TSV-based metal residue detection and analysis method and system

Abstract

The invention provides a TSV (Through Silicon Via)-based metal residue detection and analysis method and system, relates to the technical field of semiconductor manufacturing, and aims to solve the technical problem that spatial distribution characteristics of metal residues on the inner wall of a TSV cannot be accurately and quantitatively characterized in related technologies. The method comprises the following steps: acquiring an internal image of a to-be-detected TSV; generating a two-dimensional side wall projection drawing representing the inner wall of the TSV based on the internal image; identifying a metal residual area on the side wall projection drawing; and based on the spatial distribution characteristics of the metal residue area on the side wall projection drawing, a preset field potential model is applied to determine a spatial cohesion index representing the aggregation degree of the metal residues.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of semiconductor manufacturing, and particularly relate to a TSV metal residue detection and analysis method and system. BACKGROUND

[0002] In modern semiconductor manufacturing technology, three-dimensional (3D) integrated circuits have attracted wide attention due to their ability to significantly improve the integration of chips, performance and reduce power consumption. Through-silicon via (TSV) technology is one of the key technologies for realizing 3D integration. TSV is a small hole vertically etched in a silicon wafer, which establishes electrical connection between vertically stacked chips by depositing isolation medium and conductive metal (such as copper) on its inner wall and bottom.

[0003] The manufacturing process of TSV includes deep silicon etching, insulation layer deposition, barrier layer / seed layer deposition, and metal filling, etc. In these process steps, any problem in any link may cause the presence of undesirable metal residues on the inner wall of TSV. For example, incomplete cleaning after etching, or after chemical mechanical polishing (CMP), may leave a small amount of metal particles or films on the surface of the TSV insulation layer.

[0004] These metal residues pose a serious threat to the performance and reliability of the device. First, metal residues may act as charge trapping centers, affecting the electrical insulation performance of TSV, leading to increased leakage current or even electrical breakdown. Second, in subsequent high-temperature processes, these metal ions (especially copper ions) may diffuse into the surrounding silicon material, forming deep level traps, which seriously reduce the minority carrier lifetime and cause device performance degradation. In addition, large amounts of aggregated metal residues will also affect the uniformity and continuity of subsequent metal filling, which may cause filling voids and other defects.

[0005] Therefore, accurate and reliable detection and analysis of metal residues on the inner wall of TSV are crucial for ensuring TSV process quality and final product yield.

[0006] Existing detection methods mainly include physical analysis and electrical testing. Physical analysis methods, such as scanning electron microscopy (SEM) or transmission electron microscopy (TEM), can provide high-resolution images and intuitively see the presence of residues. However, these methods are usually destructive, require cutting and sample preparation of the wafer, and can only observe a limited number of TSV samples, making it difficult to conduct large-scale online monitoring. More importantly, traditional image-based analysis stops at qualitative judgment ("yes" or "no") or simple quantitative statistics (such as calculating the total area covered by residues), and cannot effectively distinguish the spatial distribution characteristics of residues. For example, a 1% residue coverage rate is composed of a large number of dispersed nanoparticles or a large cluster, which has a completely different potential hazard to the device, and the traditional method cannot quantify this difference.

[0007] Electrical testing methods, such as measuring the leakage current or capacitance of the TSV to indirectly infer the cleanliness of the inner wall, can achieve non-destructive batch detection, but the information provided is very indirect and cannot locate the specific position and form of the residue, and is easily disturbed by other factors, with limited diagnostic accuracy.

[0008] In summary, the existing technology makes it difficult for process developers to accurately assess the effectiveness of the cleaning process and to establish a quantitative relationship between residue characteristics and device reliability. SUMMARY

[0009] Embodiments of the present application provide a TSV metal residue detection and analysis method and system based on TSV metal residue detection and analysis method and system, to improve the technical problem that the spatial distribution characteristics of TSV inner wall metal residues cannot be accurately quantified and characterized in related technologies.

[0010] To achieve the above-mentioned purpose, the embodiments of the present application adopt the following technical solutions:

[0011] In a first aspect, the present application provides a TSV metal residue detection and analysis method, comprising: obtaining an internal image of a TSV to be tested; generating a two-dimensional sidewall projection map representing the inner wall of the TSV based on the internal image; identifying a metal residue area on the sidewall projection map; and based on the spatial distribution characteristics of the metal residue area on the sidewall projection map, applying a predetermined field potential model to determine a spatial cohesion index representing the aggregation degree of the metal residue.

[0012] In a possible implementation manner of the first aspect, the step of generating a two-dimensional sidewall projection map representing the inner wall of the TSV includes: establishing a cylindrical coordinate system of the inner wall of the TSV, the cylindrical coordinate system including an angle coordinate and a depth coordinate; mapping the TSV inner wall pixel points in the internal image from a three-dimensional Cartesian coordinate system to the cylindrical coordinate system, and generating the sidewall projection map with the angle coordinate and the depth coordinate as the axes.

[0013] In a possible implementation manner of the first aspect, before the preset field potential model is applied, the method further includes: identifying a plurality of discrete residual spots from the metal residual area; and determining a mass parameter and a mass center coordinate for each of the discrete residual spots.

[0014] In a possible implementation manner of the first aspect, the mass parameter is determined by taking an area of the corresponding discrete residual spot as the mass parameter.

[0015] In a possible implementation manner of the first aspect, the step of applying the preset field potential model to determine the spatial cohesion index includes: for each of a plurality of pairs of residual spots, obtaining a pairwise interaction potential value based on mass parameters of the two residual spots in the pair and a distance between the mass center coordinates of the two residual spots; and combining all the obtained pairwise interaction potential values to obtain the spatial cohesion index.

[0016] In a possible implementation manner of the first aspect, the step of obtaining the pairwise interaction potential value includes obtaining the pairwise interaction potential value based on a product of the mass parameters of the two residual spots and the distance between the mass center coordinates of the two residual spots.

[0017] In a possible implementation manner of the first aspect, the distance between the mass center coordinates is obtained based on periodicity of the sidewall projection map in the angle coordinate axis.

[0018] In a possible implementation manner of the first aspect, after the spatial cohesion index is determined, the method further includes: applying a preset perturbation to the metal residual area on the sidewall projection map to generate at least one perturbed projection map; determining at least one perturbed spatial cohesion index based on the at least one perturbed projection map; and obtaining a corrected spatial cohesion index based on the spatial cohesion index and the at least one perturbed spatial cohesion index.

[0019] In a possible implementation manner of the first aspect, the step of applying the preset perturbation to the metal residual area on the sidewall projection map includes: generating a random displacement vector conforming to a preset probability distribution for each pixel point in the metal residual area; and applying the random displacement vector to a position of the corresponding pixel point to generate the perturbed projection map.

[0020] In a second aspect, the present application also provides a TSV metal residue detection and analysis system, comprising: an image acquisition module configured to acquire an internal image of a TSV to be detected; a projection map generation module configured to generate a two-dimensional sidewall projection map representing an inner wall of the TSV based on the internal image; a region identification module configured to identify a metal residue region on the sidewall projection map; and an index determination module configured to determine a spatial cohesion index representing a degree of aggregation of the metal residue based on a spatial distribution feature of the metal residue region on the sidewall projection map by applying a preset field potential model. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 A flowchart of a TSV metal residue detection and analysis method is provided for some embodiments of the present application;

[0022] Figure 2 A structural diagram of a TSV metal residue detection and analysis system is provided for some embodiments of the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0024] Hereinafter, the terms "first", "second", etc. are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0025] In addition, in the present application, the orientation terms such as "up", "down", "left", "right", etc. can include but not limited to the orientation defined by the relative position of the components in the drawings. It should be understood that these directional terms can be relative concepts, which are used for relative description and clarification, and can be changed accordingly according to the change of the position of the components in the drawings.

[0026] In the present application, unless otherwise specified and limited, the term "connection" should be understood broadly, for example, "connection" can be fixed connection, or detachable connection, or integral; can be directly connected, or indirectly connected through intermediate medium. In addition, the term "electrical connection" can be the mode of electrical connection that realizes signal transmission.

[0027] As used herein, "about," "approximately," or "circa" includes the recited value and reference values within a range of acceptable deviation from the particular value, characterized by a range of acceptable deviation as determined by one of ordinary skill in the art taking into account the measurement being discussed and the error associated with the measurement of the particular quantity (i.e., limitations of the measurement method).

[0028] The embodiment of the present application provides a silicon through micro-via (TSV) metal residue detection analysis method. The method realizes accurate and repeatable evaluation of the metal residue condition by unfolding the inner wall of the TSV into a two-dimensional image and quantitatively analyzing the spatial distribution characteristics of the residue through a preset model. As shown in the figure, Figure 1 The TSV metal residue detection analysis method comprises the following steps:

[0029] S100, obtaining an internal image of a TSV to be measured.

[0030] The TSV to be measured is a key structure in semiconductor three-dimensional integration, and the cleanliness of the inner wall directly affects the quality of subsequent metal filling and the reliability of the device. Metal residues may be caused by etching, cleaning and other previous processes. This step aims to obtain original image data that can reflect the surface state of the inner wall of the TSV.

[0031] The internal image can be obtained by using various imaging techniques. For example, a scanning electron microscope (SEM) can be used to image the cross section of the TSV, or an optical microscope, a confocal microscope or the like can be used to scan and image from the opening end of the TSV. Alternatively, non-destructive X-ray microscopic imaging (XRM) or X-ray computed tomography (XCT) technology can also be used to obtain three-dimensional data of the internal structure of the TSV, and then extract two-dimensional slice images of the inner wall part.

[0032] The form of the image can be a digitized gray-scale image or a color image. For color images, a gray-scale processing is usually performed in the subsequent processing to convert the three-channel color information into single-channel brightness information to simplify the calculation. Each pixel point of the image corresponds to a physical position of the inner wall or internal space of the TSV, and the pixel value (gray value) thereof reflects the material characteristics of the position. Generally, metal residues will exhibit obvious contrast difference with the silicon substrate in the electron microscope or X-ray image due to their high atomic number or different surface morphology. For example, in the SEM secondary electron image, the metal residues usually appear as a brighter area than the silicon.

[0033] Exemplarily, it is assumed that a SEM is used to image a TSV with a diameter of and a depth of The TSV was observed, yielding a series of cross-sectional images taken from different depths. These images together constitute the raw data describing the internal condition of the entire TSV. Alternatively, a raw image of the TSV's internal wall, a "developed view," can be obtained using an endoscopic optical probe. Understandably, regardless of the source, the output of this step will be one or a set of digital images containing information about the TSV's internal wall, denoted as . .

[0034] S200. Based on the internal image, generate a two-dimensional sidewall projection image representing the inner wall of the TSV.

[0035] Performing spatial analysis directly on the three-dimensional inner wall of a cylinder would be extremely complex and computationally intensive. To simplify the calculations, this application "unfolds" the cylindrical inner wall of the TSV into a two-dimensional rectangular planar diagram, namely the Sidewall Projection Map (SPM). This process is similar to projecting a world map from a sphere onto a plane, thereby simplifying the subsequent calculations and analysis of the spatial location, distance, and distribution patterns of the residue.

[0036] For example, the position of each point on the inner wall of the TSV is determined from its Cartesian coordinate system in three-dimensional space. This can be transformed into a more convenient two-dimensional coordinate system for analysis. Specifically, a cylindrical coordinate system can be established for the TSV. ,in It is the radius. It's the angle. It's the depth. Because of the radius of the TSV. It is basically constant (denoted as ). Therefore, the position of any point on its inner wall can be determined by angle. and depth The only certainty.

[0037] The steps for generating the two-dimensional sidewall projection (SPM) may specifically include:

[0038] S210. Establish a cylindrical coordinate system for the inner wall of the TSV, the cylindrical coordinate system including angular coordinates. and depth coordinates .

[0039] S220. Map the TSV inner wall pixels in the internal image from the three-dimensional Cartesian coordinate system to the cylindrical coordinate system, using the angular coordinates. and the depth coordinates Using the axis as the axis, generate the sidewall projection diagram.

[0040] In S210, it is necessary to determine the central axis and radius of the TSV. For example, this can be achieved by image processing of the cross-sectional image obtained in S100, such as fitting the circular contour of the TSV using the Hough Circle Transform to determine its center and radius. The central axis is the line connecting the centers of all the cross-sections. Once the central axis is determined, a system can be established with that axis as its axis. Cylindrical coordinate system of axes. Angles The starting point ( The depth can be defined arbitrarily; for example, it can be defined as a radial direction pointing to a specific direction on the wafer (such as the notch). The starting point ( () can be defined as the opening of TSV.

[0041] In S220, for any pixel on the inner wall of the TSV, it may have a three-dimensional coordinate in the original image. Its angle in cylindrical coordinates can be calculated using geometric relationships. and depth .

[0042]

[0043]

[0044] in, These are the center coordinates of the TSV at that depth level. The function is a two-parameter arctangent function used to handle angles in all quadrants.

[0045] After mapping is complete, a new two-dimensional image, namely SPM, is generated, denoted as The horizontal axis of this image represents angles. (range from) arrive The vertical axis represents depth. (range from) arrive ,in (This is the total depth of TSV). Any point on SPM This corresponds to a point on the original inner wall of the TSV. ,in and Proportional and They are directly proportional. The pixel values ​​of the pixels on the SPM come from the pixel values ​​of the corresponding positions on the inner wall of the TSV in the original image.

[0046] For example, suppose the radius of the TSV is ,depth It can generate a resolution of SPM of pixels. The horizontal axis of SPM represents arrive Angle in degrees (each pixel represents 1 degree), vertical axis represents arrive Depth (represented per pixel) The original image shows an area located on the inner wall of the TSV at an angle of... Depth is The pixel value of a point at a given location will be assigned to the coordinates in the SPM. The pixels. In this way, the entire inner wall of the cylinder is "unfolded" into a complete rectangular map. This map completely preserves the shape, size, and relative spatial location information of all residues on the inner wall, thus aiding in subsequent quantitative analysis.

[0047] S300. Identify the areas of residual metal on the sidewall projection map. After obtaining a clear two-dimensional SPM, it is necessary to identify on this map which areas are residual metal and which are clean silicon substrates. In one specific embodiment, this step includes:

[0048] S310. The sidewall projection image is binarized to obtain a binarized projection image.

[0049] S320. In the binarized projection image, the set of pixels representing metal residue is determined as the metal residue region.

[0050] In S310, the purpose of binarization is to convert the grayscale SPM image... This is converted to a black and white image with only two pixel values ​​(e.g., 0 and 1, or 0 and 255). One value represents the background (silicon substrate), and the other represents the foreground (metal residue). The key to achieving binarization lies in choosing an appropriate threshold. All pixel values ​​higher than Pixels that are considered metal residue are assigned a value of 1; those below this value are considered metal residue. The value that is considered background is assigned a value of 0.

[0051] Threshold selection can employ a global thresholding method, applying the same threshold across the entire image. A commonly used adaptive global thresholding algorithm is the Otsu's method (maximum inter-class variance method). The Otsu's method automatically calculates a threshold that maximizes the inter-class variance of pixels between the foreground and background after segmentation. This method works well when there is significant contrast between the foreground and background. In some cases, due to factors such as image inhomogeneity, the brightness and contrast of different regions in the image processing unit (SPM) may vary. In such cases, an adaptive local thresholding method can be used, calculating an independent threshold for each small neighborhood of the image.

[0052] In S320, after binarization processing, a binarized projection image is obtained. The set of all pixels with a value of 1 in this image constitutes the defined area of ​​residual metal, denoted as [image of the area being processed]. .

[0053] For further analysis, it is usually necessary to... Perform morphological operations such as denoising and connecting broken regions. For example, opening operations (erosion followed by dilation) can be used to eliminate small isolated noise points, and closing operations (dilation followed by erosion) can be used to fill small voids inside residual regions or connect adjacent residual parts.

[0054] After completing this step, a clear "mask" image containing only information about the metal residue is obtained. Based on this mask, various geometric and spatial features can be extracted.

[0055] For example, suppose It is an 8-bit grayscale image (pixel value range 0-255). The optimal segmentation threshold calculated using the Otsu algorithm is: So, for Each pixel in Its binary projection image Corresponding pixel in the middle The value is:

[0056]

[0057] After processing, This forms a matrix of 0s and 1s. Pixels with a value of 1 constitute the area of ​​residual metal. .

[0058] S400. Based on the spatial distribution characteristics of the metal residue area on the sidewall projection map, a preset field potential model is applied to determine the spatial cohesion index characterizing the degree of metal residue aggregation.

[0059] Traditional residue detection typically stops at calculating the total residue area, but this metric cannot effectively distinguish between two fundamentally different residue states: one is a large number of isolated, randomly distributed small particles, and the other is a few but large clusters. These two states have very different potential impacts on device performance.

[0060] To address the aforementioned issues, this application introduces the Spatial Cohesion Index (SCI) to characterize the degree of "aggregation" or "cohesion" of residues. The calculation of the SCI is based on a field potential model constructed to characterize the spatial distribution of discrete target groups. The fundamental principle of this model is that the overall cohesion of a system composed of multiple discrete targets (in this case, residue spots) can be characterized by the sum of the pairwise interactions between the targets within the system. Specifically, larger and more spatially close target pairs contribute more significantly to the overall cohesion. A highly aggregated residue system, with stronger internal interactions, will be reflected by a characteristic SCI; conversely, a highly dispersed system, with weaker internal interactions, will be reflected by different numerical ranges of this index. By calculating this index, a deterministic indicator that directly quantifies the degree of cohesion in the residue system can be obtained.

[0061] The step of determining the spatial cohesion index using a pre-defined field potential model may specifically include:

[0062] S410. Identify a plurality of discrete residual spots from the metal residual region, and determine a mass parameter and a centroid coordinate for each discrete residual spot.

[0063] S420. For each pair of complex residual spot pairs, based on the mass parameters of the two residual spots in the pair and the distance between the centroid coordinates of the two residual spots, obtain a pairwise interactive potential energy value.

[0064] S430. Based on all the acquired pairwise interaction potential energy values, obtain the spatial cohesion index.

[0065] S410, Identify the spots and determine their parameters.

[0066] For example, the binarized residual region obtained by S300 The process involves identifying all independent, unconnected residual spots (or clusters). This can be achieved using a standard image processing algorithm, Connected-component labeling. This algorithm iterates through the binarized image, assigning the same unique label to all interconnected (e.g., connected via eight-neighborhood) foreground pixels (pixels with a value of 1). After execution, the result is:

[0067] The total number of discrete residual spots is denoted as Every spot (in A set of pixels.

[0068] Based on this, each discrete residual spot can be... (in Determine its characterization parameters: mass parameters and centroid coordinates .

[0069] In this embodiment, the mass parameter The area determined to be the spot This refers to the total number of pixels that make up the spot. This setting directly correlates the spot size with its contribution to subsequent interaction calculations; therefore, larger spots will have a greater impact on the final determined spatial cohesion index. The centroid is the geometric center of the spot, representing its average position on the SPM. Its coordinates can be obtained by calculating the average of the coordinates of all pixels within the spot:

[0070]

[0071]

[0072] In this way, the image of the unevenly distributed residue is abstracted into a shape composed of... A system composed of "particles", each particle All depend on their quality and location To describe.

[0073] S420, Obtain the potential energy values ​​of each pair of interactions.

[0074] After determining the mass parameters and centroid coordinates of each discrete residual spot, it is necessary to obtain the mass parameters of any two different spots (spots) in the system. and spots The pairwise interaction potential energy values ​​between ) are denoted as The interaction potential is defined as being proportional to the product of the mass parameters of the two spots and related to the distance between their centroids. This definition aims to quantify the spatial correlation strength between two spots, where a pair of spots with larger mass parameters and closer proximity will be assigned a potential value that significantly reflects their close correlation.

[0075] In one specific embodiment, the pairwise interactive potential values It can be obtained through the following formula:

[0076]

[0077] in, and It's a spot and Quality parameters (e.g., area) and ). It's a spot and center of mass and The distance between them. and It is a preset non-negative constant. In one embodiment, it can be set to... and To simplify the computational model, at this time As a normalization factor, and The decay rate of the effect of distance on the interaction is defined.

[0078] The interaction potential energy is positively correlated with the spatial cohesion between spots. That is, the closer the two spots are or the greater their mass, the larger the absolute value of the interaction potential energy and the smaller the algebraic value, which indicates a stronger cohesive effect.

[0079] Therefore, a preferred formula for the interaction potential energy is:

[0080]

[0081] A key technical detail is The calculation. Since SPM is obtained by unfolding a cylindrical surface, its left and right boundaries ( and The two centroids are physically connected. Therefore, calculating the two centroids... and When determining the distance between them, this "surrounding" topology must be considered. That is, the centroid... and Distance between Obtaining this requires considering the periodicity of the sidewall projection (SPM) along the angular coordinate axis. Specifically, the distance... The acquisition includes:

[0082] Get Distance component along the axis (depth) ;

[0083] Get Shortest distance component in the axial (angular) direction The method of obtaining it is at direct distance and orbital distance Take the minimum value between, where The total width of the SPM along the angular coordinate axis;

[0084] ;

[0085] Based on the distance component and Obtain the distance :

[0086]

[0087] By periodically processing the distance along the angular coordinate axis, the topological structure of the inner cylindrical surface of the TSV is accurately reflected, ensuring the correct quantification of the spatial relationship between any two residual spots.

[0088] S430, Obtain the Spatial Cohesion Index (SCI).

[0089] The spatial cohesion index (SCI) of the entire residue system is defined as the sum of the interaction potential energies between all different particle pairs, which is equivalent to the total potential energy of the particle system.

[0090]

[0091] In the above formula, the summation iterates through all unique pairs of spots. The final result It is a definite value.

[0092] The smaller the value, the more negative it is, indicating the presence of large (large area) and / or closely spaced spot pairs in the system, corresponding to a state of highly aggregated residues forming large clusters.

[0093] Values ​​close to zero or small negative numbers indicate that the spots in the system are either very small in mass (small in area) or very far apart, corresponding to a state in which the residue is highly dispersed and distributed in small particles.

[0094] Through calculation This method is superior to simple area statistics, providing a dimension that reflects the spatial distribution characteristics of TSV inner wall cleanliness for evaluation. This single value can be used to set process acceptance criteria, for example, specifying... It must be greater than a certain negative threshold (i.e., it cannot be too negative) to avoid the formation of large, invasive residues that could be highly detrimental to the device.

[0095] For example, there are two TSV samples, A and B, each with a total residual area of ​​1000 pixels.

[0096] Sample A: The residue consists of 100 small spots with an area of ​​10 pixels, randomly scattered on the SPM.

[0097] because and They are all small (all 10), and The average is relatively large, so each The absolute values ​​are all very small. It will be a relatively small negative number, such as -5000.

[0098] Sample B: The residue consists of two large spots, each 500 pixels in area, that are very close together.

[0099] . It's very small. It would be a very large negative number. Even with only one interaction pair, its It could reach -25000.

[0100] Despite having the same total area, The absolute value is much greater than Therefore, this method can clearly quantify that the residue in sample B is much more aggregated than that in sample A.

[0101] In some applications, the original image may contain noise, or the binarization process in the S300 may not be perfect, resulting in the identification of residual areas. The boundaries contain minute, non-physical jagged edges or burrs. These minute morphological variations can affect the identification of connected components (e.g., a large spot might be incorrectly segmented into two closely spaced smaller spots), thus affecting... and The calculation, and thus the final result Stability. In order to improve In addition to its robustness and repeatability, the method of this application also includes a correction mechanism based on physical perturbation.

[0102] This correction mechanism is based on the fact that metallic residues are not static at the microscopic level; the atoms on their surface undergo thermal vibrations. This microscopic perturbation can smooth out random errors introduced by imaging or segmentation algorithms, resulting in a more stable and physically accurate image. value.

[0103] Therefore, in a preferred embodiment, after determining the spatial cohesion index in S400, the method further includes:

[0104] S500: Apply a preset disturbance to the metal residue area on the sidewall projection map to generate at least one disturbed projection map.

[0105] This step aims to address the initially identified residual areas. Perform small, random deformations to simulate physical perturbations. Specifically, this can be done on... The position of each pixel in the image is shifted slightly by a small random displacement. This step may include:

[0106] S510. For each pixel in the metal residue area, generate a random displacement vector that conforms to a Gaussian distribution.

[0107] S520. Apply the random displacement vector to the position of the corresponding pixel to generate the perturbation projection map.

[0108] In S510, for The first in The original coordinates of the pixel are: Generate a two-dimensional random displacement vector for it. The two components of this vector and They are independent of each other and all share a common mean of 0 and a standard deviation of 0. Randomly selected from a Gaussian distribution (normal distribution).

[0109]

[0110] Standard deviation Controlling the intensity of the disturbance can be compared to "temperature" in a physical system. The value can be set based on the physical size of the pixel and the expected perturbation magnitude; for example, it can be set to 0.5 pixel units. This means that most of the displacement will be... Within a range of pixels.

[0111] In S520, this displacement vector is applied to the pixel to obtain its new, perturbed coordinates. :

[0112]

[0113]

[0114] because and It might be a floating-point number, so it needs to be rounded to the nearest integer pixel coordinate. After performing this operation on all the original residual pixels, a new set of pixels will be obtained, which constitutes a new, perturbed residual region. Its corresponding binarized image is a perturbation projection image. .

[0115] This process can be repeated. Each time, a new set of random numbers is used to generate... A series of independent perturbation projections . The value can be selected based on the required precision and computational cost, for example... .

[0116] S600. Based on the at least one perturbation projection map, determine at least one perturbation spatial cohesion index.

[0117] For each perturbation projection map generated in the previous step (in Each of these processes completely replicates the calculation process in S400. In other words, for each... :

[0118] Re-perform connected component analysis to obtain a new set of blobs, including a new total number of blobs. Area of ​​each spot and center of mass .

[0119] Using the formulas in S420 and S430, a corresponding perturbation space cohesion index is calculated. .

[0120] After this step, you will get Cohesion index of the perturbation space These values ​​will revolve around the original. The value fluctuates, and the magnitude of the fluctuation reflects the sensitivity of the original residual form to minor perturbations.

[0121] S700. Based on the spatial cohesion index and the at least one perturbed spatial cohesion index, obtain the corrected spatial cohesion index.

[0122] The original and A disturbance Combining these factors yields a final, more robust modified spatial cohesion index. For example, take its statistical average.

[0123]

[0124] Alternatively, the average of the perturbed values ​​can be used:

[0125]

[0126] this By averaging the results under multiple minor perturbation states, noise caused by the discreteness of digital images and the artificial boundary effects of segmentation algorithms is effectively smoothed out, and the results better reflect the intrinsic physical nature of the residue aggregation state. For example, if two spots are separated by chance in the segmentation algorithm but are physically close, a minor perturbation may cause them to merge into a single spot in some instances of perturbation, thus greatly changing the current situation. Value. By averaging, the effect of this "quasi-contact" state is reasonably included in the final correction index, thus providing a more reliable assessment in this application.

[0127] To implement the above method, this application also provides a metal residue detection and analysis system. This system can be a computer system integrating corresponding software, or a dedicated processing unit embedded in a detection device (such as a SEM). The system includes:

[0128] The image acquisition module can be an interface connected to imaging devices such as SEM and XRM, responsible for receiving and storing raw TSV internal images.

[0129] The projection map generation module, which incorporates a coordinate transformation algorithm, can read the original image, identify the TSV contour, and perform a mapping from 3D coordinates to 2D SPM, outputting the projection map. .

[0130] The region identification module includes image binarization algorithms (such as Otsu's method) and connected component analysis algorithms. It receives... As input, the output is a binarized residual region mask. and related characteristic parameters ( , , wait).

[0131] The index determination module receives feature parameters output by the region identification module and has built-in functions for calculating pairwise interaction potential energy and the final... The program, in particular, includes distance calculation logic to handle boundary conditions. Its output is a single numerical value. .

[0132] In a preferred embodiment, the system may further include a correction module:

[0133] The correction module receives the values ​​calculated by the exponent determination module. and the output of the region recognition module It has the functions of generating random perturbations, repeatedly calling the exponent determination module for calculation, and finally calculating the average value, outputting the final value. .

[0134] These modules can be in the form of software program code, stored in memory, and executed by a processor. They can also be in the form of hardware, such as application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs), to achieve faster processing speeds.

[0135] Compared with existing technologies, this method can not only assess the presence and quantity of residues, but also reveal their distribution patterns (dispersion or aggregation), providing support for quality control and process optimization in semiconductor manufacturing. Furthermore, the perturbation correction mechanism proposed in this application further enhances the stability and reliability of the method.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of this application.

[0137] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0138] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0139] The units described as separate components may or may not be physically separate. A component shown as a unit can be one physical unit or multiple physical units; that is, it can be located in one place or distributed in multiple different places. Depending on actual needs, some or all of the units can be selected to achieve the purpose of this embodiment.

[0140] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware.

[0141] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for detecting and analyzing metal residues based on TSV, characterized in that, include: Acquire the internal image of the TSV under test; Based on the internal image, a two-dimensional sidewall projection image representing the inner wall of the TSV is generated; Identify areas of residual metal on the sidewall projection diagram; Based on the spatial distribution characteristics of the metal residue area on the sidewall projection map, a preset field potential model is applied to determine the spatial cohesion index that characterizes the degree of metal residue aggregation.

2. The method according to claim 1, characterized in that, The step of generating a two-dimensional sidewall projection map characterizing the inner wall of the TSV includes: Establish a cylindrical coordinate system for the inner wall of the TSV, the cylindrical coordinate system including angular coordinates and depth coordinates; The TSV inner wall pixels in the internal image are mapped from the three-dimensional Cartesian coordinate system to the cylindrical coordinate system, and the side wall projection image is generated with the angular coordinates and the depth coordinates as axes.

3. The method according to claim 1, characterized in that, Before applying the preset potential model, the following is also included: Multiple discrete residual spots were identified from the metal residual region; For each discrete residual spot, determine a mass parameter and a centroid coordinate.

4. The method according to claim 3, characterized in that, The acquisition of the quality parameter includes determining the area of ​​the corresponding discrete residual spot as the quality parameter.

5. The method according to claim 3, characterized in that, The step of determining the spatial cohesion index using a pre-defined field potential model includes: For each pair of complex residual spot pairs, a pairwise interaction potential energy value is obtained based on the mass parameters of the two residual spots in the pair and the distance between the centroid coordinates of the two residual spots. All the obtained pairwise interaction potential energy values ​​are combined to obtain the spatial cohesion index.

6. The method according to claim 5, characterized in that, The step of obtaining the pairwise interaction potential energy value includes obtaining the pairwise interaction potential energy value based on the product of the mass parameters of the two residual spots and the distance between the centroid coordinates of the two residual spots.

7. The method according to claim 6, characterized in that, The distance between the centroid coordinates is obtained based on the periodicity of the sidewall projection along the angular coordinate axis.

8. The method according to any one of claims 1 to 7, characterized in that, After determining the spatial cohesion index, the process also includes: A preset perturbation is applied to the metal residue area on the sidewall projection map to generate at least one perturbed projection map; Based on the at least one perturbation projection map, at least one perturbation spatial cohesion index is determined; Based on the spatial cohesion index and the at least one perturbed spatial cohesion index, a corrected spatial cohesion index is obtained.

9. The method according to claim 8, characterized in that, The step of applying a preset disturbance to the metal residue area on the sidewall projection image includes: For each pixel in the metal residue area, a random displacement vector conforming to a preset probability distribution is generated; and the random displacement vector is applied to the position of the corresponding pixel to generate the perturbation projection map.

10. A TSV-based metal residue detection and analysis system, characterized in that, include: The image acquisition module is used to acquire internal images of the TSV under test. The projection image generation module is used to generate a two-dimensional sidewall projection image representing the inner wall of the TSV based on the internal image. The region identification module is used to identify areas of residual metal on the sidewall projection map; The index determination module is used to determine the spatial cohesion index, which characterizes the degree of aggregation of the metal residue, based on the spatial distribution characteristics of the metal residue area on the sidewall projection map and by applying a preset field potential model.

Images