Spin dependent transport in FeCo$\vert $MgBO$\vert $FeCo magnetic tunnel junctions: Can boron in the oxide region be a good thing?

Recent experimental studies on FeCoB/MgO/FeCoB tunnel junctions have shown that boron can diffuse into the oxide region during rf-sputtering and result in the formation of crystalline MgBO regions[1,2]. These tunnel junctions still provide high tunneling magnetoresistance values as well as very low RA products[3]. Using a plane wave-pseudopotential density functional approach, I have examined potential Mg(B) oxides such as Mg$_{2}$B$_{2}$O$_{5}$ (both monoclinic and triclinic) as well at kotoite (Mg$_{3}$B$_{2}$O$_{6})$. Total energy calculations indicate that these oxides should be more stable than the formation of separate regions of MgO and B$_{2}$O$_{3}$. Kotoite (Mg$_{3}$B$_{2}$O$_{6})$ also has a boron concentration close to that found in the experimentally grown MgBO regions. In addition, kotoite provides a good lattice match with MgO and could act to template neighboring FeCo into crystalline bcc layers during annealing. This evidence suggests that kotoite is formed during the deposition process. I will also discuss the complex band structure of kotoite (Mg$_{3}$B$_{2}$O$_{6})$ and examine how this will also affect spin dependent transport from the FeCo leads. [1] J. Y. Bae \textit{et al.}, J. Appl. Phys. \textbf{99} 08T316 (2006) [2] J. C. Read \textit{et al.}, Appl. Phys. Lett. \textbf{90} 132503 (2007) [3] J. C. Read \textit{personal communication}