No abstract

We present a theory for spin relaxation of electrons due to scattering off the central-cell potential of impurities in silicon. Taking into account the multivalley nature of the conduction band and the violation of translation symmetry, the spin-flip amplitude is dominated by this short-range impurity scattering after which the electron is transferred to a valley on a different axis in k-space (the so called f -process). These f -processes dominate the spin relaxation at all temperatures, where scattering off the impurity central-cell dominate at low temperatures, and scattering with Σ-axis phonons at elevated temperatures. To the best of our knowledge, the theory is the first to explain and accurately quantify the empirically-found dependence of spin relaxation on the impurity identity. Accordingly, the new formalism fills a longstanding gap in the spin relaxation theory of n-type silicon, and it is valuable for characterization of silicon-based spintronic devices.A major quest in semiconductor spintronics is genuine electrical spin injection from ferromagnetic metals. [1][2][3][4][5][6][7][8] In spite of the fabrication challenge, electrical spin injection has no intrinsic limitation that prevents the development of deep-submicron devices. In order to electrically inject spinpolarized currents, however, one cannot use ohmic contacts between semiconductors and metals due the so-called conductivity mismatch problem. 9-11 Accordingly, electrical techniques are largely limited to spin injection by ballistic hot electrons, 4,12,13 or tunneling across thin barriers. The latter approach can be integrated with the mainstream Si technology but requires narrowing down the Schottky depletion region to a few nm in order to enable measurable tunneling of spin-polarized electrons. 1,3,5,6 Such tunnel barriers can be fabricated by introducing degenerate n-type interfaces which have the detrimental effect of enhanced spin relaxation due to electron-impurity scattering. To date, existing theories neither can quantify nor explain the spin relaxation in heavily doped n-type diamond crystal semiconductors. The lack of understanding hinders development of spintronics devices with tailored spin relaxation, thereby hampering the progress of this research field.A salient feature of spin relaxation in n-type silicon is a strong dependence on the donor atom. 14-18 For example, it has been long recognized from electron paramagnetic resonance (EPR) experiments that the spin lifetime is about 100 times shorter in heavily antimony-doped silicon (Si:Sb) than in phosphorus-doped silicon (Si:P) with comparable impurity concentration. 16,18 This finding contradicts the traditional Elliott picture for spin relaxation, in which the probability for an electron to flip its spin is governed by the spin-orbit coupling of the host material (Si in this example), whereas the identity of the scattering center is of little importance. 19,20 In addition, the predicted proportionality between mobility and spin relaxation time in the Elliott-Yafet mechanism...

We consider the spin-Hall current in a disordered two-dimensional electron gas in the presence of Rashba spin-orbit interaction. We derive a generalized Kubo-Greenwood formula for the spin-Hall conductivity σ z yx and evaluate it in an systematic way using standard diagrammatic techniques for disordered systems. We find that in the diffusive regime both Boltzmann and the weak localization contributions to σ z yx are of the same order and vanish in the zero frequency limit. We show that the uniform spin current is given by the total time derivative of the magnetization from which we can conclude that the spin current vanishes exactly in the stationary limit. This conclusion is valid for arbitrary spin-independent disorder, external electric field strength, and also for interacting electrons.

We show that the disorder-averaged conductivity tensor of a disordered two-dimensional electron gas becomes anisotropic in the presence of both Rashba and Dresselhaus spin-orbit interactions ͑SOIs͒. This anisotropy is a mesoscopic effect and vanishes with vanishing charge dephasing time. Using a diagrammatic approach including zero-, one-, and two-loop diagrams, we show that a consistent calculation needs to go beyond a Boltzmann equation approach. In the absence of charge dephasing and for zero frequency, a finite anisotropy xy ϰ e 2

We present a semiautomated computer-assisted method to generate and calculate diagrams in the disorderaveraging approach to disordered two-dimensional ͑2D͒ conductors with intrinsic spin-orbit interaction ͑SOI͒.As an application, we calculate the effect of the SOI on the charge conductivity for disordered 2D systems and rings in the presence of Rashba and Dresselhaus SOI. In an infinite-size 2D system, anisotropic corrections to the conductivity tensor arise due to phase coherence and the interplay of Rashba and Dresselhaus SOI. The effect is more pronounced in the quasi-one-dimensional case, where the conductivity becomes anisotropic already in the presence of only one type of SOI. The anisotropy further increases if the time-reversal symmetry of the Hamiltonian is broken.