6.2.2 Bulk Case

By neglecting in the six moments transport model (eqns. (2.178) to (2.183)) all terms containing derivatives, equations for the homogeneous case are obtained

$\displaystyle \ensuremath{\boldsymbol{\mathrm{J}}}_n$ $\displaystyle = \mathrm{q}\, \mu_n \, \ensuremath{\boldsymbol{\mathrm{E}}}\, n \ ,$ (6.4)
$\displaystyle \ensuremath{\boldsymbol{\mathrm{S}}}_n$ $\displaystyle = - \frac{5}{2} \, \mathrm{k}_\mathrm{B}\, \frac{\tau_S} {\tau_m} \, \mu_n \, \ensuremath{\boldsymbol{\mathrm{E}}}\, n \, T_n \ ,$ (6.5)

0 $\displaystyle = \ensuremath{\boldsymbol{\mathrm{E}}}\ensuremath{\cdot}\ensurema...
...2} \, \mathrm{k}_\mathrm{B}\, n \frac{T_n - T_\mathrm{L}}{\tau_\mathcal{E}} \ ,$ (6.6)
0 $\displaystyle = - 2 \, \mathrm{q}\, \ensuremath{\boldsymbol{\mathrm{E}}}\ensure...
...}_\mathrm{B}^2 \, n \, \frac{T_n^2 \, \beta_n - T_\mathrm{L}^2}{\tau_\beta} \ .$ (6.7)

Inserting eqn. (6.4) into eqn. (6.6) and eqn. (6.5) into eqn. (6.7) yields the equations

$\displaystyle \mathrm{q}\, \mu_n \, E^2$ $\displaystyle = \frac{3}{2} \, \mathrm{k}_\mathrm{B}\, \frac{T_n - T_\mathrm{L}}{\tau_\mathcal{E}} \ ,$ (6.8)
$\displaystyle \mathrm{q}\, \frac{\tau_S}{\tau_m} \, \mu_n \, E^2 \, T_n$ $\displaystyle = \frac{3}{4} \, \mathrm{k}_\mathrm{B}\, \frac{T_n^2 \, \beta_n - T_\mathrm{L}^2}{\tau_\beta} \ ,$ (6.9)

which can further be reduced to a bulk relation for the non-MAXWELLian parameter

$\displaystyle \beta_{n, \mathrm{bulk}} = \frac{T_\mathrm{L}^2}{T_n^2} + 2 \, \f...
...tau_\beta}{\tau_\mathcal{E}} \, \Bigl( 1 - \frac{T_\mathrm{L}} {T_n} \Bigr) \ .$ (6.10)

Unfortunately, the models for the relaxation times are not accurate enough to obtain a realistic estimate for $ \beta_{n, \mathrm{bulk}}$. Therefore the following fit to Monte Carlo data has been developed

$\displaystyle 2 \, \frac{\tau_S}{\tau_m} \, \frac{\tau_\beta}{\tau_\mathcal{E}}...
... x_1 \Bigl( 1 - \exp \Bigl( - x_2 \, \frac{T_\mathrm{L}}{T_n} \Bigr) \Bigr) \ ,$ (6.11)

with $ x_0 = 0.69$, $ x_1 = 1.34$, and $ x_2 = 1.89$. This expression is accurate for doping concentrations around $ 10^{18} \, \mathrm{cm}^{-3}$ but the doping dependence of $ \beta_{n, \mathrm{bulk}}$ is only relevant at lower temperatures (Fig. 6.3).

Figure 6.3: Kurtosis $ \beta _n$ as a function of the temperature $ T_n$ for bulk silicon with the doping concentration as a parameter together with the analytical expression for $ \beta_{n, \mathrm{bulk}}$.
\includegraphics{gpfigure/Beta_Bulk.color.eps}

However, the presented formulation for $ \beta_{n, \mathrm{bulk}}$ did not solve the problem of anomalous SOI simulation results--presumably because the decrease of the kurtosis with increasing temperature occurs too slowly.

M. Gritsch: Numerical Modeling of Silicon-on-Insulator MOSFETs PDF