Electrically Switchable and Tunable Rashba-Type Spin Splitting in Covalent Perovskite Oxides

In transition-metal perovskites ($AB{\mathrm{O}}_3$) most physical properties are tunable by structural parameters such as the rotation of the $B{\mathrm{O}}_6$ octahedra. Examples include the Néel temperature of orthoferrites, the conductivity of mixed-valence manganites, or the band gap of rare-earth scandates. Since oxides often hold large internal electric dipoles and can accommodate heavy elements, they also emerge as prime candidates to display Rashba spin-orbit coupling, through which charge and spin currents may be efficiently interconverted. However, despite a few experimental reports in ${\mathrm{SrTiO}}_3$-based interface systems, the Rashba interaction has been little studied in these materials, and its interplay with structural distortions remains unknown. In this Letter, we identify a bismuth-based perovskite with a large, electrically switchable Rashba interaction whose amplitude can be controlled by both the ferroelectric polarization and the breathing mode of oxygen octahedra. This particular structural parameter arises from the strongly covalent nature of the Bi-O bonds, reminiscent of the situation in perovskite nickelates. Our results not only provide novel strategies to craft agile spin-charge converters but also highlight the relevance of covalence as a powerful handle to design emerging properties in complex oxides.

Figure 1

Energy differences in meV per formula units (meV/f.u) between the different metastable phases as a function of the substrate lattice parameter (in Å, lower scale) or strain value (in %, upper scale). The reference energy is set to the $P-1$ structure energy (dashed horizontal black line). The vertical line represents the boundary between the ferroelectric (FE) and paraelectric (PE) phases.

Figure 2

Electronic properties of the ferroelectric ground state of ${\mathrm{SrBiO}}_3$ at 6.1% of compressive strain. (a) Orbital projected [on O $p$ (blue character) and Bi $s$ (red character) states] band structure in a monoclinic $P{2}_1/n$ setting. The Fermi level is set to 0 eV. (b) and (c) Bands projected on the $\pm {\sigma }_x$ component along the $(+1/2,0,0)$ to $(+1/2,0,1)$ direction for $+P$ (b) or $-P$ (c) spontaneous polarization. (d) and (e) Spin textures in the $(k_z,k_x)$ plan obtained at $E=+0.75\mathrm{eV}$ when projecting on the $\pm {\sigma }_x$ (d) or $\pm {\sigma }_z$ (e) spin component for a positive spontaneous polarization.

Figure 3

Evolution of the Rashba coefficient when freezing different amplitudes of the polar distortion (a) and the breathing of the oxygen cage octahedra (b). (c) and (d) Bands projected on the $\pm {\sigma }_x$ component along the $(+1/2,0,0)$ to $(+1/2,0,+1)$ direction for different amplitudes of the breathing distortion. We emphasize that all other lattice distortions are fixed to their amplitude appearing in the ferroelectric ground state.