3D volumetric inspection of semiconductor wafers with higher throughput and accuracy

Abstract

The present invention relates to a method for inspecting three-dimensional circuit patterns of an inspection volume at an inspection site of a semiconductor wafer, and more particularly to a method, computer program product, and corresponding semiconductor inspection system for determining parameters of a 3D object, such as a HAR structure, in the inspection volume of a semiconductor wafer with higher precision and accuracy. The method employs milling and imaging of multiple cross sections in the inspection volume and forming a stack of averaged image slices therefrom. Inspection parameters of the 3D object from the stack of averaged image slices are thereby determined with higher precision and robustness. The method, computer program product, and device can be used for quantitative metrology, defect detection, process monitoring, defect review, and inspection of integrated circuits in semiconductor wafers.

Description

[Technical field]

[0001] The present invention relates to a method for inspecting three-dimensional circuit patterns of an inspection volume at an inspection site of a semiconductor wafer, and more particularly to a method, computer program product, and corresponding semiconductor inspection system for determining parameters of a 3D object, such as a HAR structure, in the inspection volume of a semiconductor wafer with higher precision and accuracy. The method employs milling and imaging of multiple cross sections in the inspection volume and forming a stack of averaged image slices from the multiple cross sections. Inspection parameters of the 3D object are thereby determined from the stack of averaged image slices with high precision and robustness. The method, computer program product, and device can be used for quantitative metrology, defect detection, process monitoring, defect review, and inspection of integrated circuits in semiconductor wafers. [Background technology]

[0002] Semiconductor structures are particularly delicate man-made structures that have very few defects. These rare defects are the signatures that defect detection or defect review or quantitative metrology devices look for. Manufactured semiconductor structures are based on prior knowledge, for example from design data, and are manufactured from a limited number of materials and processes. Also, semiconductor structures are manufactured as a series of layers parallel to the surface of a silicon wafer substrate. For example, in logic-type samples, metal lines run parallel in metal layers and HAR (high aspect ratio) structures, with vias running perpendicular to the metal layers or alternating conductive / non-conductive layers. The angles between the metal lines in different layers are 0° or 90°. On the other hand, it is known that for VNAND-type structures, the cross section is on average circular and arranged in a regular raster perpendicular to the surface of the silicon wafer. During manufacturing, a huge number of three-dimensional semiconductor structures are generated on the wafer, and the manufacturing process is subject to several influences. In general, the edge shape, area or overlay location of a semiconductor structure may be sensitive to the properties of the materials involved, the lithographic exposure or other manufacturing steps involved such as etching, polishing, deposition or implantation.

[0003] Feature sizes are shrinking in the manufacture of integrated circuits. Current minimum feature sizes or critical dimensions are less than 10 nm, e.g., 7 nm or 5 nm, and will approach less than 3 nm in the near future. Recently, feature sizes of 1 nm have also been realized. Thus, measuring the edge shape of patterns with high precision and determining the dimensions or line edge roughness of structures become challenging. The measurement resolution of charged particle systems is typically limited by the sampling raster of individual image points or dwell time per pixel on the sample and the charged particle beam diameter. The sampling raster resolution is configurable within the imaging system and can be adapted to the charged particle beam diameter on the sample. Although typical raster resolutions are 2 nm or less, the raster resolution limit can be reduced without physical limitations. The charged particle beam diameter has a limited dimension that depends on the type of charged particle selected, the charged particle beam operating conditions and the charged particle lens system used. The beam resolution is limited by about half the beam diameter. The resolution can be less than 3 nm, e.g., less than 2 nm or even less than 1 nm.

[0004] As the feature sizes of integrated semiconductor circuits become smaller and as the demands on the resolution of charged particle imaging systems increase, the inspection and 3D analysis of three-dimensional integrated semiconductor structures in wafers becomes more and more difficult. A semiconductor wafer has a diameter of 300 mm and consists of a plurality of several sites, so-called dies, each die containing at least one integrated circuit pattern, for example for a memory chip or a processor chip. The semiconductor wafer passes through about 1000 process steps, forming in the semiconductor wafer about 100 or more parallel layers including layers of transistors, intermediate layers of lines and interconnects and, in memory devices, multiple 3D arrays of memory cells.

[0005] A common method for generating 3D tomographic data from nanometer-scale semiconductor samples is the so-called slicing and imaging method, which is performed, for example, by a dual beam device. The slicing and imaging method is described in WO 2020 / 244795. According to the method of WO 2020 / 244795, a 3D volumetric examination is obtained in an examination sample extracted from a semiconductor wafer. This method has the disadvantage that the wafer needs to be destroyed to obtain the examination sample. This disadvantage is overcome by using the slicing and imaging method under an oblique angle into the surface of the semiconductor wafer, as described in WO 2021 / 180600. According to this method, at least a first examination site is determined and a 3D volumetric image of the examination volume is obtained by slicing and imaging a number of cross sections of the examination volume. In a first embodiment of the precision measurement, a large number N of cross sections of the examination volume are generated, the number N being 100 or even more averaged image slices. For example, 1000 slices are milled and imaged in a volume with lateral dimensions of 5 μm and a slice distance of 5 nm. For alignment and registration of the cross-sectional averaged image slices, several different methods have been proposed. For example, reference marks or so-called fiducial marks can be employed, or feature-based alignment can be employed. However, recent demands have shown that in many applications these methods require further refinement of the alignment and registration of multiple cross-sectional image slices.

[0006] Some inspection tasks require full 3D volumetric images with high accuracy and high throughput. In one embodiment, the inspection task is to determine with high precision a set of specific parameters of a semiconductor object, such as high aspect ratio (HAR) structures inside an inspection volume. For fast image acquisition with fast scan times, the signal-to-noise ratio (SNR) is very low. Prior art techniques use, for example, long dwell times, resulting in a long dwell time per pixel. In another embodiment, a single image acquisition can be replaced by a fast scan and the addition of several images to obtain a similar SNR of a single scan with a long dwell time. These methods are associated with several weaknesses. First, image acquisition for high SNR is very time consuming, slowing down volumetric image acquisition to only a few inspection sites per hour. Second, the accumulated charge due to the primary charged particles, and therefore the large current of the primary charged particles, leads to high charging effects in the wafer sample, thereby degrading the image generation. Third, during long image acquisition times, the inspection system may exhibit significant drift. This drift becomes more and more important as the demand for resolution increases, which is currently less than 5 nm. In the near future, resolution requirements will be less than 3 nm, less than 2 nm or even smaller, so that already drifts or vibrations of the order of a few nm during image acquisition of a 2D averaged image slice will result in unacceptable shearing.

[0007] Furthermore, for some applications, the methods described in WO 2021 / 180600 do not provide sufficient information for the determination of a set of parameters of a complex semiconductor structure. Summary of the Invention

[0008] It is therefore an object of the present invention to provide a further improvement to the slicing and imaging method described in WO 2021 / 180600. In general, it is an object of the present invention to provide a wafer inspection method for inspection of three-dimensional semiconductor structures in an inspection volume with higher accuracy, more information and faster.

[0009] The above objects are solved by the present invention as described in the appended claims and clauses and in the examples shown in the embodiments of the present invention.

[0010] This patent application claims priority to U.S. Provisional Patent Application No. 63 / 328,418, filed April 7, 2022, the disclosure of which is incorporated by reference in its entirety into this patent application.

[0011] The present invention is an improvement to a slicing and imaging method for producing a 3D volumetric image of an examination volume of a wafer sample. The examination of a semiconductor object involves the reconstruction of a 3D shape of a semiconductor object of interest from a 3D volumetric image formed by a stack of 2D averaged image slices. The stack of 2D averaged image slices is obtained by imaging a series of cross sections with a charged particle beam imaging device, for example a SEM (scanning electron microscope) or a HIM (helium ion microscope), and by repeated milling of the cross sections in the examination volume of the wafer sample with a FIB. A typical stack of 2D averaged image slices of a 3D volumetric image can include hundreds of averaged image slices up to more than 1000 averaged image slices.

[0012] In one embodiment, the axis of the charged particle beam imaging device is positioned perpendicular to the surface of the wafer sample, and the 2D averaged image slice includes a cross-section of the semiconductor object of interest, e.g., a memory (VNAND) structure such as a high aspect ratio (HAR) channel. In one embodiment, the wafer sample is provided by a full wafer, and the cross-section is milled by a FIB positioned under an oblique angle relative to the wafer surface.

[0013] According to a first embodiment of the present invention, a method for imaging a stack of M averaged image slices with higher throughput and high accuracy is provided. The method for imaging a stack of M averaged image slices according to the first embodiment uses fast milling of multiple cross sections through the examination volume and acquisition of a number N of cross section images by fast image scanning. According to the first embodiment, the time-consuming process of acquiring high-quality images of multiple cross sections is eliminated. Instead, the stack of M high-quality image slices is determined according to a moving average value of the first number N of cross section images. According to the first embodiment, a shorter image scan time is used for acquiring each cross section image, and multiple cross section images each having a higher noise level are generated.

[0014] In one embodiment, the number M is smaller than the number N of cross-sectional images, M<=N multiplied by a factor A, where A>=3, for example A=5, A=7 or A=20 or more. The number N of cross-sectional images is typically between the number Q of the multiple cross-sections and twice the number of cross-sections, where 2×Q=>N=>Q. The image acquisition method of a stack of M averaged image slices according to the first embodiment relies on typical properties of the semiconductor object of interest, e.g. the HAR channel, which is assumed to vary only slightly in a given direction. By calculating a moving average or average in a given direction of a large number of cross-sectional images, noise can be effectively reduced and the time of image acquisition can be shortened. Also, the generation of even more dense cross-sections due to the smaller distance between the cross-sections allows the milling time for the generation of the cross-sections to be used more effectively.

[0015] Between each subset of cross-sectional images of cross sections at small milling distances, the change of the semiconductor object of interest is typically limited. Therefore, alignment and compensation for lateral drift are computationally simple operations by determining the best fit condition or by determining the contrast or edge gradient in each averaged image slice. By acquiring a large number N of cross-sectional images of different cross sections formed in the examination volume, the drift of the stage or the charged particle beam column can be determined from the cross-sectional images and effectively removed for the determination of the moving average value. In one embodiment, multiple cross sections are milled by FIB through the examination volume at small z-distances, e.g., 2 nm, 3 nm or 5 nm. Each cross section is imaged at least once by a fast scanning motion with a charged particle beam imaging device, resulting in at least one cross-sectional image for each cross section. After or during image acquisition, a moving average value of a subset of the cross-sectional images is determined. During the determination of the average value, the image contrast is evaluated, e.g., by calculating the gradient in the moving average image, and if an overall low gradient or low contrast is detected in a particular direction, the determination of the average value is optimized by including a lateral shift for the cross-sectional images. This reduces the effect of drift in that particular direction. During optimization of drift compensation, the maximum expected drift can be bounded to reduce computation time and increase the robustness of the method. Drift can be determined as the absolute drift of each cross-sectional image relative to a reference, or as a relative drift, for example, to the previous cross-sectional image. Different registration or alignment methods can be employed to improve the speed and performance of drift compensation. In certain cases, manual verification of drift compensation is possible and incorrect image shifts can be discarded.

[0016] In one embodiment, the texture filling in the cross section may vary. For example, a semiconductor structure may include several different stacks or decks of semiconductor features stacked on top of each other. Thus, the cross section image may include image sections of different decks. For example, a cross section through a first VNAND deck and at least a second VNAND deck by milling under an oblique angle. The location of the transition zone between the two decks varies gradually across the cross section image. Subsequent cross section images share image sections or regions from the same deck that have the same texture (e.g., a region showing a channel cross section of the upper VNAND deck), but also have regions of different texture (e.g., deck transitions). In such cases, the registration for determining the lateral shift needs to be limited to regions showing the same texture so that the moving average is not corrupted.

[0017] In one embodiment, the formation of M averaged image slices is achieved by one-dimensional (1D) numerical convolution from a number N of cross-sectional images with a convolution kernel. In one embodiment, the number N of cross-sectional images form a 3D data set with abscissas x and y for each image, with different z coordinates corresponding to a given direction. The moving average value or mean is represented by convolution with a 1D convolution kernel over the z coordinates of the 3D data set. The convolution kernel can be a function over a given number of cross-sectional images, for example a moving rectangular function or a moving Gaussian kernel in the z direction. Other convolution kernels for calculating different weighted moving average point-wise median filters are also possible. Other examples of convolution kernels correspond to filters for finding the minimum of a norm function or a matching function with a given noise model. The width of the convolution kernel can be pre-determined or adaptive according to the local noise level or required SNR of the resulting M averaged image slices. In another embodiment, a machine learning based algorithm may be applied to optimally perform the calculation of the moving average value.

[0018] According to the first embodiment, the image scan time used to acquire each cross-sectional image is reduced, and many cross-sectional images are generated. With each new image scan, milling artifacts, if any, in the previous cross-sectional image are removed, and the defects or artifacts of the milling operation may vary from cross-section to cross-section. According to the first embodiment of the invention, at least some of these milling artifacts are averaged out by forming a moving average. The depth of field of charged particle beam imaging systems is rather low, making perfect focusing difficult to achieve. By averaging multiple cross-sectional images, focusing defects of the individual cross-sectional images are averaged out, and a higher resolution averaged image slice is generated.

[0019] According to a first embodiment of the present invention, many cross-sectional images are obtained with a reduced scan time and therefore an increased noise level. Averaging forms a stack of M averaged image slices, which reduces the increased noise level of the rapidly acquired cross-sectional images and forms an averaged image slice with an increased SNR.

[0020] When determining the moving average, global drifts can be detected and compensated for, while individual deviations of the trajectory of, for example, the HAR channel can nevertheless be detected. Further alignment and registration of the cross-sectional image at a reference mark or image feature present in the semiconductor is also possible. Image acquisition can be performed by alternating image scanning and milling operations, or by interlaced imaging and milling. Calculation of the averaged image slice can be performed during image acquisition by the slice and imaging method, thereby reducing the memory required to store N cross-sectional images.

[0021] The method according to the first embodiment is further applicable to generate training data for an inspection method using machine learning: annotations can be transferred from averaged image slices with low noise levels to a set of cross-sectional images used for averaging, each having a large noise level and acquired by a high-speed image scan.

[0022] According to a second embodiment of the present invention, a method for imaging a stack of M averaged image slices with higher throughput and accuracy is provided, further improving the accuracy of imaging. According to the second embodiment, image acquisition and milling operations are interlaced and performed simultaneously, so that image acquisition of N cross-sectional images and milling of N cross-sections are performed simultaneously. In one embodiment, FIB milling is performed with a FIB beam arranged in the YZ plane at a milling angle GF with respect to the y axis. The FIB is scanned in a first direction (x direction) perpendicular to the FIB beam. At the same time, imaging charged particle beam image acquisition is performed by fast scanning of the imaging charged particle beam across the cross-section in the x direction and stepping in the y direction. This results in a number N of cross-sectional images, each cross-sectional image of each interlaced milling and imaging operation comprising a first area according to the cross-section before the actual milling operation and a second area according to the cross-section after the actual milling operation. Each cross-sectional image thus comprises an image area of ​​a different z position in the stack of cross-sectional images. The multiple N cross-sectional images again form a 3D data set with abscissas x and y for each image and different z coordinates corresponding to the two regions in each cross-sectional image. This allows further reducing the effects of drift between the different cross-sectional images. The method according to the second embodiment can be combined with the method according to the first embodiment to form a stack of M averaged image slices by taking a moving average or mean as described in the first embodiment.

[0023] According to one embodiment, further cross-sectional images can be selectively added, for example by repeated image scanning. For example, a first set of cross-sectional images can be obtained in parallel with a milling operation as described in the second embodiment, and a second set of cross-sectional images can be obtained between two subsequent milling operations.

[0024] The throughput of the method for inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample can be further increased by limiting the performance of time-consuming image scanning operations to selected regions of interest.

[0025] According to a third embodiment of the present invention, the imaging method of a stack of M averaged image slices is further improved to further increase the imaging throughput. According to the third embodiment, the time for obtaining each cross-sectional image is reduced by limiting the image scan to at least one region of interest (ROI).

[0026] Full-scale 3D volume data generation using slice and image methods provides the most complete information about the inspection volume in the wafer sample. However, slice and image techniques generate an undesirably large amount of data. On the other hand, a fully reconstructed inspection volume often contains a lot of redundant information that does not contribute to the particular semiconductor object of interest and is not necessary for a measurement or set of measurements. At the same time, image quality, resolution and pixel size in the vicinity of the feature of interest are often compromised in order to reduce the acquisition / processing time and data volume. Therefore, it is desirable to be able to adjust the imaging and milling parameters so that the image scan is limited to a cross section that intersects with the region of the volume of interest inside the inspection volume. The method according to the third embodiment further reduces the total preparation time of the relevant 3D volume data of the semiconductor object of interest.

[0027] A method for inspecting a 3D semiconductor object of interest in an inspection volume of a wafer sample according to a third embodiment includes a step of defining at least a first volume of interest in the inspection volume of the semiconductor wafer, the first volume of interest being within the inspection volume. In one embodiment, a region of interest (ROI) is formed by at least one volume of interest within the inspection volume, and the ROI of the subsequent cross-sectional image is determined according to an overlap with the at least one volume of interest. In one embodiment, the volume of interest is buried deep within the inspection volume below the surface of the wafer.

[0028] The method further includes milling a plurality of cross sections through the inspection volume and acquiring a number of cross-sectional image sections by performing an image scanning motion of the cross-sectional sections within at least the first volume of interest, hi one embodiment, the milling is performed at an oblique angle with respect to the wafer surface and at least the first volume of interest is oriented perpendicular to the wafer surface.

[0029] In one embodiment, the at least first volume of interest is defined according to CAD data. The method may further include aligning the at least first volume of interest in the inspection volume according to at least a first image scan of a first cross-section in the inspection volume. From the first image scan, a precise position of the inspection volume in the wafer is obtained and registered, allowing the position of the volume of interest to be precisely determined. In one embodiment, the aligning step includes further realignment of the at least first volume of interest in second or further image scans of at least second and further cross-sections in the inspection volume. This allows the position of the volume of interest to be precisely maintained even in the presence of, for example, milling misalignments.

[0030] For example, in a regular array of HAR channels, the region of interest may include only one column or one row of HAR channels. First, the region of interest is made accessible for imaging by performing a milling operation with the FIB beam. This forms a cross section covering the at least one ROI. After detection of the region of interest, the image acquisition with the imaging charged particle beam is reduced to the at least one ROI. This reduces the time of image acquisition. The at least one ROI may have a complex shape, for example a cross or T shape formed by columns and rows of HAR channels. The at least one ROI may also be formed by several separate ROIs distributed over each cross section. The ROI may also be a function of the z coordinate or depth of the cross section formed in the inspection volume of the wafer and may follow the 3D shape of the semiconductor object of interest. The z-dependence of the size, shape and position of the ROI may be pre-determined by prior information of the semiconductor object of interest. The tracking of the unknown semiconductor object of interest may be guided by model-based assumptions or using machine learning based techniques to track the trajectory of the structure in the inspection volume.

[0031] In one embodiment, it is assumed that the change in the semiconductor object of interest is limited to a certain region in the test volume. This assumption can be triggered by a priori information, including CAD information or information obtained from other test volumes. According to one embodiment of the method according to the third embodiment, the throughput is further increased by adaptively changing the distance between two adjacent cross sections. Thus, the method includes adjusting the distance between two adjacent cross sections based on a priori information of the semiconductor object of interest in at least the first volume of interest. In one embodiment, two adjacent volumes of interest are defined, and the alignment of different layer stacks of the semiconductor sample can be determined.

[0032] In one embodiment, the at least first volume of interest is determined during milling of a plurality of cross sections into the examination volume. According to the method, the step of defining the at least first volume of interest comprises: a) acquiring a first image scan of a first cross-section; b) selecting a first cross-sectional image section in the first image scan, the first cross-sectional image section including a cross-section of the semiconductor object of interest; c) milling a second cross-section into the control volume of the wafer sample; d) selecting a second cross-sectional image section by projecting the first cross-sectional image section onto a second cross section; e) acquiring a second cross-sectional image section of the second cross-sectional section; f) repeating steps c) through e) by milling additional cross sections and selecting and acquiring additional cross section segments until a predetermined interruption criterion is met.

[0033] The first cross-sectional image section corresponds to a cross section through the volume of interest in the first image scan. In the first image scan, the cross section through the volume of interest is determined by a cross section of the semiconductor object of interest having a boundary region at the periphery of the cross section of the semiconductor object of interest. The cross section of the semiconductor object of interest can be determined by image processing, pattern recognition, template matching, or machine learning operations or a combination of these methods and using CAD information.

[0034] In one embodiment, the center of gravity of the cross section of the semiconductor object of interest corresponds to the center of gravity of the first cross section image section. After milling of the second or further cross section, the first cross section image section is projected onto the second or further cross section to obtain an image of the second or further cross section with the first cross section image section. The cross section of the semiconductor object of interest is again determined by image processing, pattern recognition, template matching or machine learning operations or a combination of these methods to determine the center of gravity of the semiconductor object of interest. If the center of gravity of the semiconductor object of interest deviates from the center of gravity of the first projected cross section image section, the first cross section image section is adjusted to form a second or adjusted cross section image section with the center of gravity of the semiconductor object of interest coinciding with the center of gravity of the second adjusted projected cross section image section. This adjustment method is applied cross section by cross section, and the volume of interest is continuously adjusted by milling operations and image acquisitions throughout the inspection volume.

[0035] In one embodiment, the method further includes adjusting a position or size of the first cross-sectional image section according to the second cross-sectional image section, and the position and size of the cross-section of the semiconductor object of interest in the second cross-sectional image section are analyzed by image processing.

[0036] In one embodiment of step e), the projection of the first cross-sectional image segment onto the second cross-section is performed in a predetermined direction, e.g., a direction parallel to the HAR channel, in such an embodiment the predetermined direction is oriented perpendicular to the wafer surface.

[0037] In one implementation of the method according to the third embodiment, a high-speed cross-sectional image scan of a cross section is obtained by high-speed image scanning to generate a cross-sectional image with a reduced SNR. In the low SNR cross-sectional image, a region of interest of the volume of interest is determined, causing image acquisition with increased dwell time and higher SNR only for that region of interest. The steps of high-speed image acquisition and determination of the region of interest in the low SNR rapidly acquired cross-sectional image can be repeated.

[0038] In one example, the method of the third embodiment includes generating a 3D volumetric image of the 3D semiconductor object of interest. The 3D volumetric image may not be limited to a volume of interest including the 3D semiconductor object of interest, but may be formed for the entire inspection volume. In one example, the 3D volumetric image of the inspection volume includes a 3D volumetric image of at least a first volume of interest and augmented 3D image data from prior information outside the at least first volume of interest. The augmented 3D image data may be obtained from CAD data or from a previously acquired 3D volumetric image of a reference volume.

[0039] The method according to the third embodiment can be combined with the method according to the first or second embodiment described above, whereby the formation of M averaged image slices from N cross-sectional images is reduced to the determination of a moving average value in at least one volume of interest. Also, in an interlaced operation, a first volume of interest can be in the area of ​​a first z-position before the actual milling operation, and a second volume of interest can be in the area of ​​a second z-position after the milling operation.

[0040] According to a fourth embodiment of the present invention, the imaging method of a stack of M averaged image slices is further improved, and the throughput of image acquisition is further increased. According to the fourth embodiment, a sparse imaging method is applied. The effect of sparse imaging is similar to the method according to the third embodiment, but using multiple small ROIs. For example, in multiple HAR channels, image acquisition can often be reduced to a few HAR channels or even to the details of the HAR channel. The method according to the fourth embodiment relies on prior knowledge of the semiconductor object of interest in the inspection volume. During sparse image acquisition, image acquisition is reduced to sparse image sampling positions, and each averaged image slice is filled with augmented information by prior knowledge.

[0041] The augmented information can be derived, for example, according to a CAD model of the semiconductor object of interest, which is modified to best match the image data of the sparse image sampling locations. The sparse image sampling locations can be predetermined according to a priori information, for example from CAD data of the semiconductor object of interest. The method according to the fourth embodiment can be combined with any of the methods according to the first to third embodiments described above to improve the throughput of a wafer inspection operation.

[0042] The fast image acquisition method of the first embodiment of the present invention allows the imaging parameters of a charged particle imaging system to be optimized for the detection of a particular semiconductor object of interest. The optimization may include, for example, the selection of a detection method for the detection of interaction products generated in each cross section. In some implementations, the disadvantage may occur that improved detection of a particular semiconductor feature is achieved at the expense of other semiconductor features. For example, when imaging a semiconductor memory wafer, the detection of the HAR channel boundary may conflict with the parameters required to accurately detect the word line layer. Alternatively, the detection of the accurate HAR channel boundary may conflict with the ability to detect the recess etch boundary. This problem becomes more severe when performing destructive milling by slicing and imaging techniques, where the image cannot be reacquired due to the cross-sectional layer that has already been removed.

[0043] According to a fifth embodiment of the present invention, the method for fast imaging of a stack of M averaged image slices is further improved to further increase the accuracy of image acquisition. The method of the fifth embodiment employs multi-mode image acquisition, including at least a first and a second imaging operation mode of a charged particle beam imaging system, for improved accuracy of imaging of a stack of M averaged image slices. In one embodiment, different imaging settings or imaging modes can be used to differentiate imaged structures by elemental composition with less sensitivity to topographical contrast or surface effects. The imaging modes can be modified and combined, and the information in the fast acquisition image scan is augmented by information from different imaging modes.

[0044] According to a first embodiment, the detector characteristics are adjusted during at least one imaging operating mode. For example, the detector gain is adjusted to a low signal level, and during a second imaging operating mode, the detector gain is adjusted to a high signal level. This allows the signal obtained by the detector to be adjusted, for example, to different ROIs within an inspection task. According to a further embodiment, the cut-off energy of secondary or backscattered electrons is adjusted. This allows, for example, imaging to be reduced to a cross section in a sample or to a specific depth below a certain material.

[0045] According to a second embodiment, the different imaging modes are realized by using separate or different detectors. For example, the first detector and the second detector can be arranged at different angular intervals and configured to determine the topographic contrast. For example, the first detector can be configured to collect large angle scattered electrons and the second, in-lens detector can be configured to collect small angle backscattered electrons. For example, the first detector can be an electron detector and the second detector can be an X-ray detector. This multi-mode image acquisition further improves the imaging of the M averaged image slices by obtaining additional information.

[0046] In one embodiment, the method includes changing the detection mode between subsequent image scanning operations, the change in detection mode including at least one of changing the dynamic range, changing the energy range of the interaction product, or changing the type of the interaction product.

[0047] The method according to the fifth embodiment can be combined with the method according to any of the first to fourth embodiments for each imaging mode of multimode imaging. For example, a different convolution kernel can be applied to each of the z-stacks of cross-sectional images of each imaging mode. This allows a high SNR to be achieved in parallel for all imaging modes. Different multimode imaging modes can also be applied to selected ROIs or to sparse imaging locations. The additional information of multimode imaging further improves the desired information and imaging accuracy regarding the semiconductor object of interest without compromising the throughput enhancement according to the first to fourth embodiments.

[0048] According to a sixth embodiment, an inspection system for 3D inspection of a semiconductor object of interest in an inspection volume of a wafer is provided. The inspection system configured for performing the method according to any of the first to fifth embodiments comprises a wafer sample holder and a stage for holding and positioning the wafer sample. The inspection system comprises a dual beam device, comprising a FIB column and a charged particle beam imaging system, e.g. a scanning electron microscope (SEM) or a helium ion microscope (HIM). In another example, a corrected electron microscope with multiple detectors for angle-resolved imaging of the interaction products is applied. Such a corrected electron microscope is described in German patent application no. 102021212203.5, filed on 28 October 2021, and incorporated herein by reference. Both columns of the dual beam device are arranged at an angle, forming an intersection of the optical axes of the columns. The stage is configured to position and hold the sample comprising the inspection volume in the vicinity of the intersection. The dual beam device further comprises at least one detector of secondary or backscattered charged particles. In one embodiment, the dual beam device further includes at least a second detector for acquiring a second signal by a multimode imaging operation. The inspection system further includes a control unit for controlling an operation of the dual beam device. The inspection system further includes a memory for storing a software program code, an operating instruction program code, and a memory for storing the acquired digital image data. The inspection system further includes a processor configured to execute processing instructions for the determination of the moving average. The inspection system is further configured to receive prior information, e.g., CAD information.

[0049] According to an embodiment, a charged particle imaging system of a dual beam device is configured to perform different imaging modes. The different imaging modes can include different detectors or detectors configured for different detector settings. The detectors can include a grid with a variable grid voltage. This allows the detector settings to be changed by selection of a repulsive field to cut off low energy charged interaction products. The detectors can include a variable analog gain factor configured for adjustment of the analog signal before conversion to a digital signal. This allows weak signals to be acquired with greater precision and definition.

[0050] One embodiment of a charged particle imaging system includes a scan deflector and a scan controller configured to perform different imaging scan operations, including, for example, line averaging, varying dwell time, varying the scan path, for example, performing a zigzag scan technique, or a serpentine scan technique.

[0051] One example of a charged particle imaging system configured to implement different imaging modes according to the fifth embodiment includes a charged particle source and a particle-optical element configured for varying the acceleration voltage, the beam current, the charged particle beam angle, the numerical aperture of the charged particle beam, or the cross-sectional shape of the charged particle beam.

[0052] One embodiment of the charged particle imaging system includes means for correcting chromatic and spherical aberrations, thereby enabling higher resolution inspection at lower electron voltages below 1 keV, for example below 500 eV.

[0053] According to an embodiment of the present invention, the acquisition time of a 3D volumetric image of a semiconductor object of interest by a slicing and imaging method is reduced and the image accuracy is increased. This increases the throughput of the inspection operation and provides an inspection system for 3D inspection of a semiconductor object of interest with higher throughput and higher accuracy. An inspection system for volumetric inspection of a semiconductor wafer with higher accuracy and throughput is provided. The inspection system is configured for milling and imaging multiple cross sections in the inspection volume and determining inspection parameters of the 3D object from the multiple cross section images. The present invention provides a device and method for 3D inspection of an inspection volume in a wafer and determining a set of parameters of semiconductor features within the inspection volume with high accuracy and higher throughput. The method and a system configured to perform the method can be used for quantitative metrology, defect detection, process monitoring, defect review and inspection of integrated circuits in a semiconductor wafer.

[0054] The present invention described by the examples and embodiments is not limited to these embodiments and examples, and can be implemented by those skilled in the art through various combinations or modifications. The present invention will be better understood with reference to the following drawings. [Brief description of the drawings]

[0055] [Figure 1] FIG. 1 illustrates an example of an inspection system for 3D volumetric inspection with a dual beam device. [Diagram 2] FIG. 1 illustrates an exemplary method for volumetric inspection of a wafer using tilted cross-section milling and imaging with a dual beam device. [Diagram 3] FIG. 2 shows an example of an averaged image slice. [Figure 4] FIG. 2 illustrates an example of a method according to a first embodiment. [Diagram 5] FIG. 13 is a diagram showing an example of the number of slices and a data stack of slice images. [Figure 6] FIG. 2 shows a first example of the calculation of an averaged image slice. [Figure 7]FIG. 13 illustrates a second example of the calculation of an averaged image slice. [Figure 8] FIG. 13 illustrates a third example embodiment of the calculation of an averaged image slice. [Figure 9] FIG. 13 shows an example of scanning acquisition of cross-sectional images in parallel with a milling operation according to the second embodiment. [Figure 10] FIG. 13 illustrates a further example of the calculation of an averaged image slice. [Figure 11] FIG. 13 illustrates a method according to a third embodiment using a limited volume of interest. [Figure 12] FIG. 11 is a diagram illustrating a first example of the third embodiment. [Figure 13] FIG. 13 is a diagram illustrating a second example of the third embodiment. [Figure 14] FIG. 13 illustrates a further example of the third embodiment using dynamic adjustment of the region of interest. [Figure 15] FIG. 13 shows a further example of the third embodiment using determination of overlay or registration error. [Figure 16] FIG. 13 illustrates a method according to a fourth embodiment. [Figure 17] FIG. 13 illustrates an example of a fourth embodiment of sparse image acquisition. [Figure 18] FIG. 1 illustrates an example of an inspection system for 3D volumetric inspection with a dual beam device using a corrected electron microscope. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0056] The same reference numbers are used throughout the figures and description to describe identical or similar features or components. A coordinate system has been chosen in which the wafer surface 55 coincides with the XY plane.

[0057] Recently, for the inspection of 3D inspection volumes in semiconductor wafers, slicing and imaging methods have been proposed that can be applied to inspection volumes inside the wafer. This produces a 3D volumetric image of the inspection volume inside the wafer in a so-called "wedge cut" manner or in a wedge cut shape, without the need for the removal of a sample from the wafer. The slicing and imaging methods are applied to inspection volumes with dimensions of a few μm, for example with a lateral extent of 5 μm to 10 μm or up to 50 μm, in wafers with a diameter of 200 mm or 300 mm. In order to make the cross-section accessible at an oblique angle relative to the top surface, V-shaped grooves or wedges are milled in the top surface of the semiconductor wafer. The 3D volumetric image of the inspection volume is acquired at a limited number of measurement sites, for example representative sites of the dies, for example at a process control monitor (PCM) or at sites identified by other detection means. The slicing and imaging methods result in the wafer being destroyed only locally, while other dies can still be used or the wafer can still be used for further processing. The 3D volumetric imaging method and inspection system are described in WO 2021 / 180600, which is incorporated herein by reference in its entirety. The present invention is an improvement and extension of the 3D volumetric imaging method and inspection system where higher throughput is required. One of the results of the present invention of the slicing and imaging technique for semiconductor device inspection is the reduction of imaging time required to obtain multiple averaged image slices.

[0058] The present invention is applicable to semiconductor devices that consist of semiconductor elements having high aspect ratios and / or located in multiple layers within the device. The fabrication of such devices relies heavily on the ability to characterize the semiconductor elements in 3D. Full-scale 3D tomography with the improved method and apparatus of the present invention uses improved slicing and imaging techniques with higher throughput.

[0059] The inspection system of the present invention is shown in Fig. 1. According to a first embodiment, an improved wafer inspection system 1000 for 3D volume inspection is provided. The improved wafer inspection system 1000 for high throughput 3D volume inspection is configured for a slice and image method under a wedge cut geometry by a dual beam device 1. For the wafer 8, several measurement sites are defined, including measurement sites 6.1 and 6.2, in a location map or inspection list generated from the inspection means or from design information. The wafer 8 is placed on a wafer support table 15. The wafer support table 15 is mounted on a stage 155 having actuators and a position control device 21. Precision control means 21 for the wafer stage 155, such as actuators and laser interferometers, are known in the prior art. A control unit 16 is configured to receive information about the actual position of the wafer stage 155 and to control the wafer stage 155 and adjust the measurement site 6.1 of the wafer 8 at the intersection point 43 of the dual beam device 1. The dual beam device 1 includes a FIB column 50 having a FIB optical axis 48 and a charged particle beam (CPB) imaging system 40 having an optical axis 42. At the intersection 43 of the optical axes of both the FIB and the CPB imaging system, a wafer surface 55 is disposed at a tilt angle GF relative to the FIB axis 48. The FIB axis 48 and the CPB imaging system axis 42 include an angle GFE. In the coordinate system of FIG. 1, the normal to the wafer surface 55 is given by the z-axis. A focused ion beam (FIB) 51 is generated by the FIB column 50 and is incident on the surface 55 of the wafer 8 under an angle GF. At the inspection site 6.1, an oblique cross section is milled in the wafer by ion beam milling under approximately the tilt angle GF at a predetermined y-position controlled by the stage 155 and the position controller 21. In the embodiment of FIG. 1, the tilt angle GF is about 30°. The actual tilt angle of the tilted cross section may deviate by up to 1°-4° from the tilt angle GF due to beam deflection of the focused ion beam, e.g., a gallium ion beam, or due to variable material properties relative to the milling along the cross section.An image of the milled surface is obtained by a charged particle beam imaging system 40. In the embodiment of Figure 1, the charged particle beam imaging system 40 is positioned with its charged particle beam 44 perpendicular to the wafer surface 55 and parallel to the z-axis. In other configurations, the optical axis 42 of the charged particle beam imaging system 40 is positioned at an angle to the z-axis.

[0060] During imaging, a beam of charged particles 44 is scanned by a scan unit of the charged particle beam imaging system 40 along a scan path over the cross section of the wafer at the measurement site 6.1, generating secondary and backscattered particles. A particle detector 17.1 and an optional internal particle detector 17.2 collect at least a portion of the secondary and / or backscattered particles and communicate particle counts to the control unit 19. Other detectors for other types of interaction products such as x-rays or photons may also be present. The control unit 19 is connected to a control unit 16 for controlling the charged particle beam imaging column 40 and the FIB column 50 and for controlling the position of the wafer mounted on the wafer support table 15 via a wafer stage 155. An operation control unit 2 communicates with the control unit 19, which triggers the positioning and alignment of the wafer 8 at the intersection 43, e.g., at the measurement site 6.1, by wafer stage movement, and iteratively triggers the operations of FIB milling, image acquisition and stage movement. The control unit 19 and the motion control unit 2 include a memory for storing instructions in the form of software code and at least one processor for executing the instructions in operation, e.g. for performing the methods described in the first to fourth embodiments. A memory is further provided for storing digital image data. The motion control unit 2 may further include an interface with a user interface or other communication interface for receiving instructions, prior information and transferring inspection results.

[0061] Each new cross section is milled by the FIB beam 51 and imaged by a charged particle imaging beam 44, for example a scanning electron beam in a SEM or a helium ion beam in a helium ion microscope (HIM).

[0062] The motion control unit 2 is configured to perform a 3D inspection inside the inspection volume 160 in the wafer 8. The motion control unit 2 is further configured to reconstruct properties of the semiconductor structure of interest from the 3D volumetric image. In one embodiment, features and 3D positions of the semiconductor structure of interest, for example, the position of the HAR structure from the HAR centroid, are detected by image processing methods. 3D volumetric image generation, including image processing methods and feature-based alignment, is described in WO2020 / 244795, which is incorporated herein by reference.

[0063] FIG. 2 shows further details of the slicing and imaging method in the wedge cut geometry. By repeating the slicing and imaging method in the wedge cut geometry, a plurality J of cross-sectional averaged image slices are generated, including averaged image slices of cross sections 52, 53.i...53.J, to generate a 3D volumetric image of the inspection volume 160 at the inspection site 6.1 of the wafer 8. FIG. 2 shows the wedge cut geometry in an embodiment of a 3D memory stack. The cross sections 53.1...53.J are milled by the FIB beam 51 at an angle GF of about 30° to the wafer surface 55, although other angles GF are possible, for example between GF=20° and GF=60°. FIG. 2 shows the situation when the plane 52 is the last new cross section milled by the FIB 51. The cross section 52 is scanned, for example, by the SEM beam 44, to generate high-resolution cross-sectional averaged image slices. The cross-sectional averaged image slice includes first cross-sectional image features (e.g., first cross-sectional image features of HAR structures 4.1, 4.2 and 4.3) formed by intersections with high aspect ratio (HAR) structures or vias, and second cross-sectional image features formed by intersections with layers L.1...LM, e.g., including SiO2, SiN or tungsten lines. Some of those lines are also called "word lines". The maximum number M of layers is typically more than 50, e.g., more than 100 or even more than 200. The HAR structures and layers extend through most of the inspection volume in the wafer, but may include gaps. The HAR structures typically have a diameter of about 100 nm or less, e.g., about 80 nm or, e.g., 40 nm. The HAR structures are arranged in a regular, e.g., hexagonal, raster with a pitch of about 300 nm, e.g., even less than 200 nm. Thus, the cross-sectional averaged image slices contain first cross-sectional image features as intersections or cross sections of the HAR structures at different depths (Z) at each XY location. For a cylindrical vertical memory HAR structure, the resulting first cross-sectional image features are circular or elliptical structures at various depths depending on the location of the structures on the tilted cross section 52. The memory stacks extend in the Z direction perpendicular to the wafer surface 55.The thickness d or minimum distance d between two adjacent cross-sectional averaged image slices is typically adjusted to a value of a few nm, e.g., 30 nm, 20 nm, 10 nm, 5 nm, 4 nm or even less. Once a layer of material of a given thickness d has been removed by the FIB, the next cross-sections 53.i...53.J are exposed and accessible for imaging by the charged particle imaging beam 44.

[0064] A number J of cross-sectional averaged image slices thus acquired cover the examination volume of the wafer 8 at the measurement site 6.1 and are used for the formation of a 3D volumetric image with a high 3D resolution, for example less than 10 nm, preferably less than 5 nm. The examination volume 160 (see FIG. 2) typically has a lateral extent in the xy plane of LX=LY=5 μm to 15 μm and a depth LZ of 2 μm to 15 μm below the wafer surface 55. However, these extents can also be larger, for example reaching 50 μm.

[0065] 3 shows an averaged image slice 331.i produced by the imaging charged particle beam 44 and corresponding to the ith cross section 53.i. The averaged image slice 311.i includes an edge line 315 between the tilted cross section and the surface 55 of the wafer at edge coordinate yi. To the right of the edge, the averaged image slice 331.i shows several cross sections 307.1...307.S through the HAR structure that intersect with the cross section 301.i. Additionally, the averaged image slice 331.i includes several word line cross sections 313.1-313.3 at different depths or z positions.

[0066] According to a first embodiment, a method for imaging a stack of M averaged image slices with higher throughput and accuracy is provided. A data stack of M high-quality averaged image slices is determined from the average value of a first plurality N of cross-sectional images. According to the first embodiment, less image scanning time is used for the acquisition of each cross-sectional image, and many cross-sectional images are generated, each of which has a higher noise level. A high-quality averaged image slice is generated by averaging a subset of the cross-sectional images by a convolution kernel. The image acquisition method for a stack of M averaged image slices according to the first embodiment relies on typical characteristics of the semiconductor object of interest, e.g., a HAR channel, which is assumed to vary only slightly in a given direction. By calculating a moving average or average in a given direction of a large number of cross-sectional images, noise is effectively reduced and the time of image acquisition can be shortened. Also, the milling time for generating cross-sections is more effectively used for the generation of more and denser cross-sectional images, in which the distance between the cross-sections is smaller.

[0067] The method according to the first embodiment is less susceptible to errors during the milling and imaging process. A shorter image scan time is used for the acquisition of each cross-sectional image, and many cross-sectional images are generated from which an averaged image slice is calculated. With each new image scan, milling artifacts of the previous cross-sectional image, if any, are removed, and defects or artifacts of the milling operation may vary from cross-sectional area to cross-sectional area. According to the first embodiment of the invention, at least some of these milling artifacts are averaged out by forming a moving average, and polishing of the cross-sectional area is no longer necessary. The depth of field of charged particle beam imaging systems is rather low, making perfect focusing difficult to achieve. By averaging multiple cross-sectional images, focusing defects of the individual cross-sectional images are averaged out, a higher resolution averaged image slice is generated, and time-consuming fine focusing adjustments can be omitted.

[0068] The method according to the first embodiment is illustrated in FIG.

[0069] In step A1, an inspection site 6.i is selected, positioned and held at the intersection 43 of the dual beam device 1. An inspection volume 160 is defined and parameters for the acquisition of a 3D volumetric image are defined. The parameters include, for example, the number of cross sections Q to be milled, the number of cross section images N to be acquired, the size or range B of the 1D convolution kernel and the predefined direction in which the 1D convolution kernel is applied. Further parameters are the dwell time of the image acquisition and other parameters used for the acquisition of the cross section images. The parameters may further include a threshold value for interrupting or stopping the method.

[0070] In step A2, the slicing and imaging method is performed until a threshold is reached, which may be reached when a predetermined control condition is met, such as the occurrence of a particular depth within the semiconductor wafer sample, or a particular layer within a volume of interest within the semiconductor wafer sample.

[0071] In step A2.1, the above number of cross sections are milled by the FIB, and the above number N of cross section images are acquired by the charged particle beam imaging system. The cross section images are stored in a memory P.

[0072] In step A2.3, an iterative evaluation is made as to whether the control volume boundary has been reached or whether the desired information has been collected. If a predefined threshold has been reached, the slicing and imaging method is stopped.

[0073] The results of an example of a HAR channel in a 3D memory stack are shown in Fig. 5. Fig. 5a shows a number of cross sections 53.1-53.N at a tilt angle GF through an examination volume 160 having a number of HAR structures 309. Each cross section intersects with a number of HAR structures having a cross section 307. Fig. 5b shows a schematic image data stack 163 in memory P with a number of cross section images 311.1-311.N, each cross section image 311.i corresponding to an image acquired from one of the N cross sections 53.1-53.N. Each cross section image 311.i is aligned, for example, after a first alignment at a fiducial or alignment mark (not shown) or by a precision stage.

[0074] In step A2.2, a succession of subsets of the cross-sectional images 311.1...N are processed and an averaged image slice is calculated from each subset of the cross-sectional images. During this processing, a running average value is calculated for each subset of the cross-sectional images.

[0075] The formation of M averaged image slices can be described as a mathematical convolution of the 3D data image stack of cross-sectional images with a 1D convolution kernel. In one embodiment, a number N of cross-sectional images form a 3D data set with abscissas x and y for each image, with different z coordinates corresponding to a given direction. The moving average value or mean is represented by a convolution with a 1D convolution kernel over the z coordinates of the 3D data stack. The convolution kernel can be a function over a given number of cross-sectional images, for example a moving rectangular function or a moving Gaussian kernel in the z direction. Other convolution kernels for calculating different weighted moving average point-wise median filters are also possible. Other examples of convolution kernels correspond to filters for finding the minimum of a norm function or matching function according to a given noise model. The width of the convolution kernel can be determined in step A1 or can be adapted according to the local noise level of the averaged image slice or the required SNR. For example, if the noise level is too high, the range B of the convolution kernel can be widened. However, the method is not limited to classical computation of the convolution with a convolution kernel; instead, machine learning based algorithms may be applied to optimally compute the averaged image slices.

[0076] Fig. 6 shows a first embodiment of the method step A2.2 in the embodiment of Fig. 5. For a first subset 165.1 of four cross-sectional images with i=1 to l=4, the average value is calculated and a first averaged image slice 331.1 with j=1 is written into the memory S. The averaging is then continued for a second subset 165.2 of four cross-sectional images with i=2 to i=5 and a second averaged image slice 331.2 with j=2 is written into the memory S. By repeating this method step up to the last subset 165.M, a data stack 167 of averaged image slices is finally generated in the memory S, which comprises averaged image slices 331.j with j=1 to M.

[0077] In step A3, the semiconductor object of interest in the inspection volume is inspected and parameters of the semiconductor object of interest are determined. For example, the semiconductor object of interest is formed by a plurality of respective semiconductor structures, such as the HAR structures shown in FIG. 2. The parameters can be, for example, volume, lateral position, dimensions such as distance or diameter, overlay area, average values ​​of these features, and maximum deviations from the average value of the individual structures. The parameters or features of interest are ultimately attributed to the inspection site and are stored in the memory S of the control unit 2 or written to an inspection file.

[0078] The generation of the data stack 167 of averaged image slices 331.1...M according to step A2.2 can be obtained according to different embodiments. In a first embodiment according to Fig. 6, a moving average is determined, generating a number M of averaged image slices. This operation corresponds to a mathematical convolution in the direction of the index i of the data stack 163 of the cross-sectional images 311.1 to 311.N. The convolution kernel can be a weighted sum of the image intensities of each set of cross-sectional images 311 of given coordinates (x,y). The weights can be equal or different. Examples of convolution kernels are given by rectangular functions (with equal weights) or Gaussian functions. The width of the convolution kernel, i.e. the number B of each subset of cross-sectional images for which the average value is calculated, can be a number greater than B=3, preferably greater than 9 or 11. Larger numbers, for example B of about 50 or more, are also possible. By increasing the range of the average convolution kernel, the time of the fast image acquisition can be further reduced and the throughput can be further increased without increasing the image noise in the averaged image slices.

[0079] Another embodiment is shown in Fig. 7. Here, the number M of averaged image slices 311 is further reduced by performing an averaging operation on every other, every third, or generally every nth subset of cross-sectional images, n=4th in the embodiment of Fig. 7. Here, the range of the average convolution kernel is shown as B=5, and each averaged image slice 331 is calculated from a subset 165.j of B=5 cross-sectional images from the data stack 163. In an embodiment, the number B can be higher. The average convolution kernel in this embodiment still has an overlap of cross-sectional image slices from the data stack 163, such that data from some cross-sectional averaged image slices are used for the calculation of two different averaged image slices 331.j. In such an embodiment, the convolution kernel can be a weighted sum with unequal weights, such as a symmetric triangular function or a Gaussian function.

[0080] Figure 8 shows a further example. Here, the number M of averaged image slices is further reduced, and the subset of cross-sectional images for calculating the averaged image slices no longer shows any overlap.

[0081] Throughout these examples, it is not necessary to acquire the entire data stack 163 of cross-sectional images 311.1...N at once. In one example, the calculation of the averaged image slices 311.1...M is performed in parallel with image acquisition, and it is not necessary to maintain all the cross-sectional images 311.1...N in memory. After the moving average averaged image slice 331.j is calculated, the unnecessary cross-sectional image 311.i can be overwritten by a new cross-sectional image. Thereby, the memory size of the memory P may be limited.

[0082] In one example, the number M of averaged image slices is approximately equal to the number N of cross-sectional images, and the number M of averaged image slices is derived from an averaging convolution kernel having a range B. In another example, the number M is significantly smaller compared to the number N of cross-sectional images, M <= A × N, where the coefficient A >= 3, for example A = 5, A = 7, or A = 20 or more.

[0083] In the above example, the cross-sectional averaged image slice 311.i is obtained from each cross-section 53.i by high-speed scan image acquisition. However, the number N of cross-sectional averaged image slices 311.1...N can also deviate from the number Q of cross-sections 53.1...Q. For example, the number N of cross-sectional images is typically between half the number Q of a plurality of cross-sections 53 and twice the number of cross-sections 53, i.e., 2 × Q => N => Q / 2. For example, high-speed milling is applied and high-speed image acquisition is applied only every other cross-section, so N < Q. For example, each cross-section 53.1...Q is imaged at least once by a high-speed scan operation of the charged particle beam imaging device 40, and at least one cross-sectional image 311.1...N is acquired for each cross-section 53.1...Q, so N > Q.

[0084] Between each subset of cross-sectional images of cross sections at small milling distances, the semiconductor object of interest typically has very limited variation. In one embodiment, a number of cross sections 53.1...Q through the examination volume 160 are milled by the FIB 50 at small z-distances, e.g., less than 5 nm, 3 nm, or even 2 nm. By acquiring a number N of cross-sectional images 311.1...N of different cross sections 53.1...Q formed in the examination volume 160, drift of the stage 155 or the charged particle beam imaging column 40 can be determined from the cross-sectional images 311.1...N and can be effectively removed during the calculation of the moving average or mean value by determining the best fit condition of the subset of cross-sectional images. In one embodiment, the image contrast of the averaged image slice 331.j is evaluated, e.g., by calculating the gradient. If an overall low gradient or low contrast is detected in a particular direction, the determination of the mean value is optimized by including a lateral shift for the cross-sectional images of the subset of cross-sectional images. This reduces the effect of drift in that particular direction.

[0085] During optimization of drift compensation, the maximum expected drift can be bounded to reduce computation time and increase the robustness of the method. Drift can be determined as the absolute drift of each cross-sectional image relative to a reference, or as a relative drift, for example, to the previous cross-sectional image. Different registration or alignment methods can be employed to improve the speed and performance of drift compensation. In certain cases, manual verification of drift compensation is possible, and incorrect image shifts can be discarded.

[0086] According to a first embodiment of the present invention, a large number N of cross-sectional images are obtained with a reduced scan time and therefore an increased noise level. Averaging produces a stack of M averaged image slices, which reduces the increased noise level of the rapidly acquired cross-sectional images and produces averaged image slices with a higher SNR.

[0087] When determining the moving average, global drifts can be detected and compensated for, while individual deviations of the trajectory of, for example, the HAR channel can also be detected. Further alignment and registration of the cross-sectional image at the fiducial marks of the image features is also possible. Calculation of the average image slice can be performed during image acquisition by the slice and image method, thereby reducing the memory requirement for storing a plurality N of cross-sectional images.

[0088] The method according to the first embodiment can further be applied to generate training data for an inspection method using machine learning. Annotations from the averaged image slices with low noise levels can be transferred to a set of cross-sectional images used for averaging, each having a large noise level and acquired by a high-speed image scan. Thus, it is possible to provide annotations with high precision in the high-quality averaged image slices, and to transfer the annotation information to the set of cross-sectional images in order to train a machine learning algorithm to extract information from the cross-sectional images with a large noise level acquired by the high-speed image scan.

[0089] A second example further reduces the effects of drift. According to the second embodiment, image acquisition of N cross-sectional images and milling of N cross-sections are performed simultaneously in parallel. In one example, milling with the FIB is performed by scanning the FIB beam 51 in a first direction (e.g., x-direction). At the same time, image acquisition is performed by scanning the imaging charged particle beam 44 across the cross-section in the same x-direction and stepping line-scan by line-scan in the y-direction. FIG. 9 shows interlaced milling and image acquisition at three different times during the milling operation. At a first time t1, only a small portion of the cross-section 53 is milled by the FIB 51, and near the end of the milling operation, an image scan 57 is started by scanning in the x-direction and stepping in the y-direction. Thus, the image scan at time t1 is performed on a significant portion of the previous cross-section 61 before milling. At a second time t2, about half of the next cross section 63 is milled, and after milling, an image scan 57 is performed on the previous cross section 61 and the newly milled cross section 63. At a time t3, milling by the FIB 51 is nearly complete. This results in each cross section image 311 of each interlaced milling and imaging operation containing a first region 61 according to the cross section before the actual milling operation and a second region 63 according to the cross section after the actual milling operation. These regions at different z positions are separated by a line 67. Each cross section image 311 therefore contains image regions at different z positions. The multiple N cross section images again form a 3D data stack with abscissas x and y of each image and different z coordinates corresponding to the two regions in each cross section image. This allows the effects of drift between the different cross section images to be further reduced. The method according to the second embodiment can be combined with the method according to the first embodiment to form a stack of M averaged image slices by taking a moving average or mean as described in the first embodiment and indicated by the same reference numbers in Fig. 10. An example is shown in Fig. 10. The moving average or mean value of the averaged image slices 331.m of the data stack 167 of averaged image slices is formed from the subset 165.m of cross-sectional images of the data stack 163 of cross-sectional images 311.1...N.According to one embodiment, the cross-sectional image subset 165.m also includes only those portions of the cross-sectional images 311.1...N that lie within a particular z-range of the convolution kernel for the calculation of a particular averaged image slice 331.m.

[0090] According to one embodiment, further cross-sectional images can be selectively added, for example by repeated image scanning. For example, a first set of cross-sectional images can be acquired in parallel with the milling operation as described in the second embodiment, and a second set of cross-sectional images can be acquired between two subsequent milling operations.

[0091] According to a third embodiment of the invention, the throughput of a method for inspection of a 3D semiconductor object of interest within an inspection volume of a wafer sample can be further increased by limiting the time consuming performance of image scans to a selected region of interest. A method according to the third embodiment is illustrated in FIG.

[0092] Step A11 follows step A1 according to the first embodiment. In addition to step A1 of the first embodiment, a volume of interest within an inspection volume of a wafer sample is selected. The selection can be made by external input, for example from CAD data, or by user input, for example from a graphical user interface. For example, a user selects a volume of interest that includes at least one semiconductor object of interest. Step A11 can further include adjusting the distance of the cross section according to the selected volume of interest.

[0093] In step A12, a slicing and imaging method is performed until a threshold is reached.

[0094] In optional step A12.0, a first cross-sectional image is acquired and the intersection of the volume of interest with the first cross-sectional image is determined, thereby registering the location of the volume of interest.

[0095] Some "fiducial" structures of the semiconductor object in the first cross-sectional image can be used for determining and selecting the volume of interest, such as the number of word lines or other layers (see FIG. 2) or other significant features. Determining the volume of interest can also include identifying specific features (e.g., transistors, contact spots, intersections of two or more elements) using machine learning detection. Determining may also be performed by using fiducial or alignment marks.

[0096] In step A12.1, the next cross section is milled by the FIB and the region of interest is determined according to the intersection of the cross section with the volume of interest.

[0097] In step A12.2, a cross-sectional image is obtained by image scanning of the region of interest by a charged particle beam imaging system. The cross-sectional image of the region of interest in the volume of interest is analyzed, and optionally the volume of interest is adapted for a next image acquisition step. Optionally, the image scanning of the cross-sectional image of the modified region of interest in the actual cross section is repeated. The final cross-sectional image of the region of interest in the actual cross section is stored in memory S.

[0098] In optional step A12.3, an assessment is made as to whether the boundaries of the volume of interest have been reached or whether the desired information has been collected: if a predefined threshold has been reached, the slicing and imaging method is stopped.

[0099] Step A13 follows step A3 according to the first embodiment, with the difference that the inspection of the semiconductor object of interest and the determination of the parameters is limited to a selected volume of interest.

[0100] An example of limited image acquisition is shown diagrammatically in FIG. 12. A number of cross sections 53.i are shown within an inspection volume 160, which are subsequently milled by the FIB under an angle GF with respect to the surface 55 of the wafer. The inspection volume has an extent LX×LY×LZ. Within the inspection volume 160, a volume of interest 171 is determined. In this simplified example, the volume of interest has an extent LX×LYI×LZ. In this example, the charged particle imaging beam 44 of the imaging system is arranged perpendicular to the cross sections 53 (or perpendicular to the FIB axis 48 (see FIG. 1)). This allows a higher resolution to be achieved during imaging. The acquisition of the cross-sectional images by the charged particle imaging beam 44 is limited to a region of interest 173.i of each cross section 53.i (only the first region of interest 173.1 is shown). The extent of the region of interest in the y-direction is thereby limited to LYI, which is significantly smaller than L. The image scan of the charged particle imaging beam 44 is adjusted for each cross section 53.i to shorten the time interval for image acquisition.

[0101] FIG. 13a shows an example of an embedded volume of interest for the inspection of HAR channels. Here, the volume of interest is limited by two z-planes z1 and z2 and is limited to include three HAR channels. FIG. 13b shows a top view of five consecutive cross sections 53 and their corresponding regions of interest 173.1-173.5. The time interval for image acquisition is reduced. In one embodiment, lateral drift may occur, for example during the milling of cross sections 53.3 and 53.4, which is not a problem since the region of interest 173.4 can be adjusted accordingly to compensate for any lateral drift.

[0102] 14 shows an example of frequent or dynamic adjustment of the region of interest 173 to cover the volume of interest 171 inside the volume of interest 160. The volume of interest in this example is dynamically adjusted in step A12.2 to cover the semiconductor object of interest, for example the HAR channel 309. A first region of interest 173.1 (here a cross section through the HAR channel 309) is determined to cover the semiconductor object of interest in the first cross section 53.1. The region of interest 173.1 is selected to include a cross section through the HAR channel 309 with a certain boundary area to cover the expected lateral position change of the HAR channel 309. A second region of interest 173.2 is then selected according to the first region of interest 173.1, and image slices of the second cross section 53.2 are analyzed, for example by image processing or pattern recognition methods. In this embodiment, the slice through the HAR channel 309 is no longer at the center of the second region of interest 173.2, and the subsequent region of interest 173.3 is adaptively changed to a larger extent and a new center position at the center of the slice through the HAR channel 309 in the region of interest 173.2. After analyzing the slice through the HAR channel 309 within the image acquired from the region of interest 173.3, if the slice through the HAR channel 309 in slice 53.3 has not changed relative to the slice through the HAR channel 309 in slice 53.2, the size of the subsequent region of interest 173.3 is reduced again. By iteratively adjusting the size or position of the region of interest 173 slice by slice, the time-consuming operation of scan image acquisition can be reduced.

[0103] 15 shows another example of the third embodiment using two volumes of interest 171.1 and 171.2. In some applications, the relevant information extracted by the inspection operation is the spatial relationship of the two semiconductor objects of interest 309.1 and 309.2 in the inspection volume 160. In this example, a first averaged image slice 175.1 of a first region of interest in the first volume of interest 171.1 and a second averaged image slice 175.2 of a second region of interest in the second volume of interest 171.2 are generated to determine the spatial relationship between the semiconductor objects of interest 309.1 and 309.2. In this example, the spatial relationship is the alignment or overlay accuracy dy of the two HAR channels 309.1 and 309.2.

[0104] According to one embodiment, the method according to the third embodiment comprises a further step A14 (see FIG. 11 ), whereby the measurement result is displayed to a user via a user display: the measurement image data stack is displayed inside the examination volume, and missing image information from outside the volume of interest is augmented, for example by CAD information.

[0105] According to a third embodiment, a higher throughput inspection method is provided, including automatic adjustment of milling and imaging parameters during the slicing and imaging method, which allows obtaining a data stack of image slices optimized for a specific inspection task, for example, the search for a specific defect, focused on a specific IC element or part thereof, or on a specific location within a semiconductor wafer. The adjustment of the imaging parameters includes limiting the scan image acquisition to a cross-sectional section within at least one volume of interest, while faster milling can be used to form a larger cross-section. In one embodiment, image acquisition is performed with high resolution and long dwell time within the volume of interest, and image acquisition continues with low resolution and short dwell outside the volume of interest. This generates a low-resolution overview image of the inspection volume, including a high-resolution volumetric image of the volume of interest. The low-resolution overview image can be supplemented by augmentation information according to step A14.

[0106] The third embodiment can be combined with the method according to the first and / or second embodiment. This provides an inspection method with even higher throughput. To reduce the time interval between image acquisitions according to the third embodiment, it is useful to restrict the image acquisition to a small volume of interest within the inspection volume. However, in some implementations, such a restriction of the volume of interest is not possible. The method according to the fourth embodiment of the invention overcomes the problem of volumetric image acquisition of large volumes of interest.

[0107] According to a fourth embodiment, a method of sparse image acquisition is applied, which is illustrated in FIG.

[0108] Step A21 follows step A11 according to the third embodiment. In addition to step A11, step A21 further includes selecting the density of the sparse images to be acquired. For example, the sparse image acquisition can be performed for only less than 10% of the cross-sections in the examination volume. For example, the sparse image acquisition can be performed for about 30% or 50% of the cross-sections in the examination volume.

[0109] In step A22, a slicing and imaging method is performed until a threshold is reached.

[0110] In step A22.1, the first or next cross section is milled by FIB.

[0111] In step A22.2, a first or next plurality of sparse image regions are acquired from the cross section milled in step A22.1 for sparse image acquisition. The sparse images are stored in a memory S.

[0112] In step A22.3, the image regions are registered and a plurality of sparse image regions are determined. The determination can be obtained, for example, from the registered image regions such that an average distribution of the sparse image regions over the examination volume can be maintained. This maintains the coverage of the examination volume with the image regions. The sparse image regions can also be determined from prior information, for example CAD information, and can be determined to cover special features of interest within the examination volume.

[0113] In step A24, a 3D volumetric image of the inspection volume is calculated from the sparse image data obtained in step A22. The calculation can be performed by mathematical methods or by machine learning methods that are trained to fill in missing data to complete the sparse image data. The mathematical methods can include interpolation, extrapolation or averaging. In another embodiment, the missing data to complete the sparse image data is obtained from CAD information. For example, the 3D volumetric image can be derived by fitting a parametric model description of the semiconductor object of interest to the sparse image data.

[0114] Step A23 follows step A3 according to the first embodiment.

[0115] The fourth embodiment can be combined with the methods according to the first to third embodiments. Figure 17a shows an example of the method according to the fourth embodiment. Inside the inspection volume 160, a large volume of interest 171 is defined. A number of cross sections 53 (only one is shown) are milled. Figure 17b shows five cross sections 53.1 to 53.5 in the horizontal xy coordinates. For each cross section, a different set of sparse imaging areas 181 is defined and sparse image sections are acquired by a charged particle imaging beam. From this different distribution of sparse image sections a 3D volumetric image of the volume of interest is extrapolated as described above in Figure A24.

[0116] One drawback of the slice and image method is the destruction of the wafer sample inside the examination volume 160. After a cross section is milled by removing material above the cross section, it is not possible to acquire additional images of the removed area within the examination volume. This can be a challenge in some examples of embodiments of the present invention. A fifth embodiment of the present invention overcomes some of these problems of high speed image acquisition by using a multi-mode imaging technique.

[0117] The method of the fifth embodiment employs a multi-mode image acquisition comprising at least a first and a second imaging operation mode of a charged particle beam imaging system for imaging of a data stack of cross-sectional images or of averaged image slices. Images acquired by different imaging modes can be normalized and an averaged image slice can be calculated as described in the first embodiment of the invention. In another example, different detector characteristics can result in different sparse image areas according to the fourth embodiment, and the signals acquired by the detectors are adjusted to different semiconductor features. Images acquired by different imaging modes can complement each other.

[0118] According to a first embodiment, the detector characteristics are adjusted for different imaging operating modes. For example, in a first imaging operating mode, the detector gain is adjusted to a low signal level, and in a second imaging operating mode, the detector gain is adjusted to a high signal level. According to a further embodiment, the cut-off energy of secondary or backscattered electrons is adjusted. This allows, for example, imaging to be reduced to a certain depth or to a certain material under the cross section.

[0119] In one embodiment, the method includes changing the detection mode between subsequent image scanning operations, the change in detection mode including at least one of changing the dynamic range, changing the energy range of the interaction product, or changing the type of the interaction product.

[0120] In one embodiment, the accuracy of the measurement can be improved by expanding the dynamic range of the image sample by acquiring a series of images in the same field of view. Each image in the series is acquired with a different detector setting of a single detector. For example, at a given brightness / contrast (or gain / offset) of the detector, a certain range of charged particle counts can be measured by the detector. The charged particle counts are converted by an analog-to-digital converter into, for example, an 8-bit format. This causes weak signal differences to be lost due to conversion losses. On the other hand, if the detector gain is too high, a strong signal will cause the detector to overflow, and similarly, strong signal differences will be lost. To expand the range of minimum and maximum electron counts of the detector, the offset and gain can be adjusted so that different ranges of signals can be detected with high resolution. This achieves an expansion of the signal resolution of the SEM.

[0121] According to a second embodiment, different imaging modes are realized in parallel by using spatially separated or different detectors. For example, the first and second detectors can be arranged at different angular intervals and configured to determine topographic contrast. For example, the first detector can be configured to collect large angle scattered electrons and the second, in-lens detector can be configured to collect small angle backscattered electrons. For example, the first detector can be an electron detector and the second detector can be an X-ray detector. For example, the first detector can be configured to detect low energy interaction products and the second detector can be configured to detect high energy interaction products. The interaction products can be secondary electrons or backscattered charged particles in a charged particle beam imaging system. The second embodiment using the first and second detectors further improves the desired information and imaging accuracy regarding the semiconductor object of interest without reducing the throughput improvement according to the first to fourth embodiments.

[0122] Multimodal image acquisition further improves imaging of the stack of M averaged image slices by providing additional information. The method according to the fifth embodiment can be combined with any of the methods according to the first to fourth embodiments.

[0123] In a sixth embodiment of the invention, an inspection system for 3D inspection of a semiconductor object of interest is described. In FIG. 1 a first example of an inspection system 1000 for 3D inspection in an inspection volume of a wafer sample 8 is shown. The inspection system 1000 is configured for performing a method according to any of the first to fourth embodiments. The inspection system 1000 comprises a wafer sample holder 15 and a stage 155 for holding and positioning the wafer sample 8. The inspection system comprises a dual beam device 1 comprising a FIB column 50 and a charged particle beam imaging system 40, e.g. a scanning electron microscope (SEM) or a helium ion microscope (HIM). The dual beam device 1 further comprises at least one detector 17.1 for secondary or backscattered charged particles. In an example, the dual beam device 1 further comprises at least a second in-lens detector 17.2 for acquiring a second signal in a multimode imaging operation according to the fifth embodiment. The inspection system further comprises a control unit 19 for controlling the operation of the dual beam device 1. The control unit 19 is configured to give instructions to the scan unit of the charged particle beam imaging system 40. This allows the image scanning operation to be limited to a region of interest or a sparsely imaged region according to the third or fourth embodiment of the present invention. The inspection system 1000 further comprises a motion control unit 2 having a software program code, a memory for storing the motion instruction program code and a memory for storing the acquired digital image data received from the detectors 17.1 and 17.2. The motion control unit 2 further comprises a processor configured to execute processing instructions for the calculation of the moving average. The motion control unit 2 of the inspection system 1000 is further configured to receive prior information, e.g. CAD information, via an interface. Further details of the inspection system 1000 of FIG. 1 are described above.

[0124] FIG. 18 shows a second embodiment of an inspection system 1000 for 3D inspection in an inspection volume of a wafer sample 8. The same reference numbers as in the description of FIG. 1 are used and reference is made to the description of FIG. 1. In the second embodiment, a charged particle beam imaging system 40 is formed by a corrected electron microscope with a number of detectors 17.1 and 17.3 for the application of angle-resolved imaging of interaction products. Such a corrected electron microscope is described in PCT / EP2022 / 079753, filed on 25 October 2022, and incorporated herein by reference. The corrected electron microscope comprises an electron source 1301, a collimation system 1405, a beam splitter 1500 and a mirror element 1415. The mirror element 1415 corrects chromatic aberration as well as spherical aberration and field curvature. The corrected electron microscope 40 further comprises an in-lens detector 17.2 and an objective lens 1102 with a deflection scanner 35. Backscattered and secondary electrons 9 are collected by the objective lens 1102 and directed at least in part via a beam splitter to the projection system 1605 and further to the electron detectors 17.1 and 17.3.

[0125] The FIB column 50 and the charged particle beam imaging system 40 of the dual beam device 1 are disposed at an angle to form an intersection point 43 of the optical axis 48 and of the column. A stage 155 is configured to position and hold a wafer sample 8 containing an inspection volume proximate the intersection point 43.

[0126] According to an embodiment, the charged particle imaging system of the dual beam device is configured to implement different imaging modes according to the fifth embodiment. The different imaging modes can include different detectors or detectors configured for different detector settings. The detector can include a grid 1607 with a variable grid voltage. This allows the detector settings to be changed by the selection of the repulsion field to cut off low energy charged interaction products. The detector can include a variable analog gain factor set for the adjustment of the analog signal before conversion to a digital signal. This allows weak signals to be acquired with higher precision and definition. The detector can include several detector elements 17.1, 17.3 for angle resolved detection of the interaction products. The charged particle imaging system 40 formed by the corrected electron microscope allows inspection with higher resolution at lower electron voltages below 1 keV, for example below 500 eV or even below 300 eV.

[0127] The charged particle imaging system 40 according to the sixth embodiment includes a scan detector 35 and a scan controller 23 configured to perform different image scanning operations, including, for example, line averaging, changing dwell time, changing the scan path, and performing, for example, a zigzag or serpentine scan method. The scan controller 23 is further configured to perform different image scanning operations according to the first to fourth embodiments of the present invention. The different scanning configurations include a high-speed image scan according to the first embodiment, an image scanning operation perpendicular to the FIB beam 51 according to the second embodiment, a limited scan operation limited to at least one region of interest according to the third embodiment, or a limited scan operation configured for image scanning of a different set of sparse imaging regions 181 according to the fourth embodiment.

[0128] One example of a charged particle imaging system 40 configured to implement different imaging modes according to the fifth embodiment includes a charged particle source 1301 and particle optical elements 1405 and 1102 configured for changing the acceleration voltage, beam current, charged particle beam angle, numerical aperture of the charged particle beam, or cross-sectional shape of the charged particle beam.

[0129] Throughout this specification, the invention is described with respect to inspection within a test volume of a wafer sample. The wafer sample 8 can be formed by a wafer, for example a 300 mm wafer, or by a small wafer sample piece extracted from the wafer.

[0130] A typical acquisition of an examination volume with a lateral extent of about LX=10 μm×LY=10 μm and a depth of LZ=10 μM requires, for example, slicing and imaging 2000 cross-sectional image slices with a dwell time of about 2 μs per pixel. Thus, a typical time for the acquisition of a 3D volumetric image with a high lateral resolution of about 2 nm may exceed 24 hours. Typically, the image acquisition time is in the range of 12 hours and 30 hours. On the other hand, the milling time required for such an examination volume is about 1 hour or less and is almost independent of the number of cross sections to be milled. The method according to the embodiment of the present invention allows a significant reduction in the time for image acquisition, for example at least by a factor of 2 or less. For example, the dwell time is reduced by a factor of 4 and the SNR of the cross-sectional image slices is increased by averaging a set of cross-sectional images including at least four rapidly scanned cross-sectional images. Even if the number of cross-sectional images is increased, for example by a factor of 2, the total image acquisition time is reduced by a factor of 2. The dwell time per pixel can be further reduced, for example to less than 0.1 μs, and the averaging kernel can be increased to cover, for example, 25 or 30 cross-sectional images. Assuming a semiconductor object of interest that changes slowly in a given direction (e.g., the z-direction perpendicular to the wafer surface), a small loss of resolution in that given direction due to averaging is often acceptable, resulting in an averaged image slice with a high SNR. This allows for a significant reduction in image acquisition time, for example by a factor of 5, 8 or even more. The noise level is indirectly proportional to the dwell time (i.e., the square root of the dwell time), and averaging a set of cross-sectional images achieves at least a similar SNR to the image slice with a longer dwell time. Also, a fast image scan with a reduced dwell time per pixel is less sensitive to charging effects, thus further enhancing the SNR of the averaged image slice.

[0131] According to one embodiment, a volume of interest is determined and image acquisition is limited to the region of interest within each cross-section. According to another embodiment, image acquisition is performed only in sparse regions or sections of the cross-section. According to both embodiments, the time for image acquisition of high quality cross-sectional image slices is reduced by a factor of two or more. In one embodiment, fast image acquisition and averaging is combined with the limited region of interest to further reduce the image acquisition time.

[0132] As mentioned above, image registration or alignment of the multiple images to calculate the average image is performed, for example, at a fiducial mark or alignment feature. Such alignment features can be two-dimensional structures prefabricated by deposition of a metal layer and charged particle beam induced sputtering of alignment features in the metal layer. Such alignment features are typically in the much larger range of line widths, for example 20 nm or more, for example 50 nm, 100 nm or more. Other examples of alignment features include existing features present on the semiconductor wafer, such as alignment markers. Other examples of alignment features include three-dimensional alignment features, such as line structures, crosses or holes milled in the wafer.

[0133] Such larger ranges do not require high resolution image acquisition, as compared to semiconductor features with dimensions of less than 10 nm or less, e.g., 5 nm, which typically require image resolution or sampling distances of 2 nm or less, e.g., 1 nm. Between each acquisition of cross-sectional images, drift can cause lateral displacement between successive images. Even with coarsely aligned features such as those described above, the lateral displacement from image to image can be determined with high accuracy of less than 1 nm without sacrificing the speed of image acquisition.

[0134] In a first example of the first and second embodiments, an image of the alignment reference mark is obtained with the same high resolution as the semiconductor feature in each rapidly acquired cross-sectional image with a larger noise level. In order to precisely calculate the center position of the reference mark, the pixels covering the coarse reference mark can be averaged by a lateral convolution kernel. Thus, each image is locally averaged in the image area containing the image data of the alignment reference mark, and the reference position of the image is obtained with high precision from the locally averaged image data of the alignment reference mark. The lateral averaging reduces the image noise so that, for example, the center position of the alignment reference mark can be obtained with high precision. Then, an averaged image of the semiconductor feature is obtained by averaging several images with high lateral overlay precision.

[0135] In a second example according to the third embodiment, fast imaging is obtained by limiting the image acquisition of the region of interest. In a third example, multiple sparse image sections of the semiconductor feature of interest are obtained within each image scan. In addition to the image acquisition of the region of interest or sparse image area, an image area including the alignment reference mark can be included within each image acquisition. This allows the image of the region of interest or sparse image area to be aligned laterally with high precision. In one example, the speed of image acquisition is further improved by shortening the image acquisition time of the image section of the alignment reference mark. As mentioned above, the alignment reference mark is typically large compared to the semiconductor feature and does not require high resolution. Due to the statistical nature of the noise, the time average and the ensemble average are the same. Thus, a low resolution image of the alignment reference mark can be obtained by changing the scan image acquisition of the alignment reference mark to a lower resolution, by fast scanning and lateral pixel binning or averaging as described above, or by a lower scan frequency or averaging of several low resolution images of the alignment reference mark.

[0136] The present invention described by the embodiments can be explained by the following clauses.

[0137] Clause 1: A method for volumetric inspection of a 3D semiconductor object of interest in the inspection volume of a wafer sample, comprising: - obtaining a first number N of cross-sectional images by high-speed milling and high-speed image scanning of a plurality of cross-sections through the inspection volume; and - forming a second number M of averaged image slices from the first number N of cross-sectional images by calculating a moving average value of the semiconductor object of interest in a predetermined direction, wherein the moving average value of each averaged image slice is calculated from a subset of the first number N of cross-sectional images.

[0138] Clause 2: The method according to Clause 1, wherein the second number M is smaller than the first number N, M < N, preferably M <= A × N with a coefficient A >= 3, for example A between 3 and 11, A = 20 or even A of about 50. Thus, for example, any of the relationships 3 ≦ A ≦ 11 or 3 ≦ A ≦ 20 or 45 ≦ A ≦ 55 may hold.

[0139] Clause 3: The method according to Clause 1, wherein the second number M is approximately equal to the first number N. Preferably, M < N and / or M / N <= 0.95 or M / N <= 0.98 or M / N <= 0.99.

[0140] Clause 4: The method according to any one of Clauses 1 to 3, comprising performing high-speed milling of a new cross-section and high-speed image scanning for subsequent cross-section acquisition at different times.

[0141] Clause 5: The method according to any one of Clauses 1 to 3, comprising simultaneously and in parallel performing high-speed milling of a new cross-section and high-speed image scanning for cross-section image acquisition in a scanning direction of a charged particle imaging beam perpendicular to the direction of a focused ion beam (FIB), wherein the cross-section image includes a first region corresponding to a first cross-section before the actual high-speed milling operation and a second region corresponding to a second deeper cross-section according to the actual high-speed milling operation.

[0142] Clause 6: A method according to any of clauses 1 to 5, wherein the semiconductor object of interest is at least one HAR channel whose predetermined direction is oriented perpendicular to a top surface of the wafer sample.

[0143] Clause 7: The method of any of clauses 1 to 6, comprising performing high speed milling at an inclination angle relative to a top surface of the sample, wherein the inclination angle GF is 45° or less, such as 30° or less.

[0144] Clause 8: A method according to any one of clauses 1 to 7, comprising calculating a moving average value of a second number of cross-sectional images by convolving the cross-sectional images with a one-dimensional convolution kernel parallel to the predetermined direction.

[0145] Clause 9: The method according to clause 8, wherein the one-dimensional convolution kernel spans at least B=3, preferably between B=3 and B=20, consecutive cross-sectional images and has the form of a weighted sum, a rectangular function or a Gaussian distribution function. Preferably, the relationship 3≦B≦20 holds.

[0146] Clause 10: A method as described in clause 8 or 9, comprising selecting a range B of the one-dimensional convolution kernel according to the noise level in the cross-sectional image so as to obtain a predetermined signal-to-noise ratio (SNR) level in the averaged image slice, with a larger range B, e.g. B=11 for higher noise levels, B=20 or more.

[0147] Clause 11: A method according to any one of clauses 1 to 10, comprising compensating for lateral drift of a cross-sectional image among the plurality of cross-sectional images when calculating the moving average value, wherein the direction of the lateral drift is perpendicular to the predetermined direction.

[0148] Clause 12: A method as described in clause 11, comprising determining lateral drift of cross-sectional images in an averaged image slice with low gradient or low contrast in the drift direction, and optimizing calculation of the averaged image slice with maximum isotropic contrast or gradient by averaging and including the lateral shift for at least one cross-sectional image.

[0149] Clause 13: The method of clause 11, further comprising determining lateral drift from a reference position obtained from an image segment of the alignment reference mark.

[0150] Clause 14: The method of clause 13, further comprising averaging laterally by pixel binning or horizontal convolution of image segments of the alignment reference mark with an average convolution kernel.

[0151] Clause 15: A method according to any of clauses 1 to 14, wherein the method further comprises a step of adjusting a distance between two adjacent cross-sections according to prior information of the semiconductor object of interest in at least the first volume of interest.

[0152] Clause 16: A method according to any of clauses 1 to 15, further comprising adjusting a milling angle of a subsequently milled cross-section according to prior information of the semiconductor object of interest within at least the first volume of interest.

[0153] Clause 17: A method according to any of clauses 1 to 16, further comprising a step of changing the detection mode between subsequent high-speed image scans, wherein the change in detection mode comprises at least one of changing the dynamic range, changing the energy range of the interaction product, or changing the type of the interaction product.

[0154] Clause 18: A method for inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample, comprising: - defining at least a first volume of interest within a control volume of the semiconductor wafer, the first volume of interest being within the control volume; - milling a plurality of cross sections through the control volume; - acquiring several cross-sectional images by performing an image scanning operation of cross-sectional sections within at least a first volume of interest, each cross-sectional image section being determined according to a region of interest formed by an intersection of a cross section with at least the first volume of interest.

[0155] Clause 19: The method of clause 18, comprising milling at an oblique angle relative to the wafer surface, wherein at least the first volume of interest is oriented perpendicular to the wafer surface.

[0156] Clause 20: A method according to clause 18 or 19, comprising defining at least a first volume of interest in accordance with CAD data.

[0157] Clause 21: A method according to any of clauses 18 to 20, further comprising a step of aligning at least a first volume of interest within the inspection volume according to at least a first image scan of a first cross section within the inspection volume.

[0158] Clause 22: A method as described in clause 21, wherein the registering step includes further re-registration of at least the first volume of interest in at least second or further image scans of second and further cross sections within the examination volume.

[0159] Clause 23: The method according to any one of clauses 18 to 19, wherein the step of defining at least a first volume of interest comprises: a) acquiring a first image scan of a first cross-section; b) selecting a first cross-sectional image section in the first image scan, the first cross-sectional image section including a cross-section of the semiconductor object of interest; c) milling a second cross-section into the control volume of the wafer sample; d) selecting a second cross-sectional image section by projecting the first cross-sectional image section onto a second cross-section; e) acquiring a second cross-sectional image section of the second cross-sectional section; and f) repeating steps c) through e) by milling additional cross sections and selecting and acquiring additional cross section image sections until a predetermined abort criterion is met.

[0160] Clause 24: A method according to clause 23, wherein in step e) projecting the first cross-sectional image section onto the second cross-section is performed in a predetermined direction.

[0161] Clause 25: A method as described in clause 23 or 24, further comprising a step of adjusting the position and size of the first cross-sectional image section according to the second cross-sectional image section, wherein the position and size of the cross-section of the semiconductor object of interest in the second cross-sectional image section are analyzed by image processing.

[0162] Clause 26: The method of any of clauses 18 to 25, wherein the semiconductor object of interest is a HAR channel and the predetermined direction is oriented perpendicular to the wafer surface.

[0163] Clause 27: A method according to any of clauses 18 to 26, further comprising the step of generating a 3D volumetric image of the 3D semiconductor object of interest.

[0164] Clause 28: A method as described in clause 27, wherein the step of generating a 3D volumetric image of the inspection volume includes the step of generating a 3D volumetric image of at least a first volume of interest, and the step of generating augmented 3D image data outside of at least the first volume of interest.

[0165] Clause 29: A method according to clause 28, wherein the augmented 3D image data is obtained from CAD data.

[0166] Clause 30: The method according to clauses 18 to 29, further comprising adjusting a distance between two adjacent cross-sections according to prior information of the semiconductor object of interest in at least the first volume of interest.

[0167] Clause 31: The method according to clauses 18 to 30, further comprising the step of forming a number of image slice segments from the plurality of cross-sectional image segments by obtaining a moving average in a predetermined direction.

[0168] Clause 32: The method described in clauses 18 to 31, further comprising a step of changing the detection mode between two subsequent image scanning operations, wherein the change in the detection mode includes at least one of changing the dynamic range, changing the energy range of the interaction product, or changing the type of the interaction product.

[0169] Clause 33: A method for inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample, comprising: - milling a plurality of cross sections through the control volume; - selecting multiple sparse regions for each cross section; - acquiring a plurality of sparse cross-sectional image segments by performing an image scanning operation of a plurality of sparse regions of each cross-section; - generating a 3D volumetric image of the examination volume from the plurality of sparse cross-sectional image segments.

[0170] Clause 34: A method as described in clause 33, wherein the step of generating a 3D volumetric image includes image interpolation between a plurality of sparse cross-sectional image segments.

[0171] Clause 35: A method as described in clause 34, wherein image interpolation between the plurality of sparse cross-sectional image sections is based on CAD data of the 3D semiconductor object of interest.

[0172] Clause 36: The method of clause 33, wherein the selection of the plurality of sparse regions for each cross section varies from cross section to cross section.

[0173] Clause 37: A method according to any one of clauses 33 to 36, wherein the selection of multiple sparse regions of each cross section is performed according to image analysis of multiple previously acquired sparse cross section image segments.

[0174] Clause 38: A method according to any one of clauses 33 to 37, wherein the selection of the plurality of sparse regions is performed in accordance with CAD data.

[0175] Clause 39: The method according to any one of clauses 33 to 38, wherein the milling is performed at an oblique angle relative to the wafer surface.

[0176] Clause 40: The method according to any of clauses 33 to 39, wherein the method further comprises adjusting the distance between two adjacent cross-sections according to prior information of the semiconductor object of interest.

[0177] Clause 41: A method according to any of clauses 33 to 40, further comprising a step of changing a detection mode between two image scanning operations, wherein the change of detection mode includes at least one of a change of dynamic range, a change of energy range of the interaction product, or a change of type of the interaction product.

[0178] Clause 42: The method of any of clauses 33 to 41, further comprising compensating for lateral drift of each of the plurality of sparse regions of each cross-section.

[0179] Clause 43: A method as claimed in clause 42, comprising determining lateral drift from a reference position obtained from an image segment of the alignment reference mark.

[0180] Clause 44: The method of clause 43, further comprising obtaining an image segment of the alignment reference mark at a lower resolution.

[0181] Clause 45: A method for inspection of a 3D semiconductor object of interest in an inspection volume of a sample wafer, comprising: - milling a plurality of cross sections through the control volume; - scanning each cross-section with an imaging charged particle beam, during which interaction products including secondary electrons, backscattered charged particles, photons or x-rays of different energy ranges are generated; - forming a plurality of cross-sectional images from detection signals of at least the first and second interaction products.

[0182] Clause 46. The method according to clause 45, wherein the first interaction product and the second interaction product are a) a first and a second energy range of selected interaction products, or - b) first and second dynamic ranges of selected interaction products, or - c) a first and a second type of interaction product; The method is selected by either

[0183] Clause 47: A method according to clause 45 or 46, wherein in the step of forming a plurality of cross-sectional images, the cross-sectional image obtained from the first detector is augmented by information received from the second detector.

[0184] Clause 48: A method according to clause 45 or 46, wherein in the step of forming a plurality of cross-sectional images, the cross-sectional image obtained from a first image scan is augmented by information received from a second image scan.

[0185] Clause 49: A semiconductor inspection device for inspection of a 3D semiconductor object of interest in an inspection volume of a sample wafer, comprising: - a wafer sample stage for holding and positioning the wafer sample; - a dual column microscope including a focused ion beam (FIB) column and a charged particle beam imaging column forming a common intersection of an optical axis of the FIB column and an optical axis of the charged particle beam imaging column; - a control unit configured to control the dual column microscope to perform a slicing and imaging method in the examination volume; - a processor and memory having software code installed thereon, the processor configured to perform any of the method steps described above; 13. A semiconductor inspection device comprising:

[0186] Clause 50: A semiconductor inspection device as described in clause 49, further comprising at least first and second detectors configured to perform multi-mode image acquisition.

[0187] Clause 51: A semiconductor inspection device according to clause 49 or 50, wherein the charged particle beam imaging column is formed by a corrected electron microscope including correction means for the correction of chromatic and spherical aberrations.

[0188] However, the present invention described by the examples and embodiments is not limited to the above clauses, and can be implemented by those skilled in the art through various combinations or modifications thereof.

[0189] A list of reference numbers is provided. 1 Dual Beam Device 2. Motion Control Unit 4 Cross-sectional image features 6. Measurement Site 8 Wafer Sample 15 Wafer support table 16 Stage Control Unit 17 Secondary or backscattered electron detector 19 Control Unit 21 Position Sensor 23 Scan Controller 40 Charged Particle Beam (CPB) Imaging System 42 Optical axis of the imaging system 43 Intersection 44 Imaging Charged Particle Beams 48 FIB optical axis 50 FIB columns 51 Focused Ion Beam 52 Section 53 Section 55 Wafer top surface 57 Imaging beam scan path 61 Image area of ​​surface before milling 63 Image area of ​​the surface after milling 67 Lines or steps between different areas 155 Wafer Stage 160 Inspection Volume 163 Image Data Stack Subset of 165 cross-sectional images Stack of 167 image slices 171 Volume of Interest 173 Areas of Interest 175 Averaged image of region of interest 181 Sparse Image Acquisition Areas 307 Measured cross-sectional image of HAR structure 309 HAR Structure 311 Cross-sectional image slice Cross section through 313 word line 315 Edge with Surface 331 Averaged Image Slice

Claims

1. 1. A method for volumetric inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample, comprising: - acquiring a first number N of cross-sectional images by rapid milling and rapid image scanning of a plurality of cross-sections through said examination volume; forming a second number M of averaged image slices from the first number N of cross-sectional images by calculating a moving average value in a predetermined direction of the semiconductor object of interest, wherein the moving average value of each averaged image slice is calculated from a subset of the first number N of cross-sectional images.

2. 2. The method according to claim 1, wherein the second number M is smaller than the first number N, M<N, in particular M≦A×N with a factor A≧3.

3. 2. The method according to claim 1, wherein the second number M is approximately equal to the first number N, in particular M<N and M / N≦0.

95.

4. The method of any one of claims 1 to 3, comprising performing the high speed milling of a new cross section and the high speed image scanning for the acquisition of a subsequent cross section at different times.

5. 4. The method of claim 1, further comprising simultaneously performing the high-speed milling of a new cross section and the high-speed image scanning for the acquisition of a cross-sectional image in parallel with a scanning direction of a charged particle imaging beam perpendicular to a direction of a focused ion beam (FIB), the cross-sectional image including a first region corresponding to a first cross section before an actual high-speed milling operation and a second region corresponding to a second, deeper cross section due to the actual high-speed milling operation.

6. The method of any one of claims 1 to 5, wherein the semiconductor object of interest is at least one HAR channel, the predetermined direction being oriented perpendicular to a top surface of the wafer sample.

7. The method according to any one of claims 1 to 6, comprising performing the high speed milling at a tilt angle relative to a top surface of the sample, the tilt angle GF≦45°, in particular GF≦30°.

8. The method of any one of claims 1 to 7, comprising calculating a moving average value of the second number of cross-sectional images by convolving the cross-sectional images with a one-dimensional convolution kernel parallel to the predetermined direction.

9. The method of claim 8 , wherein the one-dimensional convolution kernel spans at least B=3 consecutive cross-sectional images and has the form of a weighted sum, a rectangular function, or a Gaussian distribution function.

10. 10. The method of claim 8 or 9, comprising selecting a range B of the one-dimensional convolution kernel according to a noise level in the cross-sectional images, so as to obtain a predetermined signal-to-noise ratio (SNR) level in an averaged image slice.

11. 11. The method according to claim 1, further comprising: compensating for lateral drift of cross-sectional images of the plurality of cross-sectional images during the calculation of the moving average value, the direction of the lateral drift being perpendicular to the predetermined direction.

12. 12. The method of claim 11, comprising: determining the lateral drift of cross-sectional images in an averaged image slice with low gradient or low contrast in the drift direction; and optimizing the calculation of the averaged image slice with maximum isotropic contrast or gradient by averaging and including lateral shift for at least one cross-sectional image.

13. The method of claim 11 , further comprising determining the lateral drift from a reference position obtained from an image section of an alignment reference mark.

14. The method of claim 13 , further comprising averaging laterally by pixel binning or horizontal convolution of the image segments of the alignment reference mark with an average convolution kernel.

15. The method according to any one of the preceding claims, wherein the method further comprises adjusting a distance between two adjacent cross-sections according to a priori information of the semiconductor object of interest in the at least first volume of interest.

16. The method according to any one of claims 1 to 15, further comprising adjusting a milling angle of a subsequently milled cross-section according to prior information of the semiconductor object of interest in the at least first volume of interest.

17. 17. The method of any one of claims 1 to 16, further comprising changing the detection mode between subsequent high speed image scans, wherein the change in detection mode comprises at least one of changing the dynamic range, changing the energy range of the interaction products, or changing the type of interaction products.

18. 1. A method for inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample, comprising: - defining at least a first volume of interest within a control volume of a semiconductor wafer, said first volume of interest being within said control volume; - milling a plurality of cross sections through said control volume; - acquiring several cross-sectional images by performing an image scanning operation of cross-sectional sections within said at least first volume of interest, each cross-sectional image section being determined according to a region of interest formed by an intersection of a cross section with said at least first volume of interest.

19. 20. The method of claim 18, comprising milling at an oblique angle relative to a wafer surface, wherein the at least one region of interest is oriented perpendicular to the wafer surface.

20. 20. The method of claim 18 or 19, comprising defining the at least first volume of interest according to CAD data.

21. The method of any one of claims 18 to 20, further comprising registering the at least a first volume of interest within the examination volume according to at least a first image scan of a first cross-section within the examination volume.

22. 22. The method of claim 21, wherein the step of registering comprises further re-registration of the at least first volume of interest in at least second or further image scans of second and further cross sections within the examination volume.

23. The step of defining the at least first volume of interest further comprises: a) acquiring a first image scan of a first cross-section; b) selecting a first cross-sectional image section in the first image scan, the first cross-sectional image section including a cross-section of the semiconductor object of interest; c) milling a second cross-section into the control volume of the wafer sample; and d) selecting a second cross-sectional image section by projecting the first cross-sectional image section onto the second cross-section; e) acquiring a second cross-sectional image section of the second cross-sectional section; f) repeating steps c) to e) by milling further cross sections and selecting and acquiring further cross section image sections until a predetermined interruption criterion is met.

24. 24. The method of claim 23, wherein in step e) projecting the first cross-sectional image section onto the second cross-section is performed in the predetermined direction.

25. 25. The method according to claim 23 or 24, further comprising a step of adjusting the position and size of the first cross-sectional image section according to the second cross-sectional image section, and the position and size of the cross-section of the semiconductor object of interest in the second cross-sectional image section are analyzed by image processing.

26. The method of any one of claims 18 to 25, wherein the semiconductor object of interest is a HAR channel and the predetermined direction is oriented perpendicular to the wafer surface.

27. The method according to any one of claims 18 to 26, further comprising the step of generating a 3D volumetric image of the 3D semiconductor object of interest.

28. 28. The method of claim 27, wherein the step of generating a 3D volumetric image of the 3D semiconductor object of interest comprises: generating the 3D volumetric image of the at least first volume of interest; and generating augmented 3D image data outside the at least first volume of interest.

29. 30. The method of claim 28, wherein the augmented 3D image data is derived from CAD data.

30. The method according to any one of claims 18 to 29, further comprising adjusting a distance between two adjacent cross-sections according to a priori information of the semiconductor object of interest in the at least first volume of interest.

31. The method according to any one of claims 18 to 30, further comprising forming a number of image slice segments from the plurality of cross-sectional image segments by taking a moving average in a predetermined direction.

32. 32. The method according to any one of claims 18 to 31, further comprising the step of changing a detection mode between two subsequent image scanning operations, wherein the change of detection mode comprises at least one of changing a dynamic range, changing an energy range of an interaction product, or changing a type of an interaction product.

33. 1. A method for inspection of a 3D semiconductor object of interest in an inspection volume of a wafer sample, comprising: - milling a plurality of cross sections through said control volume; selecting a number of sparse regions for each cross-section; - acquiring a plurality of sparse cross-sectional image segments by performing an image scanning operation of said plurality of sparse regions of each cross-section; generating a 3D volumetric image of said examination volume from said plurality of sparse cross-sectional image segments.

34. 34. The method of claim 33, wherein generating the 3D volumetric image comprises image interpolation between the plurality of sparse cross-sectional image segments.

35. 35. The method of claim 34, wherein the image interpolation between the plurality of sparse cross-sectional image sections is based on CAD data of the 3D semiconductor object of interest.

36. The method of claim 33 , wherein the selection of the plurality of sparse regions in each cross-section varies from cross-section to cross-section.

37. A method according to any one of claims 33 to 36, wherein the selection of the plurality of sparse regions for each cross section is made according to image analysis of a plurality of previously acquired sparse cross-sectional image segments.

38. The method according to any one of claims 33 to 37, wherein said selection of said plurality of sparse regions is made according to CAD data.

39. The method of any one of claims 33 to 38, wherein the milling is performed at an oblique angle relative to the wafer surface.

40. The method according to any one of claims 33 to 39, wherein the method further comprises adjusting the distance between two adjacent cross-sections according to a priori information of the semiconductor object of interest.

41. 41. The method according to any one of claims 33 to 40, further comprising the step of changing a detection mode between two image scanning operations, wherein the change of detection mode comprises at least one of changing a dynamic range, changing an energy range of an interaction product, or changing a type of an interaction product.

42. The method of any one of claims 33 to 41, further comprising compensating for lateral drift of each of the plurality of sparse regions of each cross-section.

43. 43. The method of claim 42, comprising determining the lateral drift from a reference position obtained from an image section of an alignment reference mark.

44. 44. The method of claim 43, further comprising obtaining an image section of the alignment reference mark at a lower resolution.

45. 1. A method for inspection of a 3D semiconductor object of interest in an inspection volume of a sample wafer, comprising: - milling a plurality of cross sections through said control volume; - scanning each cross-section with an imaging charged particle beam, during which interaction products including secondary electrons, backscattered charged particles, photons or x-rays of different energy ranges are generated; - forming a plurality of cross-sectional images from the detection signals of at least the first and second interaction products.

46. the first interaction product and the second interaction product a) a first and a second energy range of selected interaction products, or b) a first and a second dynamic range of the selected interaction product, or - c) a first and a second type of interaction product, 46. ​​The method of claim 45, wherein the compound is selected from the group consisting of

47. 47. The method of claim 45 or 46, wherein in the step of forming a plurality of cross-sectional images, the cross-sectional image obtained from a first detector is augmented by information received from a second detector.

48. 47. The method of claim 45 or 46, wherein in the step of forming a plurality of cross-sectional images, the cross-sectional image obtained from a first image scan is augmented with information received from a second image scan.

49. 1. A semiconductor inspection device for inspection of a 3D semiconductor object of interest in an inspection volume of a sample wafer, comprising: a wafer sample stage for holding and positioning the wafer sample; a dual column microscope comprising a focused ion beam (FIB) column and a charged particle beam imaging column forming a common intersection point of the optical axis of said FIB column and said charged particle beam imaging column; a control unit configured to control said dual column microscope to perform a slicing and imaging method in said examination volume; - a processor and a memory having installed thereon software code configured for the processor to perform any of the method steps described above; 13. A semiconductor inspection device comprising:

50. 50. The semiconductor inspection device of claim 49, further comprising at least first and second detectors configured to perform multi-modal image acquisition.

51. 51. A semiconductor inspection device according to claim 49 or 50, wherein the charged particle beam imaging column is formed by a corrected electron microscope including correction means for correction of chromatic and spherical aberrations.

Images