Research

Physical Sciences

Title :

Fabrication of oxide heterostructures of hybrid spin valve - resistive switching memory devices for artificial intelligence

Area of research :

Physical Sciences

Principal Investigator :

Dr. Angappane S, Centre For Nano And Soft Matter Sciences (Cens), Karnataka

Timeline Start Year :

2023

Timeline End Year :

2026

Contact info :

Equipments :

Details

Executive Summary :

ReRAMs have gained much attention in the context of the next-generation non-volatile memories (NVMs), owing to their long endurance and excellent scalability. The resistive memory is a two-terminal device with the metal-insulator-metal (MIM) stacked structure, where the insulator is referred to as the switching medium. The resistance of the switching medium or the active layer can be altered by applying the appropriate electrical stimuli. In general, the ReRAMs can be set to a low resistance state (LRS) by applying a suitable bias, and the process is called the SET process. Conversely, the device can be switched back to a high resistance state (HRS) from the LRS (RESET process). The resistive switching mechanism is mainly attributed to the migration of oxygen vacancies or the active metal ions along with the switching layer under the applied electrical bias, leading to the formation and dissolution of conductive filaments. Although ReRAMs are considered more promising in terms of the key device parameters, the stochastic nature of the conductive filaments often results in the wide dispersion of switching parameters. In addition, many of the resistive memory devices require an initial electroforming process, which consumes extremely high power. On the other hand, spintronic devices are advantageous because of their low power consumption and ultra-high switching speed. Spintronic devices have a number of unique and inherent properties that make them promising for use in neuromorphic systems. The rapid dynamics and near-infinite durability of spintronic devices are their most notable features. Spintronics-based neuromorphic systems works based on the magnetic tunnel junction (MTJ), where two ferromagnetic (FM) layers are separated by a thin insulating material. The resistive state of this system depends on the relative alignments of the FM layers. When the magnetizations of the FM layers are parallel, the electron tunneling probability is maximum, which leads to a low resistive state. Whereas, the antiparallel magnetization of the FM layers results in a high resistive state. Spintronic-based neuromorphic devices have less power consumption and faster operation. However, spintronic-based neuromorphic devices lack in small ON/OFF ratio of the magnetoresistance and critical size-dependent problems. Considering the merits of both kinds of memories, a hybrid device working with resistive switching as well as spintronics-based memory technology can have high-speed switching, low power consumption, high ON/OFF ratio, excellent endurance, and scalability. The integration of spintronics and resistive switching in a single device can be ideal for realizing highly efficient neuromorphic systems and is anticipated to revolutionize the next generation NVM and bio-inspired computing technologies.

Total Budget (INR):

16,45,650

Organizations involved