Dabin Kim1,Ziyue Yang1,Jaewon Cho1,Donggeun Park2,Dong Hwi Kim1,Jinkee Lee1,Seunghwa Ryu2,Sang-Woo Kim3,Miso Kim1
Sungkyunkwan University1,Korea Advanced Institute of Science and Technology2,Yonsei University3
Dabin Kim1,Ziyue Yang1,Jaewon Cho1,Donggeun Park2,Dong Hwi Kim1,Jinkee Lee1,Seunghwa Ryu2,Sang-Woo Kim3,Miso Kim1
Sungkyunkwan University1,Korea Advanced Institute of Science and Technology2,Yonsei University3
Piezoelectric polymer fibers are essential components in creating intelligent fabrics with adaptable shapes and energy-converting capabilities for wearable health and activity monitoring. However, producing high-performance smart polymer fibers has been challenging due to their low piezoelectric performance. In this work, high-performance piezoelectric yarns are presented that are both structurally robust and mechanically flexible. We achieved this by adding barium titanate nanoparticles (BTO, BaTiO<sub>3</sub>) to poly(vinylidene fluoride-trifluoroethlene) (P(VDF-TrFE)) during electrospinning to increase the electroactive <i>β</i>-phase formation. The optimally selected BTO-15 fiber mat-based device exhibited a substantially enhanced voltage of 167.31 V under mechanical bending, which was higher than that of its pristine counterpart. The resulting BTO-doped P(VDF-TrFE) mats were transformed into yarns, where the degree of <i>β</i>-phase crystallinity is found to be significantly higher in the yarns than in the mats, showing that the yarns could be a more favorable structural platform for piezoelectric performance. Furthermore, a BTO-15 piezo-yarn device was fabricated and tested, which exhibited an output of 16.17 V. Also, tensile tests were performed on the BTO-doped mat and BTO-doped yarns which results in the yarns being mechanically strengthened and having significantly improved elastic modulus and ductility compared to the mat. Finally, the BTO-doped piezo-yarn devices were woven into cotton socks as we denoted “piezo-socks”. This can monitor and identify body signals during seven human motion activities, such as jumping, running, and stair-climbing, by using convolution neural network algorithms designed for classification achieving a high accuracy of 99.6 %.