Experimental study of time-dependent dielectric degradation by means of random telegraph noise spectroscopy

Abstract

Time-dependent dielectric breakdown (TDDB) is commonly used to assess dielectric failures. However, TDDB provides limited insights into the physics of dielectric degradation. In this paper, we explore the potential of random telegraph noise (RTN) spectroscopy to study the physics of dielectric breakdown. RTN is a fluctuation in the dielectric leakage current due to capture/emission of injected electrons by dielectric traps. We report an RTN study of large-area alumina (${\text{Al}}_2{\text{O}}_3$) thin films. A stress experiment is performed on a fresh sample, where RTN is measured before, during and after stress. Important degradation signatures are identified in the RTN spectra. The degradation imposed by the applied stress is observed as a consistent transition between two distributions, where the RTN transitions from an initial pre-stress Gaussian, to a final post-stress exponential. A calculation of the noise entropy, which generally increases with growing material disorder, confirms the transition to an exponential distribution. Finally, we relate the RTN distribution parameters to the defectivity of the dielectric.