Low-power reconfigurable MoS <sub>2</sub>/MoTe <sub>2</sub> optoelectronic synapse for visual recognition
Doping with impurity defects or sustained multi-terminal external electric fields can enhance performance of artificial optoelectronic synapses based on two-dimensional materials. But doping causes varying degrees of damage to the original lattice structure, while external fields would increase additional power consumption. Here, we demonstrate an effective surface charge transfer doping approach sensitive to air that facilitates the fabrication of reconfigurable MoS<sub>2</sub>/MoTe<sub>2</sub> devices. MoS<sub>2</sub>/MoTe<sub>2</sub> undergoes electron transfer with surface-adsorbed O<sub>2</sub>/H<sub>2</sub>O, resulting in varying degrees of p-type doping that affects the Schottky barrier and the built-in electric field strength at the PN junction. The doping level can be reconstructed by a brief gate bias resulting in controllable photocurrent. Due to the conduction of the reverse PN junction, the low dark current and high photoelectric response result in an extremely low power consumption per detectable spike (0.73 pJ), and stability is maintained during an 80,000 s reconstruction process. Notably, hardware-level self-noise reduction is achieved through feature-based long/short-term memory, and recognition accuracy on the processed MNIST dataset improved 39%. The unique photo-electro co-modulation strategy paves a promising path for future development of artificial vision systems.
Authors
- Jinsong Wu (ORCID: https://orcid.org/0000-0001-7854-9798)
- Niannian Yu (ORCID: https://orcid.org/0000-0002-2336-569X)
- Wen Luo (ORCID: https://orcid.org/0000-0003-4250-6168)
- Xin Yan (ORCID: https://orcid.org/0000-0002-7719-0792)
- Xiuying Zhang (ORCID: https://orcid.org/0000-0001-6427-9061)
- Wen Deng (ORCID: https://orcid.org/0000-0002-8878-0835)
Publication Details
- Journal
- Nano Research
- Published
- 2025-06-30
- DOI
- https://doi.org/10.26599/nr.2025.94907741
- Citations
- 2
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 1.46
Funders
- National Natural Science Foundation of China