Building Brain-Inspired Logic Circuits from Dynamically Switchable Transition-Metal Oxides

Metal-Insulator Transitions in β′-Cu V2O5 Mediated by Polaron Oscillation and Cation Shuttling

“…8 Much attention has focused on the search for materials exhibiting large conductance switching but with relatively small lattice distortions and greater tunability of the transformation temperature. 9 Here, we identify a coupled polaron oscillation and cation-shuttling mechanism as the origin of thermally triggered and voltage/current-driven metal-insulator transitions in a family of quasi-one-dimensional (1D) tunnel-structured materials, b 0 -Cu x V 2 O 5 , where x represents the variable stoichiometry of Cu ions (0.27 < x < 0.67). Such a mechanism points to an entirely distinctive approach to the design of materials exhibiting programmable electronic instabilities as required for neuromorphic electronics.…”

“…10,[21][22][23] In contrast to the canonical MIT material, VO 2 , in which both Mott and Peierls mechanisms have been implicated as the underlying origin of the MIT (but considerable new insight has emerged from recent studies), [24][25][26] little is known about the precise mechanistic origins of the observed electronic instabilities in ternary vanadium oxides. 9,14 Cu concomitantly donates a valence electron to the V 2 O 5 lattice upon insertion into the V 2 O 5 framework. The localization of the electron on a specific vanadium site within the V 2 O 5 framework induces a pronounced local structural distortion as extensively documented for Li-ion insertion in a-V 2 O 5 .…”

The non-capacitor model of leaky integrate-and-fire VO2 neuron with the thermal mechanism of the membrane potential

“…One of the promising materials for neuromorphic devices is vanadium dioxide. Having a metal insulator transition and the effect of electrical switching, this material is a good alternative for the creation of neuromorphic, logical, and oscillatory devices [9][10][11].…”

Perovskite neural trees

“…For AI to be sustainable, it is imperative to control the memory and thereby the intelligence and learning ability of autonomous machines. Creating new types of memory, including synaptic properties, is therefore a central goal in physical sciences and engineering research intersecting materials sciences, electronics, neuromorphic computers, and AI hardware [1][2][3][4][5][6][7][8][9][10][11] . In animal brains, evolution has enabled synapses (that are responsible for memory) to reach selflimiting weights as well as branching to avoid runaway effects and prevent catastrophic breakdown of neural circuits while still retaining the ability to learn throughout their lifespan.…”