Publication date: 23rd February 2022
The development of non-Von Neumann computing depends on the realization of new devices that can be integrated into traditional circuitry. Memristive devices, which are essential components in unconventional computing schemes, can operate on resistive switching memories that display volatile or non-volatile change of resistance states when an electrical impulse is applied across the electrode terminals. In a metal/insulator/metal (MIM) stacking, a non-volatile change of resistance states can be realized based on the electronic or ionic conduction of the insulating layer. In contrast, volatile switching can be performed based on the temperature or electric field-driven insulator to metal (IMT) transition of the switching layer. Such dynamics of the resistance change base their operation on the formation of conducting filament (CF), which comes with a drawback of a high voltage forming step, a requirement of high temperature, and cycle-to-cycle variation of the resistance states.
In this context, We have studied phase transition in a manganite, La0.67Sr0.33MnO3 (LSMO), which has a rich phase diagram brought about by doping and grown on different epitaxial substrates. For this work, we have chosen LaAlO3 substrate, deposit strained thin films of LSMO using Pulsed Laser Deposition and tailored the strain to stabilize competing magnetic states in the same structure. Desirable metallic or insulating forms can be attained at the boundaries of two coexisting phases. The strong correlation between electronic and magnetic states enables a change in resistance by an electric current close to the competing phase transition temperature. Our results show Joule heating induced hysteric resistive switching and a multilevel resistive network based on LSMO/LAO heterostructures. We found that these distinct resistive states are highly stable over time and pose minimal cycle-to-cycle variability. The type of switching is exploited to show short-term memory (STM) behavior which is discussed in the context of a brain-inspired computing approach based on strongly coupled phase transition-based memristive devices.