Algorithm for 3D reconstruction of diagonally cut semiconductor logic structures

Abstract

The invention relates to a three-dimensional reconstruction algorithm for diagonally cutting semiconductor logic structures. According to some embodiments, the present application provides a method for generating a representative 3D reconstruction of a plurality of periodic structural elements of a wafer, the method comprising: obtaining a 2D image of a top view of the plurality of structural elements, the plurality of structural elements being diagonally sectioned at a compound angle that allows 3D volume sampling, the plurality of structural elements are segmented such that each of the plurality of structural elements is segmented at a different height thereof, and generating a representative 3D reconstruction of the structural elements based on the images. The method enables fast, high resolution 3D volumetric inspection of 3D structural elements, and is sufficiently non-destructive for on-line bonding.

Description

Technical Field

[0001] This disclosure generally relates to an algorithm for converting 2D SEM images of diagonally cut structures into 3D reconstructions. Background Technology

[0002] Scanning electron microscopy (SEM) produces images of samples, such as wafers, by scanning a surface with a focused electron beam. Electrons interact with atoms in the sample, generating various signals that contain information about the sample's surface morphology and composition. However, due to the trend towards 3D structures in the wafer industry and the 2D imaging nature of SEM, solutions that allow for volumetric sampling of wafers are needed.

[0003] Solutions for 3D volumetric sampling, such as CD-SAXS, optical scattering measurements, and TEM, have been developed. However, current solutions...

[0004] 1. Not covering the entire application space: CD-SAXS (logic), TEM (memory), scattering measurements (memory and some logic); and / or

[0005] 2. The results take a very long time to obtain (TEM and CD-SAXS); and / or

[0006] 3. It is destructive (the entire chip is scrapped) and cannot be performed online.

[0007] Focused ion beams, also known as FIBs, are an example of technologies specifically used in the semiconductor industry. While SEM uses a focused electron beam to image samples within a chamber, FIB setups alternatively use a focused ion beam. FIBs can also be combined with both electron beam columns and ion beam columns in a system, allowing either beam to be used to study the same features.

[0008] FIB can be operated at low beam current for imaging or at high beam current for site-specific sputtering or milling. However, unlike SEM, FIB is inherently destructive to the sample.

[0009] At low primary beam currents, very little material is sputtered, allowing the FIB system to achieve 5 nm imaging resolution. At higher primary currents, a large amount of material can be removed by sputtering, allowing samples to be precisely milled to the submicron or even nanometer scale.

[0010] Until recently, FIBs were primarily used in the semiconductor industry. Applications such as defect analysis, circuit modification, photomask repair, and site-specific transmission electron microscopy (TEM) sample preparation on integrated circuits have become common procedures.

[0011] The latest FIB systems offer high-resolution imaging capabilities; this capability, combined with in-situ profiling, eliminates the need in many cases to examine FIB-profiled samples in a standalone SEM instrument. However, the highest resolution imaging and prevention of damage to sensitive samples still require SEM imaging.

[0012] It has been publicly known that SEM columns and FIB columns can be combined into the same chamber; however, SEM-FIB imaging has been both destructive and time-consuming until now.

[0013] Therefore, there has been a need for tools that enable high-resolution 3D volumetric inspection of 3D structural elements, which are fast and preferably non-destructive enough for online assembly. Summary of the Invention

[0014] This disclosure generally relates to an algorithm for converting 2D images into 3D reconstructions. The algorithm input is a 2D image, such as a SEM image, AFM image, or EDX image (or any combination thereof) of a wafer structure cut at a compound angle. The output is a single 3D reconstruction of a unit cell of the milled wafer structure. Additionally or alternatively, the output may be a 2D cross-sectional view image (along any arbitrary plane) of the reconstructed periodic structure. Furthermore, the algorithm can perform measurements on the reconstructed structure and on the original image to characterize and / or inspect the quality of the wafer structure and detect changes that occur during the manufacturing process.

[0015] According to some embodiments, the algorithm disclosed herein can receive one or more images of a region of a wafer comprising periodic structural elements as input. These one or more images are captured after the 3D structural elements are diagonally milled at a predetermined compound angle, such that each of the periodic 3D structural elements is cut at a different height, thereby exposing layers of the periodic 3D structure at different depths. Based on these one or more images, the algorithm can advantageously reconstruct the 3D structural elements by combining / assembling layers of multiple structures, each layer showing a different depth of the structure. This advantageously allows characterizing the 3D structural elements at different depths of the structure while sacrificing only a relatively small region of the wafer. Furthermore, it allows characterizing the 3D structural elements at different depths of the structure by performing only a single diagonal cut, thereby significantly reducing the time-to-results.

[0016] Therefore, based on a single cut in one or more regions of the wafer, and a single 2D image (per region) of the 3D structure on the wafer, the structure can be examined at high resolution (about 1 nm) with only a small (negligible) area of ​​the wafer sacrificed (e.g., less than 5% of the wafer).

[0017] According to some embodiments, the algorithm disclosed herein can receive multiple images as input of a region of a wafer including periodic structural elements. In this case, an image is captured after each of multiple diagonal cuts, which is also referred to herein as "sequential delamination" and "delamination". According to some embodiments, sequential delamination can advantageously improve reconstruction resolution and / or statistics compared to a single cut.

[0018] Advantageously, once the structural elements have been reconstructed (also referred to herein as “cross-sectional views”), the algorithm can perform various volumetric measurements to determine the characteristics and / or dimensions of the 3D structural elements and / or their components (e.g., the nanosheet width or thickness of a gate-all-around (GAA) transistor).

[0019] According to some embodiments, based on volume measurements, the quality / attributes of the wafer's manufacturing process can be determined.

[0020] According to some embodiments, the algorithm can be advantageously applied online to manufacturing processes, thereby providing 3D resolution (or similar levels) to three-dimensional critical dimension (CD) SEMs in a short and efficient manner.

[0021] In other words, the method disclosed herein has the advantage of being able to fully reconstruct the 3D volume of the structure and measure it with a 3D resolution of about 1 nm.

[0022] According to some embodiments, this disclosure provides a method for generating a representative 3D reconstruction of a plurality of periodic structural elements of a wafer, the method comprising: obtaining a 2D image of a top view of a plurality of structural elements, the plurality of structural elements being diagonally cut at a composite angle that allows for 3D volumetric sampling, such that each of the plurality of structural elements is cut at a different height; and generating a representative 3D reconstruction of the structural elements based on the image.

[0023] According to some embodiments, the plane of rotation (ψ) can be selected / set before cutting. According to some embodiments, the rotation angle (ψ) can be set / selected based on one or more features and portions of the structural elements (e.g., based on their height, width, substructures, etc.). According to some embodiments, the rotation angle (ψ) can be changed / adjusted between different cuts to thereby expose its different planes of rotation.

[0024] According to some embodiments, the method further includes generating a point cloud based on the 2D coordinates of each pixel in the image and one or more known parameters associated with a plane diagonally cut relative to a plurality of structural elements, wherein each point in the point cloud represents at least the location of each pixel in 3D space and its grayscale (GL); creating at least one GL-reconstructed image of a cross-sectional view of the point cloud; and performing one or more volumetric measurements on the at least one GL-reconstructed image to characterize the plurality of structural elements or one or more components thereof.

[0025] According to some embodiments, the method includes using an image analysis algorithm to segment a GL-reconstructed image to identify one or more components of a plurality of structural elements.

[0026] According to some embodiments, imaging of the diagonally cut top view is performed using scanning electron microscopy (SEM), atomic force microscopy (AFM), or focused ion beam (FIB) imaging. Each possibility is a separate embodiment. According to some embodiments, the SEM is cold field emission (CFE)-SEM. Advantageously, using CFE-SEM allows for surface measurements with low-energy electrons and consequently provides advantageously high depth resolution and high lateral resolution SEM imaging.

[0027] According to some embodiments, creating a reconstructed image of a cross-sectional view includes: sorting points in a point cloud along a coarse-resolution 3D grid; for each pixel, finding the nearest surrounding point in the point cloud within a predefined radius; and calculating an output pixel GL based on the surrounding points. According to some embodiments, calculating the output pixel GL includes linear interpolation between the nearest surrounding points.

[0028] According to some embodiments, the method further includes preprocessing the obtained 2D image and performing measurements on it.

[0029] According to some embodiments, the method further includes generating 2D and / or 3D models of multiple structural elements based on multiple cross-sectional views derived from a point cloud.

[0030] According to some embodiments, one or more volume measurements include the width of the structure, the layer thickness, the height of the structure, the recesses of the layers, the radius of the structure, the angle of the structure, or any combination thereof. Each possibility is a separate embodiment.

[0031] According to some embodiments, the method further includes determining the quality / attributes of the wafer production line based on volume measurements.

[0032] According to some embodiments, the method is suitable for in-line implementation into the wafer production line, i.e., the method can be performed via a single tool, thereby ensuring an integrated and streamlined process.

[0033] According to some embodiments, multiple structural elements are subjected to at least two diagonal cuts, wherein each subsequent diagonal cut is performed at a plane deeper than the previous diagonal cut. According to some embodiments, obtaining a top view of the diagonal cuts includes obtaining a top view image after each diagonal cut. According to some embodiments, the point cloud is generated based on the 2D coordinates of all pixels in the 2D image obtained at each diagonal cut. According to some embodiments, the distance between each diagonal cut is approximately 5 nm to 10 nm.

[0034] According to some embodiments, the method further includes cross-registration of top-view images of each of the diagonal cuts. According to some embodiments, the registration includes utilizing one or more markers on the wafer, algorithmic correlations between the top-view images, navigation data of each of the top-view images, or any combination thereof.

[0035] According to some embodiments, this disclosure provides a system for generating representative 3D reconstructions of a plurality of periodic structural elements of a wafer, the system comprising: a processing circuit system configured to: obtain a 2D image of a top view of a plurality of structural elements, the plurality of structural elements being diagonally cut at a composite angle that allows for 3D volumetric sampling, such that each of the plurality of structural elements is cut at a different height therein; and generate representative 3D reconstructions of the structural elements based on the image.

[0036] Some embodiments of this disclosure may include some, all, or none of the advantages described above. One or more other technical advantages will readily be apparent to those skilled in the art from the accompanying drawings, description, and claims. Furthermore, while specific advantages have been listed above, various embodiments may include all, some, or none of the listed advantages.

[0037] Unless otherwise defined, all technical and scientific terms used in this document have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. In case of conflict, the patent specification (including definitions) shall prevail. As used herein, unless the context clearly indicates otherwise, the indefinite articles “a” and “an” mean “at least one” or “one or more”.

[0038] Unless otherwise specifically stated, as is clear from this disclosure, it should be understood that, according to some embodiments, terms such as “processing,” “computing,” “calculating,” “determining,” “estimating,” “evaluating,” “measuring,” etc., can refer to the actions and / or processes by which a computer or computing system or similar electronic computing device manipulates and / or transforms data represented as physical (e.g., electronic) quantities in the registers and / or memory of the computing system into other data similarly represented as physical quantities in the memory, registers, or other such information storage, transmission, or display devices of the computing system.

[0039] Embodiments of this disclosure may include devices for performing the operations described herein. These devices may be specifically constructed for the desired purpose, or may include general-purpose computers selectively activated or reconfigured by a computer program stored in a computer. Such computer programs may be stored in computer-readable storage media, such as, but not limited to, any type of disk (including floppy disks, optical disks, CD-ROMs, magneto-optical disks), read-only memory (ROM), random access memory (RAM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic cards or optical cards, or any other medium suitable for storing electronic instructions and capable of being coupled to a computer system bus.

[0040] The processes and displays presented herein are not inherently associated with any particular computer or other device. Various general-purpose systems may be used with the programs taught herein, or it may prove convenient to build more specialized devices to perform the desired methods. The desired architectures of many of these systems are clear from the following description. Furthermore, embodiments of this disclosure are not described with reference to any particular programming language. It will be understood that the teachings of this disclosure as described herein can be implemented using a variety of programming languages.

[0041] The aspects of this disclosure can be described in the general context of computer-executable instructions (such as program modules) that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The disclosed embodiments can also be practiced in a distributed computing environment, where the task is performed by a remote processing device linked via a communication network. In a distributed computing environment, program modules may reside on both local and remote computer storage media, including memory storage devices. Attached Figure Description

[0042] This document describes some embodiments of the present disclosure with reference to the accompanying drawings. The description, together with the drawings, enables those skilled in the art to understand how some embodiments can be practiced. The drawings are for illustrative purposes and are not intended to show structural details of the embodiments in more detail than necessary for a basic understanding of the present disclosure.

[0043] For clarity, some objects depicted in the accompanying drawings are not drawn to scale. Furthermore, two different objects in the same drawing may be drawn to different scales. In particular, the scale of some objects may be greatly exaggerated relative to other objects in the same drawing.

[0044] In the attached diagram:

[0045] Figure 1aA cross-section of a wafer according to some embodiments is schematically illustrated, the wafer being cut diagonally such that its structural elements are cut at different heights.

[0046] Figure 1b These are exemplary 2D images of a wafer diagonally cut according to some embodiments;

[0047] Figure 2a A cross-section of a wafer diagonally cut according to some embodiments is illustrated (assuming the structure remains unchanged on a hidden axis that is perpendicular to both axes);

[0048] Figure 2b An illustrative depiction of some embodiments Figure 2a A top view of a diagonally cut wafer, as illustrated in the image;

[0049] Figure 3 The algorithmic flow of wafer reconstruction and volume sampling according to some embodiments is illustrated.

[0050] Figure 4a These are exemplary 2D images of a wafer diagonally cut according to some embodiments;

[0051] Figure 4b An illustrative description of a method based on some embodiments is provided. Figure 3 2D cut images are used to create 3D point clouds;

[0052] Figure 4c According to some embodiments, from Figure 4a An example 3D point cloud obtained from a 2D image;

[0053] Figure 4d The reconstruction process of a cross-sectional view image from a 3D point cloud (PCD) and its grayscale is illustrated illustratively according to some embodiments.

[0054] Figure 4e An illustrative depiction of a device from which some embodiments are described Figure 4a The reconstructed section view obtained;

[0055] Figure 4f An illustrative description of the embodiments of the present invention is provided. Figure 4b Image segmentation was performed on the obtained cross-sectional view;

[0056] Figure 5 Delamination of a periodic wafer according to some embodiments is illustrated illustratively;

[0057] Figure 6 An illustrative algorithmic flow for wafer reconstruction and volumetric sampling of a periodic wafer according to some embodiments is described, including delamination of the wafer by multiple diagonal cuts to improve reconstruction resolution. Detailed Implementation

[0058] The principles, uses, and implementation methods of the teachings herein can be better understood by referring to the accompanying descriptions and figures. After reading the descriptions and figures presented herein, those skilled in the art will be able to implement the teachings herein without much effort or experimentation. In the figures, the same reference numerals refer to the same parts throughout the text.

[0059] As used herein, the term "diagonal angle" refers to the angle of incidence between the incident beam and the sample surface. According to some embodiments, the diagonal angle generally ranges from about 5° to 30° or from 10° to 25°.

[0060] As used in this article, the terms “compound angle” and “3D angle” refer to angles generated by rotating a plane about an axis that is not necessarily parallel to the principal axis.

[0061] As used herein, the terms “cutting” and “milling” are used interchangeably and refer to the exposure of the internal layers of a structural element.

[0062] As used herein, the term "site" may refer to a region of a wafer. According to some embodiments, the term "site" may refer to a chip of the wafer. According to some embodiments, the terms "site" and "region" are used interchangeably.

[0063] As used herein, the term “subset of sites” may refer to a small array (e.g., 1, 2, 3, 5, or 10 chips) of volumetric sampling and / or test sacrifice.

[0064] As used in this article, the term "point cloud" refers to a collection of data points defined in a three-dimensional coordinate system, where each point represents the spatial coordinates (x, y, and z) of a specific location on the surface of an object or within a scene. Data points may also include additional information such as color, intensity, or other attributes. Point clouds are commonly used to create highly detailed 3D models of real-world objects or environments.

[0065] As used herein, the terms “measuring” and “wafer metrology” are used interchangeably and refer to the precise measurement and analysis of various parameters on semiconductor wafers used in the production of integrated circuits and microelectronic devices. Key aspects of wafer metrology include measuring parameters such as, but not limited to, wafer thickness, surface roughness, film thickness, and composition. According to some embodiments, wafer metrology refers to measuring parameters of structural components (such as, but not limited to, transistors) on a wafer. According to some embodiments, wafer metrology includes critical dimension measurements to ensure the accuracy of features, overlay measurements for proper alignment between layers, and defect inspection for identifying and analyzing any imperfections on the wafer surface. According to some embodiments, metrology may be a volumetric sampling of the wafer, i.e., inspection of the wafer at 3D resolution.

[0066] Top-view image after diagonal cropping:

[0067] Now for reference Figure 1a The figure schematically illustrates a cross-section of a wafer 100 according to some embodiments, which is diagonally cut at a diagonal angle α (as illustrated by plane 101) so that its structural elements 105 are cut at different heights. xyz represents the coordinates of the cutting system, while x' and y' represent the coordinates of the structural (image) system. According to some embodiments, the cutting can be performed using a focused ion beam (FIB), a laser beam, or any other suitable tool.

[0068] Now for reference Figure 1b The figure shows a wafer (such as) diagonally cut according to some embodiments. Figure 1a Illustration 110 of the exemplified wafer 100. Constant depth lines are represented by dashed lines 112a and 112b. Below dashed line 112b, bulk silicon 114 is seen, while dashed line 112a represents the height of the deposited material 115 before dicing. According to some embodiments, imaging may be a scanning electron microscope (SEM), atomic force microscope (AFM), EDX, or any other imaging tool, or any other suitable imaging tool.

[0069] Now for reference Figure 2a The illustration schematically depicts a cross-section of a wafer 200 cut along a diagonal 201, such that layers 212 to 228 of the structural elements of the wafer, cut diagonally, are exposed. Figure 2b A 2D top view of a wafer 200 according to some embodiments is illustrated (illustrated by arrow 205), depicting layers (depths) 212' to 228'.

[0070] According to some embodiments, the captured image ( Figure 2b (This is an example of such an image) can be preprocessed before further analysis. According to some embodiments, preprocessing may include artifact removal, noise reduction, contrast enhancement, resolution adjustment, background subtraction, segmentation (optionally performed later), and / or any combination thereof. Each possibility is a separate embodiment.

[0071] Algorithm flow overview:

[0072] Now for reference Figure 3 This figure schematically depicts an algorithmic flow 300 of the disclosed method for wafer reconstruction and volumetric sampling according to some embodiments. It should be understood that while all steps of the method are shown as sequential, some steps may be performed in parallel or in reverse order. Those skilled in the art will readily understand which steps require sequential execution and which do not. It should also be understood that some of the steps may be optional (as indicated below).

[0073] The steps in process 300 may include:

[0074] Step 310: Use a 2D imaging tool (e.g., SEM) to collect / acquire data. According to some embodiments, the image may be a FIB-SEM image, i.e., an SEM image of a wafer diagonally cut using FIB; however, other types of imaging and / or other tools for diagonally cutting wafers are also contemplated and are therefore within the scope of this disclosure.

[0075] Step 320: (Optional): Preprocess the collected images to improve image quality, resize, rotate, etc. The main goal of this step is to make the input images conform to a standard format, such as standard SNR, standard rotation angle, etc.

[0076] Step 330: Create a 3D point cloud (PCD).

[0077] Step 340: Create a cross-sectional view from the PCD according to the desired specifications (e.g., YZ axis, conforming to X = 50nm to 60nm).

[0078] Step 350: (Optional): Perform image segmentation to separate relevant regions (e.g., nanosheets).

[0079] Step 360: (Hot Selection): Perform volumetric measurements (e.g., width of nanosheets) on the cross-sectional view image and / or on the image fragment.

[0080] According to some embodiments, process 300 may include a step of directly measuring the volume of the preprocessed image obtained in step 310. According to some embodiments, volume sampling may include statistical volume measurements of the unfolded data, such as, but not limited to, width. A non-limiting example could be the spacing of structures along an axis (e.g., the X-axis). Another non-limiting example could be performing geometric measurements of the structures. For example, the thickness of several structures in the image could be measured, and quadratic statistics (i.e., mean and standard deviation) could be calculated.

[0081] Point cloud creation:

[0082] Once the 2D image is ready, a 3D point cloud (PCD) is created.

[0083] Now for reference Figures 4a to 4c These figures illustrate the conversion of 2D images into 3D point clouds according to some embodiments. Figure 4a An exemplary 2D image 400 of a wafer diagonally cut according to some embodiments is shown. White dashed lines 401a to 401g are contour lines. The x, y, and z coordinates can be calculated based on: in the image ( Figure 4bThe x' and y' coordinates of pixels, the size of the spacing, and the geometry of the clipping and cutting planes are used to generate 3D point clouds. Figure 4c As illustrated; and prior knowledge of image parameters, such as the expected location of contour lines. x This indicates that when X is at the spacing P x The change in the z-axis value obtained when the middle is changed. y This indicates that when Y is at the spacing P y The change in the z-axis value obtained when the center changes. According to some embodiments, prior knowledge, such as the position of diagonal angles, compound angles, or contour lines, can be used to calculate the localization of each 2D pixel in a 3D point cloud.

[0084] about Figure 4c This diagram illustrates a 3D PCD, where point intensity can be set according to GL settings obtained at the X, Y, Z coordinates. Black points are unassigned. Advantageously, the PCD can be rotated according to a desired axis, thereby providing an indication of how the cross-sectional view looks on that axis.

[0085] Creating a section view:

[0086] Now for reference Figure 4d .

[0087] Since a PCD can comprise millions of pixels, creating a cross-sectional view directly from a PCD would be time-consuming and computationally demanding. Therefore, an indexing stage can optionally be used before the creation of the cross-sectional view image itself. During this optional stage, a 3D mesh can optionally be applied to index the PCD into a uniform structure with coarse resolution (e.g., 2 nm to 5 nm), such as... Figure 4d The examples given.

[0088] The table on the left, labeled 402, shows a non-restricted example of the coordinates of each point in the PCD and its corresponding grayscale (GL). The PCD points are then sorted using a 3D mesh. Thus, the black point 406 in mesh 404 represents a point in the PCD distributed within a cell of mesh 404. For example, points 1, 4, 15, and 266 (in the table) are all assigned to the same cell in mesh 404, while another cell contains only points 7 and 12.

[0089] For each specific section view (SV) pixel, such as pixel 408, denoted here as X, a sphere 410 (or other volume) centered on SV pixel 408 is created. It should be understood that sphere 410 only encompasses a sub-segment of the cell, also referred to as “correlated mesh cell” 412, as indicated by the shading of mesh 404. Only the PCD point (black dot), denoted here as 414, falling within sphere 410 is considered to determine the GL of SV pixel 408, where GL = F(SV_x,SV_y,SV_z,PcdX,PcdY,pcdZ,PcdGL). F is a function that obtains the desired SV pixel coordinates and the PCD point within sphere 410 and outputs the calculated GL, such as linear interpolation (linear interpolation) or any other suitable function.

[0090] In this way, GL computation in SV is based on only a small subset of points in the PCD located near the desired pixel, thereby reducing the computational load.

[0091] According to some embodiments, the sphere 410 may, for example, have a diameter of 2 nm to 5 nm. It should be understood that choosing a relatively small neighborhood (e.g., a sphere of 2 nm to 5 nm) is typically preferred in order to avoid long-range effects or process variation effects. As mentioned above, this also allows for reductions in computation time and power.

[0092] Now for reference Figure 4e The illustration schematically depicts the calculation of GL from each of the SV pixels (such as SV pixel 408) according to some embodiments (as per [reference]). Figure 4d The reconstructed cross-sectional view (also known as "cross-sectional view 450") obtained by the above.

[0093] Image segmentation:

[0094] A 450° section view is available. Figure 4e (As shown in the image) Image segmentation is performed, thereby facilitating the measurement of structural segments (such as segments 452a to 452c), such as Figure 4f As shown. According to some embodiments, the segmentation includes coarse segmentation. According to some embodiments, coarse segmentation can be performed using standard image segmentation algorithms. A non-limiting example could be using edge detection (e.g., the Canny edge detector) for segmentation and fitting the closed shape (of the detected lines) to a function, such as a rectangle. Furthermore, AI algorithms can also be used to segment the image. According to some embodiments, the segmentation includes fine / deep segmentation.

[0095] Volume measurement:

[0096] According to some embodiments, various volumetric and statistical measurements can then be performed on the reconstructed cross-sectional view (such as cross-sectional view 450 and portions thereof). A non-limiting example could be a measurement... Figure 4fThe thickness and width of each segment 452a to 452c in the diagram. According to some embodiments, volumetric and statistical measurements may also be performed on the original 2D input image (e.g., an input SEM image) and / or on the PCD and / or on a virtual 3D model of the structure generated from the PCD.

[0097] Remove layers:

[0098] According to some embodiments, the obtained images may include images captured after more than one round of diagonal cutting (e.g., 2 rounds, 3 rounds, 4 rounds, 5 rounds or more), which is also referred to herein as “de-layering” of the wafer, i.e. obtaining multiple images from several cuts of the same field of view (FOV).

[0099] According to some embodiments, delamination can be performed to improve the resolution of the reconstructed wafer including periodic structural elements 500, such as... Figure 5 As illustrated. This is because the same height can be imaged multiple times, thereby improving z-resolution. For example, a layer may be exposed by a first diagonal cut 510a (selected from diagonal cuts 510a to 510c) of structural element 550f (selected from structural elements 550a to 550h), and then exposed again by a diagonal cut 510b (selected from diagonal cuts 510a to 510c) of structural element 550b (selected from structural elements 550a to 550h).

[0100] Therefore, in this case, the point cloud includes far more pixels because it includes all the 2D coordinates of all pixels in the 2D image obtained after each diagonal cut.

[0101] Algorithm flow for layer removal:

[0102] Now for reference Figure 6 This figure schematically depicts an algorithmic flow 600 of the methods disclosed herein for high-resolution wafer reconstruction and volumetric sampling of periodic wafers, according to some embodiments. It should be understood that while all steps of these methods are shown as sequential, some steps may be performed in parallel or in reverse order. Those skilled in the art will readily understand which steps require sequential execution and which do not.

[0103] In step 602, an image is obtained after each of the multiple diagonal cuts of the wafer, illustratively shown here as images 1 to k (it is also possible to have more than one image per cut). According to some embodiments, the image may be a FIB SEM image, i.e., an SEM image of a wafer diagonally cut multiple times using FIB; however, other types of imaging and / or other tools for diagonally cutting wafers are also contemplated and therefore within the scope of this disclosure.

[0104] In step 604, images 1 to k may optionally be preprocessed separately, for example by artifact removal, noise reduction, contrast enhancement, resolution adjustment, background subtraction, and / or segmentation.

[0105] In optional step 605, the PCD of each of images 1 to k is created and the cross-sectional view image is calculated.

[0106] In step 606, the images (images 1 to k (optionally preprocessed) obtained in step 602 or 604 or cross-sectional views obtained in step 605) may be registered to align the structural elements therein, for example by using markings located on the wafer (e.g., manual marking), cross correlation (algorithmic correlation between images) and / or navigation data for each image (x, y coordinates of the captured image), or other methods combining algorithms or prior knowledge user information.

[0107] In step 608, the methods disclosed herein, particularly as per [the relevant information], are used. Figures 4a to 4c The described method generates a uniform point cloud (PCD) from registered images (1 to k (raw or preprocessed) or cross-sectional views).

[0108] In step 610, a section view is created from the unified PCD generated in step 608 (the final creation of the section view), as described herein, particularly regarding Figure 4d and Figure 4e Described.

[0109] In step 612, the cross-sectional view may undergo segmentation, as described herein. Figure 4f As described, and in step 614, volume measurements may be performed on the cross-sectional view and / or segment.

[0110] It should be understood that Figure 6 The described process represents possible, but not exclusive, embodiments. This disclosure also covers other options for creating PCDs from 2D input images using registration data.

[0111] As used herein, the term "substantially" can be used to specify that a first property, quantity, or parameter is close to or equal to a second or target property, quantity, or parameter. For example, the first object and the second object can be said to have "substantially identical" lengths when the length of the first object is measured to be at least 80% (or some other predefined threshold percentage) and no more than 120% (or some other predefined threshold percentage) of the length of the second object. In particular, the statement that the first object and the second object have "substantially identical lengths" also covers the case where the first object has the same length as the second object.

[0112] According to some embodiments, the target quantity may refer to an optimal parameter, which can in principle be obtained using mathematical optimization software. Therefore, for example, a value assumed by the parameter can be said to be "substantially equal to" the maximum possible value that can be assumed by the parameter when the parameter value is at least 80% (or some other predefined threshold percentage) of the maximum possible value. In particular, the statement that a value assumed by the parameter is "substantially equal to" the maximum possible value that can be assumed by the parameter also covers the case where the parameter value is equal to the maximum possible value.

[0113] As used herein, the term "about" can be used to specify that the value of a quantity or parameter (e.g., the length of an element) is within a continuous range of values ​​adjacent to (and including) a given (statement) value. According to some embodiments, "about" can specify that the value of a parameter is between 80% and 120% of a given value. For example, the statement "the length of the element is about 1 m" is equivalent to the statement "the length of the element is between 0.8 m and 1.2 m". According to some embodiments, "about" can specify that the value of a parameter is between 90% and 110% of a given value. According to some embodiments, "about" can specify that the value of a parameter is between 95% and 105% of a given value.

[0114] According to some embodiments, as used herein, the terms “substantially” and “about” are used interchangeably.

[0115] It should be understood that certain features described in the context of individual embodiments of this disclosure for clarity may also be provided in combination in a single embodiment. Conversely, various features described in the context of individual embodiments of this disclosure for brevity may also be provided individually, or in any suitable sub-combination, or suitably provided in any other described embodiment of this disclosure. Unless expressly indicated otherwise, features described in the context of an embodiment should not be considered essential features of that embodiment.

[0116] Although the operations in the disclosed methods may be described in a specific order according to some embodiments, the methods of this disclosure may include some or all of the described operations performed in a different order. The methods of this disclosure may include several or all of the described operations. Unless expressly indicated otherwise, specific operations in the disclosed methods should not be considered as fundamental operations of the method.

[0117] Although this disclosure is described in conjunction with specific embodiments thereof, it will be apparent to those skilled in the art that numerous alternatives, modifications, and variations may be made. Therefore, this disclosure covers all such alternatives, modifications, and variations that fall within the scope of the appended claims.

[0118] It should be understood that the application of this disclosure is not necessarily limited to the details of the components and / or arrangements of the elements and / or methods described herein. Other embodiments may be practiced, and an embodiment may be carried out in various ways.

[0119] The wording and terminology used herein are for descriptive purposes and should not be construed as limiting. Any references cited or indicated in this application should not be construed as an admission that such references are provided as prior art to this disclosure. Section headings are used herein for ease of understanding and should not be construed as necessarily limiting.

[0120] While certain embodiments of the invention have been illustrated and described, it will be clear that the invention is not limited to the embodiments described herein. Numerous modifications, alterations, variations, substitutions, and equivalents will be apparent to those skilled in the art without departing from the spirit and scope of the invention as set forth in the appended claims.

Claims

1. A method for generating a representative 3D reconstruction of a periodic plurality of structural elements of a wafer, the method comprising: obtaining a 2D image of a top view of the plurality of structural elements, the plurality of structural elements being dihedrally sectioned with a complex angle that allows 3D volume sampling, such that each of the plurality of structural elements is cut at different heights thereof, and generating a representative 3D reconstruction of the structural elements based on the image.

2. The method of claim 1, the method comprising generating a point cloud based on 2D coordinates of each pixel in the image and based on one or more known parameters related to a plane of the dihedral sectioning with respect to the plurality of structural elements, wherein each point in the point cloud represents at least a location of each pixel in 3D space and its gray level (GL); creating at least one GL-reconstructed image of a cross-sectional view of the point cloud; and performing one or more volume measurements on the at least one GL-reconstructed image to characterize the plurality of structural elements or one or more components thereof.

3. The method of claim 1, the method further comprising segmenting the GL-reconstructed image using an image analysis algorithm to identify one or more components of the plurality of structural elements.

4. The method of claim 1, wherein the imaging of the top view of the dihedral sectioning is performed using a scanning electron microscope (SEM), an atomic force microscope (AFM), or a focused ion beam (FIB) imaging.

5. The method of claim 1, wherein creating the reconstructed image of the cross-sectional view comprises: a. 3D ordering the points in the point cloud along a grid with coarse resolution; b. for each pixel, finding the nearest surrounding points in the point cloud within a predefined radius; and c. calculating an output pixel GL based on the surrounding points.

6. The method of claim 5, wherein calculating the output pixel GL comprises a linear interpolation between the nearest surrounding points.

7. The method of claim 1, the method further comprising pre-processing the obtained 2D image and performing measurements thereon.

8. The method of claim 1, the method further comprising generating a 2D and / or 3D model of the plurality of structural elements based on a plurality of cross-sectional views derived from the point cloud.

9. The method of claim 1, wherein the one or more volume measurements comprise a width of a structure, a layer thickness, a height of a structure, a recess of a layer, a radius of a structure, an angle of a structure, or any combination thereof.

10. The method of claim 1, the method further comprising determining a quality / property of a wafer production line based on the volume measurements.

11. The method of claim 1, the method being suitable for implementation online into the wafer production line.

12. The method of claim 1, wherein the plurality of structural elements are dihedrally sectioned at least twice, wherein each subsequent dihedral sectioning of the dihedral sectioning is performed at a deeper plane than a previous dihedral sectioning, and wherein obtaining the top view of the dihedral sectioning comprises obtaining a top view image after each dihedral sectioning.

13. The method of claim 12, wherein generating the point cloud is based on the 2D coordinates of all pixels in each diagonal cut obtained 2D image.

14. The method of claim 13, wherein the distance between each of the diagonal cuts is about 5-10 nm.

15. The method of claim 12, further comprising cross-registration of the top view images of each of the diagonal cuts.

16. The method of claim 15, wherein the registration comprises utilizing one or more markers on the wafer, algorithmic correlation between the top view images, navigation data for each of the top view images, or any combination thereof.

17. A system for generating a representative 3D reconstruction of a plurality of periodic structural elements of a wafer, the system comprising: processing circuitry configured to: obtain 2D images of top views of the plurality of structural elements, the plurality of structural elements being diagonally dissected at a compound angle that allows for 3D volume sampling such that each of the plurality of structural elements is cut at different heights thereof; and generate a representative 3D reconstruction of the structural elements based on the images.

Images