Thesis presented April 22, 2010

Abstract:
Spintronics has allowed many major breakthroughs but the remaining issue of spin injection prevents its wider spread to microelectronics. Diluted Magnetic Semiconductors (DMS) have been suggested as a possible workaround, but it appears difficult to apply to IV-type Semiconductors, because of the low solubility of transition metals, that tend to clusterize. In this work, we concentrate on the germanium manganese system, and on the atomic structure of precipitates which seem coherent with the Ge matrix, and which magnetic, chemical and structural properties don't match those of any known compound. We have used in this work ab initio simulation methods in the DFT approach, along with pseudopotential PAW formalism. First, we generate needed pseudopotentials for Ge and Mn. Then we show using a thermodynamic study that first principle methods are able to reproduce the (x,T=0) phase diagram of the Ge_xMn_1-x system. Next we study simple defects of Mn in the Ge matrix, along with their possible clusterisation, and show that a defect concentration on the diamond lattice cannot account for experimental observations. Finally, we show that some ordered compounds derived from neighbor systems are metastable in GeMn, and compatible with experimental observations. Thanks to a complete study of interfaces with diamond Ge, we can explain the stability of Mn-rich (33%) nanocolumns embedded in a pure Ge matrix. Besides, we compare the simulated properties of the proposed compounds with experimental measurements for: Curie temperature, XAS spectra, but also X-ray diffraction and transmission electron microscopy (TEM).

Keywords:
Alloy, Metastability, Spintronics, ab initio, manganese, germanium, simulation

On-line thesis.