Diesel emissions in Vienna

Atmospheric Environment Vol. 22, No, 7, pp, 1255-1269, 1988. Printed in Great Britain.

0004 6981/88 $3.00+0.00 Pergamon Press pie

D I E S E L E M I S S I O N S IN V I E N N A * H. HORVATH, I. KREINER, C. NOREK a n d O. PREINING Institute for Experimental Physics, University of Vienna, Vienna, Austria

and B. GEORGI Niedersfichsiches lnstitut ffir Radio6kolgie, Hannover, F.R.G. Abstract--The aerosol in a non-industrial town normally is dominated by emissions from vehicles. Whereas gasoline-powered cars normally only emit a small amount of particulates, the emission by diesel-powered cars is considerable. The aerosol particles produced by diesel engines consist of graphitic carbon (GC) with attached hydrocarbons (HCs) including also polyaromatic HCs. Therefore the diesel particles can be carcinogenic. Besides diesel vehicles, all other combustion processes are also a source for GC; thus source apportionment of diesel emissions to the GC in the town is difficult. A direct apportionment of diesel emissions has been made possible by marking all the diesel fuel used by the vehicles in Vienna by a normally not occurring and easily detectable substance. All emitted diesel particles thus were marked with the tracer and by analyzing the atmospheric samples for the marking substance we found that the mass concentrations of diesel particles in the atmosphere varied between 5 and 23/lg m- 3. Busy streets and calm residential areas show less difference in mass concentration than expected. The deposition of diesel particles on the ground has been determined by collecting samples from the road surface. The concentration of the marking substance was below the detection limit before the marking period and a year after the period. During the period when marked diesel fuel was used, the concentrations of the diesel particles settling to the ground was 0.0124).07 g g - t of collected dust. A positive correlation between the diesel vehicle density and the sampled mass of diesel vehicles exists. In Vienna we have a background diesel particle concentration of 11 #g m 3. This value increases by 5,5 #g m 3 per 500 diesel vehicles h- 1 passing near the sampling location. The mass fraction of diesel particles of the total aerosol mass varied between 12.2 and 33%; the higher values were found in more remote areas, since diesel particles apparently diffuse easily. Estimates of diesel particle concentration by emission inventory or by using lead concentrations as an indicator for vehicle emissions gave similar values to those obtained in this study. Using available cancer risk data and diesel particle concentration found in this study, 1-2.6 additional lung cancers per 100,000 persons yr- 1 breathing diesel emissions in the measured concentration the whole lifetime can be expected. Key word index: Soot, diesel particles, vehicle emissions, urban pollution. INTRODUCTION A non-negligible portion of the atmospheric aerosol has been found to be light absorbing, especially in large towns (e.g. Denver aerosol: Wolff et al., 1982; Davis, 1984; Hasan and Dzubay, 1983; Lewis and Dzubay, 1986; Sapporo aerosol: Ohta and Okita, 1984; Portland aerosol: Shah et al., 1984; Los Angeles aerosol: Pratsinis et al., 1984; New York aerosol: Shah et al., 1985; Arctic haze: Barrie et al., 1981, Rosen et al., 1981, and many other studies). Most of the light absorption is attributed to graphitic carbon (GC) which originates from combustion processes. Graphitic carbon, the main constituent of soot, is detectable by its light absorption (Fischer, 1970; Lin et al., 1973) or by thermochemical methods (Novakov, 1982) but it is not easy to distinguish between G C coming from different sources. * This paper has been presented in part at the 3rd International Conference on Carbonaceous Particles in the Atmosphere, 5-8 October 1987, held at Lawrence Berkeley Laboratory, University of California. AE

22-7-A

1255

Sources for G C in an urban environment are all combustion processes, such as power plants, home heating, district heating, steam generators, fireplaces and all internal combustion engines. Depending on the fuel burnt and the pollution control equipment used, emissions may vary widely. Very little G C is emitted by combustion processes using natural gas. Also, large furnaces or steam generators produce little soot, since combustion is very well optimized. In most cases, small furnaces for home heating, especially when using solid fuels, and motor vehicles are the main contributors to the G C suspended in the air over a large town. Whereas gasoline-powered engines have considerable soot emissions during starting operations only, diesel vehicles emit G C to a varying degree under all operating conditions. Since in the visible range, G C is the major lightabsorbing species contained in the atmospheric aerosol, studies of the light absorption coefficients can give information on the content of G C in the atmosphere. A long record of the light absorption coefficient of the Vienna atmosphere has shown both a daily

1256

H. HORVAT'Het al.

and a yearly variation: an absorption coefficient larger by more than a factor of two is found in winter compared with summer. Obviously, emissions by domestic heating, which in many cases is by small furnaces operating on solid or liquid fuels, are one cause of this. Both in winter and the summer the light absorption coefficient also exhibits a daily pattern which very closely resembles the traffic pattern. This indicates that traffic will be a major source of lightabsorbing particles. The particulate emissions of diesel engines are almost exclusively soot particles. They are formed during the combustion process by lack of oxygen. Soot particles consist of GC in combination with HCs which can have their origin both in the incomplete combustion of the fuel and lubricating oils. The nonGC component (organic solvent fraction) of the soot particles depends on the engine type, maintenance, and operation, and can vary widely. Reported values for the organic solvent fraction of soot particles range from 10 to 90% of the total mass. The size of the particles emitted by diesel engines ranges from sub-/~m to several/~m, with a mass mean diameter in the order of tenths ofa #m. The mass of the particles emitted by diesel-powered passenger cars is in the order of 0.2 g k m - ~ driven; the diesel population in Vienna will emit approx. 1.6 t day ~ of particll~s. Graphitic carbon itself is considered completely inert, but it has a high capability for adsorption. Since soot emitted by diesel vehicles also contains aromatic HCs, carcinogenic effects have to be considered (Rondia et al., 1983; Pederson 1983; Schuetzle et al., 1983; Stoeber, 1986). Furthermore, the major fraction of the diesel particles is in a size range which can be deposited deep in the lung. Thus diesel particles can act as a vehicle of transport of adsorbed substances in the lung (which would otherwise not get there) and deliver the absorbed substances to the body at a small rate. Under these circumstances, diesel particles represent a cancer risk. The fate of diesel particles after being emitted from the tailpipe of the cars is difficult to predict. They may sediment, be re-entrained, coagulate, diffuse to surfaces, be washed or rained out, or carried to some other location by air movement and other means. Obviously, diesel particles will contribute to the dark appearance of the dust of the filter samples. Unfortunately, soot originating from diesel vehicles cannot be distinguished from soot of other origins. Thus the amount of diesel particles cannot be determined directly. So far it could only be estimated from model considerations or from comparison with trace elements which are specific for emissions from motor vehicles (e.g. lead or bromine). In this paper we will describe the results of a study which allowed a more direct determination of the diesel particles by a tracer method. The tracer was added to all the fuel sold in Vienna and the vicinity. By measuring the amount of tracer in the atmospheric

samples, the contribution of diesel vehicle emissions to the pollution in Vienna could be determined. EXPERIMENTAL METHOD

Since no unique tracer for diesel emissions has been found up to now, it was necessary to add a tracing substance to the fuel used by diesel vehicles. Vienna has the advantage of being supplied by a single refinery, therefore all the fuel sold at the gas stations and therefore used by most of the cars in Vienna comes from the same producer. Only fuel put into the tank of a car or truck outside the area where marked fuel was sold during our experiment (which was a distance of about 150km) would not contain the tracer. Vehicles not registered in Vienna and the vicinity are scarce, therefore no large disturbances by vehicles using unmarked fuel were to be expected. The substance found most suitable by us as a marker was the rare earth element dysprosium: its concentration in the atmospheric aerosol in Vienna and other towns was below the detection limit of our method of analysis. The organo-metallic compound tris-dipivalyl-methanato-dysprosium, usually abbreviated as Dy(DPM)3 with the formula Dy(CH(COC(CH3)3)2) 3 was sufficiently soluble in diesel fuel to permit the generation of a master solution of 4 kg of the marking substance which was continually added to approx. 50,000 m 3 of diesel fuel during production. During combustion, Dy adhered to the soot particles formed, giving rise to ambient concentrations of Dy in the order of 100 pg m 3. The Dy mass concentration of the atmospheric aerosol was measured at several sites by filtering a volume of 10-40 m 3 through nuclepore or glass-fiber filters. The sampling time ranged from 12 to 48 h. The filters were weighed before and after sampling, thus the total mass of the suspended particulate matter could be obtained. For the determination of the Dy content of the samples an extraction process was necessary, since both the filter material and the atmospheric particulates contained elements which are easily activated with thermal neutrons and interfere with the detection of the Dy-7-1ines. Using four times a 10% HNO3, we could extract all the Dy but leave enough of the interfering elements in the sample, so that the background noise for the V-spectroscopy could be greatly reduced. The extracts were put into containers suited for activation analysis and subsequently vacuum dried. The ambient concentrations during the experient of around 100pgm -3 of Dy gave enough pulses to make the tracer detectable with a reasonable time effort. It was necessary to optimize the activation analysis with respect to irradiation time and detection limit; for details of the analysis, preparation of samples and extractions used, see Georgi et al. (1987).* *The authors wish to thank Professor Kiihn of the Nieders/ichsisches Institut fiir Radio6kologie for his cooperation.

Diesel emissions in Vienna Occasionally, we also collected samples of settled dust at a few locations. This was done by gathering at a predetermined area the dust on the street by means of a brush, until about 4 cm 3 was collected. The area needed to collect such a sample varied from location to location. The settled dust samples then were extracted in the same manner as the filter samples and analyzed for Dy. Samples collected 2 yr after the study and at towns upwind of Vienna showed no detectable Dy in the samples, i.e. outside the study period Dy was either not present in the dust samples and/or not extracted. During a period of 4 weeks, starting from 13 June 1984, all the fuel put into tank trucks at the refinery was marked with Dy. Since all fuel sold at gas stations in Vienna and the vicinity (up to 150 km) is supplied by tank trucks, the area of at least Vienna and Nieder6sterreich was supplied with marked fuel. Only diesel fuel carried to distant fuel depots by rail was unmarked. In the initial phase the concentration of the diesel fuel at selected gas stations was measured; it took about halfa week to rise to the value supplied by the refinery. Within the city of Vienna and upwind of the town we selected several locations where our sampling equipment was operated. The locations are shown in the map (Fig. 1). It was not always possible to find an 'ideal' site, since the availability of a building giving protection against vandalism and the need for electricity only left a few locations for choice. A short

1257

description of the sites and the traffic density at 09.00 h (which is already after the morning peak) follows.

Location 1, WfihrinoerstraBe 13 This site was located very close to the city centre on a four-lane street canyon going downtown. Forty m nearer the city was an intersection with traffic lights. The cars usually had to stop at this location. A crosssection through the canyon is given in Fig. 2. During morning and evening, rush hour traffic only proceeded very slowly and prolonged stops were frequent. At 09.00 h the traffic density in vehicles h-1 was 1542 passenger cars (gasoline), 366 passenger cars (diesel) and 228 trucks (diesel). The sampling inlet was located on a pavement at a relative height of 1 m. Thus a large fraction of resuspended coarse particles could be expected here. The buildings on both sides of the street had a relative height of 25 m.

Location 2, rooftop laboratory, Strudelhofgasse 4 The sampling was performed on the roof of the Physics Building of the University of Vienna. It was located 27 m above ground, and the sampling inlets were about 1 m above the roof. The nearest street, Strudelhofgasse, has a low traffic density, but W~ihringerstraBe, with a traffic density of 2136 vehicles h - I at 09.00 h is only 75 m to the west; the horizontal distance to station 1 is 550 m to the southsoutheast. Figure 3 shows a vertical cross-section through the measuring site.

358

k6s

0 1 2 3 4 5 krn

s42,~ ''~

489

4

XlX.

~5,~ I

--

- -. XXl.

"~a

\

I t

,/

X~l /XVUL >,

4;B XlV.

514

XUL

/

XXlll. 578

i

38B ÷ / °~t/X~'

] . iv.

~~

~8

t

I 527 'r

XXll.

X[L

08

011 ~

,~1~ 2;1'

~

Xl. 10 °

~

+

163

1~4

Fig. 1. Map of Vienna, indicating the locations of the sampling sites. Vienna is surrounded by mountains in the west and north and by plains to the east and south which extend at least to 30 km. One-digit numbers give the number of the sampling station, three-digit numbers give the elevation above sea-level and roman numerals give the district numbers. The densely populated area is circled by a dotted line.

H, HORVATH et 1258

?

ENE

\

Location 3, Neuer Markt, pedestrian zone

,SAMPLING INLET W:ihr ingerst r a r e

s'

1'o

i~,

al.

2% ~.

The location was a downtown square 120 m long and 40 m wide at the beginning of a pedestrian zone. Access to this location by motor vehicle was only possible in one direction, and through traffic was not possible. Therefore vehicles in this areas were either looking for a parking space or delivering. The traffic density at this location was 132 gasoline passenger cars, 36 diesel passenger cars and 33 trucks h -~. Unfortunately, during the study period, construction work was performed within a distance of 10 m of the measuring site, which included the use of a compressor powered by a diesel engine of approx. 30 hp. It was impossible to find out whether this engine used marked fuel or not. A section through the measuring site is shown in Fig. 4.

Fig. 2. Scheme of sampling location 1, W/ihringerstral3e 13. W

E

_E~_,r(..SAMpLlNG INLET /

aJ~ t r u

e

I

0

h o

f

W~,hringerst r abe

g a s s e

10

20

m

Fig, 3. Scheme of station 2, Strudelhofgasse 4.

I i i

PEDESTRIANS/, i ONLY

WNW

\ VEHICLES. parked

VEHICLES,

'

looking

10

for parking

20 m

Fig. 4. Characteristics of station 3, Neuer Markt.

Diesel emissions in Vienna

Location 4, Gaudenzdorfer O~rtel 73 The sampling site was close to a high-capacity highway system circling the inner part of the town on a radius of approx. 3 km. This highway has four-lanes in each direction; at the sampling site the two directions were separated by a large park. During rush hours, congestions are frequent. The sampling inlet was at a distance of 37 m from a curve of the highway, 100 m after traffic lights, and near the wall of a building. All cars decelerated before the curve and accelerated after the curve. The traffic density was 1460 gasoline passenger cars, 540 diesel passenger cars, and 330 trucks h - 1.

1259

ered passenger cars, 90 diesel passenger cars and 84 trucks and buses. Near the garage a four-lane highway with 1500 cars h - 1 (31% diesel) passed over Kendlerstra6e. A map of the area is shown in Fig. 5.

Location 6, Gallizinberg, LiebhartstalstraBe 31 Situated ,at the slope of the hills surrounding the northwest part of Vienna in a residential area on a dead-end street. Eighteen passenger cars (gasoline) and three passenger cars (diesel) h - 1 passed by. This area can be characterized by one-family homes with gardens (ca. 600 m2); the main vegetation was trees and shrubs. The sampling inlet was 10 m from the road pavement on a lawn.

Location 5, Kenndlerstral3e 40 The sampling inlet was located on a lawn 15 m off the pavement of a two-lane highway. At a distance of 350 m, a garage for 96 city buses was situated. Especially in the morning and evening a heavy bus traffic could be observed on this road. After the morning rush hour, the traffic density was 378 gasoline-pow-

Location 7, Exeiberg This sampling site was situated slightly outside Vienna, northwest of the city centre, on a hill with an altitude of 516 m. As the prevailing winds are from the northwest, this location could be considered upwind of the town. Under stable fair-weather conditions the

~

GE

FL(JTZERSTEiG

v

i

z ~J Q i,i

2

I00 m

@ Qz

W

g tm

Fig. 5. Map of the area around station 5, KendlerstraBe40.

1260

H. HORVATHet al.

direction of slow air movement normally is from the southeast, therefore under these conditions the station is normally, at least to some extent, in the urban plume. The sampling inlet was 1 m above ground 100 m from a highway. Traffic on this road was low, 90 vehicles h - l , with 33% diesel vehicles.

bread factory, having trucks continuously leaving the area. The nearby through-street had a traffic density of 954 cars h -1, with 8% passenger diesel and 18% trucks. The area west of Theodor Sickel Gasse was residential, with large complexes up to six storeys high.

Location 8, Belgradplatz

Location 10 Kaiser Ebersdorf Albernerstral3e 8

This station was situated in the southern part of the town. The sampling inlet was at a height of 2 m on a pavement 5 m from the road pavement. The street went from east to west, the traffic density was intermediate: 312 vehicles h - l (15% diesel passenger, 12% trucks).

Sampling was performed on the pavement of a street with a traffic density of 114 vehicles h-x total (20% passenger diesel, 12% trucks) in the south part of Vienna. The area was partly residential, partly industrial and partly agricultural, mainly vegetable growing; 135 m to the east of the sampling intake a four-lane freeway going to the south passed by. We counted 881 vehicles h - l , 15% of them passenger diesel cars and 21.5% trucks. Another four-lane highway passed by 350 m to the west of this location, having 580 vehicles total h 1, with 12.1% passenger diesel and 13.8% trucks.

Location 9, Laaerber9, Theodor Sickel Gasse 1 This site was located in a small forest owned by the city of Vienna, at a distance of 160 m after a junction with a through street, as can be seen from Fig. 6. The sampling inlet was closer to Theodor Sickel Gasse, which had the lesser traffic density, 372 vehicles h - t , but a large fraction of diesel vehicles (13% passenger diesel cars and 32% trucks) due to a nearby large

/

/

? 2O i

40 i

?

60m , i

Q

i/

P

?


q, ?

m

,ll

? ? <1

?

Location 11, Liesing, an den Steinfeldern 6 The sampling site was located in the southern industrial area of Vienna, 10 m from the road pavement in a lot without buildings. Almost half of the vehicles driving in the immediate vicinity were diesel vehicles: 192 vehicles h - 1 (27% trucks, 18% passenger diesel vehicles) on the adjacent street (an den Steinfeldern) and 594 vehicles h -~ (23% trucks, 23% passenger diesel vehicles) on the street at the nearest intersection (BrunnerstraBe). Adjacent to the lot with the sampling location was a truck manufacturing plant and a truck yard with 50 trucks parked (30% of which had running motors using power for loading and unloading with hydraulic cranes). A map of the area is given in Fig. 7. A cross-section in a vertical plane between several locations is given in Fig. 8. In addition to the analysis for the tracer element, the following data were collected: the light absorption coefficient was determined on Nuclepore filters, using the integrating plate method (Lin et al., 1973) at locations 1, 2, 3, 6, 7 and 9. Mass size distributions of the atmospheric aerosol were obtained by using Berner low-pressure impactors (Berner, 1984) at stations 1, 2, 3, 6 and 9. The size-separated particles were deposited on plastic foils; the weight increase of the foils was determined with an analytical balance. No grease was used to enhance the sticking of the particles to the foil. The sampling time was 24 h.

? ?

?

Fig. 6. Map around sampling station 9, Theodor Sickel Stral3e 1.

V E H I C L E P O P U L A T I O N A N D E N E R G Y C O N S U M P T I O N IN VIENNA

The city of Vienna is located in central Europe at a latitude 48.12 ° N, longitude 16.35 ° E. The city lies at an altitude between 170 and 250 m; in the northwest it is surrounded by mountains of an altitude of approx. 500 m. Winds are generally from north to west, but

Diesel emissions in Vienna

1261

I

TRUCK MANUFACTURING PLANT

N

TRUCK LOAD A N D UNLOAD GRASS

o3

d

I

° SAMPLING LOCATION

AN DEN STEINFELDERN

//

0

50

i

100 m

i

i

Fig. 7. Map of the area around station 11, Liesing.

E

7 8

SE

Do

o

z o

LIJ

E

NW

0 i

1

2

3

4

i

i

i

5

5

7

8

9

10

i

i

~

i

i

HORIZONTAL DISTANCE

Fig. 8. Vertical cross-section along lines, connecting sampling stations. Coming from the west, the mountains surrounding Vienna are easily visible (top). The slope down to the city centre is shown in the middle, and the fairly fiat terrain southeast of the centre at the bottom.

under stable fair-weather conditions light winds from the south to east are frequent. Most of the jobs of the 1.7 million inhabitants of Vienna are in trade, administration and public services, and no heavy industry exists in the town or in the vicinity. Most of the industrial operations concentrate on processing of semi-finished products. The only major exception is the refinery at Schwechat, located approx. 3 km to the southeast of sampling site

10 (see Fig. 1). A short characterization of production and industry of Vienna is given in Table 1, and a summary of the energy consumption is given in Table 2. These tables show that particulate pollution can be expected mainly from space heating and traffic. From April to September, heating is not necessary, thus traffic is expected to be the major source for pollution. Only 7% of the vehicles registered in 1982 in Vienna

1262

H. HORVATHet al.

Table 1. Characteristics of Vienna Population: Area:

1,700,000 414 km 2

Elevation Location

170-250m 48.12°N,16.35°E

Gross product (%) Agriculture Energy, water Industry, trade Building Services Public services

Industry (%) 1.4 1.7 24.6 9.6 41.6 11.1

Chemical Food Electrical Textile Machinery Paper, iron, glass

17 20 16 10 11 0

Table 2. Energy consumption in

Vienna, 1982

Solid fuels Liquid fuels Automotive fuels Electricity District heating Natural gas

TWh

%

1.2 7.0 7.0 5.7 1.7 7.1

4 23 23 20 6 24

Table 3. Vehicles registered in Vienna and

fuel consumption (1982) Motor vehicles Total 520,914 (100%) Diesel-powered 36,718 (7%) Diesel-powered vehicles Total 100% Passenger cars 32% Station wagons 5% Trucks and buses 63% Trucks Total 100% 0-1 ton 29% 1-2 ton 24% 2-5 ton 20% Over 5 ton 27% Fuel consumed Total 592,000 ton (100%) Regular 13% Premium 55% Diesel fuel 32%

are powered by a diesel motor, but they consume 32% of the fuel, since diesel vehicles are mainly commercial vehicles. This is in agreement with traffic density determinations at the different sampling sites, where normally about 30% of the counted vehicles were diesel powered. The majority of the trucks have a gross weight below 5 tons. A summary of the relevant data for traffic is given in Table 3.

DETERMINATION OF THE DIESEL PARTICLE CONTENT OF A PROBE FROM ITS DY CONTENT

In the cylinder of a diesel engine air is compressed to a pressure of approx. 45 bar and heats up to a

temperature of approx. 600°C. Diesel fuel is injected under high pressure, leaving the injection nozzle as small droplets. The fuel rapidly evaporates and combustion takes place both in the gaseous mixture of diesel fuel vapor and air and at the surface of the fuel droplets. The droplets heat up and cracking of the fuel occurs. Since at no time does a homogeneous mixture of fuel and air exist, there are always areas with lack of oxygen, where GC is generated. This is especially the case at high load, where the fuel-air mixture is very inhomogeneous. The primary soot particles are small plates approx. 50-100 C atoms thick and 2-3 nm wide. They are formed within several #s. Within a timespan of 50 #s these primary particles coagulate to spherical particles with a mean diameter of 26 rim. Subsequently, these spheres form chain aggregates, which can have lengths up to 30 #m; their aerodynamic mass mean diameter is in the range of 0.08 #m. Besides GC the soot particles also contain HCs which originate from incomplete combustion of the fuel and lubricants. A variety of compounds has been found so far: aldehydes, sulfates, polycyclic aromatic hydrocarbons (PAH) and nitro PAHs have been reported. Depending on the age, the state of maintenance, and the operation of the diesel engine, the fraction of HCs adsorbed to the GC (organic solvent fraction) can vary between 10 and 90%. Small organic solvent fractions can be found in emissions from new, well-maintained engines under optimum load, whereas soot with a high organic solvent fraction is emitted by old, unmaintained engines and by all engines operated under varying loads. When diesel fuel, containing Dy (DPM)3 in solution, is used, the high flame temperatures will cause the oxidation of the Dy. Dy203 is an inert refractory material with a melting point of 2300°C and during the rapid coagulation process of the primary and secondary GC particles the Dy203 is expected to be incorporated in the soot particles. We have found that all of the Dy delivered to the motor with the fuel is contained in the emitted particulates (Georgi et al., 1987). For a vehicle with a known emission factor for particulates it is thus very simple to calculate the concentration of the Dy in the emitted soot. Let c be the concentration of Dy in the fuel (e.g. g f 1) and e the emission factor of the vehicle (e.g. in g emitted particulates f - 1 of fuel); the concentration of Dy in the emitted soot is given by c/e. When determining the Dy content of a sample, the soot content in this sample can therefore be automatically obtained if the emission factor and the concentration of the marker in the fuel are known. The only crucial point is the emission factor for the 37,000 different diesel vehicles in Vienna. Obviously, it is not possible to obtain an emission factor for each vehicle; only an average emission factor appears appropriate. Fortunately, the emission factors for trucks and passenger cars are almost alike, since larger vehicles emit a larger quantity of particulates km-1 but also have a higher fuel consumption.

Diesel emissions in Vienna

1263

>_ e,~ ~,

~. tO

11. o

I

I

I

I

I

I p,j

AERODYNAMIC

DIAMETER

(

Hm

~

e,l

}

Fig. 9. Size distributions of diesel engine particulate emissions. The gravimetric mass of the particulates emitted by a standard passenger car diesel engine during ECE 15 H test cycle is shown for two identical test runs. The mass per stage of the impactor is given in per cent of the total sampled mass, and the centrepoint of the size interval sampled by each stage is given on the abscissa. The insoluble fraction is given as broken line.

We have determined the soot emissions of standard diesel motors for passenger cars with cascade impactors. Obviously, the size distributions of the emitted particles depend on many factors such as speed, load, acceleration or deceleration, motor temperature, habits of the driver and other factors. Besides this, it is known that identical tests or identically manufactured motors give different emissions under the same conditions (see, for example, Barsic, 1984). This is demonstrated in Fig. 9: For two identical test cycles ECE 15 H the mass size distributions for the same motor is shown. For both tests one can see a maximum of the size distribution on the stage with a mass mean diameter of 0.087 pm. After weighing the foils with the deposited particles, all the samples were dried under vacuum and thus the HCs were removed from the GC. The organic insoluble fraction is given as a broken line in the figure. F r o m these tests an emission of 0.2 g k m 1 has been found. This is in agreement with other data available. For the diesel vehicle population an average fuel consumption of 8.22 {/100 km was used, giving an emission factor of 2.43 g soot { - 1 of fuel for passenger cars and station wagons. Trucks have both a higher emission of soot km 1 as well as higher fuel consumption. Typical values for European trucks are 0.4-0.6 g k m - 1 emitted particulates with a fuel economy of 20-25 f/100 km, which gives an emission factor of 2.0-2.4 g f - 1, which is almost identical with the value found for passenger cars. Emission factors published by the U.S. EPA (1973) have the following

values: 4.23 g { i for a pre-1973 passenger diesel car and 1.62 g f t for heavy-duty trucks under American driving conditions and a low fuel efficiency of 50 f/100 km. For diesel locomotives an emission factor of 3 g E - 1 is reported. The 1986 average European diesel engine particulate emissions were 0.8 g per horsepower h* (Pischinger, personal communication). With a density of diesel fuel of 850 kg m - 3, a calorific value of 41.8 M J k g - 1 and an average efficiency of 0.27 for the diesel engine under varying load, the emission factor for truck diesel engines is 2.9 gE 1. Rykowski and Brochu (1986) use a value for the emissions of the average 1984 European truck of 0.705 g per horsepower h, which gives an emission factor of 2.55 g f ~. All the values for emission factors listed in this paragraph, both for very small European passenger car motors, truck motors and large locomotive engines, are in a very narrow range and actually are independent of the size of the engine. We have used the value of 2.4 g E- ~, which was based on our own experiments. The standards which will be enforced in the near future will reduce the emissions by diesel vehicles considerably. We have measured emission factors of newer developments of small passenger diesel engines which were a factor of 0.6 lower than most of the engines in use at present. This will probably be sufficient to meet the future standards. Larger motors will need filters which remove the soot particles to

*1 horsepower h=2.685 MJ.

H. HORVATHet al.

1264

meet the standard. The same applies for the engines used in trucks, where emission standards for the particulates of 0.1 g per horsepower h are under consideration. The concentration, c, of the Dy marker in the diesel fuel was 1 kg per 50,000 m 3, the average emission factor, e, of the diesel vehicles was 2.4 g f ~; therefore the conversion factor from collected Dy to diesel particulate mass was e/c = 120,000.

MEASUREMENT ERRORS

During the time when marked fuel was used by all diesel vehicles, the atmosphere contained approx. 1 0 0 p G D y m - 3 . The filter samples contained 1-4 ngDy, and 400-4000/~g of suspended matter, which is a ratio of 1 in 400,000 to 1,000,000. The crustal concentration of Dy is 4.6 x 10 -6, which is comparable with the concentration of Dy in our samples. Therefore a strong interference with naturally occurring Dy could be expected. Fortunately, this was not the case for several reasons: (1) the contribution of crustal Dy to the atmospheric aerosol in a town is very small; the major constituents of the aerosol are sulfates, nitrates and organic matter, and (2) the extraction technique applied by us apparently was very efficient for Dy emitted by diesel vehicles and poor for crustal Dy (if present in the atmosphere at all). This was tested in two ways: 10 samples of diesel emissions obtained directly from the exhaust of a vehicle using marked fuel were extracted three times with diluted H N O 3 in the standard way described by Georgi et al. (1987) and the expected quantities of Dy were found. The accumulated fourth extracts of these samples contained no measurable Dy. Thus three extracts were already sufficient to collect all the Dy of the samples from diesel emissions. Before the experiment, and 1 yr after, the experiment atmospheric samples were collected and analyzed in exactly the same way as during the experiment. No Dy could be detected in these samples. The theoretical detection limit for Dy in a sample with no interfering substances lies between 0.00019 and 0.018 ng: the values mainly depend on the available neutron flux, the counting time and the detector used. In our case, other substances present in the atmospheric aerosol were also activated, giving rise to a background which made it impossible to detect these small quantities in samples; besides this, it was not possible to use the optimal irradiation flux and counting time, since a considerable quantity of samples had to be analysed. Under these conditions, a detection limit of 1 ng was achieved. For a 40 m 3 sample this meant that the lowest detectable mass of Dy was 25 pg m - 3 of material extracted from the filter. If the

natural background of Dy after extraction had a value below this figure, it would not have been detected by us. The natural background (without extraction) at the nuclear research center of Karlsruhe was found to be 50 pg m - 3 (Vogg and Hartel, 1976); no Dy in algae has been found by Bowen (1966). The samples drawn from the atmosphere therefore contained amounts of Dy, which were three to eight times larger than the detection limit. Obviously, we encountered large statistical fluctuations, which were caused by the low quantities to be detected and by unexpected interference by activated substances, but mostly by fluctuations caused by changing meteorological conditions. Therefore in this paper we will only present the results of the accumulated data over the total measuring period. The accuracy of the values for the concentration of diesel particles in the atmosphere is 35% of the value given.

RESULTS

Suspended particulates The measured mass concentration of diesel particles was obtained from the measured Dy content of the samples by multiplying by the factor 120,000. Since it was not possible to detect Dy masses below 25 pg m - 3 and the possibility of a natural background could not be excluded, we have subtracted this value, which corresponds to 3/~g diesel particles per m - 3, from all values. In Table 4 the results for the stations are given. Besides the mass concentrations of the total suspended particulates and the mass concentration of the diesel particles, the light absorption coefficient of the aerosol and the mass of the accumulation and coarse mode are also listed at several stations.

Table 4. Measured mass concentration, M, mass concentration of diesel particles, D, light absorption coefficient, L, mass concentration of the accumulation Mode, A, and mass concentration of the coarse mode, C, for the different sampling stations M Station 1

2 3 4 5 6 7 8 9 10 11

L

Day Night 172 39 67

94 34 31 99 83

33 28

-23 71

54

43 94 77

D 15.2 12.7 15.2 22.4 13.9 10.3 5.5 12.7 15.2 11.5 23.6

Day Night

A

110 82 17 21 49 24 -. . . 9 9 5

37" 60 22 8 23 35

25

19 --

C

. 14

7

28 23 - ....

The mass concentrations are given in yg m- 3, the absorption coefficient in M m - 1. All data are averages over the total measuring period.

Diesel emissions in Vienna

1265

3

A,0

z o

5

2'0

2; JUNE

;

,'o JULY

1; 198/,

Fig. 10. Time dependence of the fraction of diesel particles in dust samples collected on the street. The numbers are the locations where the samples were collected. Data points marked with an A are from samples collected at location A in Fig. 1. The samples were taken at the days indicated. The connecting lines have been added for convenience and do not imply that the concentration varied continuously between the data points.

Sedimented particulates The time dependence of diesel particles collected on the horizontal road pavement at a few locations is given in Fig. 10. The data were obtained by collecting a certain a m o u n t of dust on the street by brushing at a predetermined location until approx. 4 cm 3 were collected. The mass of diesel particles per total mass is given in this figure. DISCUSSION

This section is divided in two parts; first, specifics of each measuring station will be discussed, then interrelationships and comparisons with other studies will be treated.

Specifics of the sampling sites Station 1 (street canyon) This is obviously heavily influenced by the traffic. A high concentration of the suspended particulates was found compared with other sites, because the inlet was at only 1 m above street level and resuspended particles are easily collected in this way. The large difference of a factor of 1.8 in mass concentration between day and night reflects less traffic at night. The same also applies to the light absorption coefficient, but here the difference is only a factor of 1.3, indicating slightly different sources for the light-absorbing aerosol and the main mass of the aerosol. When inspecting the filter samples we found both a homogeneous deposit (which is caused by sub-/~m particles) and dark large particles which probably show sufficient light absorption but are brought into the airborne state due

to resuspension. The coarse mode has twice the mass of the accumulation mode; this again is caused by the sampling at the 1-m level and the resuspension. The concentration of diesel particles of 15.2/~g m - 3 appears fairly low (12% of the suspended particulates) in comparison with stations where a much lower mass concentration was found.

Station 2 (rooftop laboratory) This was only at a distance of 0.5 km from the measuring site discussed above, situated on the same street, but at an elevation of 27 m. In comparison with the sampling site in the street canyon the coarse-mode particles have decreased drastically whereas the accumulation mode shows no substantial change. There is no large difference between day and night, indicating an aerosol which is probably of an average age over some tens of hours. The light absorption coefficient is slightly higher during the night. This has been observed already for several years, but cannot be explained so far and might be due to increased emissions of large sources during the night. The mass concentration of the diesel particles is not much less than at the station located at street level on the same street. This indicates that the diesel particles, being in the size range of the accumulation mode, can easily be transported to an elevation of 30 m.

Station 3 (downtown pedestrian zone) At this station, unfortunately, a diesel compressor was operated during the sampling period. The impact of this can well be seen by the differences of more than a factor of two between the day and night values both of the light absorption coefficient and the mass

1266

H. HORVATHet al.

concentration. The concentration of diesel particles is fairly high considering the small number of vehicles driving in this area; most of the diesel particles probably originate from the compressor.

Station 4 (high-capacity highway) This was located at 37 m from a highway with heavy traffic. Due to the relatively large distance and the sampling inlet at 2.5 m above street level, resuspended coarse-mode particles obviously were not sampled, therefore the mass concentration was lower than at station 1, whereas the concentration of diesel particles was second highest.

Station 5 (street with intermediate traffic density near bus garage) The mass concentration and the diesel particle content are in the intermediate, range. An expected influence of the bus garage could not be found. This is probably due to the specially developed engines used by most of the city buses: they operate on a dual fuel system, using liquified gas (major component) and diesel fuel (minor component). By this means it was possible to reduce all emissions of the buses, and exhaust is invisible, even during acceleration on steep grades.

Station 6 (residential, low traffic at the northwest slope of the surrounding mountains) Due to noise pollution it was not possible to operate the equipment during the night-time. Since no local sources for particulates existed, the night values are expected not to be much different from the day values. Mass concentration, diesel particles and light absorption are very low; the coarse mode has a value comparable with the data from the roof of the Physics Building (station 2).

Station 7 (upwind Vienna in the Vienna Woods) Although this station is upwind of Vienna for the prevailing winds from the northwest, several occasions with winds from the southeast occurred, which could bring pollution from the town. Therefore this station could not be considered as a true background but only as a station with very low influences from local sources. Due to the expected low Dy content of the filter samples, we always extracted two Nuclepore filters simultaneously, which tended to stick together and could not be separated during the extraction process. It may be that the values for the diesel particles given in the table are too low. The values for the mass concentration are slightly lower than for station 6, which also was fairly undisturbed from local sources but closer to the town. The light absorption coefficients at locations 6 and 7 are the same.

Station 8 (street with intermediate traffÉc density in the southern part of Vienna) Mass concentration and diesel particle concentration have intermediate values.

Station 9 (residential area in the south part of Vienna) Compared with the residential area of location 6, a much higher light absorption coefficient was found, with higher values during the daytime. Since no industrial operations of importance were located in the vicinity, only traffÉc can cause this. This is also demonstrated by the higher value for the diesel particles. The accumulation mode mass is comparable with that at the other locations; the coarse-mode mass has a low value, since the sampling site was at some distance from the street.

Station 10 (street in southern part of Vienna) The probable sources for the particles measured at this location are emissions from local sources, which could also be soil, since the area around the station is used for agriculture. The freeway at a distance of 135 m, but normally downwind, may to some extent also give contributions.

Station 11 (southern part of Vienna, adjacent to a truck yard) Here an unexpected high concentration of diesel particles was measured in comparison to the traffic density encountered in the vicinity. Therefore we suspect that the local emissions of the trucks in the immediate vicinity gave the major contribution.

Sedimentation probes Since the samples were collected on an irregular basis and occasional street cleaning took place, the scatter of the data is considerable. Nevertheless, one can see that the mass fraction of the diesel particles is between 1 and 7%, whereas the fraction of the diesel particles in the airborne state is between 12% (station 1) and 31% (station 6). This can easily be explained by the size of the diesel particles, where the major fraction of the mass is < 1 pm (Fig. 9), whereas the major portion of the sedimented particles comes from the coarse mode. At station 1, where the sampling inlet was at a height of 1 m, the fraction of diesel particles is the smallest, since a considerable portion of the particles sampled came from the coarse mode. The low concentration of diesel particles in the sedimented particles at station 6 (residential, west-northwest part of Vienna) is due to both the lower concentration of the diesel particles there and local conditions which cause an increased deposition of coarse particles on the road, since the shoulder was bare soil and turbulence behind cars produced enough air movement for some soil particles to become airborne.

General The data for the diesel particle concentration at the different measuring sites show two general trends: (1) with increased traffic density an increase in concentration can be seen, but (2) even at locations with a little traffic, a concentration which is much higher than the detection limit was found. This suggests that

Diesel emissions in Vienna there is a background of diesel particles all over the town, and at locations with high pollution from traffic, the diesel particles generated there cause the higher values. This can easily be understood, since most of the diesel particles emitted by the motor vehicles have sizes for which the residence time in the lower troposphere is approx. 10 days (Jaenicke, 1980) due to lack of efficient removal mechanisms. The particles thus move with air and can spread. This is clearly seen by comparing the diesel particle concentration at stations 1 (at street level) and 2 (27 m above the same street). The diesel particle concentration at 27 m is 83% of the street level value; the decrease could be caused by dilution and by the fraction of diesel particles which is larger than a few #m and thus already has a considerable settling velocity. In Fig. 11 the measured concentration of diesel particles as a function of traffic density in the near vicinity of the sampling inlet is drawn. The near vicinity is defined here as within a distance of 100 m. If several streets with considerable traffic were within this distance, the vehicles counted were added. One clearly can see increasing diesel particle concentrations with increasing traffic density. One can also see that four stations do not follow this trend, but this

(0)11

(O)

0~

3

~50610

(0)7 I

I

8

(0)2 1012

t i I I ~ 200 400 600 DIESEL VEHICLES PER HOUR

I

I 800

1267

could be expected for reasons given in detail already above: elevation of 27 m above street level (location 2), diesel compressor used during construction work (location 3), station outside of Vienna (location 7) and truck yard (location 11); these data points are given in parenthesis. For all the other stations a linear least square regression has been performed, giving a straight line. The correlation coefficient is 0.93, the slope of the line is 0.011 #gm-3/(vehicle h 1), the intercept is 11.0 # g m 3. The slope determined by the regression is different from 0 on a 99% level of significance. Thus diesel vehicles are the cause of the positive slope. This means that the general background (also due to diesel vehicles) is 11 /~gm -3 [almost identical to the value found at the residential, low-traffic area (location 6)-I, and, for example, 500 vehicles h-L cause an additional yield of diesel particles of 5 . 5 # g m -3. Strong local sources of diesel particles, as occurred at locations 3 and 11, can cause higher values. A value which frequently is given for diesel particles, is the mass absorption coefficient, i.e. the ratio of the light absorption coefficient and the mass concentration of the diesel aerosol. Obviously, this value strongly depends on the organic solvent fraction of the diesel particles. For GC particles a value of 5 m 2 g t has been found by Roessler (1984); we have found a value of 6.6 m 2 g - ~ for vacuum-dried diesel emissions. From the measured light absorption coefficients and the mass of the diesel particles, it is not possible to derive a mass absorption coefficient, and only an upper limit can be given. The values can be found in Table 5. With the exception of location 1, where we expect also a light-absorbing coarse-mode aerosol, the mass absorption coefficients are lower than the values for dry diesel aerosol. This is an indication that the particles emitted by diesel vehicles under urban traffic conditions also contain a considerable organic soluble fraction. (The organic solvent fraction of diesel emissions can vary between 10 and 90% depending on the operation of the engine.) The fraction of diesel particles in the total suspended particles has been determined by weighing the filters, which have been later analyzed for Dy. The average values for the measuring period are given in Table 6. For locations 3 and 11, where a local source

+---

Fig. 11. Mass concentrations of diesel particles measured at the sampling sites in dependence of the traffic density in the vicinity (up to 100 m). Data points where considerable influence from local sources were expected are in parenthesis. The solid line is a least-square regression.

Table 5. Ratio of light absorption coefficient of the atmospheric aerosol and the mass concentration of diesel particles at the different measuring sites Location Ratio (m2 g~ 1)

1 6.3

2 1.4

3 2.4

6 0.9

Table 6. Percentage of diesel particles in the mass of the suspended particles Location Fraction (%)

1 2 3 4 5 6 7 8 9 10 11 12.2 3 3 . 0 3 1 . 0 2 2 . 6 16.7 31.2 2 1 . 2 17.9 3 1 . 6 12.2 30.6

9 1.4

H. HORVATHet al.

1268

Table 7. Estimate of the mass of emitted particles in Vienna for the time period June-August Energy yr- 1 (TWh)

Fraction used (%)

Liquid fuels Automotive fuels

7.0 7.0

Electric energy (50% gas 50% oil) District heating Natural gas Total

5.7

7.5 17.0 8.0 15.0

1.7 7. I

2.5 7.5

Source

caused a large contribution of diesel particles, the fraction is above 30% as there is a source close by. F o r the other locations, the fraction of diesel particles varies between 12 and 32%. The highest values were found at locations 2, 6 and 9; in all these cases the sampling inlet was at least 30 m from a road, so that only a small coarse-mode fraction of the aerosol was sampled. This, again, is an indication of the fact that the diesel aerosol can spread easily. Location 1, although being in the street canyon with heavy traffic, has a low fraction, since the sampling inlet was 1 m above the street pavement. The average fraction of diesel particles of the other locations (2, and 4-10) is 23.3%. It is interesting to compare this fraction with an estimate of the emissions in Vienna. F r o m the energy consumed it was possible to estimate the mass of emitted particles in Vienna for. the three summer months June, July and August. Emission factors were taken from U.S. EPA (1973) with the exception for diesel emissions, where we have used the value given above. The masses of emitted particles are listed in Table 7. F r o m this, a fraction of 26% of the total emitted particles can be expected to be diesel particles, which is in the same range as the values found with the tracer technique. Since no other study of this kind is known to the authors, a direct comparison with results of other authors is not possible. Recently, an estimate of diesel particles has been reported by Rykowski and Brochu (1986). They used available data on ambient Pb concentrations, Pb content in the fuel, amount of diesel fuel and gasoline used and emission factors for the diesel and gasoline vehicle population to estimate the mass concentration of diesel particles in several European towns. Their values for the urban areas of the following cities were: Napoli (Naples) 7.2-35.1 /~g m - 3, Birmingham 3.2-20.3 pg m - 3 and Stockholm 2.5-16.4 # g m -3. Again, these values are in the same range as the values found in this study.

CONCLUSIONS With the use of Dy as tracer it was possible to determine the concentration of emitted diesel particles

Emission factor (g kWh - 1)

Mass of emitted particles (kg)

0.22 0.13 (gasoline) 0.26 (diesel) 0.14

115,500 154,700 145,600 120,897

0.22 0.014

9350 7455 553,502

in the air of Vienna directly. We found mass concentrations between 10 and 26 p g m -3 in the urban atmosphere of Vienna. F r o m the available data it was possible to conclude that diesel particles spread easily in the town. Other indirect estimates of diesel particles obtained similar results. Using data available in the literature for cancer risk factors for diesel emissions [values range between 0.007 and 0.3, a typical value being 0.1 annual lung cancers deaths per 100,000 people per # g m -3 of particles breathed over the lifetime has been reported by Cuddihy et al. (1984)] and the mass concentration of diesel particles obtained in this study, the following additional deaths per year per 100,000 people can be expected due to diesel emissions: when living away from highways 1; when living for a whole lifetime near a high-capacity highway 2.6. For comparison, the average annual lung cancer risk for non-smokers is 7 per 100,000, for smokers about 80 per 100,000. Acknowled#ements--This work was financed in part by the

Fonds zur FSrderung der wissenschaftlichen Forschung in Osterreich and Gemeinde Wien, MA 22. The co-operation of the Osterreichische MineralSlverwaltung, Automotorenversuchsanstalt List, Institut fiir Verbrennungskraftmaschinen der Technischen Universit~it Wien and the Atominstitut der Osterreichischen Hochschulen is greatly appreciated. We are very much indebted to Professor Kummer, who allowed us to use his garden as a sampling location.

REFERENCES

Barrie L. A., Hoff R. and Daggupaty S. M. (1981) The influence of mid-latitude pollution sources on haze in the Canadian arctic. Atmospheric Environment 15, 1407-1419. Barsic N. J. (1984) Variability of heavy-duty diesel engine emissions for transient and 13 mode steady-state test methods. SAE 840346. Berner A. (1984) Design principles of the AERAS low pressure impactor. In Aerosols: Science, Technology and Industrial Applications of Airborne Particles (edited by Liu B. Y. H., Pui D. Y. H. and Fissan H. J. F,), pp. 139-142. Elsevier, New York. Bowen H. J. M. (1966) Trace Elements in Biochemistry. Academic Press, New York. Cuddihy R. C., Griffith W. C. and McClellan R. O. (1984) Health risks from light-duty diesel vehicles. Environ. Sci. Technol. 18, 14-21, and Rep. LMF 89, Ref. 8 in this paper.

Diesel emissions in Vienna Davis B. L. (1984) X-Ray diffraction analysis and source appointment of the Denver aerosol. Atmospheric Environment 18, 2197-2208. Georgi B., Horvath H., Norek C. and Kreiner I. (1987) Use of rare earth tracers for the study of diesel emission in the atmosphere. Atmospheric Environment 21, 21-27. Hasan H. and Dzubay T. G. (1983) Apportioning light extinction coefficients to chemical species in atmospheric aerosols. Atmospheric Environment 17, 1573 1581. Jaenicke R. (1980) Atmospheric aerosols and global climate. J. Aerosol Sci. 11,577 588. Lewis C. W. and Dzubay T. G. (1986) Measurement of light absorption extinction in Denver. Aerosol Sci. Technol. 5, 325 336. Lin C. I., Baker M. and Charlson R. J. (1973) Absorption coefficient of the atmospheric aerosol: a method for measurement. Appl. Opt. 12, 1356-1363. Ohta S. and Okita T. (1984) Measurement of particulate carbon in urban and marine air in Japanese areas. Atmospheric Environment 18, 2429-2455. Perderson T. C. (1983) Biologically active nitro-PAH compounds in extracts of diesel exhausts particulate. In Mobile Source Emissions Including Polyeyclic Organic Species (edited by Rondia D., Cooke M. and Haroz R. K.), pp. 227-245. D. Reidel, Dordrecht. Pratsinis S. Novakov T., Ellis E. L. and Friedlander S. K. (1984) The carbon-containing component of the Los Angeles aerosol: source apportionment and contribution to the visibility budget. J. Air Pollut. Control Ass. 34, 643-648. Roessler D. M. (1984) Photoacoustic insights on diesel exhaust particles. Appl. Opt. 23. Rykowski R. A. and Brochu A. J. (1986) Diesel particulate emissions: the United States experience, with extrapolations to Europe and Japan. In Aerosols (edited by Lee S. D., Schneider I., Grant L. D. and Verkerk P. J.). Lewis, Chelsea, MI. Rondia D., Cooke M. and Haroz R. K. (editors) (1983)

1269

Mobile Source Emissions, Including Polycyclic Organic Species. D. Reidel, Dordrecht. Rosen H., Novakov T. and Bodhaine B. A. (1981) Soot in the arctic. Atmospheric Environment 15, 1371-1374. Schuetzle D., Paputa M., Hampton C. M., Marano R., Riley T., Prater T. J., Skewes L. and Salmeen L. (1983) The identification of potential sources of nitrated polynuclear aromatic hydrocarbons (nitro-PAH) in diesel particulate extracts. In Mobile Source Emissions Including Polycyclic Organic Species (edited by Rondia D., Cooke M. and Haroz R. K.), pp. 299-312. D. Reidel, Dordrecht. Shah J. J, Kneip T. J. and Dasey J. M. (1985) Source appointment of carbonaceous aerosol in New York City by multiple linear regression. J. Air Pollut. Control Ass. 35, 541-544. Shah J. J., Watson J. G., Cooper J. A. and Huntzicker J. J. (1984) Aerosol chemical composition and light scattering in Portland, Oregon; the role of carbon. Atmospheric Environment 18, 234-240. Stoeber W. (1986) Experimental induction of tumors in hamsters, mice and rats after long-term inhalation of filtered and unfiltered diesel engine exhaust. In Cancero9enic and Mutagenic Effects of Diesel Engine Exhaust. (edited by Ishinishi N. et al.). Elsevier, Amsterdam. U.S. EPA (1973) Compilation of emission factors. U.S. Environmental Protection Agency, Office of Air and Water Programs, Office of Air Quality Planning Standards, Research Triangle Park, NC. Vogg H. and Hartel R. (1976) Experience in the analysis of atmospheric samples at the Karlsruhe Nuclear Research Centre. In Modern Trends in Activation Analysis, Proc. Miinchen, Vol. 1, p. 569. Wolff G. T., Ferman M. A., Kelly N. A., Stroup D. P. and Ruthkosky M. S. (1982) The relationship between the chemical composition of fine particles and visibility in the Denver metropolitan area. J. Air Pollut. Control Ass. 32, 1216-1220.