Equivalent electrical circuit modelling of a TaOx/HfOx based RRAM with optimized resistance window and multilevel states

In recent years, bilayer RRAMs based on conductive metal-oxide (CMO)/HfOx have gained ground in the field of analogue in-memory computing. Compared to conventional metal/HfOx based systems, bilayer RRAM structures show reduced switching stochasticity, enhanced symmetry of the set and reset transition, and improved endurance [1]–[3]. Despite their remarkable capabilities to implement neuromorphic systems, especially for the training of neural networks, the resistive switching mechanism, as well as the role of the CMO, is not yet fully understood [4]. In this work we developed and studied TaOx/HfOx RRAM devices, varying the TaOx properties such as the resistivity and thickness. We interpret the switching by means of a dedicated impedance spectroscopy experiment.