High-Resolution Electron Microscopy of Microcrystalline, Partially

Images, taken with a point-to-point resolution of ca 2.4 8, of three specially selected samples provide direct evidence for the occurrence of differen...
0 downloads 0 Views 814KB Size
J. Phys. Chem. 1981, 85,3007-3010

3007

High-Resolution Electron Microscopy of Microcrystalline, Partially Crystalline, and Amorphous Silicates L. A. Burslllt and J. M. Thomas’ Department of Physical Chemistry. University of Cambridge. Cambridge CE2 IEP, Unned Kingdom (Received: March 10, 1981; In Final Form: June 22, 1981)

Images, taken with a point-to-point resolution of ca 2.4 8, of three specially selected samples provide direct evidence for the occurrence of different, theoretically identified, ultrastructural characteristics in amorphous solids. A thin film of uranium oxide on a uranyl-exchanged Y zeolite conforms to the microcrystalline model, with mean correlation length of ca. 10 A. A partially crystalline A zeolite is composed of both crystalline and essentially amorphous regions, whereas a sample of silica gel apparently approachesthe classical random network model.

Some of the difficulties experienced in obtaining and interpreting high-resolution electron microscope (HREM) images of amorphous structures have been reviewed by H0wie.l The most crucial problem is that of overlap, and it is argued’ that, since the scattered amplitude is essentially a superposition of two-dimensional atomic amplitudes, and is hence insensitive to the height coordinates of the atoms, there may be many possible structures corresponding to a given observed image. Furthermore, when the specimen thickness is several times greater than the maximum distances over which correlations in amorphous structures are thought to extend (ca. 15 A), the images will be dominated by superpositions between atoms which are not correlated. Under such circumstances any structural signal reflecting bond lengths and angles will increasingly be lost in a background characteristic of a completely random materiaL2v3 In the preceding paper4 it was shown by computer simulation that the overlap problem is even more severe than most electron microscopists had hitherto supposed. Thus, for a random network model of As, atomic positions may be deduced directly from an optimally focussed HREM image (2.4-A resolution) only if the specimen film is at most 3-6-A thick. This result explains why little, if any, definitive structural data have so far emerged directly from HREM studies, where estimates of film thicknesses refer to values from 20 to 200 A. On the other hand, any departures from randomness which lead to alignment of atom centers along the projection axis are readily detected in the image, since the contrast from overlapping atoms is greatly enhanced relative to the intensity from individual atoms. (Even then the image interpretation is not straightforward for specimens thicker than ca. 20 A (at 200 kV), and precise knowledge of the lens transfer function and the electron optical operating conditions is r e q ~ i r e d . ~ ) We now present three distinct examples of HREM images, recorded at ca. 2.5-A resolution on a JEOL-2OOCX electron microscope, and discuss the extent to which reliable conclusions may be reached regarding their ultrastructure. The examples chosen cover a range of silicate specimens which exhibit apparently microcrystalline, partially crystalline, and apparently random network structures.

‘School of Physics, T h e University of Melbourne, Parkville, 3052, Victoria, Australia.

Uranyl-Exchanged Y Zeolite. An Example of a Microcrystalline Model Crystals of Na-Y zeolite6 (idealized formula NassA156Sil,60381.250H20) may be imaged at ca. 3-A resolution if they are first dehydrated.’~~Even so they are rapidly converted, usually within minutes, into an aluminosilicate glass upon examination in an electron beam with a dose (-lo6 electrons A-2) required for HREM. (Chemical exchange of alkali ions by uranyl (U022+)ions effectively stabilizes Y-type zeolites with respect to electron irradiation, provided also that the preparations are adequately d e h ~ d r a t e d . ~However, observation of as-exchanged specimens, prior to drying, leads to rapid amorphization, and the concomitant growthlo of a thin surface film, effectively supported on an amorphous zeolitic substrate.) Figure 1shows an HREM image of this film. Note that point spacings of ca. 2.7 A predominate in the image. Arrays of black (or white) spots and fringes a pear which are correlated over distances up to ca. 15-20 indicating a microcrystalline texture. The arrays of spots are variously pseudorhombic (symmetry approximately 2 mm), pseudosquare (4 mm), or pseudohexagonal (6 mm). Optical transforms from a number of areas ca. 150 A diameter all showed similar features and a typical example is inset in Figure 1. Note the almost continuous diffuse ring indicating that the predominant spacin s in the image range from 2.3 to 3.4 A, peaking at ca. 2.7 ,whereas the width

1,

x

(1)A. Howie, J. Non-Cryst. Solids, 31, 41 (1978). (2) 0. Krivanek and A. Howie, J. Appl. Crystallogr., 8, 213 (1975). (3) M. V. Berry and P. A. Doyle, J. Phys. C, 6 , 46 (1978). (4) L. A. Bursill, L. G . Mallinson, S. R. Elliott, and J. M. Thomas, J. Phys. Chem., preceding article in this issue. (5) D. F. Lynch, A. F. Moodie, and M. A. O’Keefe, Acta Crystallogr., Sect. A, 31, 300 (1975). (6) Y-type zeolites are derived from the parent faujasite structure by leaching out A13+ ions by complexation or appropriate acid treatment, thereby increasing the Si/Al ratio and hence the thermal stability of the structure, which, on X-ray evidence, is still essentially that of faujasite (see D. W. Breck, “Zeolite Molecular Sieves”, Wiley-Interscience, New Ynrk. -_ ..,.19741. - - . .,.

(7) J. M. Thomas, G. R. Millward, and L. A. Bursill, Phil.Trans. R. SOC. (London), Ser. A, 300, 43 (1981). ( 8 ) L. A. Bursill, E. A. Lodge, and J. M. Thomas, Nature (London), 286, 111 (1980). (9) L. A. Bursill, J. M. Thomas, and K. J. Rao, Nature (London),209, 157 (1981). (10) This could well become the basis of a new technique for the controlled growth, by exsolution under electron irradiation, of metallic or metal oxide f h s onto an aluminosilicate substrate. Such “supported” materials are often of great value as highly specific and active heterogeneous catalysts (see P. A. Jacobs, “Carboniogenic Activity of Zeolites”, Elsevier, Amsterdam, 1977).

0022-365418112085-3007$01.25/0 0 1981 American Chemical Society

3008

BursiU and Thomas

The Journal of phvscal Chemistry. Vol. 85. No. 20, 1981

Flswe 1. High-resolution electron microscopic image, and corresponding optical hansform (inset), of a thin surface film produced in situ In lhe coin8 of elechoo-baam k?a&t+m of W,r..exaansed Y-type zeolite. Note nAcrmstaRne textue, vWb in lhe image. with cwrelstbns extern% up to ca.20 A. and the almost continwus diffuse rina of the o & a l transform. In M area labaied A M r e is an amorphous zeoutlc film wimout an overlapping surface film.

of the diffuse ring indicates a mean correlation length of ca. 9 A. Note that the surface film is discontinuous; for example, the area labeled A in Figure 1 gives contrast typical of amorphous zeolite. A survey of the various phases of uranium oxide, and examination of a-U03 and UOz+. phases by HREM (Bursill, Pring, and Thomas manuscript in preparation) revealed that U-U separations, when projected normal to common low index zone axes, should cluster around 3.5 for a-UOa, around 2.2-3.4 A for UOz+,, and around 2.5-2.9 A for a-U metal. Thus we tentatively identify the surface film as a microcrystalline form of UOz+z,with cubic fluorite derived structure, containing short-range correlations up to 15-20 A and mean correlation length ca. 10 A. We note that, on the basis of the optical transform, and presumably also for electron and/or X-ray patterns from much larger areas, this specimen could reasonably have been classified as amorphous. The use of HREM has, in this case, yielded valuable information. It is impossible to estimate the film thickness, hut since U has an extremely large scattering factor for electrons, it seems reasonable to suppose it may be less than 10-8, thick. It is clear that structural information has been retained in the image despite the fact that the film is supported on an amorphous zeolitic substrate, which itself may be several times thicker than the surface film. This emphasizes the point made above and in ref 4 that alignment of atom strings or planes parallel to the projection axis produces intensity which dominates the observed contrast. It further suggests that HREM images of crystalline materials are perhaps not as sensitive to defects, such as missing

atoms, substituted atoms, or even interstitial atoms, as is usually assumed. Thus, in the present example, strings of a few (say 5) uranium atoms produce sufficient intensity to mask the effect of rather larger numbers of presumably randomly arranged Si", AI3+, and 0" ions in the amorphous substrate. A-Type Zeolite. An Example of a Partially Crystalline Solid High-resolution images of Na-A zeolite" (idealized formula Na,zA11zSiIz0,~27HzO)have already been reported,8 and three mechanisms of transformation from the crystalline to amorphous state identified! Figure 2 displays the penultimate stages of this transformation whereby islands of Na-A structure (10L104 A2 in projected area) are immersed in essentially amorphous aluminosilicate. Having identified the contrast features characteristic of the tunnels and the corresponding cubooctahedra of A-type zeolite (see Figure 1of ref 8 for comparison of experimental with computer-simulated images, and inset Figure 2 for structural projection) the eye may now readily detect a number of isolated tunnels (indicated in Figure 2). Thus departures from randomness are again evident in the image, allowing partially crystalline regions, and even the penultimate stage of retention of a single column of the original crystal, to be identified. The latter (11) Specimens of Na-A, like apecimens of Na-Y type zeolites, may r e a d i be synthesized in aqueous environments. Na-A samples used here were prepared by the method of J. F. Charnel1 (J. Crytt. Growth, 8, 291-4 (1971)).

HlgMiesolution Electron Microscopy of Silicates

TM Journal of Physical Chamishy, Vol. 85. No. 20. 1981 3008

m u r e 2. High-resolution image of A-type zeolite, showing Islands of crystalline structure immersed in an amorphous zeolitic matrix. Note the single isolated tunnel (square outline) visible in the amwphous region. The inset shows a polyhedral prolectim drawing of the A-typB zeolite down (100). See ref 8 for further details.

%re 3. HWesoluUon image, and correspondingo p t h i transfwm (inset). of a Mmmerdal siiica gel specimen. indicating no apparent departure from randomness.

3010

J. Phys. Chem. 1981, 85, 3010-3014

is entirely dependent on the alignment of atoms along the projection axis.

Silica Gel. An Approach to a Random Network Model Figure 3 shows a HREM image of a silica ge112specimen. The corresponding optical transform (inset Figure 3) indicates that the Scherzer defocus condition is well-satisfied, with spacings in the range 2.4-6.3 A appearing in the image. In this case there is no evidence for departures from randomness, either in the image or its optical transform, and it might be argued that this is consistent with a random network model for silica (see, for example, ref 13). However, our considerations in ref 4 imply that no structural information is available from such an image, since the film thickness is unknown. If it could be established that the film is