We present a novel multilayered architecture for memristive devices which provides an alternative to conventional conductive filament switching. In conventional resistive switching, conductive filaments form and extend stochastically under applied electrical bias, with longer filaments being subjected to magnified electric fields that amplify their growth rate, producing a spatially localized and highly non-uniform conduction front of filaments. This produces devices with large variations in resistive and capacitive properties that are difficult to tune. Here, we simulate a multilayered device structure with alternating ionic mobility that predicts the development of a quasi-uniform conduction front which amplifies memcapacitive properties of the device and reduces device-to-device variability. Furthermore, this novel structure is predicted to enable fine-tuned control of switching events, an important property for analog (multibit) memory and neuromorphic computing applications.
Read full abstract