Real-time dynamics of VCMA-assisted switching of magnetic tunnel junctions

Abstract

Voltage control of magnetic anisotropy (VCMA) induced by charge accumulation is typically considered as an ultrafast process, enabling energy-efficient and high-speed magnetization switching in spintronic devices. In this work, we investigate the real-time dynamics of VCMA-assisted switching of magnetic tunnel junctions via relaxation in a magnetic field. We show that device-dependent charging effects and magnetic granularity in the free layer limit the switching speed at applied voltages close to the critical switching threshold. Increasing the voltage or the applied magnetic field reduces the incubation delay and total switching time to below a few ns. Micromagnetic simulations incorporating the finite charging times of the tunnel junction and the granularity of the magnetic film reproduce the experimental results, providing critical insights into optimizing VCMA-driven magnetization control for memory and logic applications.

Figures (5)

• Figure 1: (a) Schematic of the MTJ stack and experimental setup used for the dc- and time-resolved measurements. A pulse generator sends a voltage pulse that is probed by a pick-off tee before being fed to the top MTJ electrode. A bias-tee allows for dc-biasing of the junction. (b) Hysteresis loop of an exemplary MTJ demonstrating typical TMR, coercivity and resistance of the investigated devices. (c) Impact of applied voltage $V_\text{dc}$ on the coercivity of the junction due to the VCMA effect.
• Figure 2: (a) Switching probability as a function of $V_\text{p}$ for several $t_\text{p}$ for the P-AP transition at $B_\text{z}= -75.2$ mT. Each data point is the result of 100 switching attempts. The magnetization was reset to the P-state before every attempt. (b) Critical switching voltage as a function of $t_\text{p}$ for both switching directions. The lines show logarithmic fits for $t_\text{p}>$1.5 ns. (c) Raw voltage traces recorded in the oscilloscope for pulses that do not induce switching in the P (blue) and AP (green) state stabilized by $B_\text{z}= \pm 100$ mT. The black line is the difference of these two traces magnified by 20 and the red dashed line is an exponential fit. (d) Equivalent circuit of the MTJ showing the junction resistance, capacitance and passivation layer capacitances.
• Figure 3: Representative time-resolved traces for successful P-AP switching events. Single-shot (colored, thin) and averaged (colored, bold) traces are shown for varying values of $V_\text{p}$ (a-d) and $B_\text{z}$ (e-h). The bold black trace is the difference of the voltage traces measured for fixed AP and P states, representing the full switching amplitude [see Fig. \ref{['fig2:pp']}(c)]. Increasing switching probabilities are indicated from left to right. (i) Statistical distribution of the incubation time for varying $V_\text{p}$ (i) and $B_\text{z}$ (j) obtained by fitting sigmoid functions to the single-shot switching traces.
• Figure 4: Successful switching traces for varying $V_\text{p}$ (a) and $B_\text{z}$ (b) averaged over 500 pulses. The solid lines are sigmoidal fits to the data. Gray traces indicate failed averaged traces. (c) Times at which $50\%$ of the switching is achieved according to the fits shown in (a,b). The data are plotted as a function of $V_\text{p}$ (circles) and $B_\text{z}$ (diamonds). The dashed lines are exponential fits to the data. (d) Slope of the fitted sigmoidal functions at $50\%$ switching as a function of $V_\text{p}$ (circles) and $B_\text{z}$ (diamonds). The dashed lines are fits to the data (see text). The error bars in (c,d) indicate the slopes at $40\%$ and $60\%$ switching.
• Figure 5: (a) Temporal profile of the voltage pulse used in the micromagnetic simuations. (b) Transient behavior of the $z$-component of the magnetization in response to voltage pulses of varying amplitude. The gray shaded areas indicate the time ranges inaccessible to real-time measurements before and after the pulse. (c) Spatial distribution of the magnetization during the reversal process at the times indicated by letters in (b) for slow (top row) and fast switching (bottom row).