Machine learning enhanced electrohydrodynamic printing of heteroatom-doped composite electrodes for high-performance flexible supercapacitors

Abstract

Flexible supercapacitors are promising power sources for wearable electronics, but single-component electrodes often provide insufficient capacitance or poor rate performance. Integrating materials with different charge-storage mechanisms while controlling electrode structure remains challenging. Here, a machine-learning-enhanced electrohydrodynamic printing (MLEP) strategy was developed to fabricate multi-material composite electrodes with controlled resolution and thickness. Petroleum-coke-derived activated porous carbon (APC) served as the electric double-layer capacitive framework. Egg yolk–K 3 [Fe(CN) 6 ]-derived carbon material (YFC) provided additional pseudocapacitive active sites, while PEDOT:PSS improved interfacial contact and electron transport. A multi-objective model related the main printing parameters to electrode resolution and thickness, allowing more reliable selection of printing conditions. The APC/YFC electrode delivered 561.91 F g −1 at 0.5 A g −1 . PEDOT:PSS increased the capacitance of APC from 345.93 to 377.11 F g −1 at 1 A g −1 . The assembled flexible device achieved 214.65 F g −1 at 0.5 A g −1 , retained more than 88% of its initial capacitance after 10,000 cycles, and remained stable at bending angles from 0° to 180°. This work offers a practical route for preparing multi-component electrodes for flexible energy-storage devices.