RELATIONSHIP BETWEEN HYPERFINE PARAMETERS AND THE GEOMETRY OF DEFECTS IN NONMETALLIC SOLIDS

We present a critical assessment of a standard formula relating electron-spin-resonance hyperfine parameters to the geometry of threefold-coordinated defects in insulators and semiconductors. As test systems, we use the E center in SiO2, the Pb center at the Si/SiO2 interface, and the silyl and carbyl radicals. We find that this formula, based on orbital orthogonality and directionality, is in striking disagreement with a variety of self-consistent calculations. The crucial assumption that the molecular orbitals point along bond directions is at variance with simple group-theory considerations. We provide an alternative formulation through a tight-binding Hamiltonian that yields a similar qualitative picture, but without using the offending assumption. This Hamiltonian yields predictions that are much closer to those obtained from self-consistent calculations, and includes the important effects of s- to p-promotion energy and of ionicity. We also discuss the role of ionicity in determining the s and p character of the defect wave function from experimental hyperfine-parameter data. In the case of the E center, we argue that neutral-atomic wave functions are not appropriate. We obtain a better estimate using wave functions from Si(+1). © 1990 The American Physical Society.