Transmission Electron Microscopy in Micro-nanoelectronics

Introduction

The MOS (Metal Oxyde Semiconductor) transistor is the key component driving the electronic logic revolution for the past 50 years ever since what has become known as Moore’s law was first published [MOR 65]. Moore claimed that the number of components inside a single chip would rise exponentially, increasing by a factor of two every year and a half. After 50 years and 30 technology nodes, and despite the fact that some physicists had predicted a real MOS limit for 50 nm gate lengths and below, Moore’s law still does not show any inflexion. The transistor gate length has continued to decrease from a few microns to a few tens of nanometers and the number of components per chip has crossed over the billions. This trend continues at a constant speed, respecting the initial Moore’s law. Why then are the limitations predicted in the literature still not observed? First, these limitations were based on the idea that evolution was only a matter of scaling and that ultimate transistors would look like the old transistors, that is planar, mostly made up of conventional Si and SiO₂ and fabricated using basically the same processes of that in the 1980s. In fact, transistors still evolve because new materials are being integrated; they are built following new architectural rules and fabricated using different, alternative, processes.

Although the “scaling down” evolution was accompanied, and sometimes even guided, by process simulations that were based on robust, well-understood ...