J. Am. Ceram. SOC.

, 71 [I21 1062-67 (1988)

Processing of Highly Concentrated Aqueous a-Alumina

Suspensions Stabilized with Polyelectrolytes
JOSEPH CESARANO III* and ILHAN A. AKSAY*
Department of Materials Science and Engineering, and Advanced Materials Technology Program, Washington Technology Center,
University of Washington, Seattle, Washington 98195

Stability and rheology of aqueous aY-A1203 suspensions with brief discussion pertains to the particular aqueous system of alu-
poly(methacry1ic acid) and poly(acry1ic acid) polyelectrolytes minum oxide particles and PMAA. I
were studied as a function of pH, solids loading, and molecu- The structure of PMAA is shown in Fig. 1. For PMAA alone,
lar weight. Past work has found polyelectrolyte-stabilized sus- cach molecule has many carboxylic acid sites or functional
pensions to be fairly pH independent at low (e.g., 20 ~ 0 1 % ) groups, and, depending on the solvent conditions (i.e.. pH and
solids loadings. However, we now show that the effective pH ionic strength), the fraction of functional groups that are dissoci-
range to provide dispersed and fluid suspensions narrows as ated (i.e., COO-) and those that are nondissociated (i.e., COOH)
the concentration of solids increases as related to interparticle will vary. As the fraction dissociated ( a ) increases from =O to
forces. At high solids levels, the presence of excess polymer in = 1 .O, the polymer surface charge varies from relatively neutral
solution has detrimental effects on stability, which increases to highly negative. In general, for PMAA and other polyelec-
as the molecular weight increases. Finally, with knowledge trolytes which have carboxylic acid groups as the functional
of these concepts, highly concentrated fluid suspensions of groups, the dissociation and negative charge characteristics of the
60 vol% A1,03 (>85 wt%) with submicrometer-size particles polymer increases as the pH increase^.^,^ For PMAA at pH 2 8 . 5 ,
have been processed. Higher consolidated densities and the molecules are effectively negative with a = 1. In this condi-
reduced sintering temperatures are attained when compared tion, experimental evidence shows that the PMAA molecules are
with more conventional processing techniques. in the form of relatively large, expanded, random coils in solu-
t i ~ nThis
. ~ results from electrostatic repulsion between the nega-
I. Introduction tively charged surface sites. As pH is decreased, the number of
negatively charged sites is continually decreased until the PMAA

P OLYELECTROLYTES are widely used in industrial applications

to prepare highly concentrated (>50 ~ 0 1 % )ceramic suspen-
sions which are subsequently fabricated into dcnsc components
is effectively neutral near pH 3.4 and a + 0. In this condition,
the PMAA chains approach insolubility and form relatively small
coils or c ~ u m p s . ~
by sintering. Although a variety of polyelectrolytes are commer- In general, for oxides, the relative charge density ((T) on the
cially available and arc used effectively in the preparation of sus- surface of the particles also varies with pH, but in a different
pensions, their role in colloidal stabilization has not been clearly manner from the charging of anionic polyelectrolytes. For ox-
understood. ides, at every pH, there are large numbers of negative, neutral,
In this paper, we intend to provide a clearer understanding of and positive surface sites. The cr value gives the overall net
particle dispersion with adsorbed polyelectrolytes and relate it to charge density. At low pH values, positive sites are in the vast
the chemistries of the suspended-particle surfaces and polyelectro- majority, and, therefore, (T is positive. At high pH values, nega-
lytes and interparticle forces. Aqueous oxide suspensions will be tive sites are in the majority, and cr is negative. At the zero point
considered as a general example. The particular emphasis will be of charge (zpc), which occurs when the number of positive sites
on the dispersion of a-Al2O3suspensions with poly(methacry1ic equals the number of negative sites, CT = 0. For the relatively
acid) (PMAA) and poly(acry1ic acid) (PAA). pure a-A120? powders used in this study, the zpc is at pH 4 . 7 . ’
It is also an objective of this paper to describe the methodology The comparison of the charging behavior of PMAA and Al2O3
of using polyelectrolytc stabilization for the processing of very with pH leads to an obvious conclusion that the saturation limit
highly concentrated (>60 ~ 0 1 % )aqueous ceramic suspensions. of PMAA adsorption on Al,03 should be significantly affected
This was studied in model experiments with submicrometer-sized by pH. Figure 2 is a plot of the amount of PMAA adsorbed at the
alumina which determined the guidelines for the selection of a saturation levels versus pH and simultaneously the amount of
proper pH range and related rheological behavior to molecular PMAA necessary for suspension stability. The amount of PMAA
weight. Finally, the sintering behavior of compacts prepared by which must be adsorbed to achieve a stable suspension increases
these techniques is compared with more conventional methods with decreasing pH. Regions above the solid curve are stable and
of processing. dispersed. Regions below the curve are unstable and show an
onset of flocculation. The regions below pH 4 are discussed in
11. Background Information Ref. 1 and thus will not be discussed in this paper.
Figure 2 was developed for 20 vol% A1203suspensions; there-
For detailed information concerning polyelectrolytes, surface fore, in order to expand the concepts for different solids concen-
charging of oxides, and rclatcd adsorption behavior of polyelec- trations, pH, and polyelectrolyte molecular wcight and to use
trolytes on oxides, the reader is referred to Ref. 1. The following these concepts for the processing of these suspensions, more
experiments were conducted using polyelectrolytes and col-
loidal a-A1203.
Manutcript No. 199880. Received November 8. 1987; approved April 19. 1Y8X.
Presented at the 87th Annual Meeting of the American Ceramic Society. Cincin-
nati, OH, May 7, 1985 (Basic Science Division, Paper No. 95-B-85). 111. Experimental Procedure
The work summarized in the Background Information \ection of this paper was
sponsored by the Air Force Oftice of Scientific Research (AFOSR) and the Ad- ( I ) Materials and Chemicals
vanced Research Projects Agency of the Department of Defense and was monitored The ceramics used in this study were of very-high-purity
by rhe AFOSK under Grant No AFOSR-083-0375. The work discussed in the
main body of this paper was sponsored by the Oak Ridge National Laboratory under (>99.995%), submicrometer-size, and relatively monodisperse
supontract No. 19X-27458C. (0.2 to 1 .O p m ) a-AI2O3powders (AKP 20 and AKP 30).* The
Member, the American Ceramic Society
*Sumitorno Chemical America, New York. average diameter measured by an X-ray absorptioniscdimentation
-Model 5000 Sedigrdph, Micromeritics Instrument Corp., Norcross. GA. techniquet was 0.52 p m for AKP 20 and 0.41 p m for AKP 30.
1062
December 1988 Processing Concentrated Aqueous a-Alumina Suspensions Stabilized with Polyelectrolytes 1063

PMAA PA A I ' I ' I ' 1 -

o based on adsorption data

06 n
3
Fig. 1. Schematic of PMAA and PAA polymer U
segments.

03
The average surface area as measured by standard BET N2 ad-
sorption' was 4.5 m2/g for AKP 20 and 6.8 m2/g for AKP 30.
The polyelectrolytes used were the Na' salt of poly(methacry1ic
acid) (PMAA-Na) with an average molecular weight (MW) of
15000 and PAA of various molecular weights (1800, 5000, and 4
50000).s These polyelectrolyte structures are shown in Fig. 1 . PH
The water used was distilled and deionized. The pH was ad-
Fig. 2. Stability map showing the amount of adsorbed PMAA required
justed with standardized analytical-grade HC1 and NaOH solu- to form stable suspensions of 20 vol% AKP 30 a-AlzO, a5 a function
tions (1 .ON and 0. IN). Analytical-grade NaCl was used to adjust of pH.
the ionic strength as desired.
(2) Experimental Methods
( A ) ViscosiQ Measurements: Viscosity measurements were librium supernatant to keep polymer adsorption constant, Prior to
conducted with a rotary viscometer7 with a small sample cham- taking the measurements, the new dilute suspensions were very
ber. To determine the relative magnitude of viscosity and whether briefly ultrasonicated and magnetically stirred to ensure that only
the suspensions were flocculating or stabilized, the following singlet particles wcre measured. At least 10 measurements were
rheological measurement method was used: completed for each sample to ensure accuracy.
(i) Each suspension was initially sheared at a relatively high Interparticle repulsive potentials (V,) were calculated from
shear rate of 93 s-'. This ensured that permanent particle clusters Deryagin-Landau and Verwey-Overbeek (DLVO) theory' ac-
(if any) were dispersed. cording to
(ii) The shear rate was then decreased every 30 s in eight incre-
ments to 0.46 s-' and the viscosities recorded at each shear rate. V, = -2maT; In [ I - exp(-KH)] (1)
(iii) The suspensions were then undisturbed for 10 min so where E and a are constant, H is the particle separation distance,
that nucleation and growth of particle clusters could occur. and T,can be approximated by 5 to yield
(iv) The shear ratc was then increased every 30 s in eight in-
crements to 93 s-' and the viscosities recorded again at each CV, = L2 In [ I - cxp(-d)] (2)
shear rate.
Large differences between the initial decreasing-rate sweep and where c is a constant. For every sample studied, the van der Waals
the later increasing-rate sweep were assumed to be due to floccu- attractive potentials can be assumed to be equivalent; therefore, it
lation (i.e., the nucleation and growth of particle clusters). The is only necessary to compare the values of V, for qualitative argu-
magnitude of the viscosities was taken as a relative measure of ments concerning interparticle potentials since at any distance
the fluidity of the suspensions as related to the particle interaction a higher V, indicates a relatively larger electrostatic repulsion
energy and the degree of particle clustering. between particles. For 40 vol% dispersed suspensions of 0.41-
( B ) Zeta Potential and Interparticle Repulsive Potential p m particles, the average separation distance is -70 nm (700 A).
Calculations: Zeta potential (5) was calculated by using the Therefore, in order to only examine the effects of V, when inter-
measured electrophoretic mobility** and Henry's equation5 with particle interactions become important, comparison? of V, are
Overbeek's correction, if necessary.6 This correction takes into made at a particle separation distance of 10 nm (100 A).
account electrophoretic retardation due to ions in the mobile part ( C ) Processing of Suspensions ,for Sintering: AKP 30
of the double layer. This effect is dependent on the value of KU , a-A1203powder was dry-pressed, slip cast, and simply air cast
where K is the inverse screening length and a is the particle (dried) in plastic molds. The dry-pressed pellets were approxi-
diameter. In general, as the ionic strength of a suspension in- mately 2.54 cm in diameter and pressed biaxially at 51.8 MPa.
creases, KU increases. For the purpose of this paper, the actual All suspensions were prepared with commercially available PAA-
polymer chains were not included in the calculation of ~ ahow- ; NH: (MW of 5000)" (0.22% (dry-weight basis (dwb))) and
ever, their associated ions were included in the KU determination. the pH adjusted to -8.8 with small amounts of citric acid
This is based on the assumption that the relatively large bulk an- (0.065% (dwb)). The slip-cast samples were prepared from
ionic polymer molecules do not appreciably contribute to specific 50 vol% suspensions cast between two slabs of plaster. The air-
conductivity in such low concentrations. These measurements cast samples were made by pouring 5 mL of 40 vol% and highly
can only be completed on very dilute suspensions. Therefore, it concentrated ( ~ 6 ~00 1 % )suspensions into plastic molds and
was very important that the suspensions were centrifuged and the allowing them to dry. All of the samples were then sintered in
cquilibrium supernatant carefully decanted into a beaker. Then a air at 990" to 1500°C. The samples were held at the maxi-
portion of the sediment was removed and readded to the equi- mum temperature for 1 h. The heating and cooling rates were
300"C/h. The sintered densities were measured by a standard
alcohol immersion method.

'Quantisorb, Quantachroine Co., Svosset. NY.

*Polysciences,~Inc.,Pittsburgh. PA: IV. Results and Discussion
'Model RVTD Dlgital Viscomerer, Brookfield Engineering Laboratories, Inc.,
Stoughton, MA. (1) Viscosity versus pH
**Micro-Particle Electrophoresis Apparatus Mark 11, Rank Brothers. Bottisham, Figure 3 shows the effect of pH on viscosity for stabilized sus-
Camhridge, England.
' R , T. Vanderbilt Co , Norwalk. CT. pensions at various solids loading. Using Fig. 2 for reprcscnta-
1064 Journal of the American Ceramic Society-Cesarano and Aksay Vol. 71, No. 12

For pH <8, the problems related to excess polyelectrolyte in

solution can be avoided because of the high-affinity-type adsorp-
tion behavior.' However, the viscosity increases shown in Fig. 3
occur because of two other related phenomena: as the pH de-
creases, the amount of polyelectrolyte necessary to provide
stabilization increases (Fig. 2), while at the same time the poly-
0.6
electrolyte chains become less dissociated and behave more like
neutral polymers.' This results in the AI2O3particles being cov-
ered with increasing amounts of less negatively charged polyelec-
trolytes and leads to two problems. First, with less dissociation,
the negative charge characteristics of the polyelectrolyte decrease
and the relative solubility of the polymer also decreases.' This
leads to a situation where the adsorbed polyelectrolyte has a
smaller number of negatively charged functional groups extend-
0 ing into solution and, therefore, decreases the magnitude of the
4 5 6 7 8 9 1 0
PH electrosteric repulsive barrier between particles.'.* Second, as
the amount of polyelectrolyte increases, the concentration of ions
Fig. 3. Viscosity at 9.3 sC1 versus pH for associated with the polyelectrolyte and ions due to pH adjustment
PMAA-stabilized AKP 30 a-A1203suspensions also increases and, therefore, reduces the electrostatic portion of
from 20 to 58 vol% solids concentrations. The particle repulsion due to high ionic strengths. These results are
amounts of PMAA added were enough to
provide complete surface adsorption. summarized in Table I.
These destabilizing effects become more appreciable at higher
solids concentrations since the particle repulsion necessary to
provide stability increases with increasing solids concentrations
tion, all of these suspensions are in the stable region, but still (e.g., Fig. 3).'.1° For a stable 20 vol% suspension at pH 5.5,
relatively close to the stability boundary. the volume percent solids is relatively low, and under these
In 20 vol% A1203suspensions, all of the stabilized systems conditions of polyelectrolyte dissociation and ionic strength, the
have very low viscosities and the difference between suspensions binding energy between particles is low enough to prevent the
stabilized at pH 4.5 and pH 10 is minimal. As the volume per- nucleation and growth of particle clusters and the resultant vis-
cent of solids increases, the magnitude of the viscosity also in- cosity remains low. In contrast, for a stable 40 vol% suspension
creases, whereas the effective pH range gets smaller (Fig. 3). In at pH 5.5, the binding energy between particles is low enough
all cases, the minimum viscosity occurs at pH 4 . 8 with viscos- to prevent permanent clusters, but high enough to induce a large
ity increasing for both increases and decreases in pH. Curiously. increase in viscosity. To have more fluid and stable suspensions
this pH coincides with the zpc of this AI2O3, where one might at higher solids concentrations requires adjustment to a more
expect maximum viscosities. However, upon introduction of optimum pH where the adsorbed polyelectrolyte is more fully
polyelectrolytes, the surface chemistry is governed by the ad- dissociated and ionic strengths lowered so that a more effective
sorbed polyelectrolytes and not the bare A1203surfaces. electrosteric repulsion can exist between particles.
For pH >9.0, the viscosity increase is due to the presence of These concepts will be revealed in more detail in the following
excess polymer in solution which is not adsorbed on the A1,03 section; however, it should be noted that for this system the opti-
particles. The presence of this excess polyelectrolyte is inevitable mum seems to be pH 4 . 8 . This pH is where the minimum vis-
because of the adsorption behavior of the polyelectrolytes above cosity occurs (Fig. 3) and coincides with the point at which the
pH 9.' Above pH 9, "high-affinity-type" adsorption does not polyelectrolyte is fully charged and maintains maximum solubil-
occur; therefore, in order to reach the saturation adsorption limit ity while high-affinity-type adsorption still occurs and the ionic
(and complete stabilization), there has to be an equilibrium strength is relatively low. Under these conditions, the amount of
amount of polyelectrolyte in solution. This leads to high viscosi- polyelectrolyte in solution is low and the binding energy between
ties in concentrated suspensions where the amount of water is particles is small.
small and the concentration of excess polyelectrolyte in solution Although the shapes of the viscosity curves shown in Fig. 3
becomes appreciable. The effects of excess polymer in solution are primarily due to polyelectrolyte chemistry and interparticle
will be discussed in a following section. forces (discussed in detail below) there is undoubtedly a sec-

Table I. Relation between Interparticle Force and Rheological Behavior*

8. I B 0.76 139 - 35 1.4 220-405 F

B 0.95 152 -41 1.0 60-205 F
1.2 164 -37 0.5 37-37 S
1.9 198 -33 0.1 50-70 S
8. I" 0.95 265 -26 0.0 4 15-645 F
C 1.2 27 I -27 0.0 55-82 SF
C 1.4 219 -22 0.0 50-60 S
9.5 B 0.55 100 -31 7.3 55-1 15 F
1.2 135 -32 1.4 85-87 S
1.7 163 -32 0.4 205-2 15 S
9.0 1.3 143 - 34 1.1 20-20 S
8.8 A 1.37 152 -32 0.6 20-20 S
6.5 A,C 1.2 213 - 24 0.0 45-50 S
5.3 A 1.1 265 -17 0.0 205-230 SF
4.6 A 1.3 325 -13 0.0 1300-1700 F
*40 vol% AKP-30 a-Al,O, suspensions with PMAA. 'Polymer ratio is the actual amount or polymer added to the amount needed for complete surface coverage (from Fig. 2).
Refer to text for meaning of A, B , and C. %a is the inverse Debye-Huckel length times the particle diameter. is the zeta potential. 'CV, is the relative measure of the
F, SF mean stable, flocced, slightly
electrostatic repulsion between two particles. =*q(,to q l ois the initial viscosity and viscosity after = I 0 min at a shear rate of 9.3 s - I . '3,
flocced. $'With 0.1M NaCI.
December 1988 Processing Concentrated Aqueous a-Alumina Suspensions Stabilized with Polyelectrolytes 1065

ondary effect due to changes in the adsorbed polyelectrolyte con-

figuration as a function of pH. With changes in polyelectrolyte
configurations come associated changes in the effective hydrody-
namic radii of the particles. These effects may also show minima
in viscosity and become more appreciable as the particle concen-
tration and the adsorbed polymer molecular weight increase.
However, changes in hydrodynamic radii will not be discussed in
this paper.
(2) Relating Interparticle Forces to Rheological Behavior
Table 1 shows a summary of data relating interparticle forces to J
rheological behavior €or 40 vol% AKP 30 suspensions stabilized
with PMAA. The polymer ratio is a measure of the amount of
polymer added in relation to that necessary for adsorption satura-
tion according to Fig. 2. The value of KU is a relative measure of
the ionic strength of the solution. As KQ increases, the ionic
strength increases, and, as a reference for this system, a ~a value
of 213 is equivalent to a 0 . l M NaCl solution. The cV, values are
a relative measure of the repulsive potential between particles due
solely to calculatFd electrostatic repulsion at a separation distance
of 10 nm (100 A). The r), to r),o values are a measure of the
change in viscosity over a period of =10 min at a shear rate of the resulting zeta potential of stabilizcd suspensions tabulated in
9.3 s-I, and the comments S, F, and SF are qualitative observa- Table 1 can be directly correlated to the charge on the A1203par-
tions meaning stable, flocced, and slightly flocced, respectively. ticles, amount of polyelectrolyte dissociation, and the amount of
The information in Table I can be used to provide some pre- polyelectrolyte adsorbed on the surface. This concept is repre-
liminary conclusions relating to interparticle forces. First, from sented in Fig. 4, which shows the zeta potential ((1 and relative
pH 8.8 to pH 4.6, a decrease in the magnitude of the negative number of dissociated COO- polymer sites (RNS) versus pH.
charge characteristics of the adsorbed polyelectrolyte causes less The RNS was calculated by multiplying the amount of adsorbed
repulsion between particles and induces an increase in viscosity. PMAA (Fig. 2) times the percent of COOH groups dissociated
This is shown in Table I by the data points labeled A. It is shown (measured in Ref. 1). This quantity effectively represents the
that as pH is decreased from 8.8 to 4.6, the zeta potential number of negative sites which are available to charge a particle
decreases as the polyelectrolyte begins to behave more like a surface once adsorbed. If there is no electrical interaction be-
'
neutral polymer. Also, the ionic strength (KU values) increases <
tween the charged polymer and the particle surface, then should
because of the presence of counterions. These two effects result ideally correspond directly with RNS; i.e., as the amount of ad-
in a decrease in repulsion between particles and a corresponding sorbed charged polymer increases ( increases. However, in this
increase in viscosity with the eventual loss of stability. system, the polymer has electrical interactions with the particle
Sccond, the data points labeled B in Table I indicate that the surface; therefore, Fig. 4 shows that the combination of RNS and
most dominant cause for flocculation in polyelectrolyte-stabilized the influencc of particle surface charge determines the rcsulting
suspensions is incomplete adsorption, which provides a driving zeta potential.
force for particles to cluster together in an effort to form saturated <
Figure 4 was constructed so that RNS and cross over at
monolayers. This is clearly shown in Table I in that, in all cases pH ~ 8 . since
7 this pH is very close to the zpc of the A120,,
where the polymer ratio is < 1 .O, flocculation resulted even and, therefore, the overall net charge of the A1,03 will have a
though the zeta potential and electrostatic repulsive forces are minimal contribution to the observed <. As pH decreases to 8.1,
relatively high. This is especially true for pH c8.8, where high- RNS increases and 5 becomes more negative, as expected. How-
affinity-type adsorption occurs. ' However, as expected, at <
ever, at pH 6.5, RNS is at a maximum whereas decreases in
pH >8.8, where high-affinity-type adsorption does not occur, the magnitude appreciably. From this it may be concludcd that, be-
driving force for flocculation due to incomplete adsorption is not low the zpc, the excess positive surface sites on the A1,03 begin
as strong as at lower pH. This is particularly evident for the to neutralize excess negative sites on the polyelectrolyte and the
sample at pH 9.5, which has a high cV, value due to low ionic <
effective is lower than expected. This behavior is shown to con-
strength. Even though this sample has a polymer ratio of only tinue to lower pH values; however, it is expected that for pH <3,
0.55, it exhibits a lower viscosity than the sample at pH 8.1 with <
the will be determined only by the A1203surface since RNS ap-
a polymer ratio of 0.76. <
proaches 0. However, for pH <3, the will probably still be
It may also be concluded that, once enough polymer has been lower than expected because of the physical presence of an ad-
adsorbed, there is a stabilization effect due solely to the polyelec- sorbed neutral polymer layer.
trolyte, even in the absence of appreciable repulsion due to elec- At pH values above the zpc, 6 is slightly more negative than
trostatics. This is especially apparent for the data labeled C for expected from the RNS values. This suggests that, above the zpc,
pH 8.1 with NaCl and pH 6.5 where V, -- 0, but stable fluid sus- the excess negative surface sites on the A120, become additive to
pensions resulted. Further evidence for the stabilizing effects due the negative sites on the polymer. It is emphasized that, for this
to the polyelectrolyte is that, for the same 40 vol% system in the system, ( is a strong contributing factor to the overall suspension
absence of polymer, an electrostatic value of cV, = 3 is viscosity, but not the sole determining factor. Maximum ( is at
necessary to induce comparable levels of stability and viscosity pH 8.1; however, the best suspensions in terms of stability and
as those found for PMAA-stabilized suspensions at pH 8.8 with a viscosity are observed at pH 8.6 to pH 9.0.
cV, value of 0.7."
Finally, even though Table I shows that, for 40 vol% SUS- (3) Viscosity versus Amount of Polyelectrolyte
pensions, polyelectrolyte stabilization can exist, even under Figure 5 shows the viscosity of 50 vol% AKP 20 A1203sus-
conditions of high ionic strength and low electrostatic repulsion, pensions (pH 9) versus the amount of PAA added for various
it is still necessary to have a 151 of >20 mV. This value indicates molecular weights. For each sample, the viscosity was recorded
that the polyelectrolyte is still charged and has negative soluble at 0 and 10 min at a shear rate of 9.3 s-', Increases in viscos-
functional groups extending into the solvent so that an enthalpic ity (connected by tie lines) show that flocculation is occurring in
steric stabilization mechanism can be functional with an electro- the suspensions. For each molecular weight, there is a critical
static stabilization mechanism.' amount of polymer that must be added before stabilization and
Another important aspect of polyelectrolyte stabilization is that low viscosities result. This corresponds to the saturation adsorp-
1066 Journal of the American Ceramic Society-Cesarano and Aksay Vol. 71, No. 12

--8 100 -
-
90-
>r
+
.-
rn
-
5
U
80-
-
.- 70-
c
? -
60-
f

002 0.04 0.06 0.08 0.10 0.12 0.14

PA A (%)
Fig. 6 . Sintering behavior of AKP 30 a-A120,
Fig. 5 . Viscosity versus the amount of PAA added for various processing techniques.
(dwb of AI,O,) for various molecular weights for
50 ~019%AKP 20 a-A120? suspensions at pH 9. (o),
(c),and ( A ) values are at 0 min and (+), (m), and (A)
are at 10 min. of 6.8 m'/g, this is a very low sintering temperature. In contrast,
filtration channels cause the development of excess voids in slip-
cast microstructures and yield a consolidated density of 62.1% of
tion limit of the polyelectrolytes on A1,03 and is in agreement theoretical, and higher sintering temperatures are required for the
with the previous discussion concerning the stability map in elimination of these voids.I4 As expected, the local density varia-
Fig. 2. tions and the inefficient particle packing caused by dry-pressing
Further additions of polymer past the adsorption saturation yield a consolidated density of 55.0% of theoretical and even
limit only serve to provide excess polymer in solution. As shown higher temperatures for complete void elimination.
in Fig. 5 , the rheological effects due to excess polymer are much
more drastic as the molecular weight increases. This is probably
due to two reasons. First, the viscosity of polymer solutions is V. Conclusions
proportional to the size of the individual polymer coils . 4 There-
fore, as the molecular weight increases, the viscosity of the In highly concentrated (>50 ~ 0 1 % )A1203 suspensions stabi-
aqueous solution drastically increases. Second, i t is observed lized with PMAA, a viscosity minimum is observed at pH 4 . 8 .
that, as the molecular weight increases, a more pronounced thixo- This conicides with the pH at which the polyelectrolyte is fully
tropic effect occurs due to the excess polymer. This is the result charged and maintains maximum solubility, whereas high-
of a destabilization mechanism called depletion affinity-type adsorption still occurs and the ionic strength is rela-
Basically, depletion flocculation is the result of an osmotic pres- tively low.
sure increase due to the polymer in solution, thereby forcing Below pH =8.8, the viscosity increases because of a decrease
particles together. This effect increases with increasing molecu- in the negative charge characteristics of the adsorbed polyelec-
lar weight. trolyte and a decrease in the electrosteric repulsive barrier be-
Therefore, for processing concerns, the use of lower molecular tween particles. Above pH -8.8. viscosity increases because
weight polyelectrolytes (i.e., MW of 1000 to 10000) seems to high-affinity-type adsorption behavior is lost and excess polymer
produce more "forgiving" dispersed suspensions with a wider in solution becomes appreciable. Excess polymer can also lead to
range of stability. The amount of polyelectrolyte necessary for depletion flocculation and this effect becomes greater as molecu-
stabilization increases as the molecular weight increases; this is lar weight increases.
an indication that the adsorbed layer thickness may also be in- Polyelectrolyte stabilization can exist even in the absence of
creasing with molecular weight. This may have substantial ef- appreciable electrostatic repulsive forces due to conditions of
fects in highly concentrated suspensions. relatively high ionic strength. However, even though polyelec-
trolyte stabilization can exist with low levels of electrostatic repul-
(4) Processing with Polyelectrolyte-Stabilized Suspensions sion, it is necessary that an appreciable zeta potential be present
To show the advantage of increasing the solids concentration in on the particles. The actual zeta potential built up on particles
suspension, some samples were air cast in plastic molds from 40 due to polyelectrolyte adsorption is a function of the amount of
to 60 vol% suspensions and then sintered at 1400°C for 45 min. polyelectrolyte adsorbed, the fraction of polyelectrolyte dissocia-
The suspensions were fabricated at pH 4 . 8 for optimum disper- tion, and the surface charge of the bare particle. For anionic
sion properties. Because of the reduced volume fraction of solids PMAA and A1203at pH values below the zpc of A1203, the ac-
and longer casting times for the 40 vol% suspensions, it is ex- tual zeta potential is less than expected from the adsorbed PMAA
pected that long-range particle segregation took place. l 3 This because of the cancellation effect between the negative PMAA
resulted in slightly lower green densities for the 40 vol% suspen- and positive A1203surfaces. At pH values above the zpc, there is
sions. This also became apparent after sintering, when all of the an additive effect between the negative PMAA and the negative
samples made from the 40 vol% suspensions consistently resulted surface which yields a larger zeta potential than expected from
in 3% to 6% lower sintered densities. the amount of PMAA adsorbed.
To more clearly show the possible benefits of processing with In general, by understanding the chemistries of the polyelec-
highly concentrated suspensions, two powder systems of high- trolyte and particle surface, polyelectrolyte adsorption behavior,
purity a-A1203were processed by pressing, slip-casting 50 vol% and polyelectrolyte rheological effects, one may control the vis-
suspensions, and casting 2 6 0 vol% suspensions in plastic molds. cosity and rheology for all solids levels. If it is desired to work at
The sintering behaviors and the results for AKP 30 A1203 are low solids, but still prevent problems due to segregation, then the
shown in Fig. 6 . Figure 6 shows clearly that processing from proper polyelectrolyte conditions can be tailored for a high-
highly concentrated polyelectrolyte-stabilizedsuspensions leads viscosity thixotropic suspension. If dispersion and high solids are
to higher consolidated densities and that these benefits carry over desired, then the polyelectrolyte conditions can again be tailored.
in the form of reduced sintering temperatures. The samples To date, 62 vol% (>86 wt%) AKP 30 a-AI2O3suspensions have
processed from 60 vol% suspensions had a consolidated density been fabricated with viscosities of - 4 0 0 mPa . s at 93 s-'. These
of 65.8% of theoretical and sintered to 99% of theoretical at are very low viscosities for such highly concentrated suspensions
1350°C. For an A1203powder with a relatively low surface area of 0.41-pm-size particles.
 December 1988 Reactions of Silicon-Based Ceramics in Mixed Oxidation Chlorination Environments 1067
References Academic Press, London, 1083
‘I. A. Aksay and R . Kikuchi, “Structures of Colloidal Solids”; pp. 513-21 in Sci-
‘J. Cesarano 111, I. A. Aksay, and A. Bleier, “Stability of Aqueous a-A120, Sus- ence of Ceramic Chemical Processing. Edited by L. L. Hench and D. P. Ulrich.
pensions with Poly(methacry1ic acid) Polyelectrolyte,” J. A m . Ceram. So< , 71 [4] Wiley, New York, 1986.
250-55 (1988). ’OW. Y. Shih, I. A. Aksay, and R. Kikuchi, “Reversible-Growth Model:
2 P Molyneux, Water-Soluble Synthetic Polymers: Properties and Behavior. Cluster-Cluster Aggregation with Finite Binding Energies,” P h u . Rev. A , 36 [ 101
Vol. 11; pp. 1-7. CRC Press. Boca Raton, FL, 1984. 5015-19 (1987).
‘J. PapenhuijLen, H. A. Van der Schee, and G. J. Fleer, ”Polyelectrolyte Adsorp- “.I Cesarano 111 and I. A. Aksay; unpublished data.
tion I. A New Lattice Theory,” J . Colhid Inte~anceSci., 104, 540-61 (1985). I’M. Yasrebi, J. Cesarano 111, and 1. A. Aksay, ”Effect ol Free Polyelectrolytes
‘R. Arnold and J . Th. G. Overbeek, “The Dissociation and Specific Viscosity of on Colloidal Stability”; unpublished work.
Poly(methacry1ic acid),” Rec. Truv. Chim. Pays-Bus, 69, 192-206 (1950). I3C. Han, 1. A. Aksay, and 0.J . Whittemore. “Chardcteriiation of Microstmc-
‘0.C. Henry, “The Cataphoresis of Suspended Particles. Part I-The Equation tural Evolution by Mercury Porosimetry”; pp. 339-47 in Advances in Material5
of Cataphoresis,” Pro(. R . SOC.London. A , 133, 106-29 (1931). Characterization 11. Edited bv R. L. Snvder.
, . R . A. Condrate. Sr.. and P. F. Johnson
‘P. H. Wiersema, A. L. Loeb, and I . Th. G . Overbeek, J . CoiJoid Znirtface Sci., Plenum Press, New York, 1685.
22, 78-99 (1966). “I. A. Aksay and C. H. Schilling, “Colloidal Filtration Route to Uniform Mi-
’D. J. Shaw, Introduction to Colloid and Surface Chemistry, 3d ed; pp, 183-96. crostructures”: DD. 483-91 in Ultrastructure Processine of Ceramics. Glases. and
Butterworths, London, 1980. Composites. Ehited by L. L . Hench and D . R . Ulricch. Wiley-Interscience. New
*D. H. Napper, Polymeric Stabilization of Colloidal Dispersions; pp. 138-413. York, 1984. 0

J . A m . Cerum. SOC., 71 [12] 1067-73 (1988)

Reactions of Silicon-Based Ceramics in Mixed Oxidation

Chlorination Environments
JOHN E. MARRA*’* and ERIC R. KREIDLER*
Department of Ceramic Engineering, The Ohio State University, Columbus, Ohio 432 10

NATHAN S. JACOBSON*,‘ and DENNIS S. FOX*

Lewis Research Center, National Aeronautics and Space Administration, Cleveland, Ohio 44135

The reaction of silicon-based ceramics with 2% CI2/Ar and During the operation of these systems, the materials are often
1% C12/1% to 20% 02/Ar at 950°C was studied with ther- exposed to severe oxidative and/or corrosive gases. Marine-based
mogravimetric analysis and high-pressure mass spectrometry. turbine engines ingest sodium chloride from sea salt spray,
Pure Si, S O 2 , several types of SIC, and Si3N4were examined. making the mixed oxidationichloride reactions of these materials
The primary corrosion products were SiC14(g) and SiO,(s) important. Similarly, Si-based ceramics used in heat exchanger
with smaller amounts of volatile silicon oxychlorides. The re- technology are exposed to a variety of halide gases (typically
actions appear to occur by chlorine penetration of the SiO, fluorine and chlorine) when used in the refining of Ti and Zr and
layer, and gas-phase diffusion of the silicon chlorides away ’
the remelting of aluminum. Silicon-based ceramics arc inherently
from the sample appears to be rate limiting. Pure SiO, shows unstable in oxidizing atmospheres, but obtain their oxidation re-
very little reaction with Clz. S i c with excess Si is more reac- sistance by the formation of a protective silica layer at the gas-
tive than the other materials with C12, whereas S i c with ex- solid interface. The mixed oxidationichlorination behavior of
cess carbon is more reactive than the other materials with these materials is further complicated by the volatility of the sili-
CI2/O2. Si3N4shows very little reaction with Clz. These dif- con chloride compounds.
ferences are explained on the basis of thermodynamic and The mixed oxidationichlorination behavior of metals and
microstructural factors. superalloys at high temperatures has been studied by various re-
Jacobson et have made use of a high-pressure
I. introduction mass spectrometer sampling system and conventional thermogravi-
metric analysis to study these reactions. The former analytical

T HE recent developments in advanced ceramic materials have

made these materials candidates for applications previously
restricted to metals and superalloys. In particular, Si-based ce-
technique allows the volatile corrosion products to be unambigu-
ously identified. This identification, coupled with kinetic infor-
mation, leads to a further understanding of the mechanisms of the
ramics are prime candidates for use in advanced heat engines corrosion process.
and heat exchangers because of their excellent high-temperature Although the behavior of metals in high-temperature environ-
strength and oxidation resistance. ments containing both chlorine and oxygen has been examined,
little work has been done to investigate the behavior of silicon-
based materials in these environments. Chlorine, however, is
widely used in the thermal oxidation of high-purity silicon for the
Manuscript No. 199319. Received February 5 , 1988; approved June 3, 1988.
Presented at the 89th Annual Meeting of the American Ceramic Society, Pitts- production of gate oxide devices. The chlorine acts to limit the
burgh. PA, April 27, 1988 (Basic Science Division, Poster No. 69-BP-87). amount of sodium impurities present in the growing oxide and
Support for the initial work done at The Ohio State University provided by the thereby increases the performance characteristics of the device.
Army Research Office under Contract No. DAAG29-82-K-0149; support for the
later srages of study provided by the Edward Orton Junior Ceramic Foundation. As a result of this practice, the role of chlorine in the thermal
‘Member, the American Ceramic Society. oxidation of silicon has been widely studied;”’ however, the ma-
*Now with Savannah River Laboratory, E. I . du Pont de Nemours and C o . ,
Aiken, SC 29808. jority of the previous work has focused on the characteristics of
Author to whom correspondence should be addressed. the oxide film and more specifically on chlorine incorporation