the influence of a cerium additive on ultrafine diesel …heejung/publications/5.pdfthe influence...

e

rticlesperatureumber-tion in400 rpm

mheexper-

teor. Thesed to sometalyticl particu-ed. Twopectrom-siderably

um oxide

Combustion and Flame 142 (2005) 276–288www.elsevier.com/locate/combustflam

The influence of a cerium additive on ultrafine dieselparticle emissions and kinetics of oxidation

Heejung Jung, David B. Kittelson, Michael R. Zachariah∗

Departments of Mechanical Engineering and Chemistry, University of Minnesota, 111 Church Street SE,Minneapolis, MN 55455, USA

Received 9 March 2004; received in revised form 2 November 2004; accepted 2 November 2004

Available online 30 April 2005

Abstract

The influence of a cerium additive on the kinetics of oxidation and size distribution of ultrafine diesel pawas studied using a high-temperature oxidation-tandem differential mobility analysis method over the temrange 300–700◦C. The addition of cerium to the diesel fuel was observed to cause significant changes in nweighted size distributions, light-off temperature, and kinetics of oxidation. The peak number concentrathe accumulation mode decreased 50 and 65%, respectively, for 25 and 100 ppm dosing levels under 1and 75% engine load. The light-off temperature was reduced by 250 and 300◦C, respectively, for 25 and 100 ppdosing levels. The oxidation rate increased significantly (×20) with the addition of cerium to the fuel; however, trate was relatively insensitive to dosing level. The activation energy for cerium-dosed oxidation was, withinimental error, equivalent to that for undosed fuel (Ea = 100–110 kJ mol−1). From a phenomenological kinetic raperspective, the increase in oxidation rate was attributed solely to an increase in the preexponential factresults suggested that diesel particles using regular, undosed diesel fuels were already metal-catalyzeextent, most likely from metals in the lube oil. The addition of cerium likely increased the number of casites but had no effect on the overall activation energy due to the presence of other metals in the dieselate matter coming from lube oil. The characteristics of cerium-laden diesel particles were also investigatprincipal types of aggregates were found using transmission electron microscopy and energy-dispersive setry analysis. The first was composed mainly of agglomerates of carbonaceous spherules and a few, consmaller cerium oxide nanoparticles. The second consisted of metallic aggregates composed mainly of cerinanoparticles and some carbon. 2005 Published by Elsevier Inc. on behalf of The Combustion Institute.

Keywords:Cerium; Diesel; Additive; Oxidation; Kinetics; Emissions

w-ctiv-

is-ar-

theehi-eiv-st isernolo-

1. Introduction

The increasing use of diesel combustion for poering passenger cars has led to considerable a

* Corresponding author.E-mail address:[email protected](M.R. Zachariah).

0010-2180/$ – see front matter 2005 Published by Elsevier Indoi:10.1016/j.combustflame.2004.11.015

ity in methods for the reduction of particulate emsions. About one-third of the new passenger car mket in Western Europe is diesel-powered. Even ingasoline-dominant United States, modern diesel vcles with new emission reduction systems are recing increased interest. Part of this renewed interecentered on a reduction of emissions from moddiesels and new abatement strategies and techn

c. on behalf of The Combustion Institute.

http://www.elsevier.com/locate/combustflame

mailto:[email protected]

H. Jung et al. / Combustion and Flame 142 (2005) 276–288 277

for-

nsthatInedu-tialda-of

iednate

thetoheex-use

a-usedor

or-oi--ap-ortal

ph-ntlyma-

di-heingar-eM;de,

xi-and

there-

eith

of

e-thero-e insfer

s-nsicofnd

lateakeec-

dedel)ldherata-oreseluel.

-

uty,sTF-erakof

oadderea-

at

necle–)n-in

ilu-ekreas

gies, e.g., diesel particulate filter, fuel additivesparticulate matter reduction, and NOx adsorber catalysts for NOx reduction.

Diesel fuel additives, used to lower PM emissioand to enhance oxidation rates, are an approachshows promise in improving emission reductions.this article, we report on studies of a cerium-basfuel additive and the resulting particle size distribtion. We also report on the use of a tandem differenmobility analyzer to determine size-selected oxition rate and the role of cerium oxide on the ratesoot oxidation.

2. Background of soot abatement, oxidation, andthe use of additives

Diesel particulate filters (DPFs) have been studsince the 1980s as a means to reduce or to elimidiesel particulate matter (PM) emissions[1–3]. TheDPF, however, requires periodical regeneration offiltering material by oxidation of the filter depositprevent plugging. As the ignition temperature of tdiesel PM bed is considerably higher than thehaust gas temperatures of the diesel engine, theof catalytic combustion to lower the ignition temperture has attracted interest. Most research has focon either employing a filter coated with a catalystadding a catalyst additive directly to the fuel. The fmer method is limited by the problem of catalyst psoning from the exhaust gas[4] and poor contact between the PM and catalytic surface. The additiveproach is being studied in combination with DPFsbulk soot samples. In these works, a variety of meadditives have been tested. Miyamoto et al.[5,6] in-vestigated the effect of Ca, Ba, Fe, and Ni nathenates. They found that Ca and Ba most efficiereduced the soot, by both suppressing soot fortion and enhancing soot oxidation. Valentine et al.[7]studied the catalytic effect of bimetallic Pt/Ce adtives in an attempt to lower the dosing level of tmetal additive down to 4 ppm to reduce ash loadon the DPF and the emission of metallic ultrafine pticles. Skillas et al.[8] studied the effect of Ce and Fon the size distribution and composition of diesel Pthey observed a reduction in the accumulation mobut an increase in ultrafines. Lahaye et al.[9] studiedthe catalytic effect of Ce on simulated diesel PM odation and observed that the cerium was both onwithin the soot as cerium oxides. Kasper et al.[10]added ferrocene to the fuel and speculated thatenhancement of oxidation was more effective inducing soot than a suppression effect.

Many studies[11,12] have focused on the changin ignition temperature of the diesel PM bed wdifferent metal additives. While these results are

obvious practical benefit, it has been difficult to dcouple the role of the catalyst on the increase inoxidation rate from other effects that may also pmote an increase in oxidation rate and a decreasignition temperature (e.g., the heat and mass traneffect, particle size).

In view of the complex interrelationships dicussed above, we have attempted to extract intrioxidation kinetics in this article, to assess the rolethe catalyst in reducing the ignition temperature aenhancement of the oxidation rate.

3. Experimental

3.1. Cerium additive, fuel, and lubrication oil

The cerium additive used was a nanoparticucerium oxide dispersed in an organic solvent to mit directly miscible with diesel fuel (product namEolys DPX-9), and was provided by Rhodia Eletronics & Catalysis. The manufacturer recommenusing a dosing level of 25 ppm (of cerium in the fuby weight in combination with a DPF for a real-worapplication. For our studies, we also used a higdosing level of 100 ppm to assess the effect of clyst loading on the oxidation of diesel particles. Fthese studies, a standard EPA No. 2 on-road difuel (300–500 ppm sulfur) was used as a base fThe lubricating oil used in this study was SAE15W40 (John Deere TY6391).

3.2. Engine, sampling, and dilution system

The engine used in this study was a medium-ddirect injection, four-cylinder, four-cycle, mid-1990turbocharged diesel engine (John Deere T04045250). It had 4.5-L displacement, with a peak powoutput of 125 HP (93 kW) at 2400 rpm and a petorque output of 400 Nm at 1400 rpm. The outputthe engine was coupled to a dynamometer for lcontrol; the engine was operated at 1400 rpm un75% load (300 Nm). The exhaust temperature msured at the exhaust manifold was∼560◦C and themeasured air–fuel ratio was 27 for regular fuel1400 rpm under 75% load (300 Nm).

Fig. 1 is a schematic diagram of the engibench, sampling system, scanning mobility partisizer (SMPS)[13], and high-temperature oxidationtandem differential mobility analyzer (HTO-TDMAsystem[14]. The exhaust was sampled 25 cm dowstream of the turbocharger exit and was diluteda two-stage, air ejector, variable residence time dtion tunnel, similar to that described by Abdul-Khalet al. [15]. For the oxidation study, particles weextracted after the first stage of dilution, which w

278 H. Jung et al. / Combustion and Flame 142 (2005) 276–288

s.
Fig. 1. Schematic diagram of the experimental setup. The bold line indicates the path taken by the diesel particle
ilu-.byasere

atthe

resat-C)

lu-to

ec-rod-

agehedhe

ns

ta

rti-tsre-

tsareao-n-arertze–in

re-ichpo--2.o-

chda-om)

operated with a fixed residence time of 0.8 s at a dtion ratio of 26:1 to 28:1 for 75% load at 1400 rpm

The first-stage dilution ratio was determinedcomparing the NOx concentration in the exhaust gand that in the diluted exhaust gas; corrections wmade for the background NOx concentration. Thefirst-stage dilution air temperature was maintained32◦C by preheating and temperature stabilizingcompressed air to the ejector.

For all size distribution studies, samples wetaken after the second stage of dilution to preventuration of the condensation particle counter (CP(Model 3025A, TSI Inc.). The second-stage dition ratio was determined by flow measurementsbe 31:1. Thus, the overall dilution ratio after the sond stage was 810:1 to 870:1, calculated by the puct of dilution ratios from each stage.

3.3. Oxidation experiments and size distributionmeasurement

The sampled aerosol stream from the first stof the dilution tunnel was either introduced into tHTO-TDMA for the oxidation study and/or subjecteto second-stage dilution, prior to sampling by tSMPS as illustrated inFig. 1. The HTO-TDMA was

described in detail in our previous study by Higgiet al.[14].

Briefly, in the HTO-TDMA, the sample is senthrough a bipolar diffusion charger to establishknown charge distribution, then, a specific pacle size is selected with DMA-1. The DMA selecmono-area particles based on electrical mobility. Pvious studies[16,17], employing DMA size selectionwith TEM analysis, showed that the DMA selecfor mono-area aggregates, and that the surfacedetermined by a DMA was equivalent to their prjected area, for particles with mobility diameters uder about 200 nm. These mono-area particlesoxidized in air by passing through a heated quatube for specified residence time and known timtemperature profile. The particles are rechargeda bipolar diffusion charger and the size changesulting from the high-temperature processing (whincludes oxidation, thermal restructuring, and evaration processes) is measured with scanning DMAThree initial particle sizes of 41, 92, and 132 nm mbility diameters were selected at DMA-1 to matthe previous study on diesel nanoparticle oxition [14]. Furnace temperature settings ranged frroom temperature (25–34◦C, depending on the dayto 700◦C.


for

er-tedrst-us

fornethe

en-ula-arti-othnu-tile-sed

en-ablesultex-in aex-tionenht

ternce

or-e

m-ad.tri-

nif-er,ly

enginevem-ionas5%en-ly

theno

hein-

theedtla-, as

Fig. 2. Size distributions of diesel particles with base fuelvarious engine loads at 1400 rpm.

To check for particle size changes due to thmal effects, additional experiments were conducby using nitrogen as a carrier gas in both the fistage dilution and DMA sheath flows in our previostudy[14].

4. Results and discussion

4.1. Influence of cerium additive on the sizedistribution

Fig. 2 shows the size distributions (correcteddilution ratio) of diesel soot as a function of engiload at 1400 rpm using regular diesel fuel. Fordiesel engine used in these tests, particles at highgine load (75%) are mainly agglomerates (accumtion mode) over the whole size range, whereas pcles at light engine load (10%) are composed of bnuclei and accumulation mode particles. Theseclei mode particles are primarily composed of volaorganics (for sizes below∼30 nm), while the accumulation mode particles are agglomerates compoprimarily of solid carbonaceous materials[18]. Asthe accumulation mode increased with increasinggine load, more particle surface area became availfor vapor phase compounds to condense. As a reof the greater available surface area and higherhaust temperature, higher engine loads resultedlower saturation ratio during the gas phase at thehaust, and a lower rate of homogeneous nucleaor new particle formation. The latter point has beinvestigated in a study in which particles at ligengine loads were sent to a catalytic stripper[19].The catalytic stripper is a small catalytic converthat is maintained at constant temperature to enhaboth the evaporation and the oxidation of volatileganics[19]. Volatile organics diffuse to and oxidizon the catalyst-coated wall of the stripper. Ng[20]

(a)

(b)

Fig. 3. Size distributions of diesel particles with ceriudosed fuel for various dosing levels at 1400 rpm, 75% lo(a) Number distributions. (b) Geometric surface area disbutions.

showed that nuclei mode particles could be sigicantly reduced by employing the catalytic strippindicating that the nuclei mode was overwhelmingcomposed of volatile materials.

Fig. 3 shows the effect of the cerium additivon the particle size distribution for various dosilevels under fixed engine conditions (75% engload (300 Nm) at 1400 rpm). The cerium additiclearly had a significant effect on reducing the nuber concentration of particles in the accumulatmode. A 50% reduction in peak concentration wobserved at the 25 ppm dosing level, and a 6reduction at the 100 ppm level. The relative inssitivity of the results to dosing level was qualitativeconsistent with the work of Skillas et al.[8], whoobserved a significant reduction of particles inaccumulation mode at 20 ppm cerium, but saweffect with further increases in cerium. While taccumulation mode clearly decreased, a dramaticcrease in the nuclei mode was observed withaddition of cerium, which has also been confirmby Skillas et al. [8]. This behavior is consistenwith particle physics. A decrease in the accumution mode, and therefore available surface area


e-ion,

ofar-

weto a

didnu-

an-lasssrti-

ndth-ion.thess

taiumen-erre

re;ida-

turenm

nted

ate-tse tosed

Inas-

r-

le

shown inFig. 3b, reduces scavenging of particle prcursors, thus promoting homogeneous nucleatwhile at the same time decreasing coagulationnuclei mode particles with accumulation mode pticles.

To understand this increase in the nuclei mode,sent the undosed and 100 ppm-dosed particlescatalytic stripper, operating at 300◦C. Fig. 4 showsthat the nuclei mode particles were reduced, butnot disappear altogether; this suggests that theclei mode was composed of more than volatile orgics. This is consistent with the observation of Skilet al.[8], who, using inductively coupled plasma maspectrometry (ICP-MS) found that nuclei mode pacles are composed mostly of cerium.

(a)

(b)

Fig. 4. (a) Size distributions of diesel particles before aafter the catalytic stripper at 1400 rpm, 75% load wiout transport (diffusion and thermophoresis) loss correct(b) Size distributions of diesel particles before and aftercatalytic stripper at 1400 rpm, 75% load with transport locorrection. CS, catalytic stripper.

4.2. HTO-TDMA results

Fig. 5a shows representative HTO-TDMA dafor 90-nm particles generated under 25 ppm ceradded fuel, under fixed engine conditions (75%gine load (300 N m), 1400 rpm). Results for othinitial particle sizes at 25 and 100 ppm cerium asimilar to those inFig. 5a, but are not presented hehowever, the results were used to deduce the oxtion rate presented later in the article.Fig. 5a illus-trates that the particles shrink as furnace temperaincreases; the mobility diameter decreased by 1at 300◦C and by 7 nm at 650◦C. The total particlesize decrease, due to nonoxidative effects at 500◦C,measured using nitrogen as the carrier gas, amouto ∼1 nm at 75% load using regular diesel[14];this is consistent with the shrinkage we observed300◦C for cerium loading. The initial shrinkage blow 300◦C is inconsistent with the kinetic resulobserved at higher temperatures and is likely duthe evaporation of semivolatile materials condenon the diesel particles. Sakurai et al.[21] have alsoreported such shrinkage. Above 300◦C, most of thediameter change results from particle oxidation.addition to a decrease in particle size with incre

(a)

(b)

Fig. 5. TDMA diesel particle oxidation results in air for funace settings of 30–700◦C: (a) 92-nm initial particle sizewith 25 ppm cerium-dosed fuel. (b) 41-nm initial particsize with 100 ppm cerium-dosed fuel.


load at
Fig. 6. TEM images of 50-nm mobility diameter, cerium-laden diesel particles. Particles were sampled under 75%1400 rpm using a 100 ppm cerium doping level. (c) and (e) were taken at higher magnification.
easeen-ity,be-

pmereeased in.g.,

akiorereo-

e anar-

ltedister

npm,ouriOlnt

t is

of

r-

eenral,jor-

deco-

ing oven temperature, we also observed a decror loss of particles. This decrease in number conctration was not associated with chemical reactivbut, rather, thermophoretic transport losses, whichcome most important at higher temperatures[22]. InFig. 5b, we see results for 40-nm particles at 100 pcerium. At higher oven temperatures, two modes wobserved. In this case, rather than a steady decrto smaller sizes by the primary peak, as observeall our previous studies, here we saw a mode (eat 600 and 650◦C) near 40 nm, and another peat about 34 and 32 nm, respectively. This behavmay be attributed to the fact that some particles wcomposed primarily of cerium compounds that hmogeneously condensed from the vapor and havassociated carbon coating. The mixture of 40-nm pticles, predominately either cerium or carbon, resuin a bimodal size distribution after oxidation. Thisconsistent with our TEM observation discussed lain the article.

Figs. 6 and 7are TEM images of cerium-ladediesel particles under 75% engine load, 1400 rat 100 ppm cerium. The procedure detailed inprevious study was used to sample particles on S2-coated nickel TEM grids[16]. In brief, the aerosoflow after the first stage of the dilution tunnel is seto a DMA and the resulting monodisperse outpusent to a low-pressure impactor (LPI)[23]. Particlesare collected on the TEM grid at the bottom stagethe LPI.

In Fig. 6are shown 50-nm mobility diameter paticles. Please note thatFigs. 6a, 6b, and 6dwere takenat the same magnification, whereasFigs. 6c and 6earehigher magnifications ofFigs. 6b and 6d, respectively.The dark regions within the aggregates have bidentified as predominantly cerium oxide. In genewe observed two classes of particles. The vast maity, i.e., those greater than∼80%, look morphologi-cally like those shown inFigs. 6a–6c. These particleswere characterized by carbonaceous aggregates


load at
Fig. 7. TEM images of 130-nm mobility diameter, cerium-laden diesel particles. Particles were sampled under 75%1400 rpm using a 100 ppm cerium doping level. (b) and (c) were taken at higher magnification.
ar-hepre-r-thehavem-

r-ve

thereen

cles

onted

ile

be-

ti-erti-ges,

ontes).

forTheandhe

e as

rated on the surface with fine cerium oxide nanopticles, nominally less than 5 to 7 nm in diameter. Tother, much less prevalent class of particles is resented byFigs. 6d and 6e. These particles were chaacterized as having a larger fraction of cerium inaggregate. In addition, these particles seemed tomuch smaller primary particles (5–7 nm), as copared with the more prevalent class (∼20 nm), whilehaving a significantly larger fraction of cerium. Paticles composed mainly of metallic compounds haalso been observed in the nuclei mode when eilube oil (which contains metals) or ferrocene has badded to the diesel fuel[24].

We also observed these two classes of partiin our HTO-TDMA studies. Referring toFig. 5b,we observed a bimodal distribution after oxidatiat 700◦C. The first mode consisted of preselec

40-nm particles that remained after oxidation, whthe other mode reflected shrinkage to∼20 nm. This isconsistent with some particles exiting the engineing composed primarily of metal, as seen inFig. 6d.

TEM images of 130-nm mobility diameter parcles are shown inFigs. 7a–7c. These particles werfar less compact than 50-nm mobility diameter pacles. While the contrast is not as great in these imacompared with those inFig. 6, very fine cerium par-ticles, also in the range of 5 to 7 nm, can be seenclose inspection. The other class (metal aggregaof particles was not observed at this larger mobility

In Fig. 8 are two kinds of aggregates usedenergy-dispersive spectroscopy (EDS) analysis.one on the left is the carbonaceous aggregate,the other on the right is the metal aggregate. Telectron beam diameter is approximately the sam


way from
Fig. 8. TEM images of cerium-laden diesel particles for EDS. Background noise from the substrate was measured athe particles (spots c and e) for correction of the spectrum.
les.kedr

(C),x-d.aks,

Mforres-re-orcon-thatt d,Ce

Ce.a ofas-bleen.par-andas

um,t thedewasby

m,

d bye

gre-

sootus

.tedtal-gu-

edco-rtic-inery

ag-iontra-his

ing, be-ases

oms ofr to

, weure

ius

n-n

the circles shown inFig. 8. Background signals froma substrate were measured away from the particFig. 9shows the EDS spectrum for each spot marin Fig. 8. In all EDS spectra, without correction fobackground signal, there are peaks for carbonoxygen (O), silicon (Si), and nickel (O). This is epected, as SiO2-coated nickel TEM grids were useSpots a and d indicate very distinguishable Ce pewhereas other spots do not. Unfortunately, the Cepeak (the third peak from the left in the spectrumspot d) occurs at the same energy level (within theolution of our EDS system) as the Ni L peak. Thefore, the Ni L peak is what is most visible, except fspots a and d, where the large Ce L peaks (threesecutive peaks in the middle of spectrum) indicatethere should be appreciable Ce M lines visible. Spofor the metal aggregate, exhibits high counts onpeaks; this indicates that it is composed mainly ofThe background signal was removed for the spectrspots a, b, and d. Spot a, which was a soot particlesociated with Ce nanoparticle, shows distinguishapeaks for carbon, cerium, and low counts for oxygSpot b, which was a mainly carbonaceous sootticle, exhibits a higher carbon peak than spot aa low oxygen peak. Spot d, which was identifieda metallic aggregate, shows high peaks for cerioxygen, and carbon. These spectra suggest thaCe in the particles was in the form of cerium oxirather than pure cerium and that some carbonalso present. This is consistent with the prior studySkillas et al.[8], who found Ce present as CeO2 in theash of diesel PM (DPM) and proposed that ceriulike iron, reaches its highest oxidation state (Ce4+)very early, during the particle formation phase.

These metallic aggregates were also examineelectron diffraction. The diffraction pattern for thparticle shown inFig. 10a is shown inFig. 10b. Thewell-defined pattern suggests that these metal ag

gates were composed of crystallites, rather thanparticles, which are normally identified as amorphoand result in no pattern.Fig. 10c is a dark-field imageThe bright area is where the electron beam diffracto show a pattern. Because orientations of cryslites are not homogeneous, the pattern is not relar.

The prominent feature of DPM from cerium-dosfuels is that fine cerium oxide nanoparticles derate the exterior of soot agglomerates. A nanopaulate cerium oxide mixed with fuel should remaunchanged during diesel combustion due to its vhigh melting point (2200–2400◦C) (in other words,very low vapor pressure). There may be someglomeration of particles as well as a partial reductof Ce(IV). As dosing level increases, the concention of accumulation mode particles decreases. Tresults in more cerium oxide nanoparticles survivthe scavenging by carbonaceous particles. Finallycause the total mass of particulate matter decreon the addition of cerium, as evidenced inFig. 4,enhanced oxidation, which of course proceeds frthe exterior of the particles, results in aggregatesoot with the cerium decorating the surface, similawhat was observed in the TEM.

4.3. Determination of the oxidation rate

Using the measured decrease in particle sizeobtained kinetic parameters following the proceddetailed in our previous study[14,22]. The size de-crease rate is modeled using a modified Arrhenexpression,

(1)Dp = −AnmT 1/2 exp

[− Ea

RT

],

whereAnm is an initial size-dependent preexponetial factor andEa is a size-independent activatio


d to spots
Fig. 9. EDS spectra of cerium-laden diesel particles with and without background noise correction. Legends corresponmarked inFig. 8. Minus sign in the legend means background noise correction.
ces

od.flow

energy. Values ofAnm and Ea were determined byintegrating

(2)�Dp =X∫

0

Dp(x)

u(x)dx,

wherex is the horizontal position in the tube,X isthe length of the tube, andu is the flow velocity overthe heated length of the flow tube. The differenbetween calculated and measured�Dp values wereminimized using a nonlinear least-squares methThe dependence of the size decrease rate and


ron

e-

ma of

ubepmgu-

ined

ousetivitydi-

re-1-,

Fig. 10. (a) TEM image of a cerium aggregate. (b) Electbeam diffraction pattern. (c) Dark-field image.

velocity on horizontal position is a result of their dpendence on temperature as seen in Eq.(1) and

(3)u(x) = 4

3um

T (x)

T0,

whereum is the mean flow velocity calculated frothe volume flow rate and the cross-sectional arethe flow tube, and4

3um is the peak volumetric flowvelocity assuming laminar flow.

(a)

(b)

(c)

Fig. 11. Particle size change as a function of peak flow ttemperature for the three initial particle sizes: (a) 25 pcerium-dosed fuel, (b) 100 ppm cerium-dosed fuel, (c) relar base fuel[14].

Size decreases due to oxidation were determrelative to the size of the particles at 300◦C. This islower than the temperature we used in our previstudies, 500◦C [14,22]. It was necessary to lower thbase temperature because of the increased reacof the diesel particles on the addition of metal adtives.

4.4. Kinetic rates

Fig. 11shows experimentally determined sizeductions at various oxidation temperatures for 4


r

n.thisanyob-

withs-

a-er.

ityrmalivede-ethethate isionAllis

size

offm-

ra-bewnnds-

ob-anden-

d--

ishe25iva-nd

r-fuelasen

sis-

ionun-ionshattiva-er,

tote

b-GAda-

icsisred

uchddi-tionnifi-

cat-hishatrgythatas

tlts,

in-ted

,le

arti-us

ects

et ac-d on

erns,

es

onon-ted

92-, and 132-nm particles at 25 ppm (Fig. 11a)and 100 ppm (Fig. 11b) cerium doping levels. Focomparison purposes, we also show inFig. 11c re-sults from our prior work on undosed diesel oxidatioOne of the obvious differences seen betweenwork and the undosed results is the absence ofsize-dependent oxidation. In the undosed case, weserved an increase in the apparent oxidation rateincreasing initial particle size. That result was invetigated in our previous study[14] to correct the ratesfor the effective density of soot, which were mesured[25] using an aerosol particle mass analyzPark et al.[25] observed that the effective densincreases as particle size decreases and that noization to the effective density results in an effectoxidation rate that was essentially particle size inpendent[14]. At this point, we do not have a definitivexplanation of the differences observed betweendosed and undosed cases; however, it is possiblethe nature of the soot structure in the dosed casdifferent and it is also possible that the mass fractof cerium in the soot particles is size dependent.that can be concluded without further investigationthat the apparent oxidation rate for dosed diesel isinvariant.

For practical convenience, we define the light-temperature of oxidation as the peak flow tube teperature, where the particle shrinks∼1 nm (in addi-tion to the shrinkage arising from thermal evapotion). The light-off temperature was determined to∼840◦C for the undosed diesel particles as shoin Fig. 11c, but decreased significantly to 540 a590◦C, respectively, for the 25 and 100 ppm doing levels, as shown inFigs. 11a and 11b. The resultssuggest that the catalyst fulfilled a major, practicaljective of decreasing the thermal budget of a DPFthat, after a threshold concentration, there is little befit to operating at higher doping levels.

The data inFigs. 11a and 11bwere fitted to an Ar-rhenius expression (Eq.(1)) and are shown as solilines in the plot.Table 1 lists the Arrhenius parameters (frequency factors,Anm, and activation energy, Ea) obtained, andFig. 12 shows an Arrheniusplot of the surface-specific oxidation rates for thstudy in comparison with selected prior studies. Tactivation energies of diesel particle oxidation forand 100 ppm cerium dosing were essentially equlent, within experimental error, and equaled 107 a

Table 1

25 ppm 100 ppm

Ea (kJ mol−1) 107 102A40 (105 nm K−1/2 s−1) 6.6 5.0A90 (105 nm K−1/2 s−1) 7.3 7.0A130 (105 nm K−1/2 s−1) 8.6 5.8

-

102 kJ mol−1, respectively. More interesting, pehaps, is that the activation energy of the dosedwas not any different from that of the undosed c(108 kJ mol−1), even though the absolute oxidatiorate in the range of temperature studies was some∼20times faster! These results are qualitatively content with the study by Stanmore et al.[26]. Using thethermogravimetric method, they compared oxidatrates of cerium-dosed diesel particles with that ofdosed diesel particles under various engine condit(idle, medium speed, high speed). They found tthe presence of the catalyst did not change the action energy, consistent with our observation; howevtheir measured activation energy (210 kJ mol−1) wasa factor of 2 higher than our observation. Miyamoet al. [6] also found an increase in the oxidation ra(Fig. 12g) for Ca-catalyzed DPM. However, they oserved a two-stage oxidation process. In their Tstudy, they observed an initial stage of rapid oxition with very low activation energy (Fig. 12h), fol-lowed by slower and higher activation energy kinet(Fig. 12g) similar in magnitude to other studies. Thrapid early stage is similar to an observation referto asauto-accelerationby Lahaye et al.[9]. Whileour experiments are unable to resolve transients sas described above, we do note that the general ation of metals does not seem to change the activaenergy, but does increase the oxidation rate sigcantly.

This corroborated observation that the use of aalyst did not reduce the activation energy is, at tpoint, a mystery. We have previously observed tdiesel soot oxidation has a lower activation enethan flame-generated soot; it has been speculatedsmall quantities of metals from lube oil may servean inadvertent catalyst[14,22]. This would at leasprovide a consistent picture for the observed resualthough further work is still needed to establish if,deed, sufficient metals from lube oil are incorporainto soot. Stanmore et al.[26] observed the fall-offin reactivity above 600◦C for cerium-dosed particleswhich was not observed in our work. It is possibthat the so-called auto-acceleration may be anfact of the TGA measurements, which are notoriofor being corrupted by mass and heat transfer effas demonstrated by Mahadevan et al.[27]. From apractical point of view, the similarity of Ca and Csuggests that other metals may serve as oxidancelerators, leaving one to choose the dopant basecost, ease of use, and environmental/health concas well as oxidation rate.

Literature on the health effects of cerium additivis limited. A report[28] from the Health Effect In-stitute was the only extensive study we could findthe health effects of cerium additives. The report ccluded that the risk of inhaling cerium at the estima


nt priorphite rod.ith regular

l)

Fig. 12. Arrhenius plot of the surface-specific rates of diesel particle oxidation for the current study and for other relevastudies. Previous studies include oxidation rates of various soots (diffusion flame soot and carbon black) and pyrogra(a) Current study (25 ppm cerium-dosed case). (b) Current study (100 ppm cerium-dosed case). (c) Diesel particles wfuel [14]. (d) Diffusion flame-generated soot[22]. (e) Nagle and Strickland-Constable[29] using a pyrographite rod. (f) TGAof uncatalyzed diesel soot[6]. (g) TGA of catalyzed (Ca added to fuel) diesel soot[6]. (h) TGA of catalyzed (Ca added to fuediesel soot, rapid stage oxidation[6]. (i) Flow reactor study of Printex-U flame soot from Degussa AG[30]. (j) TGA of dieselparticle[31]. (k) Flow reactor study of diesel particles[32].

aab-eiringofto

easear-is

up-in

n-t ofofoc-

icsea-ofs in

er-

on-nd

os-leimin-

50m

r,ngsh-bym-or

lionre-atmeer-

lyea-

cles,adym

worst-case ambient level (1.2 µg/m3) arising fromheavy diesel traffic, using a cerium additive withparticulate trap, appears to be small. However, thesence of more complete information precluded thfully assessment of the possible health effects of uscerium as a fuel additive. They mentioned that riskchronic exposure was more difficult to estimate duethe lack of adequate studies and the possible incrin cerium emissions during regeneration of the pticulate trap. It should be further noted that no oneproposing the use of Ce or any metal as a soot spression additive on its own. It would only be usedcombination with a particle filter. Used in this maner Ce has a double benefit: it reduces the amounsoot collected in the filter, and thus the frequencyregeneration, and facilitates regeneration when itcurs.

5. Conclusion

The influence of a cerium additive on the kinetof ultrafine diesel particle oxidation has been msured using a HTO-TDMA method. The additioncerium was observed to cause significant changenumber-weighted size distributions, light-off tempature, and the kinetics of oxidation.

• Cerium addition decreased the peak number ccentrations in the accumulation modes by 50 a

65%, respectively, at the 25 and 100 ppm ding levels (1400 rpm, 75% engine load). Nucmode particles, most likely occurring as ceriuoxide, appeared as the cerium dosing ratecreased.

• The light-off temperature was reduced by 2and 300◦C, respectively, at 25 and 100 ppcerium.

• The oxidation rate increased significantly (×20)with addition of cerium to the fuel. Howevethe rate was relatively insensitive to dopilevel, suggesting that, beyond a certain threold level, no further benefit can be obtainedhigher doping. The activation energy for ceriudosed oxidation was within experimental errand equivalent to that of undosed fuel (Ea =100–110 kJ mol−1). From a phenomenologicakinetic rate perspective, the increase in oxidatrate is attributed solely to an increase in the pexponential factor. In a prior study, we found ththe oxidation rates of diesel particles and flasoot[14,22] are very different. In particular, thactivation energy for the oxidation of diesel paticles was 108 kJ mol−1, significantly lower thanthe value of 164 kJ mol−1 for flame soot; despitethis, diesel soot actually oxidized more slowthan flame soot over the temperature range msured. These results suggest that diesel partiusing regular undosed diesel fuels, are alremetal catalyzed to some extent, most likely fro


mb-butuem-

n--

Artd

S

AE

er

13

on,7–

-

l

E

r,

er,

C.

i-a-

s,per

13

ts,3)

-cleper

st.

ci.

e-ng36

rti--

S.n,

ts,03.i-

-us-m-

i-

m-cal

ah,

m,all,n,n,

-

on

e,

9)

metals in the lube oil. The addition of ceriulikely increases the number of catalytic sites (oserved as an increase in frequency factors)has no effect on the overall activation energy dto the presence of other metals in the DPM coing from lube oil.

Acknowledgments

Authors gratefully acknowledge Rhodia Electroics & Catalysis for supplying us with their ceriumbased fuel additive. The authors also thank Dr.Miller (NIOSH) for his review and comments, anDr. McKernan (University of Minnesota) for EDanalysis.

References

[1] J.J. Tutko, S.S. Lestz, J.W. Brockmeyer, J.E. Dore, STechnical Paper 840073 (1984).

[2] K.J. Baumgard, D.B. Kittelson, SAE Technical Pap850009 (1985).

[3] G. Lepperhoff, G. Kroon, SAE Technical Paper 8500(1985).

[4] J. van Doorn, J. Varloud, P. Meriaudeau, V. PerrichM. Chevrier, C. Gauthier, Appl. Catal. B 1 (1992) 11127.

[5] N. Miyamoto, H. Zhixin, A. Harada, H. Ogawa, T. Murayama, SAE Technical Paper 871612 (1987).

[6] N. Miyamoto, H. Zhixin, O. Hideyuki, SAE TechnicaPaper 881224 (1988).

[7] J.M. Valentine, J.D. Peter-Hoblyn, G.K. Acres, SATechnical Paper 2000-01-1934 (2000).

[8] G. Skillas, Z. Qian, U. Baltensperger, U. MatteH. Burtscher, Combust. Sci. 154 (2000) 259–273.

[9] J. Lahaye, S. Boehm, P.H. Chambrion, P. EhrburgCombust. Flame 104 (1996) 199–207.

[10] M. Kasper, K. Sattler, K. Siegmann, U. Matter, H.Siegmann, J. Aerosol Sci. 30 (1999) 217.

[11] K. Pattas, N. Kyriakis, Z. Samaras, T. Manikas, A. Mhailidis, W. Mustel, P. Rouveirolles, SAE Technical Pper 980543 (1998).

[12] N. Kyriakis, Z. Samaras, E. Vouitsis, T. ManikaT. Seguelong, G. Blanchard, SAE Technical Pa2002-01-0429 (2002).

[13] S.C. Wang, R.C. Flagan, Aerosol Sci. Technol.(1990) 230–240.

[14] K.J. Higgins, H. Jung, D.B. Kittelson, J.T. RoberM.R. Zachariah, Environ. Sci. Technol. 37 (2001949–1954.

[15] I. Abdul-Khalek, D.B. Kittelson, F. Brear, The Influence of Dilution Conditions on Diesel Exhaust PartiSize Distribution Measurements, SAE Technical Pa1999-01-1142 (1999).

[16] H. Jung, D.B. Kittelson, M.R. Zachariah, CombuFlame 136 (2003) 445–456.

[17] S.N. Rogak, R.C. Flagan, H.V. Nguyen, Aerosol STechnol. 18 (1993) 25.

[18] D.B. Kittelson, J. Aerosol Sci. 29 (1998) 575–588.[19] I. Abdul-Khalek, D.B. Kittelson, Real Time Measur

ment of Volatile and Solid Exhaust Particles Usia Catalytic Stripper, SAE Technical Paper 9502(1995).

[20] I.P. Ng, Aethalometer Performance with Diesel Pacles, in: Mechanical Engineering, University of Minnesota, Twin Cities, 2002.

[21] H. Sakurai, H.J. Tobias, K. Park, D. Zarling, K.Docherty, D.B. Kittelson, P.H. McMurry, P.J. ZiemanAtmos. Environ. 37 (2003) 1199–1210.

[22] K.J. Higgins, H. Jung, D.B. Kittelson, J.T. RoberM.R. Zachariah, J. Phys. Chem. A 106 (2002) 96–1

[23] S.V. Hering, S.K. Friedlander, J.J. Collins, L.W. Rchards, Environ. Sci. Technol. 13 (1979) 184–188.

[24] A. Miller, G. Ahlstrand, D.B. Kittelson, M.R. Zachariah, The Fate of Metal (Fe) During Diesel Combtion: Nanoparticle Formation, Morphology and Cheistry, 2005, in preparation.

[25] K. Park, C. Feng, D.B. Kittelson, P.H. McMurry, Envron. Sci. Technol. 37 (2003) 577–583.

[26] B. Stanmore, B.J.-F., P. Gilot, The Ignition and Cobustion of Cerium Doped Diesel Soot, SAE TechniPaper 1999-01-0115 (1999).

[27] R. Mahadevan, D. Lee, H. Sakurai, M.R. ZachariJ. Phys. Chem. A 106 (2002) 11,083–11,092.

[28] M.G. Costantini, R. Henderson, D.S. GreenbauH. Greim, E.D. Pellizzari, R. Sawyer, J. Warren, J. BS. Cadle, J.M. Davis, N. Englert, D.B. KittelsoA. Mayer, J. McAughey, R. McClellan, R.F. PhaleG. Simon, M. Waalkes, Health Effects Inst. Commun. 9 (2001).

[29] J. Nagle, R.F. Strickland-Constable, in: Fifth CarbConference, Pergamon, Oxford, 1962.

[30] J.P.A. Neeft, T.X. Nijhuis, E. Smakman, M. MakkeJ.A. Moulijin, Fuel 76 (1997) 1129.

[31] K. Otto, M.H. Sieg, M. Zinbo, L. Bartosiewicz, SAETechnical Paper 800336 (1980).

[32] A.F. Ahlström, C.U.I. Odenbrand, Carbon 3 (198475.

the influence of a cerium additive on ultrafine diesel …heejung/publications/5.pdfthe influence...

Documents