Modeling and Simulation of Wide Bandgap Semiconductor Devices

State-of-the-art silicon carbide (SiC) devices have matured to powerful prototypes demonstrating the promising properties of SiC as basic material for high-power, high-temperature, and high-frequency applications. Apart from further technological progress in this field, numerical simulations based on accurate device models are more and more required for design and optimization of SiC devices. In this work, an extended electrothermal drift-diffusion model formulated within the framework of phenomenological transport theory is described with emphasis on pointing out the differences and extensions compared to conventional semiconductor transport models. Based on a commercially available general-purpose multi-dimensional device simulation tool originally developed for simulation of Si devices, various physical models particularly relevant to SiC are implemented on the source code level. They include the Poole-Frenkel generation-recombination mechanism, a generic model describing anisotropic material properties, and general impurity kinetics of shallow and deep traps. The implications of these models on quasi-stationary and transient device behavior are investigated by numerical simulation of ``real'' state-of-the-art 4H- and 6H-SiC devices as well as by experimental characterization, thus allowing for immediate comparison and calibration of the model parameters. All required model parameters are systematically discussed and summarized to a template parameter set for 4H- and 6H-SiC device simulation. The pn-junction, being the basic building block for semiconductor devices, is analyzed based on this set of parameters including the implemented new models. It is found that Shockley-Read-Hall statistics excellently model the characteristics of forward biased pn-junctions. The physical mechanisms leading to excess reverse leakage currents, on the other hand, are still unclear and further theoretical work based on detailed experimental investigations is required for a progress in explaining the measured data. The concept and the limits of the quasi-static approximation of impurity kinetics are revisited. For device operation beyond these limits, the consistent coupling of impurity kinetics to the electrothermal transport model reveals various effects arising from the dynamic ionization of dopants and traps. These are governed by corresponding ionization time constants which have been measured using deep level transient spectroscopy (DLTS) and thermal admittance spectroscopy. The basic impact on device characteristics results from a dynamically enlarged extension of depletion regions if the rise time of a reverse bias pulse is equal or smaller than the characteristic ionization time constant. The measurement results prove that transient incomplete ionization is negligible for dopants like nitrogen and aluminum, at least within today's high-power device operation areas, whereas boron significantly influences the dynamic device characteristics even above room temperature. Related possible device failure mechanisms are discussed including dynamic punch-through which may occur in back-to-back junction configurations and a dynamically reduced blocking capability of pn-junctions. Additionally, a detailed numerical analysis of the DLTS and thermal admittance spectroscopy measurement methods provides new insights concerning the interpretation of the measured data which can not be obtained by applying the conventional analytical theory of these methods. Furthermore, the influence of serial resistances on the evaluation of the measured signal is discussed in detail. The basic aspects of anisotropic material properties, in particular the free carrier mobility, are discussed and their impact on the device performance are demonstrated. The studies show that, depending on polytype and device structure, it can become indispensable to accurately model the anisotropic effects, because they may significantly alter the internal device behavior and the terminal characteristics. Finally, it is shown that the concept of inverse modeling is applicable to state-of-the-art SiC junction field effect transistors (JFET) by carefully analyzing several device characteristics at various bias conditions and temperatures.