Studies on the surface area of zeolites, as determined by physical adsorption and X-ray crystallography

Transcription

1 Studies on the surface area of zeolites, as determined by physical adsorption and X-ray crystallography D. J. C. YATES Central Basic Research Laboratory, Esso Research and Engineering Cotnpatzy, Linden, New Jersey Received September 18, 1967 The determination of the surface areas of zeolites is discussed. It is shown that it is incorrect to use the multilayer isotherm method of Brunauer, Emmett, and Teller for solids where only little more than one monolayer can be adsorbed, in cavities little larger than the adsorbed molecules. The areas of such materials can, however, be determined from the beginning of the linear portion of their isotherms (point B). In addition, X-ray spectra can provide an independent method of measuring changes in the surface areas of zeolites. Canadian Journal of Chemistry, 46, 1695 (1968) Introduction Although there is now a very considerable volume of work dealing with adsorption on zeolites, there is very little of it that is concerned with the problem of the determination of the surface area of such materials. In general, although there has been considerable criticism of its theoretical basis (see review in refs. la, 2, 3a), the equation derived by Brunauer, Emmett, and Teller (B.E.T.) (4) has gained wide acceptance as a valid method of determining the surface area of solids. Nevertheless in some instances, it is considered that the B.E.T. equation, at least as regards surface area determination, has its main utility in being a convenient analytical method (lb, 2a) of locating point B (5), the point on the isotherm where a monolayer is adsorbed. Most solid surfaces have physical adsorption isotherms of type I1 in Brunauer's classification (6), and their geometrical structure is such that there is no physical restriction on the number of layers of gas that can be adsorbed. Such multimolecular layers cannot form on zeolites (and some charcoals), as their very small cavities are of similar dimensions to molecules which are adsorbed (7-9). Under such conditions, the isotherms are expected to be of type I in nature. Such isotherms have been reported by many workers using the natural zeolites analcite (9), chabazite (7-9), erionite (lo), and mordenite (7, ll), and with the synthetic zeolites faujasite (7, 12-17), mordenite (IS), and type A (7, 19). Some values of surface areas have been given in the above work (10, 15, 19) and in reviews (20, 21). There has, however, been no agreement on the best method of area determination, some workers using the B.E.T. method (15, 19, 22, 23), others (10, 19) using the Langmuir equation (2b). Barrer has introduced (20, 21) the concept of monolayer equivalent area for zeolites. Some data on a synthetic faujasite are presented, where the area values obtained with the B.E.T. equation are compared with those obtained by the point B method. It is stressed that the B.E.T. equation should not be used for materials with very small pores, such as zeolites. As zeolites are crystalline, the possibility exists of measuring relative surfaces areas by X-ray crystallography. This has been examined, and it is shown that this method can easily be used to monitor surface area changes in zeolites. Experimental Apparat~ts and Materials All isotherms were determined volun~etrically, using argon at 77 OK. For most of the work, conventional Pyrex glass cells and vacuum systems were used, the pressures being measured with mercury manometers (24). In some cases, however, a quartz Bourdon gauge was used to measure pressures (17). The X-ray equipment consisted of a Phillips diffractometer, using copper Ka radiation, with a Geiger counter detector. The zeolite samples were exposed to air while their spectra were being measured. The synthetic faujasite used was an experimental material donated by the Davison Division of W. R. Grace & Co., and had a silica-to-alumina ratio similar to that of zeolite Y of the Linde Co. (25, 26). The structure of this zeolite is similar to that of synthetic zeolite Linde X, which is fairly well known (26-28). The material contained some 15% of a binder, but the nature and surface area of the binder were not given by the manufacturer. The argon, of purity 99.5%, was obtained from the Matheson Co., East Rutherford, New Jersey. Procedure In addition to the fresh material, 11 samples were available which had been used, under varying degrees of

2 1696 CANADIAN JOURNAL OF CHEMISTRY. VOL. 46, 1968 severity, for drying gases. All samples were handled identically. For the surface area measurements I g was loaded into the Pyrex cells, and evacuation begun. The cell was then heated to 150 "C while pumping. After a good vacuum was reached at this temperature, the temperature was increased to 370 "C. Pumping was then continued until a vacuum of 2 x Torr or better was obtained. The cell was then closed, cooled to 77 OK, and argon added. After the isotherm was determined, the argon was evacuated while the sample was warmed to room temperature. The cell was then recooled to 77 "K and helium added to calibrate the cell. After this, the sample was taken out of the cell and weighed. All of the surface area values are given per gram of anhydrous zeolite. The samples were ground in a mortar and pestle for X-ray examination, and that portion which passed through a 300 mesh screen was used. The powder was pressed into a holder using McCreery's procedure as discussed in detail by Klug and Alexander (29). Results Surface Area Values As referred to earlier (la, lb, 2a, 3a, 4-6), for physical adsorption isotherms, point B corresponds to the adsorption of a monolayer ofmolecules. If a molecule of known size such as nitrogen or argon is used, the surface area of the solid can be calculated from a knowledge of the number of moles of gas adsorbed, per gram of solid, at point B on the isotherm. There are two widely accepted methods of determining point B (la, 2). The first is simply to measure the isotherm in sufficient detail so that the shape of the "knee" near point B can be determined accurately. The other method is to use the B.E.T. equation and to determine two or more points on the isotherm in the relative pressure (pip,) region between 0.05 and As the B.E.T. plot gives a straight line, the slope and intercept of which give the monolayer capacity, a minimum of two isotherm points is needed. In practice, for an accurate determination, it is usual to measure 3 or 4 points on an isotherm. For nlolecular sieves, as they contain such small cavities, the isotherm is not type 11, and the B.E.T. equation should not be used. The only way, therefore, that the surface areas can be determined is by measuring the isotherm in a fairly detailed fashion. This has been done in this work, and argon isotherms at 77 OK are shown in Figs. 1 and 2. From the first of these isotherms, it was determined that the point B occurred at a pressure of 1.5 cm Cp/po = 0.071), and this pressure was used to define the point B on subsequent isotherms. This meant that the isotherm I I I I I I I pressure, crn Hg FIG. 1. Argon isotherms at 77 "K on a series of sodium faujasites. FIG. 2. Argon isotherms at 77 OK on a series of sodium faujasites.

3 YATES: STUDIES ON THE SURFACE AREA OF ZEOLITES 1697 TABLE I Surface areas of Na-Y zeolites by two methods Monolayer Surface area Ratio of areas capacity from Surface area Surface area using the point B from point B as % of B.E.T. equation l3.e.t- equation Sample (cn13//s) (m2/g) fresh sample (m2/g) point B method Fresh A B 181.O C D E F G H I J K only had to be closely defined in this region, and that the isotherms did not have to be measured at pressures higher than about 10 cm. After the experiment was finished, the sample was taken out of the cell and rapidly weighed before it could adsorb water from the air. Using this dry weight, the surface areas were calculated, and these values are given in Table I. The areas have also been expressed as a percentage of the area of the fresh sieve. The area of the adsorbed argon atoms needed to calculate the surface area from the monolayer capacity has been taken as 14.6 A2, following Livingston (30). In order to show the inapplicability of the B.E.T. equation to zeolites, we have also constructed a B.E.T. plot for every isotherm. In most cases, satisfactory B.E.T. plots were obtained, but the areas obtained from them (see Table I) were always found to be considerably lower than those obtained by the point B method, except for the samples of extremely low area (H-K). The latter had obviously lost most of their zeolitic character. As the area of the zeolite increased (i.e. going towards the fresh material), the B.E.T. equation gave areas with lower and lower values relative to the areas obtained from point B. It will be seen that the use of the B.E.T. equation with a normal zeolite would cause an underestimate of the area by about 20%. Deterlnination of Changes in Crystallinity The unique characteristic of zeolites as adsorbent~ is that they are crystalline. Most silicas are an~orphous, and some aluminas are partly crystalline, but the very developmen6 of the ad- sorptive property of a zeolite depends on its longrange order, as shown by its regular array of tetrahedrally coordinated cubo-octahedral structural units. With specific reference to faujasite, a considerable amount of work on its structure and properties has been published (12,26-28). In particular, its crystal structure has been determined by a three-dimensional Fourier analysis of its X-ray diffraction patterns by Broussard and Shoemaker (27). While their work was on the 13X faujasite, the structure of 13Y faujasite is quite similar (26). It has been found that various batches of synthetic zeolites, of the Na-X type, have variations in crystallinity depending on the variations in their manufacture.' These differences are proportional to the peak height at a given diffraction angle. Hence, if the instrumental conditions of the X-ray diffractometer are kept constant, changes in crystallinity can be easily followed with reference to the original zeolite. This has been done with all the samples used here. Their X-ray diffraction spectra have been determined, all under identical instrumental conditions, and the decrease in crystallinity has been determined with reference to the peak heights of the fresh zeolite. The peaks with Miller indices 331, 533, and 555 (see ref. 27) have been used. These have d-spacings of 5.70, 3.78, and A, and occur at 28 values of about 15.5, 23.5, and The complete spectra of Na-X over a range of 20 values from 5 to 55 degrees is given in Fig. 1 of ref. 27. In this work ID. J. C. Yates. Unpublished observations.

4 1698 CANADIAN JOURNAL OF CHEMISTRY. VOL the spectra were also measured over the same range of 28 values. However, to make the diagram simpler, typical spectra obtained for our samples are shown in Fig. 3 only for 28 values between 14 and 34 degrees. In all cases the three peak heights were measured and expressed for each of the peaks as a percentage of the peak height of fresh zeolite. The average of these three peaks was then taken and used as the best indication of the overall crystallinity of the sample. The data are given in Table 11. I I I I I I el d8prell) FIG. 3. X-ray diffraction patterns of a series of faujasites of decreasing surface area. Discussion Adsorption Measurements It has been known for some considerable time that most zeolites have type I (6) adsorption isotherms (7-9, 11-14, 18, 19), which are characteristic of monolayer adsorption, and capillary condensation is absent (2,3a). The B.E.T. theory, on the other hand, applies to the case of multilayer adsorption (la, 2,3a), most commonly seen as type I1 isotherms. Under these circumstances, it is not surprising that the application of the B.E.T. equation to type I isotherms yields an incorrect value of urn, the monolayer capacity. In this work, on a faujasite containing a binder, the TABLE I1 Crystallinity changes in Na-Y zeolites d-spacings - Average 5.7 A 3.78 A 2.85 A crystallinity (% of (% of (% of Sample (%) fresh) fresh) fresh) - - Fresh A B H I, J, and K all have zero crystallinity B.E.T equation gave urn values about 20% less than those obtained directly from point B. Similar discrepancies can be found in the literature. For example, Eberly (15) used the B.E.T. equation to determine the area of Linde NaX faujasite and obtained a value of 760 rn2ig. In some isotherms on NaX using argon at 77 OK, when the point B was at 3.0 cm, we obtained monolayer capacity values of 228 cm3/g (of dehydrated zeolite) for Lot No and 230 cm3/g for Lot No R. These materials have been used in earlier work (17, 31, 32). Assuming an argon area of 14.6 A2, these capacities correspond to surface areas of 895 and 903 m2/g. Again the B.E.T. value is too low. In later work on erionite, Eberly (10) used both the B.E.T. equation and the Langmuir equation and considered the latter more reliable. Here, again, the B.E.T. values were considerably lower than the Langmuir values (10). Similar effects had been observed considerably earlier with A type zeolites by Breck and co-workers (19). Other values obtained (23) by the B.E.T. method seem low, although a range of values was given (60S800 m2/g) for zeolites A, X, and Y, with no further details. For some considerable time it has been recognized that isotherms in zeolites can be represented by the Langmuir equation (7-14, 18, 19, 33, 34). In many cases, this equation has been used to obtain zeolite monolayer capacities, and hence surface areas. However, there are many problems in the use of the Langrnuir equation to obtain v, values (3b). For instance, there are numerous isotherms which give excellent straight-line Langmuir plots, but which fail to give consistent urn values (3c). When consideration is given to

5 YATES: STUDIES ON THE SURFACE AREA OF ZEOLITES 1699 the assumptions underlying the Langmuir equation, it seems very unlikely. that they could apply to adsorption in zeolitic cavities. In fact, it has been stated by Young and Crowell (2c): "Charcoal and chabazite are the only adsorbents characterized by type I isotherms and it is certain that adsorption on these two porous solids does not even approximate the severely simple Langmuir picture." With the addition of other zeolites, the above seems a fair summary of the situation. From the above discussion, it will be seen that the B.E.T. multimolecular isotherm gives, as might be predicted for a type I isotherm, incorrect surtace areas for zeolites. The Langmuir equation applies to type I isotherms. It does not necessarily follow, however, that the vm values derived from the equation will be correct (2c, 3b, 3c). This would seem to eliminate all methods, except that of the recognition of point B as the completion of the monolayer. As the zeolite isotherms are type I with no rise in the amount adsorbed at saturation pressure (p,), the possibility exists of using this point as a measure of surface area. For example, with the Na-X sample, which had an argon vm value of 230 cm3/g (corresponding to a surface area of 903 m2/g), the amount adsorbed at saturation was 247 cm31n. Very 'hilar values (248 cm3/g) have been reported by Barrer and Sutherland (12). If this value is taken as equivalent to a monolayer, the area would then be 970 m2/g. On balance, the use of the point B, which is very well established with adsorbents with type I1 isotherms, is considered the most reliable. It is also relatively insensitive to the particular pressure chosen for point B, as shown by the data in Table I11 for argon at 77 OK on NaX (Lot No R). If point B is taken at 3.0 cm, v,, is 230 cm3/g. However, if it were to be taken anywhere in the pressure region from 1.5 to 5 cm, the v, value would be within k2.5 % of that at 3 cm. This is very much less than the uncertainty in the area (om) which the argon atom occupies (30) in the monolayer (14.6 A'). The problems in assign- ing a unique om value to a given molecule have been discussed recently (24. Even nitrogen (at -195 "C), which has probably been used more than any other adsorbate, has been assigned om values ranging from 15.4 A2 to 16.2 A2 (lc, 30). In conclusion, providing that no molecular sieve effects interfere (e.g. neither A nor N, are Pressure (cm Hg) TBLE I11 Argon isotherm at 77 O K on Na-X Corresponding Volume adsorbed surface area (cm3/g) (m2/g) l(sat. press.) adsorbed (19) on Na-A at -195 "C) it is proposed that the determination of point B on an isotherm of a simple, small, nonpolar molecule such as argon, nitrogen, or oxygen offers the most accurate means of measuring the surface areas of zeolites. Other adsorbents which give type I isotherms with the above gases (for example some kinds of charcoal (35)) should also be treated in the same fashion. Some remarks in a recent paper (36) also draw attention to the difficulties of using the B.E.T. equation for zeolites, and a procedure is suggested which seems similar to that used here, although point B was not determined explicitly. X-Ray Measurements Although some considerable effort has been put into using low-angle X-ray scattering as a method of measuring surface areas (Id), it has not been too successful. The difficulty is that only crystallite sizes are given, rather than total surface area. This limitation does not apply here, as the unique characteristic of zeolites as adsorbents is that they are entirely crystalline and normal X- ray techniques can be used. Despite this, there has apparently been only one attempt made very recently (36) to use the X-ray spectra of zeolites as a measure of their surface area. The only details given were : "The changes in surface area as determined by N2 adsorption and in crystallinity as determined by X-ray analysis were always found to be in general agreement." Before this work was started, there was very little information available in the literature about the usefulness of the X-ray spectra as a measure of the decreasing surface areas of zeolites. It should also be stressed that there are almost no data on, and less understanding of, the reasons

6 1700 CANADIAN JOURNAL OF < :HEMISTRY. VOL. 46, 1968 for the breakdown of zeolites at elevated temperatures. For instance, it is conceivable that the outer portions of the sieve crystals might decompose most readily. If the products of this decomposition were amorphous silica and alumina, it is possible that this amorphous material would block the very small (13 A) holes in the outside of the crystal. It is then probable that the surface area would drop considerably, as there would be a much restricted access to the interior of the crystals. However, such a process would not readily be detected by X-rays, as the overall crystallinity of the material would be but little affected. Data given in Tables I and I1 and Fig. 4 show that the above process only occurs to a small 1 I I I I I + '/, Average cryslsllin~ly from X-ray speclra FIG. 4. Relation between average crystallinity and surface area for a series of faujasites. extent under the conditions used here. If the surface area and crystallinity decreased to exactly the same extent, the points would all fall on the 45" line shown in Fig. 4. This line is defined by the fresh zeolite (100% surface area and 100% crystallinity) and the origin, when all surface area and all crystallinity are lost. The points in Fig. 4 lie fairly close to this line, considering the difficulties in measuring the crystallinities. However, there are 6 points below the line (greater loss in area than in crystallinity) and 3 above it, so there is some slight evidence that surface area is lost to a greater degree than is crystallinity. It is concluded that, under the above condi- tions, surface area changes in zeolites can be measured equally well by gas adsorption or by X-ray crystallinity measurements. Nevertheless, the gas adsorption method is the most generally applicable, as there are times when the X-ray spectra vary for other reasons than surface area changes. For instance, if it is desired to ascertain whether the zeolite structure has not been damaged as a result of ion exchange, the X-ray spectra cannot readily be employed to do this. The expected changes in surface area per gram of dehydrated zeolite can be calculated, and values close to these have been found (17) for the Ag-X and Li-X. On the other hand, if a series of zeolites with the same structure and cation but varying areas are under study, the X-ray method may be faster than the adsorption method, although the latter method is probably the most accurate. Absolute Areas Zeolites are unique among highly dispersed materials, as they are crystalline in nature. If its structure is f~~lly understood, and its dimensions established, a given zeolite can then have an "ab~olute" surface area, that is, an area defined crystallographically. Such an area will be independent of the errors and uncertainties inherent in area values derived from physical adsorption isotherms. For the X form of faujasite, many X-ray measurements of its structure have been reported (12, 2&28,37), and the volumes of its supercages have been calculated and converted into surface area values (38, 39). It is known that gases such as argon and krypton do not enter the sodalite cages (17, 40), so that area values obtained from isotherms of these gases can be directly compared with values calculated from the size of the supercages. For argon, with the point B method, the sample of Na-X used in this work has a surface area of 903 m2/g. One of the calculations (38) gave a value of 1400 m2/g for Na-X, which seems rather high. Other calculations (39) gave values between 960 and 1132 m2/g, depending on the model used. In view of the fact that most samples of zeolites probably contain some am;unt of noncrystalline material, surface area values calculated from crystallographic data should be higher than values obtained from isotherm measurements. With this proviso, it is felt that the experimental value (903 m2/g) is in

7 YATES: STUDIES ON THE SURFACE AREA OF ZEOLITES 1701 quite good agreement with the calculated (39) values ( m2ig). As calculations become more precise in the future, it should be possible to use the difference between the experimental and calculated surface areas as a measure of absolute degree of crystallinity of a given sample. At present, no samples of known crystallinity seem to be available, so that X-ray spectra cannot be used to measure the absolute crystallinity of zeolites. The value of 100% average crystallinity was assigned to the fresh Nay zeolite in Table I1 solely to enable relative comparisons to be made, and should not be taken to imply that this material is entirely crystalline (apart from its binder content, whose nature is unknown). 1. P. H. EMMETT. In Catalysis. Vol. 1. Reinhold Publishing Corp., New York, N.Y (a) p. 31; (b) p. 40; (c) p. 38; (d) p D. M. YOUNG and A. D. CROWELL. Physical adsorptionofgases. Butterworths and Co. Ltd., London (a) p. 190; (b) p. 183; (c) p. 109; (d) p S. BRUNAUER, L. E. COPELAND, and D. L. KANTRO. InThesolid-gas interface. Vol 1. Edited by E. A. Floo~. M. Dekker Inc., New York, N.Y (a) p. 77; (b) p. 80; (c) p S. BRUNAUER, P. H. EMMETT, and E. TELLER. J. Am. Chem. Soc. 60, 309 (1938). 5. P. H. EMMETT and S. BRUNAUER. J. Am. Chem. Soc. 59, 1553 (1937). 6. S. BRUNAUER. The adsorption of gases and vapors. Princeton University Press, Princeton, N.J W. E. ADDISON and R. M. BARRER. J. Chem. Soc. 757 (1955). 8. P. H:-EM~ETT and T. W. DEWITT. J. Am. Chem. Soc. 65, 1253 (1943). 9. R. M. BARRER. Proc. Roy. Soc. London, Ser. A, 167, 393 (1938). 10. P. E. EBERLY. Am. Mineralogist, 49, 30 (1964) R. M. BARRER. Trans. Faradav Soc (1944). 12. R. M. BARRER and J. W. SUTH~RLAND. ' roc. ROY. Soc. London, Ser. A, 237,439 (1956). 13. R. M. BARER and W. I. STUART. Proc. Roy. Soc. London, Ser. A, 249,464 (1959). 14. R. M. BARER and P. J. REUCROFT. Proc. Roy. Soc. London, Ser. A, 258,431 (1960). 15. P. E. EBERLEY. J. Phys. Chem. 65, 68 (1961). 16. H. W. HABGOOD. Can. J. Chem. 42,2340 (1964). 17. D. J. C. YATES. J. Phys. Chem. 70, 3693 (1966). 18. R. M. BARRER and D. L. PETERSON. Proc. Roy. Soc. London, Ser. A, 280, 466 (1964). 19. D. W. BRECK, W. G. EVERSOLE, R. M. MILTON, T. B. REED, and T. L. THOMAS. J. Am. Chem. Soc. 78, 5963 (1956). 20. R. M. BARRER. 10th Colston Symposium on the structure and properties of porous materials. Butterworths, London p R. M. BARRER. Brit. Chem. Eng. 4, 267 (1959). 22. P. B. VENUTO, E. L. WU, and J. CATTANACH. Paper presented at Molecular Sieve Symposiunl, Soc. Chem. Ind., London, April R. L. MAYS and P. E. PICKERT. Paper presented at Molecular Sieve Symposium, Soc. Chem. Ind., London. Aoril 1967., - a D.-J. C. YATES, W. F. TAYLOR, and J. H. SINFELT. J. Am. Chem. Soc. 86, 2996 (1964). 25. D. W. BRECK. U.S. Patent No (1964). 26. D. W. BRECK. J. Chem. Educ. 41,678 (1964). 27. L. BROUSSARD and D. P. SHOEMAKER. J.. Am; Chem. SOC. 82, 1041 (1960). 28. R. M. BARRER. Endeavour, 23, 122 (1964). 29. H. P. KLUG and L. E. ALEXANDER. X-ray diffraction procedures. J. Wiley and Sons, Inc., New York, N.Y p H. K. LIVINGSTON. J. Colloid Sci. 4, 447 (1949). J. L. CARTER, P. J. LUCCHESI, and D. J. C. YATES. J. Phys. Chem. 68, 1385 (1964). D. J. C. YATES. J. Phvs. Chem (1965). P.-CANNON. J. Phys. khem. 63, 160 (19%). ' P. CANNON and C. P. RUTKOWSKI. J. Phys. Chem. 63, 1292 (1959). S. BRUNAUER and P. H. EMMETT. J. Am. Chem. Soc. 59, 2682 (1937). C. V. MCDANIEL and P. K. MAHER. Paper presented at Molecular Sieve Svmoosium.. Soc. -them. Ind.. London. Aoril R. M. BARRER, F.-W. BULTITUDE, and J. W. SUTHER- LAND. Trans. Faraday Soc. 53, 1111 (1957). 38. M. M. DUB~IN. Izv. Akad. Nauk SSSR Ser. Khim. 209 (1964). 39. A. V. KISELEV and A. A. LOPATKIN. Kinetika i Kataliz, 4, 786 (1963). 40. L. V. C. REES and C. J. WILLIAMS. Trans. Faraday SOC. 60, 1973 (1964).