A Controllable Method to Integrate Single GaN/Ga 2 O 3 Nanowire into Ultralow‐Energy‐Consumption Synaptic Nano‐Device for Neuromorphic Computing

Abstract Synaptic nano‐devices hold substantial promise in the realms of computation, storage, and learning, hence qualifying as indispensable constituents for building neuromorphic computing systems. In this work, the dielectrophoretic alignment technique is found to be a controllable method to integrate a single GaN/Ga 2 O 3 nanowire into an ultralow‐energy‐consumption synaptic nano‐device. The dielectrophoretic alignment technique is simple to carry out by applying an appropriate alternating current. Thanks to this cost‐efficient method, a single GaN/Ga 2 O 3 nanowire is able to be aligned at the predetermined position on the metal electrodes within 10 min. The primary determinant is proposed to be the relative polarizability of the nanowire and the medium. With stimuli of 255 nm light, this artificial synaptic nano‐device exhibits the plasticity, memory, and learning‐forgetting‐relearning capabilities, matching those found in biological synapses. Notably, the energy consumed by this synaptic nano‐device is only 6.23 × 10 −13 J during a single synaptic event, which is close to that in the human brain. When applied in neuromorphic computing, the device delivered accuracy levels above 90% in digit and image recognition, underscoring its exceptional learning and recognition proficiencies. Therefore, the research may open up new avenues for the efficient and low‐cost development of brain‐inspired chips and artificial intelligence systems requiring low energy consumption.