Finding the Good Stuff in Hrem Images

Abstract

Except in the most carefully designed installations, the commercial availability of TEM,s with point resolution near to or below 2Å has resulted in a class of instruments whose contrast transfer at Scherzer defocus is limited by instability damping, but not by zeros in the contrast transfer function. This means that in HREM images from these instruments there is often much lattice-scale information about periodicity phases at the exit surface of the specimen, along with the usual lattice information about spacings and interspot angles. Moreover, this information is not scrambled by zeros in the contrast transfer function. Because of the large amount (e.g. tens of megabytes) of data in individual HREM negatives, however, guidelines on finding as well as interpreting the useful phase information may be helpful.In work with micro-crystalline periodic structures (even those only a few unit cells across), our guidelines are the same as those used in setting up HREM zone axis images of single crystals: Look for regions of maximum symmetry in the intensity of lattice periodicities. The difference is that the search in HREM images is commonly done in direct space, whereas the search while tilting for a single crystal zone axis is commonly done in diffraction (i.e. reciprocal) space. We illustrate here with an example taken from a study of the complicated (and beam sensitive) family of PVO catalysts. Fig. 1 shows a region 2.3mm on a side from a 240kx negative of a fine-grained (VO)2P2O7 catalyst whose structure was not conclusively known, in part because of the difficulty of growing large crystals. The region, which represents only 7×10−4 of the negative area, was chosen as part of a promising crystal by inspection. The power spectrum of this image has a 7×7 array of spots which index nicely with the <010> projection of the candidate structure shown in Fig. 3a. A 2x closeup of Fig. 1 after Fourier window filtering of these periodicities is shown in Fig. 2. If the region with periodicities most nearly equal in strength is chosen again (here the lower left corner of Fig. 2), and the result is rotated to match the model orientation, Fig. 3b results. The correlation between image and model structure is striking. Some words of caution, of course, are in order. The periodicities observed are present in the specimen crystal, but the periodicity phases at best represent a map of exit surface electron properties. The rigorous connection between image and specimen atomic structure therefore requires knowledge of specimen thickness at least. Nonetheless, without an ability in direct space to recognize such regions of high symmetry, the correlation (rigorous or otherwise) would have remained hidden in a stack of TEM negatives.