Facile fabrication of micro-nano structure on Polydimethylsiloxane film surface for enhancing material recognition accuracy of machine learning-assisted triboelectric nanogenerator

One common challenge faced by previous TENG-based tactile sensor proposed for recognizing material types by a single sensor is the humidity and material stability interference, therefore, bioinspired design has been widely applied in the construction of moisture-resistant TENG-based device. Analyzing the structure-function relationship of functionalized triboelectric materials under environmental conditions and improving the sensing stability and accuracy through the design of surface hydrophobic structure remain major challenges in the development of intelligent sensing networks. To address these challenges, we designed and constructed elastic micro-nano structure on material surface base on the facile template method, achieving both hydrophobicity and stability of triboelectric material. The formation of micro-nano structure combined with the long hydrophobic alkyl chains of the material itself confer surface more outstanding hydrophobicity, which will facilitate the generation of triboelectric charges with differentiated signal waveform toward improving the sensitivity, stabilization and accuracy of intelligent sensors. As expected, the charge density and output voltage are enhanced by almost 17.4 % and 33.4 %, respectively. Additionally, the waveform characteristics of triboelectric signal also exhibit durability for subsequent machine learning-assisted recognition after immersing in polar solvent for 10 days and impacting for 9000 cycles. More importantly, the separation and compression between substrate and functionalized triboelectric material led to the flow of electrostatic electrons and the formation of unique output signals and self-powered power. Hence, moisture-resistant TENG-based sensor, display and data processing modules are integrated toward fabricate a material perception system for real-time monitoring with approximately 93 % (iron), 100 % (mask), 85 % (paper) and 91 % (plank) in the natural environment. Finally, this work provides a reliable strategy for designing smart sensors.