Abstract

The underlying physical mechanisms of NBTI, arguably the most important ULSI reliability problem, are not well understood. Until recently reaction diffusion (RD) models were widely and quite justifiably accepted to explain NBTI. However, recent studies by several independent groups indicate that the RD models are incorrect or at least seriously flawed. Recently, Penn State and Texas Instruments researchers proposed a quite different NBTI model (for pure SiO(2) gate devices) in which the instability is triggered by electron tunneling from an oxide oxygen vacancy to a silicon valence band hole. A new "on the fly" magnetic resonance technique provides results strongly supporting this proposal.