Ferroelectric Polarization-Enhanced Performance of Flexible CuInP2S6 Piezoelectric Nanogenerator for Biomechanical Energy Harvesting and Voice Recognition Applications

2D piezoelectric materials have strong intrinsic piezoelectricity and superior flexibility, which are endowed with huge potential to develop piezoelectric nanogenerators (PENGs). However, there are few attempts to investigate the energy harvesting of 2D ferroelectric materials. Herein, an enhanced output performance is reported by ferroelectric polarization in a PENG with exfoliated 2D ferroelectric CuInP2S6 (CIPS). Specifically, the polarized CIPS-based PENG produces a short-circuit current of 760 pA at 0.85% tensile strain, which is 3.8 times higher than that of unpolarized CIPS-based PENG. Systematical PFM and Raman analysis reveal that the ferroelectric polarization remarkably reinforces the effective piezoelectric constant of CIPS nanoflakes and boosts the in-plane migration and out-of-plane hopping of copper ions, which is the main reason for the enhancement of output performance. Furthermore, the CIPS-based PENG can not only be utilized to harvest biomechanical energy such as wrist joints movement, but also exhibits a potential for a voice recognition system integrated with deep learning technology. The classification accuracy of a series of letter sounds is as high as 96%. This study commendably broadens the application scope of 2D materials in micro-nano energy and intelligent sensors, which will have profound implications for exploring wearable nanoelectronic devices.

Automated summary

2D ferroelectric CuInP2S6 (CIPS) is used to enhance the output performance of a piezoelectric nanogenerator (PENG), resulting in a 3.8 times increase in short-circuit current compared to unpolarized CIPS-based PENG. The ferroelectric polarization reinforces the effective piezoelectric constant of CIPS nanoflakes and enhances the migration and hopping of copper ions, improving the output performance. Additionally, the CIPS-based PENG shows potential for voice recognition integrated with deep learning technology, achieving a high classification accuracy of 96% for letter sounds.