4.4.1.3 Integral Form of the Nonlinear Boltzmann Equation

To show the generation of the distributions $ G^{+}$ and $ G^{-}$ the integral representation of the stationary Boltzmann equation (4.10) is used. First, the scattering operator in (4.10) is reformulated as:
$\displaystyle Q[f_{s}]=$   $\displaystyle [1-f_{s}(\vec{k})]\int f_{s}(\vec{k}^{'})S(\vec{k}^{'},\vec{k})\,...
...lpha(\vec{k}^{'})\lambda(\vec{k}^{'})\delta(\vec{k}-\vec{k}^{'})\,d\vec{k}^{'}-$  
    $\displaystyle -f_{s}(\vec{k})\biggl\{\int[1-f_{s}(\vec{k}^{'})]S(\vec{k},\vec{k}^{'})\,d\vec{k}^{'}+\alpha(\vec{k})\lambda(\vec{k})\biggr\},$ (4.63)

where the self-scattering rate $ \alpha(\vec{k})$ has been introduced. Note that the delta function guarantees that the self-scattering does not change an electron state. Free-flight times are generated using the total scattering rate $ \lambda(\vec{k})$. Thus the self-scattering rate has to satisfy the equality

$\displaystyle \lambda(\vec{k})=\int[1-f_{s}(\vec{k}^{'})]S(\vec{k},\vec{k}^{'})\,d\vec{k}^{'}+\alpha(\vec{k})\lambda(\vec{k}).$ (4.64)

This gives for the self-scattering rate the following expression:

$\displaystyle \alpha(\vec{k})=\frac{1}{\lambda(\vec{k})}\int f_{s}(\vec{k}^{'})S(\vec{k},\vec{k}^{'})\,d\vec{k}^{'}.$ (4.65)

Further, an additional differential scattering rate $ \widehat{S}(\vec{k},\vec{k}^{'})$ is introduced

$\displaystyle \widehat{S}(\vec{k},\vec{k}^{'})=[1-f_{s}(\vec{k}^{'})]S(\vec{k},\vec{k}^{'})+\alpha(\vec{k})\lambda(\vec{k})\delta(\vec{k}-\vec{k}^{'}),$ (4.66)

$\displaystyle \int \frac{\widehat{S}(\vec{k},\vec{k}^{'})}{\lambda({\vec{k}})}\,d\vec{k}^{'}=1.$ (4.67)

Now taking into account (4.65) and (4.67) the scattering operator (4.64) takes the conventional form:

$\displaystyle Q[f_{s}]=\int f_{s}(\vec{k}^{'})\widehat{S}(\vec{k}^{'},\vec{k})\,d\vec{k}^{'}-f_{s}(\vec{k})\lambda(\vec{k}).$ (4.68)

Using the Neumann series of the forward equation the second iteration term (4.47) is derived as an example:

$\displaystyle f^{(2)}_{\Omega}=$   $\displaystyle \int_{0}^{\infty}\,dt_{2}\int_{t_{2}}^{\infty}\,dt_{1}\int_{t_{1}...
...a}_{2}\int\,d\vec{k}^{a}_{1}\int\,d\vec{k}_{i}\cdot\{f_{0}(\vec{k}_{i})\}\times$  
    $\displaystyle \times\biggl\{\exp\biggl(-\int_{0}^{t_{2}}\lambda[\vec{K}_{2}(y)]...
...\vec{K}_{2}(t_{2}),\vec{k}_{2}^{a})}{\lambda(\vec{K}_{2}(t_{2}))}\biggr\}\times$ (4.69)
    $\displaystyle \times\biggl\{\exp\biggl(-\int_{t_{2}}^{t_{1}}\lambda[\vec{K}_{1}...
...\vec{K}_{1}(t_{1}),\vec{k}_{1}^{a})}{\lambda(\vec{K}_{1}(t_{1}))}\biggr\}\times$  
    $\displaystyle \times\biggl\{\exp\biggl(-\int_{t_{1}}^{t_{0}}\lambda[\vec{K}(y)]...
...c{K}(t_{0}))\biggr\}
\Theta(t-t_{1})\Theta_{\Omega}(\vec{K}(t))\Theta(t_{0}-t).$  

Here $ \Theta(t)$ is the step function and $ f^{(2)}_{\Omega}=\int f^{(2)}(\vec{k},t)\Theta_{\Omega}(\vec{k})\,d\vec{k}$. From (4.71) it is seen that if the free-flight time is calculated from the exponential distribution according to the scattering rate $ \lambda(\vec{k})$, the conditional probability density for an after-scattering state $ \vec{k}^{'}$ from the initial state $ \vec{k}$ is equal to $ \widehat{S}(\vec{k},\vec{k}^{'})/\lambda(\vec{k})$.

Within the algorithm presented in [82] the before-scattering distribution function is equal to $ \lambda(\vec{k})f_{s}(\vec{k})/\langle\lambda\rangle_{s}$ which gives the distribution $ G^{+}$. In order to find the distribution function of the after-scattering states the before-scattering distribution function should be multiplied by the conditional probability density for an after-scattering state and this product is integrated over all before-scattering states. Using (4.67) and (4.66) one obtains for the after-scattering distribution:

    $\displaystyle \int\biggl\{\frac{\lambda(\vec{k})f_{s}(\vec{k})}{\langle\lambda\...
...frac{[1-f_{s}(\vec{k}^{'})]S(\vec{k},\vec{k}^{'})}{\lambda(\vec{k})}\,d\vec{k}+$  
    $\displaystyle +\frac{\lambda(\vec{k}^{'})f_{s}(\vec{k}^{'})}{\langle\lambda\ran...
...frac{[1-f_{s}(\vec{k}^{'})]S(\vec{k},\vec{k}^{'})}{\lambda(\vec{k})}\,d\vec{k}+$ (4.70)
    $\displaystyle +\int\biggl\{\frac{\lambda(\vec{k})f_{s}(\vec{k})}{\langle\lambda...
...\}
\frac{f_{s}(\vec{k}^{'})S(\vec{k}^{'},\vec{k})}{\lambda(\vec{k})}\,d\vec{k}=$  
    $\displaystyle =\int\biggl\{\frac{\widetilde{\lambda}(\vec{k})f_{s}(\vec{k})}{\l...
...}=
\frac{E_{s}}{E_{im}\langle\widetilde{\lambda}\rangle_{s}}G^{-}(\vec{k}^{'}).$  

Note that the after-scattering distribution is normalized to unity. Now it is obvious that the initial distributions $ G^{+}$ and $ G^{-}$ can be generated by introduction the main trajectory which is constructed using the algorithm from [82] to solve (4.10). Then for each main iteration two carrier ensembles with initial distributions $ G^{+}$ and $ G^{-}$ evolve in time according to (4.11) for the secondary trajectories.

S. Smirnov: