Memristor augmented ReRAM cell for cross-bar memory architecture

Abstract

Memristor (the so called Resistive Random Access Memory (Re-RAM), is an emerging next generation non-volatile memory, which shows promise towards achieving faster operation speed and also various advantages such as non-volatility, low power consumption, most importantly lesser density and latency. It can store information and can also switch between different states. It is a two terminal device. This type of memories would not lose its data even when the power is switched off. Recently Memristor's applications lie even in complex and interesting areas like Artificial Intelligence. Memristor's can be used to model human brain since its properties is more similar to synapses. Therefore with the help of synapse as Memristor and neurons as a CMOS control circuit, the entire brain can be modeled and fabricated on a single chip. Memristor can replace the power consuming transistors which can be productive in creating a logic circuit. This allows flexibility in using a circuit both for storage purpose and logical operations simultaneously. Memories are usually designed based on the crossbar architecture, where a single switching cell (1Memristor in our case) is placed at the cross-points of word line and bit line. The main finding of this work is that, when this model is applied to a crossbar structure, there is no resistance change except the desired one because of its voltage control nature in comparison to other models. As a result we are avoiding the undesired current the so called sneak current along with reducing the circuit elements i.e. the technique involved in a complete arrest of sneak path current like complementary resistive switching or connecting a diode for each cell as in other models (Linear ion Drift model, TEAM model) for memory operation. The reduction in circuit elements has helped in enhancing the density, making the design less complex and reduction in die area.