Sintering of titania-silica powder compacts with a bimodal pore-size distribution

Sintering of titania-silica powder compacts with a bimodal pore-size distribution

]OURNA Journal of Non-Crystalline Solids 147&148 (1992) 582-587 North-Holland L OF NON-CRYSTALLINE SOLIDS Sintering of titania-silica powder compa...

1MB Sizes 1 Downloads 35 Views

]OURNA

Journal of Non-Crystalline Solids 147&148 (1992) 582-587 North-Holland

L OF

NON-CRYSTALLINE SOLIDS

Sintering of titania-silica powder compacts with a bimodal pore-size distribution W.T. Minehan, M.R. S c h a e f e r a n d G . L . M e s s i n g Department of Materials Science and Engineering, The Pennsylvania State Uniuersity, University Park, PA 16802, USA

The sintering of 7.3% TiO 2-SiO 2 powders that were synthesized by the spontaneous emulsification of metal alkoxides was studied. The particles were ~ 130 nm in diameter and consisted of primary particles of 32 nm diameter. The bimodat pore size distribution of powder compacts resulted in a two-stage sintering process with the individual particles sintering to full density at 1075°C and the interparticle porosity requiring 1200°C for complete removal. By using the compact density at the beginning of the interparticle stage of densification, it was determined that the sintering of the glass powders fits Scherer's open pore, viscous sintering model. The viscosities of the TiO2-SiO 2 glass derived from the model are in reasonable agreement with literature reports for titania-silica glasses.

1. Introduction Glass p o w d e r compacts densify by viscous flow at t e m p e r a t u r e s well below their melting point. Scherer [1] modified the M a c K e n z i e - S h u t t l e worth [2] closed pore densification model to account for o p e n porosity and non-spherical particles in an effort to derive a model m o r e applicable to p o w d e r c o m p a c t densification. T h e primary feature of the model is the description of the porous b o d y in terms of a cubic array of solid intersecting cylinders whose radii are e q u a t e d to the average particle size in the compact. Densification is related to a r e d u c e d time equal to K ( t - to), w h e r e t is the isothermal densification time, t o is a fictional time w h e n the cylinder radius is e q u a ! to zero, and K= (y/~

l

o)(Ps/Po)

1/3

,

(1)

w h e r e y is the surface energy, r/ is the glass viscosity, P0 is the initial c o m p a c t density, ps is the theoretical density, and l 0 is the initial cell dimension calculated from w d 2 / 4 = (I 0 - 2 a ) ! ,

(2)

where d is the interparticle p o r e diameter and a is the average particle diameter.

T h e application of this model to c o m p a c t densification requires the initial pore size of the compact, density and particle size. U n d e r isothermal sintering conditions, the c o m p a c t density at a given time is c o m p a r e d with the theoretical curve to obtain r e d u c e d time. T h e relationship b e t w e e n r e d u c e d time and sintering time is linear if the model applies over the entire densification range. If linear, a value for K can be d e t e r m i n e d and glass viscosity estimated as a function of sintering temperature. Scherer also derived a viscous densification model for compacts with bimodal pore distributions. H e predicted the densification of such compacts is governed by the large pore size distribution of the compacts [3]. J o h n s o n et al. [4] densified colloidal silica gel compacts p r e p a r e d by the double dispersion method. T h e p o r o u s aggregates pack to p r o d u c e large interaggregate porosity and thus a bimodal pore size distribution. In this case, rapid initial densification was followed by a slower densification rate. T h e y concluded that densification kinetics were not described by Scherer's model due to the presence of a large pore size distribution and the p o o r correlation b e t w e e n the actual p o r e g e o m e t r y and that assumed in the model.

0022-3093/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved

W. Minehan et al. / Sintering of titania-silica powder compacts

In addition to geometric and structural changes, several other factors affect sintering behavior, including atmosphere, residual carbon and crystallization. Water vapor can have a direct influence on the viscosity and surface tension of the glass [5,6] and thus its sintering behavior. If gas is entrapped during the closed pore stage of densification, its relatively slow diffusivity in the glass limits the attainment of full density. This paper presents the densiflcation behavior of porous TiO2-SiO 2 particles synthesized by the spontaneous emulsification of the metal alkoxides [7,8]. Because the particles are porous after synthesis, their compacts have a bimodal pore size distribution consisting of fine intraparticle pores and larger interparticle pores. The densification of these compacts is of general interest since most gel-derived particles [9] are porous.

2. Experimental procedure Particles were produced by spontaneous emulsification of partially hydrolyzed alkoxides as previously described [7]. The TiO2-SiO 2 powders with 7.3 wt% TiO 2 were synthesized by reacting a 0.944 : 1.3 : 9.0 : 0.0027 molar ratio of Si(OC2H5)4:-

583

H 2 0 : C 2 H s O H : H C I for 90 min in a closed container at 25°C before the addition of 0.056 mol titanium iso-propoxide. After an additional 120 min at 25°C, emulsification was induced by adding two volumes of 3% NH4OHaq to one volume of the above composition. Gelation of the droplets results in a 2.7 wt% dispersion of porous TiO 2SiO 2 particles. Although compacts can be produced by conventional methods from these powders [8] densification studies were performed on samples produced by colloidal gelation. By drying the colloidal dispersion, the particles pack to a homogeneous microstructure with a high packing density. The samples were dried under ambient conditions. For the sintering studies, all compacts were prepared from the same powder, since minor variations in the synthesis conditions can influence green body microstructure. Compacts between 0.2 and 0.4 g were isothermally sintered in a mullite tube furnace. Twenty to thirty samples were placed in a platinum boat and heated in dry, flowing air at 10°C/min. All heating schedules included a 2 h dwell at 900°C to allow structural relaxation prior to densification. Heating rates were resumed at 10°C/min to the isothermal sintering temperature. For sintering kinetics, two samples were withdrawn from

Fig. 1. SEM micrograph of porous particle compact.

W. Minehan et al. / Sintering of titania-silica powder compacts

584

the sample boat after it was briefly removed from the hot zone of the tube furnace• Because of the thermal mass of the samples and the ceramic boat, the relatively short time out of the hot zone and corrections for the time out of the hot zone, the quoted sintering times are reasonably accurate except for sintering times of < 1 min. Each density reported is the average of two samples. Samples were sintered in air at 1075, 1150, 1200 and 1250°C. A sintering time of zero is defined as the m o m e n t the furnace reached the sintering temperature. The bulk density of compacts was determined dimensionally, by mercury porosimetry and Archimedes' method. Surface areas were measured by single point BET. Pore size and the distribution were measured by mercury porosimetry, and particle sizes were determined from SEM and T E M micrographs. The initial skeletal density of the glass particles was measured by helium pycnometry.

3. Particle and compact characteristics The colloidal particles are composed of primary particles 32 nm in diameter [7,8] and have an initial skeletal density of 2.06 g / c m 3 (theoretical density of the titania-silica glass is 2.2 g/cm3). As determined by TEM, no residual gel from the synthesis process exists between the particles. The average particle size is 130 nm and the powder

100 -

unllU

nu u n .

u

1250°C • 1200°C

nn m

•~

Nm

90'

[]

• 80'

EI~I [] []

[3ran

Din[]

[] []

1175oc

N

Oo

[]

70"

• :.4; " e ' "' ~o7+°c

60'

50

i

i

100

200

300

Time (minutes) Fig.

2. C o m p a c t

density as a function

of sintering

t w e e n 1075 a n d 1 2 5 0 ° C .

time be-

350 300 ea~

250 -

g

200 -

<

150 m

100 rs3

mm

50-

rn

[]

0

i

i

100 200 Time (minutes) F i g . 3. R e l a t i o n s h i p

between

300

surface area and sintering time

at 1075°C.

surface area is 310 m 2 / g . The particles have a density of 1.3 g / c m 3 or ~ 63% of theoretical and randomly pack in the colloidal compact to a bulk density of 0.90 g / c m 3 ( ~ 44% of theoretical based on a skeletal density of 2,06 g / c m 3) (fig. 1). For later comparison, if the particles were fully dense glass and there was no interparticle sintering, then the compact would have a density of 1.52 g / c m 3 or 69% of theoretical density. The compact has a bimodal pore size distribution with modal pore diameters of 6 nm and 50 nm due to the intraparticle and interparticle pores, respectively.

4. Sintering kinetics Figure 2 shows that rapid densification occurs at >~ 1200°C, while lower t e m p e r a t u r e sintering appears to occur in two stages as evidenced by the slope change in the densification data. The kinetics of surface area reduction of the compacts at 1075°C are plotted in fig. 3. The loss in surface area to ~ 21 m 2 / g after 100 min corresponds approximately to the surface area for dense or closed porosity particles. In fig. 2, the compact is seen to be ~ 70% dense for this sintering conditions. SEM micrographs of samples after sintering for 160 min at 1075°C show no neck formation between the individual particles. The T E M micrograph (fig. 4) of particles sintered for 105 min at 1075°C clearly demonstrates that the particles are dense at this stage. From these data, we

W. Minehan et al. / Sintering of titania-silica powder compacts

585

probably a result of particle dehydration during the 2 h isothermal hold at 900°C. During the second stage of sintering, compact densification occurs solely by interparticle pore shrinkage. SEM analysis of compacts at 84% density show them to be primarily composed of a matrix of fine porosity with a few large (0.1 Ixm) elliptical pores. At this stage, the particles have undergone significant necking. As densification continues to 96% relative density, fine pores are eliminated and the large pores reduced in size, but are still irregularly shaped. The larger pores may be due to agglomerates formed prior to compact gelation. Because of the long sintering times at lower temperatures, crystallization prevented full densification (i.e., 100%) below 1200°C, but at higher temperatures, compacts were sintered to transparency and densities > 99.5%.

Fig. 4. TEM micrograph of the titania-silica glass particles after sintering at 1075°C for 105 min.

conclude that densification at 1075°C can be attributed solely to intraparticle sintering. Due to the shrinkage of the particles, the pore size distribution of the compact becomes distinctly narrower and the median pore diameter decreases to 21 nm. At 1150°C, the slope change occurs at ~ 75% relative density instead of 70% because of interparticle densification. The slope change is expected because interparticle densification kinetics are slower as a result of the larger pore size ~than intraparticle densification. Above 1200°C, no densification rate change is obvious due to the rapid and complete shrinkage of intrapartMe porosity during the heating of the compact to the sintering temperature. Thus, at these temperatures, the intraparticle porosity is removed within times too short to measure using individual samples. It is interesting to note that, at these temperatures and rapid sintering rates, there was no evidence for trapped gases or gas diffusion-limited densification of the individual particles. This is

5. Densification analysis and viscosity determination

Densification kinetics were analyzed with Scherer's open pore model. Reduced time for each sintering condition was used to derive the relationship for each temperature, as presented in fig. 5. As shown above, the interparticle and intraparticle densification stages increasingly overlap with increasing temperature. However,

2.8

[]

,

~

1250°C



~" 2.6 • 2.4 ~



1200°C

•••

OO

n •• •

U• ••

2.2

mmmmm IN



ZxZ~ A

A A A A/~ZX

2.0

0

50

A

1150oc aaa a 10750C

A

100 150 200 Time (minutes)

250

300

Fig. 5. Reduced time as a function of time for densification of compacts between 1075 and 1250°C.

586

W. Minehan et al. / Sintering of titania-silica powder compacts

after the slope change, densification only occurs by interparticle sintering. The linear relation between the reduced time and sintering time at 1075°C indicates that Scherer's model is applicable for the intraparticle stage of densification. Unfortunately, we do not have other data at this or lower temperatures and, thus, could not further analyze the process of intraparticle densification. At 1200 and 1250°C, intraparticle densification is so rapid that it is complete within the first 10 rain. Thus, the linear relation between reduced time and sintering time at 1200 and 1250°C is for interparticle densification alone. These results imply that densification of the porous particle compacts can be modeled as two separate compact geometries, correlating to the densification of the individual porous particles and the densification of the compact. Unfortunately, the rapid sintering kinetics of the intraparticle porosity and the experimental overlap of the two stages at higher temperatures complicate the analysis of the interparticle porosity with Scherer's model. Clearly, dilatometer experiments would be more precise in separating the two sintering stages. The second densification stage is solely due to densification between particles, and, thus, data at 1200 and 1250°C can be analyzed to derive glass viscosity. It is assumed that the initial stage of sintering has no effect on the second stage kinetics because the particles are completely dense after the first stage. The initial physical parameters used in the viscosity analysis are as follows: l 0 is 259 nm and P0 is 75% of theoretical density. The surface tension was assumed to be 280 d y n / c m , the same as silica, since the presence of titania in the glass has little influence on the surface tension [5]. From the slope, K, the calculated viscosities are 2.4 X 10 l° Poise at 1200°C and 5.2 X 10 9 Poise at 1250°C. Corning Inc. reports a temperature difference of 80-85°C in the viscosity curves between titania-silica and silica glasses over the temperature range of 1000 to 1500°C [10]. The viscosity curve for fused silica differs from our calculated viscosities by 55-145°C. Therefore, the experimentally derived viscosities are within the expected viscosity range of the glass.

Using data for only interparticle densification, the activation energy, E, for viscous flow between 1200 and 1250°C was calculated from ln~7 = ln~70 +

(E/kT).

(3)

The activation energy for viscous flow during the second stage of sintering is estimated to be t36 kcal/mol. This correlates well with activation energies between 170 k c a l / m o l and 122 k c a l / m o l reported by Hetherington et al. [6] for silicas produced by various techniques.

6. Conclusions

The densification kinetics of compacts produced by random packing of porous titania-silica spherical particles has been characterized. The porous particles result in an initial bimodal pore size distribution with modal sizes centered at ~ 6 nm and 50 nm, and consequently two distinct regions of densification. The initial stage of sintering involves intraparticle densification and the second stage is due to the densification of the dense particles. The second stage of densification has been analyzed by Scherer's open pore model for viscous sintering. Calculated viscosities ranged from 2.4 × 10 l° Poise at 1200°C to 5.2 × 109 Poise at 1250°C. These values correlate well with documented values for titania-silica glasses [10]. These results clearly demonstrate that porous particle sintering can be analyzed with Scherer's model by simply renormalizing the sintering data for the onset of the second stage of densification between the dense particles. Our results suggest that a reassessment of the densification kinetics of double dispersed gels by first eliminating the intraparticle sintering stage and then renormalizing the density to take into account the densification of the aggregates would also result in good agreement with the Scherer model.

References [1] G.W. Scherer, J. Am. Ceram. Soc. 60 (1977) 236. [2] J.K. MacKenzie and R. Shuttleworth, Proc. Phys. Soc. 62 (1949) 833.

IV.. Minehan et al. / Sintet~ng of titania-silica powder compacts [3] G.W. Scherer, J. Am. Ceram. Soc. 67 (1984) 709. [4] D.W. Johnson Jr., E.M. Rabinovich, J.B. MacChesney and E.M. Vogel, J. Am. Ceraxn. Soc. 66 (t983) 683. [5] N.M. Parikh, J. Am. Ceram. Soc. 41 (1958) 18. [6] G. Hetherington, K.H. Jack and J.C. Kennedy, Phys. Chem. Glasses 5 (1964) 130. [7] W.T. Minehan and G.L. Messing, J. Non-Cryst. Solids 121 (1990) 375.

587

[8] W.T. Minehan, G.L. Messing and C.G. Pantano, J. NonCryst. Solids 108 (1989) 163. [9] G.L. Messing and W.T. Minehan, J. Ceram. Soc. Jpn. 99 (1991) 1036. [10] D.C. Boyd and D.A. Thompson, in: The Encyclopedia of Chemical Technology, Vol. 11 (Wiley, New York, 1980) p. 8O7.