Scholarly article on topic 'Exhaust particles of modern gasoline vehicles: A laboratory and an on-road study'

Exhaust particles of modern gasoline vehicles: A laboratory and an on-road study Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Atmospheric Environment
Keywords
{"Exhaust particles" / "Gasoline direct injection" / Nucleation / Soot / "Particle emissions"}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Panu Karjalainen, Liisa Pirjola, Juha Heikkilä, Tero Lähde, Theodoros Tzamkiozis, et al.

Abstract Vehicle technology development and upcoming particle emission limits have increased the need for detailed analyses of particle emissions of vehicles using gasoline direct injection (GDI) techniques. In this paper the particle emission characteristics of modern GDI passenger cars were studied in a laboratory and on the road, with the focus on exhaust particle number emissions, size distributions, volatility and morphology. Both during acceleration and steady conditions the number size distribution of nonvolatile exhaust particles consisted of two modes, one with mean particle size below 30 nm and the other with mean particle size approximately 70 nm. Results indicate that both of these particles modes consisted of soot but with different morphologies. Both in laboratory and on the road, significant emissions of exhaust particles were observed also during decelerations conducted by engine braking. These particles are most likely originating from lubricant oil ash components. The semivolatile nucleation particles were observed in the laboratory experiments at high engine load conditions. Thus, in general, the study indicates that a modern gasoline vehicle can emit four distinctive types of exhaust particles. The differences in particle characteristics and formation should be taken into account in the development of emission control strategies and technologies and, on the other hand, in the assessment of the impact of particle emissions on environment and human health.

Academic research paper on topic "Exhaust particles of modern gasoline vehicles: A laboratory and an on-road study"

Contents lists available at ScienceDirect

Atmospheric Environment

journal homepage: www.elsevier.com/locate/atmosenv

Exhaust particles of modern gasoline vehicles: A laboratory and an on-road study

Panu Karjalainen a, Liisa Pirjola b, Juha Heikkila a, Tero Lahde b, Theodoras Tzamkiozis c, Leonidas Ntziachristos c, Jorma Keskinen a, Topi Rönkkö a' *

a Aerosol Physics Laboratory, Department of Physics, Tampere University of Technology, Tampere, Finland b Department of Technology, Metropolia University of Applied Sciences, Helsinki, Finland

c Laboratory of Applied Thermodynamics, Department of Mechanical Engineering, Aristotle University of Thessaloniki, Thessaloniki, Greece

CrossMark

HIGHLIGHTS

GRAPHICAL ABSTRACT

> Four types of exhaust particles were observed in the exhaust of GDI vehicles.

Nonvolatile particle size distribution consisted of two modes. GDI vehicles emitted particles also during engine braking conditions. Semivolatile nucleation particles were in the exhaust at high load conditions.

Particle emissions were in real-world qualitatively similar as in the laboratory.

ARTICLE INFO

Article history: Received 4 March 2014 Received in revised form 11 August 2014 Accepted 13 August 2014 Available online 13 August 2014

Keywords:

Exhaust particles

Gasoline direct injection

Nucleation

Particle emissions

ABSTRACT

Vehicle technology development and upcoming particle emission limits have increased the need for detailed analyses of particle emissions of vehicles using gasoline direct injection (GDI) techniques. In this paper the particle emission characteristics of modern GDI passenger cars were studied in a laboratory and on the road, with the focus on exhaust particle number emissions, size distributions, volatility and morphology. Both during acceleration and steady conditions the number size distribution of nonvolatile exhaust particles consisted of two modes, one with mean particle size below 30 nm and the other with mean particle size approximately 70 nm. Results indicate that both of these particles modes consisted of soot but with different morphologies. Both in laboratory and on the road, significant emissions of exhaust particles were observed also during decelerations conducted by engine braking. These particles are most likely originating from lubricant oil ash components. The semivolatile nucleation particles were observed in the laboratory experiments at high engine load conditions. Thus, in general, the study indicates that a modern gasoline vehicle can emit four distinctive types of exhaust particles. The differences in particle characteristics and formation should be taken into account in the development of emission control strategies and technologies and, on the other hand, in the assessment of the impact of particle emissions on environment and human health.

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

* Corresponding author. Aerosol Physics Laboratory, Department of Physics, Tampere University of Technology, P.O. Box 692, FI-33101 Tampere, Finland. E-mail address: topi.ronkko@tut.fi (T. Ronkko).

http://dx.doi.org/10.1016/j.atmosenv.2014.08.025

1352-2310/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article

1. Introduction

In the development of gasoline passenger cars the increased attention on global warming and the greenhouse gas emissions has

under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

led to a widespread use of gasoline direct injection (GDI) engines (Alkidas, 2007). In general, GDI technologies offer better fuel economy compared to port fuel technologies and thus lower emissions of CO2. Also the use of alternative fuels is increasing, both in diesel and in gasoline vehicle fleets. New technologies most likely affect both the regulated emissions like NOx, hydrocarbons and particulate mass, and unregulated emissions like the amount or characteristics of emitted nanoparticles. The changes in the emissions other than CO2 can be either advantageous or harmful (see e.g. Heikkila et al. (2009) and Lahde et al. (2011)). In the case of the GDI technology, the direct fuel injection can increase the risk for increased particulate emission due to incomplete fuel volatilization, partially fuel rich zones, and impingement of fuel to piston and cylinder surfaces (Maricq et al., 1999b; Bonandrini et al., 2012; Sementa et al., 2012).

Particle emissions of vehicles are restricted by emission standards which have significant variation depending on the country. In the US, since 2004 same standards have been applied to vehicles regardless of the fuel and thus the limits for the particulate mass emission have covered also the gasoline vehicles. In the European Union, a particulate mass emission limit for direct injection gasoline engines took effect in 2009 (Euro 5), and the first restrictions for particle number emissions will come into effect in 2014 (Euro 6). Thus, globally the particle emission limitations for gasoline vehicles are under strong development (Dieselnet, 2014). Especially the European particle number emission limit for GDI engines may enforce the vehicle industry to change their emission reduction technologies and methods.

The relative importance of particle emissions of gasoline vehicles has increased because of forthcoming particle emission regulations, and because port-fuel injection (PFI) has been widely replaced by GDI technologies. On the other hand, the significance of gasoline particle emissions is now higher because of low emission level of modern diesel passenger cars. The fraction of the GDI vehicle in vehicle fleet is forecasted to grow significantly during the next years (CARB, 2010). It is known that the GDI technologies offer lower fuel consumption and NOx emission (Alkidas, 2007). However, the knowledge related to gasoline vehicle exhaust particles is not at the same level as the knowledge of diesel exhaust particles. The disadvantage of GDI technologies is an increase in particle number emission compared to PFI technology (Aakko and Nylund, 2003; Mohr et al., 2006; Braisher et al., 2010). If compared to diesel exhaust particle number concentrations, the GDI exhaust number concentrations are typically significantly lower than the concentration of diesel engine exhaust particles without a diesel particu-late filter (DPF) but higher than concentrations with a DPF (Mathis et al., 2005). The study of Maricq et al. (2012), conducted for exhaust particles of a light-duty truck with a GDI engine, indicates that the particulate matter emission of a GDI engine is dominated by elemental carbon (EC) whereas organic carbon (OC) constitutes only a small fraction. Several studies (e.g. Maricq et al., 1999a; Harris and Maricq, 2001; Khalek et al., 2010) indicate that the GDI exhaust particles are (in number) mainly in particle sizes below 100 nm. In addition, the size distribution has been observed to be bi-modal (Barone et al., 2012; Sementa et al., 2012; Sgro et al., 2012; Maricq et al., 1999a). The mode of smaller particles (mean particle size between 10 and 20 nm) has previously been observed to consist of spherical amorphous carbon (Sgro et al., 2012; Barone et al., 2012) and to be partly charged indicating their formation at high temperatures (Sgro et al., 2012). Although the volatility characteristics of the smallest particles have not been inspected in all the previous studies, some studies (e.g. Mathis et al., 2005; Li et al., 2013) indicate that the GDI exhaust can contain semivolatile nucleation particles too. In contrast to small amorphous carbon particles (Sgro et al., 2012), and in general, nonvolatile core

particles formed during diesel combustion (e.g. Laahde et al., 2009), the entirely semivolatile nucleation mode is formed in the atmospheric dilution and cooling process of the diesel exhaust (Roankkoa et al., 2006; Laahde et al., 2009).

In this study the focus is on the physical characteristics and emissions of particles emitted by modern gasoline passenger cars. Results of particle number emission, size distribution, volatility and morphology are presented. Measurements were conducted not only on a chassis dynamometer in the laboratory but also on the road. On-road studies provide a real-world driving environment and make possible to gather information from real-world exhaust dilution and dispersion processes in the atmosphere. For instance, for diesel vehicles the exhaust nanoparticle concentrations have been reported to be affected by sampling and dilution parameters used in the laboratory study (e.g. Ronkko et al., 2006). Thus, to get comprehensive and real information on exhaust particles also real-world studies are required. Also, real-world studies produce the most relevant information from the viewpoint of human exposure on particle emission. It should be noted that in the future also the vehicle emission legislation may shift towards the real-world measurements, e.g. due to the requirements for portable emission measurement systems.

2. Experimental

2.1. Experimental procedure on chassis dynamometer

The test vehicle was a modern gasoline passenger car made in 2011 (vehicle 1). The GDI engine of the test vehicle (1.8 l displacement) was turbocharged and used fuel stratified injection below about 3000 rpm. In the stratified mode, the global average air to fuel ratio is stoichiometric but due to stratified operation there are local rich and lean zones in the combustion chamber. The exhaust aftertreatment was performed with a three-way catalytic converter (TWC). The engine ran with low sulfur (<10 mg/kg) 95-octane gasoline—ethanol blend fuel where ethanol concentration was below 10%. The lubricant oil was viscosity grade 5W-30 which contained phosphorus, sulfur, calcium and zinc, 900 mg/kg, 2780 mg/kg, 3200 mg/kg and 920 mg/kg, respectively.

Experimental routine consisted of test cycles and different engine load steady points controlled by the chassis roll resistance. Before the test series, the vehicle was warmed up during a New European Driving Cycle (NEDC). During this warm-up run the emissions were also measured. The NEDC test cycle was repeated in total eight times. Selected steady-state tests were driven at the wheel speed of 80 km/h in fifth gear and at wheel powers 5 kW, 10 kW and 20 kW, controlled by the chassis roll brake.

The exhaust gas sample was extracted from an exhaust transfer tube from a sampling point that located 2 m after the tailpipe end. Exhaust dilution was conducted using a partial exhaust flow dilution system (Ntziachristos et al., 2004) consisting of a porous tube diluter, a short aging chamber and a secondary diluter. The dilution system has been observed to mimic relatively well the real-world cooling and dilution processes, especially from the viewpoint of exhaust nanoparticle formation (Ronkko et al., 2006; Keskinen and Ronkko, 2010). The primary dilution ratios of the porous tube diluter and secondary diluter (Dekati Diluter) were approximately 12 and 4.5, respectively. Both the primary dilution ratio and the total dilution ratio (~50) were calculated based on the CO2 concentrations of the raw exhaust and diluted exhaust. After the secondary dilution the diluted exhaust sample was at room temperature of about 25 °C.

Particles were measured with an EEPS (Engine exhaust particle sizer, model 3090, TSI Inc.), an UCPC (Ultrafine condensation particle counter, TSI Inc. model 3025) and an ELPI (Electrical low

pressure impactor, Dekati Oy). The ELPI was used with a filter stage (Marjamoaki et al., 2002) and an additional impactor stage for nanoparticles (Yli-Ojanpera et al., 2010). In order to study particle morphology and elemental composition, the exhaust particles were collected from the diluted sample on holey-carbon grids (Agar Scientific) by a flow-through sampler where flow of 1 l/min was used. The collected particles were then analyzed by shape with a Transmission electron microscopy (TEM) and by elemental composition with an energy dispersive spectrometry (EDS). During half of the measurements (4 NEDCs), a thermodenuder (TD) with low nanoparticle losses was employed in order to estimate particle volatility characteristics with the UCPC and ELPI. In the TD, the continuous exhaust sample was first heated up to 265 ° C, and after that, conducted through the denuder part in order to decrease the concentration of volatilized compounds. The data was corrected for the nanoparticle losses in the TD which vary between 32% and 50% for mobility sizes 10 and 4.5 nm, respectively (Heikkila et al., 2009). More information of the instruments and experimental setup is reported in the Supplementary information.

2.2. Experimental procedure on road

The gasoline engine (1.8 l displacement) of the test vehicle in on-road experiments (vehicle 2) was also turbocharged. However, the engine was newer generation with differences in the fuel injection system compared to the engine used in the laboratory tests. The engine adopts a combination of GDI and port fuel injection (PFI) technologies, in order to reduce the soot emissions. Under low load conditions only the PFI is used. In stratified GDI operation, the fuel is injected during both intake and compression strokes. Also this engine operated under global stoichiometric combustion conditions. Due major differences in the fuel injection compared to vehicle 1, smaller particle emissions were expected. The fuel and lubricant used were from exactly the same batches as in the laboratory experiments.

On-road tests consisted of controlled acceleration/deceleration routines (Table 1) and constant speed chasing tests. Experiments were performed in Alastaro (Finland) in a low-traffic road section far away from highly populated areas. Thus, the rural background particle concentrations were low, 2000—3000 1/cm3. Before the test series, the vehicle was driven around 200 km at typical highway/freeway speeds of 80—120 km/h. The experiments were performed under sunny and dry conditions (temperature 23—24 °C, wind speed 1—2 m/s, RH 40—60%).

The test vehicle was chased with the "Sniffer" mobile laboratory van (Supplementary information; Pirjola et al., 2004; Ronkko et al., 2006). The inlet probe located 0.5 m from ground level above the front bumper of the van. The chasing distance between the car and van was kept at around 12 m, although during accelerations and decelerations the constant distance was difficult to maintain exactly. The particle number concentrations and size distributions

Table 1

Acceleration/deceleration test routines in the on-road experiments with the amount of repetitions.

Mode Speed range Gear Repetitions

Acceleration 20-50 km/h 1st 12

Deceleration 50-20 km/h 1st 12

Acceleration 30-70 km/h 2nd 16

Deceleration 70-30 km/h 2nd 14

Acceleration 30-70 km/h 3rd 15

Deceleration 70-30 km/h 3rd 12

Acceleration 30-90 km/h 2nd 5

Deceleration 90-30 km/h 2nd 4

of the exhaust plume were measured with similar instruments as in the laboratory study: a UCPC, a UCPC after the TD, an ELPI and an EEPS.

3. Results

3.1. Laboratory study of particle emissions

Time-resolved particle emissions were determined from measured exhaust particle concentrations by taking into account the simultaneous exhaust flow mass rate. Four NEDCs were measured without the TD and four NEDCs with the TD upstream of the ELPI and UCPC. The EEPS was always used without the TD. Averaged temporal particle emissions (particles per second, 1/s) with standard deviations are shown in Fig. 1 for the whole NEDC. Results for the UCPC (a), the ELPI (b) and the EEPS (c) are shown separately. The emissions for particles larger than 23 nm in diameter (d) and the particulate mass emissions (e) are shown too, both calculated from the size distribution data measured using the EEPS. Mass emission was calculated for particles in the size range of 5.6—560 nm assuming particle density of 1 g/cm3. Overall, instruments showed a very good repeatability between the NEDCs with an exception that the standard deviation was relatively high during the last 100 s of the NEDC. In general, particle emissions depended a lot on the driving condition; e.g. at the end of idle modes the concentration level was low, in the particle size range covered by the ELPI and the EEPS even near zero if the detection limits of the instruments are taken into account, while during accelerations the particle number emission increased to high values, at first accelerations even larger than 1012 1/s. It should be noted that the particle emissions increased again also under deceleration conditions, especially when the speed decreased from 120 km/h to 0 km/h at the end of the test cycle but also at decelerations starting from lower speeds. These emissions can be seen in particle sizes larger than 23 nm (Fig. 1d) as well as in smaller particles. The decelerations were conducted by engine braking, i.e. in conditions where the fuel was not injected into the cylinders.

In addition to time-resolved emissions, we determined the particle number emission factors (1/km) particulate mass (mg/km) for the Urban driving cycle (UDC) and Extra-urban driving cycle (EUDC) as well as for the whole NEDC. These can be seen in Table 2. In general, the total particle number emission factors were smaller during the UDC compared to the EUDC when the semivolatile particles were included in the numbers, i.e. the measurement was conducted without the TD. When looking at the nonvolatile particles (TD) the situation was reversed; the emissions were higher during the UDC than the EUDC. Thus, when the results are studied from the viewpoint of mileage, the measurement shows that the semivolatile particles are emitted mainly during the EUDC of the NEDC (higher speeds) whereas the UDC contributes more on the nonvolatile particle emissions. This can also be seen from Fig. 1; the TD treatment did not affect significantly the particles during the first 1100 s of the NEDC. Instead, the particle emissions measured by the ELPI and the UCPC were decreased strongly by the TD treatment when the vehicle speed was over 100 km/h. The similar effect was seen also in the following deceleration and idle. At highest speeds the particle emissions measured by the UCPC without the TD treatment were higher than the concentrations measured by the ELPI and the EEPS which indicates that significant fraction of the emitted particles was 2.5—6 nm in diameter. Also, the emission of these small and semivolatile particles depended on the driving history; this was seen e.g. when the particle emissions over the EUDC were studied from an NEDC to another. During the first EUDC the particle emission measured by the UCPC without the TD was 3.0 1013 1/km, in the following EUDCs 2.3 1013 1/km,

Fig. 1. Time series of particle emissions over the NEDC cycles for (a) particle number measured with the UCPC (w & w/o TD), (b) number with the ELPI (w & w/o TD), (c) number with the EEPS (only w/o TD), (d) number of particles over 23 nm calculated from the size distributions measured by the EEPS and (e) particulate mass, calculated also from the EEPS data. "Shadowed" area indicates the standard deviation.

Table 2

Total particle number emission factors (1/km) for UDC, EUDC and NEDC, measured using the UCPC, ELPI and the EEPS. Emission factors for particle number >23 nm in diameter (EEPS, >23 nm) and for particulate mass (EEPS, mass) were calculated from the EEPS data. TD indicates the measurement with the thermodenuder.

Instrument UDC EUDC NEDC

UCPC (1/km) 6.87E+12 1.55E+13 1.23E+13

UCPC (1/km), TD 7.33E+12 4.58E+12 5.60E+12

ELPI (1/km) 5.64E+12 7.36E+12 6.73E+12

ELPI (1/km), TD 6.84E+12 3.23E+12 4.56E+12

EEPS (1/km) 5.54E+12 1.04E+13 8.58E+12

EEPS, >23 nm (1/km) 3.92E+12 1.75E+12 2.55E+12

EEPS, mass (mg/km) 4.36E+02 1.29E+03 7.51E+02

5.0$ 1012 1/km and 4.5 1012 1/km. The emission of these small semivolatile particles was related to the situation of high exhaust gas temperature (above 700 °C in the catalyst). Because also the driving history affected the emissions, the formation of these particles seems to be caused by release of semivolatile compounds from the after treatment system, engine or exhaust line surfaces.

The emission of particles larger than 23 nm, obtained from the EEPS data, was higher during the UDC compared to the EUDC. The overall emission of these particles was about 2.5 x 1012 1/km. In perspective of the European legislation, the upcoming Euro 6 limit for GDI vehicle particle number emission will be at start 6 x 10121/ km and change after three years to 6 x 10111/km. So the emission level of the tested vehicle is somewhere between these limits.

However, one must keep in mind that the official measurement protocol differs from the measurement method here. Fig. 2 shows the raw exhaust particle number size distributions (a) and size-segregated emissions (b). The particle size distributions were bimodal during acceleration and steady speed conditions; the geometric mean diameter (GMD) of the smaller particle mode was ~10 nm while for the larger mode the GMD was ~70 nm, according to the EEPS data. During accelerations, i.e. when the engine was most heavily loaded, the concentrations of larger particles were typically higher than concentrations in the smaller particle mode. Instead, under deceleration conditions and following idle, the particles were generally smaller than 20 nm in diameter, and only one particle mode existed. The largest concentrations of sub-20 nm particles were observed during deceleration from 120 km/h to 0 km/h. Note here that the changes in the exhaust temperature and mass flow cause changes in the residence time of exhaust in the vehicles tailpipe system within a transient cycle (Ronkko et al., 2014). Thus, the small particles measured under idle conditions were most likely emitted from the engine during deceleration. Also, it should be noted that the real contribution of particle emissions during deceleration was smaller than indicated based on the concentration results shown in Fig. 2a. For instance, during the last deceleration from 120 km/h to 0 km/h the concentrations of sub-20 nm particles were very high (Fig. 2a), but when exhaust mass flow is taken into account and the results are looked at emission point of view (Fig. 2b), their role diminishes. Similar trend is seen in decelerations from lower speeds also. In order to confirm the observations from transient driving, we measured the size distributions also at steady state driving conditions (speed 80 km/h, wheel

powers 5 kW, 10 kW and 20 kW), with and without the TD. These can be seen in Fig. 3. We can conclude that the TD treatment (at 265 °C) did not cause any significant change in the size or number of the exhaust particles in the particle size range measured by the ELPI. However, there was a very significant decrease in the concentration measured by the CPC during the 20 kW driving mode when the sample was passed through the TD (not plotted). In that case the majority of particles vanished in the TD treatment indicating that particles below 6 nm were semivolatile.

3.2. Particle morphology, laboratory study

The TEM images of collected particles are shown in Fig. 4. Two clearly distinct particle types were observed for samples collected over the whole NEDC. Firstly, around 10—20% of collected particles were nearly spherical (Fig. 4a,b,d), often containing internal structure of lighter and darker areas. The size of those particles varied from 10 nm to even larger than 200 nm. The EDS analyses indicated that these particles were composed of at least oxygen, zinc, phosphorous and calcium where the metals are compounds of engine oil but not of fuel. Thus in general, the results are in line with our previous study (Ronkko et al., 2014), which indicated that under transient driving conditions or, more detailed, under engine braking conditions gasoline vehicles can emit particles consisting of the lubricant oil originating compounds. Related to that, Fig. 2 shows that the particles are emitted during acceleration and steady speed conditions (during combustion) but also during engine braking when the fuel is not injected into the combustion chamber.

Fig. 2. Mean particle size distributions measured by the EEPS during the NEDC (8 repetitions) cycle in terms of (a) dilution ratio corrected particle concentrations (dN/dlog Dp) and (b) particle emissions per second (dN/t). Black line shows the wheel speed profile.

-1—I I I Illl|-1—I I I Illl|

10 100 Da (nm)

1000 1

10 100 Da (nm)

1000 1

10 100 Da (nm)

Fig. 3. Effect of thermodenuder treatment (TD at 265 °C) on the exhaust particle size distributions measured by the ELPI at steady state driving modes: (a) 5 kW, (b) 10 kW and (c) 20 kW.

The second particle type was agglomerated soot consisting of elemental carbon but also oxygen, zinc, phosphorous and calcium. Several soot particles included spherical darker parts similar to individual "lube oil originated" particles mentioned above. It should be noted that also very small nearly spherical soot-like particles were observed (Fig. 4b), possibly giving explanation for the bi-modal size distributions during acceleration and steady state driving. However, the accumulated particles can agglomerate also on the grid which prevents the direct comparison of number of collected particles with particle size distributions.

3.3. Particle emissions in the on-road study

The on-road study of particle emissions of the vehicle 2 was conducted at circumstances with very low particle concentrations in ambient air. Measurements were made in order to see the effects of on-road driving conditions and exhaust dilution on particle emissions but also to verify what are the particle concentrations and characteristics in the real exhaust plume. Because of the differences in technology between the vehicles 1 and 2, the exact and quantitative comparison to the laboratory results is not meaningful. On the qualitative manner, results largely confirmed many findings of the laboratory study.

The on-road study consisted of steady speed (low load) tests and transient acceleration/deceleration routines. During constant speeds, the total particle concentrations in the exhaust plume were at the background concentration level (~2000—3000 1/cm3) indicating very low particle emissions. Instead, in the transient tests the particle emissions were significant both in accelerations and decelerations. This can be seen in Fig. 5 where the exhaust plume CO2 concentration, total particle number concentrations measured by a UCPC (TSI model 3776), nonvolatile particle number concentrations (downstream the TD) measured by a UCPC (another TSI model

-UCPC -UCPC, TD ......Vehicle speed -CO2

Fig. 5. Total particle number (UCPC), nonvolatile particle number (UCPC, TD) and CO2 concentrations in the exhaust plume of vehicle 2 during the acceleration and deceleration routine between 30 km/h and 70 km/h (2nd gear).

3776), and vehicle speed for the second gear test between 30 km/h and 70 km/h are shown.

The highest particle concentrations were observed during accelerations where the total number concentrations of the exhaust plume reached the level of 40,000—50,000 1/cm3, which is about 20-fold compared to the background aerosol. During accelerations, particle plume concentration followed well the trends of CO2 concentrations. Instead, small but distinguishable exhaust plume concentration peaks (particle concentration between 3000 and 5000 1/cm3) associated with the start of deceleration (engine braking) did not follow the concentration of CO2. Because CO2 can be used as a tracer for combustion originated particles, the result shows that those particles emitted during engine braking are not originated from combustion. To give a rough estimation for exhaust emission factors by estimating the dilution ratio from CO2 concentrations of diluted sample and raw exhaust during 70 km/h

constant speed and taking account the exhaust flow rates, the vehicle-out emission factors were during the deceleration and acceleration 3.1 x 1010 1/s and 7.7 x 1011 1/s, respectively. In any case, the on-road study showed that the emission of engine braking particles is a real-world phenomenon for gasoline vehicles affecting human exposure to particles and, in general, air quality. Comparison of exhaust plume particle concentrations at acceleration—deceleration routines are shown in Fig. 6. All the repetitions of experiments (Table 1) were analyzed. The concentrations presented here are the average maximum concentrations reached during acceleration/deceleration, measured by the UCPC. The error bars indicate the standard deviation of maximum concentrations. It can be seen that the increased engine torque and lower engine speed increased the particle number concentration in the exhaust plume; when accelerated from 30 km/h to 70 km/h the concentrations were larger in 3rd gear than in 2nd gear. Also, the particle emissions during engine braking were higher when the deceleration started at higher engine rpm compared to lower rpm situations. Finally, although the engine braking increases the total particle number emission of the studied vehicle, it should be noted that the total emission was dominated by particles emitted during acceleration conditions.

The UCPC measurement showed clearly that the particle concentrations increased during the decelerations but, however, the concentrations were too low and/or the size of emitted particles was too small for the size distribution measurements. Instead, during accelerations the size distribution measurement was possible. Results of maximum concentrations of accelerations from 30 to 90 km/h are shown in Fig. 7. Like in the laboratory study, the particle number size distribution of real-world size distribution was dominated by ultrafine particles. The mean size of the emitted particles was approximately 20 nm for the ELPI. In addition, the bi-modality of particle size distributions was observed at certain acceleration but, however, not as clearly as in the laboratory experiments (Figs. 2 and 3).

4. Summary and discussion

The study shows that the gasoline fueled vehicle exhaust contains particles with very different concentration, size, morphology and chemical composition. Driving conditions were observed to have very significant effect on these particles, not only on the concentrations but also on the characteristics of the particles. The real-world study conducted on the road shows that the observations are relevant also from the viewpoint of the real emissions of GDI vehicles. It should be noted that, overall, the GDI technologies

100000

S loooo

Gear 1 Gear 1 Gear 2 Gear 2 Gear 3 Gear 3 Gear 2 Gear 2

20-50 50-20 30-70 70-30 30-70 70-30 30-90 90-30

km/h km/h km/h km/h km/h km/h km/h km/h

Acc. Dec. Acc. Dec. Acc. Dec. Acc. Dec.

S 4.0E+04

2.0E+04

0.0E+00

100 Da (nm)

Fig. 6. Exhaust plume total particle concentrations (UCPC measurement) during accelerations and decelerations in the on-road experiment. Error bars indicate the standard deviation (only mean + stdev) of mean concentrations between several consecutive tests. Also, the background aerosol concentration is shown.

Fig. 7. Particle number size distributions (ELPI measurement) during the repetitions of acceleration tests from 30 km/h to 90 km/h. Background particle size distribution was measured few minutes after the test on the same test road.

are diverse, even more diverse than PFI or diesel, with multiple degrees of freedom in choosing the operation parameters. Differences in technologies and operation parameter choices and as well as in drivers' behaviors can lead to different emission profiles than reported here which should be taken into account in the further use of the results.

In the laboratory part of the study, particle size distribution during the combustion of fuel (accelerations, steady state) consisted of two distinctive nonvolatile particle modes with the mean mobility diameters of ~10 nm and ~70 nm, respectively. In the on-road measurements this bi-modality was not as clear. In the laboratory, bi-modal size distributions were also measured under steady speed conditions for vehicle 1 while they were absent under steady speed conditions in the on-road experiments of vehicle 2. We suggest that these particles were soot similarly as diesel soot but with a difference that the soot was now divided into two distinctive modes. Also our results related to particle morphology, gathered from TEM analyses, support that. Note that the number-weighted particle concentrations of the modes were in this study at similar level which usually is not the case in diesel exhaust between the nucleation and soot mode particles. In diesel exhaust the nucleation mode tends to dominate the particle number concentrations when observed (e.g. Heikkilao et al., 2009). Also, in diesel exhaust the differences between the mean diameters of particles in different nonvolatile modes have been reported to be larger (e.g. Lahde et al., 2011). Also Xe et al. (2012) and Barone et al. (2012) have observed that gasoline vehicle exhaust particle size distributions can be bi-modal. Barone et al. (2012) have reported that the smaller particles may consist solely of carbonaceous primary spheres whereas the larger particles are agglomerates.

Based on the NEDC tests, the particle emissions during deceleration conducted by engine braking are repeatable and systematic. In addition, results indicate that these particles consist of both nonvolatile and semivolatile compounds (see Fig. 1). The on-road part of this study shows that the emission of these particles is a real-world phenomenon. Also, the oil-originated compounds found from the TEM samples and the on-road observation that the emissions of these particles do not follow the CO2 concentration of the exhaust, indicate that the particles are not originated from fuel combustion, but instead, from lubricant oil. The mechanisms how the lubricant oil compounds can end up to exhaust are discussed e.g. in Ronkko et al. (2014). However, it should be kept in mind that other sources cannot be surely ruled out and the particle composition can be affected also e.g. by compounds releasing from engine

or exhaust system (see e.g. Karjalainen et al., 2014). Compared to the particles emitted during acceleration, both the laboratory and on-road studies indicate that during deceleration the particles are smaller in diameter. These particles inevitably increase the particle number emissions during transient test cycles and in real-world driving but for the vehicles studied here, the total emissions were about an order of magnitude lower than the concentration of particles emitted during acceleration. The on-road study shows that for these emissions the particle emission factor calculation with respect to CO2 fails because of simultaneous zero CO2 emissions. This should be taken into account e.g. in the source appointment studies conducted in the vicinity of traffic. Also, the study indicates that the emissions of lubricant oil originated particles are strongly linked with certain driving conditions, possibly affecting their spatial existence in the urban environment.

During the high speed part of the NEDC and 20 kW steady loading, semivolatile particles were observed in the laboratory experiments. The number emission of these particles varied from an NEDC to another, indicating the strong role of driving history. Due to the volatility and small particle size, the chemical composition of these particles cannot be solved within this study. However, the results are relatively similar than previously reported for diesel exhaust where semivolatile particles, typically smaller than 20 nm, are found under high load conditions. In diesel exhaust, the formation of semivolatile nanoparticles is strongly linked to exhaust gaseous sulfuric acid concentration (Arnold et al., 2012; Ronkko et al., 2013), or at least, to fuel sulfur content (Karjalainen et al., 2014). The effect of driving history seems to be linked with the storage and release of compounds in the tailpipe system; components such as gaseous sulfuric acid, SO3 or heavy organics can be stored on engine, tailpipe or catalyst surfaces. These components may later get released when temperature is increased causing formation of new particles in the exhaust dilution and cooling process (Giechaskiel et al., 2007; Karjalainen et al., 2014). It should be noted here, according to authors' knowledge, that similar particle formation has not been reported for gasoline vehicles before. In this study, the semivolatile particle mode was not observed during the on-road tests where the maximum vehicle speed was 90 km/h and the engine loading was relatively low. If higher engine loading and thus higher exhaust temperatures are achieved in on-road testing e.g. driving on a freeway or uphill, also semivolatile nucleation mode formation is expected.

The upcoming legislation limit for particle number will enforce modifications for technologies related to GDI engines. Based on this study a combination of PFI and GDI technologies can be advantageous when the solid particle number is considered. The legislation will focus on solid particles larger than 23 nm in diameter. This study indicates that a major share of solid particles in the modern gasoline vehicle exhaust can be below this particle size limit, and during high engine load, vehicles can emit also small semivolatile particles.

Acknowledgments

Aleksi Malinen and Kaapo Lindholm from Metropolia together with Matti Happonen and Sampo Saari from TUT are acknowledged for their contribution to the experiments. The study was part of the TREAM project funded by Ecocat Oy, Neste Oil Oyj, AGCO Power, Oy Nanol Technologies Ab and Tekes (the Finnish Funding Agency for Innovation).

Appendix A. Supplementary information

Supplementary information related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2014.08.025.

References

Aakko, P., Nylund, N., 2003. Particle emissions at moderate and cold temperatures using different fuels. SAE Tech. Paper, 2003-01-3285.

Alkidas, A., 2007. Combustion advancements in gasoline engines. Energy Convers. Manag. 48, 2751—2761.

Arnold, F., Pirjola, L., Ronkko, T., Reischl, U., Schlager, H., LShde, T., Heikkila, J., Keskinen, J., 2012. First online measurements of sulfuric acid gas in modern heavy-duty diesel engine exhaust: implications for nanoparticle formation. Environ. Sci. Technol. 46, 11227—11234.

Barone, T., Storey, J., Youngquist, A., Szybist, J., 2012. An analysis of direct-injection spark-ignition (DISI) soot morphology. Atmos. Environ. 49, 268—274.

Bonandrini, G., Di Gioia, R., Papaleo, D., Venturoli, L., 2012. Numerical study on multiple injection strategies in DISI engines for particulate emission control. SAE Tech. Paper, 2012-01-0400.

Braisher, M., Stone, R., Price, P., 2010. Particle number emissions from a range of European vehicles. SAE Tech. Paper, 2010-01-0786.

California Air Resources Board (CARB), 2010. Preliminary Discussion Paper — Proposed Amendments to California's Low-emission Vehicle Regulations — Par-ticulate Matter Mass, Ultra Fine Solid Particle Number, and Black Carbon Emissions. Workshop report.

Dieselnet, www.dieselnet.com, information retrieved: 14.1.2014.

Giechaskiel, B., Ntziachristos, L., Samaras, Z., Casati, R., Scheer, V., Vogt, R., 2007. Effect of speed and speed-transition on the formation of nucleation mode particles from a light duty diesel vehicle. SAE Tech. Paper Series, 2007-01-1110.

Harris, S., Maricq, M., 2001. Signature size distributions for diesel and gasoline engine exhaust particulate matter. J. Aerosol Sci. 32 (6), 749—764.

HeikkilS, J., Ronkko, T., LShde, T., Lemmetty, M., Arffman, A., Virtanen, A., Keskinen, J., Pirjola, L., Rothe, D., 2009. Effect of open channel filter on particle emissions of modern diesel engine. J. Air Waste Manag. Assoc. 59 (10), 1148.

Karjalainen, P., Ronkko, T., Pirjola, L., Heikkila, J., Happonen, M., Arnold, F., Rothe, D., Bielaczyc, P., Keskinen, J., 2014. Role of storage and release of sulphur compounds in diesel exhaust nucleation mode particle formation. Environ. Sci. Technol. 48, 2336—2343.

Keskinen, J., Ronkko, T., 2010. Can real-world diesel exhaust particle size distribution be reproduced in the laboratory? A critical review. J. Air Waste Manag. Assoc. 60, 1245—1255.

Khalek, I., Bougher, T., Jetter, J., 2010. Particle emissions from a 2009 gasoline direct injection engine using different commercially available fuels. SAE Int. J. Fuels Lubr. 3 (2), 2010—01-2117.

Li, T., Chen, X., Yan, Z., 2013. Comparison of fine particles emissions of light-duty gasoline vehicles from chassis dynamometer tests and on-road measurements. Atmos. Environ. 68, 82—91.

Lahde, T., Ronkko, T., Virtanen, A., Schuck, T.J., Pirjola, L., Hameri, K., Kulmala, M., Arnold, F., Rothe, D., Keskinen, J., 2009. Heavy duty diesel engine exhaust aerosol particle and ion measurements. Environ. Sci. Technol. 43, 163—168.

Lahde, T., Ronkko, T., Happonen, M., Soderstrom, C., Virtanen, A., Solla, A., Kyto, M., Rothe, D., Keskinen, J., 2011. Effect of fuel injection pressure on a heavy-duty diesel engine nonvolatile particle emission. Environ. Sci. Technol. 45, 2504—2509.

Maricq, M., Podsiadlik, D., Brehob, D., Haghgooie, M., 1999a. Particulate emissions from a direct-injection spark-ignition (DISI) engine. SAE Tech. Paper Series, 1999-01-1530.

Maricq, M.M., Podsiadlik, D.H., Chase, R.E., 1999b. Examination of the size-resolved and transient nature of motor vehicle particle emissions. Environ. Sci. Technol. 33, 1618—1626.

Maricq, M., Szente, J., Jahr, K., 2012. The impact of ethanol fuel blends on PM emissions from a light-duty GDI vehicle. Aerosol Sci. Technol. 46 (5), 576—583.

Marjam3ki, M., Ntziachristos, L., Virtanen, A., Ristimoki, J., Keskinen, J., Moisio, M., Palonen, M., Lappi, M., 2002. Electrical filter stage for the ELPI. SAE Tech. Paper Series, 2002-01-0055.

Mathis, U., Mohr, M., Forss, A., 2005. Comprehensive particle characterization of modern gasoline and diesel passenger cars at low ambient temperatures. Atmos. Environ. 39 (1), 107—117.

Mohr, M., Forss, A., Lehmann, U., 2006. Particle emissions from diesel passenger cars equipped with a particle trap in comparison to other technologies. Environ. Sci. Technol. 40, 2375—2383.

Ntziachristos, L., Giechaskiel, B., Pistikopoulos, P., Samaras, Z., Mathis, U., Mohr, M., Ristim3ki, J., Keskinen, J., Mikkanen, P., Casati, R., Scheer, V., Vogt, R., 2004. Performance evaluation of a novel sampling and measurement system for exhaust particle characterization. SAE Tech. Paper Series, 2004-01-1439.

Pirjola, L., Parviainen, H., Hussein, T., Valli, A., H3meri, K., Aalto, P., Virtanen, A., Keskinen, J., Pakkanen, T.A., Mokem, T., Hillamo, R.E., 2004. "Sniffer" — a novel tool for chasing vehicles and measuring traffic pollutants. Atmos. Environ. 38, 3625—3635.

Ronkko, T., Virtanen, A., Vaaraslahti, K., Keskinen, J., Pirjola, L., Lappi, M., 2006. Effect of dilution conditions and driving parameters on nucleation mode particles in diesel exhaust: laboratory and on-road study. Atmos. Environ. 40, 2893—2901.

Ronkko, T., Lahde, T., Heikkila, J., Pirjola, L., Bauschke, U., Arnold, F., Schlager, H., Rothe, D., Yli-Ojanpero, J., Keskinen, J., 2013. Effects of gaseous sulphuric acid on diesel exhaust nanoparticle formation and characteristics. Environ. Sci. Technol. 47,11882—11889.

Roonkkoo, T., Pirjola, L., Ntziachristos, L., Heikkiloa, J., Karjalainen, P., Hillamo, R., Keskinen, J., 2014. Vehicle engines produce nanoparticles even when not fuelled. Environ. Sci. Technol. 48, 2043—2050.

Sementa, P., Vaglieco, B., Catapano, F., 2012. Thermodynamic and optical characterizations of a high performance GDI engine operating in homogeneous and stratified charge mixture conditions fueled with gasoline and bio-ethanol. Fuel 96, 204—219.

Sgro, L., Sementa, P., Vaglieco, B., Rusciano, G., D'Anna, A., Minutolo, P., 2012. Investigating the origin of nuclei particles in GDI engine exhausts. Combust. Flame 159, 1687—1692. Xe, H., Ratcliff, M., Zigler, B., 2012. Effects of gasoline direct injection engine operating parameters on particle number emissions. Energy Fuels 26, 2014—2027. Yli-Ojanpera, J., Kannosto, J., MarjamSki, M., Keskinen, J., 2010. Improving the nanoparticle resolution of the ELPI. Aerosol Air Qual. Res. 10 (4), 360—366.