Memristor device modeling and circuit design for read out integrated circuits, memory architectures, and neuromorphic systems

Significant interest has been placed on developing systems based on the memristor, which was physically recognized in 2008. The memristor is a nanoscale non-volatile device with a large varying resistance range. Voltage pulses can be applied to the memristor to change its resistance, and the last programed resistance remains until another voltage pulse is applied. The unique properties present in this device give it the potential to further the advancement of many electronic systems, such as read out integrated circuits (ROICs) for digital cameras, high-speed on-chip memory circuits, and neuromorphic circuits capable of parallel analog computation. This work first describes the different memristor modeling techniques that have been proposed, followed by a new memristor model that this capable of reproducing the I-V curves and switching characteristics of many physical memristor characterizations very accurately. Circuits are then designed using this model for the three applications previously mentioned (ROICs, memory arrays, and neuromorphic computation). It is demonstrated that a memristor based ROIC circuit can significantly reduce the area of a unit cell given that a very small memristor element is used to replace a large integrating capacitor. It is also shown that the memristor can be used to reduce the total area of on-chip microprocessor memory. Given the nonvolatile property of memristors, they can also be used to store information without consuming power to hold their memory state. Lastly, memristors can be used to implement neuromorphic circuits where parallel computations are performed in the analog domain. Just as chemical pulses alter synaptic weights in brain tissue, voltage pulses can be applied to memristors to alter their conductivity. This work is completed in SPICE, which handles many low level circuit details using a very detailed memristor model. Novel circuit designs for three different applications are not only presented in this work, they are also simulated with a level of accuracy not accounted for in the current literature