Compositional engineering of HfAlOx interlayers for defect-mediated synaptic plasticity in IGZO memtransistors for neuromorphic computing

Three-terminal memtransistors provide a promising platform for neuromorphic computing by enabling precise gate-controlled modulation of synaptic weights. Here, we demonstrate three-terminal indium–gallium–zinc oxide (IGZO)-based memtransistors incorporating composition-engineered HfAlO x (HAO) interlayers, where defect chemistry is systematically controlled through a co-sputtering approach to achieve tunable resistive switching and synaptic functionalities. By varying the HfO 2 target power while maintaining a fixed Al 2 O 3 target power, the Al/(Hf+Al) cation ratio was continuously tuned from 31.4% to 68.0%, leading to a corresponding modulation of the oxygen vacancy fraction from 24.5% to 10.6%. The composition-controlled devices exhibited bipolar resistive switching characteristics with an on/off current ratio exceeding 10 4 . Furthermore, the IGZO memtransistors demonstrated programmable synaptic behaviors including potentiation/depression and volatile memory characteristics. A multilayer perceptron neural network trained using experimentally extracted potentiation/depression characteristics achieved a handwritten digit recognition accuracy of 89.0%. These results establish composition engineering of oxide alloys as an effective strategy for controlling defect-mediated neuromorphic functionalities and provide a CMOS-compatible pathway toward hardware-based synaptic computing systems.