J Am. Cerum. Soc.

, 71 [lo] 841-44 (1988)

Low-Temperature Sintering of Aluminum Oxide

TSUNG-SHOU YEH* and MICHAEL D. SACKS*
Ceramics Division, Department of Materials Science and Engineering, University of Florida, Gainesville, Florida 326 11

A fine-sized (-0.1 pm), agglomerate-free A1203 disper- shear rate, j ~ flow

, behavior) were determined with a concentric
sion was used to prepare homogeneous green bodies with
-69% relative density and -10-nm median pore radius.
''
cylinder viscometer. Viscosity values, 7.were obtained using
the relationship
Samples could be sintered at 1150°C to a relative density
>99.5% and an average grain size of 0.25 pm. 77 = 7IY
Electrophoresis was used to determine the suspension electro-
I. Introduction kinetic behavior. A concentrated suspension (-38 ~ 0 1 % )was
centrifuged (-30 min at -15000 rpm) to obtain an essentially
M ESSING and co-workers have recently described a method
for producing high relative density (-99%) a-A1203at sin-
tering temperatures of = 1200°C. The fabrication process in-
particle-free supernatant. A representative dilute suspension
was prepared by mixing the supernatant with a small amount
( S O . 1 ~ 0 1 % of
) concentrated suspension. In some experiments,
volved the preparation of boehmite sol-gel samples containing the solution ionic strength was controlled by adding potassium
small additions ( 5 2 . 0 wt%) of fine (-0.1 pm) a - A l z 0 3 seed chloride (3.2 X 10-3M). The dilute suspensions were used to ob-
particles. The a-AIzO3particles acted as nucleation sites for the serve particles in a microelectrophoresis apparatus.it For each
8- to a-A1203phase transformation, resulting in a significant sample, a minimum of 80 individual measurements were taken
enhancement of the transformation kinetics. Furthermore, the (at several field strengths) to obtain an average electrophoretic
microstructure after the transformation consisted of relatively mobility value. The zeta potential was determined using the mo-
aggregate-free, fine-grained a-AlZO3and sintering occurred at bility and the analysis of Wiersema, Loeb, and O ~ e r b e e k . ~
much lower temperatures compared to unseeded sol-gel samples. Green compacts were prepared by slip casting. Aqueous sus-
However, it was reported that preparation of boehmite sol-gel pensions with pH =4 and -25 vol% solids concentration were
monoliths was difficult; gels undergo large shrinkages during poured into plastic rings set on absorbent plaster blocks. Cast
drying and cracking tends to occur. In the present paper, it samples were dried at -90°C for 24 h and subsequently calcined
is shown that high relative density (>99%) a-A1203can be at 500°C for 1 h before sinterin . Powder compacts were charac-
produced at temperatures < 1200°C by controlled suspension terized by mercury porosimetry!' The pore channel radii distribu-
processing of a-A1203. Shrinkage during drying and sintering is tion was obtained using standard values for the mercury surface
considerably reduced compared to the boehmite sol-gel process. energy (484 erg/cmz (4.84 X J)) and the contact angle
(140"). The total porosity, TP (i.e., the volume percentage of
11. Experimental Procedure porosity), was calculated using the relationship
The aluminum oxide used in this study was obtained by sedi-
mentation of a commercially available powder* in order to obtain
an aggregate-free, fine fraction of the material. Sedimentation
was carried out with dilute (-3 vol% solids concentration), well- where Vs is the specific volume (cm3/g) of pores (i.e., the
dispersed aqueous suspensions. Good dispersion was achieved by specific volume of mercury intruded into the porous samples) and
(1) using ultrasonication' to break up powder agglomerates and p, is the theoretical density of the alumina powder (3.98 g/cm3).
(2) adjusting the pH to =4 in order to generate a high zeta poten- The relative (or percentage of theoretical, % p , ) green density of
tial on the alumina particles. powder compacts was calculated from the relationship
The alumina powder was characterized by a variety of tech- %pt = 100 - TP
niques. Specific surface area was measured using nitrogen gas
adsorption' (multipoint BET method). Particle size distribution Samples were sintered under isothermal conditions at 1150°C
was determined by a centrifugal sedimentation method* and by in air. Bulk density values were determined by the Archimedes
scanning electron microscopyn (SEM). A thin gold-palladium displacement method using distilled water. Relative density val-
coating was deposited before making microscope observations. ues were determined by dividing the bulk density by the alumina
Phase analysis of the alumina was carried out by X-ray diffrac- theoretical density. The grain sizes of sintered samples were de-
tion (XRD) using a diffractometer** with Ni-filtered CuKa radia- termined by measuring -600 grain intercept lengths on scanning
tion and a scanning rate of l"/min. electron micrographs of thermally etched polished sections. After
Suspension rheological characteristics (i.e., shear stress, 7 , vs correcting for magnification, the average intercept size was
multiplied by a statistical factor (1.56) to obtain the average
grain size.'
Manuscript No. 199591. Received September 15, 1987; approved May 3, 1988.
Presented at the 1st International Conference on Ceramic Powder Processmg
Science, Orlando, FL, November 2, 1987 (Basic Science Division, Poster 111. Results and Discussion
No. 2-BP-87F).
Supported by the National Science Foundation, Division of Materials Research, ( I ) Powder Characterization
Ceramics and Electronic Materials program (DMR-8410494 and DMR-8451916) The area-basis median particle size determined by the centrifu-
and by the following NSF-PYI sponsors: E. I. du Pont de Nemours & Co., General
Eltctric Co., General Motors Corp., and Sohio Engineered Materials Co. gal sedimentation method was 0.10 pm. When the data are trans-
Member, the American Ceramic Society. formed to a number basis, the median diameter is 0.07 p m . The
*AKP-50, Sumitomo Chemical America, Inc., New York, NY.
'Model W-375, Heat Systems-Ultrasonics, Inc., Farmingdale, NY. number-basis average diameter determined from SEM measure-
$Model 0s-7, Quantachrome Corp., Syosset, NY. ments was 0.06 p m . The powder specific surface area was
$Model CAPA-700, Horiba Instruments, Inc., Irvine, CA.
"Model JSM-35CF, Japan Electron Optics Co., Ltd., Tokyo, Japan. 23.8 m2/g, which gives an equivalent surface area diameter of
**Model APD 3720, Philips Electronic Instruments Co., Mt. Vernon, NY. 0.06 p m . The good agreement between the equivalent surface
"Model RV-100/CV-100, Haake, Inc., Saddle Brook, NJ. area diameter and the other particle diameters indicates that the
"Mark 11, Rank Brothers, Cambridge, England.
**ModelSP-100, Quantachrome Corp., Syosset, NY. powder is essentially aggregate-free.
841
842 Journal of the American Ceramic Society-Eh and Sacks Vol. 71, No. 10
120
tions give repulsive energy maxima in the range 38kT to 106kT.
t I (In these calculations, the effective Hamaker constant for A1203
l o o t in water was assumed to be 4 X lo-,’ J.” Calculations were
- ,0.1 vm A12 O3 (38vol% Solids) made using the various zeta potentials (77 to 103 mV) and Kr
80 - values (1.7 to 9.3) discussed earlier for suspensions with and
- without KCl additions. It should be emphasized, however, that
rheology and slip casting experiments were carried out using sus-
60 -
-
pensions without KCl, i.e., under conditions in which the inter-
particle repulsive force was maximized.) Another indication that
40 - the -0.1-pm A1203 suspension in Fig. 1 is well dispersed is the
- observation that slip cast samples have a very high packing den-
20 - sity (-69 % p , ) . This is discussed in more detail below in the
- section on green-body characterization.
01 I I I I I IIII 1 I I I IIIII I I I I 1 Ill1 The shear thinning behavior in Fig. 1 can be attributed to two
10 100 1000
phenomena that become increasingly important as the particle
SHEAR RATE (s-’) size decreases -Brownian motion and electroviscous effects.
The effect of Brownian motion on viscosity was clearly demon-
Fig. 1. Plots of viscosity vs shear rate for suspensions prepared with the strated by Krieger et al.”-I4 using “neutrally stable” suspensions
indicated solids concentration and particle size of A1203. of monodisperse latex particles. (In these experiments, the sur-
face charge and ionic strength were adjusted to minimize the ef-
fects of interparticle forces on the suspension rheology.) Below
-20 vol% solids concentration, Krieger et ~ l . ‘ ~observed
.‘~ New-
X-ray diffraction shows that the powder is predominantly tonian flow behavior over a wide range of shear rates. At higher
a-alumina (a-A1203), although a trace amount of &alumina solids concentrations, suspensions showed shear thinning behav-
(O-AI,O,) was detected.6 (The amount was estimated to be -1% ior with low-shear and high-shear Newtonian limits. In the
or less. The most intense 6-Al2O3peaks were approximately present study, the -0.1-pm A1203suspension also appears to be
the same height as those a-AlZ03peaks having intensities of 1% approaching low- and high-shear limiting viscosities (Fig. l), al-
or less of the most intense a peak.’) though measurements over a broader range of shear rates are
(2) Suspension Characterization needed to confirm this. In the study by Krieger et a1.,13.14 the
Electrophoresis measurements on a sample with 3.2 X 10-3M low-shear viscosity was approximately twice the high-shear vis-
KCl and pH =4 gave an average mobility of 3.9 X cmz* cosity for suspensions with 50 vol% solids and -0.15-pm
s-’ . V-’. In order to calculate the zeta potential from the mobil- particle size. The difference between the two limiting viscosity
ity, we need the value of K r , where K is the Debye-Huckel values decreased rapidly with decreasing solids loading and
parameter and r is the particle radius. Since the ionic strength increasing particle size. In Fig. 1, the viscosity for -0.1-pm
was known, the value of K was determined readily (1.9 X A1203 suspension (with only 38 vol% solids) ranges from
lo6 cm-I). Two different particle radii were used in the zeta po- -83 mPa.s at 10 s-’ to -27 mPa.s at 1000 s-I, i.e., an ap-
tential calculation: the area-basis median radius from the centrifu- proximately threefold difference. These results indicate that the
gal sedimentation measurement (0.05 pm) and the number-basis shear thinning behavior cannot be attributed entirely to the effect
average radius from the microscopy measurement (0.03 pm). of Brownian motion, especially since the limiting viscosity at
Both radii were used in order to provide an upper and lower limit high shear rate is probably much lower than the value at 1000 s-’.
for the calculations. Thus, the K r values used were 9.3 and 5.6, The other factor which accounts for the shear thinning behav-
respectively. Using the treatment by Wiersema er at.,4 the calcu- ior in Fig. 1 is the second electroviscous effect.” Electroviscous
lated zeta potentials are 77 and 87 mV, respectively. effects become increasingly important as the zeta potential in-
A sli htl hi her electrophoretic mobility (4.0 X cm2. creases, the surface charge density increases, the ionic strength
g y g .
s-’ . V - ) was deterrmned for a suspension with no KCI addition. decreases, and the particle size decreases. I6-I8 Electroviscous
The Debye-Huckel parameter was crudely estimated by assum- contributions to the viscosity can be much larger than effects due
ing that the ionic strength was controlled by the nitric acid addi- to Brownian motion; changes in viscosity of several orders of
tion which was used to adjust the pH. (This approach neglects magnitude between low and high shear rates are p ~ s s i b l e . ” ~ ~ *
other ions of unknown concentration, such as aluminum-bearing Yield stresses are also observed if the ionic strength is decreased
ions that result from the limited solubility of the A1203 particles ~ufficiently.~’~’~
Based on the limited data collected in the present
in water at pH =4. Therefore, the calculated zeta potentials study, we cannot quantitatively assess the relative contributions
probably are overestimates.) Based on a nitric acid concentration of electroviscous effects and Brownian motion to the viscosity of
of -3 X 10-4M, the Kr values for the two particle radii estimates the -0. I-pm A1203suspension. This can be done in a relatively
(0.05 and 0.03 pm) are 2.8 and 1.7, respectively. This gives zeta straightforward fashion, however, by following procedures de-
potentials of 103 and 100 mV, respectively. scribed by Krieger et ~ l . ‘ ~ ~ ‘ ~
Figure 1 shows a plot of viscosity vs shear rate for a suspen- As noted earlier, Brownian motion and electroviscous contri-
sion prepared at pH =4 with -38 vol% of the -0.1-pm AI2O3 butions to the viscosity become increasingly significant as the
particles. The suspension shows shear thinning behavior; i.e., the particle size decreases. Figure 1 also shows viscosity vs shear
viscosity decreases with increasing shear rate. Althou h this be- rate data which we obtained previously19 for a pH =4.0 sus-
havior can often be attributed to particle f l o c ~ u l a t i o nalterna-
~~~ pension prepared with 45 vol% A1203 of median particle size
tive explanations must be considered in this case because the ~ 0 .pm.4 As expected, suspensions are less shear thinning and
electrokinetic results indicate that the suspension should be well viscosities are much lower compared to the -0.1-pm A1203sus-
dispersed. The high zeta potential and moderate ionic strength in pension. This occurs despite the higher solids loading in the
this suspension result in relatively strong repulsive forces be- -0.4-pm A1203suspension.
tween the particles. For example, DLVO’’ pair potential calcula- The results in Fig. 1 illustrate one of the practical problems as-
sociated with processing ultrafine particles. As the particle size
~
decreases, it becomes increasingly difficult to prepare suspen-
?‘This effect is not limited to electrostatically stabilized suspensions, such as used sions with high solids loading while still maintaining suitable
in the present study. In sterically stabilized systems, the ratio o f the thickness of the rheological properties for processing operations (e.g., casting).nq
adsorbed polymer layer to the radius of the particle tends to increase as the particle
size decreases. Thus, the “effective” volume ( i t . , hydrodynamic volume) fraction For example, suspensions with 54 vol% solids were easily pre-
of solids in suspension tends to increase with decreasing particle size. pared and slip cast using the -0.4-pm A1203.” In contrast, it is
October 1988 Low-Temperature Sintering of Aluminum Oxide 843
I
I
0
r 60,

-
0 SUSPENSION PROCESSED a-AI2O3 (This Study)

o a-AI2O3- SEEDED BOEHMITE (Ref. 3)

100 200 300 400

SlNTERlNG TIME (min)

PORE RADIUS (nm)

Fig. 3. Plots of relative density vs sintering time at 1150°C for suspension-

Fig. 2. Plot of specific pore volume frequency function vs pore radius processed samples in this study and boehmite sol-gel samples from
obtained by mercury porosimetry for a sample prepared by slip casting Ref. 3 .
the -0.1-pm A1,0, suspension.

difficult to prepare castable suspensions beyond -45 vol% with graph of a slip cast sample which was sintered for 2 h at 1150°C.
the -0.1-pm A1203used in this study. cross-sectioned and polished, and thermally etched for 4 h at
1150°C. The average grain size is only 0.25 p m .
(3) Green-Body Characterization The enhanced densification for the slip cast samples can be at-
Slip cast samples were prepared using suspensions with tributed, at least in part, to the improved green microstructure;
pH =4. Previous studies with concentrated (40 vol% solids) sus- i.e., the particle packing density is higher and the microstructure
pensions of -0.5-pm A1203 showed that the viscosity was is more homogeneous. In the gel samples, the “packing units” are
minimized and the slip cast green density was maximized for actually boehmite aggregates. The size of these “packing units’’ is
suspensions prepared at pH -4.21322At higher pH values, suspen- on the order of -20 to 100 nm, but specific surface area measure-
sions were less stable against flocculation because of lower zeta ments indicate that the primary particle, or crystallite, size is on
potentials. At lower pH values, suspension stability was dimin- the order of -5 to 10 nm.’,’ These porous aggregates are proba-
ished by higher ionic strengths which resulted from the large acid bly the main reason for the relatively low green density of the
additions needed to adjust the pH. dried gels (-50% to 60%). In contrast, the high green density
The green density of slip cast samples was determined by two achieved in slip cast samples (-69%) indicates that aggregates
methods -mercury porosimetry and Archimedes displacement. were effectively eliminated during the powder preparation (i.e.,
In the latter technique, compacts had to be calcined at 500°C sedimentation) process.
for 1 h in order to have sufficient handling strength for the The sol-gel samples also develop heterogeneities during heat
measurement. A relative density of -70% was obtained by the treatment. Boehmite undergoes a decomposition reaction (i.e.,
Archimedes method, while mercury porosimetry gave a value of dehydroxylation) and a succession of phase transformation^:^
-69%. These high green densities are consistent with our earlier
boehmite -+ y- -+ 8- + 13- + a-A1203
argument that the -0.1-pm A1203suspension discussed in Fig. 1
is well dispersed. The values are also similar to previous results Thus, the dried gel undergoes substantial changes in volume (be-
(-66 %p,) obtained by slip casting a coarser fraction (-0.15-pm fore sintering) due to weight loss (i.e., water removal) and
median size) of the same a-A1203powder.” (The slightly higher increases in true density as the phase transformations occur.
density values in the present case may reflect more complete
elimination of hard agglomerates by the sedimentation process .)
A similar green density (-70 % p l ) was also obtained by slip
casting the -0.4-pm A1203suspension discussed in Fig. 1.19
Figure 2 shows the pore size distribution (obtained by mercury
porosimetry) for the slip cast sample with -0.1-pm A1203.The
median pore radius is only -10 nm. In addition, almost all the
pore radii are less than 20 nm, indicating that particle packing in
the compacts is very homogeneous.
It should be noted that one practical difficulty associated with
slip casting of extremely fine particles is that consolidation rates
are relatively slow. (Liquid filtration rates through the cast layer
decrease rapidly with decreasing particle size .) Casting rates can
be enhanced, however, by applying pressure during consolidation
(e.g., pressure filtration or centrifugal casting).
(4) Sintering
Figure 3 shows plots of relative density vs sintering time at
1150°C for the slip cast samples prepared in this study and the
a-Al,O,-seeded boehmite samples prepared by Messing et d 3It
is evident that the slip cast samples have a considerably higher
densification rate. In addition, these samples reach a higher “final” Fig. 4. Scanning electron micrograph of a polished section of a
density, i.e., 3.95 5 0.01 g/cm3 (-99.2 % p , ) after 2 h and suspension-processed sample which was sintered for 2 h at 1150°C and
3.97 2 0.01 (-99.7 % p , ) after 6 h. Figure 4 shows a micro- etched for 4 h at 1150°C.
844 Journal of the American Ceramic Society--Yeh and Sacks Vol. 71, No. 10

Differential shrinkages arise during these volume changes due to of ultrafine particles (i.e., smaller than used in the present study)
local variations in the kinetics of the decomposition reaction should focus on improving our understanding of the factors
and the phase transformations. Thus, Messing et al.2z3observed which determine the magnitude of the interparticle forces.
“clusters” of -0.05- to 0.10-pm particles and “cracklike struc-
tures” in samples which were transformed to a-A120,. These
defects should affect densification the same way as powder ag- Acknowledgments: We thank P. Fernando and M. Pohl of Horiba In-
glomerates; particles in the regions of locally dense packing tend struments Inc. for providing the particle size analysis by the centrifugal sedimenta-
tion method.
to sinter rapidly, whereas large voids “open up” in the regions of
more porous packing. 23,24 The latter development is particularly
detrimental since these porous regions have much lower densifi-
cation rates. Hence, Messing et al., found that the “cracklike References
structures” persisted to relative densities as high as 99%. ‘M. Kumagai and G . L. Messing, “Enhanced Densification of Boehmite Sol-
Gels by a-Alumina Seeding,” J. Am. Ceram. Soc., 67 [ I l l C-230-C-231 (1984).
Another factor which could contribute to the different den- 2M. Kumagai and G. L. Messing, “Controlled Transformation and Sintering of a
sification rates observed in Fig. 3 is differences in particle size. Boehmite Sol-Gel by @-Alumina Seeding,” J. Am. Ceram. Soc.. 68 [91 500-505
Unfortunately, this factor cannot be evaluated readily since quan- (1985).
’G. L. Messing, M. Kumagai, R. A. Shelleman, and J. L. McArdle, “Seeded
titative particle size measurements were not available for the gel Transformations for Microstmctural Control in Ceramics”; pp. 259-71 in Science of
samples. However, Messing et uZ.~., did report that the a-A1203 Ceramic Chemical Processing. Edited by L. L. Hench and D. R. Ulrich. Wiley,
New York, 1986.
particle size was -0.05 to 0.10 p m for gel samples in which the 4P. H. Wiersema, A. L. Loeb, and J. Th. G. Overbeek, “Calculation of the Elec-
0 + a transformation was nearly complete. The particles used trophoretic Mobility of a Spherical Colloid Particle,” J. Colloid Interface Sci., 22,
in the present study are in the same size range, suggesting that 78-99 (1966).
‘M. I . Mendelson, “Average Grain Size in Polycrystalline Ceramics,” J . Am.
the enhanced densification for the slip cast samples is due almost Ceram. Soc., 52 [ 8 ] 443-46 (1969).
exclusively to the higher green density and more homogeneous bPowder Diffraction File, Card No. 11-517. Joint Committee on Powder Diffrac-
tion Standards, Swarthmore, PA.
green microstructure. ’Powder Diffraction File, Card Nos. 10.173, 11-661. Joint Committee on Powder
The low sintering temperature and small grain size obtained in Diffraction Standards, Swartbmore, PA.
the present study with a-A1203are similar to results reported by “. D. Sacks, “Rheological Science in Ceramic Processing”; pp. 522-38 in Sci-
ence of Ceramic Chemical Processing. Edited by L. L. Hench and D. R. Ulrich.
R h ~ d e using
s ~ ~ high green density (-74 % p t ) Y20,-stabilized Wiley, New York, 1986.
ZrO, powder compacts prepared from well-dispersed suspensions 9M.D. Sacks, C. S. Khadilkar, G . W. Scheiffele, A. V. Shenoy, J . H. Dow, and
R. S. Sheu, “Dispersion and Rheology in Ceramic Processing”; pp. 495-516 in Ad-
of fine, unagglomerated particles. In the latter study, samples vances in Ceramics, Vol. 21, Ceramic Powder Science. Edited by G. L. Messing,
were sintered at 1100°C (1 h) to a relative density of 99.5% and K. S . Mazdiyasni, 1. W. McCauley, and R. A. Haber. American Ceramic Society,
Westerville, OH, 1987.
average grain size of 0.2 pm. “J. Th. G. Overbeek, “The Interaction Between Colloid Particles”; pp. 245-77 in
Colloid Science. Edited by H. R. Kruyt. Elsevier, Amsterdam, 1952.
“I. Visser, “On Hamaker Constants: A Comparison Between Hamaker Constants
IV. Summary and Concluding Remarks and Lifshitz-Van Der Waals Constants,” Adv. Colloid Interface Sci., 3 , 331-63
(1972).
An agglomerate-free, fine-sized A1203suspension was pre- I’M. E. Woods and I. M. Krieger, “Rheological Studies on Dispersions of Uni-
form Colloidal Soheres: I. Aoueous Disuersions in Steadv Shear Flow.” J . Colloid.
pared by sedimentation of a commercially available powder. Interpace Sci., 34 [I] 91-99 (1970).
Good dispersion was achieved by adjusting the pH to =4, so that I3Y.S. Papir and I. M. Krieger, “Rheological Studies on Dispersions of Uniform
Colloidal Spheres: 11. Dispersions in Nonaqueous Media,” J . Colloid Interface Sci.,
the zeta potential was >75 mV and a moderate ionic strength 34 { I ] 126-30 (1970).
was maintained. Homogeneous green bodies with -69% relative ’ I . M. Krieger, “Rheology of Monodisperse Latices,” Adv. Colloid Interface
Sci., 3, 111-36 (1972).
density and -10-nm median pore radius could be prepared ”B. E. Conway and A. Dobry-Duclaux, “Viscosity of Suspensions of Electrically
by slip casting the well-dispersed suspensions. Samples sintered Charged Particles and Solutions of Polymeric Electrolytes”; pp. 83- 120 in Rheol-
at 1150°C to a relative density >99.5% and a grain size of ogy: Theory and Applications, Vol. 3. Edited by F. R. Eirich. Academic Press, New
York. 1960.
0.25 pm. Densification was enhanced and the average grain size “J’. Stone-Masui and A. Watillon, “Electroviscous Effects in Dispersions of
was smaller compared to sol-gel processed samples. Monodisperse Polystyrene Latices,” J. Colloid Interface S c i . , 28 [2] 187-202
( 1968).
Presumably, the sintering temperature for densifying A1203 I7J. G. Brodnyan and E. L. Kelly, “The Effect of Electrolyte Content of Synthetic
could be further reduced if particles with smaller size were avail- Latex Flow Behavior,” J. Colloid. Sci., 20, 7-19 (1965).
”1. M. Krieger and M. Equiluz, “The Second Electroviscous Effect in Polymer
able and if a homogeneous, high relative density green structure Latices,” Trans. Soc. Rheol., 20 [ I ] 29-45 (1976).
was maintained after consolidation. *** When the particles are I9T.-S. Yeh and M. D. Sacks, ”Effects of Particle Size Dishibution and Particle
smaller than the size used in the present study (i.e., 50.05 to Packing in the Sintering of Alumina,” Paper 2-BP-87F. presented at the First Inter-
national Conference on Ceramic Powder Processing Science, Orlando, FL,Nov. 2,
0.10 pm), the shape-forming method is generally referred to as a 1987.
“sol-gel” process. Xerogels formed by sol-gel processing invari- 2nM.D . Sacks, H . W. Lee, and 0. E . Rojas, “Suspension Processing of
AI,O,/SiC Whisker Composites,” J. Am. Ceram. Soc., 71 [ 5 ] 370-79 (1988).
ably have much lower relative densities than the values obtained 2’M.D. Sacks, H. W. Lee, and 0 .E. Rojas, “Suspension Processing of S i c
in the present study. This suggests that the magnitude of the inter- Whisker-Reinforced Ceramic Composites,” Paper 22-BP-87F, presented at the First
International Conference on Ceramic Powder Processing Science, Orlando, FL,
particle repulsive forces that o rate in the sol-gel systems are in- Nov. 2, 1987; pp. 440-51 in Ceramic Transactions, Vol. 1, Ceramic Powder
adequate. Several researchersg2’ have demonstrated that particle Science IIA. Edited by G. L. Messing, E. R. Fuller, and H. Hausner. American
consolidation under conditions of low repulsive force (e.g., rapid Ceramic Society, Westerville, OH. 1988.
’?M. D . Sacks, H . W. Lee, and 0 . E. Rojas, “Pressureless Sintering of S i c
coagulation) tends to produce more porous and highly ramified Whisker-Reinforced Composites,’’ Paper 38-C-88F, presented at the 12th Annual
structures (with lower fractat dimension), while more compact Conference on Composites and Advanced Ceramics, Engineering Ceramics Division
Meeting, Cocoa Beach, FL, Jan. 19, 1988; to be published in Ceram. Eng. Sci.
structures (with higher fractal dimension) tend to form under con- Proc.
ditions with higher interparticle repulsive forces (e.g., reaction- ,’M. D. Sacks and J. A. Pask, “Sintering of Mullite-Containing Materials: 11, Ef-
fect of Agglomeration,” J . Am. Ceram. Soc., 65 [2] 70-72 (1982).
limited coagulation). Thus, future investigations into processing ”M.D. Sacks and T.-Y. Tseng, “Preparation of SiO, Glass from Model Powder
Compacts: 11, Sintering,” J. Am. Ceram. Soc., 67 [S] 532-37 (1984).
”W. H. Rhodes, “Agglomerate and Particle Size Effects on Sintering Yttria-
Stabilized Zirconia,”J. Am. Ceram. Soc., 64 [ l ] 19-22 (1981).
?‘D. A. Weitz, J. S. Huang, M. Y. Lin, and 1. Sung, ”Limits of the Fractal
Dimension for Irreversible Kinetic Aggregation of Gold Colloids,” Phys. Rev. Lett.,
***This assumes that shrinkage-producing mechanisms, i.e., grain-boundary 54 [I31 141619 (1985).
and/or volume diffusion, still dominate as the particle size is reduced and the sinter- 27D. W. Schaefer and K. D. Keefer, “Fractal Aspects of Ceramic Synthesis”;
ing temperature is lowered. However, surface diffusion tends to become more im- pp. 277-88 in Materials Research Society Symposia Proceedings, Vol. 73, Better
portant under these conditions and this can lead to particle coarsening, instead of Ceramics Through Chemistry, 11. Edited by C. J. Brinker, D. E. Clark, and D. R.
densification. Ulrich. North-Holland, New York, 1986. 0