Neuromorphic Silicon-Based Capacitive-Tunneling Junction

The growing demand for artificial intelligence and deep learning technologies has led to a desperate need for energy-efficient neuromorphic computing systems capable of processing large datasets. Here, silicon capacitive tunneling junctions (SCTJs) that leverage the synergistic effects of capacitive coupling and quantum tunneling in Si-compatible devices are presented. These devices demonstrate high-speed switching, low energy consumption, and the ability to emulate neurobiological synaptic behaviors. By using pulse-programmed signal as "stimuli," the SCTJs modulate charge accumulation and dissipation at the Al₂O₃/n-Si interface, simultaneously facilitating rapid electrons/holes transfer through the direct tunneling effect, resulting in high-performance bidirectional and bilingual multimodal postsynaptic behavior of SCTJ, with ultrafast response times of 10 ns and energy consumption as low as 1 fJ. Additionally, the SCTJs are integrated into a system capable of monitoring and recognizing the movement trajectories of objects. These findings offer valuable insights into interface gating mechanisms in capacitive tunneling, which has great significance in constructing the next-generation multifunctional silicon-based neuromorphic computing network.