- Email: [email protected]
Washing of multicomponent gels prior to drying
J O U R N A L OF
ELSEVIER
Journal of Non-Crystalline Solids 215 (1997) 169-175
Washing of multicomponent gels prior to drying M a r i - A n n Einarsrud a.*, M a y Britt Kirkedelen b Jon Samseth b Kell Mortensen c Tor Grande a S t ~ e Pedersen a a Department oflnorganic Chemistry, Norwegian Universi~ of Science and Technology, 7034 Trondheim. Norway b IFE, 2007 Kjeller. Norway c Rise National Laboratory, 4000 Roskilde, Denmark
Received 7 October 1996; revised 12 February 1997
Abstract The sol-gel route is attractive for the preparation of multicomponent gels because homogenization on an atomic level might be possible. Prior to the drying of the wet gels it is sometimes an advantage to wash or rinse the gel using different types of solvents, i.e., for the preparation of aerogels. In this work we have studied the effect on composition of washing cordierite gels prepared using two different routes in four different solvents. Magnesium and aluminum were leached from the wet gels during washing in especially ethanol and acetone and the composition of the gel was severely altered relative to cordierite. The amount of magnesium and aluminum removed from the gels was shown to be dependent on the structure of the gels. © 1997 Elsevier Science B.V.
I. Introduction One of the major advantages of the sol-gel process for preparation of ceramic materials and glasses is the possibility to obtain a homogeneous mixing at an atomic length scale already at room temperature. This advantage has been used to prepare different multicomponent ceramics and glasses from mixtures of metal alkoxides, multiple alkoxides, salt solutions or mixed alkoxides with salt solutions. Prior to the drying of gels, the pore liquid has in some cases been exchanged with other liquids by a washing or rinsing procedure [1-4]. This exchange of pore liquid might be beneficial in order to reduce the shrinkage of the xerogel by using a pore liquid
* Corresponding author. Tel.: +47-7 359 4002; fax: +47-7 359 0860.
with a lower surface tension, i.e., n-heptane, or to remove traces of catalyst, unreacted monomers, etc. [2,3]. During preparation of aerogels by supercritical drying, a washing step is most often used to exchange the pore liquid to CO 2 [5]. It is also usual to exchange the pore liquid to acetone prior to the supercritical drying with CO 2. However, after washing of gels with cordierite composition in acetone prior to supercritical drying from CO 2, only mullite was observed after crystallization of the gels [1]. Clearly, the Mg and A1 concentration had been altered during the preparation of these gels giving a severe impact on the chemical composition. In this paper we present an investigation of the effect of a washing procedure on the properties of multicomponent gels. The main purpose of the work was to investigate chemical changes due to washing of the gels and if compositional changes due to washing could be related to the gel structure.
0022-3093/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0022-3093(97)0011 0-5
170
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
We have chosen to study a gel with the composition of cordierite because this is a three component gel. The formation of cordierite gels has also previously been investigated and the crystallization behavior of washed gels are reported in the literature [ 1,6]. In addition, cordierite is an interesting material for industrial applications. Two different synthesis routes were applied for the preparation of monolithic gels using tetraethoxysilane and aluminum s e c - b u t o x i d e as precursors for silicon and aluminum but with two different magnesium sources; magnesium acetate tetrahydrate and magnesium nitrate hexahydrate. The aluminum alkoxide was in both cases complexed with acetyl acetonate to reduce its reactivity. For washing of the gels four different solvents were chosen; acetone, ethanol, n-heptane and liquid CO 2. The composition of the washed gels was investigated and is discussed in relation to the structure of the gels prepared by the two different routes.
2. Experimental procedure Gels with cordierite composition were prepared using two different synthesis routes, the main difference is the type of magnesium source and the solvent. Gels of type A with magnesium acetate tetrahydrate (MAH) as magnesium source, were prepared using a slightly modified procedure taken from Selvaraj et al. [6]. Tetraethoxysilane (TEOS) and aluminum s e c - b u t o x i d e (ASB) were used as precursors for silicon and aluminum and the solvent chosen was 2-methoxyethanoi. During the preparation of the gel, ASB was first modified with acetyl acetonate (acac) to decrease the reactivity. ASB was initially dissolved in 2-methoxyethanol and acac with a molar ratio of ASB/acac/2-methoxyethanol equal to 1:1:9.7. This solution was refluxed in nitrogen atmosphere at 130°C for 24 h. After cooling this solution to room temperature a stoichiometric amount of TEOS was added and the solution was refluxed at 130°C for 12 h. Again after cooling to room temperature, a solution of MAH, 2-methoxyethanol and distilled water was added, mixed for 15 min and finally cast into Teflon tubes with an inner diameter of 8.6 mm and kept at 50°C for 34 h for gelation. The total molar ratio M A H / A S B / T E O S / a c a c / H 2 0 / 2 methoxyethanol was 2:4:5:4:64:51 which gives a theoretical cordierite gel density of 0.075 g / c m 3.
Gels of type N with magnesium nitrate hexahydrate (MNH) as the magnesium source were prepared using a procedure similar to one reported by Heinrich et al. [1]. ASB was initially dissolved in isopropanol and acac with a molar ratio ASB/acac/isopropanol of 1:1:6.2. A slight opacification of the solution occurred when ASB was added. The solution was refluxed in nitrogen atmosphere at 70°C for 24 h. After cooling to room temperature a stoichiometric amount of TEOS was added and refuxed at 70°C for 12 h. Again, after cooling to room temperature, a solution of MNH, isopropanol, distilled water and 1.0 M HCI as catalyst was added, mixed for 15 min and finally cast into Teflon tubes with an inner diameter of 8.6 mm and kept at 50°C for 24 h for gelation. The total molar ratio M N H / A S B / T E O S / a c a c / H 20/isopropanol/HC1 was 2:4:5:4:32:35:0.005 which gives a theoretical cordierite gel density of 0.10 g / c m 3. After gelation, the monolithic gels were washed using procedures reported in Table 1 to investigate the influence on gel composition. Washing was performed by exchanging the solvent surrounding the gel 4 times within 24 h. The volume ratio gel:washing solution was 1:3.1. After washing, the last solvent used was removed from the gels either by drying at ambient pressure or at the supercritical conditions of CO 2. The drying temperature and time are included in Table 1. During drying at ambient pressure the samples were placed in partly covered test tubes. The aerogels were prepared by placing the wet gel in an autoclave then flushing with liquid CO 2 twice an hour for 2.5 h at 15-20°C and 60 atm. The chamber was then closed and heated over a period of one hour to supercritical conditions (40°C, 100 atm), then the vapor was bled out manually over a period of 16 h. In the sample code the first letter refers to type A or N gel, the second letter to the drying procedure used (X = xerogel and A = aerogel) and the subsequent letters refers to the different washing media used (E = ethanol, A = acetone and H = n-heptane). For the aerogels, the last solvent was naturally liquid CO 2. All the gels were heated to 500°C (gel type N) or 600°C (gel type A) to remove organics from the gel preparation. Very limited shrinkage of the gels were observed in this temperature region. The aluminum, magnesium and silicon content in
M.-A. Einarsrud et al./ Journal of Non-Crystalline Solids 215 (1997) 169-175
171
Table 1 Washing and drying procedures for the different gels Sample
Washing solvent
AX AXE AXA AXEH
None a Ethanol Acetone Ethanol Heptane None ~ Ethanol Acetone None b Ethanol Acetone Ethanol Heptane None b Ethanol Acetone Ethanol Acetone
AA AAE AAA NX NXE NXA NXEH NA NAE NAA NAEA
Washing temp. (°C)
Washing time (h)
50 RT 50 50
4 4 4 4
X x X X
within 24 within 24 within 24 within 24
h h h h
50 RT
4 x within 24 h 4 X within 24 h
50 RT 50 50
4 4 4 4
x X X x
within within within within
24 24 24 24
h h h h
50 RT 50 RT
4 4 4 4
x x x X
within within within within
24 24 24 24
h h h h
Drying cond.
Drying temp. (°C)
Drying time (h)
Xerogel Xerogel Xerogel Xerogel
90 70 50 90
72 48 48 48
Aerogel Aerogel Aerogel Xerogel Xerogel Xerogel Xerogel
40 40 40 75 70 50 90
16 16 16 72 48 48 48
Aerogel Aerogel Aerogel Aerogel
40 40 40 40
16 16 16 16
a Dried from the mother liquor (2-methoxyethanol). b Dried from the mother liquor (isopropanol).
the ethanol washing medium were measured using Inductively Coupled Plasma (ICP) (AtomScan 16/25 spectrometer, Thermo Jarrell Ash). The washing medium was diluted with water (80 to 90 vol%) prior to the measurements. The chemical composition of the samples were determined by sintering the samples at 1350°C (type A) or 1400°C (type N) for 6 h for complete crystallization and determining the phase content by quantitative X-ray diffraction (XRD) using the results from the ICP of the washing medium as reference. Details about the sintering of these gels are published in a separate paper [7]. For the calculation of hydraulic radius, the density of the heat treated gels was calculated from weight and volume. The surface areas were determined by the BET method using a Micromeritics ASAP 2000 nitrogen sorption apparatus. The hydraulic radii were calculated from the formula
2vp rh = S A ' where SA is the measured surface area and Vp the pore volume calculated from the aero- or xerogel density and a skeletal density of 2.5 g / c m 3 [7]. Neutron scattering measurements using neutron
wavelength from 3 to 9 A were performed with the SANS instrument at Rise National Laboratory, Denmark. The monochromator of this instrument is a mechanical velocity selector. The detector was an area-sensitive 3He multiwire detector with a diameter of 60 cm. The distance between the sample and the detector was varied from 1 to 6 m and the q-range covered by the measurements was 0.004 to 0.3 A -~ . The samples were kept at 25°C in vacuum during the measurements.
3. Results
The concentration of AI and Mg in the ethanol used as washing solvent was applied to calculate the compositional changes of the corresponding gels during washing. The results are shown in Fig. 1. Some silicon might also be lost from the gels during washing because of unreacted monomers. However, the amount of silicon removed from the gels during washing was measured by ICP to be less than 2% for gels of type A and less than 5% for gels of type N. Fig. 1 demonstrates that the concentration of magne-
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
172
Table 2 Estimated composition based on quantitative XRD of the gels after complete crystallization at 1350°C (type A) or 1400°C (type N) given as the molar ratio Mg:AI:Si. Removal of small amounts of Si during the washing step is corrected for. The errors in the Mg and AI mole constant are estimated to be within + 10% and -5% Sample
Mg
A1
Si
A1/Mg
AX AXE AXA AXEH AA AAE AAA NXE NXA NXEH NAE NAA NAEA Cordierite
1.92 1.49 1.86 1.49 1.92 1.45 1.86 0.69 0.75 0.65 0.65 0.67 0.61 2
3.84 3.12 3.72 3.10 3.84 3.03 3.72 2.74 2.80 2.75 2.78 2.76 2.70 4
5 5 5 5 5 5 5 5 5 5 5 5 5 5
2.00 2.09 2.00 2.08 2.00 2.09 2.00 3.97 3.73 4.23 4.28 4.12 4.43 2.00
3.5
AI con~nt
3 ©
2.5 2 Mg content
1.5
~--E] 1-
1
0.5 0
5
10 15 Time [hi
20
25
Fig. 1. Calculated cumulative plot of magnesium and aluminum content in gels versus washing time in ethanol. The ethanol surrounding the gels was exchanged every 4 to 10 h. A molar silicon content of 5 is chosen as a basis to reflect the cordierite composition Mg:AI:Si equal to 2:4:5. Open symbols: type A gels (+2%), closed symbols: type N gels (+5%).
sium and aluminum in the type A gels is larger than in the type N gels. XRD measurements of the gels after complete crystallization at 1350°C and 1400°C for gels of type A and N, respectively, [7] demonstrated that the nonand acetone-washed gels of type A (AX, AXA, AA, AAA) all crystallized to ct-cordierite with possible traces of cristobalite [7]. Only minor changes of the composition occurred for these gels during the washing. The ethanol-washed gels of type A (AXE, AXEH, AAE) all crystallized to et-cordierite (major phase), cristobalite (small amount), and mullite (traces) demonstrating Mg and A1 deficiencies in these gels [7]. All the washed gels of type N (NXE,
NXA, NXEH, NAE, NAA, NAEA) crystallized to ot-cordierite (major phase), cristobalite (major phase) and mullite (major phase) [7]. A significant reduction of the A1 and Mg concentrations is therefore evident for these gels. The estimated composition given as the molar ratio Mg:Al'Si of all the gels after complete crystallization are summarized in Table 2 together with the calculated A1/Mg ratio. The density, surface area and hydraulic radius of the gels are reported in Table 3. The surface area was fairly independent of the solvent and drying procedure which is in accordance with literature as long as no sintering occurs [8]. A log-log plot of the SANS measurements for sample NAA and AAA is shown in Fig. 2. Each curve is characterized by two crossovers and two
Table 3 Density, surface area and hydraulic radius of the gels after heat treatment Sample
Density + 3% ( g / c m 3)
Surface area + 5% (m2/g)
Hydraulic radius + 6% (,~)
AX AXE AXA AXEH AAE AAA NXEH NAEA
1.04 1.20 1.19 0.76 0.22 0.22 0.91 0.27
434 495 428 534 526 468 611 591
26 18 20 34 159 182 23 113
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
AEROGEL
106
'
'
'
' ' " [
.
.
.
.
4. Discussion
' ' " 1
'
'
'
'
' ' ' '
105
104.
103
I
£ 102 o
10°
10-1
10-2
10-3
,
~
,
,
,,,11
i
i
I
i
173
i i i i I
0.001 0.002 0.005 0.01 0.02 0.05 0.1 q [~-I]
i
0.2
~
i
i i i i
0.5
Fig. 2. Scattered intensity as a function of w a v e vector, q, m e a s u r e d by S A N S o f the aerogels A A A a n d N A A , w a s h e d in acetone prior to supercritical d r y i n g f r o m C O 2 .
straight sections. The crossovers are located at qcl = 0.01 and qc2 = 0.1 ~-1 for NAA gel and at qcl = 0.008 and qc2 = 0.12 ,~-1 for AAA gel. The slope in the intermediate region, - 2 , is the same for both samples whereas the slope at high q is different, 3.65 and 2.78, respectively. The data from the SANS measurements reported in Fig. 2 are given in Table 4.
The cordierite gel network formed would ideally consist of a silicon and aluminum containing framework where both Si and A1 are four-coordinated and with magnesium as charge compensating ions. From Table 2 we can conclude that for the type A gels, washing in ethanol removes especially Mg and A1 to a stoichiometry of about 1.5:3.1:5, however, the A1/Mg ratio is quite constant. The concentration of both Mg and A1 have been reduced to a factor of approximately 3/4. Washing in acetone or n-heptane (subsequent to ethanol) does not change the composition of the gels significantly. The obtained stoichiometry for the gels washed in acetone or liquid CO 2 is about 1.9:3.8:5. Supercritical drying from CO 2 following the washing in acetone does not change the composition at all; compare, i.e., AX and AA or AXA and AAA. Mg acetate is very soluble in ethanol [9], but is not soluble in either acetone or n-heptane which might explain this difference if Mg is coordinated by acetate inside the gel. It should also be mentioned here that the washing in acetone was performed at room temperature to reduce the vapor pressure of acetone which might have decreased the effect of the washing compared to washing in ethanol performed at 50°C. On the other hand, for the gels of type N, the composition of all the gels changed dramatically during washing independent of solvent and washing procedure. There is no significant difference observed between using ethanol or acetone. There might also be a small change in composition of the gels of type N due to the supercritical drying. Mg nitrate is highly soluble in both ethanol and acetone [9], which might explain the high loss of Mg from gels containing Mg nitrate. However, since Mg acetate also is highly soluble in ethanol and less Mg was removed from the gels of
Table 4 Results obtained f r o m the S A N S data for the aerogels A A A a n d N A A Sample
2"rr/qcl + 9 0 (A)
Slope
2 ~r/q~2 4- 6 0 (,%)
Slope + 0.04
Fractal dimension
AAA NAA
785 630
- 2 - 2
52 63
- 3.65 - 2.78
D s = 2.35 D v = 2.78
174
M. -A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
type A using the same washing procedure, solubility considerations alone cannot explain the difference in behavior. The large change in composition of the type N gels during washing might also explain why Heirvich et al. [5] observed only mullite in crystallized gels expected to be of cordierite stoichiometry. The present observed change in composition is expected for their gels since our type N recipe is quite similar to their recipe. The time dependence of the washing in ethanol was also studied for the two types of gels, and the composition of the gels calculated as a function of washing time was reported in Fig. 1. It is clearly observed that Mg and A1 are leached out of both types of gel during this washing in ethanol at 50°C, however, for the gels of type A only minor amounts of Mg and A1 are removed. The concentration of A1 and Mg in the gel was shown to stabilize after about ten hours washing (exchanging the ethanol twice). The preceding discussion shows that there are significant differences in the behavior of the two types of gel during washing in the solvents chosen. The difference in the effect on Mg concentration of washing in ethanol and acetone could not be explained from solubility considerations of the Mg salts only. Solubility considerations for aluminum is more complex because a large range of complexes/coordinations can be formed inside the gel/solutions. In all the gels, yellow tiny crystals were observed after gelation, mostly at the end of the gel rods. XRD and NMR showed that these crystals were aluminum acetyl acetonate. The crystal formation indicates that some A1 is present in the pore liquid after gelation and is probably not a part of the gel network. The aluminum acetyl acetonate crystals were soluble in all washing media used in this work, except n-heptane. All of the gels washed in n-heptane were washed in ethanol prior to n-heptane and the aluminum acetyl acetonate will therefore be removed from all gels during the washing step. This fact was confirmed because all gels changed color from yellow to transparent during washing. Comparing the two types of gels, there seem to be more crystals precipitating in the type A gels which cannot explain the highest loss of AI from the type N gels. We therefore believe the difference in loss of Mg and AI during washing is caused by the different recipes used and the different gel structures and homo-
geneities obtained. Differences in the preparation of the two types of gels that could influence the gel structure and hence the effect of washing are (1) the different Mg sources used, acetate for the type A gels and nitrate for the type N gels, (2) the difference in amount of water added during the synthesis (twice as much in molar ratio for gels of type A compared to type N), (3) the inorganic acid catalyst, HC1, used during the preparation of the N type gels, and (4) the difference in the amount of solvent used giving a difference in theoretical density of the two gels, 0.1 g / c m 3 for gel type A and 0.075 g / c m 3 for gel type N. The effect of these differences on the chemistry of this complicated three component system is hard to predict, however, the two recipes were chosen because it was assumed by considering the sol-gel chemistry that the type A gels have a three-dimensional network consisting of more 'condensed' particles while the type N gels consist of a three-dimensional polymeric network. This assumption is supported by the facts that the type N gels have the smallest hydraulic radius and the highest surface area as was shown in Table 3. The structure of the two types of gels was verified from the SANS measurements given in Fig. 2. The scattering intensity, I, follows the power law l~q
-D '
where q is the scattering vector and D is the fractal dimension highly dependent upon the geometry of the scattering object. If the object is a mass or volume fractal, the volume fractal dimension D v = D. For volume fractals, D v < 3 and is increasing with decreasing porosity of the scattering volume. On the other hand, for a fractally rough surface, D > 3 and is related to the surface fractal dimension, D S, by Ds = 6 - D. For a smooth surface Ds = 2 while it increases with the roughness of the surface and has a maximum value of 3 [10]. In this case, Ds describes the surface roughness of the particles. From the observed slopes, see Table 4, we conclude that the NAA gel is volume fractal with a fractal dimension Dv = 2.78 __+0.04 which indicates that the particles are porous. The AAA gel is a surface fractal with fractal dimension D S = 2 . 3 5 _ 0.04 which shows that this sample has particles with
M. -A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
a rough surface and a highly compact interior. The slope in the intermediate q-region of both samples is - 2 , which indicates that the matrix is a cross linked polymer network. The crossover points indicate the length scales in the system. From the results of Fig. 2, the 2~ro/qc1 can be related to the cluster sizes of 785 + 90 A for AAA, and 630 + 90 ,~ for NAA. A larger cluster size for AAA is consistent with a larger hydraulic radius observed for this type of gels. The crossover at higher q, 2'rr/qc2 °gives particle sizes of about 52 + 6 A and 63 + 6 A for the AAA and NAA gels, respectively. Hence, the particles of the AAA gel have a compact interior, their surface is rough and they are somewhat smaller than the porous particles of the NAA gel. Studying the results in Table 2 in more detail, we see that the A1/Mg ratio is close to 2 as in cordierite for the type A gels even both A1 and Mg are leached out during the washing procedure. These results indicate that the structure of the gel consists of an A1 containing Si network (AI and Si are four-coordinated) and with Mg as compensating ions. We propose that removal of the Mg ions from the gel network is difficult as long as they are compensating the charge deficiency given when AI is introduced into the Si-O network. This proposed gel structure is also in agreement with the fact that the type A gels crystallized to ix-cordierite at temperatures below 1000°C [7], which is a phase with quartz structure forming solid solution with quartz [11]. On the other hand, for the type N gels the A1/Mg ratio was about 4 for all the gels and relatively more Mg is removed from the gel during washing compared to A1. The high Mg deficiency in these gels compared to A1 indicates that the type N is more inhomogeneous on a medium range order. Since at least some AI probably is not in four-coordinated positions in the Si-O network the removal of Mg (and A1) from the gel during washing is easier compared to the type A gels.
5. Conclusions Washing or rinsing of wet gels prior to drying may change gel composition and hence structure and
175
homogeneity of the wet gel depending on the recipe. Washing cordierite gels in acetone and ethanol was shown to remove Mg and AI from the final gels. Washing in n-heptane or liquid CO 2 subsequent to acetone or ethanol does not change the composition significantly. The recipe type A presented in this work is more suitable for the preparation of stoichiometric cordierite than recipe type N. Washing of the gels prior to drying at ambient pressure significantly decreased the cracking of the gels during drying. The present investigation demonstrates gel stoichiometry may easily be influenced by a simple washing or rinsing procedure. This is particularly important in connect to synthesis of powders by the sol-gel route where the diffusion distance is very short.
Acknowledgements The authors are grateful to engineer A.L. Bye for performing the ICP measurements. Borgestads Legat is acknowledged for financial support.
References [1] T. Heinrich, W. Tappert, W. Lenhard, J. Fricke, J. Sol-Gel Sci. Tech. 2 (1994) 921. [2] G.W. Scherer, S. Haereid, E. Nilsen, M.-A. Einarsrud, J. Non-Cryst. Solids 202 (1996) 42. [3] S. H~ereid, E. Nilsen, M.-A. Einarsrud, J. Porous Mater. 2 (1996) 315. [4] R. Deshpande, D.M. Smith, C.J. Brinker, patent WO 94 25149. [5] T. Heinrich, U. KleU, J. Fricke, J. Porous Mater. 1 (1995) 7. [6] U. Selvaraj, S. Komarneni, R. Roy, J. Am. Ceram. Soc. 73 (1990) 3663. [7] M.-A. Einarsrud, S. Pedersen, E. Larsen, T. Grande, "Sintering and crystallization of gels in the system AI203-MgOSiO2', submitted to J. Am. Ceram. Soc. (1996). [8] D.M. Smith, G.W. Scherer, J. Anderson, J. Non-Cryst. Solids 188 (1995) 191. [9] Gmelin Handbuch tier Anorganischen Chemie, Magnesium, Vol. 27 (Verlag Chemie, Berlin, 1939). [10] A. Emmerling, J. Fricke, J. Non-Cryst. Solids 145 (1992) 113. [11] W. Schreyer, J.F. Schairer, Z. Kristallogr. 116 (1961) 60.
ELSEVIER
Journal of Non-Crystalline Solids 215 (1997) 169-175
Washing of multicomponent gels prior to drying M a r i - A n n Einarsrud a.*, M a y Britt Kirkedelen b Jon Samseth b Kell Mortensen c Tor Grande a S t ~ e Pedersen a a Department oflnorganic Chemistry, Norwegian Universi~ of Science and Technology, 7034 Trondheim. Norway b IFE, 2007 Kjeller. Norway c Rise National Laboratory, 4000 Roskilde, Denmark
Received 7 October 1996; revised 12 February 1997
Abstract The sol-gel route is attractive for the preparation of multicomponent gels because homogenization on an atomic level might be possible. Prior to the drying of the wet gels it is sometimes an advantage to wash or rinse the gel using different types of solvents, i.e., for the preparation of aerogels. In this work we have studied the effect on composition of washing cordierite gels prepared using two different routes in four different solvents. Magnesium and aluminum were leached from the wet gels during washing in especially ethanol and acetone and the composition of the gel was severely altered relative to cordierite. The amount of magnesium and aluminum removed from the gels was shown to be dependent on the structure of the gels. © 1997 Elsevier Science B.V.
I. Introduction One of the major advantages of the sol-gel process for preparation of ceramic materials and glasses is the possibility to obtain a homogeneous mixing at an atomic length scale already at room temperature. This advantage has been used to prepare different multicomponent ceramics and glasses from mixtures of metal alkoxides, multiple alkoxides, salt solutions or mixed alkoxides with salt solutions. Prior to the drying of gels, the pore liquid has in some cases been exchanged with other liquids by a washing or rinsing procedure [1-4]. This exchange of pore liquid might be beneficial in order to reduce the shrinkage of the xerogel by using a pore liquid
* Corresponding author. Tel.: +47-7 359 4002; fax: +47-7 359 0860.
with a lower surface tension, i.e., n-heptane, or to remove traces of catalyst, unreacted monomers, etc. [2,3]. During preparation of aerogels by supercritical drying, a washing step is most often used to exchange the pore liquid to CO 2 [5]. It is also usual to exchange the pore liquid to acetone prior to the supercritical drying with CO 2. However, after washing of gels with cordierite composition in acetone prior to supercritical drying from CO 2, only mullite was observed after crystallization of the gels [1]. Clearly, the Mg and A1 concentration had been altered during the preparation of these gels giving a severe impact on the chemical composition. In this paper we present an investigation of the effect of a washing procedure on the properties of multicomponent gels. The main purpose of the work was to investigate chemical changes due to washing of the gels and if compositional changes due to washing could be related to the gel structure.
0022-3093/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0022-3093(97)0011 0-5
170
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
We have chosen to study a gel with the composition of cordierite because this is a three component gel. The formation of cordierite gels has also previously been investigated and the crystallization behavior of washed gels are reported in the literature [ 1,6]. In addition, cordierite is an interesting material for industrial applications. Two different synthesis routes were applied for the preparation of monolithic gels using tetraethoxysilane and aluminum s e c - b u t o x i d e as precursors for silicon and aluminum but with two different magnesium sources; magnesium acetate tetrahydrate and magnesium nitrate hexahydrate. The aluminum alkoxide was in both cases complexed with acetyl acetonate to reduce its reactivity. For washing of the gels four different solvents were chosen; acetone, ethanol, n-heptane and liquid CO 2. The composition of the washed gels was investigated and is discussed in relation to the structure of the gels prepared by the two different routes.
2. Experimental procedure Gels with cordierite composition were prepared using two different synthesis routes, the main difference is the type of magnesium source and the solvent. Gels of type A with magnesium acetate tetrahydrate (MAH) as magnesium source, were prepared using a slightly modified procedure taken from Selvaraj et al. [6]. Tetraethoxysilane (TEOS) and aluminum s e c - b u t o x i d e (ASB) were used as precursors for silicon and aluminum and the solvent chosen was 2-methoxyethanoi. During the preparation of the gel, ASB was first modified with acetyl acetonate (acac) to decrease the reactivity. ASB was initially dissolved in 2-methoxyethanol and acac with a molar ratio of ASB/acac/2-methoxyethanol equal to 1:1:9.7. This solution was refluxed in nitrogen atmosphere at 130°C for 24 h. After cooling this solution to room temperature a stoichiometric amount of TEOS was added and the solution was refluxed at 130°C for 12 h. Again after cooling to room temperature, a solution of MAH, 2-methoxyethanol and distilled water was added, mixed for 15 min and finally cast into Teflon tubes with an inner diameter of 8.6 mm and kept at 50°C for 34 h for gelation. The total molar ratio M A H / A S B / T E O S / a c a c / H 2 0 / 2 methoxyethanol was 2:4:5:4:64:51 which gives a theoretical cordierite gel density of 0.075 g / c m 3.
Gels of type N with magnesium nitrate hexahydrate (MNH) as the magnesium source were prepared using a procedure similar to one reported by Heinrich et al. [1]. ASB was initially dissolved in isopropanol and acac with a molar ratio ASB/acac/isopropanol of 1:1:6.2. A slight opacification of the solution occurred when ASB was added. The solution was refluxed in nitrogen atmosphere at 70°C for 24 h. After cooling to room temperature a stoichiometric amount of TEOS was added and refuxed at 70°C for 12 h. Again, after cooling to room temperature, a solution of MNH, isopropanol, distilled water and 1.0 M HCI as catalyst was added, mixed for 15 min and finally cast into Teflon tubes with an inner diameter of 8.6 mm and kept at 50°C for 24 h for gelation. The total molar ratio M N H / A S B / T E O S / a c a c / H 20/isopropanol/HC1 was 2:4:5:4:32:35:0.005 which gives a theoretical cordierite gel density of 0.10 g / c m 3. After gelation, the monolithic gels were washed using procedures reported in Table 1 to investigate the influence on gel composition. Washing was performed by exchanging the solvent surrounding the gel 4 times within 24 h. The volume ratio gel:washing solution was 1:3.1. After washing, the last solvent used was removed from the gels either by drying at ambient pressure or at the supercritical conditions of CO 2. The drying temperature and time are included in Table 1. During drying at ambient pressure the samples were placed in partly covered test tubes. The aerogels were prepared by placing the wet gel in an autoclave then flushing with liquid CO 2 twice an hour for 2.5 h at 15-20°C and 60 atm. The chamber was then closed and heated over a period of one hour to supercritical conditions (40°C, 100 atm), then the vapor was bled out manually over a period of 16 h. In the sample code the first letter refers to type A or N gel, the second letter to the drying procedure used (X = xerogel and A = aerogel) and the subsequent letters refers to the different washing media used (E = ethanol, A = acetone and H = n-heptane). For the aerogels, the last solvent was naturally liquid CO 2. All the gels were heated to 500°C (gel type N) or 600°C (gel type A) to remove organics from the gel preparation. Very limited shrinkage of the gels were observed in this temperature region. The aluminum, magnesium and silicon content in
M.-A. Einarsrud et al./ Journal of Non-Crystalline Solids 215 (1997) 169-175
171
Table 1 Washing and drying procedures for the different gels Sample
Washing solvent
AX AXE AXA AXEH
None a Ethanol Acetone Ethanol Heptane None ~ Ethanol Acetone None b Ethanol Acetone Ethanol Heptane None b Ethanol Acetone Ethanol Acetone
AA AAE AAA NX NXE NXA NXEH NA NAE NAA NAEA
Washing temp. (°C)
Washing time (h)
50 RT 50 50
4 4 4 4
X x X X
within 24 within 24 within 24 within 24
h h h h
50 RT
4 x within 24 h 4 X within 24 h
50 RT 50 50
4 4 4 4
x X X x
within within within within
24 24 24 24
h h h h
50 RT 50 RT
4 4 4 4
x x x X
within within within within
24 24 24 24
h h h h
Drying cond.
Drying temp. (°C)
Drying time (h)
Xerogel Xerogel Xerogel Xerogel
90 70 50 90
72 48 48 48
Aerogel Aerogel Aerogel Xerogel Xerogel Xerogel Xerogel
40 40 40 75 70 50 90
16 16 16 72 48 48 48
Aerogel Aerogel Aerogel Aerogel
40 40 40 40
16 16 16 16
a Dried from the mother liquor (2-methoxyethanol). b Dried from the mother liquor (isopropanol).
the ethanol washing medium were measured using Inductively Coupled Plasma (ICP) (AtomScan 16/25 spectrometer, Thermo Jarrell Ash). The washing medium was diluted with water (80 to 90 vol%) prior to the measurements. The chemical composition of the samples were determined by sintering the samples at 1350°C (type A) or 1400°C (type N) for 6 h for complete crystallization and determining the phase content by quantitative X-ray diffraction (XRD) using the results from the ICP of the washing medium as reference. Details about the sintering of these gels are published in a separate paper [7]. For the calculation of hydraulic radius, the density of the heat treated gels was calculated from weight and volume. The surface areas were determined by the BET method using a Micromeritics ASAP 2000 nitrogen sorption apparatus. The hydraulic radii were calculated from the formula
2vp rh = S A ' where SA is the measured surface area and Vp the pore volume calculated from the aero- or xerogel density and a skeletal density of 2.5 g / c m 3 [7]. Neutron scattering measurements using neutron
wavelength from 3 to 9 A were performed with the SANS instrument at Rise National Laboratory, Denmark. The monochromator of this instrument is a mechanical velocity selector. The detector was an area-sensitive 3He multiwire detector with a diameter of 60 cm. The distance between the sample and the detector was varied from 1 to 6 m and the q-range covered by the measurements was 0.004 to 0.3 A -~ . The samples were kept at 25°C in vacuum during the measurements.
3. Results
The concentration of AI and Mg in the ethanol used as washing solvent was applied to calculate the compositional changes of the corresponding gels during washing. The results are shown in Fig. 1. Some silicon might also be lost from the gels during washing because of unreacted monomers. However, the amount of silicon removed from the gels during washing was measured by ICP to be less than 2% for gels of type A and less than 5% for gels of type N. Fig. 1 demonstrates that the concentration of magne-
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
172
Table 2 Estimated composition based on quantitative XRD of the gels after complete crystallization at 1350°C (type A) or 1400°C (type N) given as the molar ratio Mg:AI:Si. Removal of small amounts of Si during the washing step is corrected for. The errors in the Mg and AI mole constant are estimated to be within + 10% and -5% Sample
Mg
A1
Si
A1/Mg
AX AXE AXA AXEH AA AAE AAA NXE NXA NXEH NAE NAA NAEA Cordierite
1.92 1.49 1.86 1.49 1.92 1.45 1.86 0.69 0.75 0.65 0.65 0.67 0.61 2
3.84 3.12 3.72 3.10 3.84 3.03 3.72 2.74 2.80 2.75 2.78 2.76 2.70 4
5 5 5 5 5 5 5 5 5 5 5 5 5 5
2.00 2.09 2.00 2.08 2.00 2.09 2.00 3.97 3.73 4.23 4.28 4.12 4.43 2.00
3.5
AI con~nt
3 ©
2.5 2 Mg content
1.5
~--E] 1-
1
0.5 0
5
10 15 Time [hi
20
25
Fig. 1. Calculated cumulative plot of magnesium and aluminum content in gels versus washing time in ethanol. The ethanol surrounding the gels was exchanged every 4 to 10 h. A molar silicon content of 5 is chosen as a basis to reflect the cordierite composition Mg:AI:Si equal to 2:4:5. Open symbols: type A gels (+2%), closed symbols: type N gels (+5%).
sium and aluminum in the type A gels is larger than in the type N gels. XRD measurements of the gels after complete crystallization at 1350°C and 1400°C for gels of type A and N, respectively, [7] demonstrated that the nonand acetone-washed gels of type A (AX, AXA, AA, AAA) all crystallized to ct-cordierite with possible traces of cristobalite [7]. Only minor changes of the composition occurred for these gels during the washing. The ethanol-washed gels of type A (AXE, AXEH, AAE) all crystallized to et-cordierite (major phase), cristobalite (small amount), and mullite (traces) demonstrating Mg and A1 deficiencies in these gels [7]. All the washed gels of type N (NXE,
NXA, NXEH, NAE, NAA, NAEA) crystallized to ot-cordierite (major phase), cristobalite (major phase) and mullite (major phase) [7]. A significant reduction of the A1 and Mg concentrations is therefore evident for these gels. The estimated composition given as the molar ratio Mg:Al'Si of all the gels after complete crystallization are summarized in Table 2 together with the calculated A1/Mg ratio. The density, surface area and hydraulic radius of the gels are reported in Table 3. The surface area was fairly independent of the solvent and drying procedure which is in accordance with literature as long as no sintering occurs [8]. A log-log plot of the SANS measurements for sample NAA and AAA is shown in Fig. 2. Each curve is characterized by two crossovers and two
Table 3 Density, surface area and hydraulic radius of the gels after heat treatment Sample
Density + 3% ( g / c m 3)
Surface area + 5% (m2/g)
Hydraulic radius + 6% (,~)
AX AXE AXA AXEH AAE AAA NXEH NAEA
1.04 1.20 1.19 0.76 0.22 0.22 0.91 0.27
434 495 428 534 526 468 611 591
26 18 20 34 159 182 23 113
M.-A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
AEROGEL
106
'
'
'
' ' " [
.
.
.
.
4. Discussion
' ' " 1
'
'
'
'
' ' ' '
105
104.
103
I
£ 102 o
10°
10-1
10-2
10-3
,
~
,
,
,,,11
i
i
I
i
173
i i i i I
0.001 0.002 0.005 0.01 0.02 0.05 0.1 q [~-I]
i
0.2
~
i
i i i i
0.5
Fig. 2. Scattered intensity as a function of w a v e vector, q, m e a s u r e d by S A N S o f the aerogels A A A a n d N A A , w a s h e d in acetone prior to supercritical d r y i n g f r o m C O 2 .
straight sections. The crossovers are located at qcl = 0.01 and qc2 = 0.1 ~-1 for NAA gel and at qcl = 0.008 and qc2 = 0.12 ,~-1 for AAA gel. The slope in the intermediate region, - 2 , is the same for both samples whereas the slope at high q is different, 3.65 and 2.78, respectively. The data from the SANS measurements reported in Fig. 2 are given in Table 4.
The cordierite gel network formed would ideally consist of a silicon and aluminum containing framework where both Si and A1 are four-coordinated and with magnesium as charge compensating ions. From Table 2 we can conclude that for the type A gels, washing in ethanol removes especially Mg and A1 to a stoichiometry of about 1.5:3.1:5, however, the A1/Mg ratio is quite constant. The concentration of both Mg and A1 have been reduced to a factor of approximately 3/4. Washing in acetone or n-heptane (subsequent to ethanol) does not change the composition of the gels significantly. The obtained stoichiometry for the gels washed in acetone or liquid CO 2 is about 1.9:3.8:5. Supercritical drying from CO 2 following the washing in acetone does not change the composition at all; compare, i.e., AX and AA or AXA and AAA. Mg acetate is very soluble in ethanol [9], but is not soluble in either acetone or n-heptane which might explain this difference if Mg is coordinated by acetate inside the gel. It should also be mentioned here that the washing in acetone was performed at room temperature to reduce the vapor pressure of acetone which might have decreased the effect of the washing compared to washing in ethanol performed at 50°C. On the other hand, for the gels of type N, the composition of all the gels changed dramatically during washing independent of solvent and washing procedure. There is no significant difference observed between using ethanol or acetone. There might also be a small change in composition of the gels of type N due to the supercritical drying. Mg nitrate is highly soluble in both ethanol and acetone [9], which might explain the high loss of Mg from gels containing Mg nitrate. However, since Mg acetate also is highly soluble in ethanol and less Mg was removed from the gels of
Table 4 Results obtained f r o m the S A N S data for the aerogels A A A a n d N A A Sample
2"rr/qcl + 9 0 (A)
Slope
2 ~r/q~2 4- 6 0 (,%)
Slope + 0.04
Fractal dimension
AAA NAA
785 630
- 2 - 2
52 63
- 3.65 - 2.78
D s = 2.35 D v = 2.78
174
M. -A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
type A using the same washing procedure, solubility considerations alone cannot explain the difference in behavior. The large change in composition of the type N gels during washing might also explain why Heirvich et al. [5] observed only mullite in crystallized gels expected to be of cordierite stoichiometry. The present observed change in composition is expected for their gels since our type N recipe is quite similar to their recipe. The time dependence of the washing in ethanol was also studied for the two types of gels, and the composition of the gels calculated as a function of washing time was reported in Fig. 1. It is clearly observed that Mg and A1 are leached out of both types of gel during this washing in ethanol at 50°C, however, for the gels of type A only minor amounts of Mg and A1 are removed. The concentration of A1 and Mg in the gel was shown to stabilize after about ten hours washing (exchanging the ethanol twice). The preceding discussion shows that there are significant differences in the behavior of the two types of gel during washing in the solvents chosen. The difference in the effect on Mg concentration of washing in ethanol and acetone could not be explained from solubility considerations of the Mg salts only. Solubility considerations for aluminum is more complex because a large range of complexes/coordinations can be formed inside the gel/solutions. In all the gels, yellow tiny crystals were observed after gelation, mostly at the end of the gel rods. XRD and NMR showed that these crystals were aluminum acetyl acetonate. The crystal formation indicates that some A1 is present in the pore liquid after gelation and is probably not a part of the gel network. The aluminum acetyl acetonate crystals were soluble in all washing media used in this work, except n-heptane. All of the gels washed in n-heptane were washed in ethanol prior to n-heptane and the aluminum acetyl acetonate will therefore be removed from all gels during the washing step. This fact was confirmed because all gels changed color from yellow to transparent during washing. Comparing the two types of gels, there seem to be more crystals precipitating in the type A gels which cannot explain the highest loss of AI from the type N gels. We therefore believe the difference in loss of Mg and AI during washing is caused by the different recipes used and the different gel structures and homo-
geneities obtained. Differences in the preparation of the two types of gels that could influence the gel structure and hence the effect of washing are (1) the different Mg sources used, acetate for the type A gels and nitrate for the type N gels, (2) the difference in amount of water added during the synthesis (twice as much in molar ratio for gels of type A compared to type N), (3) the inorganic acid catalyst, HC1, used during the preparation of the N type gels, and (4) the difference in the amount of solvent used giving a difference in theoretical density of the two gels, 0.1 g / c m 3 for gel type A and 0.075 g / c m 3 for gel type N. The effect of these differences on the chemistry of this complicated three component system is hard to predict, however, the two recipes were chosen because it was assumed by considering the sol-gel chemistry that the type A gels have a three-dimensional network consisting of more 'condensed' particles while the type N gels consist of a three-dimensional polymeric network. This assumption is supported by the facts that the type N gels have the smallest hydraulic radius and the highest surface area as was shown in Table 3. The structure of the two types of gels was verified from the SANS measurements given in Fig. 2. The scattering intensity, I, follows the power law l~q
-D '
where q is the scattering vector and D is the fractal dimension highly dependent upon the geometry of the scattering object. If the object is a mass or volume fractal, the volume fractal dimension D v = D. For volume fractals, D v < 3 and is increasing with decreasing porosity of the scattering volume. On the other hand, for a fractally rough surface, D > 3 and is related to the surface fractal dimension, D S, by Ds = 6 - D. For a smooth surface Ds = 2 while it increases with the roughness of the surface and has a maximum value of 3 [10]. In this case, Ds describes the surface roughness of the particles. From the observed slopes, see Table 4, we conclude that the NAA gel is volume fractal with a fractal dimension Dv = 2.78 __+0.04 which indicates that the particles are porous. The AAA gel is a surface fractal with fractal dimension D S = 2 . 3 5 _ 0.04 which shows that this sample has particles with
M. -A. Einarsrud et al. / Journal of Non-Crystalline Solids 215 (1997) 169-175
a rough surface and a highly compact interior. The slope in the intermediate q-region of both samples is - 2 , which indicates that the matrix is a cross linked polymer network. The crossover points indicate the length scales in the system. From the results of Fig. 2, the 2~ro/qc1 can be related to the cluster sizes of 785 + 90 A for AAA, and 630 + 90 ,~ for NAA. A larger cluster size for AAA is consistent with a larger hydraulic radius observed for this type of gels. The crossover at higher q, 2'rr/qc2 °gives particle sizes of about 52 + 6 A and 63 + 6 A for the AAA and NAA gels, respectively. Hence, the particles of the AAA gel have a compact interior, their surface is rough and they are somewhat smaller than the porous particles of the NAA gel. Studying the results in Table 2 in more detail, we see that the A1/Mg ratio is close to 2 as in cordierite for the type A gels even both A1 and Mg are leached out during the washing procedure. These results indicate that the structure of the gel consists of an A1 containing Si network (AI and Si are four-coordinated) and with Mg as compensating ions. We propose that removal of the Mg ions from the gel network is difficult as long as they are compensating the charge deficiency given when AI is introduced into the Si-O network. This proposed gel structure is also in agreement with the fact that the type A gels crystallized to ix-cordierite at temperatures below 1000°C [7], which is a phase with quartz structure forming solid solution with quartz [11]. On the other hand, for the type N gels the A1/Mg ratio was about 4 for all the gels and relatively more Mg is removed from the gel during washing compared to A1. The high Mg deficiency in these gels compared to A1 indicates that the type N is more inhomogeneous on a medium range order. Since at least some AI probably is not in four-coordinated positions in the Si-O network the removal of Mg (and A1) from the gel during washing is easier compared to the type A gels.
5. Conclusions Washing or rinsing of wet gels prior to drying may change gel composition and hence structure and
175
homogeneity of the wet gel depending on the recipe. Washing cordierite gels in acetone and ethanol was shown to remove Mg and AI from the final gels. Washing in n-heptane or liquid CO 2 subsequent to acetone or ethanol does not change the composition significantly. The recipe type A presented in this work is more suitable for the preparation of stoichiometric cordierite than recipe type N. Washing of the gels prior to drying at ambient pressure significantly decreased the cracking of the gels during drying. The present investigation demonstrates gel stoichiometry may easily be influenced by a simple washing or rinsing procedure. This is particularly important in connect to synthesis of powders by the sol-gel route where the diffusion distance is very short.
Acknowledgements The authors are grateful to engineer A.L. Bye for performing the ICP measurements. Borgestads Legat is acknowledged for financial support.
References [1] T. Heinrich, W. Tappert, W. Lenhard, J. Fricke, J. Sol-Gel Sci. Tech. 2 (1994) 921. [2] G.W. Scherer, S. Haereid, E. Nilsen, M.-A. Einarsrud, J. Non-Cryst. Solids 202 (1996) 42. [3] S. H~ereid, E. Nilsen, M.-A. Einarsrud, J. Porous Mater. 2 (1996) 315. [4] R. Deshpande, D.M. Smith, C.J. Brinker, patent WO 94 25149. [5] T. Heinrich, U. KleU, J. Fricke, J. Porous Mater. 1 (1995) 7. [6] U. Selvaraj, S. Komarneni, R. Roy, J. Am. Ceram. Soc. 73 (1990) 3663. [7] M.-A. Einarsrud, S. Pedersen, E. Larsen, T. Grande, "Sintering and crystallization of gels in the system AI203-MgOSiO2', submitted to J. Am. Ceram. Soc. (1996). [8] D.M. Smith, G.W. Scherer, J. Anderson, J. Non-Cryst. Solids 188 (1995) 191. [9] Gmelin Handbuch tier Anorganischen Chemie, Magnesium, Vol. 27 (Verlag Chemie, Berlin, 1939). [10] A. Emmerling, J. Fricke, J. Non-Cryst. Solids 145 (1992) 113. [11] W. Schreyer, J.F. Schairer, Z. Kristallogr. 116 (1961) 60.