FIB-tomography was implemented at Empa several years ago in the context with research on transport in porous materials (Holzer et al., 2006a; Holzer et al., 2004; Holzer et al., 2006b). Already in these early studies statistical image analysis of the 3D data was a special focus in order to establish microstructure-property relationships on a quantitative level (Holzer et al., 2006c; Münch et al., 2006a; Münch et al., 2006b) and it has been improved constantly over the last years. Both methodologies (FIB-tomography and the algorithms for 3D image analysis) are currently adapted and further developed for microstructural studies of SOFC electrodes. In the present contribution these activities are briefly summarized
FIB-tomography: In the last few years FIB-tomography has developed rapidly and it has become a well established method for 3D microstructure analysis. The recent methodological improvements consider the FEG-SEM technologies (dualbeam and/or crossbeam) and special detectors (i.e. for low kV BSE), system stability (i.e. suppression of drift phenomena), increasing size of image window (data cubes with voxel matrices up to 20003) and user friendly automation. These developments, which were reviewed in a recent publication (Holzer and Cantoni, 2010), open new possibilities for the study of materials with fine grained microstructures such as modern SOFC electrodes.
Quantitative image analysis: In order to establish correlations between microstructural features with the macroscopic properties of interest the microstructures need to be described with suitable geometrical pa-rameters. In this context specific algorithms for 2D and 3D characterization have been developed at Empa. Already a simple description by means of particle size distribution from 3D-images is quite chal-lenging, for example because the definition of discrete objects (particles or pores) from percolating phases is not straight forward. In order to circumvent these problems so-called 'continuous Phase Size Distributions' (PSDC) are determined from percolating structures (e.g. from the pore network) by a 3D-simulation of a liquid intrusion process which is similar to the mercury porosimetry (Münch and Holzer, 2008). In this way PSDs can be determined without complicated object recognition and stereological correction procedures. In contrast the typical characteristics of granular textures are described by statistical analysis of each discrete particle. The corresponding parameters include particle number density, 'discrete Particle Size Distribution' (PSDD), pair correlation function and principal axes of inertia. For this type of analysis special algorithms were developed for discrete feature recognition by splitting, for stereological correction of boundary truncation effects and for subsequent quantification (Holzer et al., 2007; Holzer et al., 2006a; Holzer and Münch, 2009; Münch et al., 2006a). Surface areas (e.g. total nickel surface) are determined from triangulated 3D-data. Thereby also specific types of phase boundaries such as the pore-nickel interface can be extracted (i.e. the surface which is catalytically active). In the context with transport in porous media recent developments also include the topological characterization by means of skeletonization and statistical analysis of associated graph parameters such as node coordination number, branch length, branch width (min. and max.) and branch vector orientations (Keller et al., submitted). These topological parameters enable a new characterization approach on a microscopic level which represents an alternative to the traditional 'macroscopic concepts' such as the formation factor, the tortuosity and the constrictivity (see e.g. Carman-Kozeny equation).
Loading...
9:59
Lorenz Holzer EMPA (2 of 3)by jacobrbowen86 views
Link to this comment:
All Comments (0)