On a quasilinear degenerate system arising in semiconductor theory. Part II: Localization of vacuum solutions (Q4239779)

The paper under review is concerned with the following system of PDEs in a cylinder \(Q_{T}= \Omega\times(0,T)\) \((\Omega \subset \mathbb{R} ^{d}\) a bounded domain \((1 \leq d \leq 3))\): \[ \begin{alignedat}{2} (1)&& \qquad\frac{\partial n}{\partial t}- \nabla \cdot \big( \nabla r (n) - n \nabla V \big) &= - R (n,p),\\ (2)&& \frac{\partial p}{\partial t} - \nabla \cdot \big( \nabla r (p) + p \nabla V \big) &= - R (n,p),\\(3)&& \Delta V &= n - p - C(x),\end{alignedat} \] where \(n=\) electron density, \(p=\) hole density, \(r= r(\xi)=\xi^{\alpha}\) \((\xi \in [ 0, + \infty)\), \(\alpha > 1)\) pressure function, \(R(n,p)=\) recombination-generation rate, \(V=\) electric potential and \(C=\) doping profile. Equs. (1)--(3) are supplemented by physically motivated boundary conditions on \(n, p\) and \(V\), and initial conditions on \(n,p.\) The authors study the evolution in time of the vacuum sets \(\{ x \in \Omega\mid n (x,0)=0\}\) and \(\{x \in \Omega \mid p(x,0)=0 \}\). For this, estimates from below are imposed upon \(R\). Then the main results are as follows: Let \(x_{0}\in \Omega, 0 < \rho_{0} < \text{dist} (x_{0}, \partial \Omega)\), and assume that \(n(x,0)=p(x,0)=0\) for all \(x \in B_{\rho_{0}}(x_{0})\). Then there exist \(T_{1}> 0\) and a non-increasing function \(\rho : [ 0, T_{1}) \rightarrow (0, + \infty)\) such that \(\rho (0)= \rho_{0}\) and \[ n(x,t)= p(x,t)=0\quad\text{ for a.e. } x \in B_{\rho(t)}(x_{0}), t \in (0,T_{1}) . \] Under an appropriate ``flatness condition'' upon \( n(\cdot ,0)\) it is proved that there exists \(T_{2}\in (0,T)\) such that \(n(x,t)=0\) for all \(\{ x,t\} \in B_{\rho_{0}}(x_{0})\times (0, T_{2})\). Making use of the energy \(D_{n} (Q_{T})\) of \(n\), the next result is that there exist \(M > 0\), \(T_{3}\in (0,T)\) and \(\gamma , \kappa \in (0,1)\) such that if \(D_{n}(Q_{T})\leq M\) then \[ n(x,t)= 0\quad\text{ for a.e. } x \in B_{\rho(t)}(x_{0}), t \in (T_{3}, T), \] where \(\rho (t)=\gamma(t-T_{3})^{\kappa}\). The proof of these results is based on a local energy method for free boundary problems, originally introduced by S. Antontsev.