Provided are an inspection device that detects with high precision and classifies surface unevenness, step batching, penetrating blade-shaped dislocations, penetrating spiral dislocations, basal plane dislocations, and stacking defects formed in an SiC substrate and an epitaxial layer; and a system. In the inspection device using charged particle beams, a device is used that has an electrode provided between a sample and an objective lens, the device applies a positive or negative voltage to the electrode and obtains images. A secondary electron emission rate is measured and energy EL and EH for the charged particles are found. A first image is obtained using the EH and positive potential conditions. A second image is obtained using the EL and negative potential conditions. A third image is obtained at the same position as the second image, and by using the EL and positive potential conditions.

A dielectric microscopic observation is possible, which suppresses image flow regardless of scanning speed. There are provided a sample chamber 120 holding a sample 200 between a first insulating layer 121 on which a conductive layer 211 to be irradiated with a charged particle beam is laminated and a second insulating layer 122, an amplifier 141 that amplifies a potential change that occurs at an interface between the first insulating layer and the sample as the conductive layer is irradiated with the charged particle beam, and outputs the amplified result as a measurement signal, a main control unit 142 that converts the measurement signal from the amplifier into image data, and corrects the image data with a deconvolution filter 302 to generate corrected image data, a display unit 144 including an observation image display unit 501 and a filter adjustment unit 502 that displays setting information of the deconvolution filter, and an information processing device that displays the corrected image data on the observation image display unit, and when the setting information of the deconvolution filter displayed in the filter adjustment unit is changed, adjusts the deconvolution filter according to the changed setting information.

A method of calibrating analog-to-digital converters, ADCs, of a charged particle-optical device comprises: providing, for each of the ADCs, image data of charged particles detected from a sample output by the ADC; calculating, for each of the ADCs, at least one statistical value from a distribution of the image data output by the ADC; and changing at least one setting of at least one of the ADCs based on the calculated at least one statistical values so as to compensate for any mismatch between the at least one statistical value of the ADCs.

A system and a method for measuring of parameter values of semiconductor objects within wafers with increased throughput include using a modified machine learning algorithm to extract measurement results from instances of semiconductor objects. A training method for training the modified machine learning algorithm includes reducing a user interaction. The method can be more flexible and robust and can involve less user interaction than conventional methods. The system and method can be used for quantitative metrology of integrated circuits within semiconductor wafers.

The disclosure relates to systems and method for processing images. The method includes selecting a predetermined reference structure, the predetermined reference structure having a known feature size/shape. The method also includes obtaining a reference image of the predetermined reference structure, and capturing a calibration image of the predetermined reference structure using an observation device. The calibration image includes a plurality of features. Additionally, the method includes identifying at least one portion of the plurality of features of the calibration image that include a feature size/shape substantially similar to the known feature size and shape of the predetermined reference structure. Finally, the method includes combining the identified portion of the plurality of features of the calibration image to form a stacked feature image, and determining a point spread function (PSF) of the observation device by comparing the obtained reference image with the stacked feature image.

In the present invention, at the time of measuring, using a CD-SEM, a length of a resist that shrinks when irradiated with an electron beam, in order to highly accurately estimate a shape and dimensions of the resist before shrink, a shrink database with respect to various patterns is previously prepared, said shrink database containing cross-sectional shape data obtained prior to electron beam irradiation, a cross-sectional shape data group and a CD-SEM image data group, which are obtained under various electron beam irradiation conditions, and models based on such data and data groups, and a CD-SEM image of a resist pattern to be measured is obtained (S102), then, the CD-SEM image and data in the shrink database are compared with each other (S103), and the shape and dimensions of the pattern before the shrink are estimated and outputted (S104).

Provided are an image evaluation method and an image evaluation apparatus to evaluate a two-dimensional shape and a change in shape of a pattern side wall of a semiconductor pattern based on a SEM image, thus estimating an exposure condition. To this end, a method and a device include a storage unit that stores a model indicating a relationship between a feature amount that is obtained by creating a plurality of outlines from a SEM image and an exposure condition, and outline creation parameter information corresponding to the model; an outline creation unit that creates a plurality of outlines from a SEM image using the outline creation parameter information; and an estimation unit that uses a feature amount that is found based on the plurality of outlines created by the outline creation unit and the model to find an exposure condition.

Improved is the reliability of sample analysis performed using a charged particle beam apparatus.

The charged particle beam apparatus includes region setting means for setting an irradiation region for irradiating a sample with an electron beam and an irradiation prohibited region for prohibiting the irradiation of the sample with the electron beam using a low-magnification image of the sample captured under low vacuum. In addition, the charged particle beam apparatus includes captured image acquisition means for selectively irradiating the irradiation region with the electron beam with the inside of a sample chamber under high-vacuum and acquiring a high-vacuum SEM image of the irradiation region based on the secondary or backscattered electrons emitted from the irradiation region.

This method includes: generating a backscattered-electron image of a multilayered structure (400) including a plurality of patterns formed in a plurality of layers by a scanning electron microscope (50); classifying a plurality of regions of a virtual multilayered structure (300) including a CAD pattern created from design data of the plurality of patterns into a plurality of groups according to CAD pattern arrays in a depth direction of the virtual multilayered structure (1300); performing a matching between at least one of the plurality of patterns on the backscattered-electron image and a corresponding CAD pattern; calculating a brightness index value of a region on the backscattered-electron image corresponding to a region belonging to each group; and determining that there is a pattern defect in the region on the backscattered-electron image when the brightness index value is out of a standard range.

To shorten a time required for evaluation of a recipe while suppressing an increase in a data amount. A charged particle beam device includes a microscope that scans a charged particle beam on a sample, detects secondary particles emitted from the sample, and outputs a detection signal and a computer system that generates a frame image based on the detection signal and processes an image based on the frame images. The computer system calculates a moment image between a plurality of the frame images, and calculates a feature amount data of the frame image based on a moment.