Accurate 3D microstructure
characterisation of porous materials by X-ray microtomography
A need for advanced and specialised image processing tools
Erwan Plougonven
The Department of Applied Chemistry
Applied Sciences Faculty
3 groups in the department (~55 people)
Analysis and Synthesis of Chemical Systems –
Cryotechnology laboratory
Nanomaterials, Catalysis, Electrochemistry
LGC – Environment, Energy, Reactors, separations
Tomography, a long history at the LGC Skyscan-1074 portable µCT 0.4 0.9 1.4 1.9 2.4 2.9 0 500 1000 1500 2000 we igh t (g) Time (s)
Micro and macro-tomography Skyscan-1074 portable µCT Skyscan-1172 desktop µCT PVC foam Metallic foam Human bone Extrudates bed
Examples of analysis
Nanocomposite polymer foams
(FNRS – ARC project)
Macadamia nuts
(Collaboration with UNSW)
Sintered soda-lime powder
Nanocomposite polymer foams Or what to do with poor quality 3D images ?
Polymer foams – characteristic length Statistical method 3D autocorrelation 2 )] )( [( ) ( E X µ X µ R t t R(x, y,z)
0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 Ave rag e co rrel ati on Distance 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 Ave rag e co rrel ati on Distance
Polymer foams – characteristic length
10 Statistical method 3D autocorrelation 2 )] )( [( ) ( E X µ X µ R t t R(x, y,z)
Polymer foams
Polymer foams – characteristic length -0.2 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 A verag e autoco rrel ation Distance (µm) PMMA sample 23 µm -0.2 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 A verag e autoco rrel ation Distance (µm) PCL sample 24 µm
Polymer foams – anisotropy PCL sample Co rr el ation -0.1 0.1 Co rr el ation -0.1 0.1 PMMA sample
Macadamia nuts Image analysis AND taste test
Macadamia nuts - Segmentation
Reconstruction
Morphological closing
Threshold Biggest connected component
Morphological closing
Gradient Active surface
Substraction Threshold
Combination Gaussian blur
Kernel segmentation : active surface
Hervé Delingette, Modélisation, déformation et reconnaissance d'objets tridimensionnels à l'aide de
Active surface
Hervé Delingette, Modélisation, déformation et reconnaissance d'objets tridimensionnels à l'aide de maillages simplexes, PhD thesis, Ecole centrale de Paris, 1994
Active surface
Hervé Delingette, Modélisation, déformation et reconnaissance d'objets tridimensionnels à l'aide de maillages simplexes, PhD thesis, Ecole centrale de Paris, 1994
Sintering of soda-lime glass Or what is a pore ?
Sintering and transfer properties
40 min
90 min
Pore network models
Classical definition: "part of the pore space bounded by the solid and planes erected where the hydraulic radius is minimal"
Before skeletonisation
Pore delimitation
Resulting pore decomposition
Sintering time: 150 minutes
Pore network model
Assumption: Flow between two pores controlled
only by their separation
Pore-network model Pore positioning graph
Pairwise pore connexions Branch ↔ Throat
Pore network models
Existence of non-pairwise connexions
Multiple connexions
Conclusions
Always the same basic problems (at first): segmentation, separation, labelling
No method is universal
Different material / acquisition / application different methodology
