Technology computer-aided design (TCAD) mimics semiconductor device fabrication, design, and operation by modeling, analysis, and simulation approaches based on fundamental physics or empirical observations. One of the most important uses of TCAD tools is to explore new device technologies, where many exploratory simulations are performed in order to give the device engineers a better understanding of the possible benefits and drawbacks of a potential technology.
Conventional industrial development of new devices based on experimental evaluations involves several iterations of trial and error in fabrication, until a specified goal in terms of design conditions is reached. Therein lies the importance of TCAD, that is to substantially decrease development time and costs of a new semiconductor technology. The fact that computer resources are becoming cheaper, compared to drastically increasing costs of experimental investigations, is further underlining the importance of TCAD-based predictions [51]. Therefore, TCAD plays a crucial role in the evolution of semiconductor technologies by replacing many cost- and time-intensive experiments with computer simulations.
The first numerical modeling was suggested in 1964 for a 1D bipolar transistor [52]. This approach was further developed and applied to PN-junction diodes and junction isolated, double-diffused, and bipolar transistors in the late 1960s. These devices and technology were the basis of the first ICs, but had many scaling issues and process variabilities, i.e., when various devices of the identical fabrication process exhibit significant differences in the electrical or mechanical properties [51]. With the development of electronic devices and ICs in the past decades, TCAD evolved along the way into a very strong branch of electronic design automation. The fast-moving progress in semiconductor technologies thus demands continued research in all areas related to TCAD in order to keep up with the rapid developments. The individual simulation steps involved in TCAD are process, device, and circuit simulations and are discussed in the following.
The accuracy and robustness of the process technology, its variability, and operating conditions of the ICs are critical in determining performance, yield, and reliability of devices. Therefore, process simulations are necessary to accurately predict, for example, the active dopant distribution, the stress distribution, and the device geometry. The development of models and methods for a better representation of the actual physical processes is the key driver for continued advances in process simulations. Model development is typically based on fundamental physics and/or on empirical observations. Process simulation is thus an interdisciplinary field where scientists from chemistry, physics, computer science, mathematics, and engineering collaborate in developing new models involved in predicting the fabrication of various semiconductor devices.
Industries typically prefer simple modeling approaches which approximate the desired properties and simultaneously require negligible computational efforts, i.e., empirical modeling approaches. In order to obtain reliable parameters for empirical modeling it is necessary to calibrate the parameters relative to experimental data. In summary, the ultimate goal of process simulations is modeling and simulation of the fabrication processes, such as oxidation, implantation, and annealing, in order to provide doping profiles and geometries for consecutive device simulations.
In order to be able to predict device characteristics, models for the behavior of electrical devices are necessary. These models, i.e., compact and physics-driven models, are at the core of device simulations, which require device geometries and doping levels as input. Device simulations are particularly useful for predictive parametric analysis of novel device structures, i.e., conducting vast parameter studies to obtain, for example, current density, threshold voltage, on-resistance, and breakdown voltage. Device modeling and simulation enables to obtain a better understanding of conceived properties and behavior of the semiconductor devices and to improve their reliability and scalability. Furthermore, device simulation increases development speed and reduces risks as well as uncertainties. The final output of device simulations is the electrical characteristics of a device. The obtained device properties are forwarded to circuit simulations, which predict the behavior of the final products, such as amplifiers, filters, inverters, and rectifiers.
Circuit simulations use mathematical models with accurate modeling capabilities to replicate the behavior of an actual circuit, i.e., combination of electronic components, such as resistors, transistors, capacitors, inductors, and diodes. The ultimate goal of circuit simulations is to improve the design and the overall operation efficiency of circuits before they are actually fabricated. Typical circuit simulation types are: Analog, digital, mixed-signal, and piecewise linear, which differ in their underlying algorithms. A circuit simulator can be used for various types of simulations, such as: transient, noise, and Monte Carlo analysis, which all provide different kinds of information about the circuit.
TCAD simulation tools are critical to address the full complexity of semiconductor technologies including design rules, parasitic effects, operation in harsh environments, and more. Currently available tools include 1D, 2D, and 3D simulators and support a large variety of application scenarios along the entire TCAD simulation chain, starting from process simulations over device simulations and ultimately to circuit simulations. Current major suppliers of TCAD tools are: Cogenda, Crosslight, Global TCAD Solutions, Silvaco, Synopsys, and Tiberlab. In this thesis, Silvaco’s Victory Process [53] and Victory Device [54] simulators are used.
Victory Process is a general purpose process simulator for applications including etching and deposition, implantation, annealing, and stress simulation. Etching and deposition can be performed via geometrical models for fast structure prototyping or with physical models for detailed process analysis. The simulator provides fast analytical models as well as a very accurate Monte Carlo method for ion implantation. Annealing steps include a comprehensive set of doping diffusion models and a hierarchy of oxidation models.
Victory Device is a simulator, which is used to predict electrical, optical, and thermal behavior of semiconductor devices. It provides a physics-based modular and extensible platform to analyze direct and alternating current as well as the time domain responses for devices manufactured with various materials. It offers an advanced tetrahedral meshing engine for fast and accurate discretization of complex 3D geometries for rapid-prototyping.