Nucleation and crystal growth of analcime from clear aluminosilicate solutions Geoffrey S. Wiersema and Robert W. Thompson Department of Chemical Engineering, Worcester Polytechnic Institute, Worcester, Massachusetts 01609, USA Hydrothermal syntheses of the zeolitic mineral analcime have been carried out in clear aluminosilicate solutions with batch composition A1,0, 84S10, 87Na20 2560H20 in the temperature range 130-160 "C and under autogenous pressures Observation of the changes of the crystal sizes with time indicated that the analcime crystals grew at a constant rate, dependent on the synthesis temperature, and that the rate decreased when the particles settled to the bottom of the autoclave At that point in the synthesis, a second population of nuclei was observed to form and grow in the solution above the settled crystals It was demonstrated that when pure forms of silica were used in these syntheses, fewer crystals were formed These results supported the concept that nucleus formation is promoted by impurities in the silica source Molecular sieve zeolites are crystalline aluminosilicates which have numerous industrial uses owing to their very regular pore openings on the molecular level The term 'molecular sieve' stems from their ability to separate molecules based on their physical size in relation to the pore openings in the zeolites Catalytic properties of zeolites are due to acid sites in the crystal framework, which arise from the aluminate ions in the lattice when the charge is compensated by a proton While there are significant deposits of natural zeolites which have some commercial importance, especially in ion-exchange pro- cesses, most zeolites used commercially are synthetic, having the advantages of purity and uniformity Zeolites generally are synthesised in hydrothermal systems at elevated temperatures at solution pHs in the range 9-13 5, depending on the system, although some low-pH syntheses have been reported Typical crystallizations from solution involve formation of nuclei of the crystalline phase by any of several well researched mechanisms,2 followed by the growth of those crystals by the incorporation of solute from the solution phase The relevant issues pertaining to molecular sieve zeolite synthesis are the mechanism by which zeolites nucleate to form the smallest fragment of crystalline material, and whether the rate of crystal growth is limited by diffusion from the bulk solution to the crystal surface or by the kinetics of incorporation at the crystal surface In spite of over four decades of research and progress in zeolite synthesis techniques and experience, there is still much uncertainty regarding the relevant mechanisms of zeolite nucleation and crystal growth While most syntheses of impor- tance occur in the presence of an amorphous aluminosilicate gel which forms soon after mixing the ingredients, there have been several reports of synthesising various zeolites from clear aluminosilicate solutions 21 Clear solutions provide a means of evaluating these crystallisation systems in situ by optical microscopy9 lo or quasi-elastic light scattering spectroscopy (QELSS)," l8 r e, without having to invade or interrupt the process Additionally, since most clear solution syntheses pro- duce a single burst of nuclei in a short time, the crystal populations from such systems are monodisperse, making crystal size and crystal growth-rate determinations easier than in polydisperse systems Analcime itself is not a commercially used zeolite, however, it is a common by-product from sodium aluminosilicate zeolite synthesis systems, and is a stable end-member, since it does not redissolve in its synthesis solution to form more stable phases, as some other more useful zeolites do Consequently, it was found to be a convenient crystal phase to use to investigate the fundamental nucleation and crystal growth aspects of zeolite syntheses In this work, individual autoclaves were removed from the synthesis ovens, the syntheses were quenched, and the products were analysed ex sztu to study the features of the zeolite synthesis process which are reported here In recent investigations using the same batch composition and a synthesis temperature of 150OC19 it has been reported that bimodal crystal-size distributions formed, and were the result of extended ageing at ambient temperature The size of the two populations of crystals formed could be regulated by altering the ageing time of the mixed solution prior to synthesis at elevated temperature, with increasingly smaller crystals forming with increasingly longer ageing times In the current study, the ageing time was kept as short as possible (hours rather than days) and experiments were conducted at various temperatures to note the effect of changing that synthesis variable Experimenta1 Analcime was synthesised in this study nominally from the batch composition reported previously,19 which was given by A1,0, 84S10, 87Na20 2560H20 The reagents used in these experiments were aluminium wire (Aldrich, 99 999%), sodium hydroxide (Aldrich, 99 99%), Cab-0-Si1 (Eastech, 99 + YO), deionized water (Barnstead NANOpure 11, > 17 0 MR cm-'), puratronic silica (Johnson Matthey, 99 999%), tetraethyl orthosilicate (Aldrich, 99 999%), sodium silicate pentahydrate (Sigma) and sodium silicate non- ahydrate (J T Baker) Elemental analyses were not available for these reagents, but they were all research-grade materials and, therefore, were expected to have impurities present in trace amounts, as noted previously 22 The aluminium wire was dissolved in gently boiling sodium hydroxide and water in a HDPE Nalgene flask with a reflux condenser, a portion of the water and sodium hydroxide in the batch were used This step took approximately 1 h The remaining portions of the water and sodium hydroxide solution were used to dissolve the silica source, which was allowed to mix unheated overnight Each of the silicate solu-tions were prepared with identical compositions, neglecting any impurities which might have been present The two solutions were mixed at room temperature, followed by stirring for 15 min The entire mixture was then filtered through a 0 20 pm polysulfone membrane to remove particu- late matter A 'clear' aluminosilicate solution was produced following this procedure The solution thus prepared was distributed among 8 cm3 Teflon-lined steel autoclaves, which were sealed and placed in a convection oven at the desired temperature (130-160 "C) J Muter Chem, 1996, 6( lo), 1693-1699 1693 Individual vessels were removed from the oven at predeter- mined times and quenched in cold water to effectively stop the synthesis reactions The contents of the autoclaves were filtered through 0 20 pm polysulfone membranes, rinsed several times with deionized water, and dried at 80 "Cfor at least 4 h Products were weighed to determine a solid yield from each autoclave Powder X-ray diffraction was used to determine the crystalline phase present in each sample, with Cu-Ka as the X-ray source Observations were made on a JSM 840 scanning electron microscope, from which particle-size distributions were determined Results and Discussion The first aspect of this study was to determine the effects of changing the synthesis temperature on the analcime crystal growth rate This task was accomplished by monitoring the evolution of crystal sizes with time at synthesis temperature, using M-5 Cab-0-Si1 as the silica source In these experiments the water and sodium hydroxide were divided approximately equally between the aluminate solution and silicate solution during preparation The results of this part of the study are shown in Fig 1, in which the change in the linear dimension of the crystals is shown to have increased at a constant rate at each temperature The best lines drawn through the data points at each tempera- ture shown indicated that each crystal growth process began quite early in the synthesis, I e, each line passes through the origin, with the exception of the line passing through the 150°C data The crystal growth rates computed from the slopes of the respective lines are reported in Table 1 It was observed that the crystal growth rates were essentially constant during the times reported for the experiments in Fig 1 Zeolite crystal growth rates can be expected to depend on the batch composition, the solution pH and the reaction temperature While the temperature was constant, one might expect reagent concentrations to change during crystal growth owing to consumption of material from the clear solution However, the conversion of aluminium was only approximately 10% during these experiments, while all other reagents were present in significant excess Therefore, the driving force for crystal growth could be assumed to be essentially constant during the synthesis times noted in Fig 1 If the crystal growth rate can be assumed to depend on 80 0 5 10 15 20 25 synthesis time/h Fig.1 Change in the linear dimension of analcime crystals at various synthesis temperatures (A)130°C, (A) 140"C, (0)150°C and (B) 160 "C, using M5 Cab-0-Si1 in the standard batch composition Table 1 Temperature dependence of the linear growth rate of analcime synthesis temperature/"(= linear growth rate/pm h ~~ 130 23 140 50 150 72 160 110 synthesis conditions in the following way G=k(T)g(C,,C2,C3, , pH, ) (C, =composition n) that is, if the temperature effect can be isolated in a separate function, then the temperature dependence can be evaluated by computing an Arrhenius-type activation energy by plotting the natural logarithm of the crystal growth rate against the reciprocal of the absolute temperature A straight line fit of the data from the experiments at different temperatures was obtained, as shown in Fig 2 An activation energy of 75 kJmol-' was computed from the data obtained, which suggests a surface kinetics controlled rate-limiting process rather than a diffusion-limited process 23 Table 2 shows the value obtained from this work compared to values of the activation energy for zeolite crystal growth reported in the literature for several other systems While the value obtained here for analcime is slightly higher than some of the other reported values, it is certainly of the same order of magnitude, and clearly points to surface kinetics being the rate-determining step rather than diffusion Schoeman and co-workers16 l7 used the results of a chronomal analysis to demonstrate that the growth-limiting step in silicalite synthesis from a clear solution was a first-order surface reaction step Effects of the silica source In the following experiments with different silica sources, the solutions were divided in such a way that approximately 12% of the water and 3 5% of the sodium hydroxide were used to prepare the aluminate solutions, while the remaining water and sodium hydroxide were used to prepare the silicate solu-tion The final batch composition was the same as in previous experiments, and the combined solution mixing and filtration were conducted in the same way as before Syntheses were conducted at 160 "C Fig 3 shows the linear crystal growth curve for the first generation sf analcime crystals formed in this synthesis when M-5 Cab-0-Sil was used as the silica source While the evolution of the largest crystals occurred at a constant rate during the synthesis time shown, the linear growth rate was ' -21.50 I 2.30 2.35 2.40 2 45 2 50 103 KIT Fig.2 Plot of the logarithm of the analcime crystal growth rate us reciprocal absolute temperature to determine the activation energy for crystal growth Table 2 Comparison of activation energies for several zeolites zeolite system activation energy/kJ mol ref NaA 46 24 NaX 59 NaY 63 mordenite 46 25 NaX 63 26 NaA 44 27 silicalite 96 11 silicali te 45 16 analcime 75 this work 1694 J Mater Chew, 1996, 6(10), 1693-1699 0A-t 0 2 4 6 synthesis time/h Fig.3 Change in the linear dimension of analcime crystals at 160°C using four different silica sources in syntheses using the standard batch composition. Symbols represent the silica sources: (0)Cab-0-Sil, (A) puratronic silica, (0)sodium silicate nonahydrate and (+) sodium silicate pentahydrate. computed to be 13.4 pm h-l, i.e., slightly higher than in the previous experiments. It is conjectured that the preparation technique used in this series of experiments resulted in a slightly higher silica concentration in solution, increasing the driving force for growth by a small degree. Fig. 3 also shows the linear crystal growth curves for synth- eses carried out the same way as the Cab-0-Sil experiment, and with the same batch composition, but using different silica sources. It is obvious that one line drawn through the Cab-0- Sil data adequately fits all the data in the figure, and that the linear growth rate was the same for the crystals in all four synthesis batches.These results indicate that the driving force for zeolite crystal growth was the same in all four experiments, and was, therefore, a function of the material in the clear aluminosilicate solutions. Fig. 4 shows the increase of the yield of analcime with time at constant temperature in these experiments. Note that the mass of analcime increased rather rapidly at early times, and slowed down after about 20 h. The final yield was close to 5 g of analcime per kg of synthesis solution in all of the experi- ments, although there is some scatter around that value.For reasons that will be discussed below this scatter is to be expected, and it is more informative to examine the yield during the first few hours of synthesis. Note that the time scales in Fig. 3 and 4 are different, and that most of the mass of analcime added to the solid product occurred well beyond the time scale of the linear crystal growth results shown in Fig. 3. Between about 5 and 100 h of synthesis, the first generation of analcime crystals, which had grown to approximately 50 pm in dimension, settled to the bottom of the autoclave, and began to grow into a coalesced mass, losing their individual crystal identity, and another generation of nuclei formed in the solution phase above this settled popu- lation.Fig. 5 shows scanning electron photomicrographs of several samples taken from this experiment, some at early .” 0 20 40 60 80 100 120 synthesis time/h Fig. 4 Yield of analcime us. time using the same four silica sources, designated by the same symbols, as noted in Fig. 3 times, and some much later. Clearly the early samples contain one population of individual crystals, while the later samples showed evidence of a bimodal size distribution and subsequent agglomeration of larger crystals from the first generation. Fig. 5(d) shows that several of the smaller second generation of analcime crystals settled on top of the mass of agglomerated first generation crystals and began to become overgrown by the subsequent growth of the agglomerates, by a mechanism described elsewhere.28 Therefore, it might be expected that at later times, when the growing crystals have agglomerated randomly and lost their identity, the growth of analcime mass would be difficult to describe quantitatively and might proceed differently from one experiment to the next as noted in the previous paragraph.One would not expect the thermodynamic yield to be different in these experiments. However, focusing on the growth of the first population should provide infor- mation about the first nucleation event, that is the nucleation event governed by the constituents in the original synthesis solution. Fig. 6 shows the analcime yield during the first 6 h of synthesis for the same experiments reported in Fig.3 and 4. While the yield was rather difficult to determine experimentally, and especially so during the early periods owing to the small amounts formed, it is demonstrated that the yield using puratronic silica was lower than that for Cab-0-Sil, while the yield using sodium silicate pentahydrate was higher. The sodium silicate nonahydrate data are somewhat more difficult to evaluate, but that yield also became higher than the Cab- 0-Sil yield as time progressed. The individual crystals grew at the same rate, as shown in Fig. 3, but the yields were different during that time period. These results suggest that the number of crystals formed in each case was different. Table 3 shows values for the number concentration of crys- tals formed in several of these experiments, computed from the following relationship: (sample mass) =(analcime density) x (crystal volume) x (number of crystals) where the sample mass was that measured prior to 5 h of synthesis to account for only the first generation of crystals which formed.The uniformity of the crystal sizes made the use of a single average size to represent the whole population a very good assumption. In the control experiment using Cab- 0-Sil, for example, 5.19 x lo5 crystals (kg reaction solution)-’ nucleated in the first generation. This crystal number concen- tration is about six orders of magnitude less than the number concentration of silicalite crystals reported by Twomey et d.” in their clear solution syntheses, and their final crystal size was of the order of 1pm in dimension.Analcime and silicalite are fundamentally different zeolites, and their syntheses cannot be compared to any great extent, other than to note that these limited results indicate that, in these respective clear solutions, analcime nucleated far fewer crystals than silicalite. The number concentration of crystals nucleated using the other silica sources was slightly different than when using Cab- 0-Sil, as shown in Table 3. Thus, when using puratronic silica, the absolute number of crystals formed by nucleation was somewhat smaller (approximately one order of magnitude), but the driving force for crystal growth of individual crystals, i.e., the driving force for the surface kinetics step, was essentially the same in both cases.These results suggest that there is something inherent in the silica sources which affected zeolite nucleation in clear aluminosilicate solutions, but not zeolite crystal growth, since all other reagents were common to these experiments. These ‘things,’ which are unidentified at the moment, may be impurities or colloidal matter small enough to pass through the 0.20 pm membrane filter, and may promote nucleation in some manner as yet not completely determined. .~~Hamilton et ~1 reached similar conclusions in their study of J. Muter. Chem., 1996, 6(lo), 1693-1699 1695 Fig. 5 SEM photomicrographs of analcime crystals synthesised at 160 C from the standard batch using M5 Cab-0-Si1, at (a) 1 h and (b) 3 h showing one uniform sized population (c) Sample after 5 h at temperature shows a second population (d) Samples from the bottom of the autoclave with crystals from the second population overgrown by the agglomerated first population 1696 J Muter Chem , 1996, 6(lo), 1693-1699 -L 1.00 t .-0 c2 0.80 $ 0) 0.60 0)\U5 0.40.-)r Q) .-E 0.20 -0 az 0.00 0 2 4 6 a synthesis time/h Fig.6 Analcime yield from Fig. 4, but at early synthesis times, with the same symbols as in Fig. 4 Table 3 Nuclei concentration from various silica sources" number of nuclei formed/ silica source (kg reaction solution)-' Cab-0-Sil 5.19 x 105 puratronic 3.75 x 104 Na2Si03 .9H20 5.01 x 105 Na2Si0, .5H,O 1.04 x lo6 TEOSb 9.1 x 105 Cab-0-Silc 2.9 x 107 "Measurements made after 3 h of synthesis; only one population evident.bTEOShydrolysed for 10 days at room temperature. 'Ethanol added to the solution to mimic the equivalent amount generated owing to the hydrolysis reaction when using TEOS. zeolite NaX synthesis, and correlated their results to metal impurities in the silica source, but the variation in their results was greater with different silica sources. The fact that there were differences noted in this study, while the only parameter varied was the silica source, is at least consistent with their conclusions. It is unlikely that nucleation of analcime in these systems was by the classical nucleation mechanism.It is improbable that the second population of crystals was nucleated by that mechanism once the first population had grown to such a large size and settled to the bottom of the autoclave, thereby removing crystal mass from the solution and reducing the supersaturation. Of the order of 10% of the A1 was converted by this time, and far less of the silica was converted on a fractional basis (quantitatively, if 0.10 moles of A1 were con- verted, leaving 0.90 moles, then 0.20 moles of Si were converted, leaving 83.80 moles, based on the starting composition.) Twomey et al." physically removed the first population of silicalite crystals from the synthesis solution by filtration, and also observed that a second generation of nuclei formed in the remaining solution. It is more likely that impurities were present in sufficient concentration to catalyse the nucleation of the second population of analcime crystals.Effects of using tetraethylorthosilicate Another source of pure silica is tetraethyl orthosilicate (TEOS), carried out at room temperature, with stirring, for 10 days, after which time the synthesis mixture was prepared as before. A modified Cab-0-Sil synthesis solution was prepared by adding an equivalent amount of ethanol to simulate the by- product of the hydrolysis reaction with TEOS, and to serve as a control. It should be noted that in each of these four examples, two separate liquid phases persisted, only one of which supported the zeolite precipitation reactions.Fig. 7 shows the evolution of the maximum analcime crystal size with time during the syntheses described above. The maximum crystal size was measured in these experiments, since there was a rather broad crystal size distribution formed in these systems. The linear growth rate in the first 6 h in the Cab-0-Sil-ethanol system was 4.5 pm h-l, while the crystal growth rate in the TEOS system (considering the first four data points) was 1.4 pm h-l. These initial growth rates were appreciably slower than the analcime growth rates noted in the experiments without ethanol. The growth in the Cab-O- Sil-ethanol system slowed considerably after the first 10 h owing to settling, and a second population formed. The growth rate in the TEOS system appeared to remain essentially constant for up to 50 h, even though settling had occurred, and a second population formed.Fig. 8 shows the yield from the two systems just described. Several features are noted in comparison to Fig. 4 and 6. First, the two systems containing ethanol both exhibited a longer induction time than in the absence of ethanol. This result could be simply a dilution effect due to the partitioning of silica between the two liquid phases, and would not be expected to be due to lowered solubility of aluminosilicates, since there was no turbidity observed owing to gel formation, or zeolite, since that would have increased the yield. It appears from the data in Fig. 8 that the yield may have continued to increase in the Cab-0-Sil-thanol system at longer synthesis times, but we have no way of predicting the final yield in that system.100 1 1 80 c + +I+El 0.-+Otv) 0' 1 0 20 40 60 80 100 synthesis time/h Fig. 7 Change in the linear dimension of analcime crystals synthesised at 160 "C with Cab-0-Sil and added ethanol and tetraethylorthosilic- ate, using the standard batch composition. Symbols designate silica sources: (0)Cab-0-Sil with added ethanol; (+) TEOS. 7 I h c .-0 c3-3-28a liquid silica source which can be hydrolysed to yield Si02, but which then contains ethanol as a by-product. When the hydrolysis reaction was allowed to occur in situ with the synthesis reaction, i.e., simultaneously with the synthesis, the final zeolite synthesis product contained predominantly unidentified impurities, suggesting that the hydrolysis was incomplete, and that the solution composition was not the same as in the previous cases.When the hydrolysis was carried out in a separate vessel overnight at room temperature and used in the synthesis, rather slow conversion to analcime was observed using the standard synthesis conditions. There was virtually no difference noted when the hydrolysis reaction was 0,Y Y 0 20 40 60 80 100 120 140 synthesis time/h Fig. 8 Analcime yield from the experiments noted in Fig. 7, with the same symbols as in Fig. 7 J. Mater. Chem., 1996,6( lo), 1693-1699 1697 The yield from the TEOS system appeared to have stopped increasing after about 40 h, and to have produced only slightly more than 0 4g of analcime per kg of reaction solution, about an order of magnitude less than in the systems described in Fig 4 and 6 In spite of the reduced yield from the TEOS system, it was rather dramatic that the particle size from that system was the largest noted in all the expenments conducted This statement must be qualified by indicating that a much broader crystal- size distribution was observed in these experiments than in the ethanol-free systems, so these particle dimensions clearly do not represent the average crystal size The reason for the broader size distribution was probably that the silica in these expenments was distributed between two liquid phases, and that silica was transferred to the aqueous phase as the synthesis proceeded Thus, this system behaved more like a gel system in which nucleation typically occurs over a longer time as the gel particles dissolve, leading to a broader crystal-size distri- bution The identity of the first population was lost when settling and overgrowth occurred, i e, it is very likely that in the ethanol-free system, crystals would have grown to a much larger size had they not formed an agglomerated mass at the bottom of the autoclave Table 3 also shows the crystal number concentration for these experiments, computed as before from the yield and the average crystal size before the formation of the second popu- lation It appears that the systems with ethanol present nucleated more crystals than the ethanol-free systems Among the various explanations for this result (e g ,impurities, silicate concentration, reduced zeolite solubility) the most reasonable might be to suggest that the number of nuclei formed is related to the Al/Si ratio, and that that ratio would be higher in the liquid phase in which the precipitation occurred compared to the ethanol-free system, owing to partitioning of the silica between the liquid phases This result was not expected, and will be the subject of continued investigation While it may seem inconsistent that the TEOS system, which exhibited the lowest yield and produced the largest crystals, also nucleated a rather large number of crystals, it is not fair to compare the results of the ethanol-free systems to these, especially at long synthesis times The broad crystal-size distri- bution compared to the unimodal crystal population in the ethanol-free systems, and the settling, overgrowth and sub- sequent nucleation of new populations cloud the comparison It is only realistic to compare these systems at very early times, when the yields are low and only one population existed which is, in fact, how the crystal number concentrations were deter- mined Therefore, one must conclude that the nucleation levels reported in Table 3 are representative of the earliest stages of these processes Conclusions Synthesis studies of the molecular sieve analcime have been conducted Syntheses carried out at various temperatures demonstrated that analcime crystal growth was constant in the early stages of synthesis The temperature dependence of the linear crystal growth rate in the range 130-160°C was calculated to be 75 kJ mol-l, a value consistent with prior values computed for other zeolite systems, and one which suggests that the limiting step in the growth process is surface kinetics rather than diffusional transport limitations Syntheses carried out with four different silica sources at 160 "C showed that the linear crystal growth rate was the same for all silica sources tested These results indicated that the driving force for growth was the same in these experiments, and probably had to do with the silicate anion oligomer distribution in the solution, which has been reported previously not to vary with the silica source22 The yields of analcime from these expenments were different, however, and pointed 1698 J Muter Chem, 1996, 6(10), 1693-1699 to different nucleation rates in each system These differences were suggested to stem from materials inherent in the silica sources, since the water, sodium hydroxide and alumina used in all of these experiments were the same Syntheses with the same batch composition, but using a hydrolysed silicate in which ethanol was a by-product, showed that the linear crystal growth rates were lowered, as were the final yields Nucleation rates in these systems, however, were reported to be greater, most likely owing to the partitioning of silica between two liquid phases and the increased Al/Si ratio This matter is still being investigated, however Typical of this system was the observation that a population of crystals was nucleated very early in the process and in a very short time period, resulting in a unimodal size distribution As these crystals grew, they settled to the bottom of the reaction vessel, where overgrowth occurred, as reported pre- viously,28 forming a membrane-like mass of agglomerated crystals Subsequently, a second population of crystals formed in the solution above the agglomerating mass These, too, eventually settled to the bottom and became incorporated in the agglomerating analcime mass This is perhaps the second report, the other being in ref 11, of formation of a second distinct population of zeolite crystals nucleated late in the synthesis in an unstirred system A similar observation was made in a stirred system," but was attnbuted to a possible secondary nucleation mechanism involving collisions References 1 D W Breck, Zeolite Molecular Sieves Structure Chemistry and Use, John Wiley & Sons, New York, 1974 2 A D Randolph and M A Larson, Theory of Particulate Processes, Academic Press, San Diego, 2nd edn ,1988 3 S Ueda and M Koizumi, Am Mineral, 1979,64,172 4 S Ueda and M Koizumi, Am Mineral, 1980,65, 1012 5 S Ueda and M Koizumi, Proc First Int Symp Hydroth React, 1983,p 695 6 S Ueda, T Sera, Y Tsuzuki, M Koizumi and S Takahashi,J Clay Scz , 1983,23,60 7 S Ueda, N Kageyama and M Koizumi, Proc 6th Int Zeol Conf, ed D Olsen and A Burio, Butterworth, Guildford, 1984, p 905 8 P Wenqin, S Ueda and M Koizumi, Proc 7th Int Zeol Conf, ed Y Murakawi, A Iijima and J W Ward, Kodansha/Elsevier, Tokyo/Amsterdam, 1986, p 177 9 J C Jansen, C W R Engelen and H van Bekkum, ACS Symp Ser ,ed M L Ocelli and H E Robson, American Chemical Society, Washington, DC, 1989,398,257 10 C S Cundy, B Lowe and D Sinclair, J Cryst Growth, 1990, 100,189 11 T A M Twomey, M Mackay, H P C E Kuipers and R W Thompson, Zeolites, 1994,14, 162 12 B J Schoeman, J Sterte and J-E Otterstedt, J Chem SOC Chem Commun , 1993,994 13 B J Schoeman, PhD Dissertation, University of Goteborg, Sweden, 1994 14 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 110 15 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 208 16 A E Persson, B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994,14, 557 17 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 568 18 B J Schoeman, J Sterte and J-E Otterstedt, Stud Surf Sci Catal, No 83, 1994, Elsevier, Tokyo 19 F DiRenzo, R Dutartre, P Espiau, F Fajula and M-A Nicolle, Crystallization of Zeolites in the Microgravity Environment of Space 8th Euro Symp on Muter & Fluid Sci in Space, Brussels, April 12-16,1992, ESA SP-33, p 691 20 A A Brock, G N Link, P S Poitras and R W Thompson, J Muter Chem ,1993,3,907 21 F DiRenzo, F Fajula, P Espiau, M-A Nicolle and R Dutartre, Zeolites, 1994,14,256 22 K E Hamilton, E N Coker, A Sacco, Jr, A G Dixon and R W Thompson, Zeolites, 1993,13,645 23 R W Thompson, in Molecular Sieves-Science and Technology, 24 25 ed.H. Karge and J. Weitkamp, Springer-Verlag, Heidelberg, 1996, ch. 1, p 23. D. W. Breck and E. M. Flanigen, Molecular Sieves, 1968, Society of Chemical Industry, London, p. 47. D. Domine and J. Quobex, Molecular Sieves, 1968, Society of 27 28 S. P. Zhdanov and N. N. Samulevich, Proc. 5th Conf. on Zeolites, 1981, Heyden, London, p. 75. S. Gonthier and R. W. Thompson, in Advanced Zeolite Science and Applications, ed. J. C.Jansen, Elsevier, Amsterdam, 1994, p. 43. 26 Chemical Industry, London, p. 78. S. P. Zhdanov, Ado. Chem. Ser., 1971,20,101. Paper 6/02429G; Received 9th April, 1996 J. Muter. Chem., 1996,6( lo), 1693-1699 1699