We present a new numerical method to calculate the correlation functions in semiconductor materials by a direct solution of the Boltzmann equation. The correlation function is calculated solving a set of time-dependent Boltzmann equations corresponding to different initial conditions. As application, we present the calculation of the correlation function of velocity fluctuations and the noise temperature for holes in silicon at T=300 K. The results are in good agreement with Monte Carlo calculation as well as with experimental data.

We use a full band iterative matrix method to study the high-field transport in n-type GaAs for the first time using three non-parabolic valleys (G, L, and X). The new simulation of transient electrons in GaAs shows that, even if there exist carriers in the X-valley beginning from an electric field E ¼ 10 kV/cm, the two-valley model (G and L) remains still valid until E ¼ 20 kV/cm. We present results for the distribution functions, in the transient and permanent regimes in GaAs in the G-, L-, and X-valleys, for electric fields in the range 1 to 60 kV/cm. These functions allow to calculate all parameters of the first order in both regimes. We obtain good agreement with the experimental velocity-field curves.

We present a new Monte Carlo algorithm, using a Scattered Packet technique to calculate transport parameters in semiconductors. This algorithm allows calculations with a great precision and a significant gain in CPU time as compared to the standard Monte Carlo scheme. As application we report the calculation of the correlation function of velocity fluctuations and noise temperature for holes in silicon at T ¼ 300 K. The results are in good agreement with standard Monte Carlo calculations as well as with experimental data.

International audienceWe extend the iterative matrix method for the solution of the Boltzmann transport equation in multi-band semiconductors to the calculation of second order parameters such as correlation functions and the noise temperature. The method allows also the extraction of the diffusion coefficient in each valley. The results for the case of GaAs with two valleys show that the diffusion in this material is dominated mainly by interactions in the Γ valley. The calculated noise temperature is in good agreement with Monte Carlo result