Available online at www.sciencedirect.

com

Ceramics International 38 (2012) 787–793

www.elsevier.com/locate/ceramint

Pore size control of Al2O3 ceramics using two-step sintering

Toshihiro Isobe *, Asami Ooyama, Mai Shimizu, Akira Nakajima
Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan
Received 14 May 2011; received in revised form 15 July 2011; accepted 1 August 2011
Available online 10th August 2011

Abstract
Porous Al2O3 ceramics were prepared using a two-step sintering method. Green bodies were prepared using the slip cast method. The average
pore size and porosity of the obtained samples were, respectively, 73 nm and 42%. The porosities and the average pore size of the samples prepared
using a one-step sintering method decreased concomitantly with increasing sintering temperature. The porosity was mostly constant at
temperatures higher than 1300 8C. The T1 and T2 temperatures were set, respectively as 1150 8C and 1000 8C for two-step sintering tests.
Under these conditions, the pore size was controllable by sintering time without porosity change. The average pore size of the obtained samples was
61–76 nm without porosity change.
Crown Copyright # 2011 Published by Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: Ceramics; Nanostructures; Structural materials; Microstructure

1. Introduction show excellent microstructures and properties. Consequently,

porous ceramics are used in many end products.
Porous ceramics are widely used in industrial, catalyst, Partial sintering, an extremely straightforward and simple
and environmental fields. Various methods for their technique, is another method used to prepare porous ceramics
preparation have been studied [1]. The replica method [20–24]. The powder compacts are heated at less than
includes the impregnation of a porous structure with ceramic optimum sintering temperature for preparation of the dense
slurries or a precursor solution to prepare porous ceramics body. The pore size and porosity of the porous ceramics
[2–8]. This method is available to prepare porous ceramics depend on the particle size and porosity of the green body.
having a simple, complicated, or unique structure. Using a They are also controllable using a heating program. The
polymeric sponge as the template, the obtained porous ceramics sintered at high temperatures generally have low
ceramics have 40–95% porosity and 0.2–3 mm pore or hole porosity and small pores because of matrix densification.
size [2–5], which is probably suitable for high-throughput Therefore, the sintering program is an important factor to be
filtration. Natural woods can be processed using this method considered in preparing porous ceramics. This method also has
[6–8]. In the sacrificial template method, organic additives an important shortcoming: pore size control of 10 nm order is
are mixed with ceramic powders and are burned out during difficult.
sintering of the molded green bodies. Synthetic organic Two-step sintering is a remarkable sintering program
powders [5,9–13] and fibers [12–14] have been used as pore [25,26]. This technique uses the difference of reaction
formers. The foaming method incorporates air into ceramic velocities between the diffusion of atoms and the grain
slurries or liquid precursor and maintains the structure that is boundary migration. The resultant ceramics have high relative
constructed by the air bubbles [15–19]. The pore size and density and low grain growth. Diffusion of the atoms and grain
porosity depend on the rheological properties of the boundary migration are thought to be independent parameters
precursors. Porous ceramics prepared using these methods during sintering by this program. Therefore, this method is
useful to control the pore size or porosity precisely. In this
study, porous Al2O3 ceramics were prepared using two-step
* Corresponding author. Tel.: +81 3 5734 2525; fax: +81 3 5734 3355. sintering. The relations between the heating program and the
E-mail address: isobe.t.ad@m.titech.ac.jp (T. Isobe). porosity and pore size were investigated.
0272-8842/$36.00. Crown Copyright # 2011 Published by Elsevier Ltd and Techna Group S.r.l. All rights reserved.
doi:10.1016/j.ceramint.2011.08.005
788 T. Isobe et al. / Ceramics International 38 (2012) 787–793

2. Experimental procedure Table 1

Sintering program for a two-step sintering study.

2.1. Preparation of green bodies T1 (8C) T2 (8C) t [h]

A 1000–1300 T1 minus 150 12
First, 30% (vol) high purity a-Al2O3 (TM-DAR; Taimei B 1000–1250 1000 12
Chemicals Co. Ltd., Nagano, Japan) was mixed with 70% (vol) C 1150 1000 2–12
distilled water and 0.62% (mass) poly(ammonium acrylate)
(PAA) (Ceruna D-305; Chukyo Yushi Co. Ltd., Aichi, Japan) as
a dispersant. Mixing was done using a planetary-type mixer 12 h. Regarding program B, the effect of the T1 for the porous
[27] without stainless steel mesh for 1 min, rotated at 800 rpm, structure of the samples was evaluated. The values of T2 and t
and revolved at 2000 rpm for 1 min. Then, the mixtures were were, respectively, 1000 8C and 12 h. The t dependence was
treated for 10 min using a mixer with the mesh to disperse the evaluated using the program C. The values of T1 and the t were,
a-Al2O3 particles in the slurry. After mesh treatment, the respectively, 1000 8C and 2–12 h.
suspensions were re-mixed without mesh for 1 min to remove After sintering, the surfaces of all samples were polished
bubbles. The respective rotation and revolution rates were 60 using commercial #400 and #1500 emery. The samples were
and 2000 rpm. The resulting slurries were slip-casted into a washed ultrasonically for 10 min in de-ionized purified water.
16 mmf gypsum mold for 5 min. They were dried at room Then the densities and porosities of the samples were measured
temperature for 24 h and were heated at 800 8C for 2 h to burn using Archimedes’ technique with water. The linear shrinkage
out the dispersant. The microstructures of the present samples was calculated using Eq. (1).
were observed using a field emission scanning electron
Db  Da
microscope (FE-SEM, S4500; Hitachi High-Technologies  100 (1)
Db
Corp., Tokyo, Japan).
In Eq. (1), Db and Da, respectively denote the diameters of the
2.2. Sintering of ceramic bodies sample before and after sintering. The pore size distribution of
the samples was measured using mercury intrusion porosime-
In the case of single-step sintering, the green bodies were try (Pascal 140 and Pascal 240; CE Instruments Ltd., Italy).
sintered at 1000–1400 8C for 2 h. The heating and cooling rates The weight of the sample used for the measurement was 0.7 g.
were 100 8C/h. The sintering program of the two-step sintering Their microstructures were observed using a field emission
is presented in Fig. 1. Temperatures of the first sintering step scanning electron microscope (FE–SEM, S4500; Hitachi
(T1) and the second sintering step (T2) are presented in Table 1. High-Technologies Corp., Tokyo, Japan).
Several combinations of temperatures of T1, T2, and keeping
time of T2 (t) were used for this study. The heating rate of up to 3. Results and discussion
the T1 for all programs was 100 8C/h. The respective cooling
rates from T1 to T2 and from T2 to room temperature were 3.1. Single-step sintering
15 8C/min and 100 8C/h. For program A, the T1 was 1000–
1300 8C and the T2 was set the 150 8C lower than T1. The t was Fig. 2 shows the pore size distribution and cumulative
volume of the green bodies before sintering. The average pore
(a) size was about 73 nm. It shows a sharp peak at the average
T
Temperatur e

150
100°C/h
2h
e [mm3/g]

100 °C/h 100

Time
Volume

(b)
T1 15°C/min 50
Temperatu rre

T2
100°C/h
t
0
0.01 0.1 1 10
100°C/h
Pore size [µm]
Time
Fig. 2. Pore size distribution and cumulative volume of the sample before
Fig. 1. Heating programs of (a) single and (b) two-step sintering methods. sintering.
 T. Isobe et al. / Ceramics International 38 (2012) 787–793 789

value. The microstructure of the present sample is also depicted 60 80

in Fig. 3. The a-Al2O3 particles were packed randomly. No
large pore or cracking was observed. The porosity measured 50

Average pore size [nm]

60
using Archimedes’ technique was about 42%. Fig. 4 and Table
40

Porosity [%]
2 presents the average pore size and porosity as a function of
sintering temperature. The pores of all samples were mainly
30 40
open pores. With heating at less than 1000 8C, the porosities
and the average pore size were mostly constant. With heating at 20
temperatures greater than 1100 8C, the average pore size 20
decreased with the increasing of the sintering temperature. The 10
sample sintered at 1250 8C has 22 nm average pore size. The
pore size distribution of the samples sintered at temperatures 0 0
800 1000 1200 1400
higher than 1300 8C could not be measured because the
Heating temperature [°C]
samples have low porosity. The porosity change was observed
at a slightly lower temperature. Porosity reached 2% at more Fig. 4. Porosity and average pore size of the samples sintered using the single-
than 1300 8C. It is considered that these changes cause step sintering method.
shrinkage during sintering.
The linear shrinkage rate of the samples was estimated. predicted using the following equation.
Fig. 5 shows the relation between the sintering temperature and p ffiffiffiffiffiffiffiffiffiffi p ffiffiffiffiffiffiffiffiffiffi
linear shrinkage. The linear shrinkage of the samples was not 3
1=r1  3 1=r2
b¼ p ffiffiffiffiffiffiffiffiffiffi (2)
changed significantly after heating below 1100 8C. However, it 3
1=r1
was increased significantly after heating to 1100 8C or higher
temperatures. This behavior showed agreement with that Therein, r1 and r2, respectively denote the relative density of
described in previous reports [28]. The validity of these results the green body and the samples heated at various temperatures.
was confirmed using calculations. The linear shrinkage, b, was The curve calculated using Eq. (2) is shown as a dotted line in
Fig. 5. This curve fits well with the measured value, which
indicates that no severe defects and orientation of the Al2O3
particles were present in the samples.

3.2. Optimum T2 for two-step sintering

Fig. 6 shows the average pore size and porosity of the samples
sintered using the program A. The average pore size and porosity
of the sample sintered at 1100 8C as T1 were, respectively, 74 nm
and 36%. The pore size of the sample sintered at 1150 8C as T1 is
about equal to the value shown above. However, it decreased
concomitantly with increasing T1.
When the ceramics are prepared using two-step sintering
method, T1 is the temperature at which the grain boundary
migration is active. Actually, T2 should be set to a temperature
at which the diffusion of atoms is sufficiently active without
movement of the particles. These temperatures can be inferred

Table 2
Porosity and average pore size of the a-Al2O3 ceramics prepared using single-
step sintering.
Sintering temperature (8C) Porosity (%) Average pore size (nm)
800 42 73
1000 40 71
1050 38 70
1100 34 68
1150 26 64
1200 14 53
1250 5 22
1300 <2 –
1350 <2 –
1400 <2 –
Fig. 3. FE-SEM micrographs of the samples before sintering.
790 T. Isobe et al. / Ceramics International 38 (2012) 787–793

3.3. Optimum T1 for two-step sintering

The samples prepared using program B are presented in

Fig. 8. The respective porosity and the average pore size of the
samples sintered at 1050 and 1100 8C as T1 were about 42%
and 73 nm, which were the same as those of the samples before
sintering. They also decreased concomitantly with increasing
T1 at 1150 8C and higher temperatures.
Porosity and the average pore size changes are considered
to cause sintering shrinkage. This behavior resembles that of
samples sintered using program A. However, the behaviors did
not accord completely. Decreases of the pore size and porosity
are regarded as originating in T1. The sintering shrinkage
showed slightly different behavior because the samples were
cooled promptly after having arrived at T1. The micro-
structures of the present samples are presented in Fig. 9. Data
Fig. 5. Measured shrinkages of the samples sintered using single-step sintering for 1150 8C as T1 are displayed in Fig. 7(a). Particle adhesion
method and shrinkage calculated using Eq. (2). was confirmed in both micrographs. However, the particles of
these samples did not grow larger than the particles of the
samples prepared using program A. This microstructural
from the shrinkage curve of the sample sintered using one-step difference supports discussion of the slight difference between
sintering. In Fig. 5, the samples shrank gradually at Figs. 6 and 8.
temperatures lower than 1050 8C and changed dynamically In the two-step sintering method, excessive energy is given
at 1050 8C and higher temperatures, which indicated that grain to the particles at T1 and is used for the diffusion of atoms at T2
boundary migration was active at 1050 8C and higher instead of that originally used for the grain boundary migration
temperatures. Therefore, when T1 and T2 were higher than at T1. The porosity and pore size of the samples after sintering
the temperature of 1050 8C, both the pore size and porosity at 1050 and 1100 8C as T1 corresponded with the green body,
decreased, as they did in the general two-step sintering method. which suggests that both temperatures are not sufficient to
To obtain the results described in Section 3.3, T2 was set as cause the diffusion of atoms. It also indicates that the promotion
1000 8C. of the diffusion of atoms using one-step sintering is difficult. In
The sample microstructures were also verified. The FE– contrast, samples sintered at 1150 8C as T1 have a somewhat
SEM micrographs of the samples sintered at 1000 8C as T2 are low value of the porosity. This temperature is expected to be
portrayed in Fig. 7(a). The Al2O3 particle shape and size did not ample for leading the initial driving force of the diffusion of
change much from those observed before sintering. At 1100 8C atoms. Consequently, T1 was selected as 1150 8C for this study.
as T2, some particles were round. The particles that mutually
adhered are portrayed in Fig. 7(c). Sintering of the powder 3.4. Control of the pore size using two-step sintering
compacts was promoted at 1200 8C as T2; grain growth was
observed (Fig. 7(d)). It is inferred that this microstructural The green bodies were sintered at 1150 8C of T1 and 1000 8C
change well explained the results, as depicted in Fig. 6. of T2 for 2–12 h. The obtained data are presented in Fig. 10. The
average pore size of the sample sintered for 2 h was about
50
61 nm. It increased concomitantly with increased sintering
time, becoming about 73 nm after sintering for 12 h. However,
80
the porosity was almost 35%.
Average pore size [nm]

40
In fact, T1 of 1150 8C is the temperature at which the grain
60 boundary migration and the diffusion of atoms with the
Porosity [%]

30 movement of the particles are active. The porosity of the

obtained samples decreased slightly from 42% to 35%.
40
20 Furthermore, keeping temperatures at T2 does not work for
densification. This process is regarded as similar to the initial
10 20 stage of sintering [29]. In this stage, the atomic diffusion is
explained by the evaporation–condensation mechanism, sur-
face and volume diffusion. The following equation shows the
0 0
900 1000 1100 1200 evaporation–condensation mechanism [29].
T2 [°C]
 pffiffiffi 3=2 1=3
x 3 pgM p
Fig. 6. Porosity and average pore size of the samples sintered using two-step ¼ pffiffiffi 3=2 3=2 2 r 2=3 t1=3 (3)
program A. r 2R T d
 T. Isobe et al. / Ceramics International 38 (2012) 787–793 791

Fig. 7. FE-SEM micrograph of the samples sintered using two-step program A: T2: (a) 1000 8C, (b) 1050 8C, (c) 1100 8C, and, (d) 1150 8C.

50 120 Therein, x, r, g, M, P, R, T, d, and t, respectively represent the

neck size, particle radius, surface tension, molecular weight of
Average pore size [nm]

40 100 the vapor, vapor pressure, ideal gas constant, absolute temper-
ature, density, and sintering time. The equation shows that the
osity [%]

30 long sintering time engenders neck growth and the decrease of

80 the particle size. Furthermore, the distance between particles
Poro

does not change because the atoms merely move from the
20
60 surface of the neck. Similarly to the evaporation–condensation
mechanism, the surface and volume diffusion do not decrease
10
the porosity [29].
40
Based on these three sintering programs, we infer the
0 following results. When T1 and T2 were set to temperatures
1000 1100 1200 1300
higher than the temperature at which the grain boundary
T1 [°C]
migration, the pore size and porosity decreased concomitantly
Fig. 8. Porosity and average pore size of the samples sintered using two-step with increased sintering temperature. When T2 was set lower
program B. than this temperature, the pore size and porosity decreased
concomitantly with increased sintering temperature. If T1 is

Fig. 9. FE-SEM micrograph of the samples sintered using two-step program B: T1: (a) 1200 and (b) 1250 8C.
792 T. Isobe et al. / Ceramics International 38 (2012) 787–793

50 References
100

Average pore size [nm]

40 [1] A.R. Studart, U.T. Gonzenbach, E. Tervoort, L.J. Gauckler, Processing
routes to macroporous ceramics: a review, J. Am. Ceram. Soc. 89 (2006)
80 1771–1789.
Porosity [%]

30 [2] J. Saggio-Woyansky, C.E. Scott, W.P. Minnear, Processing of porous

ceramics, Am. Ceram. Soc. Bull. 71 (1992) 1674–1682.
60 [3] J. Luyten, I. Thijs, W. Vandermeulen, S. Mullens, B. Wallaeys, R.
20
Mortelmans, Strong ceramic foams from polyurethane templates, Adv.
Appl. Ceram. 104 (2005) 4–8.
10 40 [4] J.T. Richardson, Y. Peng, D. Remue, Properties of ceramic foam catalyst
supports: pressure drop, Appl. Catal. A: General 204 (2000) 19–32.
[5] J. Luyten, S. Mullens, J. Cooymans, A.M. De Wilde, I. Thijs, New
0 20 processing techniques of ceramic foams, Adv. Eng. Mater. 5 (2003)
5 10 15
715–718.
t [h] [6] T. Ota, M. Takahashi, T. Hibi, M. Ozawa, S. Suzuki, Y. Hikichi, H. Suzuki,
Biomimetic process for producing SiC ‘‘Wood’’, J. Am. Ceram. Soc. 78
Fig. 10. Porosity and average pore size of the samples sintered using two-step
(1995) 3409–3411.
program C.
[7] T. Ota, M. Imaeda, H. Takase, M. Kobayashi, N. Kinoshita, T. Hirashita,
H. Miyazaki, Y. Hikichi, Porous titania ceramic prepared by mimicking
insufficiently high, then neither the pore size nor porosity silicified wood, J. Am. Ceram. Soc. 83 (2000) 1521–1523.
[8] M. Mizutani, H. Takase, N. Adachi, T. Ota, K. Daimon, Y. Hikichi, Porous
changed. When T1 and T2 are set higher and lower than the grain
ceramics prepared by mimicking silicified wood, Sci. Technol. Adv.
boundary migration, the pore size is controllable by the Mater. 6 (2005) 76–83.
sintering time without porosity change. [9] P. Colombo, E. Bernardo, L. Biasetto, Novel microcellular ceramics from
a silicone resin, J. Am. Ceram. Soc. 87 (2004) 152–154.
4. Conclusions [10] I. Thijs, J. Luyten, S. Mullens, Producing ceramic foams with hollow
spheres, J. Am. Ceram. Soc. 87 (2004) 170–172.
[11] D.M. Liu, Densification and anisotropic shrinkage of die-pressed SiC of
Porous Al2O3 ceramics were prepared using a two-step different granular forms, Ceram. Int. 23 (1997) 135–139.
sintering method. Porous ceramics having various pore size and [12] T. Isobe, Y. Kameshima, A. Nakajima, K. Okada, Y. Hotta, Gas perme-
porosity can be prepared precisely using the sintering program ability of the porous alumina ceramics with unidirectionally aligned pores,
from the simple powder compact. The following results were J. Eur. Ceram. Soc. 27 (2007) 53–59.
[13] T. Isobe, Y. Kameshima, A. Nakajima, K. Okada, Y. Hotta, Extrusion
obtained:
method using nylon 66 fibers for the preparation of porous alumina
ceramics with oriented pores, J. Eur. Ceram. Soc. 26 (2006)
1. The porosities and the average pore size of the samples 2213–2217.
prepared using a one-step sintering method decreased [14] T. Isobe, T. Tomita, Y. Kameshima, A. Nakajima, K. Okada, Preparation
concomitantly with increasing sintering temperature. The and properties of porous alumina ceramics with oriented cylindrical
pores produced by an extrusion method, J. Eur. Ceram. Soc. 26 (2006)
porosity was mostly constant at temperatures higher than
957–960.
1300 8C. The porosity and average pore size curve as a [15] J. Zeschky, T. Hofner, C. Arnold, R. Weissmann, D. Bahloul-Hourlier, M.
function of the temperature shows typical sintering behavior. Scheffler, P. Greil, Polysilsesquioxane derived ceramic foams with gradi-
2. For the two-step sintering method, the T2 temperature is set ent porosity, Acta Mater. 53 (2005) 927–937.
to 1000 8C based on evaluation of the results of the samples [16] E.J.A.E. Williams, J.R.G. Evans, Expanded ceramic foam, J. Mater. Sci.
prepared using program A. If the T2 is higher than this 31 (1996) 559–563.
[17] T. Tomita, S. Kawasaki, K. Okada, A novel preparation method for foamed
temperature, then both porosity and pore size decease during silica ceramics by sol–gel reaction and mechanical foaming, J. Porous
sintering. Mater. 11 (2004) 107–115.
3. The T1 temperature is set to 1150 8C from the results of the [18] P. Sepulveda, J.G.P. Binner, S.O. Rogero, O.Z. Higa, J.C. Bressiani,
samples prepared by program B. When the T1 temperature is Production of porous hydroxyapatite by the gel-casting of foams and
cytotoxic evaluation, J. Biomed. Mater. Res. 50 (2000) 27–34.
lower than this temperature, the pore size and porosity do not
[19] F.S. Ortega, J.A. Rodrigues, V.C. Pandolfelli, Strength of gelcast ceramic
change. foams, Am. Ceram. Soc. Bull. 83 (2004) 9501–9506.
4. When T1 and T2 are set to 1150 and 1000 8C, which are the [20] R.L. Coble, Initial sintering of alumina and hematite, J. Am. Ceram. Soc.
optimum temperatures, the pore size is controllable by the 41 (1958) 55–62.
sintering time without porosity change. The average pore [21] S.J. Glass, D.J. Green, Permeability and infiltration of partially sintered
size of the obtained samples was 61–76 nm. ceramics, J. Am. Ceram. Soc. 82 (1999) 2745–2752.
[22] S.Y. Shan, J.F. Yang, J.Q. Gao, W.H. Zhang, Z.H. Jin, R. Janssen, T. Ohji,
Porous silicon nitride ceramics prepared by reduction–nitridation of silica,
J. Am. Ceram. Soc. 88 (2005) 2594–2596.
Acknowledgments [23] I.H. Arita, V.M. Castano, D.S. Wilkinson, Synthesis and processing of
hydroxyapatite ceramic tapes with controlled porosity, J. Mater. Sci.:
We are grateful to Professor E. Sakai of Tokyo Institute of Mater. Med. 6 (1995) 19–23.
[24] D. Hardly, D.J. Green, Mechanical properties of a partially sintered
Technology for permitting the use of mercury porosimetry. FE- alumina, J. Eur. Ceram. Soc. 15 (1995) 769–775;
SEM micrographs were taken at the Center for Advanced I.-W. Chen, X.-H. Wang, Sintering dense nanocrystalline ceramics with-
Materials Analysis (CAMA) of Tokyo Institute of Technology. out final-stage grain growth, Nature 404 (2000) 168–171.
 T. Isobe et al. / Ceramics International 38 (2012) 787–793 793

[25] K. Maca, V. Pouchly, P. Zalud, Two-step sintering of oxide ceramics with [27] T. Isobe, A. Ooyama, Y. Kameshima, A. Nakajima, K. Okada, Instrument
various crystal structures, J. Eur. Ceram. Soc. 30 (2010) 583–589. for ceramic particle dispersion, Powder Technol. 200 (2010) 25–29.
[26] Z. Razavi Hesabi, M. Haghighatzadeh, M. Mazaheri, D. Galusek, S.K. [28] T. Isobe, Y. Hotta, Y. Kinemuchi, K. Watari, Preparation and mechanical
Sadrnezhaad, Suppression of grain growth in sub-micrometer alumina properties of the Al2O3 ceramics from wet-jet milled slurry, J. Ceram. Soc.
via two-step sintering method, J. Eur. Ceram. Soc. 29 (2009) Jpn. 115 (2007) 738–741.
1371–1377. [29] W.D. Kingery, Introduction to Ceramics, Wiley, New York, 1960.