fabrication of reaction-bonded sic ceramics by slip casting of sic/c suspension
TRANSCRIPT
A
odlsa©
K
1
ctcgasreotmd
wiosip
0d
Materials Science and Engineering A 483–484 (2008) 676–678
Fabrication of reaction-bonded SiC ceramics by slipcasting of SiC/C suspension
Yuan Li ∗, Jing Lin, Jiqiang Gao, Guanjun Qiao, Hongjie WangState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong University, Xi’an 710049, PR China
Received 6 June 2006; received in revised form 7 December 2006; accepted 20 December 2006
bstract
The influence of particle size and size distribution on slip casting performance of SiC and SiC/C suspension was investigated. Two kindsf commercial grade SiC powder and petroleum coke powder were used as the starting powder. The results showed that suitable particle sizeistribution is helpful to obtain low viscosity and relative high packing density. Smaller particles were used to prevent sedimentation velocity of
arger particles. Ternary suspension containing petroleum coke and two kinds of silicon carbide powder exhibited higher viscosity than binaryuspension containing only two kinds of silicon carbide powder. The right amount of petroleum coke could lead to suitable viscosity for slip castingnd high flexural strength. The maximum flexural strength of sintered products was 274 ± 15 MPa in this work. 2007 Elsevier B.V. All rights reserved.nsion
ds
2
(cFruap
Nfipp
eywords: Silicon carbide; Slip casting; Particle size distribution; Binary suspe
. Introduction
Preparation of reaction-bonded silicon carbide (RBSC)eramics requires that the green body should contain a cer-ain amount of carbon or carbon source [1,2]. Therefore, siliconarbide and carbon source particles should be dispersed homo-eneously together in aqueous medium when RBSC ceramicsre prepared by slip casting. Slip casting has been reported as aimple and inexpensive consolidation process to produce mate-ials with high green density and homogeneous microstructureven for complex geometries [3,4]. According to the researchf Paik [5], in which carbon black powder (80 nm) was used ashe carbon source, carbon black is difficult to disperse, and theixture suspension was prone to segregate during the consoli-
ation.In this work, petroleum coke power, which could disperse
ell in water, and silicon carbide powder were used as the start-ng powder. The aim of this work is to seek suitable fractionf each raw powder, prepare slurries with suitable viscosity for
lip casting, and get RBSC products with relative high mechan-cal performance. The influence of particle size distribution andetroleum coke fraction in suspension upon viscosity, packing∗ Corresponding author.E-mail address: [email protected] (Y. Li).
wcSpcif
921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.msea.2006.12.162
ensity, and ultra flexural strength of sintered products were alsotudied.
. Experimental procedure
The starting powders used in this study were petroleum cokePC) powder with mean diameter of smaller than 40 �m, and twoommercial grades of SiC powders (W14 and W3.5, Dongtanine Ceramics Co., China) with mean diameter of 14 and 3.5 �m,espectively. Carboxymethyl cellulose Na salt (CMC-Na) wassed as the dispersant as well as binder. Distilled water workeds the dispersing medium and ammonia water was used to adjustH value of the suspension.
In order to prepare suspension suitable for slip casting, CMC-a salt solution with 0.3 wt.% concentration was prepared atrst. Then suspension with 41.5 vol.% solid loading was pre-ared at pH value of 10 [6]. SiC powder and petroleum cokeowder were continuously added to the CMC-Na salt solutionith hand stirring. The resulting suspensions were ultrasoni-
ated for 20 min followed by mechanical stirring for 60 min.ubsequently, such suspension was aged to improve the dis-
ersion and stability, and then used to fabricate green bodies byasting the suspension in a suitable plaster mould. After demold-ng, the green bodies were kept at room temperature for 8 h,ollowed by completely drying in an oven at 100 ◦C. Finally, thengineering A 483–484 (2008) 676–678 677
ga
(odMTtC
3
fvtatoaasaia
pss(ttitfldbba
F(
Ft
iWnwstpcssIbplpdar
Y. Li et al. / Materials Science and E
reen bodies were reaction-bonded at 1550 ◦C under vacuumtmosphere.
Suspension viscosity was measured by rotational viscometerNDJ-1, Cany precision instruments Co., China). The densityf green and sintered body was determined by Archimedes’isplacement method in kerosene and water, respectively.icrostructure was observed using an optical microscope.
hree-point flexural strength of sintered RBSC samples wasested using electrical multi-test instrument (CMT5104A, Sansio., China).
. Results and discussion
The suspension viscosity of single raw SiC particles could beound in Fig. 1. The suspension of W14 particles showed loweriscosity but were more prone to sedimentate than the W3.5 par-icles. PC particles showed good stability in aqueous medium,nd its suspension viscosity was 1150 mPa s, which was betweenhe suspension viscosity of W14 and W3.5 particles. The naturef the commercial SiC particle surface is mainly originated fromthin SiO amorphous layer covered with hydroxyl groups [7,8],nd the PC particles may contain hydrophilic groups on itsurface, so that these two kinds of particles could disperse inqueous medium. The difference in suspending behavior orig-nated mainly from different surface chemical properties, sizend density of the raw material powder.
Suspension viscosity and relative packing density of binaryarticles (W14 + W3.5) are also shown in Fig. 1. Compared withingle type of particle, when using mixture of binary particles,uspension viscosity decreased evidently. A minimum viscosity215 mPa s) was indicated at 50 wt.% W3.5 fraction, and fur-her increasing of W3.5 fraction resulted in sharp increasing ofhe viscosity. During mechanical stirring, smaller particles werentroduced and distributed to the interstices of the larger par-icles in suspension [9]. Such distribution would afford lowerow resistance and decrease the viscosity. Moreover, if such
istribution was preserved during the consolidation of the greenody, close-packed structure and high packing density shoulde achieved. The highest relative packing density, 61.2%, waschieved at 50 wt.% of W3.5 fraction, corresponding to the min-ig. 1. Viscosity and relative packing density vs. W3.5 fraction in binary systemW14 + W3.5).
aistipdttf
tTvimspwr
ig. 2. Relative packing density and viscosity vs. petroleum coke fraction inernary system (W14 + W3.5 + PC).
mum viscosity. Relative packing density was 57.2% at 75 wt.%3.5 fraction, which may be due to the high viscosity. It was
oticeable that relative packing density at 25 wt.% W3.5 fractionas only 53.4%, which was even lower than the packing den-
ity of green body containing 75 wt.% W3.5 particles, althoughhere was only minimal difference between the viscosity of sus-ensions containing 25 and 50 wt.% W3.5 particles. This resultonfirmed that uniform distribution of particles in suspensiontate had not been preserved during the consolidation due to theegregation of particles when W3.5 fraction was not sufficient.n addition, it was observed that not only sedimentation velocityut also segregation tendency decreased when fraction of W3.5articles added, i.e. smaller particles prevented sedimentation ofarger particles. Under this condition, smaller particles and largerarticles can deposit at the same velocity, so that homogeneousistribution was preserved and high packing density could bechieved. Considering both the casting performance and cost, theatio of W3.5/W14 was fixed at 1/2 in the following experiment.
Fig. 2 shows the evolution of suspension viscosity and rel-tive packing density as a function of petroleum coke fractionn ternary system (W14 + W3.5 + PC). Single PC particles haveuspension viscosity of 1150 mPa s. As a whole, ternary sys-em exhibited higher viscosity than the binary system, butts viscosity was much lower than each of the single startingowder suspension. Viscosity of ternary component suspensionecreased with the PC fraction within the range of 15–25 wt.%;hen increased with further addition of PC fraction. In con-rast, relative packing density increased continually with the PCraction.
In suspension, part of CMC-Na molecules was absorbed ontohe particle surface, and the rest was distributed in liquid media.he former could improve its dispersion and resulted in the loweriscosity by electrosteric stabilization mechanism, and the latterncreased the suspension viscosity due to interaction of long
olecule chains [10]. On one side, ternary system had smallerpecific surface area, since ternary system contained more larger
articles than the W14 + W3.5 suspension; at a fixed CMC-Naeight concentration, decreasing of specific surface area wouldesult in more CMC-Na molecules in the liquid media, thus,
678 Y. Li et al. / Materials Science and Engineering A 483–484 (2008) 676–678
F
vodspt
wcicpth
sFp2pstwprtbcsatcfoimAc
g
Fo
pSpgsfs
4
ppWiovatstrPw
R
Powders, Elsevier Science B.V., Amsterdam, 1983, p. 735.[8] M.N. Rahaman, Y. Boiteux, Am. Ceram. Soc. Bull. 65 (1986) 1171–1176.
ig. 3. Three-point flexural strength vs. petroleum coke fraction in green body.
iscosity would increased. On the other side, gradually addingf larger particles in ternary system could result in the viscosityecreasing as smaller particles were inserted exactly into the freepace between the larger ones, thereby creating relative denseacking. The viscosity was determined by these two opposingendencies so that viscosity plot as shown in Fig. 2 was achieved.
The evolution of relative packing density versus PC fractionas somewhat amazing, since it was usually accepted that the
losed packing structure is achieved at a low viscosity. Thisndicated that the packing density is determined not only by vis-osity but also by other factors. It was supposed that a greateracking efficiency could be achieved by optimization of the par-icle size distribution when different sized particles sedimentateomogeneously in spite of different viscosity.
Fig. 3 shows the variation of three-point flexural strength andintered density with the petroleum coke fraction in green body.lexural strength improved firstly followed by decreasing asetroleum coke fraction increased. The maximum strength was74 ± 15 MPa when PC fraction was 25 wt.%. In general, highacking density would lead to better sinter ability and highertrength. However, such opinion may not be correct for reac-ion bonding process of RBSC ceramics. The strength of RBSCas mainly determined by carbon fraction, with minor by theacking density. During the reaction bonding, the melted siliconeacted with carbon to form new silicon carbide, which bondedhose original silicon carbide particles together [11]. If the car-on fraction was insufficient (<25 wt.%), new-formed siliconarbide was insufficient to fill the space around carbon particleso that free silicon was too much; when carbon fraction wasppropriate (∼25 wt.%), new-formed silicon carbide could fillhe space around the carbon particles and join the original sili-on carbide particles together, which led to the least free silicon;urther adding of carbon fraction (>25 wt.%) led to the presencef free carbon because the carbon particles are quickly closedn new-formed silicon carbide during reaction-bonding so that
elt silicon could not contact and react with the carbon particles.
s a result, neither insufficient nor excessive amount of carbonould lead to perfect sintering and enough sintered strength.Fig. 4 shows the microstructure of RBSC obtained from the
reen body with weight ratio of PC/W14/W3.5 of 25/50/25. The
[[
ig. 4. Microstructure of RBSC prepared from starting powder with weight ratiof PC/W14/W3.5 of 25/50/25.
arts of gray color represented the original SiC and new formediC particles and those of the white color indicated silicon. Theores between particles were filled with silicon. Relative homo-eneous distribution could be observed, and neither large area ofilicon nor residual carbon was found. Suitable starting powderraction and slip casting process ensured relative high flexuraltrength of RBSC products.
. Conclusions
RBSC ceramics could be fabricated via aqueous slip castingrocessing using SiC and petroleum coke powder as startingowder and CMC-Na salt as the dispersant. A certain amount of3.5 particles were helpful to lower viscosity, improve the pack-
ng density, and prevent sedimentation of W14 particles. Addingf PC particles to SiC suspension resulted in the increasing ofiscosity due to decreasing of specific surface areas. Viscositynd packing density changed with particle size distribution andotal particle surface areas in ternary system. However, sinteredtrength and density was extremely determined by carbon frac-ion, since perfect reaction bonding could realize only at theight amount of carbon fraction. The most suitable fraction ofC particles was 25 wt.% and the maximum flexural strengthas 274 ± 15 MPa.
eferences
[1] Q. Huang, G. Qiao, Rare Metals Eng. 30 (2001) 150–152.[2] O.P. Chakrabarti, P.K. Das, J. Mukerji, Mater. Chem. Phys. 67 (2001)
199–202.[3] X. Xu, M.I.L.L. Oliveira, J.M.F. Ferreira, J. Mater. Sci. Lett. 20 (2001)
2043–2044.[4] V.A. Hackey, U. Paik, J. Am. Ceram. Soc. 80 (1997) 1781–1788.[5] U. Paik, Hyeon-Cheo, Mater. Sci. Eng. A 334 (2002) 267–274.[6] I. Varga, F. Csempesz, G. Zaray, Spectrochim. Acta B 51 (1996) 253–
259.[7] M. Person, L. Hermnsson, R. Carlsson, in: P. Vincenzini (Ed.), Ceramics
[9] S.M. Olhero, J.M.F. Ferreira, Powder Technol. 139 (2004) 69–75.10] J. Cesarano, I.A. Aksay, J. Am. Ceram. Soc. 71 (1988) 1062–1067.11] J.N. Ness, T.F. Page, J. Mater. Sci. 21 (1986) 1377–1397.