Quantum correlation of channel-confined ions in graphene-based transistors for energy-efficient neuromorphic chips

The rapid progress of artificial intelligence has exposed the inherent limitations of the conventional chip technology, particularly the high energy-consumption, driving the emergence of neuromorphic chips and ionics. Using K + ion-filled graphene channels, we investigate the mechanism underlying the graphene-based ion transistors by ab initio molecular dynamics simulations. Here we show that graphene electrons enable long-range correlation of confined ions, which provides a basis for the sensitive responses of transistors to the channel ion density (as modulated by a gate voltage). The ON/OFF switching effect specifically results from the competition between π-π stacking and cation-π interaction in the channels with different ion-filling densities. The nonlinear increasing of transport efficiency (i.e., signal amplification effect) is due to the ion density-depended collective oscillation of channel-confined ions. Additionally, resonance between channel-outside and channel-confined ions triggers rapid ion dehydration, enabling the transistor’s ultrahigh ion diffusivity. These atomic-level insights as a design principle for the ultralow energy-consumption neuro-morphic chips.