# Dataset of ultrasonic frequency – domain signals and machine – learning outputs for parameterising lithium – ion battery electrodes’ coating and calendering processes

## TL;DR

This paper provides an open-access dataset of ultrasonic signals and machine learning outputs to improve lithium-ion battery electrode manufacturing.

## Contribution

The work introduces the first curated frequency-domain ultrasonic dataset at the electrode level for battery manufacturing.

## Key findings

- The dataset includes FFT frequencies and magnitudes along with process metadata.
- It supports non-destructive inspection of electrode microstructural changes.
- The data can benchmark signal-processing methods and develop quality control models.

## Abstract

Inline, non – destructive diagnostics are essential for controlling electrode quality in roll – to – roll battery manufacturing to minimise the cost and wasting valuable minerals. Ultrasonic pulse – echo is sensitive to thickness, density, and porosity, but open, process – aware datasets especially at electrode level remain limited and scarce. This data article disseminates open – access ultrasonic frequency – domain datasets, aligned manufacturing process metadata and machine – learning outputs, acquired during coating and calendering of lithium-ion battery electrodes. Unlike prior studies that reported time – domain analyses, this work releases the first curated frequency – domain ultrasonic dataset ath the electrode level. The data repository includes FFT frequencies and magnitudes, with mass, thickness, density, and full design of experiment factors. These data enable inspection of manufacturing – induced microstructural change without destruction and support benchmarking of signal-processing pipelines as well as development of inline, physics-informed quality control models for battery – electrode manufacturing.