Trade-off between data retention and reset in NiO RRAMS

NiO-based resistive-switching memory (RRAM) is a promising new technology for high-density non-volatile storage. The main obstacles to practical application in nonvolatile memories are the variability of program/erase voltages, the large and hardly scalable programming current and the cell reliability. We have investigated data retention in RRAM samples with NiO as active switching material. Temperature-accelerated bake experiments show that data retention limited by oxidation of the conductive filament (CF) obey an Arrhenius law, while the retention time decreases for decreasing size of the CF. The results are interpreted by a physical model for CF dissolution in an oxidizing environment, which can be applied to both data retention extrapolation at long times and low temperatures, and the reset operation in the ns/ms regime. The model is verified with experimental data, and the tradeoff between data retention and reset current is finally discussed.