Study Of Electric And Magnetic Transport In Micro-nano Systems

In this thesis we will investigate electric and magnetic transport properties of samples at micro-nano scale both by establishing network models and employing ab initio approaches based on non-equilibrium Green's function (NEGF) techniques respectively. The purpose of this work is to reveal: how the smaller size influences transport properties, how to design particular patterns to enhance magnetoresistance and how to control the spin filter effects (SFE) through the length of molecule and contacts between the ending groups and electrodes in 1-dimensional molecular devices. The main results of our study are listed as follows.We have proposed the random fuse network model (RFN) to investigate the electric transport properties in manganites, and well explained the steplike negative differential resistance (NDR) observed in spatial confined manganite structures. We assume that a fuse with thermal power above a threshold Pth would be broken immediately and irreversibly by self-heating, which corresponds to the turn-off of weak links between ferromagnetic domains. When the conductance distribution is broad enough to form an intrinsic percolation 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 traditional Kirchhoff iterative solutions. This is the notorious critical slowing down(CSD) for solving numerically the random resistor network (RRN) problem. 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. By adopting this effective algorithm, we investigated the detailed evolution of the breaking process in a random fuse network by calculating a sequence of external voltages against the number of broken bonds. We found the occurrence of catastrophic crack growth is observed in weak disordered RFN, while a gradual breaking process will be obtained in the strong disordered network. Secondly, NDR multisteps are reproduced in a random fuse network of small size and strong disorder and the corresponding evolution of current morphology is present. Finally, we found the steplike NDR responses are suppressed gradually with increasing network size. Furthermore, we found the abrupt steps of I-V curves exhibited in simulations on small-size networks gradually melt into a nonlinear smooth curve after L /κν< 1, where L ,κ,νare the network size, the disorder strength and the critical exponent of the percolation correlation length. All these suggest that the NDR is a universal phenomenon in the strong disordered micro-nano systems.We have studied the effects of size and current path structures on magnetotransport in micro-nano samples by performing simulations on clusters of finite size. We proposed magnetotransport networks based on tunneling mechanisms on fractal structures including the Koch curves and the percolation backbones. 2D and 3D backbone structures are extracted from regular lattices by the direct electrifying algorithm and the Koch curves are yielded by traditional iterative algorithms. Our simulations show that the finite-size effect is nontrivial even for magnetotransport network of larger size. The traditional quasi-one-dimensional simplification of current paths for backbone transport is not sufficient, and the details of the cluster must be considered in the study of magnetoresistance. The role of the cluster structures is mainly exhibited in the different contributions of the magnetic moments on the multi-connected sites in bulbs. For systems of larger size, this effect may be covered after averaging. But for small systems, such as magnetotransport materials at micro-nano scales, the MR ratio shows an obvious structure-dependence of the MR fluctuations due to magnetic disorder. We have compared the MR behaviors of the Koch curves and the corresponding string-parts from the first to the seventh generation. We found the effect of loops in the Koch curves structures is twofold. For generation 4-7, the peaks of the probability distributions of magnetoresistance are broadened and pushed to the left slightly after the loops are added to the cluster. However, for the lower generations, where the number of bonds is less than 50, the long tails of the string-MR distributions extending into the lower-MR regime are suppressed by the loops, resulting in a narrowed peak in the total-cluster results. This work has provided an insight into the important role of size and structure patterns on magnetotransport of mesoscopic samples. Better MR behaviors can be obtained by designing the structure patterns of clusters"properly".We have investigated the electronic spin transport through (CpFeCpV)_n multidecker wire sandwiched between magnetic Nickel (Ni) electrodes in the linear response regime by the nonequilibrium Green's function (NEGF) technique. Because the molecular device is an open system so than the traditional density functional theory (DFT) cannot deal with the transport properties in molecular devices. The software of the nonequilibrium Green's function (NEGF) technique is available nowadays and the validity has been argued by large numbers of work. We constructed a two-probe system of a finite (CpFeCpV)_n multidecker wire connected to two magnetic fcc Nickel (Ni) electrodes on their (001) surfaces. And then the spin transport simulation is carried out with the help of the Atomistix Toolkit after a structure optimization for the system is performed by the Vienna ab initio simulation package (VASP) to estimate the structure parameters. Firstly, we showed the spin filter efficiency, the total linear conductance and the transmission spectra for molecular wires with different combination of anchoring groups at their two ends. We found higher SFE in cases with metal anchoring groups. The different anchoring groups can change the sign of the spin filter efficiency. Increasing molecule length results in decreasing spin filter efficiency and conductance in most of the studied cases where the electronic states at the Fermi energy is usually localized. Secondly, we simulate the spin transport through a Ni-CpFeCpVCpFe-Ni structure when the anchoring Fe atom is absorbed on the hollow or the top position. Through the analysis of the transmission spectra, SFE and the molecular projected self-consistent Hamiltonian (MPSH), we found the top absorbing induced the higher SFE. The decoupling between the molecule and the electrode breaks the transport channel for spin-up electrons and results in higher SFE when the anchoring Fe atom is absorbed on the top position. In summary, our result shows that the contact effect can manipulate the performance of spin filters made of this molecule, including the amplitude and the sign of the spin filter efficiency and the total conductance. For the same type of contact, increasing molecule length results in decreasing spin filter efficiency and conductance in most of the studied cases where the electronic states at the Fermi energy is usually localized. We show that the localization and matching of electronic states plays a critical role in the contact effect as well as the molecule length dependence.