Figure 1

(a) Experimental TMR curve for the device shown here taken with a 5-μA current bias. The coupling field $H_{\mathrm{coup}}$ (=−0.9 mT at $T$ $=$ 303 K) between the CoFeB layers has been subtracted, as it is in all subsequent data. In all following data, we define $H_{\mathrm{Eff}}$ $=$ ($H_{\mathrm{applied}}$ −$H_{\mathrm{coup}}$). (b) A small section of an experimental time trace showing two-state switching for ${\mu }_0{H}_{\mathrm{Eff}}$ $=$ 0 mT. The voltage across the device corresponds to $V^{\mathrm{AP}}$ $=$ 8 mV and $V^{\mathrm{p}}$ $=$ 6 mV. The device is incorporated into a voltage divider circuit, so the difference in the voltages for the P and AP states does not correspond to the TMR value of the device. The full time trace consists of 32 Mpts and captures a roughly 4×10${}^4$ transitions. (measurement schematic) The sampling rate of the scope is set so that a high number of transitions is obtained, while the sampling rate is a minimum of 200 times faster than the relaxation rate, so that short dwell time events are not missed. (c) A histogram of the full time trace of the measured voltage across the device shown in (b). The bimodal structure indicates that the device is well described by simple two-state switching with no evidence of an intermediate state being present. (d) Representative data showing the histogram of the time (${\mu }_0$$H_{\mathrm{Eff}}$ $=$ −0.2 mT) the device remains in the P and AP states. The fits to the data give τ${}^{\pm }$ through Ln($N$) ∼ −$t$/τ${}^{\pm }$ where $N$ is the number of events.Reuse & Permissions

Figure 2

(a) Measured relaxation times for both τ${}^{+}$ [the P-to-AP (■) transition] and τ${}^{-}$ [the AP-to-P (□) transition] as functions of field for $T$ $=$ 283 K and 363 K (other temperatures are not shown for clarity). (b) Ratio of the relaxation times as a function of field for $T$ ranging from 283 K to 363 K along with linear fits. We define ${\mu }_0{H}_{\mathrm{Eff}}$ $=$ 0 mT as the field for which τ${}^{+}$ $=$ τ${}^{-}$. The value of ${\mu }_0{H}_{\mathrm{Coup}}$ changes slightly with $T$ (≈0.002 mT/K), requiring different values of $H_{\mathrm{Coup}}$ to be subtracted from $H_{\mathrm{App}}$ for each temperature. (c) The values of τ${}^{\mathrm{Equal}}$ vs 1/$T$. The slope of this line gives $E_0$ and its intercept Ln(τ${}_0$). (d) $d$$($Ln(τ${}^{+}$/τ${}^{-}$)$)$/($d{\mu }_0$$H_{\mathrm{Eff}}$) vs 1/$T$, the error bars are determined from the fits in (b).Reuse & Permissions

Figure 3

(a) Summary of the measured relaxation times at $T$ $=$ 303 K for both τ${}^{+}$ (the P-to-AP (■) transition) and τ${}^{-}$ (the AP-to-P (□) transition) as a function of $H_{\mathrm{Eff}}$ for several different applied voltages, as indicated in the plot, along with linear fits. (b) $d$$\left(\mathrm{Ln}\right({\tau }^{+}$/ τ${}^{-}\left)\right)/($$d{\mu }_0$$H_{\mathrm{Eff}}$) for the same applied voltages along with linear fits.Reuse & Permissions

Figure 4

(a) Measured slopes of the data in Fig. 3b, $d$$\left(\mathrm{Ln}\right({\tau }^{+}$/ τ${}^{-}\left)\right)/($$d{\mu }_0$$H_{\mathrm{Eff}}$) vs applied voltage along with a linear fit for |$V$| < 0.2 V. (b) The measured value of $H_{\mathrm{Eff}}$-intercept of the data in Fig. 3b along with a quadratic fit. The error bars in determining ${\mu }_0{H}_{\mathrm{Eff}}$(τ${}^{+}$ $=$ τ${}^{-}$) are ± 0.025 mT and are smaller than the data points.Reuse & Permissions