18) Zhang, K. M.; Wexler, A. S. Modeling the number distributions of urban and regional aerosols: theoretical foundations. Atmos. Environ. 2002, 36, 1863-1874. (19) Jacobson, M. Z.; Seinfeld, J. H. Evolution of nanoparticle size and mixing state near the point of emission. Atmos. Environ. 2004, 38 (13), 1839-1850. (20) Charron, A.; Harrison, R. M. Primary particle formation from vehicle emissions during exhaust dilution in the roadside atmosphere. Atmos. Environ. 2003, 37, 4109-4119. (21) Kittelson, D.; Watts, W.; Johnson, J. Nanoparticle emissions on Minnesota highways. Atmos. Environ. 2004, 38, 9-19. (22) Fa¨rnlund, J.; Holman, C.; Ka˚geson, P. Emissions of ultrafine particles from different types of light duty vehicles; Swedish National Road Administration: Borla¨nge, 2001. (23) Shi, J. P.; Evans, D. E.; Khan, A. A.; Harrison, R. M. Sources and concentration of nanoparticles (
