High Resolution Tem of Semiconductor Defects

Computational techniques are now well developed for the determination of the electronic structure of semiconductor surfaces and their line and planar defects. These calculations depend sensitively on the atomic structure assumed for the defect, and so provide an outstanding challenge for high resolution electron microscopy, while it is unlikely that the details of the charge redistribution which controls the electronic structure of defects will be directly imaged for many years, the use of a-priori chemical knowledge may frequently greatly limit the number of likely defect structures. New high resolution imaging methods for the study of surface roughness, an area of almost total ignorance, are described elsewhere in this volume (1). It seems likely that the electronic structure of the line and planar defects will only be solved when all the information available from such techniques as EPR, SDP Hall effect measurements, CTS and EBIC are considered (see (2) for a review of recent work). The unique contribution of high resolution electron microscopy (HREM) is its non-statistical capability of analysing isolated, well characterised defects. The outstanding problems which semiconductor defects offer for HREM are (i) the difficulty of obtaining three dimensional defect structure information (ii) the need to extract information beyond the point resolution limit of current machines (but within their information resolution limit (3)) (iii) the need for chemical (atomic number) information. These three problems are briefly discussed below in the light of recent work.