Multiscale Modeling in Semiconductors: From Defect Formation to Device Performance

by Yu Jin

Institution: University of Washington
Year: 2018
Keywords: Defects; Modeling; Semiconductor; TCAD; Electrical engineering; Nanotechnology; Electrical engineering
Posted: 02/01/2018
Record ID: 2215968
Full text PDF: http://hdl.handle.net/1773/40877


This work is aimed to build a model framework to predict device performance based on the formation of defects in order to meet the demand for higher-performance integrated circuits and solar cells. We use a multiscale modeling technique to investigate the properties of some important defects. Those defects play important roles in the study of precipitation, diffusion and recombination in semiconductors. Ab initio (density functional theory, DFT) calculations are used to extract critical parameters at atomic scale and to verify key mechanisms, while continuum modeling is conducted to describe the defects kinetics and interactions at device scale. Combining process/device simulation and the fundamental understanding at atomic scale, we can gain insight about how process conditions can affect defect formation and therefore device performance. Thus, this multiscale modeling framework can provide useful guidance in performance optimization and cost reduction. Based on this approach, we have developed models for carbon clustering and associated metal gettering, which can be used to reduce noise in advanced silicon CMOS image sensor. We have also advanced models for oxygen precipitation in silicon by considering morphology evolution, dynamic interactions with point defects, and doping dependency. The carbon and oxygen precipitation processes are modeled using the reduced moment-based model (RKPM) with improved computation efficiency. The impact of charged grain boundaries on device performance, as well as electron beam induced current (EBIC) imaging measurement, of CdTe solar cell has been investigated in detail. Based on our simulation results, we propose that passivation with accumulated grain boundaries will be more beneficial to the performance of CdTe solar cell, while depleted grain boundaries generally degrade performance. We also conduct a series of DFT calculations to investigate the light induced degradation (LID) related defects in silicon solar cell. Based on these calculations, a comprehensive model for light induced degradation is proposed which matches experimental observation under full range of conditions.Advisors/Committee Members: Dunham, Scott (advisor).