# README — Micro-CT and SEM Image Data of Mechano-Fusion Coated Particles ## General Remark ## This dataset is associated with the following publication "Analyzing the 3D morphology of particle systems coated by mechano-fusion using micro-CT image data" and contains micro-computed tomography (micro-CT) and high-resolution scanning electron microscopy (SEM) image data of a model battery particle system. The particle system consists of alumina host particles coated with polytetrafluoroethylene (PTFE) guest particles via mechano-fusion, a high-intensity mixing, dry particle coating process. **Analyzing the 3D morphology of particle systems coated by mechano-fusion using micro-CT image data** Abstract: *Particle coating is a widely used strategy for tailoring particle properties to specific applications, with the morphology of the coating layer strongly influencing the effective properties of the resulting hetero-aggregates. Dry particle coating by means of mechano-fusion (MF) has gained increasing attention in applications such as battery material manufacturing, where coating layers surrounding host particles can significantly enhance performance and lifetime. To improve the understanding of process–structure relationships and consequently optimize the coating process, detailed characterization of the resulting hetero-aggregates is essential. In this paper, several geometric descriptors are introduced for the quantitative analysis of coating structures, including mean coating thickness, surface area coverage, and contact loss. In addition, correlations between these descriptors are investigated to provide a more comprehensive understanding of the interactions that govern the coating process. The analysis is based on micro-computed tomography (CT) images of a model particle system consisting of PTFE-coated alumina particles produced by MF. To reduce artifacts arising from the embedding of coated particles in epoxy, a hierarchical segmentation pipeline was developed for micro-CT image processing.* Authors: Philipp Rieder, Phillip Gräfensteiner, Judith M. Seyffer*, Tom Kirstein, Nora Brachhold, Christos G. Aneziris, Volker Schmidt, Urs A. Peuker *Corresponding author for dataset: Technische Universität Bergakademie Freiberg Institute of Mechanical Process Engineering and Mineral Processing Agricolastraße 1 D-09599 Freiberg Germany ## Sample Material and Sample Preparation ## ### Pristine Materials Info on the pristine particles used as host and guest particles: - **Host particles** Alumina (DAW-45, Denka Chemcial GmbH, Düsseldorf Germany) - median diameter d50=43.3 μm according to manufacturer, spherical particle shape: https://www.denka.co.jp/eng/product/detail_00040/ - **Guest particles** PTFE (FP30-PD-000110, Goodfellow GmbH, Hamburg, Germany) - mean diameter d=6...9 μm according to manufacturer ### Mechano-fusion Process Coated particles were produced by mechano-fusion using the Angmill Mechano-Fusion system (AMS) of the **Hosokawa Picobond** module connected to the **machinery platform Picoline (Hosokawa Alpine AG, Augsburg, Germany)** at 5000 1/min for 10 minutes. ### Sample preparation A representative sample of the coated particles after MF was produced by sample splitting using the **rotary micro riffler (QuantaChrome Instruments, Boynton Beach, USA)**. The coated particles were embedded in an epoxy resin. From the cured sample, a cuboid of 1mm × 1mm ×2mm was cut and two scans were performed with the micro-CT (upper and lower field of view (FOV)). The same sample was reembedded and polished to obtain cross sections (**LaboForce-100, Struers ApS, Ballerup, Denmark**). On those cross sections, high resolution SEM images were acquired. ## Micro-CT Equipment, Acquisition and Reconstruction Parameters ## All X-ray tomography scans were performed using a **Zeiss Xradia 510 VERSA (Carl Zeiss Microscopy GmbH, Oberkochen, Germany)** system. ### Acquisition Settings * Voltage: ~80 kV * Current: ~87 µA * Exposure Time: 18 s * Optical Magnification: 40× * Detector-to-rotation-axis distance: = 8.003 mm * Source-to-rotation-axis distance: = 13.003 mm * Pixel Size: 0.4099 µm * Number of projections: 1601 * Data Type: 16-bit ### Reconstruction Settings #### Upper_FOV * Center Shift: 10.34 * Beam hardening: Standard Beam Hardening Correction * Beam hardening constant: 0.05 * Rotation angle: 0.0 * Recon filter: Smooth * Sigma: 0.5 * Recon scaling: Custom (Min: -4.072113, Max: 3.801693) #### Lower_FOV * Center Shift: 11.152 * Beam hardening: Standard Beam Hardening Correction * Beam hardening constant: 0.05 * Rotation angle: 0.0 * Recon filter: Smooth * Sigma: 0.5 * Recon scaling: Custom (Min: -4.072113, Max: 3.801693) ## SEM Equipment and Acquisition Parameters ## All high-resolution SEM images were acquired with a **Tescan Amber (Tescan, Brno, Czech Republic)** ### Acquisition Settings * Voltage: 15 keV * Detector: Back scattered electron (BSE) detector * Field of View: 1mm (overview image) or 500 µm * Pixel Size: 0.12 µm * Data Type: 8-bit ## Dataset overview ## The micro-CT image data consists of two separat datasets (scans) of two different field of views (FOV) of the same sample. The FOVs represent the upper and the lower half of the prepared sample cuboid. The folders for each FOV (`Lower_FOV` and `Upper_FOV`) contain a preview screenshot `Preview.png` from the Zeiss 3D Viewer and subfolders for the image slices (in Tif formate) of the raw reconstructed image stack `Raw_images` (16-bit) as well as the different segmentation outputs. `Segmentation_phasewise` (8-bit) is the segmentation result after the two-step Dragonfly segmentation (see section "Phase-wise segmentation" of the Paper for more information). A distinct gray value is assigned to the four phases: background (0), host (1), guest coating (2) and air inclusion (3). In `Segmentation_final_aggregatewise` (16-bit) each aggregate (host-guest) is represented by a unique gray value as a result of the aggregate-wise segmentation. `Segmentation_hosts` contains only the host particles from the the previous aggregate-wise segmentation where again each host particle is represented by a unique gray value (16-bit). The TIFF images of the segmented datasets may appear fully black in standard image viewers because of their intensity range. The underlying data are intact, and the segmentation channels can be visualized by adjusting the threshold or display range, for example in ImageJ. The high-resolution SEM images are stored in the subfolder `Images` further subdivided into folders by polished and imaged cross section `Cross-section-x` (x=1...5). Additionally, the trained ilastik pixel classification model for image 0.1796.tif from cross section 2 is provided as `Pixel-seg.ilp`. The classifier contains pretrained three pretrained classes (alumina-host, PTFE-guest and background). The free software ilastik can be downloaded from https://www.ilastik.org/download for batch processing of further SEM images. ## Dataset Structure ## The dataset is provided as three zip archives with the following folder structure: ``` micro-CT-lower-FOV.zip ├── Raw_images/ ├── Segmentation_final_aggregatewise/ ├── Segmentation_hosts/ ├── Segmentation_phasewise/ └── Preview.PNG micro-CT-upper-FOV.zip ├── Raw_images/ ├── Segmentation_final_aggregatewise/ ├── Segmentation_hosts/ ├── Segmentation_phasewise/ └── Preview.PNG high-res-SEM.zip ├── Pixel-seg.ilp └── Images/ ├── Cross-section-1/ ├── Cross-section-2/ ├── Cross-section-3/ ├── Cross-section-4/ └── Cross-section-5/ ```