Percolation-enhanced Magnetoresistance In High Spin-polarized Oxides

Recently, magnetoresistance(MR) effects have been widely investigated for their fundamental importance and the potential technological application. High spin-polarized oxides with nearly 100% spin-polarization exhibit great magnetoresistance effects in their heterogeneous structures. Among these structures, half-metallic granular systems are outstanding. Many experiments have shown that magnetotransport properties in these granular systems are dependent on the microscopic configurations. Particularly, the magnetoresistance can be greatly enhanced in half-metallic/insulator composites close to the percolation threshold. Study of percolation behavior of magnetotransport network in these systems is not only of theoretical importance but also valuable for the technological application. The purpose of this work is to systemically study the percolation behaviors in these materials and to investigate the methods to enhance magnetoresistance in the spin-polarized conductance network. The main results of our study are listed as follows.We have composed an algorithm to deal with the notorious critical slowing down(CSD) for solving numerically the random resistor network problem. The conventional relaxation methods for solving Kirchhoff's equations of RRN are not effective when the system approaches its percolation threshold: its performance decreases rapidly at the critical region of a metal-insulator-like phase transition -a so-called critical slowing down. The same problem exists in the simulation of the networks with a broad distribution of conductance. When the conductance distribution is broad enough to form an intrinsic percolative path, the fractal geometry leads to a multifractal distribution of voltage drops across the network and in this way reduces the speed of convergence in iterative solutions. In order to deal with the RRN with a broad conductance distribution, we compose a new algorithm based on those applied to the binary distribution RRN. First, we compute the "connected cluster" using modified Hoshen-Kopelman algorithm. Next we extract current-carrying backbone and store the coefficient matrix with row-indexed compact storage strategy. Finally we use the Biconjugate Gradient Method to solve the matrix. In our work, this algorithm is proved to be accurate and efficient for the RRN with a broad conductance distribution and some useful results have been obtained.We have discussed the enhanced magnetoresistance caused by different percolation...