Epitaxial Cu₂₋ₓSe Thin Films Exhibiting Memristive and AntiferroelectricSwitching for Neuromorphic Function
ORAL
Abstract
Materials exhibiting resistive switching are of great interest for applications in neuromorphic
computing and energy-efficient electronic devices. Here, we demonstrate that epitaxial Cu₂₋ₓSe
thin films grown on Al₂O₃ substrates by molecular beam epitaxy can stabilize the superionic
phase of copper selenide—normally observed only above ~400 K—at room temperature. By
precisely tuning the growth conditions, we control the stoichiometry and thereby induce either
the superionic or the antiferroelectric phase. Room-temperature current–voltage (I–V)
measurements reveal memristive switching with ON/OFF ratios exceeding five orders of
magnitude and antiferroelectric switching with ratios up to three orders of magnitude. We further
demonstrate how the memristive and antiferroelectric behaviors can serve as building blocks for
artificial neuron architectures. Moreover, the memristive devices exhibit plasticity features,
including short-term memory transitions, emulating synaptic learning rules. These findings
establish Cu₂₋ₓSe as a versatile platform for nanoscale switching that combines low leakage
current, large ON/OFF ratios, and neuromorphic functionality—positioning it as a promising
candidate for bio-inspired and reconfigurable electronic systems.
computing and energy-efficient electronic devices. Here, we demonstrate that epitaxial Cu₂₋ₓSe
thin films grown on Al₂O₃ substrates by molecular beam epitaxy can stabilize the superionic
phase of copper selenide—normally observed only above ~400 K—at room temperature. By
precisely tuning the growth conditions, we control the stoichiometry and thereby induce either
the superionic or the antiferroelectric phase. Room-temperature current–voltage (I–V)
measurements reveal memristive switching with ON/OFF ratios exceeding five orders of
magnitude and antiferroelectric switching with ratios up to three orders of magnitude. We further
demonstrate how the memristive and antiferroelectric behaviors can serve as building blocks for
artificial neuron architectures. Moreover, the memristive devices exhibit plasticity features,
including short-term memory transitions, emulating synaptic learning rules. These findings
establish Cu₂₋ₓSe as a versatile platform for nanoscale switching that combines low leakage
current, large ON/OFF ratios, and neuromorphic functionality—positioning it as a promising
candidate for bio-inspired and reconfigurable electronic systems.
*Supported in part by the MURI program, grant # FA9550-19-454963, and the University of CaliforniaNational Laboratory Fees Research Program (LFRP).
–
Presenters
-
Becker Sharif
- University of California, Santa Cruz