Nanowire Growth Process Modeling and Reliability Models for Nanodevices

Abstract

Nowadays, nanotechnology is becoming an inescapable part of everyday life. The big barrier in front of its rapid growth is our incapability of producing nanoscale materials in a reliable and cost-effective way. In fact, the current yield of nano-devices is very low (around 10 %), which makes fabrications of nano-devices very expensive and uncertain. To overcome this challenge, the first and most important step is to investigate how to control nano-structure synthesis variations. The main directions of reliability research in nanotechnology can be classified either from a material perspective or from a device perspective. The first direction focuses on restructuring materials and/or optimizing process conditions at the nano-level (nanomaterials). The other direction is linked to nano-devices and includes the creation of nano-electronic and electro-mechanical systems at nano-level architectures by taking into account the reliability of future products. In this dissertation, we have investigated two topics on both nano-materials and nano-devices. In the first research work, we have studied the optimization of one of the most important nanowire growth processes using statistical methods. Research on nanowire growth with patterned arrays of catalyst has shown that the wire-to-wire spacing is an important factor affecting the quality of resulting nanowires. To improve the process yield and the length uniformity of fabricated nanowires, it is important to reduce the resource competition between nanowires during the growth process. We have proposed a physical-statistical nanowire-interaction model considering the shadowing effect and shared substrate diffusion area to determine the optimal pitch that would ensure the minimum competition between nanowires. A sigmoid function is used in the model, and the least squares estimation method is used to estimate the model parameters. The estimated model is then used to determine the optimal spatial arrangement of catalyst arrays. This work is an early attempt that uses a physical-statistical modeling approach to studying selective nanowire growth for the improvement of process yield. In the second research work, the reliability of nano-dielectrics is investigated. As electronic devices get smaller, reliability issues pose new challenges due to unknown underlying physics of failure (i.e., failure mechanisms and modes). This necessitates new reliability analysis approaches related to nano-scale devices. One of the most important nano-devices is the transistor that is subject to various failure mechanisms. Dielectric breakdown is known to be the most critical one and has become a major barrier for reliable circuit design in nano-scale. Due to the need for aggressive downscaling of transistors, dielectric films are being made extremely thin, and this has led to adopting high permittivity (k) dielectrics as an alternative to widely used SiO₂ in recent years. Since most time-dependent dielectric breakdown test data on bilayer stacks show significant deviations from a Weibull trend, we have proposed two new approaches to modeling the time to breakdown of bi-layer high-k dielectrics. In the first approach, we have used a marked space-time self-exciting point process to model the defect generation rate. A simulation algorithm is used to generate defects within the dielectric space, and an optimization algorithm is employed to minimize the Kullback-Leibler divergence between the empirical distribution obtained from the real data and the one based on the simulated data to find the best parameter values and to predict the total time to failure. The novelty of the presented approach lies in using a conditional intensity for trap generation in dielectric that is a function of time, space and size of the previous defects. In addition, in the second approach, a k-out-of-n system framework is proposed to estimate the total failure time after the generation of more than one soft breakdown.