When does atomic resolution plan view imaging of surfaces work?

Surface structures that are different from the corresponding bulk, reconstructions, are exceedingly difficult to characterize with most experimental methods. Scanning tunneling microscopy, the workhorse for imaging complex surface structures of metals and semiconductors, is not as effective for oxides and other insulating materials. This paper details the use of transmission electron microscopy plan view imaging in conjunction with image processing for solving complex surface structures. We address the issue of extracting the surface structure from a weak signal with a large bulk contribution. This method requires the sample to be thin enough for kinematical assumptions to be valid. The analysis was performed on two sets of data, c(6×2) on the (100) surface and (3×3) on the (111) surface of SrTiO₃, and was unsuccessful in the latter due to the thickness of the sample and a lack of inversion symmetry. The limits and the functionality of this method are discussed.