High performance memristive device based on metal oxide oxygen concentration gradient and its fabrication

Abstract

The invention discloses a high-performance memristive device based on a metal oxide oxygen concentration gradient and its preparation, wherein the device unit of the memristor includes an upper electrode, a functional layer and a lower electrode from top to bottom, and the functional layer is a metal Oxygen, the oxygen content in the functional layer changes in a gradient. The present invention improves the internal composition of the key functional layer in the memristor and its preparation method, and uses the metal oxide with a gradient change in oxygen content as the functional layer, so that the on-off of the localized conductive filament depends on the high oxygen content. The area suppresses the random on-off of the conductive channel, which can improve the stability, consistency and switching speed of the memristive device. It can increase the high / low resistance state while reducing the current limit to reduce power consumption. The functional layer based on the oxygen concentration gradient is easy to form a tapered conductive channel, and the cone top of the conductive channel with high oxygen content has the characteristics of easy on-off, which can realize high-speed resistance switching under low-voltage operation.

Description

technical field [0001] The invention belongs to the technical field of microelectronic devices, and more specifically relates to a high-performance memristive device based on a metal oxide oxygen concentration gradient and its preparation. The memristor can utilize oxygen concentration gradients such as TaOx, HfOx, AlOx, and CuOx The dielectric material is a resistance switching functional layer, which realizes the device function of the memristor. Background technique [0002] According to their different performance characteristics, memristors can be divided into continuously adjustable resistance gradual change and resistance sudden change. The former has proven to have great application prospects in simulating neuron synapses, and the latter can be applied to data storage; due to its non-volatile, low power consumption, and high switching characteristics at the nanosecond level, it can not only be used as a download to replace Flash A generation of memory, and has great...

AI Technical Summary

Problems solved by technology

However, due to the difficulty in controlling the on-off of the conductive thread, the on-off of the localized conductive path has great randomness, which leads to discrete distribution of on / off voltage and high and low resistance values, which can be used for integrated crossbar structures due to electrical crosstalk ( The problem of crosstalk) is likely to cause misreading and increased power consumption caused by leakage current. It is usually necessary to design complex peripheral circuits to improve the recognition accuracy, which not only increases the cost but also increases the circuit area;

[0004] In order to improve the stability, the limiting current can be increased to the mA level to obtain a lower resistance state, but at the same time the power consumption is increased, which makes it difficult to realize the commercialization of the device with high power consumption in terms of integration; moreover, according to The energy requirements of memristor switching characteristics, that is, voltage pulse width × voltage amplitude, when the pulse width drops to 100ns or even 10ns, the voltage amplitude will increase significantly, making it difficult to achieve low power consumption of the chip, which is not conducive to the integration of high-speed chip circuits and the fusion of computing and storage

Images