Bioinspired PEDOT:PSS-PVDF(HFP) flexible sensor for machine-learning-assisted multimodal recognition

Abstract

Poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) and polyvinylidene fluoride-hexafluoropropylene copolymer (PVDF(HFP)) are famous materials for developing flexible sensors, their conductivity, wettability, and piezoelectricity are affected by solvents. However, challenges remain in constructing multi-signals sensor by integrating PEDOT:PSS and PVDF(HFP) through solvent induced strategy, and existing sensors lack the intelligent extraction of stimulus information and accurate recognition of multimodal applications. In this work, a bioinspired flexible sensor with strain-pressure-humidity triple-signals is constructed based on PEDOT:PSS-PVDF(HFP) film (P-P film) through the synchronous induction effect of dimethyl sulfoxide (DMSO). The P-P film with double-layer structure presents excellent conductivity (800 S/cm), piezoelectricity (output voltage, 0.13 V-0.17 V), and humidity sensitivity (response/recovery time, 1.15 s/0.42 s). This phenomenon is resulted from the conformational transition of PEDOT, β-phase content enhancement of PVDF(HFP), and removal of insulated PSS under induction effect of DMSO. The data analysis, mining, and classification are performed through machine learning, which contribute to the recognition rates for different change degree of strain, pressure, and humidity are 93.35 %, 94.59 %, and 99.95 %, respectively. Importantly, a high-precision recognition (97.46 %) for multimodal applications is achieved by intelligently extract the features of strain-pressure-humidity triple-signals. This work is beneficial for improving the performance of existing sensors and developing new high-performance materials with the assistance of machine learning.