상세 보기
초록
This work demonstrates a robust and reliable memristive heterostructure device comprising a zinc oxide (ZnO) buffer layer incorporated with a solution-processed hexagonal boron nitride (hBN):polyvinyl alcohol (PVA) nanocomposite. The fabricated device supports memory functionality, exhibiting low-voltage SET and RESET processes with operating voltages below 2 V with a R-OFF/R-ON ratio greater than 200. Furthermore, the device demonstrates a 2-bit multilevel cell (MLC) storage capability and exceptional thermal robustness, maintaining stable performance and retaining the data at 110 degrees C for over 10(4) s. In addition, the fabricated device architecture also exhibits a stable analog switching behavior, emulating biological long-term potentiation (LTP) and long-term depression (LTD) with high cycle-to-cycle uniformity and low device-to-device variability, achieving a functional device yield of 72%. The device nonlinearity is analyzed and fit, and system-level simulations using the MNIST dataset achieve a high classification accuracy of 91.4%, demonstrating the efficacy of trap-mediated weight updates. These results highlight the strong potential of 2-D nanomaterial-polymer heterostructures for scalable, high-density, and thermally stable neuromorphic hardware applications for edge computing.
키워드
- 제목
- Highly Efficient ZnO/hBN:PVA-Based Multibit Memristor Array for Edge Computing
- 저자
- Ranjan, Harsh; Singh, Chandra Prakash; Singh, Vivek Pratap; Pal, Parthasarathi; Pandey, Saurabh Kumar
- 발행일
- 2026-06
- 유형
- Article
- 권
- 73
- 호
- 6
- 페이지
- 3896 ~ 3902