Convolutional neural network prediction of the photocurrent–voltage curve directly from scanning electron microscopy images

Abstract

In the pursuit of efficient and sustainable energy conversion, high-performance photocatalytic devices show promise. A key characteristic of these devices is the photocurrent density vs. applied voltage (J-V) curve, providing...

A robust methodology for PEC performance analysis of photoanodes using machine learning and analytical data

“…58,59 In general, the bandgap position has been paid attention; however, our result indicates that the morphology of the materials is the dominant reason for the performance, and this was also suggested from our recent results that predict the PEC performance from SEM images. 50…”

“…Here, the photocurrent value is specified at 1.23 V, but this method is not limited to this value, and the models can be built for other targets such as turn-on voltage, 49 PEC curve types 48 and shapes. 50…”

“…We prepared numerous samples following the same procedure and analyzed them using various methods, including X-ray diffraction (XRD), Raman spectroscopy, UV/Vis absorption spectroscopy, and photoelectrochemical impedance spectroscopy (PEIS). Descriptors extracted from the analytical data, such as peaks and patterns, were used to predict photocurrent values, turn-on voltage, 49 photocurrent curve types 48 and shapes, 50 ultimately identifying the dominant factors contributing to performance. This “data-driven” approach streamlines the process, as it does not require a comprehensive understanding of all analytical data features, making it a valuable strategy for resolving the reproducibility challenges frequently encountered with photocatalytic materials.…”

Subtraction Descriptors in Machine Learning for Optimizing the Cocatalyst Effect of Cobalt Phosphate on Hematite Photoanodes

“…Our strategy confronts the paucity of data head-on by integrating analytical data with ML approaches, a method that has seen success in several photocatalytic materials and devices. − The studies generated dozens of samples using a consistent method and analyzed them through techniques such as X-ray diffraction (XRD), Raman spectroscopy, UV/vis absorption spectroscopy, and photoelectrochemical impedance spectroscopy (PEIS). By using data descriptors such as peaks and patterns from these analyses, the research successfully predicted photocurrent values and pinpointed key factors affecting performance.…”