Site Group Interaction Effects in Zeolite-Y Part 2.-Na-Ag Selectivity in Different Site Groups BY ANDRE MAES* AND ADRIEN CREMERS Katholieke Universiteit Leuven, Centrum voor Oppervlaktescheikunde en Colloi'dale Scheikunde, De Croylaan 42, B-3030 Heverlee, Belgium Received 13th October, 1976 The influence of the nature of the ion neutralizing a certain site group on the selectivity in another group is studied in zeolite-Y. The high Ag to Na preference in the small cages in the binary system is reduced by filling the large cavities with Cs+ or NH: ions. The selectivity data are fitted in terms of the occupancy and selectivity coefficient of 4 (binary system) and 2 (ternary system) types of sites. The calculated occupancies in each site agree with the observed ones, which are obtained from X-ray studies.The thermodynamic ion exchange formalism for the case of several groups of sites is discussed. One of the characteristic features of most zeolitic ion exchangers is the association of the framework charge with various types of cationic sites. Very often, the number of available crystallographic sites exceeds the number of negative charges to be neutralized and changes in the occupancy factors are expected, depending on the nature of the neutralizing cati0ns.l. Such site heterogeneity, combined with the fact that ion exchange reactions are rarely thermodynamically ideal and that certain ions may only be partially exchanged, for steric or other reasons, leads to ion exchange isotherms of great complexity which are often difficult to rationalize in terms of the properties and extents of the various site groups.In a few cases, ion exchange isotherms have been sucessfully analysed in such terms 3 9 and these analyses indicated that a given group of sites may be assigned a characteristic set of thermodynamic state functions for some arbitrary pair of cations. In fact, this is the point taken by Barrer and Klinowski in their theoretical approach to the case of several groups of homogeneous sites. Admittedly, these authors recognize that each group of sites may fail to behave ideally, and this is formally taken care of by assigning to each group a characteristic non-ideality pattern of its own. By an appropriate choice of combinations, they were able to produce some complex Kielland plots which agreed closely with experimental data.More recently, Barrer, Klinowski and Sherry extended this approach to cases in which part of the zeolitic ions are not exchangeable. One of the most crucial points in these papers is the assumption of the existence of a given equilibrium constant for some arbitrary exchange reaction in a given set of sites. The purpose of the present paper is twofold: first, to elaborate on some conceptual difficulties involved in formalizing thermodynamically the exchange of ions in ion exchangers with interacting site groups, and secondly to show experi- mentally that such site group interactions do in fact exist. Synthetic zeolite-" is an excellent system for comparing a given exchange reaction in the small cages in the absence and presence of a cation which is unable to penetrate these areas of the crystal.In particular, we present a study of the Naf-Agf equilibrium, in the presence of a large excess of cesium ions, which confine the 136A . MAES AND A . CREMERS 137 Na+-Ag+ equilibrium to the small cages, and show that such an equilibrium is significantly different from what is observed in the binary system. THEORETICAL Confining, for simplicity, our attention to monovalent ion exchange reactions, the overall thermodynamic constant for the ion exchange reaction between a solid (z) and a solution (s) is defined as &+A, f A,+B, (1) Azf K , = - Bzf BaA' A,, B, represent the equivalent fractions of the ions in the exchanger and fA/fB the activity coefficient ratio in the solid; aB/aA represents the activity ratio in solution which, at sufficiently low ionic strength, may be identified with the molality ratio.The overall selectivity coefficient K , is defined by B A Kc = Ktf If (3) As usual, the standard and reference states are the homoionic solids, in equilibrium with an infinitely dilute solution of the corresponding ions. In the case of an exchanger containing n types of site, Barrer and Klinowski and Barrer, Klinowski and Sherry define an equilibrium constant for the ith set by in which At, BL represent the equivalent fraction of ions in set i andf:lfy the activity coefficient ratio, which formally takes care of the non-ideality effects within the corresponding site group. Evidently, if Xi represents the equivalent fraction of the ith set (with respect to the ion exchanger capacity), then n n A, = XiAi and B, = XiBi.1 1 The individual thermodynamic constants are related to the overall thermodynamic constant by n n K , = n K f i or AGO = zXiAGP ( 5 ) 1 1 in which AGO stands for the overall standard free energy change and AG; for the standard free energy change within the ith set. Implicit in the foregoing approach is the definition of the chemical potential pq in the ith set: ,up = ',u;+RT 1dA'f A. ' (6) Of course,fq -+ 1 when At -+ 1. The activity coefficient f:, or more generally, the activity coefficient ratio ftfl? within the ith set is some complex function of the ion exchanger composition and carries the burden of all interaction effects resulting from changes in composition between ions positioned in sites belonging to all different groups.From a rigorous thermodynamic point of view, this approach may be criticized138 SITE GROUP INTERACTION I N ZEOLITES on the following grounds. Eqn (5) is based upon the view that the standard free energy content of one equivalent of zeolite may be expressed as a sum of terms, GP, referring to the standard free energy contents of the various site groups (including the interaction terms between the ions positioned in the site group concerned). However, such a sum should contain a number of cross terms referring to interaction energies between different site groups, and these terms are not necessarily identical for two different ionic forms. Therefore, the meaning of the concept of a standard free energy content of a given site group is not clearly delineated and is ambiguous.The difficulties may be analysed in a more detailed fashion. In studying the thermodynamic properties of a given set of sites, one is assigning to it a thermodynamic status of its own, and one is therefore faced with the logical consequence that the activity coefficients are interrelated through the Gibbs-Duhem equation. This cannot be expected to be the case for the following reason. In attempting to express the composition dependence of the activity coefficients, one has a choice between two alternatives : the first is to express it as a polynomial in terms of composition changes of all different sets ; the second alternative, apparently taken by Barrer and Klinowski, is to rely on a ‘‘ simple concentration dependence ”, for example in terms of composi- tion changes in the site group concerned.The second alternative may be justified in view of the fact that all site groups are in equilibrium with each other, (being in equilibrium with the liquid phase) so that a given composition in one site group automatically fixes the composition of all other groups. Whichever choice is made, the coefficients in such a polynomial must carry the interdependence of composition changes between the different site groups. This interdependence, however, is primarily governed by differences in affinities of the competing ions for the different site groups and not by cation-cation interactions. It is therefore to be expected that the activity coefficients are not interrelated through the Gibbs-Duhem equation.These coefficients, and consequently the thermodynamic equilibrium constant for a particular site group, are not to be considered as thermodynamic quantities in the true sense. There may be no reasonable alternative to the one presented by Barrer and Klinowski and the case of interacting site groups may not be treated in a rigorous thermodynamic manner altogether. In any case, the foregoing criticism should not prevent an attempt to analyse ion exchange equilibria in terms of “equilibrium constants ” or preferably selectivity coefficients for the various site groups. One should keep in mind, however, that these quantities only yield quantitative measures of differences in ion-site affinities and are not to be identified with thermodynamic quantities in the formal sense.EXPERIMENTAL The zeolite used in this study is the Na-Y faujasite sample of unit cell formula Na54A154Si1380384 241 H20. Preparation methods and details of experimental procedure can be found el~ewhere.~ The Ag+-Na+ binary equilibria were studied at 25°C and 0.01 mol dm-3 total concentration, and the silver and sodium distributions were monitored using y-emitting isotopes llomAg and 22Na. The Ag+-Naf equilibria in the presence of cesium and ammonium ions were studied using the following procedure : first, a separate study was made of the ion exchange behaviour of Cs+ in Na-Y at high Cs+ loading in order to verify the maximum exchange limit and selectivity behaviour. The Ag+-Na+ equilibria were then studied in the presence of an excess of 3.55 x lo3 and 1.8 x lo3 ions/u.c.Csf ions and an excess of 7.1 x lo3 ions/u.c. NHZ ions. The corresponding molarities in the equilibrium solution are 0.3 and 0.15 (Cs+) and 0.6 (NH1;) mol dm-3. Preliminary tests showed this excess to be sufficient to exclude both Na+ and Ag+ from the supercages and confine these ions exclusively to the small cages.A. MAES AND A . CREMERS 139 RESULTS AND DISCUSSION The silver-sodium binary equilibria and those in the presence of a high excess of cesium or ammonium are summarized in fig. 1, which shows the variation of In K,(Ag+-Na+) as a function of the overall silver loading. It should be emphasized that, in calculating selectivity coefficients, the Na+/Ag+ molality ratios have been identified with the activity ratios in both binary and ternary systems.This simplification requires no comment in the binary study which was carried out at 0.01 mol dm-, total concentration, but requires some justification for the case in the presence of a high excess of CsN0,. The problem relates to obtaining activity coefficients of traces of AgN03 and NaN03 in a relatively concentrated CsNO, solution (0.15 or 0.3 mol dm-3). Harned and Robinson * showed that the activity coefficient of a trace of a given salt in an excess of another salt is intermediate 4 G 3 22 C - 2 \ I I I 025 050 0.75 ZA, FIG. 1.-Surface composition dependence of in $K, in the binary system (O), and in presence of an excess of 3.55 x lo3 (O), 1.8 x lo3 (A) Cs+ ions/u.c., and 7.1 x lo3 (A) NH: ions/u.c. between the activity coefficient of its own solution and the activity coefficient of the other salt at the same ionic strength, but generally closer to the latter.Taking the 0.3 mol dm-3 CsNO, case as an example, the activity coefficients of the three electro- lytes at 0.3 ionic strength are 0.602 (CsNO,), 0.606 (AgNO,) and 0.666 (NaNO,). It is apparent that the behaviour of AgN0, and CsNO, is entirely similar and that, therefore, the trace activity coefficient of AgN0, in a 0.3 mol dm-, CsNO, solution will be given by the numbers shown. For NaNO,, the activity coefficient of which differs by some 10 % from the CsNO, value, its trace activity coefficient is expected to differ by < 5 % from the value of CsNO,, in view of the argument presented above. Therefore, the Na+/Ag+ molality ratio may safely be identified with the activity ratio.The standard free energy for the Ag+-Na+ system, as obtained by graphical integration,1° is - 5.43 kJ mol-l. This figure shows that the silver-to-sodium]I 40 SITE GROUP INTERACTION IN ZEOLITES selectivity is significantly lowered in the presence of Cs+ or NH: ions, the effect being roughly the same in both cases. The ion exchange data for Cs+ in Na-Y are summarized in fig. 2, the upper part of which shows that the maximum exchange limit is 68 %, i.e. some 36.5 ions/u.c., a result which is in agreement with the one obtained by Sherry.ll The lower part of this figure shows the normalized selectivity coefficient log K: (Cs+-Na+) at high cesium loadings. These data are also in excellent agreement with those of Sherry, i.e.the limiting value of K,” (Cs+-Na+) at high Cs+ loading is ~ 3 . SCS 0.8 0.9 I I I 0.5 - k2 M 0.4- 0 - ZCS FIG. 2.-Na+-Cs+ exchange isotherm showing the limit of maximum Cs+ exchange (upper part) and the log E K c as a function of the normalized surface composition (lower part). Data of Sherry (solid line) are shown for comparison. Inspection of fig. 1 shows clearly that the Ag+ to Naf selectivity in the presence of Cs+ or NHZ ions is significantly lower than in the binary system. Before analysing these differences quantitatively, it is imperative to show that, in the presence of these other ions, the Ag+-Na+ exchange is restricted to the small cages. A first indication is provided by the data in fig. 3, which shows that the sum total of Ag++Na+ never exceeds the limit of 17.5 i0nsju.c.The small cage occupancy in the presence of NH: is lower and may be understood in terms of a slightly different neutralization pattern in small and large cavities.I2 A second indication is to be found in the Cs+ selectivity at high loading. As far as Na+ is concerned, the Cs+/Na+ ratio in the equilibrium solution is x 100 (in the case of a Cs+ excess of 3.55 lo3 ions/u.c.) irrespective of the amount of Ag+ added. Since the K, (Cs+-Na+) value is m3 at high Cs+ loading (see fig. 2) it follows that the Na+ occupancy in the supercages is 0.15 i0nslu.c. at most. When the Cs+ excess is reduced to 1.8 x lo3, this would amount to some 0.3 ions/u.c.A . MAES AND A . CREMERS 141 The Ag+ occupancy in the supercages can be estimated on a similar basis. On the basis of calculations to be presented below, the selectivity coefficient K, (Cs+-Ag+) in the supercages is near unity.Therefore, at low Ag+ loading, when the Cs+/Ag+ ratio in the equilibrium solution is very high (=lo4), the Ag+ occupancy in the supercages is entirely negligible. At higher Ag+ loadings, the Cs+/Ag+ ratio in the equilibrium solution is 200 to 300, which would correspond to 0.3 Ag+ ions/u.c. Similar calculations can be made in the case of NHZ; the selectivity coefficient KF(NH+,-Na+) at high loading is =4,l2 and the NHZ excess is even larger. Consequently the involvement of the supercage in the Ag+-Na+ exchange is restricted to roughly 1 % of its capacity and we may be confident in analysing the data in the presence of Cs+ or NHZ in terms of ion competition for the small cage sites.I I I I I 2 4 6 8 Aga& ions1u.c. FIG. 3.-Determination of the small cage capacity in an excess of 0, 3.55 x lo3 Cs ions1u.c. ; and A, 7.1 x lo3 NH; i0nslu.c. In order to analyse the selectivity data quantitatively, we used the X-ray data, presented in the Part 1 ; as a guideline, we assume that the total charge in the small cages is 17.5 and that the maximum number of silver ions which can be accommodated in site I is =4 (no Na+ ions are found in site I).2 In attempting to assign the remaining 30.5 ions to the supercages, we may choose from two alternatives : 10 ions/u.c. can be assigned to site 11, leaving the remaining 26.5 ions unlocalized. This choice could be justified on the basis of the finding of X-ray work, that no more than 10 Na+ ions could be localized in site IL2 An excellent fit of the Ag+-Naf binary data, as shown in fig.4, is then obtained by using the K , data listed in table 1. Another alternative would be to assign 20 ions/u.c. to site 11. The motivation for this choice is found in the fairly high selectivity of Ag+ ions in the supercages as indicated by the data of Theng, Vansant and Uytterhoeven l2 and the existing parallelism between selectivity and extent of ion 10calization.l~ This possibility is confirmed by the fact that in Y-zeolite, more K+ ions are localized, 20 in site IIY1 than Na+ ions.2 This1 42 SITE GROUP INTERACTION I N ZEOLITES choice necessitates a slight shift in the K, value assigned to I’ and I1 and gives an equally good description of the data. In any case, the shape of the Kielland plot at increasing Ag+ loading is not very sensitive to the choice of combinations for the supercage and it would be logical to expect shifts in the course of the exchange.Either choice leads to values for the overall AG values [using eqn (5)] which are in 1 I I I I 5 10 15 Ag i0nslu.c. FIG. 4.-Comparison of the experimental (same symbols as in fig. 1) and calculated selectivities as a function of the number of Ag+ ions/u.c. in the binary (solid line = fit a ; broken line = fit b) and the ternary system. excellent agreement with the result obtained by integration: -4.94 (fit 1) or -5.28 (fit 2) kJ mol-l. The values are slightly more negative than those obtained by Sherry,ll which were derived only on the basis of data at high loading. The data in table 1 are used to calculate the way in which the various site groups are occupied by Ag+ ions as the silver content is increased in the equilibrium solution.The results of these calculations are summarized in fig. 5 which shows the relative occupancy of the site groups as a function of the overall silver loading. Apparently, TABLE 1.-Two SETS (a AND b) OF PARAMETERS (SELECTMTY COEFFICIENT Kc AND MAXIMUM NUMBER OF IONS IN EACH SITE GROUP/U.C.) USED TO FIT THE BINARY Na+-Ag+ EXCHANGE EQUILIBRIUM. THE FRACTION OF TOTAL CAPACITY (Xi), In Kc AND THE FREE ENERGY ASSOCIATED WITH THE RESPECTIVE SITE GROUPS (- Xi RTln Kc) ARE ALSO TABULATED. ma. -Xi RT In Kc site KC ions1u.c. (Xc) In Kc jkJ mol-1 I a 2000 4 b 2000 4 I’ a 2 13.5 b 4 13.5 I1 a 20 10 b 10 20 unloc. a 4 26.5 b 4 16.5 0.074 0.074 0.25 0.25 0.185 0.37 0.49 0.25 7.601 7.60 1 0.693 1.386 2.996 2.303 1.386 1.386 1.41 1.41 0.44 0.88 1.39 2.14 1.70 0.88A .MAES AND A . CREMERS 143 the most important conclusion to be drawn from these data, is the very high affinity of silver ions for site I. An illustration of the nature of the agreement of X-ray data with those obtained by least squaring the ion exchange data is shown in table 2. The K, values listed in the tables represent only a quantitative measure of ion site affinities. However, it is worth emphasizing that the K, value, assigned to site I, 0 50 0.40 I00 x +.I ." 0.30 0.75 8 Q +a .C1 v) 4-r 0 .C1 Y 0 0.20 (d Ct: 0.10 425 5 10 15 Ag ions/u.c.FIG. 5.-Fraction of each site occupied by Agf- as a function of the total Ag+ content in ions/u.c. in site I (A, A), site I' (0, 0), site I1 (a, 0) and unlocalized (+, 0). Filled and open symbols respectively refer to the first (fit a) and second assumption (fit b). is ambiguous, as it really refers to a displacement reaction of Na ions from site I' by silver ions taking up positions in site I. The Ag+-Na+ equilibrium data for the small cages in the presence of Cs+ or NH: are impossible to fit in terms of the same occupancy numbers used for the binary data, and a significant shift of Ag+ ions has to be postulated from site I -+ 1'. The TABLE 2.-cOMPARISON BETWEEN THE POPULATION OF Ag+ IONS IN EACH SITE GROUP DETER- MINED BY X-RAY DIFFRACTION TECHNIQUES AND CALCULATED BY USE OF THE PARAMETERS OF FIT a AND b I I' I1 U total Ag+/ (ions/u.c.) tit a fit b X-ray fit a fit b X-ray fit CI fit b X-ray fit a fit b X-ray 2.0 2.00 2.00 1.9 0.01 0.03 - 0.1 0.1 - 0.05 0.03 - 7.25 3.84 3.84 3.9 0.3 0.61 0.3 1.90 2.05 1.8 1.19 0.74 1.25 14.00 3.92 3.92 4.4 1.16 1.89 1.8 4.8 5.9 6.0 4.08 2.31 1.8144 SITE GROUP INTERACTION I N ZEOLITES results are sumniarized in table 3.It is apparent from the nature of the fit, shown in fig. 4, that the parameters shown in table 3, give a close description of the ion exchange data. Two important observations can be made at this point : first, the presence of Cs+ (or NH;) ions in the supercages leads to a significant reduction in the silver affinity for site I ( ~ 4 kJ mol-I). The free energy loss associated with the exchange of Na+ for Ag+ ions in the small cages in the binary system can be calculated from the parameters in table I and is compared in table 3 with the data from the ternary system.The Ag+ to Na+ preference in the small cages is reduced from 5.72 kJ (fit a) or 7.05 kJ (fit b) to 4.6 kJ mol-I when the large cavities are respectively occupied by mostly Naf and Cs+ ions. Therefore, it appears that the free energy change for a given ion exchange reaction in a particular set of sites, (leaving aside the basic difficulties treated in the theoretical section) is very much dependent on the composi- tion of other site groups. Perhaps the nature of the effects described here can be qualitatively understood in terms of a strong interaction between the poorly hydrated TABLE 3.-cOMPARISON OF THERMODYNAMIC DATA FOR THE Na+-Ag+ EXCHANGE IN THE SMALL CAGES ONLY IN THE BINARY AND TERNARY SYSTEMS fraction of max.small cage -Xi R T h Kc site K, i0nslu.c. C.E.C. in Kc /kJ mol-1 ternary I 400 1.75 0.1 5.99 1.49 I‘ 4 15.75 0.9 1.356 3.11 binary I a 2000 4 0.229 7.601 4.37 b 2000 4 0.229 7.601 4.37 I’ a 2 13.5 0.774 0.693 1.34 b 4 13.5 0.774 1.386 2.70 and very polarisable Cs+ (and perhaps NH4) ions and the oxygen framework; this may lead to a weakening of the coordination bond of the silver ions in the hexagonal prisms. The second important aspect is the significant shift of Ag+ ions from the hexagonal prisms to the cubo-octahedra (I + 1’) upon introducing Cs+ ions into the supercages. Evidently, there is some logic in this since, as explained on theoretical grounds by Mortier,14 a change in energy difference between two sites, I and I’ in our case, should be reflected by a corresponding change in the ratio of occupancy numbers at these sites.In conclusion, we have presented experimental evidence that a change in the composition of the crystal may lead, as has sometimes been invoked in the past, to a relocation of some ions from one type of site to another. In addition, it is shown that the selectivity pattern for a particular site may be affected by the nature of the ions which are used to neutralize the electrical charge in other site groups, as was hypothesized in a recent paper.15 Acknowledgment is made to the Belgian Government (Programmatic van het Wetenschapsbeleid) for financial support. W. J. Mortier and H. J. Bosmans, J. Phys. Chem., 1971, 75, 3327. M. L. Costenoble, W. Mortier and J. B. Uytterhoeven, J.C.S. Paraday I, 1976, 72, 1877 E. Gallei, D. Eisenbach and A. Ahmed, J. Catalysis, 1974, 33, 62.A . MAES A N D A . CREMERS 145 * R. M. Barrer and W. M. Meier, J. Inoug. Nuclear Chem., 1966,28, 629. R. M. Barrer and J. KIinowski, J.C.S. Faraday I, 1972, 68, 71. R. M. Barrer, J. Winowski and H. S. Sherry, J.C.S. Faraday II, 1973, 69, 1669. A. Maes and A. Cremers, Adv. Chem. Ser., 1973,121,230. R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 1959). * H. S. Harned and R. A. Robinson, Multicomponent Electrolyte Solutions (Pergamon, 1968). lo G. L. Gaines and H. C. Thomas, J. Chem. Phys., 1953,21,714. l 1 H. S. Sherry, J. Phys. Chem., 1966,70,1158. l2 B. K. G. Theng, E. F. Vansant and J. B. Uytterhoeven, Trans. Faraday Soc., 1968, 64, 3370. l3 M. Costenoble and A. Maes, J.C.S. Faraduy I, 1978, 74, 131. l4 W. J. Mortier, J. Phys. Chem., 1975, 79, 1447. A. Maes and A. Cremers, J.C.S. Faraday I, 1975, 71,265. (PAPER 6/1920)