,.:$AE TECHNICAL.

PAPER SEJirlES

Per Stobbe

Henrik Guldberg PetersenStobbe Engrg. A/S

Jakob Weiland Høj and Spencer C. SorensonTechnical University of Denmark

1993 International Off-Highway &Powerplant Congress & Exposition

Milwaukee, WisconsinSectember 13-15. 1993

The Engineel-ing SocietyFo'r Adval1cirrg Mobility

J~ir (~nd Space@INTERNATJ'ONAL

400 Commonwealth Drive, Warrendale, P A 15096-0001 U.S.A. T el: (412)776-4841 Fax:( 412)776-5760

932495

Per Stobbe

Henrik Guldberg PetersenStobbe Engrg. A/S

Jakob Weiland Høj and Spencer C. SorensonTechnical University of Oenmark

ABSTRACT

Many of the materials which have been developed foruse as particle filters in the exhaust of diesel engines havecharacteristics which give rise to significant problems inpractical use. Due to its special characteristics, it is shownthai SiC is very well suited for use as the base material forparticulate filters. The physical and thermal properties ofporous SiC substrate material as applied to diesel particulatefilters have been determined and are presented. Experimen-tal results from several types of filter regeneration processesin exhaust gas systems confirm the improvements in the areaof thermal load and reduction in temperature level duringregeneration. The reduction in temperature during regener-ation is shown to be consistent with the high thermal con-ductivity of SiC.

INTRODUCTION

In recent years, concern over the adverse health effectsof particulate matter in diesel engine exhaust gasses has ledto increasingly stringent standards for exhaust emissions. Inaddition to engine modifications and improved engine con-trol, aftertreatment with particulate filters has been investi-gated for the reduction of particulate emissions from diesel

engines.There is a variety of filter types thaI have been pro-

posed and tested on engines and in vehicles, basically dif-fering in the type of material used and the physical structureof the filter. aften these two variables are interrelated.

ane of the most common fyres of filters to date hasbeen the so-called "wall flow filter", which is normallyconstructed of a honeycornb structure made of a porous,sintered, ceramic material. Although most filters could bethought of as a kind of wall flow filter, this designation

nonnally refers to the filter with the honeycomb structurewith plugged altemate ends (1)*. The most common versionof this type of filter is made of Cordierite, although recentlySilicon Carbide (SiC) has also been used as a filter material(2,3). This type of filter has the advantage of a high filtrationefficiency, which remains high as long as the total structuralintegrity of the filter in maintained. It is possible to coat thesurface of the filter with materials that are catalyticallyactive in order to reduce the ignition temperature of theaccumulated soot (4).

Another type of particulate filter that has been inves-tigated is the ceramic foam filter, in which the filteringelement is a hollow tube made of a porous ceramic rnaterial,Mullite in Reference (5) and SiC in Reference (6). This typeof construction suffers from a relatively low surface tovolume ratio, which leads to higher filter volume and theneed for more frequent regeneration.

A somewhat sirnilar concept is the use of stainlesssteel tubes with a series of holes in them, which are woundwith fine glass fibers to make a filtration element (7). Thistype of filter can have a high filtration efficiency, but sincethe filtration structure itself is not bonded, it is susceptibleto long tenn mechanical degradation due to vibration s andaltemating thennalloading. This type of filter is often usedwith some kind of fuel additive or ignition prornoter that isperiodicallY sprayed onto the filter to reduce the regener-ation temperature (8).

The use of a stainless steel wire mesh has also beeninvestigated as a filter material (9). This type of filter hasonly been used in connection with catalyticallY active coat-ings to combine the filtration process with catalysis in theso-called "Catalytic trap oxidizer". ane of the primarydisadvantages of the systems has been a low filtration effi-ciency compared to other systems.

* Numbers in parentheses indicate Reference at end ofpaper.

The low themial conductivity is also a problem inelectrical regeneration that is aften used for fleet and offroad vehicles. An e~ple is the "City Filter" concept,where a filter is regenerated arter typically a day's serviceby electrically heating the filter of a parked vehicle to atemperature adequate to ignite the particulate matter (10).The low thermal conductivity of the Cordierite makes itmore difficult to heat the filters.

SIC AS A FILTER MA TERIAL

Silicon carbide is a material that has the potential tosalve same of the problems that arise during the use of thewall flow filter. Table 1 presents a comparison ofthe proper-ties of pure polycrystalline 'SiC, Cordierite, and Mullite. Itcan be seen that for the case of pure, solid materials, the SiChas a decomposition temperature of 2300°C, while theCordierite melts at a temperature of 1350°C, and Mullite at1850°C. In practice with actual filters, these respectivetemperatures are more nearly 1600, 1200 and 1400°C. Thisis ane of the main advantages of SiC, since a material withsuch a high decomposition point should be able to withstandalmost any kind of regeneration process, intended or unin-tended, without being destroyed. Such a characteristicshould enable the use of a cheaper and less complicatedcontrol or security arrangement, since a pure thermal de-struction of an SiC filter is highly unlikely.

af all these systems, the ane that has probably shownthe greatest potential and amount of study is the wall flowfilter based on the honeycomb structure. This is due mainlyto its compact structure, high filtration efficiency, andgeneral mechanical stability. There are same problems thathave not been completely sol ved with this filter. These aremainly due to the type of material that has traditionally beenused. Cordierite is a low cost ceramic material, but has sameproperties that are undesirable in a diesel particulate filter.The first of these is a low mel ting point, which can beexceeded in practical engine conditions if a filter regener-ation occurs at an unintended time. This can result is a veryrapid destruction of the filter. A large portion of the devel-opment effort for Cordierite wall flow filters has been todevelop systems that are preventedfrom regenerating undersuch conditions.

An additional characteristic that contributes to thisproblem is the low thermal conductivity. This results inlocally high filter temperatures because the heat transferthrough the filter substrate is tao low to reduce the tempera-ture resulting from the combustion of the particulate matter.This is especially undesirable in connection with the rela-tively low melting temperature. Filter meltdown aften oc-curs in low flow conditions, where there is not sufficientflow to remove the energy released by combustion throughconvection.

Propert y for Pure M,aterials Silicorl Carbide Cordierite Mullite

Formula SiC 2MgO2A12035SiO2 3AlZO32SiOz

2.93.21 2.1Density, p

>380 195 200_~~xural_~!r:en~~, 0", MPa

Young's Modulus, E, GPa 410 132 152

0.20.15 0.25Poission's ratio, v

Specific Heat, J/kg_OC--- 670 1000

-63.6 -4.5 x 10 0.3 -1.0 x 10-6

850

4.2 x 10-6Thermal Expansion, aT, 11°C 25-500 °C

100 2.8Thermal conductivit~ ~ 4.5

0.25 2.5MOR

EaT(å1)0.3Thermal Shock Parameter, TSPI,

15000 8000 1300Therrnal Shock Parameter, TSP3, AMOR(I-V)EaT

Decomposition Point °C 2300 1350 (melt)

< 1200

1850 (me1t)

< 1500Safe Working Temperature unloaded, °C < 1600

Table 1 -A summary of different propertiesfor 3 types of dense, sintered polycrystalline ceramics. lt should be emphasized thatthese properties are for pure, dense substances, and not in the porous fonn encountered in filters. The three materials havemechanical and thermal properties which make them suitableforapplication to diesel particulatefilters. All three are commercialproducts, SiC from Ceradyne, lnc., Cordierite from Asahi Glass Company Ltd., and Mullite from McDanel Refractory company.

(Il)

Stobbe2

Fig. 1 -Filter substrate I1wnufactured out of Silicon Carbide. The diameter of the units are 170 mm and the wall thickness is1 mm. Length 130, 275 and 380 mm.

cated by the therrnal shock parameter, which is a combina-lian ofthe therrnal expansion coefficient and the mechanicalstrength of a material. If the therrnal conductivity is takeninto account (TSP3), iben the SiC has a higher therrnal shockparameter than Cordierite and Mullite. If it is not taken intoaccount (TSPI) iben it is lower than Cordierite and aboutthe same as MuUite. As will be shown later on in the paper,the therrnal conductivity of the SiC plays a significant folein its behavior during uge as a diesel particulate filter sub-strate. It is therefore felt thai TSP3 is more relevant. In thecomparison of these three materials, the Mullite is thepoorest choice of the three materials, since it has hightherrnal expansion, but a low strength.

From the point of view of the basic physical propertiesof SiC, iben it can be seen thai there are same significantpotential benefits associated with the uge of SiC as thesubstrate material for a wall flow type diesel particulatefilter. The work of the authors, and thai of others show thatit is possible and practical to manufacture a substrate withsuitable porosity and strength, to be used as a diesel filter(2,3). Reasonably conventional manufacturing techniquescan be used to construct these structures. These techniquesinclude an extrusion process, as well as high temperature

processing.

Another propert y tlllat is advantageous for the SiCfilter is a high thermal conductivity. From Table 1 it can beseen that the thermal conductivity of polycrystalline SiC isabout 30 times higher tharl that of Cordierite, and about 20times higher than that of Mullite. A consequence of this, aswill be shown later in the paper, is a reduction in thetemperature gradients, as well as the peak temperaturesencountered during the re!~eneration process.

Not all of the properties for SiC are improved withrespect to Cordierite. Tabll~ l shows that the thermal expan-sion coefficient for SiC is about five times larger than thatof Cordierite, and about the same as Mullite. For a giventemperature difference, th~~n, the thermal stresses in an SiCstructure would be the same amount greater than those williCordierite and about the same as those of Mullite. A miti-gating factor is found in the thermal conductivity, however,since this tends to reduce the local temperature gradients,which would lowerthe thelmal stresses. Forthe overall filterstructure at a uniform temlJerature, however, the SiC wouldhave a higher thermal stress than the Cordierite.

A compensating factor to this is the higher strength ofthe SiC, which is at least t'Nice as great as that of Mullite orCordierite. Since it is stronger than the Cordierite, it is thenable to withstand higher stresses. The compensation is indi-

Stobbe3

(foam filters, woven fiber fabric and pressed, sintered fi-brous substrate). This advantage is important when therelatively limited space on a vehicle is taken into account.Furthermore, for a given filter volume it is not necessary toregenerate a wall-flow-filter as frequently as is necessaryfor a depth filter. Another advantage of the WFF's is thatthe filtration efficiency generally is larger, typically 80-90%compared to 40-60% for a foam filter (6)- and the efficiencyis not sensitive to exhaust gas flow variations. A problemwith depth filters is "blow off', where the trapped soot isblown off the filter when the gas velocity through the filteris too high.

Generally the factors that increase the filtration effi-ciency such as reduced mean pore size, larger wall thicknessand reduced porosity will exhibit a negative influence on thepressure drop, thus giving higher pressure drop (or backpressure) leading to increased fuel consumption of the ve-hicle.

Theoretical Considerations -The permeability for afilter substrate is a measure for what the pressure drop willbe at a given gas flow rate. That is, the higher the perme-ability, the lower the pres sure drop at a fixed flow.

For a given substrate an increase in wall thickness, gasflow rate and/or in gas viscosity will increase the pressuredrop over the filter. These characteristics are seen for flowin the laminar region from Darcy's law:

æ=to1l0Ya (1)

In terms of cost, thl~ basic raw material used in thefilters is commercial grindling grain SiC, which is relativelycheap. The extrusion techniques used are no more expensivethan those involved willi titte production of substrates out ofCordierite or other ceramics. Since much highertempera-tures are involved willi the high temperature processing,however, processing cos1:s will be higher than for othermaterials. Extruder wear mayaiso be increased because ofthe abrasiveness of SiC. Since currently only prototypefilters are being produced, ane cannot sar with certaintywhat this additional cost will be in a practical production

system.Figure ane shows photographs of a substrate material

made out of SiC which h~LS been manufactured using thesetechniques. They are curre:ntly limited by the size limitationof the extruding equipment. The pie shape sections can beconveniently grouped to make larger structures. Reference(2) has shown that the use of smaller elements combined tomake a larger unit gives same structural advantages. Theauthors' experience to datl~ has not indicated cracking prob-lems willi the configuration shown in Figure ane.

Because ofthe good electrical properties SiC ceramicsare used as the basic material for making heating elements.Another common application of SiC is the so-called "kilnfumiture" which is a type of structural ceramic used tosupport e.g. porcelain or Cordierite during sintering. Here itis possible to take advantaJ~e ofthe very high strength at hightemperatures, the good the~rmal cycling properties and goodoxidation resistance (can be used up to 1600°C in air con-

tinuously).The last mentioned application of SiC as a structural

ceramic for high temperatures willi thermal cycling leadamong other things to tl1e conclusion, that it would besuitable as a substrate for diesel particulate filters.

where: ~ is the pressure drop, t is the filter thickness,11 is the dynamic gas viscosity, V is the linear gas flow rateand a is the permeability.

If the mean pore size is reduced, the permeability ofthe filter substrate will be reduced hence leading to higherpressure drop. This is concluded by Kitagawa (12) and canalso be concluded from an empirical formula (13).

PROPERTlES OF POF!OUS SUBSTRATES

(2)

where: E is the fractional porosity, <1> is the sphericityofthe substrate grains (defined as the surface ofthe spherewith the same volume as the grain divided by the surface ofthe grain i.e. <1> < 1), and D is the grain diameter.

Kitagawa (12) proposes that it is preferable to have avery narrow distribution of pore sizes in arder to gel the hestpossible filtration efficiency at a given pressure drop. Awide distribution of pore sizes with both very large poresand very small pores will give a relatively low efficiencysince the soot particles will "slip" through the larger pores.

It is very difficult to obtain a narrow pore size distribu-tion in an oxide ceramic like Cordierite, since the poreforming material shrinks and since grain growth during

The previous material has indicated that on the basisof the properties of pure polycrystalline material, SiC is avery good candidate for use as a diesel particulate filtersubstrate material. Howe~'er, due to forming and processingtechniques, the properties of porous filter materials are notexpected to be the same as those for pure, solid materials.Therefore, a series of experiments have been performed tocompare the relevant properties ofSiC to Cordierite. Mullitewas not included in the comparison, since the solid phaseproperties ofthe material INere notjudged to be significantlyattractive.

FLOW PROPER1"lES -An ideal particulate filterhas zero pressure drop and a filtration efficiency of 100%.This is not obtainable in practice, of course, but intensivedevelopment work on the SiC filter substrates has givenpromising results such as ;a very low pressure drop and goodtrapping efficiency in the same filter.

The honeycomb type wall-flow-filters (WFF) have alarger filtering surface per unit of volume than depth filters

Stobbe4

25 30 35 45 50 5540Pore Size (micron)

Fig. 3 -Permeability as afunction ofmeanpore sizeforSiCfilter substrate material.

Fig. 2 -A Scanning Electron Microscope picture of thestructure of an SiC based substrate for a diesel particulatefilter. The photograph is magnified by a factor of 100.

Fig. 4 -Filter pressure drop as a function of filter facevelocity for a Cordierite filter and an SiC filter substrate.

The perrneabilities obtained were:

4.3 x 10-13 m2 = 0.43 DarcyCordierite (33~ pores):

SiC (401l pores): 2.1 x lO-12m2 = 2.1 Darcy

sintering is difficult to control (2). In the SiC filter substratethere is no such problem because of a different sinteringmechanism, i.e. the distribution of the pore size is verynarrow because of a very narrow distribution of grain sizesin the substrate. Another advantage in the SiC substrate isthat there is no closed porosity and the pores have a capillarystructure. Figure two shows a SEM photograph of a typicalmicro structure.

The perrneability is very much dependent on the po-rosity of the filter substrate.The porosity of the present SiCsubstrate is 50% which gives a good combination of mech-anical, flow and filtration properties. This value is thereforechosen as a fixed value for different pore sizes, which can-at least to some degree -be chosen freely. These qualitiesgive the possibility of obtaining both a high filtration effi-ciency and a very low pres sure loss through the design ofthe filter substrate.

Experimental Results -The above mentioned advant-ages of the present SiC filter substrate become quite clearwhen it is compared to a typical Cordierite filter substrate.Flow tests were made both on thin disks ofthe two types ofmaterial and on whole filters without soot load. The per-meability for the two different types of filter substrate wasobtained from the measurements on the thin disks. The pore

size of the SiC material was 40Jl and that of the Cordieriteslightly smaller at 33Jl. It was found that the SiC filtersubstrate has a perrneability about rive times as large as thatobtained for Cordierite substrates with the same filtration

efficiency.

Figure three shows how the permeability varies withmean pore size for the SiC substrate and it is seen thai thepermeability has approximately a parabolic relationshipwith pore size, as indicated in Equation (2) above.

When the pressure drop was measured on clean, wholefilters of the same materials, the tendency was the samealthough the difference in pressure drop was not as large asmight immediately be expected from the permeabilitiesalone. This can be seen from Figure four.

The reason for this is thai the wall thickness of the SiCfilter is l mm while it is 0.43 mm for the Cordierite filterand thai the cell density is 8 cells per cm2 for the SiCsubstrate and 16 cells per cm2 for the Cordierite substrate.For the measured permeabilities and wall thicknesses, it is

Stobbe5

-~ .-

)(

MORA

MaRc

~

A-Crush

B-Crush

Fig. 5 -Orientations of strength test sampies for determination of MOR and crush strenght.

where: 0"0 is a normalizing factor and m is a materialconstant, the Weibull modulus.

The strength of the material is then given by the stresswhere the probability of failure is 50%.

The equation above can be altered to:

expected thai the pressure drop through the Cordieritewould be 2.1 time greater than through the SiC, in goodagreement with the measured difference of 1.8. Enginedynamometer test showed a similarrelationship between thetwo types offilter substrate for a gas temperature of250°C.

For filtration efficiencies of about 75 to 85%, thesmaller pressure drop (or back pressure) for the SiC sub-strate compared to thai of the Cordierite substrate is clearlyan advantage, since the increase in fuel consumption of thevehicle will be lower while the filtration efficiency is thesame. This gives the possibility of increasing the filtrationtime and thus increasing the time between filter regener-ations.

MECHANICAL PROPERTlES -Ceramics are brit-tIe materials and because of the low ductilitY, the compres-sive (crush) strength is greater than the tensile- and bendingstrengths (flexural strength or Modulus Of Rupture, MOR).This is a consequence of the fact thai the strength of theceramic materials is limited by the largest defect in thestructure. For porous materials such as filter substrates, ibismeans thai the largest pore limits the strength and theimportance of a narrow pore size distribution thus becomesquite clear. In more general terrns, the strength and stiffnessof the ceramic material will decrease as porositY and poresize increase.

SiC has lang been known to have excellent mechan-icaI properties at low as well as at high temperatures. Theroom temperature flexural strength of dense sintered SiChas been reported as high as 720 MPa (14), but a flexuralstrength of about 400 MPa is more typical (see Table 1). Asthe matter of fact, the strength of the SiC increases slightlywith increasing temperature (15). This fact and the goodtherrnal properties and corrosion resistance male SiC asuitable material for many high temperature applications.

Strength measurements -The mechanical propertiesof the porous SiC were investigated in arder to evaluate thesuitability as a diesel particulate filter substrate. A materialwas chosen which simultaneously exhibits a high filtrationefficiency and a low back pressure. This material has aporosity of 50% and a mean pore size of 40Jl. It consists ofan extruded honeycomb structure with a wall thickness of l

mm and a cell size of 2.5 x 2.5 mm.The foliowing strengths were measured: MORA,

MORB, M°Rc, A-Crush and B-Crush. Figure five showsthe sampies which were tested and the orientations involvedfor the different stresses. The MOR values were measuredby 3 point bending and the crush strengths were measured

using a hydraulic press.When the strength of brittie materials such as SiC and

Cordierite are evaluated, it is necessary to use the so caJledWeibuJl statistics (14). For a component of a given size,surface finish, and microstructure the probabiiity of failure,

S, when a stress, 0", is applied is given by:

10(10(1/(1-5» = m lo (0"/0"0) (4)

When the above equation is plotted, it will give astraight line with slope ffi. This type of plot is called a

Weibull plot.s = l -exp {-( O"/O"O)ffi (3)

Stobbe6

Fig. 6 -Weibu/l plots of MORA for SiC and Cordierite. Fig. 8 -A comparison ofthe MOR and crush strengthforSiC and Cordierite as a function of mean pore size.

Fig. 7 -Weibull plors af MORB and MORcfor SiC.

The probability for failure for the i'th ranked samplein a group ofN is found from:

Fig. 9 -A Scanning Electron Microscope picture of afracture in lhe SiC substrate structure. Note that the frac-tures occur in the grains. and not necessarily at the grainboundaries. The photograph is magnified by afactorof 300.

Si = i/(i+N) (5)

is a good result since it is an indication that the materials arehomogenous in structure.

The obtained strengths for SiC are up to 5 times greaterthan those of Cordierite.

This was also to be expected from the values for thedense materials (Table l). The higher strength of SiC makesit more resistant to the mechanical stresses and vibrations towhich a diesel particulate filter wiU inevitably be exposed.It also makes the SiC material more resistant to therrnalstresses during filtration and regeneration even though SiChas a larger elasticity modulus than Cordierite.

A typical value for the Weibull modulus is between 5and 10 for dense sintered ceramics. For porous filter sub-strates the value normally will be lower.

Experimental results -The Weibull plots for MORAfor SiC and Cordierite are given in Figure six and the MORBand MORc for SiC are given in Figure seven.

For the purpose of comparison, the MORA was meas-ured for both SiC and Cordierite. The strengths for the othervalues for Cordierite was taken from Gulati and Sherwood(16). The tested Cordierite has a porosity of 46% and a mean

pore size of21J.1m.The Weibull modulus for all sampies is generally

higher than expected (m = 7.5 for MORA for both SiC andCordierite and m = 4.5 for MOR B and MORc for SiC). This

Stobbe"7

ceramic due to the high thennal conductivity, hence thennalstresses are suppressed and a high durability is achieved.Even with soort tenn and severe regeneration no localtemperature peåks like those in Cordierite (12) are observed.The upper limit for soot load and the distribution of the sootin the SiC substrate is thus not limited by the thennalproperties. The "working window" (or flexibility) duringfiltration and regeneration is therefore larger for SiC thanfor other substrates.

The high thennal conductivity of the SiC will alsomake it easier to have a unifonn distribution of temperaturesin the filter when it is heated (electrically or with a burner)prior to the regeneration. Hence thennal stresses caused bylarge thennal gradients in the filter from an uneven cømbus-tion of the soot are unlikely.

Thennal Shock -The resistance to thennal shock fora material can be quantified empirically through strength,stiffness, thennal expansion and thermal conductivity in theso called Thennal Shock Parameter (TSP). This parametercan be calculated in at least 3 different ways (15, 16, 18)

MOR

E-aT-A.TTSPl = (6)

MOR. (l-v)E.aT.~T

TSP2 = (7)

(8)

In these formulas, A is Poisson's ratio. E is the elas-

ticity modulus, aT is the thermal expansion coefficient, A is

the thermal conductivity, and åT is the temperature changeofthe sample. (In table two, åT = lOOO°C -25°C)

Depending on which formula is used to calculate theTSP, either Cordierite or SiC can be shown to have thegreater value. The higher the value of the TSP, the better thethermal shock resistance. Since the mechanical and thermalproperties depend on the orientation of the material, theTSP's are given in intervals in two.

SiCMaterial Cordierite

TSPI

TSP2

TSP3

0.009 -0.05

0.008 -0.04

0.21 -0.72

0.15 -0,54

30-460 74-259

Table 2 -The different thermal shock parametersfor porousSiC and Cordierite.

To illustrate how the pore size influences the strength,the values for MORA, A-Crush, and B-Crush are plottedagainst mean pore size for SiC and Cordierite in Figureeight. The porosity of the SiC sampies is 50% and theporosity ofthe Cordierite (16) is 50%,46% and 43% forthe3 different pore sizes, 13Jlm, 21Jlm and 33Jlm respectively.The data for Cordierite are taken from Gulati (16).

Figure eight shows thai the strength of the SiC isgenerally much larger than for Cordierite independent ofpore size. The high strength of the SiC filter substrate is dueto very good contact between the individual ceramic grainsin the material. This is seen by examining the fracturesurface. Figure nine shows a SEM photomicrograph of thefracture through a SiC sample willi 50Jlm pores. This typeof fracture -transgranular -is typical for the SiC materialand is an indication of a strong grain boundary in thai it isthe material itself which breaks and not the band between

grains.THERMAL CONDUCTIVITY -For the application

as substrate for a diesel particulate filter, it is necessary thaithe material used is able to withstand high temperatures,thennal cycling, and severe thennal shocks. It is well knownthai both SiC and Cordierite have excellent resistance tothennal shock, and since these two materials have otherdesirable properties they are therefore suitable for the appli-cation.

Maximum Temperature -Cordierite is an excellentmaterial for diesel particulate filters. It has, however, anebig disadvantage which has been widely described in theliterature: The substrate melis down and/ar cracks if theregeneration of the filter is not perfonned carefully underintensive control offlow, temperature etc. This is due to thevery low thennal conductivity and low mel ting temperatureof the Cordierite. The thennal conductivity of Cordierite ison the arder of 1-2 W/m-oC for pure dense material ascompared to about 100 for the SiC. This makes the filtersensitive to both a maximum soot load and an unevendistribution of the soot. With electrical regeneration, a highsoot load (more than 4 grams /liter filter) (17) on a 12.5 lCordierite filter makes it necessary to regenerate the filtervery slowly (3-8 hrs) with a very well defined air flow. Thiscan be acceptable for a vehicle which can be put in a garageover night for regeneration (i.e. a city bus) but it is clearly aproblem for vehicles used continuously, or for on-roadvehicles driven over lang periods oftime.

Nonnally, the temperature during filtration and re-generation must be kepi below 1000-1 100°C for a Cord-ierite filter. Higher temperatures and/ar large temperaturegradients willlead to either cracks or to local melt downwhich again willlead to a large drop in filtration efficiency.When ibis occurs, the filter must be replaced.

These problems are not seen in the SiC filter substratebecause of its high thennal conductivity which has beenmeasured to be Il W/m-oC at 60°C with 50% porosity andhigh service temperature (1600°C). During regeneration,the energy related will be evenly distributed throughout the

Table two shows thai when the thennal conductivityis taken into account (TSP3), the TSP's for SiC and forCordierite are of comparable magnitude. The conclusionfrom these empirical values is, that Cordierite is somewhat

Stobbe8

HIGH TEMPERATURE CORROSION OF SICBASED DIESEL FILTERS

Olle possible concem of the use of SiC as a dieselparticulate substrate may be that of high temperature corro-sion. The authors have not yet observed any noticeableeffects of this process, but since data are not yet availablefor very long term testing, it is thought that the processwhich could lead to corrosion should be clarified.

SiC oxidizes relatively easily in connection with at-mospheric air according to the reaction [1], thereby forminga tight layer 9f amorphous silica on the surface of thematerial.

SiC + 02 ~ Si02 + CO2 [1]Fig. 10 -Bending strength of sampies of SiC and Cordieritequenched in 20°C water for different quenching tempera-tures. The ratio ofthe two strengths is also shown.

betler to withstand therrnal shock than SiC. It should, how-ever, be taken into account that the therrnal conductivity ofthe SiC material is so high that therrnal gradients are muchsmaller with SiC. This can be seen from the regenerationtests.

At high temperatures, (about 1400°C) this layerrecry-stallizes to form cristobalite. Further oxidation of the SiCbelow this layer can then only occur through diffusion ofoxygen through the reaction layer. The rate of oxidationoccurs according to a parabolic "power law" and is tempera-ture dependent according to the Arrhenius kinetic equation.The activation energy, Ea, in pure oxygen has been deter-mined to be 84.6 kj/mol for amorphous silica and 65.3kj/mol for cristobalite (21). At 900°C in pure oxygen, a40~m grain will be oxidized to a depth of < 0.05~m. Thiscorresponds to a reduction of the contact area (SiC-SiC)between the sintered grains of approximately 0.2%. At1500°C, the reduction will be about 1 %. For extreme tem-perature excursions, the layer can crack and therefore allowlocal free passage of oxygen to the SiC, and a consequentincrease in the rate of oxidation.

The presence of water vapor (50% rh at 20°C) and/olCO% in the corroding gas accelerates the process signifi-cantly (22). The rate of corrosion at 900°C in a "wet"atmosphere is equal to the rate of corrosion at 1500°C inpure dry oxygen. This is caused by a shift in the corrosionmechanism from the pacifying oxidation discussed above toan active oxidation in which the foliowing reactions arethought to participate:

SiC + 202 ~ SiO(g) + CO [2]

SiC + 4H20 -+ SiO2 + CO2 + 4H2 [3]

In order to test the thennal shock characteristics, small

sampies (l x 2 x 7 cm ) were cut from filters of SiC andCordierite. These sampies were annealed in air for 20minutes at temperatures from 70 to 620°C and were thenquenched in room temperature water. After this rathersevere treatment the 3 point bending strength (MORA) wasmeasured for all the sampies (6 sampies of each material ateach temperature. These strengths are plotted againstquenching temperature in Figure 10. In addition, the ratioof the strength of SiC to that of Cordierite is also shown.

The figure shows that both materials begin to beaffected by the quenching process at temperatures of 200-300°C. The relative strength of the cordierite decreasesmore rapidly than that of the cordierite. For the Cordierite,the strength af ter quenching from 620°C to 20°C is reducedby 43%, while for the SiC, the reduction is 74%. In tenns ofabsolute strength, the MORA of the SiC is 2.4 times higherthan that of the Cordierite when both quenched from 620°Cto 20°C.

Quenching from 220°C to room temperature in lessthan a second does not affect the strength of either materialeven though this is an extremely severe treatment. Theresults for SiC are in very good agreement with the resultsof Koumoto for 20 to 30% porous fine grained SiC (19). Asummary of the thennal and physical properties of the twoporous substrate materials SiC and Cordierite is given in

three.

SiOz + Hz ~ SiO(g) + HzO [4]

The reaction products are gaseous. Gasses such asS02, C~, and H2S will have the same effect.

The worst conceivable case is the deposit of largeamounts of combustion products, ash, which could form analkaline melt on the surface ofthe SiC grains. The protectivelayer of Si02 is easily soluble in the alkaline melt with a rateof diffusion for oxygen through the deposit to the SiCsurface which is about 104 times greater than through solid

Stobbe9

SiO2. The process is now referred to as accelerated corro-sion (22). The content of alkaline sulfates and carbonates inthe ash will therefore be very important for the use of SiCbased filter material. At 900°C for example with free accessto Na2S04 deposits, the corrosion rate will correspond to a

reduction in the grain size of 50 ~m/hr, which correspondsto a complete melting of the material.

An evaluation of the importance of the corrosionproces ses for the lire time of the filters must therefore beconducted on the basis ofthe expected operating conditionswilli respect to~ the combination of temperature, time, and

atmospheric composition. The operating conditions can bedivided up into two periods: particle accumulation andcontrolled regeneration. In addition to this is the worst case

Propert y of Filter Substlrate Silicon CarbideSiC

Cordierite

2MgO2AI2035SiO23

1.6Bulk density, p, g/cm

Porosity, E, %2Cells per cm

Cell size, mm

50 46 -(16)

8 16

2.5 x 2.5 2.1 x 2.1Watt thickness, t, mm

0.43

0.63Filte~ace per liter m2~ 0.42

Mean pore ~m 40 33 -(16)

75 -85 -(16)

0.43

Filtration efficiency, % 75 -90

Permeability ~Darcy 2.1

Back pressur~~ (velocity = 2cm/~an) 20 35Modulus of Elasticit~GPa 85 5 -(16)

0.26Poisson's ratio, v~ 0.16

~ompressive

strength~itudinal, (JA, MPa 25 9 -(16)

_Compressive

streng~nsverse, 0"8, MPa 12 3.4 -(16)

Rending str~MORA, MPa 19.5 3.5

~g

strength, MORB~~a 2.2

Bend~ength, MaRC, M~ 3.4 1 -(16)

> 10-11Electrical r~ity, .Q cm l

Specific heat, Cp, J~C.'at 25°C -(20) 750 600

<0.5Thermal conductiv~~W/m-oC, at 25°C Il

~~ennal

conductivity ,~~_oC, at 630°C 7 <0.5

1.0 x 10-64.6 X 10-6The~ansion coefficient, ~~C

150 -850Thennal Shock Parameter, TSP3, AX MORx(I-V~ 150 -750EX<XT

Decomposition/Me~temperature. °C 2300 -1200

Critical regeneration temperature limit °C-- -1600 -1000

Max. safe loading, g/lit~~ 20 [back pressure] 7 [thennalload] -(17)

Max. observed temperature during "Worst Case" regeneration, °C -900 > 1400

Table 3 -Properties for two rypes of ceramic substrates as used in a wall flow filter. Where no reference is given, the propertieshave been measured by the authors.

Stobbe10

Fig. 11 -Bosch smoke number at the intet and out/et of anSiC filter, and increase in back pressure during steady stateaccumulation of particulate matter.

rical unit. The volume ofthe filters were adjusted by chang-ing the length of the unit.

FILTRATION -Figure Il shows the filtration effectsfor a clean filter on a 9 liter turbocharged diesel engine. Theengine was run at a constant speed and load, and particulatematter accumulated. The effect of the filter was monitoredby comparing the Bosch smoke number at the inlet andoutlet of the filter. The tests were started with a clean filter.The Bosch smoke number initially was 0.95 at the filter inletand about 0.25 at the outlet of the filter. After olle hour theinlet Bosch number increased to 1.25 due to the increasingback pressure, while the smoke number out of the filterdecreased to under O.l. This indicates the increasing effi-ciency due to the increase in the mass of particles on thefilter. While the Bosch number cannot be translated directlyinto a particle collection efficiency, the numbers shown areconsistent with a trapping efficiency in the range of 70 to90%.

REGENERATION -

Exhaust ~as driven re~eneration -SiC particulatefilters have been constructed and tested for a variety ofengine types and regeneration strategies. Figure 12 showsengine back pressure, speed and load, and filter tempera-tures for regeneration of an uncoated SiC filter. Althoughthe authors do not have experience willi catalytically coatedSiC substrates, it is believed that it is technically possible toapply a catalytic coating. Specifications for the engine andfilter used for the test are shown in Tables 4 and 5 on thenext page. Particles are accumulated on the filter at a me-dium load, and then the regeneration is started by the highexhaust temperature resulting from increasing the speed andload of the engine. No auxiliary regeneration energy wasused. When the filter inlet reaches about 550°C, the regener-ation starts, as can be seen by the reduction in the backpressure in the exhaust manifold.

situation which results from a failure of the regenerationcontrol system. During normal operation and accumulationof soot, the temperature will not exceed 600°C and the filterwill remain completely unaffected. For slow regenerationprocesses, the temperature will remain lower than 800°C.During a rapid controlled regeneration, the temperature willonly increase to about 900°C depending on the amount ofaccumulated particles. The filter will be exposed tb thistemperature for a very short period of time, approximately30 seconds, arter which the temperature falls. For filterusage during continuous driving it will be necessary toregenerate about 3-4 times a day. For a filter lifetime ofabout 2 years, this corresponds to a lire cycle regenerationtime of 20-30 hours at 900°C. Excluding corrosion fromalkaline ash deposits, the corrosion will be minimal. For theworst scenario, where large amounts of particulate matterare accumulated, experience with Cordierite filters shows ashort term increase in temperature to over 1200°C. Corro-sion could be significant but not catastrophic under theseconditions, and the lire of the filter will depend upon howoften these conditions are encountered. However, results todate with regeneration of SiC filters indicate that the tem-peratures are much lower, nearly always lowerthan 900°C.

The problem with alkaline melt becomes of import-ance for temperatures about the eutectic temperature for agiven mixture of alkali sulfates, carbonates, chlorides, etc.This temperature caD be as low as 300-400°C. How muchdamage, if any, might occur will depend on the quantity andcomposition of the deposit/ash from the fuel and particulate.A point of interest in the future evaluation of long termdurability of SiC filters is then the composition of thelubricant additives, since they may remain on the filter arter

regeneration.A recent study ofthe effects oflubricating oil and fuel

composition on ash accumulation has shown that arter 3000hours of operation with conventional fuels and lubricants ina 240 kW engine, 1.27 kg of ash were accumulated in aparticulate filter (23). The ash contained a variety of metal,some of which might give long term difficulties. By switch-ing to low sulfur fuel and ashless lubricant, the amount ofash accumulated was reduced to 0.06 kg, and the elementswhich might cause a long term corrosion problem were notpresent. On the contrary, a large portion of silicon wasobserved in the deposits, which certainly presents no prob-lem for the filter.

It caD then be stated that the general demand for alltypes of filters, that low ash lubricant and low sulfur fuel beused, will contributeto a reduction or elimination of corro-sion potential on SiC substrates.

ENGINE TEST RESUL TS

In all the engine tests reported in this section, the filterswere constructed of material corresponding to the charac-teristics given in two. They were constructed by sealing 4quadrants ofthe type shown in Figure 1 into a round cylind-

Stobbe11

1.6

75.5

86.4Stroke -mm

Cylinders

Combustion System

lntake System

Maximum Power -kW

Rated Speed -rpm

æ~, e!"!2I ea.

"C"O,..c:IV~

4

lndirect Injection

Bonnally _Aspirate?

35

4000

TabTe 4 -Engine Specifications for the engine regenerationtests shown in Figures 12-15.

Fig. 12 -Filter temperature distribution and exhaust backpressure during the exhaust temperature driven regener-ation of an SiC filter under steady state conditions. Engineandfilter specifications are given in Tables 4 and 5.

Table 5 -Filter specifications for the filter regenerationtests shown in Figures 12-15.

Fig. J 3 -A sequence of regenerations for the conditionsshown in Figure J J.

sufficient air to maintain the regeneration, but is marginalin terms of energy removal.

Figure 14 illustrates ane of these tests. In this case, anuncoated SiC filter with the 2 liter capacity and the samecharacteristics as the filter of Table 5 was compared to theCordierite filter shown in Table 5. The filters were loadedat a medium speed and load, then heated by an increase inengine speed and load, and when the regeneration starts, theengine was rapidly put into idle.

Figure 14 shows that the back pressure developmentduring the entire test was essentially the same for the twofilters. This is due to the small size of the SiC filter, andindicates that the accumulation and regeneration proceededin essentially the same manner for both filters. However, amarked difference was noted in the temperature develop-ment. For the Cordierite filter, the temperature at the centerof the filter increased dramatically from the initial tempera-ture of 680°C to a maximum in excess of l 100°C. Thistemperature is very near the melting point of the Cordierite,and in fact this is a typical failure mode for the Cordieritefilter, in that a regeneration is started and the engine flow is

With the engine maintained at constant speed andload, it can be seen that the regeneration proceeds in acontrolled fashion, over a period o~ about 10 minutes, witha very uniform temperature distribution within the filter.The filter temperatures are low, and in this case do notexceed 600°C anywhere during the entire course of theregeneration. Figure 13 shows a sequence of 4 ofthese testsconducted in series. It can be seen thai the back pressuredrops to essentially its initial value, confirming the com-pleteness of the regeneration. The uniformity of temperatureis observed in all the cycles.

"Worst Case" Regenerations -In arder to demonstratethe additional margin of safety provided by the use of SiCsubstrate material, experiments were performed whichwould be expected to lead to the thermal destruction of othertypes of filters. The basic concept is to start a regeneration,and then put the engine into a condition which supplies

Stobbe12

o 20 40 60 60Regenemtion time (sec)

100 120

Fig. 14 -"Worst Case" Regeneration of an uncoated SiCfilter in engine exhaust gasses. The filter is loaded at amedium speed and load, then heated by an increase inengine speed and load, and when the regeneration starts,the engine is rapidly put into idle.

Fig. 15 -Filter temperature distribution during a "worstcase» regeneration of an SiC filter. The filter was heavilyloaded and ,regeneration occurred during idling conditions.Engine and.filter specifications are given in Tables 4 and 5.The temperature locations are the same as in Figure 12.

with the "City Filter" concept(lO). This type ofregenerationoccurs witbi a filter which is located at a position in theexhaust gas system where regeneration conditions are neverencountered. Therefore, the particulate matter is collectedfor a lengtbl of time and then burned while the vehicle isparked. This technique, which is quite sui for off-road andservice vehicles with a reasonably fixed daily schedule,would allow regeneration under carefully controlled condi-tions.

drastically reduced. A Cordierite filter with a higher particu-late loading would have melted in the center. Due to the lowthermal conductivity, the local temperature becomes veryhigh at the location where the regeneration front is found

(24).For the SiC filter however; the center temperature of

the filter does not exceed 800°C, even though the regener-ation proceeds in an equivalent manner to that of the Cord-ierite filter. Thus, the advantages,ofthe use of SiC as a filtersubstrate are clearly demonstrated. Not only is the meltingpoint of the SiC much higher than that of the Cordierite, itsuse reduces the maximum filter temperature encounteredunder a regeneration, further increasing the margin of

safety.

The n~generation of an SiC filter by this technique isshown in Fiigure 16. Two 6 liter filters were used with thestructural Sl>ecifications given in Table 5. The regenerationis controlleci by a combination of the electrical heating rateand a controlled flow of air through the system. A 6 kWelectrical heater was used, and the air flow was 300 litersper minute. The heating element was located below and justupstream oj' the filter. The air flow through the filter was inavertical djrection from the bottom to the top.

Figurc~ 16 shows the time development of the filtertemperaturf~s and the oxygen concentration at the outlet ofthe filter. As was the case with the exhalist gas regenerationprocess shown above, the use of SiC as a filter materialresults in a ]low maximum temperature, in this case approxi-mately 800°C. Under similar flow conditions and loading,a 12 liter siingle Cordierite filter was found to melt in thecenter of the filter.

EFFE:CT OF THERMAL CONDUCTIVITY -In theexperimental results, it has been shown that the uge of SiCsubstrate rf~sults in lower and more uniform temperaturesunder regeneration. In order to verify the influence of ther-mal conductivity on the temperature development, a simpli-fied mathe:matical model was used. In the model, it wasassumed th:at a filter consists of a one-dimensional structure,

An additional "worst case" type of regeneration wasalso investigated with this engine and a 3 liter SiC filterdescribed in Tables 4 and 5. In this case, an uncoated SiCfilter was loaded with appr9ximately 40 grams of particulatematter at an intermediate speed and load. At this loading, asignificant deterioration in engine performance was ob-served. Foliowing this, the engine was stopped, and the filterwas electrically heated to a temperature of about 600°C. Theengine was then started and idled at 1100 rpm, at which timea very vigorous regeneration was observed. The tempera-ture development throughout the filter during this rapidregeneration is shown in Figure 15. It can be seen that thefilter temperature does not exceed 900°C at any time orlocation. It can also be observed that the temperature timehistories are quite similar in nature to those generated by thesimplified heat transfer model described below. No physicalchange could be detected in the filter.

Electrical Regeneration -Tbe benefits of the use ofSiC canalso be observed during the electrical regeneration

Stobbe13

Fig. 17 -The effect ofthermal conductivity on the simulatedtemperatuJ"e development in a simulate porous material,with space and time dependent energy release, and convec-

rive heat transfer.

Fig. 16 -Filter temperature and outlet oxygen concentra-tion for an SiC filter regeneration by the use of electricalheating and controlled air flow on a parked vehicle.

in which there is an internat energy release which is timeand space dependent. In this case, it was assumed that acombustion wave, composed of a sinusoidal shape with

width, ~, traverses the structure at a constant velocity, c,regardless of the substrate material. In addition, it wasassumed that the combustion energy released in the materialwas carried away by a convective process, corresponding to

flow through a porous medium.The model results in the following partial differential

equation, which was solved by conventional finite dif-

ference techniques (24):

a2e

aE.2

7t to Hu .7tL to't+f.- --sm [- (~--)] =2 tr CpTo 8 1r

(9):lae ~(e-eG)

7+ A

were chosen based on Tables one and two. The product ofthe convec:tive heat transfer coefficient and the surface areaof the pon)us structure was arbitrarily chosen such that thetemperature of the first part of the filter to be heated de-creased to approximately the inlet gas temperature at the endof the enelrgy release. The convective heat transfer was thenmaintained constant, and the thermal conductivity of the

material v"as varied from l to 500 W/m.oC.The results are shown in Figure 17. It can be seen that

as the thermal conductivity increases, the temperature riseis much more gradual, and the magnitude of the peaktemperatulre encountered is reduced. Similar results are alsoobtained 'with slower regenerations, provided that the con-vective h(~at transfer rates are adjusted to maintain a similarbalance b'etween energy release and removal rates. In thissimple model, the effects of the convective heat and masstransfer on the reaction rate are not included. A more de-tailed model of the complete process, including chemicaireactions, can be found in Reference (25).

Thf: trends calculated with this simplified model arein good qlualitative agreement with the experimental resultsfor a variety of regeneration processes, both rapid and slow,for SiC filters, as shown in the results of this work and alsoin Reference (3). This tends to confirm the beneficial effectsof a high thermal conductivity material in reducing both themaximum temperature and the thermal gradients en-

countered during filter regeneration.

CONCL.USIONS

It has been shown that SiC possesses physical andthermal properties which make it well suited for use as asubstratl~ for diesel particulate filters. When compared to

Cordierite, it exhibits higher thermal conductivity, physicalstrength, melting temperature, working temperature, elec-

Where: e is the ratio of the temperature to the initial

temperature -To, c, is the non-dimensional distance, E is thefilter loading in grams of particle per kg of filter, Hu is theheating value of the particulate matter, Cp is the specific heatof the filter material, L is the length of the filter, to = L 2/ais the reference time for thermal diffusion, tr is the time forthe reaction wave to mave from ane end of the filter to the

other, ) 't = t/to is the dimensionless time, y is the ratio ofsurface area to volume of the filter material, h is the convec-

ti ve heat transfer coefficient, A is the thermal conductivity,and the subscript, G, represents the gas.

A simulation was performed for a rapid regeneration,thai is a regeneration which occurs in about ane minute. Thecombustion wave was assumed to travel at 5 cm per secondover the filter length of 300mm. Typical material values

Stobbe14

.

trical conductivity, thermal expansion coefficient, and equi-valent filtration efficiency. The adverse effects ofthe higherthermal expansion on thermal stresses are compensated byhigher strength and other advantageous physical properties.Calculated thermal shock parameters indicate that SiC maybe worse than or equivalent to Cordierite willi regard tothermal shock. Thermal shock (quenching) experimentsindicated that a substrate material of SiC has mechanicalstrength which is at least 2.5 times greater than that ofCordierite.

When SiC filters are regenerated under differing con-ditions it is found thatwhen compared to Cordierite filtersof comparable size, local temperatures are lower in the SiCfilters, and that SiC filters can survive regenerations whichwould result in the melting of a Cordierite substrate. Langterm corrosion does not appear pose a significant problemwilli SiC filter substrate. As willi other types of filter sub-strates, the use of ashless lubricants and low sulfur fuels willgive significant improvement in filter performance.

"Worst case" testing of SiC filters indicates that evenfor very severe regenerations processes, which would meltmany other filter types, the maximum temperature of theSiC substrate does not exceed 900°C, and the SiC filtersurvives.

Simplified heat transfer calculations, which includecombined effects of thermal conductivity, internat energyrelease, and convective removal .,of energy, confirm theobserved advantages of a substrate with a high thermalconductivity. These advantages include lower peak tem-peratures and small temperature gradients during regenera-tion processes.

REFERENCES

8. Handenb,erg, H. O. and Dandel, H. L., "Particulate TrapRegeneration induced by means of Oxidizing Agents In-jected into the Exhaust Gas", SAE paper 870016, 1987.9. fnger, B. E., "Catalytic Control of Diesel Particulate",SAE Paper 820184,1982.10. Eismarl;:, J., "The Volvo City Filter-Tackling the Par-ticulate Problem", Volvo Technology Report, 1990.)r;'Watermlan, N. A., Ashby, M.S~,EIsevierMate~e~lecto~, Elsevier, London, 1991.l:];;. Kitagavva, J., Asarni, S., Uehara, K. and Hijikata, T.,"ImprovementofPose Size Distribution ofWal1 Flow TypeDiesel Particulate Filter", SAE paper 920144.13. Meyer, B.A. and Smith, D.W. "Flow Through PorousMedia: Comparison of Consolidated and UnconsolidatedMaterials", Ind. Eng. Chem. Fundam., 24, 360-368,1985):14. LemkO\V, J., and Sørensen, O. T., (eds), Materiale Kend-ska~, (Knowledge ofMaterials) p 28, Danish TechnologicalInstitute, Tiistrup, Denmark, 1992.15. Schwetz, K.A. and Lipp, A. "The Effect of Boron andAlurninum Sintering Additives on the Properties of DenseSintered Alpha Silicon Carbide", Science of Cerarnics, 10,Hausner, H. (ed.) Deutsche Keram. Gesellschaft, 1987.16. Gulati, T. S., and Sherwood, D. L., "Dynarnic fatiguefor Cordieitite Cerarnic Wall-Flow Diesel Filters", SAEpaper 91013517. Kitaga'wa, J., Hijikata, T., and Yamada, S., "E1ectricHeating Regeneration of Large Wall-Flow Type DPF',SAE paper 910136.18. Gibson, L. J., and Ashby, M. F., Cellular Solids, Perga-rnon Press, London, 1988.19. Koumoto, K., Shirnizu, H., SeO, W. S., Pai, C. H. andYanagida, H. "Thermal Shock Resistance of Porous SiCCerarnics"" British Ceramic Transactions and Journal, 90,32-33, 1991.20. Barin, I. ~oche~pata of Pure Substances,VHC Ver1ag, Berlin, 1989.21. Jorgensen, P. J., Wadsworth, M. E. and Culter, I. B.,"Oxidation! of Silicon Carbide", ,. Am. Cer. Soc., 42,12,1959, pp 613-616.22. McKee:, D. W., and Chatterji, D., "Corrosion ofSiliconCarbide in Gases and Alkaline Melts", l. Am. Cer. Soc.,59,9-10, 1~)76, pp 441-444.23. Barris, M. A., Reinhart, S. B. and Wah1quist, F. H., "TheInfluence of Lubricating Oil and Diesel Fuel on Ash Accu-mulation iin an Exhaust Particulate Trap", SAE paper910131, 1~)91.24. Garner, C. P., and Dent, J. C.," A Therma1 RegenerationModel folr Monolithic and Fibrous Diesel Particu1ateTraps", SAE Paper 880007, 1988.25. Holman, J. P., Heat Transfer, Mc Graw-Hill Book Co.,New York, 1972.

1. Howitt, J. S., Montierith, M. R., "CellularCeramic DieselParticulate Filter", SAE paver 810114, 1981.2. Itoh, A., Shimato, K., Komori, T., Okazoe, H., Yamada,T., Niimura, K. and Watanabe, Y. "Study of SiC Applica-tion to Diesel Particulate Filter (Part 1): Material Develop-ment", SAE paver 930360.3. Itoh, A., Shimato, K., Komori, T., Okazoe, H., Yamada,T., Niimura, K. and Watanabe, Y. "Study of SiC Applica-tion to Diesel Particulate Filter (Part 2): Engine Test Re-sults", SAE paver 930360.4. Engler, B., "Catalytically Activated Diesel ParticulateTraps- New Developments and Applications", SAE paver860007, 1987.5. Mihara, T., Kuwano, S., Kusuda, T., Yonemura, M.,Nakamoto, M., Takigawa, M. "Diesel Particulate Trap ofCorrugate Honeycomb Fabricated with Mullite Fiber Ce-ramics", SAE Paver 860010, 1986.6. Goto, Y., T. Abe, T. Sabo, M. Hayashida, "Study ofRegeneration of Diesel Particle Trapper by Electrical Self-Heating Type Filter", SAE Paver 920140. 19927. Handenberg, H. O., "Urban Bus Application of a CeramicFiber Coil Particulate Trap", SAE paver 870011, 1987.

Stobbe15