Metal oxide-based resistive random access memory (RRAM) has attracted a lot of attention for its scalability, temperature robustness, and potential to achieve machine learning. However, a thick oxide layer results in relatively high program voltage while a thin one causes large leakage current and a small window. Owing to these fundamental limitations, by optimizing the oxide layer itself a novel interface engineering idea is proposed to reduce the programming voltage, increase the uniformity and on/off ratio. According to this idea, a molybdenum disulfide (MoS2 )-palladium nanoparticles hybrid structure is used to engineer the oxide/electrode interface of hafnium oxide (HfOx )-based RRAM. Through its interface engineering, the set voltage can be greatly lowered (from -3.5 to -0.8 V) with better uniformity under a relatively thick HfOx layer (≈15 nm), and a 30 times improvement of the memory window can be obtained. Moreover, due to the atomic thickness of MoS2 film and high transmittance of ITO, the proposed RRAM exhibits high transparency in visible light. As the proposed interface-engineering RRAM exhibits good transparency, low SET voltage, and a large resistive switching window, it has huge potential in data storage in transparent circuits and wearable electronics with relatively low supply voltage.
Read full abstract