Reconstructions and Dynamic Processes on CdTe Surfaces Studied by UHV High-Resolution Electron Microscopy

Surface profile imaging at resolutions of better than 2Å is highly suitable for studies of surface structures and reactions. In the case of semiconductor materials, the main challenge is to prepare surfaces free of any contamination. The technique has previously been used to study surface reconstructions of Si and CdTe. In our previous observations, clean surfaces of CdTe were obtained by careful control of the incident electron beam within a JEM-4000EX high resolution electron microscope with a pressure of 10-7 torr. In the present study, observations of reconstructions and dynamic phenomena on CdTe surfaces were carried out with a Phillips-430ST, modified for Ultra-High Vacuum in the vicinity of the specimen and equipped with an in situ heating facility. The base vacuum in the region of the sample could usually reach ∼3×l0-9 torr after baking the microscope column at ∼120°C for 36 hours. The CdTe specimen was prepared by cutting a large single crystal into 3mm discs in a [110] direction, then mechanically polishing to a thickness of ∼20 microns, and finally ion milling to perforation.When viewed along a [110] projection, the CdTe sample was found to be dominated by clean or nearly clean (111) and (110) surfaces(with amorphous materials less than 5Å) whilst the (001) surface was usually very short and rough. A completely clean surface was obtained by in situ annealing of the crystal to about 200°. The (110) surface was then found to be reconstructed with a very characteristic chevron appearance in the manner described previously. Long and flat CdTe(OOl) surfaces were obtained by insitu annealing of the crystal at ∼510°C at which temperature edges of the crystal started to gradually sublime. Characterization of the surface structure was then possible when the crystal was cooled back down to temperatures below about 300°C. It was found that the (001) surface had a (2×1) reconstruction at temperatures below about 200°C which transformed reversibly into a (3×1) reconstruction over the approximate temperature range of 200°C<T<300°C. Figures la and lb show the (2×1) and (3×1) reconstructed (001) surfaces, viewed along the [110] projection, which were recorded at temperatures of 140°C and 240°C respectively. Structural models for the (2×1) and (3×1) reconstructions, obtained directly on the basis of the experimental images, are shown in Figs.2a and 2b respectively. The (2×1) reconstruction involves a 1/2 monolayer of Cd vacancies and a very large inward contraction of the remaining Cd surface atoms, which then displace the second layer of Te atoms as indicated. This model is similar to that proposed by Chadi for the Ga-rich (2×1) reconstructed GaAs(100) surface. The (3×1) reconstruction involves both the formation of surface dimers and the presence of vacancies at the surface. Every third atomic-pair is missing along the [1,-1,0] direction, and the remaining two atom pairs at the surface form the surface dimer. Although the (3×1) reconstruction has a larger number of electrons in dangling bonds, a surface with vacancies can be relaxed to reduce the strain energy due to the surface dimers. The directions of the atomic displacements away from the ideal dimer positions are indicated in the figure. Relatively large atomic displacements for several layers into the bulk are clearly visible in experimental images, as seen in Fig.lb. Further details of the surface reconstructions can be found elsewhere.