Abstract
The Xα scattered-wave method, in its spin-restricted version, has been used to calculate the electronic states associated with an ideal gallium vacancy as well as with the following substitutional transition-metal impurities in gallium arsenide: chromium, manganese, iron, cobalt, nickel, and copper. The gallium vacancy is found to behave as a simple acceptor with an ionization energy of approximately 0.50 eV. In addition, a level of a1 symmetry with considerable amplitude in the vacancy sphere is found at an energy of approximately 0.75 eV below the valence edge. The transition-metal impurities can be broken up into two categories, depending on whether or not the 3d states of the transition metal play an active role in determining the active electronic states of the defect. Those impurities to the left of (and including) cobalt are found to behave in the "standard" way, as the active impurity states in the gap have substantial d character and there is clear evidence of bond formation involving the impurity and the neighboring host ligands. By contrast, nickel and copper appear to behave as simple acceptors in the sense that the active defect levels have little d character and are rather more characteristic of the broken bonds at the impurity site. The d states of this latter group appear as "resonances" in the host valence band and play little direct role in determining the electronic properties of the material. The paper concludes with a discussion of these results and possible factors that might affect their validity.