Multi-Angle Plasma Focused Ion Beam (FIB) Curtaining Artifact Correction Using a Fourier-Based Linear Optimization Model. (7th December 2018)

• Record Type: Journal Article
• Authors: Schankula, Christopher W.; Anand, Christopher K.; Bassim, Nabil D.
• Is Part Of: Microscopy and microanalysis. Volume 24:Number 6(2018)
• Page Start: 657
• Page End: 666
• Publication Date: 2018-12-07
• Subjects: Microscopy -- Periodicals; Microchemistry -- Periodicals; 502.82
• DOI: 10.1017/S1431927618015234
• Languages: English
• ISSNs: 1431-9276

Abstract

Abstract: We present a flexible linear optimization model for correcting multi-angle curtaining effects in plasma focused ion beam scanning electron microscopy (PFIB-SEM) images produced by rocking-polishing schemes. When PFIB-SEM is employed in a serial sectioning tomography workow, it is capable of imaging large three-dimensional volumes quickly, providing rich information in the critical 10–100 nm feature length scale. During tomogram acquisition, a "rocking polish" is often used to reduce straight-line "curtaining" gradations in the milled sample surface. While this mitigation scheme is effective for deep curtains, it leaves shallower line artifacts at two discretized angles. Segmentation and other automated processing of the image set requires that these artifacts be corrected for accurate microstructural quantification. Our work details a new Fourier-based linear optimization model for correcting curtaining artifacts by targeting curtains at two discrete angles. We demonstrate its capabilities by processing images from a tomogram from a multiphase, heterogeneous concrete sample. We present methods for selecting the parameters which meet the user's goals most appropriately. Compared to previous works, we show that our model provides effective multi-angle curtain correction without introducing artifacts into the image, modifying non-curtain structures or causing changes to the contrast of voids. Our algorithm can be easily parallelized to take advantage of multi-core hardware.