Atomic-scale Tuning of Ultrathin Memristors

Memristors, with proposed applications for logic gates, non-volatile memory, and neuromorphic computing, typically consist of bilayer dielectric conducting oxides sandwiched between two metal contacts. When a voltage is applied, oxygen vacancies diffuse through the material and form a conducting filament, causing the memristor to switch from a high resistance state (HRS) to a low resistance state (LRS). Increasing the concentration of oxygen vacancies can improve the performance of memristors in critical areas such as switching speed and on/off ratio (HRS/LRS).

Using density functional theory with hybrid functionals, we investigated if Mg doping in Al₂O₃ can increase the performance of memristors. We found that Mg defects act as deep acceptors that will cause the Fermi level to be located lower in the band gap compared to undoped Al₂O₃. This creates a more resistive HRS, while at the same time lowering the formation energy of oxygen vacancies, thus promoting their formation. These findings were used to guide experiments in tuning the performance of ultrathin memristors by stacking ALD grown Al₂O₃ and MgO layers. Adding MgO layers to the stack increased both the HRS and the on/off ratio, as predicted by our calculations [1].

[1] R. Goul, A. Marshall, S. Seacat, H. Peelaers, F.C. Robles Hernandez, and J.Z. Wu. Atomic-scale tuning of ultrathin memristors. Commun Phys 5, 260 (2022).