Limits of the scaled shift correction to levels of interstitial defects in semiconductors

The state-of-the-art method to calculate defect properties in semiconductors is density-functional theory (DFT) in a supercell geometry. Standard implementations of DFT, like the local density or the generalized gradient approximation, suffer from the underestimation of the band gap, which may lead to erroneous defect level positions. One possible remedy to this problem is the use of the scissor operator, originally introduced in the case of vacancies. Here we report a case study on interstitial hydrogen in silicon and silicon carbide, which shows that the scissor correction cannot always be applied successfully for interstitial defects and can cause significant errors, especially in wide-band-gap materials.