New method of vibration isolation of scanning electron microscope

This paper presents a new image-processing approach to estimating a pointing error of the electron probe of a scanning electron microscope. An environmental disturbance causes a pointing error of the probe that is reflected upon a specimen image in a microscope. The new approach uses the deteriorated specimen image to estimate the pointing errors of the probe. The microscale is used as a specimen in this experiment, and a simple mathematical model is used to simulate a microscale image. The mathematical model is obtained by using the surfcace tilt and shadowing contrast of the microscale and by approximating the delta and step functions to differentiable functions. Simulated microscale images are identified by a least-squares procedure with measured images to estimate the pointing error of the probe. The estimated pointing errors are used to design a controller for vibration isolation of a scanning electron microscope. The designed controller is based on a transfer function from acceleration sensor outputs to the pointing errors. An acceleration sensor is situated close to the specimen stage in the microscope chamber to detect the stage motion. Sinusoidal excitation tests are performed to determine the transfer function. The sensor outputs are passed through the designed controller to compute the inputs into the image-shifting coils, and the coils move the electron probe to cancel the pointing errors. The performance of the designed controller is verified by comparing specimen images with and without control when the microscope vibrates. The comparison shows the pointing errors are significantly reduced in a region of lower frequencies.