Ceramic Foam

8/2/2019 Ceramic Foam

1/12

Journal of the European Ceramic Society 18 (1998) 1339-13500 1998 Elsevier Science Limited

PII: SO95S-2219(98)00063-6Printed in Great Britain. All rights reserved

0955-2219/98/.$-see front matter

Ceramic Foams by Powder ProcessingL. Montanaro,O* Y. Jorand,b G. Fantozzib and A. NegroaODepartment of Materials Sciences and Chemical Engineering, Politecnico, Cso. Ducadegli Abruzzi 24,10129Torino, ItalybG E M P P M. . . . *, U.M.R. 5510,I.N.S.A Lyon, Bat. 502,69621 Villeurbanne cedex, France

Abstract

Ceramic foams show a significant potenti al of devel-opment and application, essentially due to the emer-gence of environm ental preoccupat ions . A briefoverview of the state of the art in cellular ceramicapplication, preparation and characterizatio n is pre-sented in order to introduce some new data concern-ing th e elaboration of mullite and PZT foams by areplication and a bubble generat ion met hod, respec-tiv ely. Some discrepancies betw een t he theory,developed for describing the properties of open-cellfoams, and th e experimental mechanical behaviourof t hese semi-closed cell materials w ere also show n.0 1998 Elsevi er Science Limi ted. A ll rights reserved

1 IntroductionThe objective of this paper is to present anoverview of the state of the art, to give new dataconcerning the elaboration of porous materials andcomparing them to the literature.Introducing cellular solids, Ashby consideredthat when man builds large load-bearingstructures, he uses dense solids: steel, concrete,glass. When nature does the same, she generallyuses cellular materials: wood, bone, coral. It isalmost certainly that cellular materials permit thesimultaneous optimization of stiffness, strengthand overall weight in a given application.Cellular ceramics are comprised of variousarrangements of a space-filling polygons (cells) andcan be classified into two broad groups: honey-combs and foams.2In honeycomb the cells form a two-dimensionalarray, whereas foams are comprised of a three-dimensional array of hollow polygons. Foams areusually sub-divided into two further categories,depending on whether or not the individual cellspossess solid faces.*To whom correspondence should be addressed. Fax: 00391-1 -564-46-65; e-mail: [email protected]

If the solid of which the foam is made is con-tained only in cell edges, the material is termedopen-cell. If the cell faces are present, the foam istermed closed-cell and the individual cells are iso-lated from each other.There is clearly the possibility that foams can bepartly open and partly closed.These porous network structures3 have relativelylow mass, low density, and low thermal con-ductivity, and differ in the property of perme-ability, having the open-pore ones the higherpermeability. By combining the proper ceramicmaterials and processing, porous ceramics can alsohave relatively high strength, high resistance tochemical attack, high temperature resistance, highstructural uniformity.

2 Present and Future ApplicationsThese properties make ceramic foams suitable for avariety of applications. Both closed-cell and open-cell foams are used as thermal insulating materialsfor furnaces and also for aerospace applications(tiles for space shuttles), fire protection materials,low mass kiln furniture and gas combustion bur-ners. The most common applications for open-cellporous ceramics are molten metal and Dieselengine exhaust filters, catalyst supports,4 industrialhot gas filters, grease filter for commercial kitch-ens.5 Expanding applications are now also beingfound in the electronic and biomedical areas. Thefoams already commercially available or in studyare made of various materials, following the con-straints of specific applications, as cordierite;4Jjmullite;1~7~8silicon carbide;2p9 alumina;2T8-i2 par-tially stabilized zirconia;1,8,9T12 nd some compositesystems (Sic-alumina;13 alumina-zirconia;2 alu-mina-mullite;2 mullite-zirconia).2.1 Thermal insulationA principal application of these foams is in fabri-cation of thermal insulators, due to their specificcharacteristics, as thermal stability, low thermal

1339


2/12

1340 L. Montanaro et al.conductivity, low density, resistance to thermalcycling, thermal shock resistance, low gas adsorp-tion and absorption, low specific heat, and alsobecause they are available in various size andconfigurations.

Preliminary measurements of a zirconia foamreveal that thermal insulation equivalent to that ofNASA space shuttle protective thermal tiles can beachieved at 550C higher operating temperature.Many different compositions of refractory foamswere investigated (carbon, oxide and non-oxidematerials).*,i42.2 Molten metal filtrationFor production of castings, foam ceramic filters arehelping to improve quality and productivity byremoving non metallic inclusions; the filters mustbe able to resist attack at high temperatures by avariety of molten metals, which can contain suchreactive elements as aluminium, titanium, hafniumand carbon, Thermal shock behaviour is obviouslyalso important: it was found to be strongly depen-dent on cell size (increasing with increasing cellsize) and weakly dependent on density (increasingwith increasing density).The material selected depends on the material tobe filtered and is usually a metallic oxide of variouscompositions (Table 1).8,152.3 Hot gas cleanupApplication of high-performance, high-tempera-ture particulate control ceramic filters is expectedto be beneficial not only to the advanced fossil-fuelprocessing technologies, but also to high-tempera-ture industrial processes, waste incineration pro-cesses and to diesel soot filtration. Developmentand utilization of hot gas filtration depends on thecreatice design and use of new high-temperaturematerials.The criteria for successful use and operation ofporous filters as a viable advanced particulateremoval concept therefore requires not only ther-mal, chemical and mechanical stability of ceramic

materials, but also long-term structural durability( > 10 000 h) of the entire filter and high reliabilityof integrated process design features.i6$17Such filters must withstand variation in the efflu-ent gas stream chemistry, variation in the natureand loadings of the entrained fines, and oscillationsin the effluent gas stream temperature and possiblepressure, while still maintaining high particulateremoval efficiencies, high flow capacity, and rela-tively low pressure drop flow characteristics. Dur-ing operation the filter must also withstand a varietyof mechanical, vibration, and thermal stresses.The principal materials for these applicationsinclude alumina, mullite, cordierite, silicon nitride,silicon carbide. Both alumina/mullite and cordier-ite have been demonstrated to have certain advan-tages over non-oxide materials. The oxides alreadycontain stable oxide phases which do not undergofurther phase transition. They also retain theirphysical integrity during exposure to gas phasealkali: in fact long-term degradation mechanismsmay result from chemical reactions, particularlywith alkali species and/or steam, which wouldaffect the long-term durability of the system.Even if till now candle filters and cross-flow fil-ters have been mainly applied, ceramic foams arepresently investigated, above all for diesel soot fil-tration, for their high particulate removal effi-ciency, their high flow capacity and low pressuredrop generation.6,10,i8

3 ProcessingNumerous processing routes are available to rea-lize porous ceramics:i5 capsule-free HIP, bubblesgeneration into a slurry or at a green state during aspecific thermal treatment, reaction sintering, con-trol of the sintering conditions in order to achieve apartial densification, stacking of presintered gran-ules or fibres, aerogel and sol-gel methods, pyr-olysis of various organic additives, polymericsponge method.

Table 1. Some characteristics of molten metal filtersTradename Composit ion Appli cations BenefitsCeltrexCorningCerapor

Udicell

AlucelSelee

55% AIZOs. 38%Si0*.7%MgO77% AIZOs. 23% SiOzAlumina, Sic, cordierite, ZrOz

Alumina, mullite, ZTA, PSZ

92% alumina with mullite phaseAlumina, PSZ

Iron alloysCarbon low alloy,stainless steelAluminum, iron,copper, bronze,steel, zincSuperalloys,low-carbonstainless steelNonferrous alloysAluminum, iron,steel

Reduction in scrap ratePouring temperaturesup to 1675CLaminated duplex andtriplex constructionLarge volumes up to120 tonsImproved thermal shock resistance,smaller filters requiredHigh flow rates


3/12

Ceramic foam s by pow der processing 1341This paper deals on the foam ceramics processed by:l a replication method or polymeric sponge

method;l bubble generation method by a pore foaming

agent.3.1 Polymeric sponge methodThe first method, patented in 1963,19 consists in theimpregnation of a polymeric sponge with a ceramicslurry followed by a thermal treatment which leadsto the burning out of the organic portion and tothe sintering of the ceramic skeleton.Many steps must be optimized to develop a foamproduct having the desired performances, namely: thechoice of the polymeric template; the preparationof the ceramic slurry; the impregnation; the thermalcycle comprising drying, burning out of the volativecomponents, sintering of the ceramic portion.A variety of open-cell, semi-closed and closed-cellsponge materials are suited to replication process;their pore size determines the pore size of the ceramicfoam after the shrinkage linked to the firing step.The sponge should volatilize at low temperaturewithout yielding noxious by-products; in addition,it must readily soften and burn off, without indu-cing sensible residual stresses and disrupting theunsintered ceramic network. Its resiliency, itshydrophobic behaviour and its ability to be uni-formly covered are other significant properties.Many polymeric-sponge materials can satisfythese requirements, namely poly(urethane), cellu-lose, poly(viny1 chloride), poly(styrene), latex. Insome cases, spongelike polymers (like polysilanes,polycarbosilanes) have been used to prepare siliconcarbide foams by pyrolysis: these pyrolyzed porousnetworks were then immersed in ceramic (a-alu-mina) slurries to obtain a composite foam.i3One of the key steps of this process is undoubtedlythe elaboration of an appropriate ceramic slurryable to uniformly cover the polymeric walls, toeasily sinter in a dense ceramic network, able towithstand the in-use constraints.Even if limited information can be found in lit-erature, being a large number of foam elaborationprocesses patented, it is certain that a commonslurry for this application is formed by a ceramicpowder, a dispersion medium (generally water) andsome additives. Two typical slurry formulations,respectively in aqueous and non-aqueous medium,are reported in Tables 2 and 3.3 The ceramicmaterial is firstly obviously chosen depending onthe particular application and desired properties ofthe final foam.A pure, fine ceramic powder having a narrowparticle size distribution is usually requested:dimensions lower than 45 micron are common, and

Table 2. Polymeric sponge process: typical aqueous basedrecipe31, Prepare ceramic slurry

A1203C&3KaolinBentoniteColloidal aluminiumorthophospahte in waterWaterfor a total slurry content of82% solids, 18% water2 Immerse Scm thickpoly(urethane) foam inceramic slurry.3. Knead foam to remove

41% by weight13%3.5%1%14.5%

excess air; remove from slurry.4. Pass impregnated foam throughrollers to remove 80% of slurry.S.Oven dry at 125C for 1 h.6. Slowly heat at 0.5C mint to500C; hold at 500C for 1 h7.Heat to 1350C at lCmin-and hold for 5 h

Final propertiesPermeability 1425x lO~cm*Porosity 0.87Pore size 12 pores/linear cm-Thickness 5cmStructural uniformity Excellent

generally the mean particle size is close to fewmicrons. In addition, equiaxial particles shouldlead to a more homogeneous coating of the poly-meric walls.The quantity of particles that can enter a spongedepends on factors which are related to thestructure of the sponge (that is its suction force,which is optimized increasing the open porosity ofthe sponge, and its ability to retain the ceramicparticles, against gravity force), but also to theconcentration of the slurry.The slurries contain very variable solid weightpercent, but usually ranging between 50 and70wt% of solids. For higher solid concentrations,the slurry becomes more and more viscous and theparticles might then have difficulty in entering thesponge structure: therefore, the sponge loadingdecreases (Fig. 1, Ref. 20).Many additives should be used for improving thecoating performances or even the final sinteredfoam properties.Table 3. Polymeric sponge process: nonaqueous slurry com-position3 _~Alumina (RC-HPT)90 trichloroethane-10 ethanolBinder (Butvar B-76)Plasticizer (Carbowax PEG-300)Plasticizer (Ucon 5@-HB-2000)Dispersant (Sarkosyl 0)MgG

41.5%41.5%1.9%1.0%1.9%0.1%0.2%-___


4/12

1342 L. Montanaro et al.80 -32i% 7 0 -

! 6 0 -

5 5 0 -

2 0 3 0 4 0 5 0 6 0 7 0Slurry Concentration (wt. %)

Fig. 1. Relationship between particle loading and slurry con-centration.

7 0 wt% in solid were prepared, adding differentadditives, as clay, calcium and zinc salts. In thiscase, the viscosity range is very different, as shownin Figs 2 and 3. l1 The clay additions lead to anewtonian behaviour; deviations from the linearbehaviours versus time were observed starting from6 wt% additions, due to the swelling of the clay inwater. The calcium and zinc additions lead to athixotropic behaviour and to limit viscositieshigher than in the case of clay (300-600mPa sagainst 100-200 mPa s). But, unfortunately theauthors do not give the influence of these viscosityparameters on the coating performances. The onlydata which are indirectly correlated are the

A binder can provide strength to the ceramicstructure after drying and prevent collapse duringvolatization of the organic portion. Some rheologicalagent should be added to adjust the slurry viscositySometimes a thixotropic behaviour has been sug-gested for an appropriate coating: when the poly-meric sponge is impregnated, the slurry must be fluidenough to enter, fill and uniformly coat the spongenetwork and subsequently regain enough viscosityunder static conditions to remain in the sponge.In addition, being the impregnation generallyconducted by compression to remove air, immer-sion of the sponge in the slurry, and free expansionof the polymer in the ceramic suspension, it shouldbe supposed that a high viscosity of the slurryretards the movement of the sponge when it tries torecover its original shape.Different slurry formulations and optimal visc-osities are reported in literature. In the case of thecoating of a poly(urethane) sponge by a potteryclay slurry, containing sodium carbonate41 wt%related to clay-as defloculant, the evolution of theviscosity as a function of slurry concentration isreported in Table 4.2o They noted a sharp increasein viscosity for slurry concentrations higher than45 wt% and they concluded that a too much vis-cous slurry is detrimental to the formation of uni-form products. In conclusion, they assume verylow viscosity value (about IO-20mPa s) as theoptimum value.On the contrary, always for the coating of apoly(urethane) sponge by an alpha alumina slurry,slurry concentrations ranging between 60 and

Table 4. Slurry concentration and viscositySlurry concentrat ion (wt %) Viscosity (cPs.)26 1.935 2.040 4.845 12.550 1290.0

0 20 40 60 80Time (min.)

Fig. 2. Viscosity versus time curve for alumina clay slurries.1400

q SlurryD, CaO: SiO2 2110 (3 : 3 : 1)1 2 0 0 l S l u r r y, CaO:Si02 : ZnO (6 : 5 : 1)1000

010 2 0 t4 0 6 0 8 0Time (min.)

Fig. 3. Viscosity versus time curves for alumina slurries con-taining CaO, ZnO and SiOz.

l,OI1350 1400 1450 1500 1550 1600Temperature (C)

Fig. 4. Mechanical strength versus temperature for aluminafoams.


5/12

Ceram ic foam s by powder processing 1343

modules of rupture of the foams differently added(Fig. 4, Ref. 11). In another paper,12 the beneficialeffect of polyethylene oxide (PEO, average mol-ecular weight = 100 000) on the coherent andhomegeneous coating of a poly(urethane) foam hasbeen underlined. The optimum slurry formulationconsists of 40 ~01% of ceramic powder (alumina oralumina/zirconia), 2 wt% of a polyelectrolyte(Darvan C) and 1 wt% PEO (referenced to thepowder weight in slurry). However, also in this case,the influence of this addition on slurry viscositywas not reported.

Tulliani7~2 investigated the formulation of mul-lite slurries for coating poly(urethane) foams in thep.p.1. range 25-65. The aqueous slurries were sta-bilized at pH 10 by NaOH addition. Different solidconcentrations and additives were tested in orderto optimize the coating. The viscosity changes werealso evaluated. The more significant results arereported in Table 5. The polymeric foams are thenimmersed in the well dispersed slurries and arecompressed whilst submerged in order to fill all thepores. The impregnated poly(urethane) support isthen removed from the slurry and excess materialsqueezed from the foams by means of a rollingmill. Different compression ratios can be set on themill in order to achieve a well distributed coatingon the sponge support and to improve the perme-ability of the sample. Drying can be done in air orin an oven.Firing of the samples is a two stage process asshown schematically in Fig. 5.7 The first stage atabout 300C consists of slowly decomposing andburning out the poly(urethane) support withoutcollapsing the deposited mullite powder, whereasthe second stage, at high temperature (155OC inthe mullite case) is for the sintering and densifica-

T (Cl

Fig. 5. Thermal cycle for pyrolysis of the organic support andsintering of ceramic coating made of mullite.tion of the ceramic powder.Some other thermal cycles reported in literatureare the following: for alumina and alumina-zirco-nia foams12 the heating schedule consists of aheating rate of lCmin-i to 550C rapid heating(2h) from 550 to 1600C and 1 h hold at 1600C.For alumina foams the samples were heated upto 500C at lCmin-; subsequently, the tempera-ture was raised to 1400-1500C in 5-8 h. At thefinal temperature, the material was soaked for 4-8 h and allowed to cool in the furnace. For SiC-alumina foams,13 the samples were sintered for 1hat 1300C in flowing argon and 1 h at 1600C in1.5 MPa argon, at a heating rate of lOCmin-.The struts and pore walls consist of well sinteredmaterial, with a grain size of few microns(Fig. 6.21). Some typical defects in microstructureof the foams should be easily observed, as the tri-angular voids inside struts and long cracks betweenwalls (Fig. 7.21). Mercury porosimetry (Fig. 8.,Ref. 21) confirms that the pore walls and struts arewell densified: there is also a clear indication of apolymodal distribution of the pore sizes. The largersize range is undoubtedly related to the voids insidethe struts and between the pore walls whereas thesmaller is indicative of the finer porosity within thesintered walls.

Table 5. Composition, viscosity and coating performance of some mullite slurriesS olid concentration(mullite, wt%)

Add itive (wt% referred to m ullite) Viscosity (mPa s)(at 2Os-I, at 20C)

Coating

70 Polyelectrolyte (Darvan C) = 2 150 Newtonian7050

50

55

Polyelectrolyte (Darvan C) = 2Surfactant (silicone-type) = 1Polyelectrolyte (Darvan C) = 2Rheological agent(carboxymethylcellulose) = 0.8Polyelectrolyte (Darvan C) = 2Rheological agent(carboxymethylcellulose) = 1.6Polyelectrolyte (Darvan C) = 2Rheological agent

520 Newtonian1500 slightlythixotropic6300 slightlythixotropic8300 thixotropic

(carboxymethylcellulose) = 1.560 Polyelectrolyte (Darvan C) = 2Rheological agent 8600 thixotropic60 (carboxymethylcellulose) = 1.0Polyelectrolyte (Darvan C) = 2Rheological agent 130 pseudo-plastic(Polyethylene oxide, M.W. = 100 000) = 1 O

Bad,d&continuitiesBad, discontinuitiesSlightly better thanthe previous coatingsSlightly better than theprevious coatingSlightly better than theprevious coating, but difficultiesin homogeneously coating foamshaving a p.p.i. number higher than 50Good coating for foams having ap.p.i. number lower than 50Good coating for foams having ap.p.i. number higher than 50


6/12

1344 L. Montanaro et al.3.2 Bubble generation methodThe first technique using this method was patentedin 1973:22 in the majority of cases, a chemical mix-ture containing the desired constituents is treatedto evolve a gas which creates bubbles and causesthe material to foam. In the first patent, clays weremixed to propellants (calcium carbide, calciumhydroxide, aluminum sulfate, hydrogen peroxide). Inacid media, metal blowing agents, like magnesium,calcium, chromium, manganese, iron and cobalt,should be added for promoting hydrogen gas evolu-tion; in alkali media, aluminum is typically employed.Also in this case a drying and sintering step needsto develop a self-supporting ceramic network.A foaming agent, as silica gel, carbon black, talc,mica, is often added to give uniform foaming. But,more easily, freon has been dispersed as fine dro-plets in the slurry, and a surfactant agent dispersesthe freon and stabilized the gas bubbles. A similarresult can be also achieved when a sponge poly-meric network is produced simultaneously with thefoaming of a ceramic-filled slurry.Compared to the polymeric sponge method,the foaming method allows to produce small-pore-sized closed-cell foams, which cannot bemade by an impregnation technique.

Fig. 6. SEM micrography of struts and pore wall of a ceramic Fig. 7. Micropores and macro defects (triangular voids) of afoam: (a) 61 ppi; (b) 32 ppi. mullite foam.

Using this method, Boumchedda23 developed aporous PZT material for hybrid pressure sensors.In this case an adequate organic additive,which induces an expansion at low temperature inthe green body, was choosen: this allows theobtention of highly porous ceramics, havingtypically 60% open porosity and upper values ofabout 90%, and a pore size ranging between0.01 and 1 mm. The ceramic powder presented aparticle size distribution close to 1.5 micron; theadditive showed a mean particle size of about20 microns.A wet mixing of the pore foaming agent with theceramic powder was performed in ethylic alcohol,being the organic additive insoluble in this med-ium. After oven drying, the mixture was 1OOpmsieve granulated and uniaxially pressed underlOOMPa. At this step the microstructure is onlyconstituted by organic particles well dispersed withthe ceramic grains. Then a specific thermaltreatment is realized in order to transform theorganic additive into a high viscosity liquid which

(a)


7/12

Ceramic foam s by pow der processing 1345surrounds the ceramic particles. The green micro- is about 7.7 g cmp3. Concerning the microstructurestructure in an expanded state was obtained as a of the wall cell, it can be seen in Fig. 1 (a) that aresult of the decomposition gas pressure and the good densification can be obtained with thiseasiness of the green microstructure deformation. method. A typical defect observed is givenAn illustration of mechanism of transition fromthe initial green microstructure to the expandedone is given in Fig. 9.The organic additive (from glucide family) dur-ing the thermal treatment must develop a liquidhaving a sufficiently high viscosity so that theshape of the green body is maintained, no sig-nificant deformations appear and no cracks aregenerated by the gas bubble formation during itsdecomposition. Moreover, when the temperature islowered, the remaining organic additive acts as abinder.Typical microstructures of foams obtained bythis method are given in Fig. 10: the sintered den-sity of the resulting ceramic foams was lower than1 gcme3, even if the theoretical density of the PZT

VIM PoreillsMcmn)Fig. 8. Pore size distribution of a mullite foam.

Initial state Foam generation

_. _.treatment

*

0 Ceramic gramsm Adjuvant0 Porosity

Fig. 9. Scheme of the transition from the initial green micro-structure to the green foam microstructure.

Fig. 10. Typical ceramic foam microstructure obtained by thepore generation method (optical microscope, x 15).

Fig. 11. Micropores and macro defects of a P.Z.T. foamobtained by pore generation: (a) aspect of the microstructurein the cell walls; (b) crack defect.


8/12

1346 L. Montanaro et al.Fig. 1l(b). These cracks appear during the sinter-ing process, but they result from residual thermalstresses induced during an inadequate foamingthermal treatment.

4 Mechanical Behaviour of Ceramic FoamsIt is important to relate the mechanical behaviour totheir microstructure and to know the significantparameters which control this mechanical behaviour.In order to derive mechanical relationships forcellular ceramics, Gibson and Ashby have pro-posed an idealized unit cell shown in Fig. 12. Asignificant property of the foam is the relativedensity (p/ps, where p is the density of the bulkfoam and ps is the density of the cell edges andfaces). The relative density of an open foam islinked to the thickness t of the cell edges or faces,the length L of the edges as follows2 (for lowdensity):

P/W = Cl(t/q2 (1)4.1 Elastic behaviourThe cell edge bending is the essential deformationmode of an open foam. By using standard beamtheory, the deflection of the cell edges*andtherefore the elastic modulus can be determined:

E/J%= G(W4= G(P/P~)~ (2)where E and Es are, respectively, Youngs moduliof the foam and the edge.

Fig. 12. Scheme of an idealized unit cell (from Ref. 23).

Moreover, foams can show some elasticanisotropy. Experimental data exhibit good agree-ment with eqn (2) with a Cs value between 0.36 andO*5.2An example of dependence of E with the relativedensity is shown in Fig. 13 for two mullite foams25with 32 and 61 p.p.i., respectively. The exponent isnear the theoretical one (2) for the 32p.p.i. foam(l-8) whereas it is higher (2.4) for the 64p.p.i.foams. This difference is probably due to the factthat the foams are not constituted of totally opencells, but of partially closed cells. In this case, thedependence of E with the relative density is nomore given by eqn (2) but by more complexequation.244.2 Fracture toughnessThe fracture toughness of open cell foams dependson the edge strength (a~~,cell size (L) nd relativedensity as follows:26

Krc = c,flfim(p/ps)32 (3)with C4~O.65.In order to optimize the foam toughness, alarge cell foam with a high edge strength mustbe used. So the process must be improved thatstrut defects (pores, cracks...) may be removedthereby increasing u-f,. Triangular cavity appearsduring polymer removal in foams obtained bycoating and burnout of polymer foams (Fig. 7).The pores as well as other microstructural defectslike inclusions or cracks influence the strutstrength.Fracture toughness of brittle foams can be mea-sured either by using the single edge-notched beamgeometry2 or from the work of fracture fjobtained by measuring the area U under the curveload-displacement up to the maximum load.27 Thecritical potentialG1c is given by? energy release rates of the foam

Grc = 2y;lp (4)

with p = qb/( 1 - x), x = a/wa = flaw sizew = specimen width7c= geometric fonction depending on x and speci-men geometryrf= work of fracture = Up1 m-im = specimen mass1 = specimen length

Fracture toughness is obtained by the relationship


9/12

Ceramic foam s by pow der processing 13474- Y=

3,5 --3 --

3 2,5 --& 2--w 13 --

1 --075 --

0 1 II I I02 0 2 4 O P 0,32 OJ6 0,4

[ 3 2 p.p.i.1 dpr

I,41,210 30,6 li

0

i

094y = 1 , 8 0 2 8 x + 3 , 0 6 8 6 0,2

I I I I LO-I,4 -1,3 -1,2 -I,1 -1 Q9

( 3 i p . p . i , ) w e )6- 05 -- 0

0 cd -4 -0;;i

4 --g 0 e* /;----- ccw 2 -- : 0

2-0: la --e--9 0 __/------ _i__---=@ *~

: 0>8--0

0,15 416 0,17 418 0,19 -1,9 -I,85 -1,8 -I,75 -1,7 -1,65PP Mdpr)

Fig. 13. Example of dependence of E with the relative density (from Ref. 25).

KIC = JEGICThe evolution of KIc as a function of relative den-sity for two mullite foams is show in Fig. 14.25 Thefracture toughness increases with relative densityfor the 32 p.p.1. foam whereas it seems independentfor the 61 p.p.1. foam.This type of plot does not take into account anevolution of the strut strength with density asindicated by eqn (3). From eqns (1) and (3), it canbe shown that Klc should increase as (t/L)3. Breznyand Green2* have measured the strut strength byusing the mirror method. They have normalizedthe fracture toughness with the strut strength andcell size and then plotted this normalized Klc as afunction of t/L (Fig. 15). This plot is independentof the relative density as shown by eqn (1). Theslope obtained for the alumina-mullite open foamis in good agreement with the Gibson and Ashbymode1,24 the slope being near 3.

The two other materials exhibit a more rapidincrease, others factors contributing to thisincrease. Finally, for open cell foams, the fracturetoughness can be predicted correctly by eqn (3).

4.3 Tensile strength2Under tensile stress, the struts parallel to the loaddirection should fail first and should be responsiblefor the foam strength. In fact, struts are randomlyoriented and so failure of struts occurs essentiallyby bending.Tensile strength of foams can be obtained by afracture mechanics approach. Taking into accounteqn (3), the tensile strength af is given by therelation:P@*(1 2o f t = Kd f i = Gq s G ( 5 )

Cs = geometric factor ?Z 0.18a = critical flaw sizeFor a cellular structure, the lower limit of a is thecell size L.Equation (5) shows that, for a given density,tensile strength increase requires an improvementin the strut strength afS.The bend strength of two mullite foams havebeen measured as a function of relative density by


10/12

1348 L . Montanaro et al.Tulliani in three point bending.25 The Weibull dence being stronger for the 32p.p.i. foam than formoduli were estimated from the bend strength the 61 p.p.i. one. So, the mullite foams do notresults: 1.8 for the 32p.p.i. foam and 3.2 for the exhibit a good agreement with eqn (5) and this61 p.p.1. foam. These values are very low, indicat- disagreement can be due to the influence of celling a wide distribution of flaw sizes. faces, the foams being partially closed.

The effect of density on the bend strength ofthese two mullite foams is shown in Fig. 16. Onecan notice that bend strength has approximativelya linear dependence on relative density, the depen-4.4 Compressive strengthMaiti et 0-1.~~have extended the model of Gibsonand Ashby to the compressive behaviour of foams.Compressive rupture occurs by struts bending. Foropen cell ceramics, the compressive strength isgiven by:Q4 - =35 y 2 . 3 0 5x - 0 , 5 0 93 4- Rz = 0 , 7 4 22 lR

h ; 93 -- 0 *2 /,25 -- 0m l2 -- 9+*

0915 -- 0.3 0,l -- 7.0,05 -- / ,*

lO- I4202 p.p.i. 025 0,3 Q35le

0 :t .--- __----0_--

0l l

0,4

025 -r \ w - -v!5 0,15 --g6

OJ --0,05 --

0 f I0,14 0,16 0,18 020 1.p.i. p/P

Fig. 14. Example of dependence of KIT with the relativedensity (from Ref. 25).

GA model mediction

,001 I,1 1Strut thickness / length

Fig. 15. Dependence of normalized KIc with the t/L ratio.Experimental data for alumina, alumina-zirconia, alumina-mullite and theoretical line given by the G A model (fromRef. 2).

a f t = c 6 af s f0 1 2with C6 g O-65. Eqn (6) has the same form as thatfor tensile strength [eqn (5)] and thus for foams thetensile and compressive strengths are similar. Thisresult is very different from the dense ceramics one(the compressive strength of dense ceramics is anorder of magnitude higher than the tensile one).A typical load-displacement plot of a 61 p.p.i.mullite foam is shown in Fig. 17. An inital linearelastic behaviour is first observed with some strut

3T25 --

--22 1s --s I--

0 3 t

y =1 5 , 5 7 4 x - 3 . 3 2 23R== 0 , 8 4 3 2 t

* , '84

0-l 1 I, I I0,24 0,28 0,32 0,36 0,4

dcs3,5

i

y = 19,331x - 1 . 4 5 773 a

2?538, 2 0 0 _c t_/-- .0 195

t* -4 -

1 a l0,5 t04 I I II I I I

0,14 0,15 0,16 0,17 0,lS 0,19 02pps

Fig. 16. Example of dependence of the flexural strength withthe relative density (from Ref. 25).


11/12

Ceramic foam s by pow der processing 1349

i

iv

;:.

DISPLACEMENT (mm)Fig. 17. Typical load&-displacement plot (from Ref. 25).

fractures which correspond to slight drops in thestress-strain curve. The stress increases as thedamage. At the maximum load, macroscopiccracks propagate with a strong load drop. Finally,densification can occur corresponding to a loadincrease (not shown in Fig. 17).

Figure 18 shows the variation of compressivestrength of mullite foams as a function of relativedensity.25 The Weibull moduli are very low as forbend strengths: 2 for the 32p.p.i. foam and 3.5 forthe 61 p.p.i foam.The compressive strength is higher than bendstrength. There is a discrepancy between theexperimental results and that predicted by eqn (6).The exponent is about 2 for the 61 p.p.i. foam and0.3 for the 32p.p.i. foam.One reason for this disagreement between theoryand experiment can be due to the difficulty toobtain an uniform loading. Another reason is dueto the presence of faces (the foams are partiallyclosed cell ceramics) The work of fracture mea-sured during compression is an order of magnitudehigher than for tension, due to the cells collapse.Concerning high temperature behaviour, it isimportant to know the thermomechanical proper-ties of foams, particularly thermal shock resistanceand creep behaviour, Foams presenting low ther-mal conductivity and toughness, they can besignificantly damaged during thermal shock. This

6 -.5 --

c% 4 -_0 3--b

2 --1 -_

l

44

4_______--_--- -_.4

Y = 4,0877x + 2.34840 - l I! I ! I03 0,28 0,3 0,32 0,34 0,36

Cl3 2 p.p.i. dP

95 -- y = 26,694x - 206530 1 , I I I I

914 415 0,16 417 0,18 0.19CII6 1 p.p.i. p/ P

Fig. 18. Example of dependence of the compressive strengthwith the relative density (from Ref. 25).thermal shock resistance can be measured by fol-lowing the elastic modulus, the bend or compres-sive strength as function of the quenchingtemperature.2 Creep and thermal shock behaviourare two important properties for the lifetime offoams.

5 ConclusionsCeramic foams have an important potential ofdevelopment, essentially due to the emergence ofenvironmental preoccupations. Nethertheless manyothers applications may emerge in the near future.Many elaboration routes have been developed toproduce ceramic foams, but it is important to notethat, in consideration to the field of the applica-tions previously cited, low cost processes and rawmaterials are imperatively required. The twoexamples detailed here are adapted to an industrialproduction of low cost ceramic filters.Considering the mechanical properties evalua-tion of the cellular materials, one can note thatnumerous studies also exist. On the other hand,little work has been done in the field of the high


12/12

1350 L. Montanaro et al.temperature corrosion. However, it is clear that inthe case of hot filtration, each application willconstitute a particular case. So investigations inthis field have to be developed.

References1.2.

3.

4.

5.

6.

7.

8.9.

10.

11.

Ashby, M. F., The mechanical properties of cellularsolids. Metallurgical Trans. A, 1983, 14A, 1755-1769.Brezny, R. and Green, D. J., Mechanical behavior of cel-lular solids. In Mat erials Science and Technology, Vol. II,Structure and Properties of Ceramics, ed. R. W. Cahn, P.Haasen and E. J. Kramer. VCH, Germany, 1992, pp. 467-516.Saggio-Woyansky, J., Scott, C. E. and Minnear, W. P.,Processing of porous ceramics. Amer. Cer. Sot. Bull.,1992, 71(11), 16741682.Inui, T. and Otowa, T., Catalytic combustion of benzene-soot captured on ceramic foam matrix. Appl. Catalysis,1985, 14, 83-93 .Masuda, T., Tomita, K. and Iwata, T., Application ofceramic foam. In Ceramic Transactions, Vol. 31, PorousMaterials, ed. Ishizaki et al. The American CeramicSociety, Westerville, OH, 1993, pp. 285-293.Mizrah., T., Maurer, A., Gauckler, L. and Gabathuler, J.P., Open-pore ceramic foam as Diesel particulate filter.SAE Paper 890172, 1989.Tulliani, J. M., Montanaro, L. and Borello, C., Open-poremullite foams for Diesel particulate filtration. Proceedingsof the 4th Euroceramics, Vol. 4, Basic Science-Trends inEmerging Materials and Applications, ed. A. Bellosi.Faenza Editrice, Faenza, Italy 1995, pp. 493-500.Ceramic foams for molten-metal filtration. Amer. Cer.Sot. Bull., 1991, 70(7), 1149-l 150.Otzuka, M. and Akiyama, T., Structure and mechanicalbehavior of ceramic foams. In Ceramic: Charting theFuture, ed. Vincenzini. Techna, Italy, 1995, pp. 2431-2438.Tutko, J. J., Lestz, S. S., Brockmeyer, J. W. and Dore, J.E., Feasibility of ceramic foam as a Diesel particulatetrap. SAE Paper 840073, 1984.Tantry, P. K., Misra, S. N. and Shashi Mohan, A. L.,Developmental studies on porous alumina ceramics. InCeramic Transactions, Vol. 31, Porous Mat erials, ed.Ishizaki et al. The American Ceramic Society, Westerville,OH, 1993, pp. 89-99.

12.

13.

14.

15.

16.

17

18.

19.20.

21.22.23.

24.25.26.27.28.

Lange, F. F. and Miller, K. T., Open-cell, low-densityceramics fabricated from reticulated polymer substrates.Adv. Ceram. Ma ter., 1987, 2(4), 827-831.Lannguth, K., Preparation of macro-porous Sic-Al203composites with polysilanes and polycarbosilanes. Ceram.Int ern., 1995, 21, 237-242.Sherman, A. J., Tuffias, R. H. and Kaplan, R. B.,Refractory ceramic foams: a novel, new high-temperaturestructure. Amer. Cer. Sot. Bull, 1991, 70(6), 1025-1029.Sheppard, L. M., Porous ceramics: processing and appli-cations. In Ceramic Transactions, Vol. 31, Porous Materi-als, ed. Ishizaki et al. The American Ceramic Society,Westerville, OH, 1993, pp. 3-25.White, L. R., Tompkins, T. L., Hsieh, K. C. and Johnson,D. D., Ceramic filters for gas cleanup. J. Engin. for gastu rbines andpow er, 1993, 115(7), 665-669.Alvin, M. A., Lippert, T. E. and Lane, J. E., Assessmentof porous ceramic materials for hot gas filtration applica-tions. Am er. Cer. Sot. B ull, 1991, 70(9), 1491-1498.Gabathuler, J. P., Mizrah, T., Dole, M. and Bressler, H.,Filter for cleaning exhaust gases of Diesel engines. USPatent no. 4913712, 1990.Schwartzwalder, K. and Somers, A. V., Method of mak-ing porous ceramic articles. US Patent no. 3 090 094,1963.Chou, K. S., Liu, H. C. and Chang, K. L., Microstructureevolution during fabrication of a porous ceramic filter. InCeramic Transactions, Vol. 31, Porous Mat erials, ed. Ishi-zaki et al. The American Ceramic Society, Westerville,OH, 1993, pp. 101-110.Tulliani, J. M., Montanaro, L., Swain, M. V. and Bell, T.,Semi-closed cell mullite foams. Submitted to J.A.C.S.Sundermann, E. and Viedt, J., Method of manufacturingceramic foam bodies. US Pat no. 3 745 201, 1973.Boumchedda, K., Elaboration and characterization ofcomposite piezoelectric materials of 3-3 connection. Ph.D.thesis, INSA, Lyon, France, 1994.Gibson, L. J. and Ashby, M. F., Cellular Solids: Structureand Properties. Pergamon Press, New York, 1988.Tulliani, J. M., Ph.D. thesis.Maiti, S. K., Ashby, M. F. and Gibson, L. S., Scripta.Met all., 1984, 18, 213-217.Fantozzi, G., Fracture of Materials. Institut National desSciences Appliquees de Lyon, France, 1996.Brezny, R. and Green, D. J., J. Am. Ceram. Sot., 1989,72,11451152.

Ceramic Foam

Documents