Noisy Scanning Electron Microscope Research Articles

Line roughness estimation and Poisson denoising in scanning electron microscope images using deep learning

We propose the use of deep supervised learning for the estimation of line edge roughness (LER) and line width roughness (LWR) in low-dose scanning electron microscope (SEM) images. We simulate a supervised learning dataset of 100,800 SEM rough line images constructed by means of the Thorsos method and the ARTIMAGEN library developed by the National Institute of Standards and Technology. We also devise two separate deep convolutional neural networks called SEMNet and EDGENet, each of which has 17 convolutional layers, 16 batch normalization layers, and 16 dropout layers. SEMNet performs the Poisson denoising of SEM images, and it is trained with a dataset of simulated noisy-original SEM image pairs. EDGENet directly estimates the edge geometries from noisy SEM images, and it is trained with a dataset of simulated noisy SEM image-edge array pairs. SEMNet achieved considerable improvements in peak signal-to-noise ratio as well as the best LER/LWR estimation accuracy compared with standard image denoisers. EDGENet offers excellent LER and LWR estimation as well as roughness spectrum estimation.

Automatic nanohandling station inside a scanning electron microscope

Current research work on the development of automated nanohandling systems in a scanning electron microscope (SEM) is presented. Two experimental set-ups are shown in which nanohandling robots operate in the vacuum chamber of an SEM. A client—server control system that can integrate various microrobots and sensors has been developed and evaluated by automatic handling of TEM (transmission electron microscope) lamellae by two nanorobots. The robots are controlled in a closed-loop way by using images from several CCD (charge coupled device) cameras and from the SEM. Algorithms for real-time processing of noisy SEM images have been implemented and tested. The experiment on automatic handling of TEM lamellae inside an SEM is described. The other set-up contains a nanohandling robot using the probe of an atomic force microscope (AFM) as end-effector. The manipulation of individual multiwall carbon nanotubes (MWNTs) and the characterization of nanofilms by nanoindentation are the applications being investigated. The experimental set-up includes a nanopositioning piezo stage with three degrees of freedom (DoF) and a three-axis nanomanipulator. Piezoresistive AFM probes are applied as end-effectors. In this way the acting forces can be detected, allowing force feedback for the station's control system. First investigations have been carried out by bending MWNTs and calculating their elastic modulus.

Development and control of a versatile nanohandling robot cell

Current research work on the development of a nanohandling robot cell in a scanning electron microscope (SEM) is presented. An experimental setup is shown, in which two microrobots cooperate in the vacuum chamber of an SEM. One microrobot is used as manipulator of the tool and a second one is used for manipulating the object. Priority objective in future is the systematic prototyping of CNT-based nanodevices. Special algorithms for real-time processing of noisy SEM images have been implemented and tested. In addition, a 3D graphical user interface has been developed and delivers depth information which is often crucial for teleoperation of nanoobjects due to the lack of depth information in original SEM images. In order to enable automated nanohandling, a new control strategy is proposed. One key challenge for a successful automation is continuous pose estimation of tool and object which can be solved by applying the developed 2D and 3D tracking algorithms. The functionality of the robot cell is exemplified by handling and mechanical characterization of carbon nanotubes (CNTs).

Microrobot System for Automatic Nanohandling Inside a Scanning Electron Microscope

In this paper, current research work on the development of an automated nano handling station in a scanning electron microscope (SEM) is presented. An experimental setup is described, in which two mobile microrobots cooperate in the vacuum chamber of an SEM. The robots are positioned by a closed-loop controller with sensor data, which is provided by three charge-coupled device cameras and the SEM. Continuous pose estimation is carried out by processing noisy SEM images in real time. To enable the automation of complex tasks, a client-server control system that can integrate various microrobots and sensors is introduced. Finally, the overall system is evaluated by automatic handling of transmission electron microscope lamellae.

Object size measurement method from noisy SEM images by utilizing scale space

We proposed an object size measurement method from noisy scanning electron microscope (SEM) images with shot noise by utilizing the scale space approach. The proposed measurement algorithm consists of six steps: (1) specifying an object to be measured, (2) estimating the noise level, (3) constructing Gaussian and gradient scale spaces, (4) determining a starting scale for edge tracking, (5) edge tracking and (6) object size measurement. For the object size measurement, the model parameters of SEM images are estimated from regions segmented by edge positions at the finest scale. Then the appropriate scale for the object size measurement is automatically selected by estimated model parameters and an optimal scale parameter for smoothing filter in the Canny edge detector. And then, the object size is measured. We applied the algorithm to model SEM images and an SEM image. Though the processing time for our proposed method took about 13 times longer than that of the Canny detector, the range of the probability of extracting true edge positions was extended from 94% in the Canny detector to 80% in our method: the measurable region of the edge tracking was more extensive than that of the Canny detector.

DEVELOPMENT AND CONTROL OF A VERSATILE NANOHANDLING ROBOT CELL

Current research work on the development of a nanohandling cell in a scanning electron microscope (SEM) is presented. An experimental setup is shown, in which two microrobots cooperate in the vacuum chamber of an SEM. One microrobot is used as manipulator and a second one is used as sample holder. The task to be carried out is gripping of carbon nanotubes (CNT). In order to enable automatic gripping an automation strategy is introduced. One key challenge for a successful automation is the continuous pose estimation of the gripper and the CNT. Therefore, algorithms for real-time processing of noisy SEM images have been implemented and tested.

Metrics for focusing in extremely noisy scanning electron microscopy condition.

Finding a best focused image in very noisy condition is an extremely difficult task for the SEM user. If a performance, which is much higher than that of an expert in focusing, can be achieved in a computer-controlled scanning electron microscope (SEM), it will be very helpful for our field due to the many possible applications. To accomplish this work, we employ a powerful metric-the covariance obtained by a special scanning method. It can select the best focused image from a series of SEM images acquired by altering the focus of the objective lens under an extremely noisy SEM image condition. The noise immunity of the present method is quantitatively evaluated, and it is further improved based on the obtained evaluation result.