Origin of the Magnetoresistance in Oxide Tunnel Junctions Determined through Electric Polarization Control of the Interface

The observed magnetoresistance (MR) in three-terminal (3T) ferromagnet-nonmagnet (FM-NM) tunnel junctions has historically been assigned to ensemble dephasing (Hanle effect) of a spin accumulation, thus offering a powerful approach for characterizing the spin lifetime of candidate materials for spintronics applications. However, due to crucial discrepancies of the extracted spin parameters with known materials properties, this interpretation has come under intense scrutiny. By employing epitaxial artificial dipoles as the tunnel barrier in oxide heterostructures, the band alignments between the FM and NM channels can be controllably engineered, providing an experimental platform for testing the predictions of the various spin accumulation models. Using this approach, we have been able to definitively rule out spin accumulation as the origin of the 3T MR. Instead, we assign the origin of the magnetoresistance to spin-dependent hopping through defect states in the barrier, a fundamental phenomenon seen across diverse systems. A theoretical framework is developed that can account for the signal amplitude, linewidth, and anisotropy.

Popular Summary

Spintronics is a key technology for next-generation electronics, and taking advantage of functional oxides is a promising avenue for novel applications beyond those that can be achieved using traditional semiconductors and metals. A popular approach, known as three-terminal Hanle, was recently developed to characterize the fundamental properties of functional oxides such as the spin lifetime. However, there has been a major debate in the physics community on whether or not this method reflects a spin accumulation in the semiconductor that is useful for spintronics applications; limited studies have been conducted to investigate if three-terminal Hanle is responsible for the magnetoresistance of spintronics devices. By exploiting functional oxide heterostructures, which provide tunable control over the different spin-transport processes, and in combination with advanced characterization techniques and theory, we unambiguously identify the physical origin of three-terminal Hanle signals.

We use the electric polarization inherent to a class of oxide materials to engineer highly tunable interfaces and test the predictions of three-terminal Hanle models. Because different spin-dependent phenomena, including spin accumulation, rely on different electron tunneling dynamics, we can differentiate among these phenomena by tuning the band alignment between the semiconductor and ferromagnet using an epitaxial electric dipole in the form of ${\mathrm{LaAlO}}_3$ (001) tunnel barriers. Using magnetotransport and inelastic tunneling spectroscopy (resolving the energy-dependent tunneling process), we find that the measured properties reflect those of the barrier rather than the semiconductor. We propose that spin-dependent tunneling through the defect states in the barrier is the origin of the three-terminal Hanle signals, and we develop a theoretical model that correctly accounts for the experimentally observed behavior.

This demonstration of polarization control over spin transport across the interface highlights possibilities for employing functional oxides for new spintronics applications. Our results provide important guiding principles for the development and characterization of futuristic spintronics devices.

Figure 1

Charge and spin transport across $\mathrm{Co}/{\mathrm{LaAlO}}_3/\mathrm{Nb}:{\mathrm{SrTiO}}_3$ heterojunctions with interface dipole layers. (a) Schematic diagram for stacking of 1 u.c. ${\mathrm{LaAlO}}_3(001)$ inserted between Co and $\mathrm{Nb}:{\mathrm{SrTiO}}_3(001)$. (b) Schematic diagram of the measurement configuration. (c) $I_C$ vs $V_C$ for different $t_{\mathrm{LAO}}$. (d) Typical junction dc magnetoresistance ($t_{\mathrm{LAO}}=2\mathrm{u}.\mathrm{c}.$, $I_C=2.8\mathrm{mA}$) as a function of $B^{\perp }$ (black) and $B^{\parallel }$ (blue). Dashed lines indicate the Lorentzian component, while solid lines show the best fit including the background. (e) Bias dependence of the total Lorentzian MR amplitude ($\mathrm{\Delta }{V}_C=\mathrm{\Delta }{V}_C^{\perp }-\mathrm{\Delta }{V}_C^{\parallel }$) for $t_{\mathrm{LAO}}=2\mathrm{u}.\mathrm{c}.$

Figure 2

Tuning charge and spin transport via artificial dipoles. (a) Top panel: Comparison of the junction zero-bias contact resistance-area product ($R_{C}A$) (left axis, red circles) with $\mathrm{\Delta }{V}_{\mathrm{C}}$ (right axis, blue squares) measured at $V_C=0.4\mathrm{V}$ as a function of $t_{\mathrm{LAO}}$. Bottom panel: Linewidths of the Lorentzian MR. (b),(c) Interface energy band diagrams illustrating the decrease of the Schottky barrier height and depletion width by the insertion of an interface dipole in the form of a single unit cell of ${\mathrm{LaAlO}}_3(001)$.

Figure 3

Magnetic-field-modulated inelastic electron tunneling spectroscopy. (a) Magnetic field dependence of $d^{2}i/d{v}^2$ for $t_{\mathrm{LAO}}=2$ and 3 u.c. for $B^{\perp }$ (black circles) and $B^{\parallel }$ (red circles). The bias state for 2 u.c. is $V_C=0.3\mathrm{V}$, $dv=50\mathrm{mV}$, and $V_C=0.4\mathrm{V}$, $dv=50\mathrm{mV}$ for 3 u.c. (b) Top panel: $|\mathrm{\Delta }{d}^{2}i/d{v}^2|$ for the 2 u.c. (black squares) and 3 u.c. (blue triangles) ${\mathrm{LaAlO}}_3$ tunnel barriers. Measured at fixed ac voltage, $dv=50\mathrm{mV}$ and $T=5\mathrm{K}$. Bottom panel: Corresponding linewidths of the total Lorentzian contribution extracted from $d^{2}i/d{v}^2$.

Figure 4

Spin-dependent hopping as the origin for junction magnetoresistance. (a) Schematic diagram for electron hopping through a single defect between a FM and NM contact in the electron-extraction regime. The schematic is drawn for optimally dipole tuned $\mathrm{Co}/{\mathrm{LaAlO}}_3/\mathrm{Nb}:{\mathrm{SrTiO}}_3$ near the flat-band condition. (b) Calculated magnetic field dependence of the hopping current through a localized state. The linewidth is determined by the local fields acting upon the defect site, here plotted for $B_L=50\mathrm{mT}\approx \mathrm{\Delta }B/2$, which reproduces the experimental linewidths. (c) Dependence of the anisotropic components $\mathrm{\Delta }{i}^{\perp ,\parallel }$ on the tunnel coupling rate (${\gamma }_F$) normalized to ${\omega }_L=g{\mu }_B{B}_L/\hslash$. Solid lines represent $B^{\parallel }$ and dashed lines $B^{\perp }$ at different values of the coupling (${\gamma }_N$) between the defect state and the $\mathrm{Nb}:{\mathrm{SrTiO}}_3$. (d) Measured anisotropy $\mathrm{\Delta }{V}_C^{\parallel }/\mathrm{\Delta }{V}_C^{\perp }$ as a function of bias for $t_{\mathrm{LAO}}=2\mathrm{u}.\mathrm{c}.$