(Electronics and Photonics Division Award) The Impact of Defects on the Performance of Semiconductor Devices and Materials

Abstract

Since the early days of the semiconductor industry, defect control has been key to the successful development of devices and circuits. It requires a thorough understanding of their formation and the impact on the electrical material parameters. This has only been possible by the invention of powerful structural, chemical and electrical characterization tools, with the device itself perhaps as the most sensitive probe. This evolution was paralleled by the development of ab initio calculation methods, based on Density Functional Theory and more recently, also TCAD tools allowing more and more refined modelling of the impact of defects on the electrical characteristics of devices.The implementation of other materials than Si and SiO2 in CMOS through the hetero-epitaxial growth on a silicon substrate for example, has renewed the interest in defect engineering during the last two decades, leading to single-defect analysis methods like Random Telegraph Noise (RTN) [1] or novel growth schemes like Aspect Ratio Trapping (ART) [2] for the removal of extended defects from the device active regions.In this presentation, some state-of-the-art analysis techniques will be highlighted, including Deep Level Transient Spectroscopy (DLTS) [3], Generation-Recombination (GR) noise and RTN spectroscopy [1,4] and p-n diode lifetime analysis [5]. These methods will be applied to several case studies. As shown in Fig. 1, threading extended defects impact the recombination lifetime of lowly-doped n-type In0.47Ga0.53As starting from a density of a few 107 cm-2. Likewise, it will be demonstrated that the GR noise observed in GaN-on-Si MOSHEMTs (Fig. 2) most likely originates from threading dislocations [6]. It is concluded that when hetero-epitaxial layers can be grown with a sufficiently low defect density, their impact will be more on the variability of the electrical parameters rather than on the effective values. In addition, the position of the defect with respect to strategic nodes like a p-n junction or depletion region largely determines its electrical impact, as has been validated by TCAD simulations.