Because of their covalent bonding, semiconductors often have surface reconstructions that are quite complex (relative to metals, for instance, which typically only exhibit surface relaxation). Determining these reconstructions can be very important for controlling growth by MBE or CVD, especially when inhomogenous structures are desired.
A relatively untapped method of exploring the reconstructions of semiconductors is the observation with Scanning Tunneling Microscopy of the chemisorption sites of highly reactive atoms, such as halogens. In Surface Science 282, p. 37 (1999), "Atomic Structure Determination for GaAs(001)-6×6 by STM," I was lead author in research which used this technique to determine the structure of one reconstruction of gallium arsenide.
During that research, we also observed the chemisorption of chlorine on the gallium arsenide GaAs(001)-c(8x2) gallium-rich reconstruction. This was reported, for instance, in Physical Review Letters 85, p. 1488 (2000), "Anomalous Mobility of Strongly Bound Surface Species: Cl on GaAs(001)-c(8×2)." In this study we did not consider reinterpreting what the surface reconstruction geometry was, because there was a very well accepted model. However, we could never completely make sense of our data within that model. The apparent chemisorption sites just didn't line up with chemically sensible bonds.
In 2001, the Scheffler theoretical group in Germany reported the possibility the the GaAs(001)-c(8×2) surface was actually a very unusual new structure, which they dubbed the zeta reconstruction (reference needed). This was very exciting, because with this new reconstruction our data made perfect sense. This was pointed out in print in their 2002 comment (reference needed).
Using the computational chemistry package Gaussian98, I am currently pursuing calculations to further understand the energetics of chlorine on this proposed reconstruction. We observed the Cl atoms to hop across the surface with surprising speed at room temperature. Because of the complexity of the zeta reconstruction, this is a very computationally intensive task. I an currently relaxing a model cluster for the zeta reconstruction to its prefered geometry. After that, addition of chlorine atoms can take place.