You are on page 1of 10

SYNTHESIS AND SINTERING BEHAVIOR OF A

NANOCRYSTALLINE a-ALUMINA POWDER


JI GUANG LI and XUDONG SUN{
Department of Materials Science and Engineering, Northeastern University, Shenyang, People's
Republic of China
(Received 27 July 1999; accepted 23 April 2000)
AbstractNanocrystalline a-alumina powders with a primary mean particle diameter of 10 nm were syn-
thesized from aluminum nitrate and ammonia solution using a precipitation method. The presence of am-
monium nitrate (a by-product of the precipitation reaction) in the Al(OH)
3
dry gel can reduce the
formation temperatures of g-, d-, y-, and a-Al
2
O
3
during heating. The combined eect of 5 wt% a-alumina
seed crystals, 100 nm in diameter, and 44% ammonium nitrate can reduce the y-Al
2
O
3
4a-Al
2
O
3
trans-
formation temperature from 1200 to 9008C. The a-Al
2
O
3
powder milled in anhydrous alcohol has an
agglomeration strength of 76 MPa (soft agglomerated), while the one milled in deionized water has an
agglomeration strength of 234 MPa (hard agglomerated). For both the soft and the hard agglomerated
powders initial stage sintering is controlled by grain boundary diusion, with activation energies of 365
and 492 kJ/mol, respectively. The alumina ceramic produced by sintering the soft agglomerated powder at
14008C for 2 h has a mean grain size of 0.93 mm, a mean exural strength of 700 MPa, and a fracture
toughness of 4.75 MPa m
1/2
. 7 2000 Published by Elsevier Science Ltd on behalf of Acta Metallurgica Inc.
Keywords: Alumina; Nanocrystalline powder; Synthesis; Sintering mechanism; Agglomeration
1. INTRODUCTION
Synthesis of ultrane a-Al
2
O
3
powder has been the
subject of many works [14]. The formation of a-
Al
2
O
3
from g-Al
2
O
3
occurs via a series of poly-
morphic transformations on heating:
g-Al
2
O
3
4d-Al
2
O
3
4y-Al
2
O
3
4a-Al
2
O
3
, where a-
Al
2
O
3
forms at temperatures as high as 12008C [5].
During the y-Al
2
O
3
4a-Al
2
O
3
transformation, a-
Al
2
O
3
nuclei form within the ultrane y-Al
2
O
3
matrix, but rapidly grow to produce a-Al
2
O
3
colo-
nies [5]. As a result of the low intrinsic nucleation
density and the volume reduction accompanying the
phase transformation due to the lower specic
volume of a-Al
2
O
3
(0.251 and 0.276 cm
3
/g for a-
Al
2
O
3
and y-Al
2
O
3
, respectively [6]), the a-Al
2
O
3
colonies recede from the matrix and thus form a
network of pore channels and a-Al
2
O
3
``dendrites''.
The resultant coarsened and severely aggregated a-
alumina powder requires sintering at a temperature
greater than 16008C to obtain high densities.
A number of researchers have attempted to take
advantage of the nanocrystalline nature of the tran-
sition aluminas and aluminum hydroxides [79]. It
has, however, been demonstrated repeatedly that
pressureless sintering of such powders require high
temperatures (i.e. > 16008C), which is generally
attributed to the large and extensive pore network
developed during the transformation to a-Al
2
O
3
[8].
Kumagai and Messing [10, 11] demonstrated that
with 10:1 mm a-alumina seed crystals, the kinetics
of the y-Al
2
O
3
4a-Al
2
O
3
phase transformation can
be enhanced dramatically and the microstructure of
the transformed a-Al
2
O
3
can be improved signi-
cantly. Thus boehmite gels, when seeded with a-
Al
2
O
3
, can be sintered to nearly full density at
12008C. However, large bulk samples cannot be
obtained by this method because the gel often
cracks into small pieces during the drying process.
Yeh and Sacks [12] studied the possibility of low
temperature sintering of a-Al
2
O
3
. After removing
agglomerates by centrifugal sedimentation and slip
casting the powder to high green density, a 60 nm
diameter a-Al
2
O
3
powder was sintered to > 99%
relative density at about 15008C. The obvious draw-
back of this method, however, is that it is inecient
and expensive. Rajendran [13], in a study of pro-
duction of ultrane a-Al
2
O
3
powder from alumi-
num nitrate and ammonia solution, found that with
a-Al
2
O
3
seed and ammonium nitrate (a by-product
of the precipitation reaction) in the dry gel of
Acta mater. 48 (2000) 31033112
1359-6454/00/$20.00 7 2000 Published by Elsevier Science Ltd on behalf of Acta Metallurgica Inc.
PII: S1359- 6454( 00) 00115- 4
www.elsevier.com/locate/actamat
{ To whom all correspondence should be addressed.
aluminum hydroxide, the temperature of the
y-Al
2
O
3
4a-Al
2
O
3
phase transformation was low-
ered to 9508C. Due to the relatively low a-Al
2
O
3
formation temperature, exaggerated growth and
hard agglomeration of transition alumina particles
during the phase transformation could be elimi-
nated, and the resultant 60 nm a-Al
2
O
3
powder
could be sintered to > 99% relative density at
15008C. Although there have been many attempts
at modeling the sintering process (e.g. Refs [14
16]), the models usually use micrometer-sized or
sub-micrometer-sized particles and little work can
be found on the sintering mechanism of nanocrys-
talline particles.
In the present work, nanocrystalline a-Al
2
O
3
powders were synthesized through a simple precipi-
tation method, and the eects of a-Al
2
O
3
seeding
and ammonium nitrate (a by-product of the precipi-
tation reaction) on the phase transformations on
heating the Al(OH)
3
dry gel were studied systemati-
cally. The eect of milling media on agglomeration
strength, the mechanism of initial stage sintering,
the eect of agglomeration on sintering behavior
and the mechanical properties of the sintered
samples were investigated.
2. EXPERIMENTAL PROCEDURE
2.1. Powder synthesis
Aluminum nitrate (A.R.) and ammonia solution
(A.R.) were used as the starting materials.
Aluminum nitrate was dissolved in deionized water
and the insoluble impurities were ltered out. To
avoid agglomeration of the ultrane hydrous
alumina, a polyethyleneglycol (PEG) solution with
equal amounts of PEG200, PEG1540 and
PEG10000 was used as a dispersant [17]. During
precipitation, a 4.5 mol/m
3
ammonia solution was
added to the rapidly stirred aluminum nitrate sol-
ution, and the nal pH value of the system was
nine. The slurry was stirred continuously for
another 30 min to homogenize the system.
In order to study the eects of alumina seeding
and residual NH
4
NO
3
on the phase transformations
of the precursor precipitates, three specimens were
prepared: for specimen A, the wet gel was rst
washed eight times with deionized water and then
washed ve times with anhydrous alcohol to
remove the ammonium nitrate completely; for speci-
men B, the wet gel was ltered for 30 min without
washing, thus some ammonium nitrate (about 44%)
remained in the specimen; for specimen C, the prep-
aration process was the same as specimen B, except
that 5 wt% (relative to the produced alumina) dis-
persed a-Al
2
O
3
seed crystals (about 100 nm in diam-
eter) were mixed with the aluminum nitrate solution
before the precipitation process. The three speci-
mens were desiccated in an air oven at 708C. The
dry gels were then calcined at appropriate tempera-
tures to form a-Al
2
O
3
.
In order to observe the eect of milling media on
the strength of agglomerates, the a-Al
2
O
3
powder
obtained by calcining specimen C was milled in a
polytetrauoroethylene bottle with pure alumina
balls, using either anhydrous alcohol or deionized
water as the mill uid.
2.2. Compaction and sintering
The ultrane a-alumina powders, without any
binder, were pressed uniaxially at 30 MPa into pel-
lets and bars. They were then cold isostatically
pressed at 200 MPa. The pellets were 10 mm in di-
ameter and 2 mm in thickness; the bars were 6
6 35 mm
3
in size.
Conventional ramp-and-hold sintering was con-
ducted in a furnace heated by SiC elements. The
bars were sintered at selected temperatures for 2 h,
with a heating rate of 58C/min. Constant heating
rate (CHR) sintering and isothermal sintering (IS)
of the pellets were conducted in air on a Leitz
microscope equipped with a hot plate and an image
analyzer.
2.3. Microstructural characterization and mechanical
properties
Bulk densities of the specimens were determined
by Archimedes' principle. Dierential thermal ana-
lyses (DTA) and thermogravimetric analyses (TG)
of the precipitates were performed on a Shimadzu
DT30 thermal analyzer by heating 30 mg samples
to 12008C in air at a rate of 108C/min. Phase identi-
Fig. 1. TG and DTA curves of the dry gels upon heating:
(a) DTA of Al(OH)
3
; (b) DTA of Al(OH)
3
with residual
NH
4
NO
3
; (c) TG of Al(OH)
3
with residual NH
4
NO
3
; (d)
DTA of Al(OH)
3
with residual NH
4
NO
3
and 5 wt% a-
Al
2
O
3
seed crystals.
3104 LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER
cation was performed using X-ray diractometry
(XRD) on a D/MAX-RB X-ray diractometer. The
particle sizes and size distributions of the alumina
powders were observed using a Phillips EM420
transmission electron microscope (TEM).
Using an Instron-M2406 materials testing ma-
chine, the breakpoint of the green density vs com-
paction pressure curve was used to characterize the
strength of agglomerates [18]. Sintered specimens
for microstructural characterization were polished
to 1 mm nish with diamond paste and thermally
etched at 13008C for 30 min. Microstructural obser-
vations were performed using a Phillips EM505
scanning electron microscope (SEM).
By assuming that the alumina grains were perfect
spheres, the average grain size of the sintered com-
pacts was determined by
D =
2
cos(p=6)

AT
p
pN
r
(1)
where D is the average grain diameter, T
p
is the
percentage of theoretical density, and N is the num-
ber of grains in a selected representative eld with
an area A [19].
Three-point bending strength of the bar speci-
mens was measured using an Instron-M2406, with
an outer span of 25 mm and a crosshead speed of
0.05 mm/min. Eight specimens per batch were
tested. Fracture toughness of the samples was
measured by the indentation method, using the
equation given by Niihara et al. [20]:
K
IC
= 0:0089

E
H
v

0:4
P a
1
l
0:5
(0:25Rl=aR2:5)
(2)
where H
v
is the Vickers hardness, E is the elastic
modulus (E = 380 GPa for monolithic alumina), P
is the indentation load, a is the length of the half-
diagonal of the indent, l = c a, and c is the half-
length of the indentation crack.
3. RESULTS AND DISCUSSION
3.1. Phase transformations of the dry gels upon
heating
Upon heating, g-Al
2
O
3
undergoes a series of
polymorphic changes before the stable corundum
structured a-Al
2
O
3
forms [21, 22]. The g4d and
d4y transformations are displacive [22, 23], with
relatively low activation energies. The y4a trans-
formation is reconstructive and proceeds through a
nucleation and growth process [5]. Since most of
the activation energy is required for the nucleation
process, elevated temperatures are needed to nucle-
ate a-Al
2
O
3
. Normally, the y-Al
2
O
3
4a-Al
2
O
3
transformation temperature is as high as 12008C [5].
A high transformation temperature always results
in the coarsening of particles and formation of hard
agglomerates in the powder. Thus, a reduction in
the y-Al
2
O
3
4a-Al
2
O
3
transformation temperature
is crucial for the processing of highly reactive ultra-
ne a-alumina powder.
DTA results (Fig. 1) show that for the pure
Al(OH)
3
dry gel (specimen A), two endothermic
peaks exist at around 110 and 2006008C. The for-
Table 1. X-ray diraction results obtained by heating Al(OH)
3
dry gels
Sample NH
4
NO
3
(wt%) a-Al
2
O
3
seed (wt%) Temperature (8C)
440 500 600 700 800 900 1000 1050 1100 1150 1200
A 0 0 AlOOH g g g g+d d+y d+y d+y y+a y+a a
B 44 0 AlOOH+g g g+d g+d d+y d+y d+y d+y+a y+a a
C 44 5 AlOOH+g g g+d g+d d+y+a a
Fig. 2. X-ray diraction pattern of the a-Al
2
O
3
powder obtained after calcining specimen C at 9008C
for 3 min.
LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER 3105
mer is caused by the evaporation of absorbed water
and the latter is associated with continuous removal
of the structural water in the hydroxide and oxy-hy-
droxide of aluminum [8]. There are also two
exothermic peaks: the one at 11938C is due to the
y-Al
2
O
3
4a-Al
2
O
3
phase transformation whereas
the one at 6958C is probably caused by the burning
o of organic materials.
For the dry gel containing residual NH
4
NO
3
(specimen B), a sharp exothermic peak exists at
3348C. This is due to the decomposition of am-
monium nitrate which releases a large amount of
heat (210 kJ/mol) [13]. The y-Al
2
O
3
4a-Al
2
O
3
transformation temperature is reduced to 11278C.
From TG measurements, it can be concluded that
there is about 44 wt% NH
4
NO
3
in the specimen.
The existence of 5 wt% a-Al
2
O
3
seed crystals
(specimen C) can further reduce the
y-Al
2
O
3
4a-Al
2
O
3
transformation temperature to
9188C, which means that the combined eect of
44 wt% NH
4
NO
3
and 5 wt% a-Al
2
O
3
seeding can
lower the transformation temperature by 2758C.
Table 1 shows the XRD results of the dry gels
calcined at dierent temperatures. It can be seen
that pure Al(OH)
3
dry gel undergoes the following
phase transformations during heating:
Al(OH)
3
4AlOOH4g-Al
2
O
3
4d-Al
2
O
3
4y-Al
2
O
3
4a-Al
2
O
3
where the formation temperatures of AlOOH, g-
Al
2
O
3
, d-Al
2
O
3
, y-Al
2
O
3
, and a-Al
2
O
3
are 440, 500,
800, 900, and 11008C, respectively. By calcining at
12008C for 1 h, Al(OH)
3
can be transformed com-
pletely to a-Al
2
O
3
. This result is similar to those
reported by others [10, 22].
Although the presence of NH
4
NO
3
in the dry gel
does not change the sequence of the phase trans-
formations, it does reduce the transformation tem-
peratures signicantly. The formation temperatures
of g-, d-, y-, and a-Al
2
O
3
are 60, 200, 100, and 508C
lower than those of the pure Al(OH)
3
dry gel, re-
spectively. The temperature for the complete trans-
formation of y-Al
2
O
3
to a-Al
2
O
3
is 11008C, 1008C
lower than that of the pure Al(OH)
3
dry gel. This is
contrary to the ndings of Rajendran [13] who
observed that NH
4
NO
3
reduces only the tempera-
ture of the y-Al
2
O
3
4a-Al
2
O
3
phase transformation
and had no eect on the temperatures of the other
phase transformations.
It can be seen from Table 1 that for specimen C,
Fig. 3. Morphology of the a-Al
2
O
3
powder prior to milling
(TEM).
Fig. 4. SEM photomicrograph showing agglomeration of the a-Al
2
O
3
powders after milling in: (a)
anhydrous alcohol; (b) deionized water.
Fig. 5. Relationship between relative density of the green
body and compaction pressure for the a-Al
2
O
3
powders.
3106 LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER
a-Al
2
O
3
seeding had no inuence on the formation
temperatures of the transition aluminas, but it did
lower the formation temperature of a-Al
2
O
3
signi-
cantly. After calcining at 9008C for 3 min, the speci-
men transformed completely to a-Al
2
O
3
(Fig. 2).
The dramatic reduction in the formation tem-
perature of a-alumina is due to the dual eects of
ammonium nitrate and a-alumina seeding. The
release of a large amount of energy as a result of
the decomposition of ammonium nitrate in the dry
hydrous alumina gel at 3208C may destroy the nor-
mal arrangement of atoms and produce a highly
disordered structure in the lattices of transition
alumina phases, contributing to the reduction in the
activation energies of the phase transformations
and the nucleation energy of the a-Al
2
O
3
phase.
The eect of a-alumina seeding can be explained by
the heterogeneous nucleation theory [24]. Since each
seed particle can provide multiple nucleation sites
[25], the nucleation energy and thus the phase trans-
formation temperature are lowered signicantly.
Since specimen C has the lowest formation tem-
perature for the a-Al
2
O
3
phase and it is considered
to have the least tendency to form hard agglomer-
ates, only this powder is investigated further in the
present paper.
3.2. Eect of mill uid on strength of agglomerates
and green body homogeneity
After calcining specimen C at 9008C for 3 min,
the a-Al
2
O
3
powder obtained was ball milled in
anhydrous alcohol or deionized water. TEM obser-
vation shows that the diameter of the primary par-
ticles before milling is about 10 nm (Fig. 3). Figure 4
shows that after milling both powders are agglom-
erated, and the diameters of the agglomerates range
from less than 0.1 mm to about 7 mm. For both
powders the green density vs compaction pressure
curves are composed of two straight lines with
dierent slopes (Fig. 5). Assuming that the change
in slope corresponds to the agglomeration strength,
these were determined to be 76 and 234 MPa for
the powders milled in anhydrous alcohol and in
deionized water, respectively.
The strength of agglomerates has a great eect
on microstructural homogeneity of the green com-
pacts. Green bodies of the powder milled in anhy-
Fig. 6. SEM photomicrographs showing microstructure of the green bodies of the a-Al
2
O
3
powders: (a)
soft agglomerated powder; (b) hard agglomerated powder.
Table 2. Slopes (m) and intercepts (B) of the linear relationships between ln(DL/L
0
) and ln t at various isothermal sintering temperatures
Temperature (8C) Soft agglomerated a-Al
2
O
3
powder Hard agglomerated a-Al
2
O
3
powder
m B m B
900 0.3072 5.8307 0.3089 7.2310
925 0.3075 5.5825 0.3099 6.9102
950 0.3103 5.3608 0.3094 6.6058
975 0.3098 5.1399 0.3101 6.3120
1000 0.3069 4.9308 0.3099 6.0305
1025 0.3094 4.7303 0.3104 5.7584
1050 0.3072 4.5404 0.3105 5.4960
1075 0.3059 4.3504 0.3096 5.2459
1100 0.3101 4.1804 0.3088 5.0032
1125 0.3099 4.0104 0.3092 4.7701
1150 0.3104 3.8504 0.3110 4.5445
1175 0.3078 3.6797 0.3100 4.3267
1200 0.3118 3.5304 0.3087 4.1169
LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER 3107
drous alcohol are comparatively homogeneous
[Fig. 6(a)]. Since the nal compaction pressure
(200 MPa) is greater than the strength of agglomer-
ates for this powder, agglomerates can be broken
and thus the powder can be considered as being
soft agglomerated. Conversely, for the powder
milled in deionized water, agglomerates cannot be
broken down and it can be considered as being
hard agglomerated. Green bodies of the hard
agglomerated powder are not as homogeneous as
the soft agglomerated one [Fig. 6(b)]. Relative green
densities of the soft agglomerated and hard agglom-
erated powders were determined to be 36.62 and
28.94%, respectively.
3.3. Sintering mechanism of the a-alumina
nanocrystalline powders
3.3.1. Diusion mechanism during initial stage sin-
tering. Under isothermal conditions, the relation-
ship between the linear shrinkage strain (DL/L
0
) of
the compact and the residence time (t ) can be
expressed as [15]
DL
L
0
=

KgOD
kTr
p

m
t
m
(3)
where DL is the change in length of the specimen,
L
0
is the initial length of the specimen, K, p, m are
numerical constants, g is the surface energy, O is
the vacancy volume, D is the self-diusion coe-
cient, k is the Boltzmann constant, T is the absolute
temperature, and r is the particle radius. Equation
(3) predicts that the plot of log shrinkage strain
against log time should be a straight line if grain
growth does not occur. The diusion mechanism of
the initial stage sintering can be determined by the
slope (m) of the straight line [15]. For uniform
spherical particles, m = 0:31 for grain boundary dif-
fusion, and m = 0:46 for volume diusion.
Plots of ln(DL=L
0
) vs ln t exhibit linear relation-
ships between 900 and 12008C for both the soft
agglomerated and the hard agglomerated powders
(Figs 7 and 8). Slopes and intercepts were calcu-
lated using least-squares analysis (Table 2). It can
be seen that the slopes are very close to 0.31, indi-
cating that the initial stage sintering of the a-
alumina nanocrystalline powders is dominated by
grain boundary diusion.
3.3.2. Activation energy of the initial stage sinter-
ing. According to Johnson [26], if grain boundary
diusion or volume diusion alone causes densica-
tion, the relationship between linear shrinkage
strain and temperature under CHR conditions can
be approximately expressed as
Fig. 9. Relationships between ln[DL=(L
0
T )] and 1/T for
the soft agglomerated and the hard agglomerated ultrane
a-Al
2
O
3
powders during constant heating rate sintering.
Fig. 10. Linear shrinkage rate of the specimens as a func-
tion of temperature during constant heating rate sintering.
Fig. 7. Linear relationships between ln(DL/L
0
) and ln t
during isothermal sintering for the soft agglomerated
ultrane a-Al
2
O
3
powder.
Fig. 8. Linear relationships between ln(DL/L
0
) and ln t
during isothermal sintering for the hard agglomerated
ultrane a-Al
2
O
3
powder.
3108 LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER
DL
L
0
T
1
8
>
>
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
>
>
:

2:14gOD
0B
R
ckr
4
QT
2

1=3
exp

Q
3RT

grain boundary diffusion

5:34gOD
0V
R
ckr
3
QT

1=2
exp

Q
2RT

volume diffusion
(4)
where O is the vacancy volume, R is the gas con-
stant, c is the heating rate, Q is the activation
energy, D
0B
is the grain boundary diusion coe-
cient, and D
0V
is the volume diusion coecient.
In equation (4) the exponential term is more
sensitive to temperature changes than the other
temperature terms, thus the ln[DL=(L
0
T )] vs 1/T
plot of a CHR experiment should exhibit a
single slope if a single diusion mechanism oper-
ates or a combination of mechanisms with the
same activation energies is predominant over the
entire temperature range. The activation energy
can be determined from the slope of the line,
nQ=R, where n = 1=3 for grain boundary diu-
sion and n = 1=2 for volume diusion.
For the soft agglomerated powder, linear re-
lationships between ln[DL=(L
0
T )] and T are found
in the temperature range 95013008C (Fig. 9). A
change in slope can be caused by a change in mass
transport mechanism or a change in the relative im-
portance of competing mechanisms [27]. The devi-
ation from linearity below 9508C may be caused by
surface diusion with a relatively low activation
energy. Surface diusion, although contributing to
the growth of particles and the formation of necks,
does not result in obvious shrinkage of green bodies
[27]. At temperatures greater than 13008C, grain
growth may be signicant, causing deviation from
linearity, as would be expected from equation (4).
For the hard agglomerated a-Al
2
O
3
powder, a
linear relationship between ln[DL=(L
0
T )] and 1/T is
found at temperatures below 12508C.
Least-squares analysis yields the following re-
lationships:
ln[DL=(L
0
T )] = 13,150=T 0:7979556
soft agglomerated a-Al
2
O
3
powder
ln[DL=(L
0
T )] = 17,841=T 1:924625
hard agglomerated a-Al
2
O
3
powder:
Since the initial stage sintering of the ultrane
alumina powders is controlled by grain boundary
diusion, the activation energy, Q, is calculated to
be 328 kJ/mol for the soft agglomerated powder
and 445 kJ/mol for the hard agglomerated powder.
Table 3 shows the comparison of activation ener-
gies of a-Al
2
O
3
powders of dierent sizes. It can be
seen that activation energy of the nanometric a-
alumina powder is much smaller than those of the
sub-micrometer and micrometer sized a-alumina
powders.
The activation energy of the hard agglomerated
powder is much higher than that of the soft
agglomerated one. Since the agglomerates of the
soft agglomerated powder can be broken during the
compaction process, sintering proceeds mainly
between the primary particles and thus, the diu-
sion distances are much shorter, resulting in lower
activation energy.
For the hard agglomerated powder, however, the
strength of the agglomerates is much higher than
the compaction pressure, and the agglomerates can-
Fig. 11. Linear shrinkage of the specimens as a function
of temperature during constant heating rate sintering.
Fig. 12. Relative density as a function of temperature
under conventional ramp-and-hold sintering conditions.
Table 3. Activation energies for the initial stage sintering of dierent a-Al
2
O
3
powders
Diameter, d (mm) State of agglomeration Initial stage activation energy, Q (kJ/mol)
0.01 Soft agglomerated 328
0.01 Hard agglomerated 445
0.20.5 Hard agglomerated 627 [28]
12 Dispersed 481 [27]
35 Dispersed 593 [28]
LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER 3109
not be crushed completely during compaction,
resulting in poor microstructural homogeneity of
the green body. Since the unbroken agglomerates
have comparatively high densities, preferential sin-
tering occurs in these regions at a temperature
lower than the initial stage sintering temperature.
This dierential sintering can produce internal stres-
ses in the green compacts, resulting in cracking
between agglomerates and the surrounding particles
[2931]. Thus, densication of the green body is
controlled by the sintering between the already den-
sied regions, which necessitates longer diusion
distances and a higher activation energy.
3.4. Densication process during CHR sintering
Figures 10 and 11 show the linear shrinkage
strain rate and the linear shrinkage strain of the
compacts as a function of temperature, respectively.
For the soft agglomerated powder, shrinkage occurs
mainly between 1200 and 14008C. For the hard
agglomerated powder, the linear shrinkage strain
rate is relatively small in the entire experimental
temperature range. The soft agglomerated powder
has a maximum shrinkage strain rate of 3:42
10
4
=s at 13508C, with a shrinkage strain of
22.14%. Whereas, the hard agglomerated powder
has a maximum shrinkage strain rate of 1:562
10
4
=s at 14008C, only about half of that of the
soft agglomerated powder, and the corresponding
shrinkage strain is 19.57%. The relationship
between linear shrinkage strain (e ), relative density
(r), and relative green density (r
0
) is given by [32]
r = r
0
=(1 e)
3
: (5)
Relative densities at maximum shrinkage strain rate
were calculated to be 77.58 and 55.62% for the soft
agglomerated and the hard agglomerated powders,
respectively. According to the work of Lange [33],
the relative density at the maximum shrinkage
strain rate is 77%, which is consistent with the soft
agglomerated a-Al
2
O
3
powder. The occurrence of a
maximum linear shrinkage strain rate indicates a
change of densication mechanism from sintering
kinetics to coarsening kinetics [33]. Thus, sintering
kinetics dominate up to 1350 and 14008C for the
soft agglomerated and the hard agglomerated pow-
ders, respectively. At 15008C, relative densities of
the soft agglomerated and the hard agglomerated
powder are 99.0 and 80.6%, respectively, indicative
of the eect of agglomeration strength on sintering.
3.5. Conventional ramp-and-hold sintering
3.5.1. Eect of agglomerates on densication. For
the soft agglomerated powder, densication mainly
occurs in the temperature range of 110013508C
(Fig. 12). After holding at 14008C for 2 h, the green
compact reaches 99% relative density with a linear
shrinkage of 28.18% (Fig. 13) and a mean grain
size of 0.93 mm [Fig. 14(a)]. Densication tempera-
ture is about 3004008C lower than that of mi-
crometer-sized a-Al
2
O
3
powders. Under the same
sintering condition, the hard agglomerated a-Al
2
O
3
ultrane powder can only reach 73.06% relative
density (Fig. 12), and the mean grain size is about
Fig. 14. Microstructure (SEM) of the specimens obtained by sintering the soft agglomerated a-Al
2
O
3
powder (a) and the hard agglomerated a-Al
2
O
3
powder (b) at 14008C for 2 h.
Fig. 13. Relationship between linear shrinkage and tem-
perature under conventional ramp-and-hold sintering con-
ditions.
3110 LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER
3 mm [Fig. 14(b)]. The dierence in sintering kinetics
between the two powders is attributed to the fact
that the hard agglomerated a-Al
2
O
3
powder has the
lower green density under the same compaction
conditions and dierential sintering occurs between
the hard agglomerates and the surrounding matrix
resulting in the formation of gaps. Since the densi-
cation rate of the aggregates is higher than that of
the matrix, isostatic tensile stresses arise within a
spherical aggregate. Within the surrounding matrix,
the radial stress should be tensile and the tangential
stress should be compressive. The tensile stress can
be released by the formation of a crack-like internal
surface around most of the aggregate/matrix inter-
face [30, 31].
3.5.2. Mechanical properties. The three-point
bending strength of the samples of soft agglomer-
ated powder increases with sintering temperature
(Fig. 15). After sintering at 14008C for 2 h, the
mean three-point bending strength is 700 MPa.
The measured mean fracture toughness of the
samples is 4.8 MPa m
1/2
, which is higher than that
of conventional alumina (about 3.0 MPa m
1/2
). The
small grain size and rod shape of the grains may be
the main reasons for this increase.
4. CONCLUSIONS
Ultrane a-alumina powders have been syn-
thesized from aluminum nitrate and ammonia sol-
ution by a simple precipitation method. With 44%
ammonium nitrate (the by-product of the precipi-
tation reaction) in the dry gel of Al(OH)
3
, the for-
mation temperatures of g-, d-, y-, and a-Al
2
O
3
are
60, 200, 100, and 508C lower than those of the pure
Al(OH)
3
dry gel, respectively. Seeding with a-
alumina can lower the temperature of the
y-Al
2
O
3
4a-Al
2
O
3
phase transformation further.
With 5 wt% a-alumina seed crystals of 100 nm di-
ameter and 44% ammonium nitrate, the transform-
ation temperature was lowered to 9008C, and the
resultant a-alumina powder has a mean particle di-
ameter of 10 nm.
The milling uid has a signicant eect on the
strength of the agglomerates. The powder milled in
anhydrous alcohol has an agglomeration strength
of 76 MPa (soft agglomerated), while the one milled
in deionized water has an agglomeration strength of
234 MPa (hard agglomerated).
The soft agglomerated powder produces a com-
paratively homogeneous green body microstructure.
Conventional ramp-and-hold sintering shows that
after holding at 14008C for 2 h, the ceramic is 99%
theoretical density with a mean grain size of
0.93 mm, a mean exural strength of 700 MPa and a
fracture toughness of 4.75 MPa m
1/2
. Conversely,
the green body of the hard agglomerated powder is
poor in microstructural homogeneity and contains
uncrushed agglomerates that result in dierential
sintering and internal stresses in the compact.
Under the same sintering condition, the hard
agglomerated powder can only reach 73.06% theor-
etical density.
Initial stage sintering of both the soft and the
hard agglomerated powders is controlled by grain
boundary diusion, with activation energies of 365
and 492 kJ/mol, respectively. Sintering kinetics
dominate up to 1350 and 14008C for the soft
agglomerated and hard agglomerated powder, re-
spectively, with coarsening kinetics dominating at
high temperatures.
AcknowledgementsThis work was supported by the
National Natural Science Foundation of China (No.
59502007) and the Natural Science Foundation of
Liaoning Province (No. 9521035), China. The authors are
very grateful to Dr J. A. Yeomans, University of Surrey,
for many good suggestions and correction of English.
REFERENCES
1. Yoldas, B. E., Am. Ceram. Soc. Bull., 1975, 54(3),
289.
2. Blendel, E., Bown, H. K. and Coble, R. L., Am.
Ceram. Soc. Bull., 1984, 63(6), 797.
3. Fanell, A. J. and Burlew, J. V., J. Am. Ceram. Soc.,
1986, 69(8), C174.
4. Ogihara, T., Nakajima, H., Yanagawa, T., Ogata, N.,
Yoshida, K. and Matsushida, N., J. Am. Ceram. Soc.,
1991, 74(9), 2263.
5. Dynys, F. W. and Halloran, J. W., J. Am. Ceram.
Soc., 1982, 65(9), 442.
6. Wilson, S. J. and Stacey, M. H., J. Colloid Sci.
Interface Sci., 1981, 82(2), 507.
7. Tsai, D. S. and Hsieh, C. C., J. Am. Ceram. Soc.,
1991, 74(4), 830.
8. Baldkar, P. A. and Bailey, J. E., J. Mater. Sci., 1976,
11, 1794.
9. Ayral, A. and Phalippou, J., Adv. Ceram. Mater.,
1988, 3(6), 575.
10. Kumagai, M. and Messing, G. L., J. Am. Ceram.
Soc., 1985, 68(9), 500.
11. Messing, G. L. and Kumagai, M., Am. Ceram. Soc.
Bull., 1994, 73(10), 88.
12. Yeh, T. S. and Sacks, M. D., J. Am. Ceram. Soc.,
1988, 71(10), 841.
13. Rajendran, S., J. Mater. Sci., 1994, 29, 5664.
14. Cui, G. W., in Defects, Diusion and Sintering.
Tsinghua University Press, Beijing, 1990, p. 144.
Fig. 15. Three-point bending strength of the alumina cer-
amic as a function of sintering temperature.
LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER 3111
15. Johnson, D. L. and Cutler, I. B., J. Am. Ceram. Soc.,
1963, 46(11), 541.
16. Coble, R. L., J. appl. Phys., 1961, 32(5), 787.
17. Xu, D. C., Hao, X. C. and Zhu, X. H., J. Chin.
Ceram. Soc., 1992, 20(1), 48.
18. Niesz, B. E., Bennett, R. B. and Synder, M. J.,
Ceram. Bull., 1972, 51(9), 677.
19. Sumita, S., J. Japan. Ceram. Soc., 1991, 99(7), 538.
20. Niihara, K., Morena, R. and Hasselman, D. P. H., J.
Mater. Sci. Lett., 1982, 1, 13.
21. Stumpf, S. C., Russell, A. S., Newsome, J. W. and
Tucker, C. M., Ind. Engng Chem., 1950, 42, 1398.
22. Lippins, B. C. and DeBoer, J. H., Acta crystallogr.,
1964, 17, 1312.
23. Wilson, S. J., J. Solid St. Chem., 1979, 30, 247.
24. Mcardle, J. L. and Messing, G. L., Adv. Mater., 1988,
3(4), 387.
25. Shelleman, R. A., Messing, G. L. and Kumagai, M.,
J. Non-Cryst. Solids, 1996, 82, 277.
26. Johnson, D. L., J. appl. Phys., 1969, 40(1), 192.
27. Young, W. S. and Cutler, I. B., J. Am. Ceram. Soc.,
1970, 53(12), 659.
28. Johnson, D. L. and Cutler, I. B., J. Am. Ceram. Soc.,
1963, 46(11), 545.
29. Reeve, K. D., Ceram. Bull., 1963, 42(8), 452.
30. Kellett, B. and Lange, F. F., J. Am. Ceram. Soc.,
1984, 67(5), 369.
31. Evans, A. G., J. Am. Ceram. Soc., 1982, 65(10), 497.
32. Greskovich, G. and Lay, K. W., J. Am. Ceram. Soc.,
1972, 55(3), 142.
33. Lange, F. F., J. Am. Ceram. Soc., 1989, 72(1), 3.
3112 LI and SUN: NANOCRYSTALLINE a-ALUMINA POWDER

You might also like