Numerical device simulation has proven to be invaluable for characterizing the specific properties of semiconductor devices under different operating conditions. Advanced simulation tools are able to extract various figures of merit and allow detailed insight into the physical processes taking place in all regions of the devices. Furthermore, they can be finally employed to optimize the device structures regarding various targets.
Due to the computational power and the available memory in today's average computers, device simulation and optimization can be rigorously employed to pursue these goals. The economical impact is enormous, considering only the cost of one test wafer in comparison to the costs of one average workstation and the maintenance of application setups. For that reason, the development of these tools, both commercially and academically, will be continued in order to provide state-of-the art models, simulation modes, and user-interaction capabilities.
Such extended features are necessary to meet the requirements for today's
advanced radio frequency (RF) device structures, which are supposed to be
characterized also by efficient small-signal capabilities. The basic idea of a
small-signal simulation mode is to linearize the transport equations around
a steady-state operating point and apply sinusoidal contact signals. Various
important figures of merit, such as S-parameters or the cut-off frequency
of a transistor, can be extracted by means of the small-signal simulation mode.
In this introductory chapter the term RF is discussed first, followed by an overview of state-of-the-art RF devices. In addition, the respective capabilities of commercial simulators are presented and compared with academic codes, especially with the newly implemented features of MINIMOS-NT.
The second chapter contains the derivation of the analytical problem and its discretization including the nonlinear solution technique. Since the accurate simulation of advanced MOS devices requires a higher-order transport model as a replacement for the well-known drift-diffusion model, the energy-transport and six moments model provided by MINIMOS-NT are discussed. Furthermore, the small-signal systems for all three transport models are derived. The third chapter deals with the identified concepts of all small-signal capabilities which have been implemented in MINIMOS-NT based on the chosen small-signal approach. The various figures of merit are described and their results presented.
Since the small-signal simulation mode requires the ability to handle complex-valued quantities, the topic of linear equation systems becomes important. As discussed in the fourth chapter, it has been decided to completely replace the formerly applied modules handling the real-valued equation systems assembled by the steady-state and transient simulation mode. Instead, new modules have been introduced which are able to handle both real-valued and complex-valued equation systems. As this decision raised various questions regarding the design, implementation, and application of the new modules, their solutions are a fundamental part of the discussion of a small-signal simulation mode. Thus, the fourth chapter deals with the assembly module, whereas the fifth one discusses the solver modules. Both chapters include an overview of existing modules and the motivation behind each measure taken during assembling and solving linear equation systems.
The simulations presented in the sixth chapter finally demonstrate all of the capabilities discussed before. Different kinds of examples show how well these features can be used to perform advanced device and mixed-mode device/circuit simulations. In addition, higher-order transport models are evaluated for a set of devices. All simulation results are compared either with measurements or with reference results from other simulators.
A final summary and the outlook for future developments regarding the two main topics of this thesis are given in the last chapter. The appendices consist mostly of topics regarding the usability of the newly implemented capabilities. After the documentation of the input-deck interface to all features, the powerful and convenient stepping module of MINIMOS-NT is presented, which has been significantly extended in the course of this work. The third appendix discusses miscellaneous projects such as the interactive and post-processing mode of MINIMOS-NT, the newly designed and implemented library SEILIB for advanced and efficient processing of parameterized input-decks, and eventually one of its most prominent applications, the MINIMOS-NT test. The fourth appendix summarizes the formulae of two-port parameter conversions and the last one explains three different sparse matrix formats.