ARTICLE IN PRESS

1352-2310/$ - se

doi:10.1016/j.at

�Correspondfax: +3023 10 9

E-mail addr

Atmospheric Environment 39 (2005) 4925–4936

www.elsevier.com/locate/atmosenv

Effects of a catalysed and an additized particle filteron the emissions of a diesel passenger car operating

on low sulphur fuels

Leonidas Ntziachristosa, Zissis Samarasa,�, Efthimios Zervasb, Pascal Dorlheneb

aLaboratory of Applied Thermodynamics, Mechanical Engineering Department, Aristotle University Thessaloniki,

P.O. Box 458, GR-54124 Thessaloniki, GreecebRenault, 1, Allee Cornuel, F-91510 Lardy, France

Received 12 January 2005; received in revised form 5 April 2005; accepted 30 April 2005

Abstract

This paper presents the emission characteristics of a diesel passenger car operated on low sulphur fuels (8 and

38 ppm) when fitted with either a catalysed diesel particle filter (DPF) or a non-catalysed one combined with a fuel-

borne catalyst. Measurements were conducted over the New European Driving Cycle and a higher speed driving cycle

to monitor the off-cycle DPF emission behaviour. Regulated gaseous pollutants and particle mass, number and surface

were recorded. Aerosol samples were collected with a dedicated sampling system, which provided identical dilution

conditions, regardless of the vehicle configuration, and allowed a distinction between volatile and non-volatile particles.

The results showed that DPFs have the potential of filtration efficiencies which may exceed 99.5% in all airborne

particle properties measured, over the transient cycles. As a result, the cycle average particle number was reduced from

1014 to about 1011 particles km�1 when fitting any DPF and the particle mass was reduced from �40mgkm�1 to the

detection limit of the current measurement procedure. The exact particle concentration depended on the filter material

properties. However, the efficiency in reducing mass appears lower than the airborne number, which suggests a

sampling artefact of the present particulate matter measurement procedure. A nucleation mode formed at high exhaust

gas temperature with the use of the higher sulphur fuel in combination with the catalysed DPF, thus decreasing the

apparent DPF filtration efficiency. This was removed when any of the contributing factors (high temperature, higher

sulphur fuel, catalysed DPF) were not present, suggesting sulphate particle formation downstream of the filter. Finally,

results show that the DPF soot loading has an insignificant effect on particle size distribution downstream of the filter,

when operating within soot-loading limits that are typically encountered in normal on-road operation.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Exhaust aerosol; Nucleation mode; Diesel aftertreatment; Filtration efficiency

e front matter r 2005 Elsevier Ltd. All rights reserve

mosenv.2005.04.040

ing author. Tel.: +3023 10 99 60 14;

9 60 19.

ess: zisis@auth.gr (Z. Samaras).

1. Introduction

Diesel passenger car registrations represented 44.3%

of the total European passenger car market in 2003 and

reached as much as 70% of total registrations in some

countries, according to the European Automotive

d.

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364926

Manufacturers Association (ACEA, 2004). The main

driving force for this trend is the higher fuel economy

compared to gasoline and the improved drivability at

low engine speeds. Additionally, ACEA is committed to

reducing the mean CO2 from new vehicles at 140 g km�1

by 2008 (25% reduction over 1990 levels) in an effort to

meet the European targets for the Kyoto Protocol. This

reduction will basically be accomplished via increasing

the market share of diesel cars.

In order to reduce the environmental impact from

vehicles, European legislation introduced the Euro 4

step in January 2005, which, for diesel cars, corresponds

to 0.25 g km�1 of NOx and 0.025 g km�1 of particulate

matter (PM). Despite an �80% reduction in diesel car

emission standards over levels 10 years ago, the mean

NOx emission level of diesel vehicles continues to be

three times higher than their gasoline counterparts.

Gasoline vehicles are also known to emit low levels of

soot particles (Andersson et al., 2002; Mohr et al., 2000).

Hence, future emission standards are already under

consideration to further tighten emission limits for diesel

vehicles.

Diesel particle filters (DPFs) are one of the most

technically feasible solutions to reduce PM. DPFs are

fitted in the exhaust line and collect PM by deep-bed

filtration. PM accumulated in the filter is then periodi-

cally combusted by oxidizing agents in the exhaust gas in

a process called regeneration. Commercial applications

either use some kind of a catalyst to decrease the soot

ignition temperature and increase the range of exhaust

gas temperatures where regeneration can occur or apply

an oxidation catalyst to increase NO2 concentration,

which is an efficient soot oxidation agent.

DPFs have been widely fitted to new vehicles and

retrofitted mainly to buses and large construction

equipment (e.g. Sequelong et al., 2004; Ball et al.,

2004; Mayer et al., 1999). Extensive experience has

already been accumulated with the operation and

emission performance of these systems on large diesel

applications (Lanni, 2003). Passenger car manufacturers

started to offer compact DPF systems on a voluntary

basis (Joubert and Sequelong, 2004) and some first

conclusions of their operation and filtration character-

istics have been collected (Jeuland et al., 2004).

However, there are still open issues with regard to the

performance of the systems over fully transient and real-

world driving conditions of light duty vehicles, because

very little data under such conditions are currently

available (Durbin et al., 2003; Mathis et al., 2005).

In the present paper we try to address these issues by

studying the emission performance of a current technol-

ogy diesel car fitted with two different DPF systems.

Measurements are conducted over the European cold-

start certification cycle (New European Driving Cycle—

NEDC) and an additional higher speed cycle to account

for distinctly different driving conditions. The focus is

on PM and airborne particle emissions which are

measured using a specially developed sampling protocol

and methodology to separate vehicle effects from

sampling effects. Additionally, two fuels with identical

properties but different sulphur content are tested to

explore the additional benefit of eliminating fuel

sulphur.

2. Experimental

2.1. Vehicle and fuels

A 2001 model year Renault Laguna 1.9 dCi was used

in this study, equipped with a common-rail direct

injection diesel engine and meeting Euro 3 emission

standards. PM emissions in this vehicle are controlled by

high pressure fuel injection (1350 bar max common rail

pressure) and two diesel oxidation catalysts (DOCs) in

series (pre-cat and main cat), which mainly decrease the

volatile content of exhaust PM. This combustion system

and aftertreatment configuration corresponds to one of

the most widespread diesel exhaust control configura-

tions for passenger cars in Europe today (the other one

being unit injectors with DOC). The vehicle was of low

mileage (�28,000 km) and it was regularly maintained

according to manufacturer specifications. The lubrica-

tion oil used was a 15W-40 grade (ACEA A3/B3) with

6000 ppm wt sulphur content. The vehicle was driven for

about 1000 km before the measurements. ‘‘Baseline’’

emission tests were performed with the vehicle in its

original configuration (no particle filter). It should be

stressed that the engine calibration was not changed

when a DPF was installed in the exhaust line.

Two fuels were used to address the effect of fuel

sulphur on particle emissions. These fulfilled the current

EN590 specifications (monaromatics 14%, polyaro-

matics 4.3%) for automotive diesel fuel and their exact

chemical character and physical properties are found

elsewhere (Ntziachristos et al., 2004b). The only

difference in fuel composition was their sulphur content.

The higher sulphur fuel (HSF—38ppm wt) was derived

from the lower sulphur one (LSF—8ppm wt) by doping

with sulphur compounds (thiophene and di-tertiarybu-

tyl-disulphide). These fuels were fed to the engine by an

external canister to avoid sulphur contamination of the

fuel transfer lines.

2.2. Measurement protocol

Each measurement day consisted of a series of

test cycles each using the same sampling protocol.

Measurements started with the certification NEDC

and a hot-start repetition of the urban part (UDC)

followed. Then, three cycles developed in the EU

Artemis project (Andre, 2004) were conducted to

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–4936 4927

simulate operation over urban, rural and highway

conditions. The introduction of these additional cycles

in the protocol is important to characterize the emission

performance of the vehicle and the aftertreatment

systems in off-cycle conditions. Then, three steady speed

tests at 50, 90 and 120 kmh�1 were performed to obtain

SMPS particle size distributions. The net sampling time

per day was about 2 h, with the vehicle covering a

distance of about 115 km. Each measurement day was

repeated twice. The vehicle was conditioned for about

100 km before testing whenever a fuel change was

required.

2.3. Diesel particle filters

The two DPF systems examined were based on

different principles to enhance soot oxidation. The first

system utilizes a Ce-based additive (CeDPF) to reduce

soot ignition temperature and to allow regeneration at a

wider exhaust gas temperature range. The concentration

of Ce in the fuel was adjusted at 25 ppm for these

measurements. The filter material was SiC and the

monolith external dimensions were +144mm�L

152.4mm (5.6600 � 600 with a cell density of 200 cells

per square inch (cpsi) and a wall thickness of 0.38mm.

The second system was a catalysed soot filter (CSF) of

the same material and external dimensions with the first

one. The presence of Pt-catalyst in the second filter

assists soot oxidation, therefore no external additive was

required in this configuration. This second filter had a

cell density of 300 cpsi and a wall thickness of 0.30mm.

Material porosity and mean pore size were also larger

than the CeDPF (before the washcoat application).

0

2

4

6

8

10

12

14

0 100 200 300 400

Tra

p so

ot lo

adin

g (g

)

CSF+LSF

120 120

Accumul

12

0 100

Fig. 1. Calculated mean trap soot loading as a function of accumu

triangles to steady state operation. ‘‘120’’ indicates the 120 kmh�1 t

corresponds to a forced regeneration of the CeDPF at high engine spe

to the eye only.

The DPFs were installed in place of the main

underfloor catalyst, which was removed for non-baseline

measurements. CSF measurements preceded the CeDPF

ones chronologically to reduce engine memory effects

by Ce contamination. In order to estimate the filter

soot loading, two pressure transducers were placed

upstream and downstream of the DPFs to record the

exhaust gas pressure drop and a thermocouple was

placed upstream to measure the exhaust gas tempera-

ture. The exhaust gas flowrate was calculated as the sum

of fuel consumption measured externally and the engine

intake air recorded by the vehicle’s hot wire anem-

ometer.

The filter soot-loading level is one parameter that

needs to be considered in the measurement protocol. The

soot loading was calculated according to an algorithm

which takes into account the measured signals and the

filter geometry and calculates the soot layer character-

istics (Haralampous et al., 2004). A low to moderate

soot-loading level is representative of the actual on-road

filter operation and necessary to maintain a low exhaust

gas backpressure and fuel penalty. In the tests of this

study, the CSF configuration was found to oxidize the

soot accumulated and to keep or even reduce its mean

soot-loading level, with no vehicle or engine modifica-

tion. However, the CeDPF was not able to self-

regenerate, despite the additive presence, and a forced

regeneration at high engine load and speed had to be

performed for filter cleaning. Fig. 1 shows the evolution

of the calculated mean filter soot loading as a function of

the vehicle accumulated distance during the measure-

ment campaign. A different regeneration behaviour

would have been obtained by changing the engine

ated mileage [km]

CSF+HSF

0 120

CeDPF+LSF

Forced

200 300 400 0 100 200 300

lated distance. Diamonds correspond to transient cycles and

est where self-regeneration takes place for the CSF. ‘‘Forced’’

ed and load to increase exhaust gas temperature. Lines are guide

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364928

calibration, but this was not done here in order to keep

engine-out emissions at the same levels with the baseline.

2.4. Sampling

The conditions and procedures for airborne particle

sampling are critical for the interpretation of the results.

The particle sampling setup utilized in this study was

presented in detail elsewhere (Ntziachristos et al., 2004b)

and has been widely applied to characterize emissions

from passenger cars (Ntziachristos et al., 2004a) and

heavy duty engines (Thompson et al., 2004). Its main

task is to precisely condition aerosol samples drawn

from raw exhaust before measurement with several

instruments, to enable the comparison of particle

emissions from different sources (engines/vehicles) under

the same sampling conditions.

A schematic of the sampling system is shown in Fig. 2.

The porous dilutor draws aerosol samples at a semi-

constant dilution ratio ðDR ¼ 12:5� 2:5 : 1Þ and dilu-

tion air temperature ðDAT ¼ 32 1CÞ and these are

allowed to stabilize in an aging chamber for 2.5 s (a

constant residence time). These conditions are consid-

ered to maximize the potential for nucleation mode

(NM) particle formation by increasing the saturation

ratio of semi-volatile components (Mathis et al., 2004).

The diluted sample is then split into three parts: One

Fig. 2. Particle sampling system schematic. Raw exhaust is sampled d

dilutor. Aerosol samples are then measured with different instrum

downstream of the thermodenuder (TD), a condensation particle co

charger (DC) and a gravimetric impactor (DGI).

part passes through a thermodenuder (TD) operating at

250 1C to separate non-volatile (solid) particles before

they are sampled with an electrical low pressure

impactor (ELPI). This temperature is effective in

removing NM particles, without any large effect on

accumulation mode particles (Matter et al., 1999).

Another (larger) part of the diluted sample is collected

in a gravimetric impactor (DGI) which provides the

mass-weighted size distribution. Finally, a third smaller

part of the sample is further diluted with cascaded

ejector dilutors to decrease the particle concentration

within the range of the TSI CPC 3010 and TSI SMPS

3936 used. Table 1 provides the sampling conditions for

each instrument in some detail. All results presented in

this study were corrected for particle losses in the

sampling system and instrument/device related ineffi-

ciencies. No DGI results are presented here though

because of the low particle mass collected on each

impactor stage when the DPFs were tested (below

0.1mg on a �180mg filter).

Gaseous pollutants and PM mass were measured

according to the standard constant volume sampling

(CVS) method with a dilution tunnel, following the

specifications of current European regulations valid for

Euro 3 compliant vehicles. Pallflex TX40 (Teflon-

coated) filters were used for PM collection. A separate

filter was used for each of the two NEDC parts (urban

irectly from the tailpipe and is precisely conditioned in a porous

ents, including the electrical low pressure impactor (ELPI)

unter, a scanning mobility particle sizer (SMPS), a diffusion

ARTICLE IN PRESS

Table 1

Particle property measured by each instrument and aerosol sample condition upstream of the instrument inlet

Instrument Sampling property Total dilution ratio Temperature (1C) Residence time (s)

DGI Mass-weighted size

distribution

12.572.5 Tamb to Tamb+5 1C 2.5

ELPI+TD Solid particle total number

(7 nm–1mm) and size

distribution

12.572.5 Tamb+�10 1C 3.5

DC Total active surface of

airborne particles (7 nm–1 mm)

16575 Tamb 3.0

SMPS 3936L Number-weighted size

distribution (7.6–290 nm)

16575 (w. trap) Tamb 3.0 (w. trap)

19.370.5� 103 (baseline) 3.5 (baseline)

CPC 3010 Total particle number

4�5 nm

Similar to SMPS Tamb Similar to SMPS

0.5

0.6

0.7

Baseline

CSF

(g k

m-1

)NEDC

Euro 3

Euro 3

(Euro 3)

L. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–4936 4929

and extra-urban) and the Artemis cycle. Only a primary

filter (no backup) was used to avoid sampling artefacts

when testing DPFs (Chase et al., 2004). All tests were

conducted at the facilities of the Laboratory of Applied

Thermodynamics, Mechanical Engineering Department,

Aristotle University of Thessaloniki.

0.0

0.1

0.2

0.3

0.4 CeDPF

Em

issi

on r

ate

8 ppm38 ppm 8 ppm 38 ppm 8 ppm 38 ppm

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Em

issi

on r

ate

(g k

m-1

)

Artemis

NOx HC×10 CO

8 ppm38 ppm 8 ppm38 ppm 8 ppm38 ppm

NOx HC×10 CO

Fig. 3. Gaseous pollutant emission rate with different fuels (8

and 38 ppm S) and vehicle configurations. Error bars corre-

spond to the min–max of two measurements conducted at

different days. ‘‘Euro 3’’ level corresponds to the emission

standard values.

3. Results

3.1. Gaseous pollutants

Fig. 3 shows the effect of the different DPF

configurations on the regulated gaseous pollutants from

this vehicle, for the two driving cycles considered. Fig.

3a also includes the vehicle emission standard level

(Euro 3) for each pollutant. CO is at least three times

below the emission standard (0.64 g km�1) and shows no

particular dependence on aftertreatment, despite the fact

that the main oxidation catalyst was removed and

replaced with the DPFs. In particular, over the hot-start

ARTEMIS cycle, CO is at the detection limit and the

only obvious deviation is when a non-catalysed DPF is

used. This increase in CO might either come from

limited soot combustion in the DPF or from the increase

of the tailpipe levels when the catalyst is removed.

No emission standard exists for HC, but only an

emission ‘‘margin’’, derived as the difference between

the NOx þHC (0.56 g km�1) and NOx (0.50 g km�1)

emission standards. In general, emission levels are 50%

below the margin for the cold-start NEDC and are

further reduced over the hot-start cycle. There is an

increase when DPFs are used, which may be more than

double the baseline condition. Variability also increases

though, but this is not unexpected since earlier studies

have shown that the HC measurement procedure for

diesel vehicles is associated with higher variability than

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364930

the measurement procedures for other pollutants

(Zervas et al., 2005; Mamakos et al., 2004). Increases

in total hydrocarbons most probably originate from the

removal of the oxidation catalyst, a fact specific to this

study only and could be remedied in the actual

application of such a system by properly dimensioning

the DOC. Similar to the CO, HC are not a priority

pollutant from diesel vehicles; therefore their emission

increase within such small ranges is not of particular

importance.

NOx is one of the key diesel pollutants. There are two

main and one secondary mechanism by which a DPF

may affect diesel engine-out NOx emissions. The

primary mechanism is by increasing the exhaust back-

pressure which in turn increases the engine load for the

same driving conditions. This may lead to an increase of

NOx emissions. At the same time, internal exhaust gas

recirculation (EGR) increases which has the opposite

effect on NOx emissions. Finally, NO2 may be consumed

in oxidizing PM in the filter but this is only a small

fraction of the total emissions. In general, engine

manufacturers are able to tune the engine-out NOx

emissions by adjusting the combustion parameters (fuel

timing, EGR, etc.), depending on the aftertreatment

system used. No adjustments were made to the engine of

the test vehicle in conjunction with retrofitting with the

DPFs here. NOx emissions over the lower power NEDC

were at 70% of the emission standard and showed no

appreciable effect with the DPFs. Up to a 25% increase

was observed over the power demanding Artemis cycle

for our vehicle with the DPFs. Since the engine was not

recalibrated for DPF use, we may suggest that none of

the DPFs tested have a significantly negative effect on

NOx emissions, at least to the extent that it would be

impossible to correct by engine recalibration. In addi-

tion, use of a DPF reduces the design stringency on

engine-out PM levels, thus a NOx favourable calibration

could be used for the commercial application of such a

system.

3.2. PM and airborne particle properties

The effect of DPFs on particle properties (total and

solid particle number and total active surface) and PM is

shown in Fig. 4. Based on these data, Table 2

summarizes the DPF efficiency in reducing PM and

the airborne particle properties measured. Starting with

PM mass, DPFs appear to be at least 95% effective in

removing particle mass from the exhaust. In particular,

over the NEDC, PM emission levels are reduced from

�40 to �1mgkm�1. However, since these very low PM

emission levels are generally associated with poor

measurement repeatability, the mass values quoted here

should be considered approximate (Zervas et al., 2005).

Reductions are even more impressive when looking

at airborne particle properties, with emission levels

decreasing with the DPF systems by three orders of

magnitude. Solid particle number decreases from slightly

above 1014 particle km�1 to below 1011 particle km�1. Total

particle number reduction is also in the same range, but

emissions downstream of the DPF are 20% (NEDC) to

twice higher (Artemis) than the 1011 particle km�1 level.

However, it needs to be considered in the comparisons

that, when a DPF is used, the repeatability of total particle

number measurement is in the range of 20–80% (expressed

as 1.96 times the relative standard deviation—see Zervas et

al., 2005). Active surface is found at �70 cm2km�1

downstream of the DPF compared to 12–13m2km�1 at

baseline configuration.

Focusing on fuel sulphur effects on baseline emis-

sions, the first observation is that its effect is negligible

over the NEDC for all particle properties. Use of 8 ppm

instead of 38 ppm sulphur fuel results in +11% PM and

solid particle number, �10% total particle number and

no difference in the active surface area. These differences

may well be considered within the experimental un-

certainty of the measurement. However, there is a

measurable effect over the Artemis cycle with total

particle number decreasing by 60% and total surface

area by 23% when reducing the fuel sulphur content.

The increase of total particle number at high speeds

when a non-zero sulphur fuel and a DOC are used has

been frequently observed in the past (Maricq et al., 2002;

Vogt et al., 2003). This has been attributed to sulphate

formation downstream of the DPFs as the exhaust gas

temperature increases and was actually confirmed here

with an additional vehicle of the same type.

The effect of sulphur on post-DPF emissions is similar

to the baseline measurements, with effects only found

over the high-speed Artemis cycle. When the CSF is in

place, the CPC/ELPI (total over solid particle number)

ratio is in the range of 1.32–1.40 for all fuel and cycle

configurations, except over the Artemis cycle using the

higher sulphur fuel, when it actually doubles to 2.64. At

the same time, active surface recorded with the DC is

three times higher over the Artemis than over the

NEDC, when using the HSF fuel. All these occur with

no appreciable difference in the solid particle number

(ELPI) NEDC/Artemis ratio which remains at 0.6–0.7

for the CSF, regardless of the fuel sulphur level. These

results show that the fuel sulphur content may have an

effect on semi-volatile particle nucleation when using a

DPF, even in the sub-50 ppm range. However, the effect

is negligible in the type-approval cycle and limited to the

high-speed cycle, despite the fact that favourable

conditions for particle nucleation were applied.

3.3. Detailed trap operation/efficiency

Fig. 4 may further be used to compare the efficiency of

the two systems. There is a repeatable and measurable

difference in concentration downstream of the two DPF

ARTICLE IN PRESS

0

10

20

30

40

50

60

70P

M (

mg

km-1

)

38 ppm

8ppm

NEDC Artemis

38 ppm

8ppm

Baseline CSF×10 CeDPF×10

0

10

20

30

40

50

60

70

DC

- A

ctiv

e S

urfa

ce (

m2

km-1

)

NEDC Artemis

38 ppm

8 ppm

38 ppm

8 ppm

Baseline CSF ×103 CeDPF ×103

0

5

10

15

20

25

30

35

40

45

50

ELP

I - S

olid

par

ticle

num

ber

(km

-1)

NEDC Artemis

38 ppm

8 ppm

8 ppm

38 ppm

Baseline ×1013 CSF ×1010 CeDPF ×1010

0

5

10

15

20

25

30

35

40

45

50C

PC

- T

otal

par

ticle

num

ber

(km

-1)

NEDC Artemis

38 ppm

8 ppm

38 ppm

8 ppm

Baseline ×1013 CSF ×1010 CeDPF ×1010

Fig. 4. Particle mass (PM) and different airborne particle properties for the fuel and vehicle configurations studied. Error bars

correspond to min–max of two measurements conducted at different days. Multipliers (shown in the legend) were applied to DPF

emissions in order to be visible in the linear y-axis.

L. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–4936 4931

systems, in particular with regard to solid particles. The

CeDPF configuration resulted in 40–80% lower levels

than the CSF, which is beyond the repeatability

uncertainty, especially over the Artemis cycle. Although

the difference is again not important at an absolute scale

compared to the baseline emission levels, it still shows

that filtration efficiency is not the same for all DPFs.

Furthermore, the high efficiency of CeDPF in solid

particles indicates that there is no detectable escape of

solid nanoparticles due to the fuel-borne catalyst.

The real-time recordings of particle number concen-

tration may be used to better explore efficiency as a

function of the upstream exhaust gas condition. Fig. 5

shows the particle number emission rate over the

Artemis cycle, as recorded with the CPC (total number)

and the ELPI (solids number) on a logarithmic scale.

CSF has a thinner wall, higher porosity and larger mean

pore size than the CeDPF, which may all lead to

somewhat reduced efficiency. This is particularly visible

in solid number and shows that the material character-

istics have a measurable effect on emission levels

downstream of the filter.

A parameter that may influence the emission level

downstream of a DPF is its soot loading, i.e. the

thickness of the soot layer formed within the trap.

This may be studied in more detail over steady state

tests. Fig. 6 shows particle size distributions down-

stream of the CSF for which measurements at different

ARTICLE IN PRESS

Table 2

Filtration efficiency (expressed in %) of DPFs, based on different particle properties

Measurement Driving cycle CSF CeDPF

38ppm S 8 ppm S 8 ppm S

PM NEDC 96.94 98.12 98.39

Artemis 97.44 95.30 98.59

Active surface (DC) NEDC 99.96 99.93 99.95

Artemis 99.71 99.88 99.95

Solids’ number (ELPI) NEDC 99.94 99.93 99.96

Artemis 99.93 99.91 99.98

Total number (CPC) NEDC 99.92 99.89 99.87

Artemis 99.95 99.88 99.94

Total number (SMPS) 50 kmh�1 99.94 99.95 99.87

90 kmh�1 99.94 99.96 99.92

120 kmh�1 35.67 99.96 99.94

L. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364932

soot-loading levels were taken. Particle concentration

follows typical lognormal distributions at 50 kmh�1,

which are noisy due to the very low levels. There is little

to differentiate between soot loading and fuel, despite

that soot levels range between 2 and 12 g in the DPF,

which is the extent of the range reached for typical on-

road application. Nevertheless, a more careful look

reveals that concentration increases with both soot

loading and fuel sulphur. This is only marginal,

however, compared to baseline emission levels. Results

are more interesting at 120 kmh�1. At this condition

(Texh�380 1C), this catalysed filter has started to

regenerate as shown by the decreasing soot-loading

levels. This is a rather controlled regeneration, in the

sense that soot is not consumed quickly as in a flame,

but it slowly oxidizes, also assisted by the catalyst. The

effect of this process on emissions is not visible when

LSF is used and the distributions remain lognormal at

more or less the same level and mean size with

50 kmh�1. A somewhat higher emission rate may be

solely due to the higher exhaust flowrate at this speed.

A large NM is only observed when HSF fuel is used.

This is probably not due to the regeneration, but it is

rather related to the sulphur content of the fuel, as this

has been recorded also without the DPF in place. The

inset in the lower left panel shows that the magnitude of

this NM reaches or exceeds conventional diesel levels,

leading to an apparent DPF efficiency as low as 36%

(see also Table 2). The present results show that even

with a fuel of very low sulphur content (38 ppm wt) NM

may be still formed downstream of a catalysed DPF

at high exhaust gas temperature. This should not be

considered as an outcome of low filtration efficiency, but

new particle formation downstream of the DPF.

However, with the use of a near-zero sulphur fuel

(LSF—8ppm wt) or a non-catalysed DPF (CeDPF), we

did not observe NM formation (Table 2—last row).

4. Discussion and conclusions

Use of two different DPFs on a diesel passenger car

revealed that they have the potential of very high

filtration efficiencies which may exceed 99.5% in all

airborne particle properties measured, using diesel fuels

with a sulphur content which is available in Europe as of

2005 (below 50 ppm). From this perspective, both a fuel-

borne and a catalysed filter appear to be viable

alternatives for passenger car applications. Despite

sampling conditions which favour the formation of

NM particles, i.e. low dilution ratio (12.5:1) and

moderate dilution air temperature (32 1C), there was

no extensive NM formation during transient tests with

DPFs. The result of this high efficiency was that a

conventional passenger car emitting �1014 solid

particles km�1 reached 1011 solid particles km�1 with

the use of two different technology filters. It is worth

stressing that total particle emissions from the vehicle

and fuels tested were found at similar levels with solid

particles, at least over the type approval cycle, for both

baseline and DPF configurations.

With regard to PM mass, use of the DPF systems

reduced exhaust emissions from �40 to �1mg km�1.

Although this is a large reduction on an absolute scale, it

corresponds to an efficiency in the range of 95–98.5%

(Table 2), which is lower than for airborne particles.

This could be attributed to the collection of semi-volatile

material on the filter, which is normally not air-

suspended in the diluted exhaust. Chase et al. (2004)

similarly observed that the organic portion of the PM

sample depended on the sampling filter material and

that there was significant mass gain depending on the

number of cascaded filters used. Therefore, they

concluded that some of the organic material on the

filter is just an ‘‘artefact’’ of the sampling procedure.

Our measurements confirm that this material is not

ARTICLE IN PRESS

1.0E+05

1.0E+06

1.0E+07

1.0E+08

1.0E+09

1.0E+10

1.0E+11

1.0E+12

0 200 400 600 800 10000

40

120

Artemis cycle time [s]

Sol

id p

artic

le e

mis

sion

rat

e [s

-1]

Spe

ed (

km h

-1)

1.0E+07

1.0E+08

1.0E+09

1.0E+10

1.0E+11

1.0E+12

1.0E+13

1.0E+14

0

100

200

300

400

Tot

al p

artic

le e

mis

sion

rat

e [s

-1]

Tem

pera

ture

(° C

)

1.0E+13

80

CPC

ELPI

CSF

Baseline

CeDPF

Temperature

CSF

Speed

CeDPF

Baseline

Fig. 5. Real-time recording of particle number emission rate, exhaust gas temperature and vehicle speed over the Artemis cycle with

different vehicle configurations using LSF (8 ppm S). The top panel corresponds to total particle number (CPC) and the bottom one to

solid particle number (ELPI).

L. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–4936 4933

measured by the airborne particle instrumentation,

hence absorption or adsorption of gaseous semi-volatile

species on the sampling filter is a plausible mechanism.

This would imply two things: firstly, that even in

airborne particle mass terms the actual DPF efficiency

is even higher and, secondly, that the regulated

procedure for PM measurement downstream of DPF

filters may become problematic due to interference

from semi-volatile condensation following the current

procedure (see also Liu et al., 2003). This indicates that

the current procedure must be improved for future

regulations.

A typical particle size distribution downstream of the

DPFs examined is lognormal at low speeds and may

shift to bi-modal with the formation of a discrete NM at

higher speeds. The NM reaches baseline emission

concentration levels and reduces the apparent filtration

efficiency of total particles down to 36%. However, NM

develops only when favourable conditions for sulphate

formation are established. These include exhaust gas

temperature in the order of 400 1C, use of the higher

sulphur fuel (38 ppm) and a catalysed filter. No NM was

measured when any of these conditions were not met.

The same dependence of NM on DPF and exhaust gas

condition has been also shown by Vaaraslahti et al.

(2004) for a heavy duty application with near-zero

sulphur fuels. Jeuland et al. (2004) also observed

formation of an NM downstream of a DPF system at

ARTICLE IN PRESS

0.0

0.5

1.0

1.5

2.0

2.5

0.0

0.5

1.0

1.5

2.0

2.5

0.01 0.1 1

Mobility diameter (µm)

50 km/h

HS F - H

LS F- H

LS F - L

HS F - L

Tot

al p

artic

le e

mis

sion

rate

×10

11 d

N/d

logD

p (k

m-1

)

0

2

4

6

8

10

12

14

20 120 220 320 420 520 620 720

Sampling Time (s)

Tra

p so

ot lo

adin

g (g

)

50 km/h

SMPS

scansHSF - H

LS F - L

HS F - L

0.01 0.1 1

Mobility diameter (µm)

120 km/h

Tot

al p

artic

le e

mis

sion

rate

×10

11 d

N/d

logD

p (k

m-1

)

HSF - H

LSF - L

HSF - L

LSF - H

0

2

4

6

8

10

12

14

20 120 220 320 420 520 620 720

Sampling Time (s)

Tra

p so

ot lo

adin

g (g

)

120 km/h

SMPS

sc ans

HS F - H

LS F - H

LS F - L

HS F - L

1015

1014

1013

1012

1011

1010

LSF - H

Fig. 6. Particle size distributions (left) obtained at different CSF soot-loading levels (right). Upper and lower panels correspond to

50 kmh�1 and 120kmh�1, respectively. H, L correspond to ‘‘higher’’ and ‘‘lower’’ loadings, respectively. Each distribution is an

average of two scans. The ‘‘SMPS scans’’ line on the right corresponds to the start time of the SMPS scans. The inset in the bottom left

panel shows the whole distributions on a logarithmic axis.

L. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364934

high speed. Our measurements showed that solid

particle number remained at the ELPI detection limit

over all steady state speed tests, which confirms that NM

particles downstream of the DPF are due to droplet

formation. In any case, these findings show that in order

to achieve ultra-high DPF efficiency under all operating

conditions, fuel of near-zero sulphur content should be

used.

So far, only a limited number of studies looked at the

effect of soot load on DPF emission performance (e.g.

Pattas et al., 1998; Dementhon and Martin, 1997). The

results of this study show that the DPF soot loading has

an insignificant effect on particle size distributions

downstream of the filter, when the DPF soot loading

is kept at levels typical of normal filter operation, i.e. not

associated with high backpressure built-up. In fact, the

size distributions shown in Fig. 6 demonstrate that there

may be an up to three-fold increase in particle number

concentration depending on DPF soot load, but this

remains within the experimental uncertainty of the low

concentrations downstream of the DPF.

Moreover, it needs to be emphasized that no escape of

nanoparticles was monitored post-DPF when a fuel-

borne catalyst was used, despite the fact that engine-out

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–4936 4935

nanoparticle formation caused by fuel additives was

reported in the past (Burtscher et al., 1999). Never-

theless, there is the need to specifically assess the

potential for Ce emissions in both well running

and poorly running vehicles with traps, in order to

evaluate the impact of Ce release to the environment

based on plausible emissions and hence to address

concerns with respect to secondary emissions of non-

regulated pollutants.

Acknowledgements

This work was conducted in the framework of the

‘‘Particulates’’ project (Contract no. 2000RD.11091),

funded by the European Commission Directorate

General Energy and Transport. The authors would like

to acknowledge Dr. Panayotis Pistikopoulos, Barouch

Giechaskiel, Athanasios Mamakos and Argyrios Tzilvelis

for support in the experimental and computational part

of this work.

References

ACEA, 2004. New passenger car registrations in Western

Europe. Available online at www.acea.be.

Andersson, J.D., Wedekind, B.G.A., Hall, D., Stradling, R.,

Wilson, G., 2002. DETR/SMMT/CONCAWE particulate

research program: light-duty results. SAE Technology

Paper 2001-01-3577.

Andre, M., 2004. The ARTEMIS European driving cycles for

measuring car pollutant emissions. Science of the Total

Environment 334–335, 73–84.

Ball, B., Hully, D., Lausseur, Ph., Lyons, N., 2004. Experience

of continuously regenerating particulate traps on city buses

in Europe. SAE Technology Paper 2004-01-0078.

Burtscher, H., Matter, U., Skillas, G., Zhiqiang, Q., 1999.

Particles in diesel exhaust caused by fuel additives. Journal

of Aerosol Science 29, S955–S956.

Chase, R.E., Duszkiewicz, G.J., Richert, J.F.O., Lewis, D.,

Maricq, M.M., Xu, N., 2004. PM measurement artifact:

organic vapor deposition on different filter media. SAE

Technology Paper 2004-01-0967.

Dementhon, J.-B., Martin, B., 1997. Influence of various diesel

traps on particulate size distribution. SAE Technology

Paper 972999.

Durbin, T.D., Zhu, X., Norbeck, J.M., 2003. The effects of

diesel particulate filters and a low-aromatic, low-sulphur

diesel fuel on emissions for medium-duty diesel trucks.

Atmospheric Environment 37, 2105–2116.

Haralampous, O.A., Kandylas, I.P., Koltsakis, G.C., Samaras,

Z.C., 2004. Diesel particulate filter pressure drop Part 2: on-

board calculation of soot loading. International Journal

Engine Research 5, 163–173.

Jeuland, N., Dementhon, J.B., Gagnepain, L., Plassat, G.,

Coroller, P, Momique, J.C., Belot, G., Dalili, D., 2004.

Performances and durability of DPF (diesel particulate

filter) tested on a fleet of Peugeot 607 taxis: final results.

SAE Technology Paper 2004-01-0073.

Joubert, E., Sequelong, Th., 2004. Diesel particulate filters

market introduction in Europe: review and status. 10th

annual DEER Conference, Coronado, CA, USA, Septem-

ber 2004.

Lanni, Th., 2003. Fine urban and precursor emissions control

for diesel urban transit buses. Environmental Pollution 123,

427–437.

Liu, Z.G., Skemp, M.D., Lincoln, J.C., 2003. Diesel particulate

filters: trends and implications of particle size distribution

measurement. SAE Technology Paper 2003-01-0046.

Mamakos, A., Ntziachristos, L., Samaras, Z., 2004. Compar-

ability of particle emission measurements between vehicle

testing laboratories: a long way to go. Measurement Science

and Technology 15, 1855–1866.

Maricq, M.M., Chase, R.E., Xu, N., Laing, P.M., 2002. The

effects of the catalytic converter and fuel sulphur level on

motor vehicle particulate matter emissions: light duty diesel

vehicles. Environmental Science Technology 36, 283–289.

Mathis, U., Ristimaki, J., Mohr, M., Keskinen, J., Ntziachris-

tos, L., Samaras, Z., Mikkanen, P., 2004. Sampling

conditions for the measurement of nucleation mode

particles in the exhaust of a Diesel vehicle. Aerosol Science

and Technology 38, 1149–1160.

Mathis, U., Mohr, M., Forss, A.-M., 2005. Comprehensive

particle characterization of modern gasoline and diesel

passenger cars at low ambient temperatures. Atmospheric

Environment 39, 107–117.

Matter, U., Siegmann, H., Burtscher, H., 1999. Dynamic field

measurements of submicron particles from diesel engines.

Environmental Science Technology 33, 1946–1952.

Mayer, A., Matter, U., Scheidegger, G., Czerwinski, J., Wyser,

M., Kieser, D., Weidhofer, J., 1999. Particulate traps for retro-

fitting construction site engines-VERT: final measurements

and implementation. SAE Technology Paper 1999-01-0116.

Mohr, M., Forss, A.-M., Steffen, D., 2000. Particulate

emissions of gasoline vehicles and influence of the sampling

procedure. SAE Technology Paper 2000-01-1137.

Ntziachristos, L., Samaras, Z., Mohr, M., Mathis, U.,

Keskinen, J., Ristimaki, J., Mikkanen, P., Vogt, R.,

2004a. Performance evaluation of a novel sampling and

measurement system for exhaust particle characterization.

SAE Technology Paper 2004-01-1439.

Ntziachristos, L., Mamakos, A., Samaras, Z., Mathis, U.,

Mohr, M., Thompson, N., Stradling, R., Forti, L., de

Serves, C., 2004b. Overview of the European ‘‘Particulates’’

project on the characterization of exhaust particulate

emissions from road vehicles: results for light-duty vehicles.

SAE Technology Paper 2004-01-1985.

Pattas, K., Kyriakis, N., Samaras, Z., Pistikopoulos, P.,

Ntziachristos, L., 1998. Effect of DPF on particulate size

distribution using an electrical low pressure impactor. SAE

Technology Paper 980544.

Sequelong, Th., Biancotto, D., Georges, H., Blanchard, G.P.,

Macaudiere, P.J, Grellier, J.-M., 2004. Retrofit program of a

Euro 1 and Euro 2 urban bus fleet in La Rochelle,

using the ceria-based fuel-borne catalyst for diesel particulate

filter regeneration. SAE Technology Paper 2004-01-0821.

Thompson, N., Ntziachristos, L., Samaras, Z., Aakko, P.,

Wass, U., Hausberger, S., Sams, Th., 2004. Overview of

ARTICLE IN PRESSL. Ntziachristos et al. / Atmospheric Environment 39 (2005) 4925–49364936

the European ‘‘Particulates’’ project on the characterization

of exhaust particulate emissions from road vehicles: results

for heavy-duty engines. SAE Technology Paper 2004-01-

1986.

Vaaraslahti, K., Virtanen, A., Ristimaki, J., Keskinen, J., 2004.

Nucleation mode formation in heavy-duty diesel exhaust

with and without a particulate filter. Environmental Science

Technology 38, 4884–4890.

Vogt, R., Scheer, V., Casati, R., Benter, T., 2003. On-road

measurement of particle emission in the exhaust plume of a

diesel passenger car. Environmental Science Technology 37,

4070–4076.

Zervas, E., Dorlhene, P., Forti, L., Perrin, C., Momique, J.C.,

Monier, R., Ing, H., Lopez, B., 2005. Exhaust gas particle

number measurement. Inter-laboratory study using ELPI.

Aerosol Science and Technology 39, 333–346.