Effect of Engine Load on Diesel Soot Particles - ACS Publications

Mar 18, 2004 - impact on earth's radiation balance as well as on health effects of vehicle emissions. ... diesel particles depend on engine load. The ...
0 downloads 0 Views 155KB Size
Environ. Sci. Technol. 2004, 38, 2551-2556

Effect of Engine Load on Diesel Soot Particles ANNELE K. K. VIRTANEN, J Y R K I M . R I S T I M A¨ K I , KATI M. VAARASLAHTI, AND JORMA KESKINEN* Aerosol Physics Laboratory, Institute of Physics, Tampere University of Technology, P.O. Box 692, FIN-33101 Tampere, Finland

This study concentrates on characterization of nonvolatile fraction of diesel particles. These particles have an impact on earth’s radiation balance as well as on health effects of vehicle emissions. In addition to composition and size distribution of particles, an important factor affecting their health effects and properties and lifetimes in the atmosphere is their morphology. The effect of engine parameters on soot particle size distributions and also on particle morphology has been studied. It was found that the shape of the size distribution and also the structure of diesel particles depend on engine load. The number distributions were found to obey log-normal assumption. The width of the distribution increased with increasing engine load. The geometric standard deviations of measured distributions varied from 1.7 to 2.1. Simultaneously, the fractal dimension of particles decreased with increasing engine load. The values for mass fractal dimensions based on scaling of particle mass and mobility size were between 2.6 and 2.8. Both electron microscopy and measurements of aerodynamic size versus mobility size suggest that the morphology of particles in different size regimes vary, with the large particles being less compact than the small ones.

Introduction Ambient fine particles are of current interest because of their negative health effects (1). Atmospheric aerosols also influence the climate by scattering or absorbing light and by acting as cloud nuclei (2). In addition to composition and size distribution, an important factor affecting the health effects, properties, and lifetimes of particles in the atmosphere is their morphology. An important example of solid agglomerated particles is diesel soot. These carbon particles directly affect the earth’s radiation balance by absorbing solar radiation. This increases the atmospheric warming (3). Numerical aerosol models predicting the aerosol behavior in the atmosphere consider the particle residence time and optical properties. The size and morphology of nonvolatile soot particles affect both these factors (4, 5). Thus detailed study of the structure, size, and concentrations of particles emitted from different sources is needed for modeling purposes. In addition, the properties of nonvolatile soot particles also have an impact on particle health effects (68). After emission to atmosphere, the condensation of volatile materials onto soot particles or the formation of new volatile * Corresponding author telephone: +358 3 3115 2676; fax: +358 3 3115 2600; Author e-mail address: [email protected]. 10.1021/es035139z CCC: $27.50 Published on Web 03/18/2004

 2004 American Chemical Society

particles through nucleation may take place. In addition to the properties of the exhaust gas, these processes depend on the conditions (e.g., temperature, relative humidity) of the ambient air. On the other hand, the structure of the nonvolatile particle core is independent of ambient conditions, making it a more universal quantity. In this study, we concentrate on studying the structure of the nonvolatile core of exhaust particles (i.e., soot particles) emitted into atmosphere. Diesel exhaust soot particles are formed through collisions of primary soot particles typically sized between 20 and 50 nm (9). The structure of the particles depends on agglomeration conditions (i.e., conditions in the combustion chamber and in its vicinity). One way to define the structure of agglomerates is to treat them as fractal structures. There are different ways to define and measure the fractal dimension (10-15). We use the mass fractal dimension, based on scaling of particle mass, as a function of mobility size (14). If the particles are agglomerates, it is possible to evaluate the fractal dimension of the agglomerate when the scaling of the effective density or the mass and the particle size is known. The fractal dimension (df) can be defined through

Fe ∝ dbdf-3

(1)

where Fe is the effective density and db is the mobility diameter. The fractal dimension (df) defined according to eq 1 may have values from 1 to 3. Fractal dimension is 3 for compact particles, resulting in a constant, size-independent value of effective density. The fractal dimension of agglomerates is less than 3; therefore, effective density decreases as particle size increases. The looser the agglomerate structure, the smaller the fractal dimension is. For simple chain agglomerates, the fractal dimension is 1. The fractal dimension depends on the agglomeration process: monomer-cluster agglomeration results in fractal dimension close to value 3, whereas cluster-cluster agglomeration produces a lower fractal dimension (df ) 1.8-2) (16). According to earlier studies, the found fractal dimensions of diesel exhaust particles have been between 2 and 3 (1719). In earlier studies, a connection of particle structure and engine load has been found. Results presented by Skillas et al. (17) show that the fractal dimension of diesel soot particles depends on engine load. Also Park et al. (19) found a slight connection between the increasing engine load and the decreasing fractal dimension. Analyzing a large number of size distributions, Harris and Maricq (20, 21) found that the width of normalized soot particle size distributions was practically constant and independent of engine type, engine conditions, or fuel properties. According to them, the size distribution can be described with a log-normal distribution with a geometric standard deviation of ∼1.8 (20, 21). In this study, we take closer look at the distribution shape, especially with respect to distribution width. With the exception of the recent Park et al. paper (19), in most of the soot particle structure studies the particles were generated by a light-duty diesel engine with a maximum power of a few kilowatts. In this study, the size distribution and structure of particles emitted from an off-road tractor engine, a passenger car, and a city bus were studied.

Materials and Methods Vehicles and Fuels. Three diesel engines of different volumes were represented in the measurements. A passenger car (Audi A4 1.9 TDI) and a city bus (Volvo DH 10A-285) were tested VOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2551

FIGURE 1. Measurement setup for passenger car measurements (left) and city bus and tractor engine measurements (right).

TABLE 1. Vehicle and Engine Information model year odometer, km displacement, dm3 transmission cylinders combustion system max power, kW max torque, Nm aftertreatment emission level

passenger car

city bus

1995 27 500 1.89 A4 4 DI, turbocharged, intercooled 66/4000 rpm 202/1900 rpm EGR + oxidation catalyst Euro II

2000 257 000 9.6 automatic 6 DI, turbocharged, intercooled 210/2000 rpm 1200/1450 rpm oxidation catalyst Euro II

on light-duty and heavy-duty chassis dynamometers, respectively. In addition, a tractor engine (Sisudiesel 420 DSRE) was tested on an engine dynamometer. The passenger car was equipped with EGR and oxidation catalyst, the city bus was equipped with oxidation catalyst only, and the tractor engine ran without any after-treatment system. More detailed engine and vehicle information is presented in Table 1. In the passenger car measurements, two different fuels were used: 23 and 430 ppm sulfur. In the city bus measurements, the fuel sulfur content was 50 ppm; in the tractor engine measurements, the fuel sulfur content was as high as 2000 ppm. Engine Parameters. For the passenger car, steady-state measurements with three different speeds (55, 80, and 120 km/h) were done. The used engine loads were normal road loads. At 80 km/h speed, also a higher load (50% of maximum load) was studied. In the city bus measurements, the vehicle 2552

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004

tractor engine 1999 ∼600 4.2 4 DI, turbocharged, intercooled 75/2300 rpm 350/1300 none Euro I

speed was 100 km/h in lowest load point. Higher load points (25-75% of maximum load) were measured with 60 km/h speed. The engine rotation speed in all the tractor engine measurements was 2300 rpm, and the engine load was varied from 15% to 75% of maximum load. Sampling. Two different sampling methods were used (Figure 1.). In both cases, tailpipe sampling was used to avoid coagulation of particles and storage effects for gaseous components found in transfer lines and dilution tunnel (22). In the car experiments, hot exhaust gas was diluted immediately after the tailpipe with two ejector diluters. Twostage ejector dilution systems have been applied to different combustion sources in several studies (22, 23). The first dilution unit, short transfer line, and also the dilution gas were heated to 200 °C. The second diluter and the dilution air were at room temperature. The dilution ratio of both diluters was approximately 8. The setup minimizes changes

FIGURE 2. Measured SMPS size distributions for high S fuel. Number distributions with log-normal fits on the left and volume distributions on the right. in size distribution caused by gas-to-particle conversion (i.e., nucleation and condensation of gaseous components onto soot particles). In the city bus and tractor engine measurements, a porous tube diluter was used, followed by a secondary ejector dilutor. After the dilution, the volatile material was removed from particles by using a thermodenuder (24). The sampling scheme was as shown on the righthand side of Figure 1. Distribution Measurements. Both aerodynamic and mobility size distributions were measured with on-line instruments. The mobility size distribution is measured with a scanning mobility particle sizer (SMPS) (25), and the aerodynamic distribution is measured with an electrical lowpressure impactor (ELPI) (26). SMPS includes a differential mobility analyzer (DMA) (27) and a condensation particle counter (CPC). DMA classifies particles according to their electrical mobility. The classified particles are then counted with the CPC. In the ELPI, impactor particles are classified into 12 size channels from 30 nm to 10 µm, according to their aerodynamic diameter. If the filter stage is used, as is the case here, the measurement range is 7 nm-10 µm (28). Particles are first charged, and then the current carried by the charged particles is measured at each impactor stage by a multichannel electrometer. Estimation of Fractal Dimension. The methods used in this study to estimate the effective density are based on measuring the aerodynamic and mobility equivalent size of the sample particles and on the connection of these equivalent sizes and effective density of the particle:

da2Cc(da)F0 ) db2Cc(db)Fe

(2)

Here da is the aerodynamic diameter, Fe is the effective density, and db is the mobility equivalent diameter of the same particle. The slip correction factor Cc is evaluated separately for each of the equivalent diameters. F0 is the unit density. Once the aerodynamic and mobility equivalent sizes of the particles are known, the effective density can readily be calculated using eq 2 (29, 30). A widely used method to find effective density of certain particle size is to use DMA to classify particles according to their mobility size and then to measure the aerodynamic size of single mobility particles (30-32). If the measured size range is wide enough, the fractal dimension can be derived from eq 1. This method is relatively slow, thus we have used a faster distribution fitting method based on measurement of aerodynamic and mobility size distributions measured with ELPI and SMPS. The primary output of the ELPI

measurement is the current distribution as a function of the aerodynamic particle size. The SMPS number distribution can be transferred to the current distribution of the corresponding ELPI measurement (33). The difference of these two distributions is minimized by using eqs 1 and 2 (34). Thus from two simultaneous distribution measurements we gain the particle effective density as a function of particle size over the whole distribution and the corresponding fractal dimension. It should be noted here that, if the measured particles have different morphologies, the method gives average value for fractal dimension weighted with the relative concentration of different morphology types. The single mobility method described above was used as a reference method. Electron microscopy was used for qualitative analysis.

Results and Discussion Number Distributions. As an example, the measured size distributions of the passenger car for high S fuel are shown in Figure 2. The number concentration increases with increasing speed (Figure 2, left). With the higher engine load (80 km/h + 50% load), the temperature and the amount of injected fuel are highest while the air/fuel ratio is lowest. These conditions increase the initial soot formation and also accelerate the agglomeration process. Due to the fast agglomeration, the peak of the volume distribution is on larger particles in the 80 km/h + 50% load case than in the 80 km/h with normal road load case (Figure 2, right). Measured SMPS number distributions obey well the lognormal distribution assumption as can be seen in Figure 2. This is in agreement with earlier studies (20, 21). With low S fuel, the measured size distributions were comparable with distributions shown in Figure 2. The effect of engine load on the shape of particle size distribution can be clearly seen when the distributions are normalized in respect of concentration and peak diameter (20) as shown in Figure 3, where the normalized size distributions of both the vehicles and the test bench engine are shown. The load compared to the maximum load of used engine speed is marked in the labels. The distributions for the passenger car are marked with black symbols, and the city bus and the tractor engine distributions are marked with gray symbols. Distributions corresponding to different engines coincide as described by Harris and Maricq (20). As can be seen, the distribution width increases with increasing engine load. The distribution broadening seems to have the strongest effect on the right side of the distribution. The geometric standard deviations (GSDs) corresponding disVOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2553

FIGURE 3. SMPS number distributions for three different engine types normalized with respect to peak size and concentration. tributions measured with 10% loads are ∼1.8, while values for 50-75% loads are ∼2.1-2.2. The increase in the GSD can be clearly seen, when the measured GSD values are plotted as a function of the engine mean effective pressure. MEP is obtained by dividing the work per cycle by the total cylinder volume, and it is a useful engine performance measure because it scales engines of

different volumes (35). No remarkable difference for these three different engine types or after-treatment systems can be seen. Harris and Maricq (20) presented the approximation according to which the diesel particle number distributions can be described with signature size distributions with GSD value ∼1.8 independent of engine type, engine speed, or fuel. Our results show that the GSD increases as a function of engine load but seems to be insensitive for engine or fuel type. The increase of the distribution width is remarkable with high engine loads. Particle Morphology. The effect of engine load on soot particle structure was studied with the passenger car and the city bus. The fractal dimension is shown in Figure 4b as a function of engine load. For the tractor engine, the structure information is available only for highest load point (marked with a star in Figure 4b). Clearly the fractal dimension decreases as engine load increases. The dashed line in Figure 4a,b is a guide for the eye only. There is relatively large scatter in the measured values, but it seems that neither engine type nor fuel sulfur level affect the fractal dimension significantly. No effect of sulfur level is expected because the fractal dimension as measured here describes the structure of the nonvolatile soot core. Decrease of fractal dimension as a function of engine load has been also reported by Skillas et al. (17). The decrease of fractal dimension as a function of engine torque can be explained as follows. With high engine loads, the amount of injected fuel is higher than in low engine torques. Also the air/fuel ratio is lower for high engine torque values (see Table

FIGURE 4. Measured geometric standard deviation (a) and fractal dimension (b) of measured exhaust particles as a function of engine mean effective pressure. Results for passenger car with high S fuel marked with black diamonds and low S fuel marked with gray diamonds. City bus and tractor engine measurements marked with white triangles and stars, respectively.

FIGURE 5. Electron microscope image of diesel particles for the 80 km/h case. Particles 200 nm (right). 2554

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004

FIGURE 6. Fitted density profile compared to reference method. Suggestion for different morphologies in different size regimes sketched in the figure. 1). These conditions favor the soot primary particle formation. In coagulation process, the monomer cluster agglomeration dominates in the beginning, resulting in compact clusters having fractal dimension close to the value of 3 (16). Later, the agglomeration continues with cluster-cluster agglomeration resulting in a lower fractal dimension (df ) 1.8-2) (16). When the initial soot concentration and temperature are high, the cluster-cluster agglomeration starts earlier resulting in a higher amount of large particles having lower fractal dimension than the smaller ones. In Figure 5, an example of the electron microscope images of particles emitted from passenger car in the 80 km/h case is shown. As can be seen, particles smaller than 100 nm have clusterlike structure (Figure 5, left), and larger particles (db ∼ 200 nm) seem to have a more open structure implying that particles have formed through cluster-cluster agglomeration. If the larger particles indeed do have a lower fractal dimension than the small particles, the fitting method used in this study finds the average value for the fractal dimension of the particles weighted with the relative concentration of each morphology type. This could be the reason why the fractal dimension values found here are higher than those reported in the studies of Park et al. (19) and Skillas et al. (17). Park et al. (19) studied the particle structure emitted from a heavy-duty engine and found fractal dimensions from 2.33 to 2.41 depending on engine load. Skillas et al. (17) studied the morphology of particles generated by a small diesel engine with a nominal power of 3.3 kW. They found that the fractal dimension varied from ∼2.2 for high engine load to ∼2.9 for low engine load (17). The method used in those studies was based on measurement of mass or aerodynamic size of well-known monodisperse mobility size. In that case, no concentration weighting of different morphologies takes place. In Figure 6, the particle density as a function of particle size gained with our method is compared to a simultaneously measured reference method, which is in principle similar to the method of Skillas et al. (17). This means that the aerodynamic size of particles monodispersed in DMA is measured in a multistage impactor. The presented results were measured with the city bus with the highest load point. As can be seen, the agreement of the fitted density profile and reference method points is good near the peak of the distribution. When the particle size increases, the density values gained with reference method decrease more rapidly. Skillas et al. (17) distinguished two different particle “size regimes” in their measurements: one from 55 to 170 nm and the other from 84 to 270 nm. From their study, it can be seen

that the fractal dimension differs in these two size regimes: for the smaller particles, the df values varied between ∼2.3 and ∼2.9; for the larger particles, it varied between ∼2 and 2.75 depending again on engine load. This is in agreement with the assumption that larger particles have formed through cluster-cluster agglomeration and the smaller particles have formed through particle-cluster agglomeration. According to Heywood (35), soot formation takes place in many stages of the combustion process due to the non-homogeneous nature of the mixture and the duration of injection and its overlapping with the combustion process. This means that, in the combustion chamber, soot of different age exists. This supports the assumption of the existence of soot particles formed through different coagulation processes. As can be seen by comparing Figure 4, panels a and b, with increasing engine load, the GSD increases while df decreases. According to our knowledge, this connection in the case of diesel engines has not been previously published. This connection will be the topic of further studies. Summary. We have found that the shape of the size distribution and also the structure of nonvolatile diesel particles depend on engine load. The number distributions obey well the log-normal assumption. We have discovered that the width of the distribution increases with increasing engine load. Simultaneously, the fractal dimension of particles decreases with increasing engine load. The width of measured distributions varied from 1.7 to 2.1, implying that the distributions have not achieved the self-preserving state. We found that the values for fractal dimensions were between 2.6 and 2.8 depending on engine load. We have also suggested that the morphology of particles varies in different size regimes. Due to this, the fractal dimension values for diesel particles gained from the fitting method are lower than other reported values (17, 19). The electron microscopy images as well as measurements of aerodynamic size of single mobility size particles support this assumption. The reference measurement indicates that the density plotted as a function of mobility size in log-log figure cannot be presented with a single line.

Acknowledgments We thank several people from the TUT Aerosolphysics Group for their help during the measurement campaigns. The tractor engine measurements have been done in the TUT Energy and Process Laboratory. The car and city bus chassis dynamometer measurements have been done in VTT Processes, Engines and Vehicles Laboratory in Espoo. The vehicle test conditions and sampling were decided and realized in cooperation with Maija Lappi and Hannu Vesala from VTT. The measurement campaigns have been part of the Mobile2 programme (Contract 40710/99) and Fine programme (Contract 40405/02) funded by Tekes, ECOCAT Oy (formerly Kemira Metalkat Oy), FinnKatalyt Oy, and Dekati Oy.

Literature Cited (1) Dockery, D. W.; Pope, C. A. Acute respiratory effects of particulate air pollution. Annu. Rev. Public Health 1994, 15, 107-132. (2) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics. From Air Pollution to Climate Change; John Wiley & Sons: New York, 1998. (3) IPCC. Synthesis report. 2001; available at www. ipcc.ch. (4) Colbeck, I.; Appleby, L.; Hardman E. J.; Harrison, R. M. The optical properties and morphology of cloud-processed carbonaceous smoke. J. Aerosol Sci. 1990, 21, 527-538. (5) Colbeck, I.; Hardman, E. J.; Harrison, R. M. Optical and dynamical properties of fractal clusters of carbonaceous smoke. J. Aerosol Sci. 1989, 20, 765-774. (6) Dasenbrock, C.; Peters, L.; Creutzenberg, O.; Heinrich, U. The carcinogenic potency of carbon particles with and without PAH after repeated intratracheal administration in the rat. Toxicol Lett. 1996, 88, 15-21. VOL. 38, NO. 9, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2555

(7) Be´rube´, K. A.; Jones, T. P.; Williamson, B. J.; Winters, C.; Morgan, A. J.; Richards, R. J. Physicochemical characterization of diesel exhaust particles: factors for assessing biological activity. Atmos. Environ., 1999, 33, 1599-1614. (8) Gehr, P.; Geiser, M.; Hof, V. I.; Schu ¨ rch, S. Surfactant-ultrafine particle interactions: what we can learn from PM10 studies. Philos. Trans. R. Soc. London A 2000, 358, 2707-2718. (9) Burtscher, H. Review: Measurement and characteristics of combustion aerosols with special consideration of photoelectric charging and charging by flame ions. J. Aerosol Sci. 1992, 23, 549-595. (10) Forrest, S. R.; Witten, T. A. Long-range correlations in smokeparticle aggregates. J. Phys. A: Math. Gen. 1979, 12, 109-117. (11) Witten, T. A.; Sander, L. M. Diffusion-limited aggregation, a kinetic critical phenomena. Phys. Rev. Lett. 1981, 47, 14001403. (12) Samson, R. J.; Mulholland, G. W.; Gentry, J. W. Structural analysis of soot agglomerates. Langmuir 1987, 3, 272-281. (13) Kaye, B. H.; Clark, G. C.; Leblanc, J. E.; Trottier, R. A. Image analysis orcedures for characterizing the fractal dimension of fineparticles. Part. Charact. 1987, 4, 63-66. (14) Schmidtt-Ott, A. New approaches to in-situ characterization of ultrafine agglomerates. J. Aerosol Sci. 1988, 19, 553-563. (15) Rogak, S. N.; Flagan, R. C.; Nguyen, H. V. Mobility and structure of aerosol agglomerates. Aerosol Sci. Technol. 1993, 18, 25-47. (16) Meakin, P. Computer simulation of cluster-cluster aggregation using linear trajectories: results from three-dimensional simulations and a comparison with aggregates formed using Brownian trajectories. J. Colloid Interface Sci. 1984, 102, 505-512. (17) Skillas, G.; Ku ¨ nzel, S.; Burtscher, H.; Baltensberger, U.; Siegmann, K. High fractal-like dimensions of diesel soot agglomerates. J. Aerosol Sci. 1998, 29, 411-419. (18) Virtanen, A.; Ristima¨ki, J.; Marjama¨ki, M.; Vaaraslahti, K.; Keskinen, J.; Lappi, M. Effective density of diesel exhaust particles as a function of size. SAE Tech. Pap. 2002, No. 2002-01-0056. (19) Park, K.; Cao, F.; Kittelson, D.; McMurry, P. Relationship between particle mass and mobility for diesel exhaust particles. Environ. Sci. Technol. 2003, 37, 577-583. (20) Harris, S.; Maricq, M. Signature size distribution for diesel and gasoline engine exhaust particle matter. J. Aerosol Sci. 2001, 32, 749-764. (21) Harris, S.; Maricq, M. The role of fragmentation in defining the signature size distribution of diesel soot. J. Aerosol Sci. 2002, 33, 935-942. (22) Maricq, M. M.; Chase, R. E.; Podsiadlik, D. H.; Vogt R. Vehicle exhaust particle size distributions: A comparison of tailpipe

2556

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 9, 2004

(23) (24)

(25) (26) (27) (28)

(29) (30)

(31)

(32)

(33)

(34)

(35)

and dilution tunnel measurement. SAE Tech. Pap. 1999, No. 1999-01-1461. Kauppinen, E. I.; Pakkanen, T. P. Coal combustion aerosols: a field study. Environ sci. Tecnol. 1990, 24, 1811-1818. Ntziachristos, L.; Giechaskiel, B.; Pistikopoulos, P.; Samaras, Z.; Mathis, U.; Mohr, M.; Ristima¨ki, J.; Keskinen, J.; Mikkanen, P.; Casati, R.; Scheer, V.; Vogt, R. Performance Evaluation of a Novel Sampling and Measurement System for Exhaust Particle Characterization. SAE Tech. Pap. 2004, No. 2004-01-1439. Wang, S. C.; Flagan, R. C. Scanning electrical mobility spectrometer. Aerosol Sci. Technol. 1990, 13, 230-240. Keskinen, J.; Pietarinen, K.; Lehtima¨ki, M. Electrical low-pressure impactor. J. Aerosol Sci. 1992, 23, 353-360 Liu, B. Y.; Pui, D. Y. H. J. Colloid Interface Sci. 1974, 47, 155-171 Marjama¨ki, M.; Ntziachristos, L.; Virtanen, A.; Ristima¨ki, J.; Keskinen, J.; Moisio, M.; Palonen, M.; Lappi, M. Electrical filter stage for ELPI. SAE Tech. Pap. 2002, No. 2002-01-0055. Kasper, G. Dynamics and measurement of smokes. I. Aerosol Sci. Technol. 1982, 1, 187-199. Kelly, W. P.; McMurry, P. Measurement of particle density by inertial classification of differential mobility analyzer-generated monodisperse aerosol. Aerosol Sci. Technol. 1992, 17, 199-212. Hering, S. V.; Stolzenburg, M. R. On-line determination of particle size and density in the nanometer size range. Aerosol Sci. Technol. 1995, 23, 155-173. Schleicher, B.; Ku ¨ nzel, S.; Burtscher, H. In situ measurement of size and density of submicron aerosol particles. J. Appl. Phys. 1995, 78, 4416-4422 Ristima¨ki, J.; Virtanen, A.; Marjama¨ki, M.; Rostedt, A.; Keskinen, J. On-line measurement of size distribution and effective density of submicron aerosol particles. J. Aerosol Sci. 2002, 33, 15411557. Virtanen, A.; Ristima¨ki, J.; Keskinen, J. Method for measuring effective density and fractal dimension of aerosol agglomerates. Accepted to Aerosol Sci. Technol. Heywood, J. B. Internal Combustion Engine Fundamentals; McGraw-Hill Book Company: New York, 1988.

Received for review October 13, 2003. Revised manuscript received January 14, 2004. Accepted February 11, 2004. ES035139Z