@

초록

The growing demand for artificial intelligence (AI) has prompted the development of neuromorphic hardware capable of efficient, parallel, and low-power computation. To meet the requirements for integration and environmental stability in AI systems, neuromorphic transistors based on robust solid-state materials are essential. Here, we report an all-metal-oxide neuromorphic transistor that employs sodium-embedded alumina (SEA) as a solid-state electrolyte and indium-gallium-zinc oxide as the semiconducting channel. A thermal annealing process was used to tailor the chemical composition of SEA, enabling precise control over synaptic plasticity and the deterministic realization of both short-term and long-term plasticity. The long-term devices exhibited stable excitatory/inhibitory postsynaptic responses, long-term potentiation/depression, and paired-pulse facilitation. Furthermore, we demonstrated neuromorphic circuits including a reconfigurable logic gate (AND and OR), an analog comparator, and a multiply-accumulate array that performed analog signal multiplication and summation using programmable synaptic weights. This study highlights the potential of all-solid-state neuromorphic transistors for neuromorphic and analog computing relevant to future AI systems.

키워드