Reducing curtaining effects in FIB/SEM applications by a goniometer stage and an image processing method
Thomas Henning Loeber,Bert Laegel,Sandra Wolff,Sebastian Schuff,Frank Balle,Tilmann Beck,Dietmar Eifler,Jan Henrik Fitschen,Gabriele Steidl
DOI: https://doi.org/10.1116/1.4991638
2017-11-01
Abstract:In the last two decades, focused ion beam (FIB) systems have been used for sample preparation. For example, the edges of a sample can be polished for analytical measurements or continuous cross-sections can be milled for three-dimensional (3D) tomography and reconstruction. One major challenge in both procedures is the so-called curtaining effect, i.e., increasing surface roughness in the direction of the milling depth. The roughness of the cut can influence the result of the measurement and the segmentation process. In the present study, the authors report on two different methods to reduce the curtaining effect, namely, a hardware- and a software-based solution. For instance, Tescan implemented the so-called “rocking stage” in its plasma FIB. However, this is not available for other FIB systems. Therefore, for our FEI gallium FIB, an inhouse-developed goniometer stage is installed, which can be adapted as necessary. With this relatively inexpensive solution, the sample can be rotated around an additional axis and tilted by ±8°. Different sample heights are adjustable, and the sample's edge can be polished and imaged without stage movement. However, for automated milling and imaging procedures such as 3D tomography, such a tilting stage is not feasible. Therefore, as a second option, an image processing method is proposed that can be applied after the milling procedure on a whole image stack. A novel variation of this method for mathematical image processing is developed to reduce milling artifacts. Besides the curtaining effect, additional artifacts such as discontinuities caused by redeposition of previously removed materials or charging effects can be removed. The method is applied to the entire 3D dataset, and distortions are reduced by using information of their particular structure and directional dependence. The resulting new image stack can then be used to compose a 3D volume reconstruction. As an example, the geometries of silicon carbide particles reinforcing an aluminum matrix can be measured with nearly no milling artifacts.