Fresnel diffraction at {100} platelets in diamond: An attempt at defect structure analysis by high-resolution (3 Å) phase-contrast microscopy

Abstract

High-resolution (0·3 nm) images of platelets in diamond have been obtained using ultra-high resolution pole-pieces with spherical aberration coefficient C _s = 0·7 mm. A through-focal series of images of a (100) defect, viewed edge-on in the [001] zone, shows phase contract which may be interpreted, using the projected-charge-density approximation, as arising from a thin lamella 2 ± 1 atomic layers thick having charge density, as seen by electrons, only slightly less dense than diamond. Image matching, using full-scale N-beam dynamical theory and the method of periodic continuation (MacLagan, Bursill and Spargo 1977) was undertaken using a number of structural models designed to probe the width and charge density of the patelets. The calculation included the effects of beam divergence, crystal tilt and phase changes caused by spherical aberration and objective lens defocus. Models included the nitrogen platelet proposed by Lang (1964), interstitial carbon (also with hydrogen to complete tetrahedral bonds), mixed carbon/nitrogen two and three-layer plates with atomic coordinates simulating a disordered structure, and one and two-layer vacancy discs. The effects of relaxation at the core and long-range strain were investigated. All models were designed to have overall an extrinsic fault vector [100](100], as required by X-ray and diffraction contrast observations. The calculations showed that the fringe contrast is sufficiently complex and model-sensitive to allow some defect structure refinement to proceed, provided experimental parameters, such as crystal thickness and orientation, objective lens defocus and electron optical magnification, can be determined. Furthermore, it is imperative that accurate measurements of the fringe profile across the defect be made. Thus, the photographic film response must be carefully controlled and calibrated. Image matching clearly showed that the platelet involves one extra atomic layer in the space created by the expansion vector. Best agreement for both the image width and contrast was obtained for a mixed carbon/nitrogen platelet two or three atomic layers wide. However, the observations do not rule out a pure nitrogen platelet having two atomic layers in positions shifted slightly, but significantly, from the Lang model. Nor do they exclude the possibility that other minor species, including hydrogen or vacant sites, be present.