INTERACTION OF SELENIUM WITH THE GAAS(001)-(2X4)/C(2X8) SURFACE STUDIED BY SCANNING-TUNNELING-MICROSCOPY

We have used the scanning tunneling microscope (STM) to study the deposition of Se onto the GaAs(001)-(2 X 4)/c(2 X 8) surface and the formation of the GaAs(001):Se-(2 X 1) surface reconstruction. The experiments were carried out in a combined STM-molecular-beam-epitaxy (MBE) system, allowing the surface to be imaged after deposition without transferring through air. Se was deposited onto well-characterized GaAs buffer layers grown in situ by MBE. We have imaged the surface at several stages during the deposition of up to two monolayers of Se, with depositions carried out at 300-degrees-C and 470-degrees-C. We have also looked at the effect of annealing the Se covered surface at over 520-degrees-C, observing smoothing of the surface and desorption of Se from the surface. Se reacts with the GaAs surface at 470-degrees-C, displacing the surface As-and causing significant disruption to the GaAs surface. After approximately 0.25 monolayer of Se is deposited, the original (2 X 4) reconstruction is replaced by a largely disordered phase. With further Se deposition the Se-terminated (2 X 1) reconstruction is formed. The general surface morphology is similar to the original GaAs surface prior to Se deposition except that there are many small (2 X 1) islands on the terraces. When Se is deposited onto the GaAs surface at 300-degrees-C a disordered Se overlayer is formed. The (2 X 1) reconstruction is only formed by subsequent annealing at higher temperature. The resulting surface morphology is similar to that formed by Se deposition at 470-degrees-C. High-resolution STM images resolve the Se dimers that make up the (2 X 1) reconstruction; however, the individual atoms within one dimer have not been resolved. Small islands on the GaAs(001):Se-(2 X 1) surface can be removed by annealing at above 520-degrees-C for up to 20 min. The resulting (2 X 1) surface has well-ordered flat planes several hundred angstroms across, which are comparable to those on the original GaAs buffer layer. The temperature at which the islands are removed is very close to the temperature at which Se is desorbed from the (2 X 1) terraces, suggesting that the removal of the islands involves desorption of Se and is not a simple diffusion process.