Please choose your delivery country and your customer group
In this contribution the serialized reconstruction process of a porous La(0.58)Sr(0.4)Co(0.2)Fe(0.8)O(3-δ) (LSCF)-cathode will be illustrated. Thereby the focus lies on the further processing of the 3D data, obtained by FIB/SEM techniques. Possible sources of error and pre-conditions that are essential to obtain accurate results will be discussed. The influence of microstructure on electrode performance can be assessed only if the actual data on grain size, porosity, tortuosity are approximated accurately. FIB/SEM techniques provide a way to obtain highly detailed microstructure data, which can subsequently be used for the quantification of these microstructural parameters. Even more desirable would be to use the reconstructed microstructure as model geometry in a microstructure model, which is also possible with the presented method. Nevertheless, during the reconstruction process there are several sources of error, which can lead to erroneous data. In this work we have determined the most important aspects related to this point. The first sources of errors concern the FIB processing. It is important that the reconstructed cathode volume is large enough, so that the obtained data are not position-dependent. In the studied case, a minimum area covered by 7x7 particles is determined as a precondition. The resolution of the SEM images is a further important aspect, directly correlated with the particle sizes of the reconstructed microstructure. In the studied case, where a distribution of equally dimensioned particles with spherical shape was considered, the resolution should be chosen at least as high as to ensure that not less than 20-30 pixels (or voxels, respectively) correspond to the diameter of one particle. Another class of errors concerns the additional treatment of the obtained SEM images. The high precision of cutting-out and fixing the sectors of interest is a prerequisite for the quality of the reconstruction. Furthermore, the infiltration of the sample with a two component binder before FIB treatment helps to avoid material artefacts and the uncontrolled breaking away of particles during the FIB/SEM process. Under consideration of all the before mentioned aspects, an accurate quantification of valuable microstructural parameters from the reconstruction is possible. Nevertheless, it would be desirable to use the reconstructed microstructure directly in a model in order to investigate the interaction of microstructure and performance more accurately. Therefore a 3D numerical geometry model was developed, which can cope with the high number of voxels (107-108 voxels). It consists of a 3D FEM-model, with a solver based on high performance computing techniques, and a 3D stochastic geometry generator. Further details on this 3D numerical model will be the subject of a forthcoming article.