Toward Fast Calibration of Global Drift in Scanning Electron Microscopes with Respect to Time and Magnification

Abstract

It is a well-known fact that scanning electron microscopic (SEM) image acquisition is mainly affected by nonlinearities and instabilities of the column and probe-specimen interaction; in turn, producing a shift in the image points with respect to many parameters and time, in particular. Even though this drift is comparatively less in modern SEMs, it is still an important factor to consider in most of the SEM-based applications. In this airticle, a simple and real-time method is proposed to estimate the global drift from a set of target images using image phase correlation, and to model its evolution by using the recursive equations of time and magnification. Based on the developed model, it is opted to use a Kalman filter in real time for accurate estimation and removal of the drift from the images. The developed method is tested using the images from a tungsten filament gun SEM (Jeol JSM 820) and a field effect gun SEM (FEI Quanta 200). The derived results show the effectiveness of the developed algorithm and also demonstrates its ability to be used in robotics as well as in material characterization under SEM.

Spatiotemporal Calibration of Electron Microscopes

“…The imaging model of a SEM is the composition of three mappings [19,20] which are depicted in Figure 1: 1. A pixel-drift mapping from drifted to non-drifted image.…”

Fast image drift compensation in scanning electron microscope using image registration

“…This can be performed in two ways: forward mapping and inverse mapping. Using forward mapping, for each pixel p 1 ∈ I t , the corrected pixel p ∈Î is obtained directly by copying p 1 using H as given by (5).p = Hp 1…”

“…Naresh Marturi, Sounkalo Dembélé, and Nadine Piat are with Automatic control and Micro Mechatronic Systems (AS2M) department, Institute FEMTO-ST, Besançon, France. naresh.marturi at femto-st.fr [5], [6] and [7], a frequency domain phase correlation method is used. Cornille has proposed that the drift between pixels or between lines of an image is negligible; instead, the drift between the two images can be considered as a whole [3].…”

“…The first one is based on estimating the drift directly from the current image by comparing it with a reference image and correcting the current image in real time [5], [7]. The second approach is based on developing an empirical model of the drift that is then used to estimate and compensate the drift in real time [3], [4].…”

Scanning Electron Microscope Calibration Using a Multi-Image Non-Linear Minimization Process

“…[17,18] This drift can be calibrated and be compensated by some approaches. [19][20][21] However, few authors have investigated the spatial distortion for an accurate calibration of SEM. One reason might be the complexity of modeling of distortions in a high magnification, where the common model of distortion is weakened.…”