A Robust Artificial Synapse Based on Organic Ferroelectric Polymer

Memristors with history-dependent resistance are considered as artificial synapses and have potential in mimicking the massive parallelism and low-power operation existing in the human brain. However, the state-of-the-art memristors still suffer from excessive write noise, abrupt resistance variation, inherent stochasticity, poor endurance behavior, and costly energy consumption, which impedes massive neural architecture. A robust and low-energy consumption organic three-terminal memristor based on ferroelectric polymer gate insulator is demonstrated here. The conductance of this memristor can be precisely manipulated to vary between more than 1000 intermediate states with the highest OFF/ON ratio of ≈10⁴. The quasicontinuous resistive switching in the MoS₂ channel results from the ferroelectric domain dynamics as confirmed unambiguously by the in situ real-time correlation between dynamic resistive switching and polarization change. Typical synaptic plasticity such as long-term potentiation and depression (LTP/D) and spike-timing dependent plasticity (STDP) are successfully simulated. In addition, the device is expected to experience 1 × 10⁹ synaptic spikes with an ultralow energy consumption for each synaptic operation (less than 1 fJ, compatible with a bio-synaptic event), which highlights its immense potential for the massive neural architecture in bioinspired networks.