Get started
Work

Self-integrating dual oxide memristor for artificial neuron functionality

Year: 2025
Type: article
Abstract: In this work, we address the challenge of achieving stable volatile switching in metal–oxide devices for neuromorphic applications. Conventional single-oxide-layer structures often transition to a non-volatile state at high currents, limiting their ability to support the dynamic spiking behavior required for artificial neuron applications. To overcome this, we explore a double-oxide-layer device (Cu/Cu2O/2H-TaSe2/Cu2O/Cu) alongside a single-oxide-layer device (Cu/Cu2O/2H-TaSe2/Au) to compare their switching behaviors. Electrical characterization reveals that while both devices exhibit volatile threshold switching governed by conductive filament formation, only the double-oxide-layer device maintains a volatile state at applied current limits. Theoretical modeling suggests that the switching behavior is controlled by voltage-dependent saddle-node and subcritical pitchfork bifurcations, which we experimentally validate through temporal studies. These studies show that as the electric field increases, the transition time from the high-resistance state to the low-resistance state decreases. Furthermore, neuron-like oscillatory switching behavior emerges only in the double-oxide-layer device, observed during voltage ramp and pulse studies, highlighting its capability for self-sustained dynamic switching without the need for an external capacitor. Our findings demonstrate that double-oxide-layer devices effectively overcome the limitations of single-oxide-layer configurations, making them strong candidates for neuromorphic applications requiring stable volatile switching and intrinsic oscillatory behavior.
DOI: 10.1063/5.0253472
Language: English
Open Access status: diamond
OpenAlex ID: W4409049439
FWCI:
Cites: 19
Cited by:
Related to: 10