Compact modelling of pulsed reset transtions in TiN/Ti/HfO2/W memristors programmed at different low resistance states
Please login to view abstract download link
This work presents a study on the compact modeling of TiN/Ti/HfO2/W memristors, focused on the characterization of reset dynamics across a wide range of operational time scales. A critical challenge in resistive switching (RS) technology is the voltage-time dilemma [1-2], where devices must maintain non-volatile stability at low read voltages while allowing fast programming at slightly higher voltage levels. This is feasible thanks to the highly non-linear dependence of the reset mechanisms on the applied voltage. We characterize and model here the Time-to-Reset (TtR) parameter (defined as the time required for the device current to drop to half of its initial value upon the application of a constant voltage pulse, V) and the transient response (I-t) for circuit simulations. To address this, we propose two approaches: i) an analytical model based on the integration of the Stanford model dynamic equation [3], including the effects of a series ohmic resistor and a thermal resistance; ii) an enhanced SPICE simulation model, which accurately captures the shape of the reset transitions experimentally recorded, by implementing a thermal resistance that evolves according to the filament state. Fig. 1 shows the results from the experimental characterization. Fig. 2a shows a comparison of the TtR-V plots (for different initial low resistance states) obtained experimentally and with the analytical approach. The developed model incorporates essential physical effects, including Joule heating and the voltage drop across the device series resistance. Different values of the thermal resistance have been used. If TtR is plotted versus the corresponding estimated temperatures at the reset point by using these values of the thermal resistance, the curves merge into a single plot, pointing out the presence of the same reset mechanism [2]. Finally, Fig. 3 shows an example of a reset transient obtained with the developed simulation model. [1] P. Huang et al., IEEE Transactions on nanotechnology, vol. 13, no. 6, pp. 1127-1132, 2014. [2] R. Dittmann, S. Menzel and R. Waser, Advances in Physics, vol. 70, no. 2, pp. 155-349, 2022. [3] X. Guan et al., IEEE Electron device letters, 33, 1405-1407, 2012.