600 K temperature robustness and ten resistance levels memristor based on aluminum nitride films
The deployment of artificial intelligence hardware in extreme environments presents substantial technological challenges, particularly regarding multi-level data storage and operational stability. This work focuses on highly crystalline aluminum nitride (AlN) as memristive material with a large bandgap that is reported to exhibit outstanding thermal stability. Utilizing plasma-enhanced atomic layer deposition (PEALD), we have produced AlN films with low impurity concentrations through controlled plasma exposure duration, which permits the fabrication on Al/AlN/W memristor devices. The resulting memristor devices exhibit a pronounced dependence of electrical characteristics on film quality, demonstrating remarkable performance metrics including: (1) an on/off ratio surpassing 10 7 , (2) programmable 10-level resistance modulation, and (3) robust switching stability across an extensive temperature spectrum (@298-600 K) and retention time of >10 3 s (@600K). These advancements establish a viable framework for engineering high-performance, thermally robust memristors while contributing critical technical foundations for AI implementation under harsh operational conditions.
Authors
- Chenjia Xu
- Ying Yang (ORCID: https://orcid.org/0000-0001-8178-7391)
- Ping Zhu (ORCID: https://orcid.org/0000-0002-5773-8861)
- Xiaoqin Yang (ORCID: https://orcid.org/0000-0002-9539-2785)
- Hong Gu
- Jin Cai
- Lili Ma (ORCID: https://orcid.org/0000-0002-9774-8220)
- Zengli Huang (ORCID: https://orcid.org/0000-0002-1201-3509)
- Peng Zhu (ORCID: https://orcid.org/0000-0001-7345-2073)
Institutions
- Suzhou Research Institute (CN)
Publication Details
- Journal
- Chip
- Published
- 2025-11-06
- DOI
- https://doi.org/10.1016/j.chip.2025.100173
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00