J. Chem. SOC., Furuduy Trans. I , 1989, 85(4), 945-955 Isopiestic Measurement of Salt Imbibition in Zeolites Na-X and Na-Y Barrie M. Lowe Chemistry Department , University of Edinburgh, Scot land Christopher G. Pope* Chemistry Department, University of Otago, Dunedin, New Zealand Factors affecting the reliability of the isopiestic method for the measurement of the imbibition of salts from aqueous solution by zeolites have been examined. It has been found that the sharp intersection of the isopiestic curves, which is essential to the accuracy of the method, does not occur when the zeolite crystals are so small that sorption of the salt on the external surface of the zeolite is substantial. Results for the occlusion of sodium nitrate by zeolite Na-X (ca. 6 pm crystals) and Na-Y (ca.7 pm crystals) at 298.2, 313.2 and 328.2 K are reported. Those for sodium nitrate and sodium chlorate at 298.2 K are used to test the applicability of the Donnan equation. The application of the isopiestic technique to the measurements of the uptake of solutes by zeolites was introduced by Fegan and Lowe.' The method is versatile, as it can be applied to any solution containing a single involatile solute, and the experiments are simple to perform and require little equipment. The results yield the compositions of both the imbibed and external solutions directly, and it is easy to confirm that equilibrium has been established while the experiments are being performed. Fegan2 has demonstrated that accurate data can be obtained at room temperature, but the experiments tend to be rather slow and tedious.This work is concerned with a further examination of the reliability of the method, with simplifying and speeding up the procedure, and with extending the technique to higher temperatures where solvent condensation on the sample containers presents a possible problem. Experiments were carried out on the imbibition of sodium nitrate at 298, 313 and 328 K by samples of Na-X and Na-Y with different Si/Al ratios and particle size. One sample was studied in more detail at 298 K in order to examine whether the salt imbibition increased smoothly as the water activity decreased, as Fegan' has reported a sharp change of slope at the water activity of the saturated solution when sodium chloride was used, although it does not seem to be clear why such an effect should exist.Finally, experiments with sodium chlorate, sodium citrate and sodium fluoride were carried out to investigate the dependence of imbibition on the salt used. Chlorate and nitrate are of interest because of their different structure directing influence during the synthesis of sodalite and ~ancrinite,~ citrate for its charge, size and hydration, and fluoride because its reactivity with aluminium and silicon suggested that it would probably interact strongly with the zeolite lattice. 945946 Salt Imbibition in Zeolites Experimental Materials A set of Na-X and Na-Y samples were made using a common general procedure and starting materials. Only the proportions of the reactants used were changed, and this in turn resulted in different times being required to complete crystallisation.Sodium aluminate solution was prepared by dissolving AR-grade aluminium wire with AR sodium hydroxide solution in a screw-capped polythene bottle. The rest of the water required was mixed with Cabosil M-5 fumed silica in a second bottle. When the aluminium was completely dissolved, and the solution was cold, the aluminate solution was added to the silica and the mixture was blended thoroughly to a creamy consistency. This was aged for five days at 298 K, and then transferred to a water bath at 367 K where it was left unstirred until crystallisation was complete, as judged by the appearance of the suspension and optical microscope examination of the product (1-3 days). The clear supernatant liquid was then rapidly removed and the remaining slurry filtered and washed with hot water.All the preparations carried out by this method produced small irregularly shaped solid particles which appeared to be made up of agglomerates of smaller material, and were pure as judged by X.r.d. Only details of F1 and F5 at the ends of the Si/A1 composition range synthesised are given. Large crystals of essentially the same composition as F5 were made from the same starting materials using the same apparatus and temperatures by a procedure based on that reported by Kacirek and Le~hert.~ A more dilute synthesis mixture was prepared with the molar composition shown in table 1. A seeding gel was formed by following the preparative procedure used before for F5, and after a short time, a percentage of this was thoroughly mixed with the synthesis mixture, which was also at 367 K.In the product used in this work, which was pure as judged by electron microscopy and X.r.d., 1 YO of seeding mix was added after it had been aged for 26 h. Other proportions of seed and aging times also were possible, but these were not examined in detail. When the synthesis mixture was left unseeded crystallisation was much slower, and the product was a mixture of faujasite and zeolite S. Large crystals of high aluminium content were prepared by the method of K ~ h l . ~ The synthesis mixture was prepared from sodium silicate solution (1 2 % Na,O, 30 % SiO, from B.D.H.) and AR-grade aluminium wire, sodium hydroxide and potassium hydroxide. The silicate solution was added to the aluminate solution quickly, with rapid stirring at room temperature in a polythene bottle.The mixed reagents gelled in about 30 s, and were transferred to an oven at 354 K for 1 h and then to a water bath at 369 K for 4 h. The product was washed and filtered as before. Complete exchange to the sodium form was carried out by repeated treatment with sodium nitrate solution (1 mol dm-3) which had been made alkaline with sodium hydroxide (0.01 mol dm-3). Washing was completed with sodium hydroxide (0.01 mol dm-3), and finally with small volumes of water. Electron microscopy showed no impurities but the X.r.d. pattern showed small peaks (ca. 1 YO total intensity) due to zeolite A. This is designated sample K. Details of all the zeolites used in this work are shown in the table.Imbibition Experiment Procedure The apparatus differed from that of Fegan and Lowel in that simple glass specimen bottles were used to hold the zeolite samples, and these were capped with polythene stoppers after the desiccator was opened following equilibration. The desiccators were not rocked in the thermostat bath. A simple filter-paper cover proved effective at preventing condensation falling from the desiccator lid into the bottles. The solutionB. M. Lowe and C. G. Pope 947 Table 1. Synthesis and properties of zeolites composition of synthesis mixtures (mol) Si/Al in sample Na,O K,O A1,0, SiO, H,O producta size and morphology F1 75 0 15 60 4500 1.58 irregular shapes, probably agglomerates ; typical overall size ca. 0.5 pm F5 27 0 3 60 900 2.50 as F1 but larger at ca.2pm F5L 27 0 3 60 3600 2.54 spheroidal, platy surfaces typical size ca. 7 pm K 159 53 28.6 60 3550 0.98 as F5L size ca. 6pm a Estimated from the cubic unit cell parameter. D. W. Breck and E. M. Flanigen, Molecular Sieves (Society of Chemical Industry, London, 1968), p. 47. controlling the water activity was placed as close to the samples as practicable to minimise temperature differences, and wide diffusion paths were provided to further speed vapour transport. When experiments were performed above 298 K the use of an unsaturated control solution was preferred, as otherwise the change in solubility with temperature delayed the isopiestic equilibration. One of the sample bottles contained a weighed amount of control solution.Monitoring the weight of this provided an extra check on the composition of the external solution at equilibrium, Results and Discussion The use of glass bottles and the crude capping technique did not introduce errors in the weights of the equilibrated samples larger than the balance precision of f0.2 mg. Equilibration was not speeded up by allowing the control solution to evaporate from a larger surface area, but decreasing the length of the vapour diffusion paths, and avoiding constrictions were important. Equilibrium was typically established in 1-4 days, depending on the control solution composition. Reproducibility of results was similar at 298.2, 313.2 and 328.2 K, and condensation of water onto the sample containers when the desiccator was removed from the thermostat bath at higher than ambient temperature was not important.The theory of the isopiestic method, as presented by Fegan2 and Fegan and Lowe' predicts that at high salt contents the zeolite particles will be surrounded by a solution film in which the water activity is the same as that of the control solution. If the salt content is reduced, whilst the water activity is held constant, the liquid film becomes thinner but remains at the same concentration, as does the imbibed solution. Thus, the slope of this part of the isopiestic plot is constant as it is determined by the water activity, and can be predicted from literature data on the salt solution used. The observed and predicted slopes have always agreed within experimental error in both this and previously published work.', 2 , 6 The slopes of isopiestic plots at low salt concentrations, where no external film of solution is present, depend on the affinity of the zeolite for the salt and the water.Usually this part of the isopiestic curve is observed to be linear, or nearly SO,^^^^'*' and of different slope to the high salt side, although there is no thermodynamic requirement that this should be so.948 Salt Imbibition in Zeolites 0 0-04 0.08 mass ratio salt/zeolite Fig. 1. Isopiestic curves for NaNO, at 298.2 K on small crystal sample F5. The numbers on the curves are the molalities of the external solutions of NaNO,. The arrows mark the salt content at which the large crystal sample showed sharp intersection points for the same external solutions. At the intersection point the liquid film has just disappeared but the imbibed material is still at equilibrium with the external solution.Determination of the intersection point, which defines the composition of the imbibed solution is thus the basis of the isopiestic measurements of salt uptake which have been reported. Fig. 1 shows a set of isopiestic curves measured at 298.2 K in which sodium nitrate was imbibed by sample F5. The distinct curvature of the lines at intermediate salt concentrations which was also observed with F1 and other small crystal samples, made it uncertain as to how the concentration of the imbibed solution in vapour contact with the external solution should be determined. The curves did represent true equilibria, as judged by the following criteria.(i) The experimental points were unchanged after extended contact times. (ii) The slopes of the curves at high salt content were consistent with the compositions of the control solutions. (iii) Addition of water to the zeolite samples, followed by further equilibration, did not alter the final steady weights achieved. A further check on the reversibility of the data was performed by successively changing the control solution from saturated KCl to saturated KNO, and back again. Only two, reproducible isopiestic curves were obtained, irrespective of the order in which the measurements were made. Fig. 2 shows a set of results obtained with the larger crystal sample, FSL, which had essentially the same chemical composition as F5. Little curvature was observed. When data points very close to the intersection were obtained, they sometimes lay slightly above the curves, but the deviation was much smaller than with the small crystal samples, and only slightly greater than the probable experimental error.Results obtained at the same water activity with the two samples are shown for comparison in fig. 3. The intersection which is sharp, or nearly so, occurs at the same salt loading at which the small crystal data begin to display curvature. This was observedB. M. Lowe and C. G. Pope 949 0.40 0 Y .- .-.I 8 \ 92 3 'a 1 E 0 2 0.35 0.30 b 2.14 0.84 0.08 mass ratio salt/zeolite Fig. 2. Isopiestic curves for NaNO, at 298.2 K on large crystal sample F5L. The numbers on the curves are the molalities of the external solutions of NaNO,. It is possible that a small amount of curvature exists at the intercepts, and that this has led to the slope at low salt content appearing less steep with the two most dilute external solutions.0 at each water activity studied. The corresponding concentrations are marked by arrows on the curves of fig. 1. The different behaviour of the two samples seems most likely to arise from the difference in external surface area of the crystals, or from capillary-condensation effects. However, the latter explanation is inconsistent with the results, as the two zeolite samples have similar water contents and because the phenomenon occurs essentially unchanged at high and low water activities. It is possible to explain the observations if salt adsorption occurs more strongly on the external surface of the crystals than within their zeolitic pore space. If the F5 particles consist of agglomerates of material, size ca.0.1 pm, which is consistent with their appearance under the electron microscope, an adsorption of ca. one NaNO, unit per 0.3 nm2 area of external surface would be required to produce the observed effect. The external surface of crystals larger than ca. 2 pm would, on this basis, produce an almost negligible effect on the isopiestic curves. Whilst these magnitudes seem to be possible, it is surprising that adsorption should be so great, bearing in mind the low concentrations of imbibed salts which have been observed in this work, and by others.1,2i6,8,9 We believe that the local salt concentrations which occur in different parts of porous crystals, and on their external surfaces depend on the net result of ion/water and ion/ zeolite surface interactions, and on the influence of the Donnan exclusion effect.In large950 Salt Imbibition in Zeolites I 1 I 0.04 0.08 mass ratio salt/zeolite Fig. 3. Comparison of large and small crystal imbibition of NaNO, at 298.2 K. The numbers by the curves are the molalities of the external solutions of NaNO,. 0, F5L (large crystals); 0, F5 (small crystals). 0 cavities, where salt ions can only approach one part of the zeolite framework closely, the Donnan effect, and the more effective hydration which can be achieved in the external solution generally results in low imbibition. Adsorption on the outside surface of the crystals should be stronger, as Donnan exclusion does not apply there.Stronger adsorption may also occur in small cavities, where stronger ion framework interaction is possible. This probably accounts for the large concentrations of salts which become locked into sodalite and cancrinite during synthesi~,~! lo and may partly explain the profound effect that salts may exert on synthesis pr0ducts.l' We have observed that perchlorate is trapped in faujasite if it is included in the synthesis mixture, though only to the extent of about one ion per two sodalite cages in our experiments.12 Large post-synthesis imbibition of salts is not expected in aqueous systems, because ions which are small enough to enter the windows of small cages have large solvation energies which makes them favour a bulk solution environment. Our results suggest that salt imbibition experiments, whether carried out by the isopiestic or classical analysis procedures, may well give misleading information if external surface adsorption is significant.The isopiestic technique is to be preferred, because the shape of the characteristic curves shows immediately whether this is the case. Fig. 4 shows the imbibition of NaNO, at 298.2 K on F5L over an extended range of water activity. No abrupt change in imbibition behaviour is shown by this system at a water activity corresponding to a saturated external solution of the salt. Decreasing the water activity down to very low values only causes extra salt to enter the zeolite to a small extent. The isopiestic curves are all linear, with a constant slope of - 0.46 0.02B. M .Lowe and C. G . Pope 95 1 0.5 0.9 water activity 0.1 Fig. 4. Imbibition of NaNO, in F5L at 298.2 K over an extended range of water activity, Water activity in a saturated solution of NaNO, arrowed. 4 0 4 6 8 molality of external NaNQ solution 0 Fig. 5. NaNO, imbibition at 298.2 K (O), 313.2 K (0) and 328.2 K (A) on F5L and K samples. Ordinate scales displaced for clarity.952 Salt Imbibition in Zeolites 0.7 0.5 Y* 0.3 0.1 0 \ \\ I I I 4 8 12 ionic strength Fig. 6. yf values at 298.2 K. External solution values are calculated from the equation of Br0m1ey.l~ (a) external chlorate; (b) external nitrate. NaNO, in F5L (0) and in K (a). NaClO, in F5L (0). at the low salt side of their intersection points.This slope is nearly equal to the density ratio, water/solid NaNO,, (0.44) and suggests that the pore space is always full, so that salt uptake can only occur if an equivalent volume of water is expelled. The regular and simple nature of the results appears to confirm further the accuracy of the data, and that complications due to adsorption or capillary condensation are absent. Fig. 5 compares the imbibition by a low and a high aluminium content sample over a range of temperatures. There is little temperature dependence and the F5L results display the form typical of the operation of a Donnan equilibrium, as has been reported b e f ~ r e . ~ Surprisingly, the high aluminium content sample imbibed similar amounts of NaNO,, though previous work had suggested2,9 that higher salt uptake might be expected.The graphs of internal us. external molality were almost straight lines, and displayed little of the curvature which has been attributed to the operation of a Donnan effect. ** Donnan theory leads to the equation mzyf: = mi(mi+m,)(y~i)z (1) for imbibition of a 1 : 1 electrolyte. The terms on the left-hand side relate to the external solution, and those on the right-hand side to the imbibed solution. m, is the molality of the charge balancing cations. Even though plots of mi(mi + m,) us. mf have been found to be straight lines with some of the systems studied in the past, this does not constitute a sensitive or reliable test of the Donnan equation. There appears to be no reason toB. M. Lowe and C. G. Pope 953 6 0 4 0 - c e W 2 0 0 0 3 0 6 0 9 0 Fig.7. Conventional tests of the Donnan equation, NaNO, in F5L at 298.2 K, calculated with different assumed internal cation molalities. 0, Calculated from correct zeolite composition. , Calculated with pure zeolite cation concentration halved. expect that the activity coefficient ratio should be constant. Indeed, the large, varied, and different ionic concentrations outside and inside the zeolite, together with the different environments in the two locations suggest that we might expect to see the ratio vary a great deal. As yf, values are available from independent sources13 we believe that attention should be directed to the terms relating to the imbibed solution. If the zeolite contains imbibed salt, the cations from this material are not distinguishable from the original charge-balancing cations of the same chemical type, and should be included in the calculation in the same way.m, is therefore properly calculated from the chemical composition of the original zeolite, the amount of imbibed salt, and the water content at imbibition. With the values of mi, m,, me and y f , it is possible to obtain y+ from eqn (1). These values are dependent on the environment within the zeolite pores and cannot be calculated from first principles, as the problem involves all the difficulties associated with the treatment of concentrated electrolyte solutions, together with the extra complication of the influence of the zeolite pore walls. are acceptable. (1) y should change systematically with salt imbibition. (2) y f should be of similar magnitude, but probably somewhat higher than expected for an external solution of the same ionic strength.(3) y should vary systematically with the charge on the zeolite framework for a given salt and zeolite. The basis for proposition (2) is that the zeolite cavities must restrict the ability of the solvent to respond to the demands of the internal electrolyte, which must destabilise the internal salt to some extent compared with the free solution environment. Fig. 6 shows We suggest three criteria which might be used to assess whether the values of y954 Salt Imbibition in Zeolites 0 2 4 6 8 10 external molality Fig. 8. Imbibition of sodium nitrate (O), sodium chlorate (a) and trisodium &rate (0) at 298.2 K.the variation of y k and y k with ionic strength of the salt solutions. Bulk solution activity coefficients are calculated using the equations of Bromley. l4 The data for the high aluminium sample follow the bulk solution values remarkably closely. The more siliceous zeolite also yields yki values which vary smoothly with concentration, but are somewhat higher than y+ at the same ionic strength. The results appear to be consistent with the predictions which can be made from Donnan theory. The lower values obtained with the sample K may reflect some stabilisation of the salt by the charge on the zeolite framework. Fig. 7 shows the conventional test of the Donnan equation. The agreement appears convincing, until the results of using quite incorrect charge balancing cation concentrations are also examined.This illustrates quite clearly that the test is insufficiently sensitive to be very useful. The relation between salt and water uptake into initially salt-free zeolites was examined in experiments with sodium nitrate, sodium chlorate, sodium citrate and sodium fluoride. Nitrate and chlorate gave very similar isopiestic curves, and their different structure directing effect in synthesis3 was not reflected in these results. Citrate was used because of its larger size, higher charge and greater water affinity. It was found that as this salt entered the zeolite, the water content remained almost unchanged over a wide range of water activity. The space required to accommodate the salt, and the influence of the ions on the internal water packing clearly produced compensating effects.The imbibition of the three salts is shown for comparison in fig. 8. The low uptake of citrate is consistent with hydration causing the salt to favour concentration in the bulk solution.B. M. Lowe and C. G. Pope 955 Experiments with sodium fluoride were unsuccessful, as the isopiestic curves showed no sharp changes of slope. Thus no estimate of salt occlusion could be made, even though this probably occurred to some extent. The failure of the method arose as follows. Salt entering the zeolite causes little change in water content, due to the strong solvation and rather small molar volume of the salt. The low solubility of sodium fluoride means that no external solution can form except at high water activity, so that the high salt content side of the isopiestic curves is also horizontal with the control solutions used.The sample showed the same X.r.d. pattern before and after fluoride treatment, so that lattice attack was not significant during these experiments. Experience in this work, and with other large crystal material’ suggests that at high water activities and low salt contents, the slopes of isopiestic curves on the low salt side of the intercept have constant slope, or a slope which changes very slowly with water activity except perhaps for very high water activities. Larger variations may well be an indication of external adsorption being important. At low water activity, behaviour appears to be more varied, and no generalisation can be offered at the present time. C.G.P. thanks the University of Otago for study leave, and the University of Edinburgh, Chemistry Department, for its hospitality during 1987. References 1 S . G. Fegan and B. M. Lowe in Proc. 6th Int. Zeolite Conf., Reno, ed. A. Bisio and D. Olson (Butterworths, London, 1983), p. 288. 2 S . G. Fegan, Ph.D. Thesis (University of Edinburgh, 1985). 3 R. M. Barrer and J. F. Cole, J. Chem. SOC., A., 1970, 1516. 4 H. Kacirek and H. Lechert, J. Phys. Chem., 1975, 79, 1589. 5 G. H. Kuhl, Zeolites, 1987, 7 , 451. 6 K. R. Franklin, B. M. Lowe and G. H. Walters, J. Chem. Res. ( S ) , 1988, 32. 7 B. M. Lowe, C. G. Pope and C. D. Williams, J. Chem., SOC., Chem. Commun., 1988, 1186. 8 R. M. Barrer and W. M. Meier, Trans. Faraday SOC., 1958, 54, 1074. 9 R. M. Barrer and A. J. Walker, Trans. Faraday Soc., 1968, 60, 171. 10 R. M. Barrer, J. F. Cole and H. Villiger, J. Chem. SOC. A . , 1970, 1523. 11 R. M. Barrer, Hydrothermal Chemistry of Zeolites (Academic Press, London, 1982). 12 B. M. Lowe and C. G. Pope, unpublished work. 13 J. F. Zemaitis, Jr, D. M. Clark, M. Rafal, N. C. Scrivner, Handbook of Aqueous Electrolyte 14 L. A. Bromley, AZChE J., 1973, 19, 313. Thermodynamics (D.I.P.P.R., American Institute of Chemical Engineers, New York, 1986). Paper 8/02148A; Received 13th September, 1988