Quaternary Ammonium Cation Effects on the Crystallization of Zeolites of the Offretite-Erionite Family Part II. Electron Diffraction Studies J. V. Sanders, M. L. Occe 11i,* R. A. Innes,* and S. S. Po1lack+ D,v,~'on of Materials Science, University of Melbourne, Parkville, Victoria, 3052 Australia *Gu 1f Research & Development Company, Pittsburgh, PA 15230 USA +Pittsburgh Energy Technology Center, USnOE, Pittsburgh, PA 15236 USA
Electron diffraction studies of zeolites prepared using six different organic cations confirmed varying degrees of offretite and p.rionite character of the crystals. Both pure offretite and pure erionite crystals werp. identified, as well as crysta1~ that were interqrowth~ of erionite in offretite. Many of the single-phase crystals contained stacking faults of various types. The nature of the offretite-erionite zeolites is thu~ very complex. A variety of materials can be prepared within this family by changing the organic cation and other variah1es in the preparation. INTRODLICTI ON
While chemical properties of zeolites are generally determined by the presence of aluminum atoms in the skeletal structure and by the associated (exchangeable) cati ons, physi ca1 properti es depend upon the cage openings and channe 1 structu res with in the crysta 1s , Those members of the zeo1ite family having channel structures ",1 nm in size are particularly interesting catalytically because this channel size restricts the permeating molecules to a size range that is most useful in the chemical and petrochemical industry. Thus, syntheti c pentas i l zeol ites have been developed for the producti on of hydrocarbons used as liquid fuels. One pentasil, ZSM-5, contains two sets of intersecting channel systems and has a morphology that generally favors access to these channels. Furthermore, ZSM-5 crystals are free from planar growth faults that could introduce an impedance to the channels. This is not so for natural mordenite, whose adsorption properties are comnensurats with channel dimensions much smaller than those indicated by its crystal structure (hence thp. name small-pore mordenite). This indicates a structural b~ockage within the channels, which can be detected by electron diffraction. Similarly, the limited sorptive properties of gmel~nite have been attributp.d to random blocking of the l2-membered- ring channels , Preparations of ZSM-ll frequently contain some ZSM-5 in the form of intergrowths within the ZSM-ll matrix. This system has been studied, a~ fault structures have been characterized by transmission electron microscopy. The structures of erionite and offretite are closely related, allowing intergrowths to form within a crystal, as has been observed in the synthesis of TMA offretite. Stacking faults in erionite are characterized in electron micrographs by con!rast lines running parallel to the intersection of the fault with the surfaces. The offretite-erionite system has been studied in an effort to understand the extent to which the catalytic properties (such as methanol conversion) depend upon the detailed structures within the crystals (as
determined by electron diffraction and electron microscopy). A great diversity of fault structures withinSdifferent preparations, as well as variation of specific catalytic activity, was found. EXPERIMENTAL Tetramethylammonium chloride (TMA), choline chloride (CC), benzyltrimethy1ammonium chloride (BTMA), benzy1triethy1ammonium chloride (BTEA), and l,4-diazobicyc 10 (2,2,2) octane monobasic [DABCO(I)l and dibasic [DABCO(I I) 1 were used to crystallize a set of s~x zeolites of the offretite-erionite family crystal composition and properties are in a manner described e1sewher~; described in the reference given. Specimens were prepared for electron microscopy by ultrasonic dispersion of the zeo1i te powder mi xed wi th sma 11 gl ass beads ina1coho1. Some of the dispersion was collected on a continuous or holey carbon film mounted on a grid. Specimens were examined in a JEOL 100 CX transmission electron microscope fitted with top entry two-axis tilting stage and a UHR objective pol piece (C = 0.7 mm). In this situation, electron diffraction patterns can be obtain~d from crystals tilted by about ± 30° from the horizontal. However, at tilts greater than about 15°, it is not possible to insert the objective aperture, but it is nevertheless still possible to obtain phase-contrast lattice images, with the contrast transfer function being attenuated by the natural limits of the instrument. 6 7 These zeolites are most unstable in the electron beam.' Crystals were quite stable for illumination conditions required for diffraction but were found to have a very short 1ife under the more intense el ectron beam requi red for imaging at magnifications in excess of about 100,000 X. This makes it difficult to obtain lattice images of the crystals, and therefore, diffraction is the main tool used here to examine the structures of these zeolites. Isolated crystal fragments were selected and, when possible, tilted to a zone-axis orientation, and the diffraction patterns recorded at or near that orientation. RESULTS AND DISCUSSION 1. Electron Diffraction Dlffractlon patterns were indexed, and the orientation and phase established. In the two principal orientations shown in Figure 1, it is not possible to distinguish erionite from offretite, nor are faults apparent. In other orientations, however, these two zeolites can generally be distinguished, and faults det~ted by the appearance of streaks in the diffraction patterns. Bennet and Gard observed this streaking in the case of offretite and Linde T. Figures 2a and 2b show diffraction patterns from crystals identified as offretite and erionite, respectively. The patterns are essentially free from streaks (except for slight effects in b), so these two crystals are considered to be free from planar faults or intergrowths. In Figures 2c and 2d, however, there are streaks through all the reflections, which are a maximum at the reciprocal lattice points. This indicates the presence of randomly distributed planar faults in each structure; the faults are normal to (001). In other crystals, a variety of different types of faulting were detected, as shown by Figure 3. In Figure 3a, the spot pattern shows the crystal to be offretite (cf. Figure 2a), and the streaks have a maximum intensity between the points of the reciprocal lattice for offretite, i.e., where one would expect to see a reflection for erionite (Figure Zb}, This pattern therefore indicates that the offretite crystal contains many very thin (one unit cell?) slabs of intergrown erionite. In Figure 3b, the streaks are much shorter and are clearly centered on the erionite position. This crystal is, therefore, an intergrowth of thicker slabs of erionite within an offretite matrix.
J.V. Sanders et al.
In other crystals, patterns contained satellite spots or streaks at the positions of t : n ± 1/4 in the offretite reciprocal lattice. In Figure 3d, these can be seen as slightly elongated spots, and in Figure 3c, they are elongated into streaks. These patterns imply the existence of a regular set of faults within the offretite, with a mean spacing of four times that of the c-axis of offretite. The extent of the superstructures can be assessed bv the extent of the elongation in the streak. A well-ordered arrangement produces a spot, whereas a disordered arrangement, averaging a fault in four unit cells, produces a streak. Faults of this type are listed as "4c faults," and the regu l a r arrangement is 1i sted as a "4c superstructure." TABLE I.
FAULT DISTRIBUTION AMONG DIFFERENT OFFRETITE CRYSTALS
(110) Not Distinguishable Erionite Clean Eri onite Faulted Offretite Clean Offretite + Offretite Faults Offretite + Erionite Faults Offretite + 4c Superstructure Offretite + 4c Faults
3 2* 2*
3 1 6
*Some of these could be offret,te faults. The proportion of crystals containing these various types of faults in any given preparation is not known, but at least one preparation (using BTMA) contained examples of all the types illustrated in Figures 2 and 3. Other preparations contained different amounts; the results of analyses of diffraction patterns from at least ten crystals in five different preparations are collected in Table 1. While BTEA- and BTMA-containing crystals show a range of faults, DABCO(I) and DABCO(II) are both pure erionite, with some faulting in the former, which may also contain some offretite faults. 2.
Fault Structures The 1ayer structures of these material s make faul ting and intergrowths likely. Some of t§e possibilities involving a single fault are discussed by Millward and Thomas in terms of the AABAAB ••• stacking sequence in offretite, and the AABAAC ••• sequence in erionite. It should be noted that in this nomenclature, the intrusion of any C layer, such as in erionite or chabazite, into a matrix of offretite effectively blocks the 12-ring channels, whereas a cancrinite fault would leave the channel clear. For this reason, it is important that the faults be accurately characterized if the catalytic properties of these materials are to be understood. The "offretite type" faults (Figure 2c) could be cancrinite units or B layers displacing A layers, which would leave the channels clear. However, the reciprocal lattice streaks at the erionite reciprocal lattice points (Figure 3a) must indicate the presence of thin erionite units, or C layers, and hence blocked channels. The nature of the faults producing streaking at t : (n ± 1/4) is not yet resolved, but it seems possible to formulate a 12-1ayer sequence with a center of symmetry [t = (n ± 1/2) missing1 either with or without C layers. The 4c faults dominate the BTEA and BTMA-containing crystals but do not appear in the erionite-rich materials containing DABCO I, DABCO II, and TMA ions. These faults contribute intensity with a varying degree (from crystal to crystal) of sharpness at the t : (n ± 1/4) reciprocal lattice positions, accompanied by a marked low in the intensity of scattering at t = (n ± 1/2), an The allowed erionite ref1exion, in an offretite reciprocal lattice. distribution of these faults within a crystal can be seen in the lattice image in Figure 4, where the diffraction pattern (insert) comes from the whole crystal after it has been aligned with the electron beam approximately parallel to
(010). The area in the micrograph in Figure 4 is obviously tilted slightly with respect to this orientation b~ause the lattice image is dominated hy the (100) lattice fringes with an 1l.4A spacing. Thesf! f r inqes are broken into bands roughly parallf!l to (001), which wander across the crystal and must indicate the regions of of f re t i te crystal bf!tween the faults. In some pleces (pointing hand), their spacing is almost regular, with a width of 4 unit cells of offretite and, in others, most irregular, as can be seen in Figure 5 where the numbers refer to the number of unit cells of offretite between each fault. These spacings were determined from higher magnification images, whf!re the ir positions can be more readily identified than at the magnification in Figure 5. An examplf! of this arrangement of faults in a crystal of BTEA-offretitf! is shown in the lattice image in Figure 6. It is not possible to say how much of the changing appearance of the faults across the crystal is structural until Wf! have a better understanding of the effect of tilt, thickness, and defocus on the appearance of the faults. Nevertheless, these images give the impression that each fault is not restricted to a single crystallographic plane. REFERENCES Sanders, ,1. V., Zeolites, 1985, 5,81. 2. Kokotailo, G. T., and Lawton, S. L. Nature 1964, 203, 621. 3. Millward, G. R., Ramdas, S., and Thomas, J. M., J. Chem. Soc. Faraday Trans. 1983, 2, 79, 1075. Kokotailo, G. T., Sawzuk, S., and Lawton, S. L., Am. Miner. 1972, 4. 57, 439. Occelli, M. L., Innes, R. A., Pollack, S. S., and Sanders, ,1. V., 5. in Proc. 7th Int. Zeolite Conf., Tokyo (1986), Submitted. 6. Bursill, L. A., Thomas, ,I. M., and Rao, K-JoCf, Nature 198 1, 289, 157. 7. Bursill, L. A., Lodge, E. A., and Thomas, ,1. M., Nature 1980, 286, 111. 8. Bennett, J. M., and Gard, J. A., Nature 1967, 214, 1005. 9. Millward, G. R., and Thomas, J. M., J. Chem. Soc., Chem. Comm. 1984, 77.
Erionite (nk) Offretite (hk)
66 33 5 4
Figure 1. Electron diffraction patterns of erionite and offretite in tile orientations in lillic;] they cannot be distinguished.
Sanders et al.
9 8 7 6 5 4 3
2 1 1=0
1 2 3
OFFRETITE Figure 2.
tc* L a*
Electron diffraction patterns of erionite and offretite. In (a) and (b), the patterns indicate that the crystals are essentially free of faulting, whereas the streaks parallel to  in (c) and (d) indicate the presence of simple offretite and erionite, respectively.
Various types of faulting in offretite shown by diffraction patterns. (a) Very thin (single-layer) erionite faults in offretite. (b) Thin slabs of erionite in offretite (the 2 indices refer to the erionite reciprocal lattice, and odd orders are streaked). Faults of the 4c type (see text) in offretite appear as streaks for irregular faulting in (e), and spots at 2 = n + 1/4 in (d) for a more regular sequence of faults. -
J.V. Sanders et al.
Figure 4. Lattice image and diffraction pattern of part of a crystal containing BTfolA. Tile faults can be seen as a set of roughly horizontal bands in the direction of the arrows crossing the (100) lattice fringes, the vertical black lines. An area of almost regularly spaced 4c faults can be seen at the top left-hand corner, as indicated bv the hand.
Faults in a crystal containing BTi~A. The numbers indicate the numbers of unit cells of offretite between the fault planes.
Latti ce image of a BTEA-offretite crys tal showi ng "fault" pl anes. Uhen these are regularly separated by two rows of dots (as in area E) they are small bands of erionite.