No abstract

Using EPR we have resolved the question of whether the dominant Pt defect in silicon consists of an isolated platinum ion or a platinum-platinum pair. We have measured the uniaxial-stress-induced shifts in the g values and find that the stress-coupling tensor shows the defect symmetry to be C2". By selective doping with ' 'Pt and ' Pt, we demonstrate that there is no hyperfine evidence of a second platinum. We therefore conclude that the original model of Woodbury and Ludwig of an isolated substitutional platinum impurity that spontaneously distorts off center in a (100) direction is correct. We have also measured the defect realignment under stress, which occurs even at pumped-liquid-helium temperatures.We find that the sense of the alignment for this platinum defect is opposite to that for the negative vacancy. In addition, we have fully resolved the superhyperfine interaction involving the two nearestneighbor silicon atoms. Lastly, we observe a shift in the g values depending upon the nuclear isotopes of these two nearest-neighbor silicon atoms.

The vacancy model for platinum in silicon as proposed by Watkins postulates a neutral Pt atom in the 5d' electronic configuration occupying a negatively charged lattice vacancy, so that electronic properties of Pt should be similar to those of the isolated vacancy V . We show that this model, including strong Pt spin-orbit coupling and a Jahn-Teller (JT) distortion of C2"symmetry combining tetragonal and trigonal components, and having only -10% of the electronic wave function localized on the Pt, is qualitatively consistent with the results of the electron-paramagnetic-resonance (EPR) studies of Woodbury and Ludwig, which revealed an unusual form for the g tensor (nearly axial about (100) but departing strongly from the spin-only value of 2, with g j & 2 & g~~) . The model accounts also for the anisotropic Pt hyperfine interaction and for superhyperfine interaction found to involve only two of the four nearest-neighbor Si atoms. With three electrons in t2 vacancylike orbitals the JT distortion has two energetically similar forms yielding the same C2"symmetry, one of which occurs for V and the other for Pt . With this identification, opposite signs found for the experimental strain-coupling coefficients of V and Pt may be explained. The vacancy model predicts a positive value for the product g""g»g, the opposite of that given by an alternative model due to Ammerlaan and van Oosten, which predicts -70% localization in the Pt 5d shell. These models can, therefore, be distinguished by an experiment that determines this sign.

No abstract