High-field ODMR investigation of the EL2 defect in semi-insulating GaAs

The microscopic structure of the EL2 defect in semi-insulating GaAs has remained controversial. It is generally agreed that EL2 is an As antisite-related defect, but whether it is an isolated AsGa defect or an AsGa-related pair defect, has not been unambiguously decided. In a previous investigation with electron nuclear double resonance (ENDOR) detected via the magnetic circular dichroism (MCD) of the near infrared absorption of EL2+ (MCDA-ENDOR) at 24 GHz (K-band) the local symmetry could not be determined conclusively due to complicated multiple ENDOR line splittings caused by strong effects of pseudo-dipolar couplings. We report on an MCDA-ENDOR investigation at 95 GHz (W-band) where the undesired 75As ENDOR line splittings due to pseudo-dipolar coupling were suppressed. The surprising result is that the angular dependence of the nearest 75As shell shows almost Td symmetry. A small non-equivalence of the nearest neighbours is hidden in an unresolved ENDOR line width for B0 vertical bar [0 0 1] and can be a most 1.5% of the superhyperfine (shf) constants. The lines of a further 75As shell, interpreted as the fifth shell, are split into at least two subshells indicating a pair defect in spite of the almost equivalence of the nearest neighbours. It is discussed whether EL2 is an AsGa+-VGa3- pair defect following recent theoretical calculations showing that this pair defect has the observed properties of the shf interactions