2.3  Memristive Device Modeling

Beside the extensive efforts for physical implementation of resistance switching and memristive devices, significant progress has also been made regarding the modeling [129, 130, 131, 132, 133, 87, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143] as well as better understanding of the working principles and improving the performance [144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158] of the memristive devices. However, most of the presented memristor models rely on a linear ionic drift model for TiO memristive devices suggested in [69] which is not adequately accurate, especially in high voltage switching regimes, and can be used only for limited applications as will be shown in Chapter 3. In [138] a more detailed but quite complicated and computationally expensive physical model based on the Simmons tunneling barriers [159] is presented. It takes into account the asymmetric switching behavior as well as the nonlinearities observed in TiO memristive devices [147, 148, 149]. To my best knowledge, this nonlinear ionic drift model [138] (its SPICE implementation is presented in [139]) is up to now the most accurate model for the TiO memristive devices. More computationally effective and simpler models including nonlinearities of memristive devices as well as additional physical operating mechanisms for different types of memristive devices, have been presented in [140, 141, 142, 143] based on voltage/current thresholds.