Open NMC-622 Slurry Dataset: Ultrasound Measurements

The dataset is structured for research in material characterisation, signal processing, and machine learning-based slurry metrology. 📝 Dataset Description “An open ultrasonic pulse–echo dataset acquired from NMC-622 cathode slurries monitored under static conditions to characterise post-mixing structural recovery and ageing.

Slurries were prepared at solid contents of 63, 65, 67, and 69 wt.% using a Thinky ARE-250 mixing protocol, then transferred into a fixed test cup (7.5 mm height) for repeatable ultrasonic monitoring. A 5 MHz contact transducer driven by an EPOCH 650 pulser–receiver recorded A-scan waveforms over a 20 µs time window (8000 points), with repeated frame acquisition at controlled intervals. From each waveform, reflection amplitude and time-of-flight (ToF) features extracted from the slurry–air interface echo provide a compact representation of evolving acoustic response during recovery. ” 🧠 Purpose and Use Cases: 🔹 Benchmarking of signal processing pipelines for ultrasonic data collected from NMC-622 slurries. 🔹 Development of physics-informed feature extraction methods. 🔹 Training and testing machine-learning models for slurry metrology and coating readiness assessment. 🧩 Supporting Scripts 📊 data_load.py — Loads the full dataset from the HDF5 file and visualises ultrasound signals in both time and frequency domains. 📊 data_load_streamlit.py — Provides a graphical interface integrating data loading, visualisation, and analysis tools (see README for usage instructions). 🤖 trend_change_detection.py — Processes ultrasound signals to perform trend change detection, segmenting signals into distinct stages, and generates corresponding plots. 🤖 clustering_fft.py — Performs unsupervised clustering on FFT-transformed signals using methods such as K-means, with visualisations (e.g., t-SNE) to identify and confirm distinct regimes in the data.