Temperature Dependence of Spin-Transfer-Induced Switching of Nanomagnets

We measure the temperature, magnetic-field, and current dependence for the switching of nanomagnets by a spin-polarized current. Depending on current bias, switching can occur between either two static magnetic states or a static state and a current-driven precessional mode. In both cases, the switching is thermally activated and governed by the sample temperature, not a higher effective magnetic temperature. The activation barriers for switching between static states depend linearly on current, with a weaker dependence for dynamic to static switching.

Figure 1

(a), (b) Differential resistance of a nanopillar spin valve device as a function of magnetic field and (c), (d) as a function of current, measured at (a), (c) $T=295\mathrm{K}$ and (b), (d) $T=4.2\mathrm{K}$.

Figure 2

(a) Phase diagram of magnetic states for the Py free layer at $T=4.2\mathrm{K}$ obtained from measurements of $dV/dI$ as a function of $I$ at fixed $H$ (■, $I_D^{\pm }$; ▴, $I_S^{\pm }$) and as a function of $H$ at fixed $I$ (○). Stars (★) mark $I$ and $H$ values for which the dwell times in both the HR and LR states are approximately equal to 1 ms, at 16 temperatures ranging from 300 K at small $|I|$ to 4.2 K at large $|I|$. (b) The current dependence of the amplitude of two-level resistance fluctuations, normalized by the full difference in resistance between the P and AP states of the nanopillar. Inset: Two-level fluctuations between the HR and LR states for $I=1.0\mathrm{m}\mathrm{A}$, $H=956.6\mathrm{G}$, $T=4.2\mathrm{K}$.

Figure 3

Temperature dependence of dwell times for the two-level resistance fluctuations measured at fixed $I$ and $H$. (a) $\mathrm{A}\mathrm{P}\to \mathrm{P}$ (open symbols) and $\mathrm{P}\to \mathrm{A}\mathrm{P}$ (solid symbols) dwell times: (□,■) $-0.26\mathrm{m}\mathrm{A}$, 400.8 G; (▵,▴) $-0.22\mathrm{m}\mathrm{A}$, 415.3 G; (◇,◆) $-0.18\mathrm{m}\mathrm{A}$, 430.7 G; (○,●) $-0.14\mathrm{m}\mathrm{A}$, 446.4 G; (▿,▾) $-0.04\mathrm{m}\mathrm{A}$, 482.8 G. (b) $\mathrm{A}\mathrm{P}\to \mathrm{P}$ (open symbols) and $\mathrm{P}\to \mathrm{A}\mathrm{P}$ (solid symbols) dwell times: (□,■) 0.46 mA, 709.2 G; (▵,▴) 0.36 mA, 649.4 G; (○,●) 0.16 mA, 561.6 G. (c) $\mathrm{P}\to D$ (open symbols) and $D\to \mathrm{P}$ transitions (solid symbols) dwell times: (□,■) $-0.72\mathrm{m}\mathrm{A}$, 213.1 G; (▵,▴) $-0.70\mathrm{m}\mathrm{A}$, 217.8 G; (◇,◆) $-0.68\mathrm{m}\mathrm{A}$, 222.6 G; (○,●) $-0.66\mathrm{m}\mathrm{A}$, 226.8 G; (☆,★) $-0.64\mathrm{m}\mathrm{A}$, 230.9 G; (▿,▾) -0.62 mA, 235.7 G. (d) $D\to \mathrm{A}\mathrm{P}$ (open symbols) and $\mathrm{A}\mathrm{P}\to D$ (solid symbols) dwell times: (□,■) 0.98 mA, 953.5 G; (▵,▴) 0.96 mA, 950.7 G; (◇,◆) 0.94 mA, 947.5 G; (○,●) 0.92 mA, 944.7 G; (☆,★) 0.90 mA, 941.3 G; (▿,▾) 0.88 mA, 936.9 G; (◃,◂) 0.84 mA, 928.2 G. The lines are fits using Eqs. (1, 2, 3).

Figure 4

(a) Measured dwell times as a function of $I$ at the field $H_{\mathrm{e}\mathrm{q}}(I,T)$ for which the average dwell times for $\mathrm{L}\mathrm{R}\to \mathrm{H}\mathrm{R}$ and $\mathrm{H}\mathrm{R}\to \mathrm{L}\mathrm{R}$ transitions are equal, for the temperatures □ 300 K, ▵ 280 K, ○ 260 K, ◇ 240 K, ■ 200 K, ● 140 K, and ▴ 80 K. (b) The activation barriers □ ${\epsilon }_{HR}(I)$ for switching $\mathrm{H}\mathrm{R}\to \mathrm{L}\mathrm{R}$ and ● ${\epsilon }_{LR}(I)$ for switching $\mathrm{L}\mathrm{R}\to \mathrm{H}\mathrm{R}$, obtained by using Eq. (4) to collapse data such as in (a) for 16 temperatures between 4.2 and 300 K.