Magnetic tunnel junctions with superlattice barriers

Abstract

The urgent demand for high-performance emerging memory, propelled by artificial intelligence in internet of things (AIoT) and machine learning advancements, spotlights spin-transfer torque magnetic random-access memory as a prime candidate for practical application. However, magnetic tunnel junctions (MTJs) with a single-crystalline MgO barrier, which are central to magnetic random-access memory (MRAM), suffer from significant drawbacks: insufficient endurance due to breakdown and high writing power requirements. A superlattice barrier-based MTJ (SL-MTJ) is proposed to overcome the limitation. We first fabricated the MTJ using an SL barrier while examining the magnetoresistance and resistance-area product. Lower writing power can be achieved in SL-MTJs compared to MgO-MTJs. Our study may provide a new route to the development of MRAM technologies.