Figure 5

Time and temperature stability.

(a) Measured R at room temperature as a function of time for bipolar set and reset states and for NR, PR and set states. Set state and bipolar set state show stable resistance, while NR and PR states show drift as expected from SR in amorphous GST. The bipolar reset state is instead stable over time, confirming that the high R in the bipolar reset state is due to a tunneling barrier rather than an amorphous phase. (b) Measured R after an annealing bake as a function of the annealing temperature for PR, NR and bipolar reset states. R was measured at room temperature after each annealing step. NR and PR states display drift and decay due to crystallization around 150 °C, while the bipolar reset state shows a gradual decrease of R with no evidence for drift or crystallization. Calculation results based on the distributed energy model are also shown.