Deep learning-assisted piezoelectric nanogenerators based on layered SnS2/MoS2 heterostructures for interactive sensing and affective computing applications
Journal
Chemical Engineering Journal
Journal Volume
530
Start Page
173184
ISSN
1385-8947
Date Issued
2026-02-15
Author(s)
Lee, Cheng-Tse
Kuo, Hsin-Yu
Huang, Shih-Min
Chen, Chien-Chang
Chao, Ying-Chin
Ni, I-Chih
Murti, Bayu Tri
Chen, Chih-Jou
Yang, Po-Kang
Abstract
Layered metal dichalcogenides (LMDCs) have attracted widespread attention in piezoelectric systems, particularly for miniaturized energy devices and wearable sensing applications. However, insufficient output performance and limited sensing accuracy remain critical challenges. In this work, we developed a van der Waals heterostructure (vdWH) composed of molybdenum disulfide/tin disulfide (SnS2/MoS2) by using chemical vapor deposition (CVD). The enhancement of intrinsic piezoelectricity was further confirmed by piezoelectric force microscopy (PFM). Furthermore, the as-fabricated SnS2/MoS2 vdWH was assembled into a piezoelectric nanogenerator (SM-PENG) device and demonstrated as a self-powered healthcare sensor to monitor diverse human motions. A sequential fast data density functional transform (s-fDDFT) model was also employed to improve the prediction accuracy of the designed sensor. Most importantly, the SM-PENGs were integrated into a multi-channel facial expression recognition platform for affective computing applications. We believe that this SM-PENG not only shows great potential for future ultrathin active sensors, but also represents a promising step toward personalized healthcare monitoring and emotion recognition.
Subjects
Deep learning
Piezoelectricity
Recognition
Sensor
van der Waals heterostructure
Publisher
Elsevier BV
Type
journal article