High‑Temperature Image Pre‐Processing Based on ε‐Ga2O3 Photo‐Synapses

• Published in: Advanced Physics Research Vol. 4; no. 4
• Main Authors: Lin, Yingtong, Zhu, Yuhan, Wang, Feng, Cai, Yuchen, Li, Zhengqiao, Li, Shuhui, Zhan, Xueying, He, Jun, Jiang, Chao, Wang, Zhenxing
• Format: Journal Article
• Language: English
• Published: Edinburgh John Wiley & Sons, Inc 01.04.2025; Wiley-VCH
• Subjects: Big Data; Chemical vapor deposition; Extreme environments; Gallium oxides; High temperature; Image compression; Information storage; Microscopy; Neuromorphic computing; photo synapse; Power consumption; preprocessors; Room temperature; Sensors; Surface roughness; Synapses; Temperature; Thin films; Topography; ε‐Ga2O3
• ISSN: 2751-1200; 2751-1200
• DOI: 10.1002/apxr.202400115

The era of big data has brought about substantial challenges in terms of power consumption and communication bandwidth. Edge‐based neuromorphic computing with data pre‐processing can help alleviate these burdens and enhance overall system efficiency. However, existing neuromorphic devices are typically limited to operation at room temperature, restricting their applications in extreme environments. Here, the use of wide‐bandgap ε‐Ga2O3 photo‐synapses as image pre‐processors is demonstrated, even at temperatures as high as 650 K. This is possible because of the high‐quality ε‐Ga2O3 epitaxial films, which exhibit angstrom‐level surface roughness, as evidenced by the highly uniform 152 photodetector devices. These devices show enhanced photo‐synaptic behavior at elevated temperatures, enabling high‐temperature image compression capability. The 4 × 4‐pixel images are compresed into 4 × 1 vectors, achieving an image recognition accuracy of 100% at 650 K. This work demonstrates the potential of ε‐Ga2O3 artificial synapses for extremely high‐temperature computing. This work reveals the high‐temperature image preprocessors using wide‐bandgap ε‐Ga2O3 optoelectronic synapses. High‐quality ε‐Ga2O3 films is grown. 152 photodetector devices with high uniformity are fabricated. This work demonstrates high‐temperature image compressing based on ε‐Ga2O3 optoelectronic synapses. 4 × 4‐pixel images can be compressed into 4 × 1 vectors at 650 K. he recognition accuracy can reach 100% after 32 epochs at 650 K.