Comment: "Electrical injection and detection of spin accumulation in silicon at 500 K with magnetic metal / silicon dioxide contacts" [Nature Commun. 2:245 doi:10.1038/ncomms125 (2011)]

Abstract

In a recent publication, we demonstrated electrical spin injection and detection in n-type silicon at temperatures up to 500K using ferromagnetic metal / SiO2 tunnel barrier contacts in a three-terminal geometry (Nature Commun. 2:245 doi:10.1038/ncomms125 (2011)). In comparing our measured spin-voltage signal with the value predicted by theory, we followed the analysis of Tran et al, (Phys. Rev. Lett. 102, 036601 (2009)), and inadvertently propagated an error found therein. As they note in a recent erratum (arXiv:0810.4770v2), the correct expression for the spin resistance area product from the theory for a sample with a spin diffusion length LSD much less than the contact width or channel thickness (our experimental situation) is given by the product {gamma}^2 {rho} LSD, where {gamma} is the tunneling spin polarization, and {rho} is the resistivity of the semiconductor transport channel. With this correction, our measured spin voltages are much larger than those predicted by theory, rather than in good agreement as we stated. We emphasize that the basic conclusions of our paper are the same - the systematic decrease in electron spin lifetime with increasing electron density demonstrates spin accumulation in the Si channel rather than in interface states. We further show that the measured spin lifetimes are essentially independent of the tunnel barrier material (SiO2, Al2O3, MgO) or the magnetic metal used (Fe, CoFe, NiFe), demonstrating clear correlation of the measured spin lifetime with the character of the Si, and little correlation with the tunnel barrier / interface or magnetic metal.