Preparation of 33 mol% Na2)-67 mol% SiO2 glass by gel-glass transformation

Preparation of 33 mol% Na2)-67 mol% SiO2 glass by gel-glass transformation

Journal of Non-Crystalline Solids 53 (1982) 183-193 North-Holland Publishing Company 183 P R E P A R A T I O N OF 33 MOL% N a 2 0 - 6 7 MOL% SiO 2 G...

465KB Sizes 0 Downloads 12 Views

Journal of Non-Crystalline Solids 53 (1982) 183-193 North-Holland Publishing Company

183

P R E P A R A T I O N OF 33 MOL% N a 2 0 - 6 7 MOL% SiO 2 GLASS BY GEL-GLASS

TRANSFORMATION

L a r r y L. H E N C H Ceramics Division, Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA

M. P R A S S A S a n d J. P H A L I P P O U Laboratory of Materials Science and CNRS Glass Laboratory, University of Montpellier IL Place Eugene Bataillon, 34060 Montpellier Cedex, France

Received 15 June 1982

It is difficult to prepare gels of the binary Na20-SiO 2 system because the presence of OH- ions promotes quick and controlled gellation. A reproducible method for obtaining monolithic gel samples of 33 mol% Na20-67 mol% SiO2 (33N) is described based upon hydration and polycondensation reactions of SiO(CH3h and NaOCH 3. Conversion of the dried gels to glasses and glass-ceramics is followed by the evolution of physicochemical properties and by comparing infrared reflection structural features of the gel-derived glasses with conventional melt-derived glasses. Densification in gel-glass transformation in the range from 490-520°C produces non-porous glass monoliths with density and structure equivalent to melt-derived 33N glasses. At higher temperatures (520°C) uncontrolled crystallization occurs.

1. Introduction A n u m b e r of investigators (e.g. [1-7]) have d e m o n s t r a t e d that glasses can be p r e p a r e d from gel-precursors. However, the a b o v e studies have c o n c e n t r a t e d p r i m a r i l y on the p r e p a r a t i o n of glasses very high in network f o r m e r c o n t e n t a n d / o r have required the gels to be densified b y hot pressing or cold pressing a n d sintering of gel powders. T h e objective of the p r e s e n t investigation is to p r e p a r e s o d a - s i l i c a t e glasses. P r i m a r y e m p h a s i s in this p a p e r will b e d e v o t e d to the processing a n d g e l - g l a s s t r a n s f o r m a t i o n of a 33 mol% N a 2 0 - 6 7 mol% SiO 2 (33N) glass. M i x t u r e s of organometaUic liquids are used as the starting m a t e r i a l s for p r e p a r a t i o n of the gels. Careful control of d r y i n g a n d firing variables are necessary in o r d e r to prevent fracture, bloating, or crystallization before or d u r i n g the g e l - g l a s s t r a n s f o r m a t i o n for the 33N c o m p o s i t i o n . T h e effect of these variables on other c o m p o s i t i o n s in the N a O 2 - S i O 2 system will b e discussed in a later p u b l i c a t i o n .

0022-3093/82/0000-0000/$02.75

© 1982 N o r t h - H o l l a n d

L.L. Hench et aL / Preparation of N a 2 0 - S i O 2 glass

184

2. Gel preparation

Chemical-grade Si(OCH3) 4 was mixed at 0°C with CH3ONa (30% in CH3OH solution) and excess CH3OH to make a solution containing 45% organometallics. The proportions of the two organometallics were selected to yield 33 mol% N a 2 0 - 6 7 mol% SiO 2 after loss of water and excess methanol. Distilled H 2 0 was added to the mixture at a rate of approximately 7 drops per 5 s while continuing to stir vigorously and maintaining the solution at 0°C. After the final cm 3 of n 2 0 was added the solution was stirred for an additional one minute before pouring into a Teflon-lined mould. Gelling time was monitored by establishing when the surface could no longer be indented with minimal pressure. Gelling time is a very strong function of the quantity of water added to produce the hydrolysis and polycondensation reactions that take place as the gel forms, i.e. Hydrolysis reaction Si(OCH3) 4 + 4 H 2 0 ~ Si(OH)4 + 4 CH3OH;

(1)

Polycondensation reaction OH

OH

t

H O - Si

J

- O H + H O - Si

f

I

I

OH

I

- O H ---, - S i - O - S i - + H20.

I

OH

(2)

I

Fig. 1 shows the time for gellation as a function of the amount of water added per mol of Si(Ofn3) 4 for the 33N composition. The quantity of water added

2 I

6

~

3

~

2

0

I 4

A

I 6

mole H20 / mole Si(OCH3) 4

3 1

/ ~ e ~ 8

J

t 10

cc H20

Fig. 1. Effect of m o l a r r a t i o o f H 2 0 / S i ( O C H 3 ) 4 a n d w a t e r c o n t e n t o n g e l a t i o n time for a 3 3 N gel.

L.L. Heneh et al. / Preparation of N a 2 0 - SiO 2 glass

185

per 25 cm 3 of Si(OCH3) 4 is also indicated. The quantities shown are much less than necessary to achieve stoichiometric hydration, e.g. eq. (1). In fact, the hydrolysis reaction is not complete. Therefore the product of reaction (1) should be written Si(OH),,(OCH3)4_,~ where n is a variable depending on the kinetics of the hydrolysis reaction and the amount of water added. Consequently the polycondensation reaction is more accurately written as

2 Si(OH). (OCH 3 )4--n ( O C H 3 ) a - . ( O H ) ~ _ , - S i - O - S i - (OCH 3)4_ o(OH).--I

+

H20.

(3)

With the addition of CH3ONa to the solution, both the hydrolysis and the polycondensation reactions become more difficult to write [2]. It is possible that Na ÷ can replace either the organometallic radical or the hydroxyl radical in the product of eq. (3). At this stage there is insufficient evidence to specify which of these sites are occupied by the alkali ions. A further complication in defining the product of gellation is that some fraction of CH3ONa may not enter into the gel structure but remain as an imbedded liquid within the pores of the gel. Also, some of the Na + may be in the form of a dissolved salt in the water. Fig. 1 is the first of the processing curves required to form glass objects from the organometallic solution. By adding water to the solution to achieve a ratio of n(H20)/n(Si(OCH3)4) between 2.0 to 2.3, it is possible to control gelling time between 15 and 60 min, ample time for casting complex shapes or to initiate drawing operations. Experience shows that it is generally easier to maintain large shapes of gel product without cracks using the higher ratios of n(H20)/n(Si(OCH3) 4). 3. Drying

Drying of the gel can be divided into two stages: Stage l: Evaporation of the liquid inside the pores of the gel. This liquid is primarily excess CH3OH. It has a high vapor pressure at room temperature and can evaporate very rapidly, producing large drying stresses. Stage 2: Elimination of organometallic residues by oxidation. This process can be described as: [ - Si -OCH

j

3 +

0 2 ~

-

I Si - O H + CO 2 + CO + H 2 0 i

(4)

followed by a secondary condensation reaction (eq. (2)). Thus control of gel drying requires maintaining the proper combination of temperature, time, and atmosphere to permit each stage of reaction to proceed at a rate which will not lead to drying stresses so large as to fracture the material.

L,L. Hench et al. / Preparation of N a 2 0 - S i O e glass

186

The first stage of drying was controlled for the 33N gels by keeping them in a container closed with a membrane that was semipermeable to CH3OH. Very slow evaporation of CH3OH occurred from the gel for 30-60 days at room temperature. This changed the gel from a transparent solid to a white opaque one as the evaporating liquid left behind pores large enough to scatter light. As reported in another publication [14], the alkali ions also reacted with CO 2 to form carbonate groups in the surface of the pores during this period of drying. There were sufficient volatile species remaining in the room temperature dried gel that rapid heating to the gel-glass transformation range could produce entrapment of the gases and bloating. This problem was eliminated by employing a second drying step at 90°C. Fig. 2 shows the change in weight and volume occurring during the 90°C drying. There is a rapid loss of approximately 15% weight accompanied by about 13% shrinkage within the first 30 min. An additional 4% weight loss occurs over a period of several days with very little additional shrinkage. In subsequent firing studies a second 90°C drying for more than 24 h was used to compare with the room temperature dried samples to determine the effects of drying on densification and the gel-glass transformation.

4. Firing Exposure of the dried gel to high temperatures results in several simultaneous processes: oxidation and condensation reactions (eqs. (3) and (2)), transport and evaporation of the reaction products, and pore shrinkage and closure. If these competing processes occur at a temperature where the re-

30 A

c:

o t,)

weight

20

....

g

-~,. - -~- -~- - ~ - - ~ -

volume

change

~D c h a n g e

. . . . . . . .

-

_

'~D--

I0

0

II

I 5

I I0 Drying

t 50 Time

Fig. 2. Drying behavior of 33N gel at 90°C.

(hours)

i I00

I

500

IOtO0

L.L. Hench et al. /

Weight

Volume

Change

--0-- no drying

IOO

187

o f N a 2 0 - S i O 2 glass

Preparation

Change

no drying ----E3-- with 90eC

drying

90 Gloss Gloss- Ceramic

Gel ~

"c LE i

80 70

,0,°o;? ,,

60 50

Z~

40 J"

50 o

weight

20

O O

IO

0"

O @

~-O

i I00

0

I~

i 200

i 500

Firing

I 400

I 500

I 600

Temperoture (*C)

Fig. 3. Gel-glass transformation curve for 33 mol% N a 2 0 - 6 7 mol% SiO 2 (33N) samples.

sultant dense solid remains amorphous, a gel-glass transformation is considered as achieved. If the processes occur such that crystallization is not prevented a polycrystalline glass-ceramic results. The gel-glass transformation curve for 33N is shown in fig. 3. Volume and weight changes were determined on small ( - 1 cm × 0.7 cm x 3 cm) samples

------------ 5 4 0 = C

~

--540=C

o

~_ _

iiiii 4

7

0

>-_,9_,4__ 5o o"f o,cP

o

c

~, ,zOO

~

520 ° C

f

~

470°C

~

~

, IO00

Wovenumber

I a00

i

-~ I 600

L

I 400

After 25%, ~90°C, 400o¢

(cm-I)

Fig. 4. Infrared reflection spectra comparing 33N melt glass with 33N gels after 1 h firing at the temperatures indicated.

188

L.L. Hench et al. / Preparation of N a 2 0 - S i O ~ glass

f

i

\ .....~

~ "

"

"4

,,

"

~

r..

25 ° C

1700

1600

1400

1200

1000 WAVE NUMBER (cm 1 )

800

600

400

200

Fig. 5. Infrared transmission spectra of 33N dried gels and gel-derived glasses after I h firing at temperatures indicated.

of dried 33N gel heated for 1 h at the temperatures indicated. This sample size is generally small enough to prevent bloating even without 90°C drying. The solid lines summarize the data obtained on samples put directly into the furnace without a previous 90°C drying step. The dashed line shows data for samples with the secondary 90°C drying. Fig. 4 shows infrared reflection spectra (IRRS) obtained from the samples after the one hour heating at the temperatures indicated. Previous studies [8,9,10] have shown the use of IRRS as a quantitative tool for following the compositional dependence of glass surface reactions. IRRS is especially useful in the present study because it can follow the structural changes accompanying the gel-glass transformation as well as crystallization of the material. Furthermore, IRRS can be used directly on the small samples that result from the almost 80% volume shrinkage (fig. 3). Although spectra could often be obtained from the smooth as-fired surfaces, all samples were ground and polished sequentially with 400, 600, and 1000 grit dry SiC paper to ensure an equivalent reflecting surface and to avoid surface "skin" artifacts. The gel before and after drying at 90°C, or after heating to temperatures up to 470°C shows no IRRS peaks (fig. 4). This is due to the extensive scattering of the incident beam by the porosity in the material and the carbonate phase formed on the surface of the pores. Heating for 1 h in the gel-glass transfor-

L.L. Hench el al. / Preparation of N a 2 0 - SiO 2 glass

189

ioo 90 490"C

4~0 ° C

~

70

t,.

,,

60

~

5O

.540 *C~ / ~.

- - -4Z9 *-c- -~ -~- . . . . .

3o

eight 0

'\

IO

o

-o-

chanQe

'

i

I

I

2

3

2'z

Firing Time (hours) Fig. 6. Effect of temperature and time on densification of 33N gels.

mation range, 480-520°C, however, results in an IRRS spectrum (fig. 4). The IRRS spectrum for 33N glass is characterized by three dominant peaks [8]. The peak at approximately 1080 cm -1 is due to S i - O - S i stretching vibrations. These molecular vibrations are located at a lower wavenumber than

'" Vitreous Silica

80

60 ,i

3 c

40 \

:•r~ 20

! /~

~

..

~ /---'~22hr x~ -

,

'.' ,

~-. ~

. ..

I 1200

1000

800

Wovenumbers

Fig. 7. IRRS spectra characteristic o f a 33N 40 min or more.

600

400

(crn-I)

glass is developed by heating a 33N gel at 490°C

for

L.L. Hench et al. / Preparation of Na:O-SiO 2 glass

190

the same mode for vitreous SiO2 (see fig. 7) because the silica tetrahedra in the 33N glass also contain silicon-nonbridging-oxygen (NBO) bonds. The NBO bonds in the glass are associated with alkali as S i - O - N a ÷ and their molecular stretching vibration gives rise to the second peak of the 33N IRRS spectra located at 950 cm-1. The third peak at 450 cm-1 is due to S i - O - S i "rocking" vibrations. At 480°C, the onset of the gel-glass transformation, all three IRRS peaks have developed. However, the separation of the spectra into two well defined S i - O - S i and S i - O - N a + peaks characteristic of 33N glass is not completed. For comparison, the IRRS spectrum of 33N glass melted and cast using a normal molten glass preparation technique is shown as a dashed curve in fig. 4. Heating the gel to 490°C for 1 h is sufficient to complete the transformation and the spectra for the melt glass and gel-derived glass are nearly identical. The same is true for 500°C. However, the 1 h heat treatment at 520°C begins to show a difference from the 480-500°C spectra. This difference is associated with the onset of crystallization in the sample. Crystallization has proceeded so far to completion in the 540°C sample that no IRRS spectrum is observed. Similar results were obtained at 625°C. The loss of reflection intensity from crystallization is due to the scattering from the large crystals and porosity accompanying the uncontrolled crystal growth. Additional information on the chemical-structural changes occurring during the gel-glass transformation is obtained using IR transmission (fig. 5). The spectra shown in fig. 5 were made with the KBr pellet technique using powders

80 :>,

D "6 n-

60

40

tn oc rr .~55N

Melf

GlOSrrSn

. . . . -.

20

:,

1200

" ~ ........

I000

__~ .....

800

Wovenumbers

(cm-O)

Fig. 8. Changes in IRRS spectra due to heating a 33N gel at 540°C.

600

_:2'

400

L.L. Hench et at,. / Preparation of Na : 0 - SiO e glass

191

prepared by crushing and grinding the same samples used to establish figs. 3, 4 and 9. All spectra of the gels show peaks at 1430-1440 c m -1 and 860 cm -I which have been attributed to the presence of Na2CO 3 H20, NaHCO 3 and N a E C O 3 species in the dried 33N gel [14]. For samples dried at room temperature the carbonate peak is relatively small. However, the second-stage drying at 90°C increases the intensity of the carbonate peak considerably. Heating to 460°C causes the peak to decrease and finally heating in the gel-glass transformation range causes it to disappear. Fig. 5 also shows that a shoulder at 950 c m - I attributed to S i - O - N a * vibrational modes, appears concurrently with the elimination of the carbonate modes. The IR transmission from 700-200 cm-~ spectrum of the gel transformed at 500°C corresponds to that of a 33N melt glass. The effect of crystallization on processing is especially apparent when one compares the time dependence of shrinkage (fig. 6) for the various firing temperatures. The volume change occurring at 490°C is effectively complete after 1 h and shows little change during an additional 21 h of heating. The IRRS spectrum of the 33N gel begins to show evidence of a glass after just 15 min at 490°C (fig. 7). Intensity of the spectra continues to increase with heating time, 15-40 min, as the density of vibrational modes in the sample increases. After 40 min at 490°C the spectrum is equivalent to that of a 33N melt glass and remains unchanged after 22 h at 490°C. The sequence of structural changes is similar at 470°C (figs. 4 and 6), but much slower. Very little change in weight accompanies the structural changes associated with the gel-glass transformation, as shown for the 470°C heating in fig. 6. In contrast, the volume change and shrinkage at 540°C is high at short times but is much less after l h (fig. 6). IRRS analysis (fig. 8) shows the spectrum of a 33N-like glass after just 5 min immersion of a 33N sample into the 540°C furnace. The peaks are somewhat displaced from the 33N melt glass similar to the partially transformed 480°C sample (fig. 4). Increasing the firing time for the 540°C samples, though, produces a progressive loss in spectral intensity (fig. 8) due to scattering from crystal-pore interfaces.

5. Discussion of results

The density and porosity of 33N samples before and after the gel-glass transformation are shown in fig. 9 for the various temperatures after 1 h drying or firing. The shape of the density-temperature curve closely parallels the volume change curve in fig. 3. Within the gel-glass transformation range, a density is obtained which is very close to that of the melt-derived 33N glass. The porosity of some of the 33N glasses prepared by gel transformation is also zero, and can be the same as for a good melt-derived glass. A decrease of 60% porosity occurs during the transformation process. However, because the pores in the dried gel are so uniform in size the densification occurs rapidly and uniformly without initiating shrinkage stresses. Consequently, very little

192

L.L. Hench et al. / Preparation of N a 2 0 - S i O 2 glass

.

.

.

.

.

.

.

.

Gel-~, Gloss i

2.5

80

2.0

60

porosity

~

~

~

A U

c~

40

1.5

g_ g

~ s i t y

1.0

0.5 0

I I00

t 200

20

t 300

t 400

~ 500

0 600

Firing Temperature (%) Fig. 9. Density and porosity of 33N gel samples after 1 h heating at temperatures indicated.

warpage or cracking occurs, in the sample sizes studied. Also, thermal shock fracture that normally occurs during insertion of ceramic or glass samples directly into a 500°C furnace does not take place. Therefore it appears possible to use this procedure to fabricate components with very high tolerance by casting from solution or machine the soft-dried gel and subsequently shrink it to much smaller dimensions. This procedure should maintain tolerances while concurrently transforming components into a fully dense glass. The results of this study show that the gel-glass transformation is a kinetic process. The steep temperature dependence of the transformation range is evidence of viscous flow of the network undergoing the structural change. A pure SiO 2 gel requires 1150-1250°C for the transformation to be complete [l l] whereas only 480-520°C is necessary for the 33 mol% N a 2 0 - 6 7 mol% SiO 2 gel to change to a glass. The ratio of transformation temperature for the two compositions is approximately equivalent to the ratio of their liquidus temperatures. The upper temperature limit of the transformation range is established by the crystallization kinetics of the system. If the rate of crystallization is sufficiently rapid the structural change will proceed directly towards its lowest free-energy instead of the metastable glassy state, such as described previously [12], and crystal growth will result. Because there was no control of nucleation in the rapid, single-stage firing the crystal growth was uncontrolled and many cracks, pores and large crystals developed. Heating at the lower temperatures of the gel-glass transformation could induce nucleation simultaneously with

L.L. Hench et al. / Preparation of N a 2 0 - SiO 2 glass

193

the densification process. A second, s u b s e q u e n t heating step at the higher end of the t r a n s f o r m a t i o n range could then produce controlled devitrification a n d a dense glass-ceramic from the gel. The authors would like to acknowledge the e n c o u r a g e m e n t of Prof. J. Zarzycki, Director I n s t i t u t des Verres, a n d support of C N R S throughout this study. They also appreciate the laboratory c o n t r i b u t i o n s of J. Wilson H e n c h a n d M. Wilson. One of the authors (L.L.H.) acknowledges partial financial s u p p o r t of A F O S R C o n t r a c t No. F49620-80-C-0047 d u r i n g the course of this study:

References [1] R. Roy, J. Amer. Ceram. Soc. 52 (1969) 344. [2] H. Dislich, Angew. Chem. Int. Edn. 10 (1971) 363. [3] S.P. Mukherjee, J. Zarzycki and J.P. Traverse, J. Mater. Sci. 11 (1970) 341. [4] S.P. Mukherjee, J. Non-Crystalline Solids 42 (1980) 477. [5] B.E. Yoldas, J. Mater. Sci. 12 (1977) 1203. [6] S. Sakka and K. Kamiya, J. Non-Crystalline Solids 42 (1980) 403. [7] R. Jabra, J. Phalippou and J. Zarzycki, J. Non-Crystalline Solids 42 (1980) 489. [8] D.M. Sanders, W.P. Person and L.L. Hench, Appl. Spectrosc. 28 (1974) 247. [9] D.E. Clark, E.C. Ethridge, M.F. Dilmore and L.L. Hench, Glass Technol. 18 (1977) 637. [10] L.L. Hench and D.E. Clark, J. Non-Crystalline Solids 28 (1978) 83. [11] M. Prassas, Ph.D. thesis (Montpellier, France, 1981). [12] L.L. Hench, D.L. Kinser and S.W. Freiman, Phys. Chem. Glasses 12 (1971) 58. [13] J. Jacoby, Thesis (Nancy, France, 1977). [14] M. Prassas, J. Phalippou, L.L. Hench and J. Zarzycki, J. Non-Crystalline Solids 48 (1982) 79.