High-voltage and power semiconductor devices are used in many areas of modern electronic products. To satisfy the demands of customers, chip manufactures are attempting to improve the reliability of their devices. Many reliability issues under investigation are related to high energetic carriers, called hot carriers. Among others, they are responsible for interface and oxide degradation and also trigger impact-ionization. The latter leads to device break-down, latch-up, or snap-back effects, which can lead to thermal damage. This problem is especially important in smart power devices that combine low power and low voltage devices together with high power and/or high-voltage devices on a single die. Many parasitic devices are created by this combination, which can lead to device failures.
There are many highly sophisticated methods available for modeling hot carrier effects, especially for impact-ionization and its associated failure mechanisms. For modeling, the detailed shape of the carrier distribution function is essential for good results. In theory, this can be obtained by solving the Boltzmann Transport Equation (BTE). One method that can be used to solve the BTE is the statistical Monte Carlo method. It gives good results but requires a considerable amount of computational resources. Using the method of moments, the BTE can be reduced to macroscopic transport models, of which the simplest, the Drift-Diffusion (DD) model, uses only two moments. More advanced models are the Energy-Transport (ET) and the Six Moments (SM) model, using four and six moments, respectively. The more moments that are used, the better the results are for the hot carrier populations. However, using these complex models on smart power devices, which often have relatively large dimensions, leads to long simulation times or even to major convergence problems. Since the devices under investigation are relatively large, the non-local relations between the local electric field and hot carrier effects, such as impact-ionization, can essentially be neglected. This is the reason why the majority of simulations on larger devices can still be performed using the relatively robust and simple DD transport model. Work regarding reliability issues therefore focuses on achieving best results from DD to be used by device designers to improve their products in the context of reliability.