High resolution low distortion imaging using charged particle scanning microscope

Abstract

Techniques disclosed herein are to obtain microscope images with high resolution and low distortion. The conversion is applied to the distorted high-resolution image (220). This conversion may be decided based on the aggregated low-resolution images.

Description

Technical Field

[0001] The various examples generally relate to imaging using a charged particle scanning microscope, such as a scanning electron microscope. These examples are particularly relevant to digital post-processing of images obtained using charged particle microscopy to improve image quality. Background Technology

[0002] In scanning electron microscopy (SEM) or other types of charged particle scanning microscopy (such as helium ion microscopy), imaging time largely depends on the time required to scan the field of view (FOV) using an electron beam. To improve the signal-to-noise ratio (SNR), the so-called dwell time, i.e., the time the beam spends in one pixel, is typically increased. This inherently results in a longer imaging time.

[0003] On the other hand, various instabilities within the SEM lead to so-called "drift," which is a gradual shift of the field of view relative to the imaging object / sample. The root causes of drift include thermal drift, charging, plateau instability, or beam instability. This drift results in distortion of the acquired image, typically observed as a "tilt" or "shear" along the X or Y direction, the magnitude of which is proportional to the imaging time. This is illustrated in Figure 1. Figure 1 shows an image 220 obtained using a SEM with a relatively long dwell time (e.g., 1 μs or longer for a 1x1 nm pixel). In Figure 1, the distortion is clearly visible: depicted is a semiconductor structure comprising straight lines and circular structures arranged in a square pattern. The straight lines are curved, and the square pattern is tilted. The circular structures have elliptical ends at the tilted points.

[0004] Several techniques are known in the art to mitigate this type of drift-driven distortion. One approach is the so-called fast-frame averaging or drift-corrected frame integration.

[0005] In fast frame averaging, as shown in Figure 2, instead of obtaining a single image with a long dwell time (as shown in Figure 1) and high SNR but also large distortion, a series of fast images 201, 202, 203, and 204 with low signal-to-noise ratio (SNR) and low distortion are obtained within a shorter dwell time. These fast images 201, 202, 203, and 204 are then synthesized at 200. This produces a combined presentation 211 (or simply, a combined image) of these images 201, 202, 203, and 204. The challenge of fast frame averaging is the relative displacement of the field of view in the fast images due to the aforementioned drift. Therefore, it is necessary to align the fast images 201, 202, 203, and 204 relative to each other before synthesizing them to produce the final image. This is achieved using registration. Because images 201, 202, 203, and 204 have low SNR, the computational accuracy of their mutual drift is limited, resulting in imperfect alignment. Therefore, the combined 211 appears "blurry", that is, the resolution is reduced. Summary of the Invention

[0006] Therefore, advanced imaging techniques using charged particle scanning microscopy are needed to mitigate or reduce at least some of the aforementioned drawbacks. In particular, a proper balance needs to be struck between resolution loss (blurring) on ​​one hand and distortion on the other. High-resolution images with no or at least low distortion are required.

[0007] This needs to be satisfied through the features of the independent claims. The features of the dependent claims define multiple embodiments.

[0008] A computer-implemented method includes controlling a charged particle scanning microscope to load a sample. The method also includes controlling the charged particle scanning microscope using a first imaging setting to obtain one or more first images of the sample. The method further includes determining first localizations of multiple features of the sample based on the one or more first images. The method further includes controlling the charged particle scanning microscope using a second imaging setting to obtain a second image of the sample. The second imaging setting is at least partially different from the first imaging setting. The method further includes determining second localizations of the multiple features based on the second image. The method further includes determining a transition between the first localization and the second localization. The method further includes transforming the second image based on the transition.

[0009] A processor configured to load and execute program code. When executing the program code, the processor performs the methods described above.

[0010] A program code that can be loaded by a processor and executed by the processor, wherein the processor performs the methods described above when executing the program code.

[0011] It should be understood that the foregoing features and those to be explained below can be used not only in the indicated combinations, but also in other combinations or separately, without departing from the scope of the invention. Attached Figure Description

[0012] Figure 1 schematically illustrates SEM images from various examples, acquired using relatively long dwell times.

[0013] Figure 2 schematically illustrates multiple SEM images presented in combination, with the former acquired using a relatively short dwell time.

[0014] Figure 3 The illustrations depict systems based on various examples, including computing devices and SEMs.

[0015] Figure 4 The flowchart shows the methods based on various instances.

[0016] Figure 5 The first localization of the features depicted in a combined presentation of multiple first images obtained using SEM according to various examples is shown.

[0017] Figure 6 The second localization of features depicted in a second image obtained using SEM is shown based on various examples.

[0018] Figure 7 The illustrations schematically depict dual-beam devices incorporating SEM and focused ion beam apparatus according to various examples.

[0019] Figure 8 The illustrations depict layer-by-layer grinding processes based on various examples. Detailed Implementation

[0020] Some examples of the present invention typically provide multiple circuits or other electrical devices. All references to circuits and other electrical devices and the functions provided by each device are not intended to be limited to what is illustrated and described herein. While certain reference numerals may be assigned to the various circuits or other electrical devices disclosed, such numerals are not intended to limit the scope of operation of the circuits and other electrical devices. Such circuits and other electrical devices may be combined and / or separated from each other in any manner based on the type of ideal electrical implementation. It should be understood that any circuit or other electrical device disclosed herein may include any number of microcontrollers, graphics processing units (GPUs), integrated circuits, memory devices (e.g., flash memory, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electronically erasable programmable read-only memory (EEPROM), or other suitable variations thereof), and software that cooperates with each other to perform the operations disclosed herein. Additionally, any one or more electrical devices may be configured to execute program code specifically implemented in a non-transitory computer-readable medium, which may be encoded to perform any number of functions as disclosed.

[0021] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the following description of the embodiments should not be considered limiting. The scope of the invention is not intended to be limited to the embodiments or drawings described below, which are merely illustrative.

[0022] The accompanying drawings should be considered schematic representations, and the elements shown are not necessarily shown to scale. Rather, the various elements are shown to enable those skilled in the art to understand their function and general purpose. Any connection or coupling between functional blocks, devices, components, or other entities or functional units shown in the drawings or described herein may also be implemented through indirect connections or couplings. Coupling between components may also be established wirelessly. Functional blocks can be implemented in hardware, firmware, software, or a combination thereof.

[0023] The following section discloses techniques for imaging microscopic samples using charged particle imaging devices such as SEM.

[0024] The disclosed technique finds a trade-off when imaging samples such as semiconductor wafers using charged particle scanning microscopy (e.g., SEM): (i) using low-charge, fast scanning (low particle current / flux or short residence time per pixel) results in a poor signal-to-noise ratio (SNR); while (ii) using high-charge scanning (higher particle current / flux or longer residence time) produces a good SNR, but results in image distortion due to drift (including charge-induced drift). A technique is disclosed that combines both methods (i) and (ii) to produce images that are as accurate as possible.

[0025] Based on various examples, a superposition of numerous images obtained under short dwell times and / or low currents is determined. This combination presents images with low or zero distortion. This allows for the derivation of baseline real-world information regarding the relative configuration of the depicted features. This baseline real-world information can be used to obtain additional images without distortion using long dwell times and / or high particle currents. Specifically, the conversion can be determined based on a comparison of the corresponding localization of features in the two image types.

[0026] Figure 3 System 100 is schematically illustrated. System 100 includes a computing unit 150 and a charged particle scanning microscope 160. The computing unit 150 includes a processor 152 and a memory 153. The processor 152 can communicate with the charged particle scanning microscope 160 via a communication interface 154. The processor 152 can load program code from the memory 153 and execute the program code. When the program code is loaded and executed, the processor 152 can perform techniques as disclosed herein, such as controlling the charged particle scanning microscope to load samples, acquire images, etc. Moreover, the processor 152 can manipulate or digitally post-process the images acquired using the charged particle scanning microscope 160.

[0027] To acquire an image, the charged particle scanning microscope 160 includes a particle source 161, scanning optics 162 configured to direct / scan the primary particle beam emitted from the source 161, and a sample platform / sample carrier 164 configured to hold a sample (e.g., a semiconductor or biological sample). Charged particles from the secondary beam are detected by a detector 165. By appropriately controlling the scanning optics 162 to scan the sample with the primary particle beam, multiple measurement points for image acquisition can be obtained. A specific dwell time is implemented for each measurement point.

[0028] Therefore, charged particle scanning microscopy (such as SEM) produces high-resolution images by focusing an electron beam to scan a sample. Electrons, typically emitted from a tungsten filament or field emission gun, are accelerated and focused onto the sample by scanning optics (containing electromagnetic lenses). As the electrons interact with the sample, they emit various signals containing secondary electrons. These signals are detected by a detector (usually a secondary electron detector) to produce an image. In raster scanning, the electron beam moves systematically pixel-by-pixel across the sample in a grid pattern. The time the electron beam spends on each pixel before moving to the next is called the dwell time. Adjusting the dwell time affects image quality: a longer dwell time results in a better signal-to-noise ratio, but at the cost of slower scanning speed and increased distortion. The following reveals techniques that can address this trade-off by appropriately post-processing the acquired images.

[0029] Figure 4 The flowchart shows the methods based on various instances. Figure 4The method can be performed by a computer used to control a charged particle scanning microscope. Figure 4 The method can be derived from Figure 3 The processor 152 of the computing device 150 of the system 100 shown executes, for example, when loading program code from memory 153 and executing the program code.

[0030] Figure 4 The method can obtain microscope images of samples with relatively low distortion, while also having relatively high resolution and SNR.

[0031] Figure 4 The technique described here is based on dedistorting a second image obtained with a long dwell time (and therefore with large inherent distortion). This is based on a transformation that quantizes the distortion. This transformation is obtained from one or more first images obtained using a short dwell time. Features are located in one or more of the first images and compared with the corresponding locations of features in the second image.

[0032] In selective step 905, the transformation can be configured. Multiple free parameters for the transformation can be determined. The transformation parameters can be one or more of the following: scaling, rotation, shearing, and translation / displacement. For example, for a linear transformation, fewer free parameters are needed than for a nonlinear transformation of a specific order.

[0033] The conversion can be configured based on existing knowledge, such as the expected magnitude or severity of distortion. Specifically, the conversion can be configured based on imaging settings for obtaining a first image(s) and / or imaging settings for obtaining a second image and / or the sample type of the sample to be imaged and / or the hardware configuration of the microscope. Imaging settings may include dwell time. Imaging settings may include scanning pattern. Imaging settings may specify focal length. Imaging settings may include spot size.

[0034] Depending on the use case, instrument acquisition mode, and laboratory / factory environment, the exact form of conversion can be selected, especially the number of free parameters.

[0035] These characteristics influence the expected severity of distortion. More severe distortion typically requires more free parameters for the transformation to fully and accurately capture it. For example, certain sample types are prone to charge drift. Some hardware settings in charged particle scanning microscopy may be susceptible to temperature drift. Based on existing knowledge, the number of free parameters for the transformation can be appropriately configured. The number of free parameters should be chosen to be as small as possible, but can also be as large as needed.

[0036] Not all scenarios require configuring the transition at step 905. In other scenarios, the transition can be pre-configured, making step 905 optional.

[0037] At step 910, the count of one or more first images used as references can be selectively determined (i.e., one or more images can also be labeled "reference images"). That is, it can be determined how many first images to acquire.

[0038] The count of one or more first images may depend on the count of the free parameters of the transformation, as determined in step 905. For example, for free parameters with larger counts, it is generally preferable to use first images with larger counts to ensure that each parameter value can be determined with appropriate accuracy. Specifically, first images with larger counts result in higher accuracy localization of multiple features. Consequently, this enables a more accurate determination of the transformation.

[0039] For example, if a linear affine transformation is sufficient, only six free parameters are needed. This translates to a relatively finite number of first images; for example, even a single first image may suffice.

[0040] The count of one or more first images can be chosen to be as small as possible, but as large as needed.

[0041] Step 910 allows for a balance between the quality of distortion correction (i.e., the amount of residual, uncorrected distortion in the generated image) and imaging time.

[0042] Step 910 is optional. In some instances, the count of one or more first images is predefined.

[0043] In selective step 915, the sample to be imaged is loaded. For this purpose, the charged particle scanning microscope can be controlled to load the sample. The loading lock can be opened after the pressure chamber is filled; and once the sample is placed on the sample carrier, the loading lock can be closed and the pressure chamber can be emptied.

[0044] Alternatively, samples can be preloaded.

[0045] In step 920, one or more first images (also referred to herein as reference images) are obtained. This includes controlling the charged particle scanning microscope to obtain one or more first images of the sample using first imaging settings.

[0046] The one or more first images may correspond to SEM images 201-204 shown in Figure 2.

[0047] The platform does not move between acquiring multiple first images. That is, the field of view (at least nominally) is fixed. The same measurement point (e.g., platform position) can be used.

[0048] If a count of one or more first images has been set in step 910, that count is used in step 920.

[0049] If multiple first images are obtained in step 920, a combination of multiple reference images is determined in step 925. For example, pixel-by-pixel addition or multiplication or other combinations may be performed.

[0050] Such combined presentations or combined images 211 have been discussed previously in conjunction with Figure 2.

[0051] Registration can be performed between multiple first images. Then, based on the registration, the combination can be performed. Specifically, based on the registration, an alignment vector can be determined for each first image, and multiple first images can be aligned by applying the corresponding alignment vector.

[0052] The combination of multiple first images can be optimized based on an edge sharpness measurement. This means that parameters of the combination (e.g., the relative weights of each pixel or individual first images) can be adjusted to maximize edge sharpness in the combined rendering. The edge sharpness measurement can be maximized. An edge sharpness measurement can be defined, for example, as the absolute value of the average local gradient at the boundaries of each feature. More generally, the sharpness measurement can be included in an evaluation function that is optimized when computing the alignment vector. This approach allows for a reduction in the blurriness of the combined rendering and thus improves the accuracy of the reference object localization subsequently performed in step 930.

[0053] In step 930, first localizations of multiple features of a sample depicted in one or more first images are determined based on one or more first images obtained in step 920. For example, if step 925 is performed, the first localizations can be determined based on a combination of multiple reference images.

[0054] Therefore, the first localization can serve as a baseline of true information about the relative configuration of features, and thus can be used to derive a transformation for dedistorting the distorted second image. Figure 5 An example of such a first positioning 301 is shown. In this document, the center point of multiple circular features (e.g., vertical vias in a three-dimensional (3D) memory structure, such as those to be combined below) Figure 8 The points discussed are located and the set of these center points (intersections) specifies the reference location.

[0055] Please refer to the reference again immediately. Figure 4 Step 930 may therefore include the detection of a set of marked structures and the determination of their image coordinates. Since multiple first images or a combination of single first images present no distortion (or only very little distortion), the coordinates are affected only by random errors (depending on blur and signal-to-noise ratio), but not by systematic errors. In other words, since one or more first images have relatively low distortion (obtained using relatively short dwell times), the first localization of the feature can be considered a reference without significant distortion. Therefore, the first localization can also be labeled as a "reference localization".

[0056] In step 935, a second image is obtained. Similar to step 920, this involves controlling the charged particle scanning microscope to acquire the second image using the second imaging settings.

[0057] The second imaging setting is at least partially different from the first imaging setting used in step 920. Specifically, the dwell time for the second imaging setting used in step 935 is greater than the dwell time for the first imaging setting used in step 920. Therefore, the SNR for the second imaging setting used in step 935 is greater than the SNR for the first imaging setting used in step 920.

[0058] Therefore, as mentioned above, the second image is significantly affected by drift-driven distortion.

[0059] An example of this second image is image 220 shown in Figure 1.

[0060] The second image has a field of view corresponding to that of the one or more first images obtained in step 920. This means that the same features visible in the one or more first images obtained in step 920 are also visible in the second image obtained in step 935. However, the relative configuration of these features in the second image is distorted.

[0061] Therefore, in step 940, a second localization of multiple features can be determined based on additional images. This is in Figure 6 As shown in the image; in this document, the second positioning 302 is shown by crossing, and its positioning is directed at the center position of the circular / elliptical features of image 220.

[0062] In step 945, the transition between the first position determined in step 930 and the second position determined in step 940 is determined.

[0063] Step 945 may include, for example, an operational mathematical transformation in the form of a parsing map function.

[0064]

[0065]

[0066] It will be in step 940 (see Figure 6 The coordinates of the features in the second location determined by ) Transformed into coordinates of the corresponding feature in a combination of one or more first images or in a single first image. .coefficient … and , … These are the parameters of the mapping function, i.e., the free parameters of the transformation. These functions can be computed using, for example, the least squares method to minimize (best fit) the deviation between the object coordinates and the measurement. As a mapping function, a linear affine transformation can be used, for example. In some cases, higher-order polynomial transformation functions are required for accurate fitting. This increases the number of free parameters of the transformation. The number of local features used to fit the transformation should be significantly greater than the number of fitting parameters. … and , … This helps to avoid overfitting. It also helps to mitigate the impact of inaccuracies in the individual feature locations determined in step 930.

[0067] In step 950, the further image obtained in step 935 is transformed based on this transformation. This includes applying the transformation or its inverse transformation. This produces a generated image with low or zero distortion.

[0068] In summary, a technique for obtaining distortion-free or low-distortion, yet unblurred, generated images has been disclosed. To detect a set of features at “distortion-free” locations, a single low-SNR, distortion-free reference image or a combination of multiple reference images (typically blurred) can be presented. The same features can then be detected in further images obtained using longer dwell times, thus exhibiting high SNR and significant distortion. Subsequently, an analytical transformation function is computed and applied to the coordinates of features in the distorted additional images to transform them into the coordinates of corresponding features in one or more distortion-free reference images. This transformation function can ultimately be applied to the distorted, high-SNR image to eliminate or significantly reduce distortion. Thus, a “distortion-free”, high-resolution generated image is produced.

[0069] This type of technique is particularly helpful for studying 3D storage structures using slice image tomography. Slice image tomography can obtain 3D volumetric images of nanoscale semiconductor structures. The presence of these semiconductor structures in the image can be located as features at steps 930 and 940.

[0070] Possibly Figure 4Between step 920, obtaining one or more first images, and step 935, obtaining a second image, a layer of material is removed from the sample using a polishing process employing slicing and image tomography techniques. This is illustrated in step 955, where polishing occurs. In other words, the first location determined in step 930 remains valid for multiple slices of the sample imaged using the corresponding second image in each iteration of loop 959. This is because the first location can be determined based on prior knowledge of the relative configuration and extent of the corresponding semiconductor structure: such techniques are based on the finding that the vertical extent of certain semiconductor structures (such as vias in 3D memory structures), i.e., components with polished layers extending perpendicular to the material, can be used as prior knowledge about the appearance of multiple features. In particular, array structures of such vias in 3D memory structures are unaffected by polishing. Therefore, it is not necessary to redetermine the first location for subsequent slices imaged using slicing and tomography techniques. Such techniques accelerate overall imaging.

[0071] Section imaging computed tomography is performed using a dual-beam setup. In a dual-beam setup, two particle optics systems are configured at an angle (cylinder offset angle). Both particle optics systems jointly observe the sample. The two particle optics systems can be vertically oriented or oriented with a cylinder offset angle between 45° and 90°. The first particle optics system defines the imaging column. The imaging column can be visualized using charged particle scanning microscopy such as SEM (see [link to documentation]). Figure 3 This is performed using a microscope (160) or a scanning helium ion microscope (HIM). A second particle optical system defines the grinding column. The grinding column can be a focused ion beam (FIB) optical system using, for example, gallium (Ga) ions. Ga-ion FIBs are used to slice wafers one-to-one to obtain test volumes. Therefore, images depicting wafer cross-sections are obtained using the imaging column at different grinding depths; see step 935 for multiple iterations of cycle 959.

[0072] An exemplary implementation of slice and image tomography is described in Neumann, Jens Timo, et al., in the International Society for Optical Engineering, Metrology, Inspection and Process Control of Microlithography, XXXIV, Vol. 11325, entitled “3-D analysis of high-aspect ratio features in 3-D-NAND”.

[0073] Slicing and image tomography are also described in WO 2021180600 A1, which is incorporated herein by reference.

[0074] Inspection system such as Figure 7As shown. The wafer inspection system 5000 is configured for slicing and imaging in a wedge-shaped cutting geometry using a dual-beam device 5001. For wafer 5008, several measurement points containing measurement points 5006.1 and 5006.2 are defined in a position map or inspection list generated from inspection tools or design information. Wafer 5008 is placed on wafer support stage 5015 (corresponding to sample platform 164). Wafer support stage 5015 is mounted on platform 5155 having actuators and position controller 5021. Actuators and components (such as laser interferometers) for precise control 5021 of wafer platform 5155 are known in the art. Control unit 5016 receives information about the actual position of wafer platform 5155 and is configured to control wafer platform 5155 and adjust measurement point 5006.1 of wafer 5008 at intersection 5043 of dual-beam device 5001. The dual-beam device 5001 includes a FIB column 5050 with a FIB optical axis 5048 and a charged particle beam (CPB) imaging system 5040 with an optical axis 5042 (e.g., SEM or HIM; see [link]). Figure 3 (Charged particle scanning microscope 160). At the intersection 5043 of the dual optical axes of the FIB and CPB imaging systems, the wafer surface 5055 is configured to form an angle GF with the FIB axis 5048. The FIB and CPB jointly observe the sample. The FIB axis 5048 and the CPB imaging system axis 5042 include the angle GFE. Figure 7 In the coordinate system, the normal to the wafer surface 55 is given by the z-axis. A focused ion beam (FIB) 5051 is generated by the FIB column 5050 and impacts the surface 5055 of the wafer 5008 at an angle GF. At inspection point 5006.1, the inclined profile surface is ground into the wafer at a predetermined y-position at approximately an angle GF using ion beam grinding, controlled by a platform 5155 and a position controller 5021. In the illustrated example, the angle GF is approximately 30°. Due to the beam divergence of the focused ion beam (e.g., a gallium ion beam) or due to the variable material properties related to the grinding along the profile, the actual angle of inclination of the inclined profile surface can deviate from the angle GF by 1° to 4°. An image of the ground surface is obtained using a charged particle beam imaging system 5040. The aspects explained above in conjunction with Figures 1-5 apply to such imaging; that is, a trade-off needs to be struck between low SNR and distortion. Figure 7 In one example, the charged particle beam imaging system 5040 is configured such that its charged particle beam 5044 is perpendicular to the wafer surface 5055 and parallel to the z-axis. In other configurations, the optical axis 5042 of the charged particle beam imaging system 40 is configured to form an angle with the z-axis.

[0075] During imaging, the charged particle beam 5044 is scanned by the scanning unit of the charged particle beam imaging system 5040 (see...). Figure 3The scanning optics scan along a scanning path on the cross-sectional surface of the wafer at measurement site 5006.1, generating secondary particles and backscattered particles. Particle detector 5017.1 and selective internal particle detector 5017.2 collect at least some of the secondary particles and / or backscattered particles and communicate particle counting to control unit 5019. Other detectors may also be present for other types of interaction products (such as X-rays or photons) (see [link to relevant documentation]). Figure 3 (Detector 165). Control unit 5019 controls charged particle beam imaging column 5040 and FIB column 5050, and is connected to control unit 5016 to control the wafer position mounted on wafer support stage 5015 via wafer platform 5155. Operation control unit 5002 communicates with control unit 5019, and triggers, for example, the placement and alignment of measurement point 5006.1 of wafer 5008 at intersection point 5043 via wafer platform movement, and repeatedly triggers FIB grinding, image acquisition, and platform movement operations. Control unit 5019 and operation control unit 5002 include memory for storing instructions in the form of software code, and at least one processor for executing instructions during operation. The memory is further provided for storing digital image data. Operation control unit 5002 may further include a user interface or an interface connected to other communication interfaces to receive instructions, previous information, and transmit inspection results.

[0076] Each new profile surface is polished by FIB beam 5051 and imaged by charged particle imaging beam 5044 (see Figure: Step 935; Cycle 959).

[0077] Figure 8 Further details of slicing and imaging measurements in the wedge-cut geometry are shown. By repeating the slicing and imaging methods in the wedge-cut geometry, multiple J average image slices containing average image slices of cross-sectional surfaces 5052, 5053.i…5053.J are generated, and a 3-D volume image of the inspection volume 5160 at inspection site 5006.1 of wafer 5008 is generated. Figure 8 This illustrates a wedge-cut geometry in an example of a 3D memory stack. The cross-sectional surfaces 5053.1…5053.J are ground with FIB beam 5051 at an angle GF of approximately 30° to the wafer surface 5055, but other angles GF are also possible, such as those between GF = 20° and GF = 60°, and for example, GF = 36° and GF = 25°. Figure 8This illustrates the case when surface 5052 is a newly profiled surface that was last ground from FIB 5051. For example, a corresponding image slice is obtained by scanning the profiled surface 5052 using SEM beam 5044. The average profiled image slice contains first profiled image features formed by intersections with high aspect ratio (HAR) structures or vias (e.g., first profiled image features of HAR structures 5004.1, 5004.2, and 5004.3) and second profiled image features formed by intersections with layers L.1…LM, which contain, for example, SiO2, SN-, or tungsten lines. Some lines are also referred to as “word lines.” The maximum number of layers M is typically greater than 50, for example greater than 100 or even greater than 200. HAR structures and layers extend throughout most of the inspection volume in the wafer, but may also include gaps. The diameter of HAR structures is typically less than 100 nm, for example about 80 nm, or for example 40 nm. HAR structures are regularly configured, such as hexagonal gratings, with pitches approximately below 300 nm, for example even below 250 nm, or below 60 and below 40 nm (e.g., for DRAM). The appearance of the HAR structure in SEM images can be used to determine positioning to identify transitions, as previously combined. Figure 4 Explanation: Steps 930 and 940.

[0078] While the invention has been shown and described with reference to certain preferred embodiments, equivalents and modifications will be apparent to those skilled in the art upon reading and understanding the specification. The invention includes all such equivalents and modifications and is limited only by the scope of the appended claims.

[0079] To illustrate this, various examples have been disclosed above in which dwell time is adjusted between acquiring images with low SNR and high SNR, respectively. Alternatively or additionally, different imaging parameters that affect SNR, such as particle flux / particle current, can also be adjusted.

Claims

1. A computer-implemented method, comprising: - Control the (915) charged particle scanning microscope (160, 5040) to load the sample (5008); - Using the first imaging setting control (920), the charged particle scanning microscope (160, 5040) obtains one or more first images (201, 202, 203, 204) of the sample (5008); -Based on one or more first images (201, 202, 203, 204), determine the first localization (301) of multiple features of the sample (5008); - The charged particle scanning microscope (160, 5040) is controlled (930) using a second imaging setting to obtain a second image (220) of the sample (5008), the second imaging setting being at least partially different from the first imaging setting; -Based on the second image (220), determine (940) the second location (302) of the plurality of features; - Determine (945) the transition between the first position (301) and the second position (302); and - Based on this transformation, transform (950) the second image (220).

2. The computer implementation method as described in claim 1, The one or more first images (201, 202, 203, 204) contain multiple first images. This method further includes: - Determine (925) the combined presentation (211) of the plurality of first images (201, 202, 203, 204), The first positioning (301) is determined based on the combined presentation (211).

3. The computer implementation method as described in claim 2, The combined rendering (211) is determined based on the pixel-by-pixel combination (200) of the plurality of first images (201, 202, 203, 204).

4. The computer implementation method as described in claim 2 or 3, The optimization based on edge sharpness measurement determines the combined presentation (211).

5. The computer-implemented method as described in any of the preceding claims, The first location (301) is further determined based on prior knowledge of the relative configuration and / or range of the semiconductor structures (5004.1, 5004.2, 5004.3) with respect to the multiple features.

6. The computer-implemented method as described in any of the preceding claims further comprises: - Control the (955) grinding process to remove a material layer from the sample (5008) between obtaining the one or more first images and obtaining the second image. The semiconductor structure with the aforementioned features has components that extend perpendicular to the material layer.

7. The computer-implemented method as described in claim 6, The grinding process is a focused ion beam grinding process using a focused ion beam source, so that the sample (5008) can be observed together using the charged particle scanning microscope (160, 5040).

8. The computer-implemented method as described in any of the preceding claims, - Based on the count of the free parameters of the transformation, determine (910) the count of the one or more first images (201, 202, 203, 204).

9. The computer implementation method as described in claim 8, further comprising: -Based on at least one of the hardware configuration of the charged particle scanning microscope (160, 5040), the first imaging setting, the second imaging setting, or the sample type of the sample (5008), determine (905) the count of the free parameters of the transformation.

10. The computer-implemented method as described in any of the preceding claims, The second dwell time and / or particle current of the second imaging setting are greater than the first dwell time and / or particle current of the first imaging setting.

11. The computer-implemented method as described in any of the preceding claims, The second signal-to-noise ratio for the second imaging setting is greater than the first signal-to-noise ratio for the first imaging setting.

12. A processor configured to load and execute program code, wherein the processor performs the method as described in any of the preceding claims when executing the program code.

13. A system (100) comprising the processor as described in claim 12 and the charged particle scanning microscope (160, 5040).

14. Program code for being loaded by a processor and configured to cause the processor to perform the method as claimed in any one of claims 1 to 11.

Images