Full text of DOI:10.1002/1527-2648(200011)2:11<741::AID-ADEM741>3.0.CO;2-U

Raether, Springer/In-situ Measurement of Neck Formation During Sintering of Alumina
[
[
[
16] D. C. Dunand, A. M. Jansen, Acta Mater. 1997, 45, 4569.
17] A. M. Jansen, D. C. Dunand, Acta Mater. 1997, 45, 4583.
18] A. Dlouhy, N. Merk, G. Eggeler, Acta Metall. Mater.
ing. Different mechanisms for mass transport have been pro-
posed such as surface diffusion, grain boundary diffusion,
[
7]
and volume diffusion. A general critical assessment of ki-
[
8]
1
993, 41, 3245.
netic sintering models has been given by Exner. Today the
importance of the heating cycle has been recognized and it is
believed that in-situ measuring methods are best suited to
overcome the simplifications of the models and to describe
the development of microstructure in real powder com-
[
[
19] Y. Li, T. G. Langdon, Acta Mater. 1997, 45, 4797.
20] Duralcan Composites for Wrought Products: Mechanical
and Physical Property Data. Duralcan USA, San Diego,
CA 1993.
[
9]
[
21] E. Gariboldi, M. Vedani, in Proc. Advancing with Com-
posites '97 (Ed.: I. Crivelli Visconti), Milan 1997, p. 29.
22] P. E. Krajewski, J. W. Jones, J. E. Allison, Metall. Trans.
pacts. Dilatometry is the customary in-situ method for in-
vestigating sintering processes: porosity is calculated from
the measured shrinkage. Since surface diffusion causes neck
growth but does not contribute to shrinkage, it cannot be de-
tected by dilatometric measurements. The change of specific
surface area has been measured additionally to shrinkage to
obtain full information: neck formation reduces the surface
[
1
995, 26A, 3107.
Received: January 03, 2000
Final version: July 04, 2000
[
7]
area at the particle contacts. From those investigations it has
been concluded that a mixture of surface diffusion and grain
boundary diffusion determines microstructure development
during sintering of alumina where surface diffusion domi-
In-situ Measurement of Neck
Formation During Sintering of
Alumina by a Novel Thermo-
Optical Measuring Device**
[
7]
nates in the formation of sintering necks.
The specific surface area can only be measured after an in-
terruption of the heating cycle. Another disadvantage is that
a change in composition and roughness of the particle sur-
faces during heating may influence the result. It has already
been shown that thermal diffusivity and shrinkage can be de-
tected in-situ using a novel thermo-optical measuring device
By Friedrich Raether* and Ronald Springer
[
10]
called TOM. Since heat transfer in the initial sintering stage
is mainly restricted by the thermal resistance at the particle
contacts, thermal diffusivity strongly increases with neck for-
Alumina is the most important material used in technical
[
1]
ceramics: its market share is nearly 40 % worldwide.
Strength, hardness, and chemical resistance are the decisive
properties of Al O ceramics. Nowadays under suitable man-
ufacturing conditions bending strengths of much more than
4
example cracks, have to be avoided carefully: in particular
the forming and sintering procedure have to be optimized.
To establish a stable sintering process other problems have to
be solved. The green parts pass through a strength minimum
during heating when binder has been burned out and sinter-
ing necks are small, this can lead to fracture when forces act
in this period, for example gravity acting on large parts. Prop-
erties of alumina ceramics may vary with the batch of raw
material used or show seasonal fluctuations: the reasons are
[
10]
mation.
Using computer simulations it has been shown
2
3
that other heat transfer mechanisms like conductivity within
[
11,12]
[
2]
the pores and heat radiation can be neglected,
and the
00 MPa are achieved. To this end, structural defects, for
neck radius has been estimated from thermal diffusivity
[
13]
data. Sintering of different alumina powders has been in-
vestigated in the present study using the TOM method to de-
velop a better understanding of parameters influencing neck
formation in the initial sintering stage.
[
3]
Green samples have been prepared from seven high purity
alumina powders from different manufacturers. The cylindri-
cal silicon molds were filled in small portions, each carefully
pressed by a spatula. Compacts were cold isostatically
pressed for 3 min at a pressure of 250 MPa. No additives
were used. The cylindrical samples (diameter 12 mm and
height 30 mm) were cut into disks of 1.5 mm thickness with a
diamond saw. The green compacts had been stored prior to
sintering either in a drying chamber at a temperature of 90 ꢀC
or in air at 100 % humidity at room temperature for one week.
Fractions of powder A had been pretreated before compac-
tion in two different ways: stirring in 25 % HCl solution for
[
4]
often not well understood.
Sintering properties of alumina have been investigated ex-
tensively for more than four decades. Kinetic models have
been developed for neck growth, shrinkage, and grain
growth in the initial, intermediate, and final sintering
[
5,6]
stages.
The models consider an idealized microstructure
and have been applied mainly to constant-temperature sinter-
2
3
h and washing in deionized water until pH 7, stirring in
% KOH solution for 2 h and washing in deionized water un-
[
*] Dr. F. Raether, R. Springer
Fraunhofer-Institut Silicatforschung (ISC)
Neunerplatz 2, D-97082 Würzburg (Germany)
til pH 7.
Those fractions were dried, crushed in a mortar, and
sieved in a 100 lm mesh before compaction. The third frac-
tion of powder A was compacted as delivered like the pow-
ders B to G.
[
**] The authors gratefully acknowledge the help of P. Löschke,
N. Keidel, A. Heid, and Th. Freund with the measurements.
ADVANCED ENGINEERING MATERIALS 2000, 2, No. 11
1438-1656/00/1111-0741 $ 17.50+.50/0
741

Raether, Springer/In-situ Measurement of Neck Formation During Sintering of Alumina
Particle size distributions of the powders were measured
by a laser light scattering device (Malvern: type 4700) in a di-
lute aqueous solution after 10 min of soft ultrasonic agitation
detection limit of the XPS method. No Na was found on the
surface of powder D. (XPS-spectra of the two powders are al-
[
15]
ready published: powders A and D are denoted A respec-
tively B in this reference.) Further XPS-measurements demon-
strated that after both kinds of surface treatment (HCl and
KOH) Na has been removed from the particle surface.
Results of the TOM experiments with the seven powders
are shown in Figure 1: the compacts were stored in the dry-
ing chamber before heating. The corrected thermal diffusivity
increases between room temperature and 1000 ꢀC by a factor
of 5 to 10 during heating of the green compacts (Fig. 1a);
powder D shows the strongest rise. The onset of shrinkage is
between 900 and 1100 ꢀC: at much higher temperatures than
the rise of thermal diffusivity. Powder D sinters at the lowest
temperatures (Fig. 1b).
Figure 2a shows corrected thermal diffusivities during
heating of compacts of powder D prepared after dry and wet
storage of the compacts: sintering neck formation is equal in
both cases. Unlike powder D there is a distinct difference for
powder A: the compacts stored in humid air show a stronger
increase in thermal diffusivity. This difference becomes
smaller for the fractions of powder A that have been treated
in KOH and larger for those that have been treated in HCl
(Fig. 2b). From the dilatometric measurements it was found
that deviations in sintering shrinkage between the differently
treated powders are within the uncertainty of the measuring
method.
(
an increase of ultrasonic power leads to agglomeration). The
specific surface area was measured by nitrogen adsorption
using the Brunauer±Emmett±Teller (BET) method (Micro-
meritics: ASAP 2400) after heating the powders for 16 h at
1
10 ꢀC in vacuum. The particle surfaces of powder A (after
different pretreatment) and of powder D were investigated
by the XPS-method (X-ray photoelectron spectroscopy, Sur-
face Science Instruments: S-Probe).
Thermal diffusivity and shrinkage of the green compacts
were recorded simultaneously during the heating cycle using
the TOM device. Thermal diffusivity was measured by a laser
flash technique and shrinkage was measured with a non-con-
tact method by an optical dilatometer. Details of the TOM de-
[
14]
vice are given elsewhere.
The in-situ measurements were
-furnace using syn-
performed in a resistance heated MoSi
2
thetic air at a flow of 1 l/min. The green compacts were
heated to 1650 ꢀC at a rate of 5 K/min and immediately
cooled to room temperature thereafter. One densely sintered
sample of powder A has been heated a second time. Mea-
sured sample diameters and thermal diffusivities of all sam-
ples have been divided by the corresponding values mea-
sured at the same temperature with the dense sample.
Thereby the temperature influence on the material parame-
ters has been essentially eliminated and the change of micro-
structure can be seen more clearly. (Thermal diffusivity of
alumina strongly decreases with temperature due to increas-
ing phonon±phonon scattering.) The corresponding quanti-
ties are denoted corrected thermal diffusivity and corrected
diameter: they vary between 0 and 1 and have no dimension.
Density of sintered samples was measured by the Archime-
dean method.
Table 1 summarizes the powder properties and the sin-
tered densities of the alumina samples. The comparison be-
tween particle surface measured by BET and particle surface
calculated from particle size distribution indicates an agglom-
eration of the powders.
The XPS-measurement of the untreated powder A shows
the existence of Na in a thin layer on the particle surfaces.
The average concentration of Na in powder A was below the
Table 1. Particle size (d50), span ((d90±d10)/d50), specific surface area measured by BET
and calculated from particle size (sBET and scal) of powders A to G and sintered density
(r ) of the corresponding compacts.
f
Fig. 1. a) Corrected thermal diffusivity and b) corrected sample diameter versus tem-
perature during heating of green compacts in the TOM device.
742
ADVANCED ENGINEERING MATERIALS 2000, 2, No. 11

Raether, Springer/In-situ Measurement of Neck Formation During Sintering of Alumina
Fig. 2. Corrected thermal diffusivity versus temperature of green compacts a) of powder
D and b) of powder A after different pretreatments.
The strength of the compacts is relatively low at the onset
of shrinkage. Therefore, crack formation is likely when stress-
es occur during heating. Local stresses may also lead to desin-
tering and preferential densification of particle agglomerates:
Fig. 3. Ratio of shrinkage rate and approximate strength (square of corrected thermal
this results in large pores, which are difficult to eliminate in
diffusivity) versus temperature during heating of green compacts: a) compacts of differ-
ent powders stored in drying chamber, b) compacts of powder D stored under wet and
dry conditions and c) compacts of powder A after different pretreatments stored under
wet and dry conditions. (Data have been smoothed.)
[
16]
the final sintering stage. The stresses are caused by differ-
ential sintering due to density or temperature gradients or lo-
cal density fluctuations; they can also be caused by an unfa-
vorable shape or by using composite materials with different
sintering shrinkage or thermal expansion.
It was shown in a previous paper that the increase of thermal
diffusivity during heating of alumina green compacts is caused
[
10]
by the formation of sintering necks. Thermal diffusivity is
roughly proportional to the neck radius and strength is roughly
proportional to the square of the neck radius in the initial sinter-
ing stage. A relation has been established between the square of
the relative neck radius derived by thermal diffusivity mea-
surement and the tensile strength of calcined alumina green
[
10]
compacts, which is approximately proportional. All powders
show a maximum in this stress±strength ratio at temperatures
between 1100 and 1200 ꢀC (Fig. 3). The magnitude of the stress±
strength ratio varies between 0.8 and 1.6 %/min; temperature
and steepness of the leading edge of the curves are very differ-
ent, too. Large differences in the tendency for crack formation
are expected for the powders from the considerations above.
Owing to the absence of shrinkage it is quite clear that a
surface diffusion mechanism is responsible for the neck for-
mation below 1000 ꢀC in the present investigation. But the in-
crease of thermal diffusivity at very low temperatures (below
Fig. 4. Corrected thermal diffusivity of green compacts during heating (measured at
different temperatures) versus specific surface area of the respective powders
tion between thermal diffusivity of the green compacts dur-
ing heating at different temperatures and specific surface
area of the respective powders: thermal diffusivity is higher
for those powders with a higher specific surface area reflect-
ing enhanced neck formation. Alumina powder particles are
[
17]
covered with a thin layer of hydrous alumina species. It is
assumed that material flow at low temperatures is within this
disturbed surface layer where the activation energy will be
very low. Neck volume at particle contacts between convex
surfaces is initially very small: a surface layer with a thickness
5
00 ꢀC) cannot be caused by the surface diffusion of crystal-
line a-alumina species: the activation energy needed for this
[7]
process is too high. So the question is: what enhances sur-
face diffusion in those powders? Figure 4 shows the correla-
ADVANCED ENGINEERING MATERIALS 2000, 2, No. 11
743

Raether, Springer/In-situ Measurement of Neck Formation During Sintering of Alumina
[
23]
of 0.3 nm on spherical particles with a particle radius of
00 nm is sufficient to provide enough material for filling a
surement has been combined with the TOM device.
So a
2
complete picture of microstructure evolution during sintering
can be obtained using a single instrument.
[
18]
sintering neck with a relative radius of 25 % (compare ).
Therefore even monolayers at the particle surface can attrib-
ute significantly to neck growth. A humid furnace atmos-
phere enhances neck formation as was shown in a previous
[
[
1] K. M. Nutter, CIMTEC '98 ± World Ceramic Congress
Ed.: P. Vicenzini) Florence, Italy 1998, Techna Srl.,
999.
2] E. Dörre, H. Hübner, Alumina - Processing, Properties
and Applications, Springer, Germany 1984.
3] R. Lucke, W. Hennicke, cfi/Ber. DKG 1993, 70, 75.
4] N. Shinohara, M. Okumiya, T. Hotta, K. Nakahira, M.
Naito, K. Uematsu, Am. Ceram. Soc. Bull. 1999, 78, 81.
5] S. Prochazka, R. L. Coble, Phys. Sint. 1970, 2, 15.
6] R. L. Coble, J. Appl. Phys. 1961, 31, 787.
[
10]
paper; on the contrary a calcination of the alumina powder
(
1
[
19]
decreases neck formation. Storage of the green compacts in
a moist atmosphere prior to sintering can also promote neck
growth but has no influence for powder D (compare Fig. 2).
Accordingly, also the stress±strength ratio is influenced by
the storage conditions of powder A but not of powder D
[
[
(
Fig. 3b and 3c). Since the enhancement of neck growth due
to water vapor was related to the existence of sodium con-
tamination on the powder surface, it is assumed that sodium
plays an important part in the mechanism. In a previous IR
investigation a high fraction of OH-bonds not connected by
hydrogen bonds was detected in powder A but was absent in
[
[
[
7] S. H. Hillman, R. M. German, J. Mater. Sci. 1992, 27,
2
647.
[
[
8] H. E. Exner, Powder Met. 1980, 4, 203.
[
15]
powder D: this supports the hypothesis that hydrous alu-
mina species are decisive for the different sintering behavior
of the two powders. Sodium has been removed by the surface
treatment of powder A. Thereafter, the influence of water va-
por during storage of the powder on sintering neck formation
was still present (Fig. 2b and 3c) so differences in surface re-
activity between the two powders have survived the surface
treatment.
9] W. Schatt, Sintervorgänge, VDI, Düsseldorf 1992.
[
10] F. Raether, J. Zimmer, R. Springer, 9th CIMTEC: World
Ceramics Congress, Ceramics: Getting into the 2000's ±
Part B (Ed.: P. Vicenzini), Techna Srl 1999, p. 711.
11] O. Hahn, Models for the Heat Transport in Sintering Ce-
ramic Powder Compacts, Universität Würzburg 1996
12] O. Hahn, F. Raether, M. C. Arduini-Schuster, J. Fricke,
Int. J. Heat Mass Transfer 1997, 40, 689.
13] O. Hahn, F. Raether, J. Fricke, in Proceedings Werkstoff-
woche 96, Vol. 8 (Ed.: J. Hirsch), DGM-Informationsge-
sellschaft, Frankfurt, Germany, 1997, p. 239.
[
[
[
The differences in sintering neck formation between the
seven powders underline that microstructure evolution of
real compacts cannot be predicted by simple geometric mod-
els. There is a considerable influence of surface chemistry on
mass flow at the particle surfaces. Highly mobile hydrous
alumina species probably play a decisive role here. Pre-
viously it was demonstrated that a holding period at a tem-
perature before the onset of shrinkage can prevent crack for-
[
[
14] F. Raether, R. Hofmann, G. Müller, H. J. Sölter, J.
Therm. Anal. 1998, 53, 717.
15] H. Böse, TIZ Baustoff-Technik & Recycling Praxis 1996,
1
19, 9.
[
[
[
[
16] J. Zheng, J. S. Reed, J. Am. Ceram. Soc. 1989, 72, 810.
17] D. H. Lee, R. A. Condrate Sr., Mater. Lett. 1995, 23, 241.
18] E. Geguzin, Physik des Sinterns, VEB, Leipzig 1973.
19] R. Hofmann, Entwicklung einer Thermo-Optischen
Messnlage (TOM) zur in-situ Charakterisierung von Sin-
termaterialien, Universität Würzburg 1997.
[
10]
mation at higher temperatures. The benefits of this holding
[
20]
period have also been pointed out from theoretic models.
Surface chemistry may explain some of the process variations
observed in industrial sintering of alumina. So humidity dur-
ing storage of powder and green compacts and also surface
impurities like Na have to be carefully controlled. Although
in the initial sintering stage a beneficial influence of enhanced
surface diffusion is expected, in the intermediate sintering
[
20] J. T. Lin, L. C. De Jonghe, J. Am. Ceram. Soc. 1997, 80,
2
269.
[
21] R. R. Walker, J. Am. Ceram. Soc. 1955, 38, 187.
[
21]
stage water vapor has a detrimental effect on densification
[
22] J. Manara, R. Caps, F. Raether, J. Fricke, Opt. Commun.
probably because coarsening outweighs densification. To im-
prove sintering behavior, surface diffusion at low tempera-
ture may be supported by hydrolysis of the alumina powders
and a controlled usage of catalytic materials like Na.
1
999, 168, 237.
[
23] R. Springer, F. Raether, R. Caps, J. Manara, High Temp.-
High Press. 2000, 32, 385.
Received: July 07, 2000
In the final sintering stage shrinkage rate decreases and
[
16]
average pore size increases.
Analog to sintering neck for-
mation pore size cannot be determined from dilatometric
measurements. So another in-situ method has been tested:
light transmission measurements at high temperatures. Scat-
tering coefficients depend on the pore size and there is a
strong influence of pore size distribution on light transmis-
______________________
[
22]
sion data in alumina ceramics. The light transmission mea-
744
ADVANCED ENGINEERING MATERIALS 2000, 2, No. 11