Dependence of Switching Probability on Operation Conditions in Ge_xSe_1–xOvonic Threshold Switching Selectors

GeSe-Based Ovonic Threshold Switching Volatile True Random Number Generator

“…It has been reported that under a constant voltage bias, the current response varies: the device might be switched on immediately; or switched on after some time (t-on) during the pulse; or remain off till the end of the pulse (Fig. 1c) [24], [27]. 100 repetitive square pulses are applied onto an OTS selector, with amplitude of 2.7V and duration of 50 μs.…”

“…The switching speed of OTS can be shorter than 2 ns [20]. Furthermore, its dependence of switching probability on both voltage and time make it possible to conveniently reduce the pulse width to nanosecond level by increasing the switching voltage, while keeping the bimodal distribution balanced [24]. It is predictable that the throughput can be conveniently boosted to 100 MHz or higher, by increasing the operation voltage for a faster switching speed.…”

“…Particularly, the endurance of GeSe-based OTS device is more than 10 10 cycles by using a simple recovery scheme, while after composition optimization, the OTS based on Se-Ge-As-Te has achieved an excellent endurance of more than 10 11 cycles without recovery operation. Recently, it has been reported that under a constant voltage, the time-to-switch-on (t-on) follow the Weibull distribution [24]. This stochastic nature of OTS switching, being a drawback for memory applications, however can be exploited to implement TRNGs.…”

Stochastic Computing Based on Volatile GeSe Ovonic Threshold Switching Selectors

“…It is well known that the time-dependent dielectric breakdown (TDDB) follows the Weibull distribution and is triggered by the formation of a filamentary conductive percolation path [23] [25]. The Weibull distribution of tdelay in OTS can be attributed to a volatile filamentary formation process, therefore, induced by a different mechanism such as electronic defect de/localization [25][26][27].…”

“…This demonstrates that a pulse with either higher amplitude or longer width can increase the switching probability measured at the end of the pulse. The details of the Weibull plot can be found in [25] [26]. The Weibull parameters, α and β, in Fig.…”