Fast Low-Current Spin-Orbit-Torque Switching of Magnetic Tunnel Junctions through Atomic Modifications of the Free-Layer Interfaces

Future applications of spin-orbit torque will require new mechanisms to improve the efficiency of switching nanoscale magnetic tunnel junctions (MTJs), while also controlling the magnetic dynamics to achieve fast nanosecond-scale performance with low-write-error rates. Here, we demonstrate a strategy to simultaneously enhance the interfacial magnetic anisotropy energy and suppress interfacial spin-memory loss by introducing subatomic and monatomic layers of Hf at the top and bottom interfaces of the ferromagnetic free layer of an in-plane magnetized three-terminal MTJ device. When combined with a $\beta \text{−}\mathrm{W}$ spin Hall channel that generates spin-orbit torque, the cumulative effect is a switching current density of $5.4\times {10}^6\mathrm{A}/{\mathrm{cm}}^2$.

Figure 1

Hf-spacer–Hf-dusting sample annealed at $240°\mathrm{C}$. (a) The Hf-spacer–Hf-dusting sample structure and measurement schematics. Inset: SEM image shows a representative nanopillar situated on a W channel after $e$-beam exposure, development, and ion-beam etching. (b) Current-induced switching loop of the MTJ free layer showing a thermally assisted switching current of $50\mu \mathrm{A}$. The device is $190\times 30{\mathrm{nm}}^2$ and is situated on a 480-nm-wide W channel. Inset: In-plane field-switching minor loop of the free layer. (c) Current ramp rate measurement on the device of (b). Fitting to the macrospin model gives a zero-thermal-fluctuation critical current of $115\mu \mathrm{A}$ with a thermal stability factor of 35.6. (d) The free layer demagnetization field change with annealing temperature for a Hf-spacer–Hf-dusting sample compared to that of a Hf-dusting-only sample and a sample without Hf insertion as measured by flip-chip FMR. $M_{\mathrm{eff}}$ significantly decreases in the samples with Hf dusting due to enhanced interfacial perpendicular anisotropy. (e) Linewidths at different resonance frequencies (applied fields) for the Hf-dusting-only sample and the Hf-spacer–Hf-dusting sample measured by flip-chip FMR. Both samples are annealed at $240°\mathrm{C}$. The damping decreases significantly with the insertion of the 0.25-nm Hf spacer.

Figure 2

Fast and reliable pulse switching of a Hf-spacer–Hf-dusting sample. Pulse-switching phase diagrams and macrospin fits for polarities $\mathrm{P}\to \mathrm{AP}$ (a) and $\mathrm{AP}\to \mathrm{P}$ (b), respectively, with the switching probability scale bar on the right. Each point is a result of ${10}^3$ switching attempts. A characteristic switching time of approximately 1 ns and a critical voltage of 0.46 V are obtained after fitting 50% probability points (green points) to the macrospin model. (c) WER measurement results for 2-ns square pulses applied to the device (a) and (b). Each point is a result of ${10}^6$ switching attempts. WER of approximately ${10}^{-6}$ is obtained at sufficiently high-voltage (current) amplitudes for both polarities.

Figure 3

Temperature dependence of the Hf-dusting effect. (a) Flip-chip FMR measurement on two Hf-dusting-only samples annealed at $240°\mathrm{C}$ (magenta) and $300°\mathrm{C}$ (blue), respectively, showing a further reduction of $M_{\mathrm{eff}}$ at higher annealing temperature. (b) Current-induced switching of Hf-dusting-only samples annealed at two different temperatures, $240°\mathrm{C}$ (magenta) and $300°\mathrm{C}$ (blue). The spin-torque switching loops indicate a substantial reduction in critical current with the higher-temperature anneal as quantified by the results of ramp rate measurements of $I_{c0}$.