Method and system for detecting and analyzing surface contamination particles of tsv wafer after cmp cleaning

Abstract

The application provides a TSV wafer surface contamination particle detection analysis method and system based on CMP cleaning, relates to the technical field of integrated circuits, and is used for improving the technical problem that the risk assessment of contamination particles is not accurate in the related art. The method comprises the following steps: obtaining a particle distribution map of a wafer to be detected; obtaining TSV layout data of the wafer to be detected; aligning the particle distribution map and the TSV layout data based on a preset reference mark to generate fusion data in a unified coordinate system; dividing the surface of the wafer to be detected into an analysis grid composed of multiple analysis units; and based on the fusion data, generating a TSV contamination score representing the contamination risk of a target TSV according to the size information of each contamination particle around the target TSV and the spatial distance between each contamination particle and the geometric center of the target TSV.

Description

Technical Field

[0001] This application relates to the field of integrated circuit technology, and in particular to a method and system for detecting and analyzing contaminant particles on the surface of a TSV wafer after CMP cleaning. Background Technology

[0002] In 3D integrated circuit manufacturing technology, through-silicon vias (TSVs) are a key structure for achieving vertical interconnects between chips. The manufacturing process of TSVs is complex, with chemical mechanical polishing (CMP) being the core process step for achieving wafer surface planarization. However, the CMP process and its subsequent cleaning processes inevitably leave micron- or even nanometer-sized contaminant particles on the wafer surface.

[0003] If these contamination particles happen to fall on or near the TSV structure, they can cause serious electrical problems, such as short circuits or open circuits between TSVs, or crosstalk in signal transmission, which can directly affect the performance and reliability of the chip, or even render the entire chip unusable.

[0004] In existing technologies, the detection of wafer surface contamination after CMP primarily relies on whole-wafer surface scanning equipment. These devices can quickly detect the number, size, and location distribution of surface particles and generate particle distribution maps. However, traditional analytical methods typically determine wafer cleanliness based on the total number or density of particles, or simply define a "kill zone" to determine if any particles have fallen directly above the TSV (Through-Seam Vessel). The drawback of this approach is its inability to finely assess the specific risk differences posed by particles of different locations and sizes to a particular TSV structure. For example, a small particle close to a TSV may pose a far greater risk than a larger particle far from all TSVs. This one-size-fits-all approach can easily lead to misjudgments of wafer quality, resulting in yield losses or potential reliability risks. Summary of the Invention

[0005] This application provides a method and system for detecting and analyzing contaminant particles on the surface of a TSV wafer after CMP cleaning, which improves the technical problem of insufficient accuracy in risk assessment of contaminant particles in related technologies.

[0006] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: In a first aspect, this application provides a method for detecting and analyzing contaminant particles on the surface of a TSV wafer after CMP cleaning. The method includes: acquiring a particle distribution map of the wafer to be inspected, the particle distribution map including the position coordinates and size information of each contaminant particle on the wafer surface; acquiring TSV layout data of the wafer to be inspected, the TSV layout data including the geometric center position coordinates of each TSV on the wafer surface; aligning the particle distribution map and the TSV layout data based on a preset reference mark to generate fused data in a unified coordinate system; dividing the surface of the wafer to be inspected into an analysis grid composed of multiple analysis units; and for each target TSV, based on the fused data, generating a TSV contamination score characterizing its contamination risk according to the size information of each contaminant particle around the target TSV and the spatial distance between each contaminant particle and the geometric center of the target TSV.

[0007] In one possible implementation of the first aspect, aligning the particle distribution map with the TSV layout data includes: identifying a first reference marker in the particle distribution map and a second reference marker in the TSV layout data; obtaining a coordinate transformation matrix based on the coordinates of the first reference marker and the second reference marker; and applying the coordinate transformation matrix to the position coordinates of the particle distribution map to align the coordinate system of the particle distribution map with the coordinate system of the TSV layout data.

[0008] In one possible implementation of the first aspect, the step of generating a TSV contamination score for the target TSV includes: for any contamination particle around the target TSV, determining its basic influence value based on the size information of the contamination particle; obtaining a distance attenuation factor based on the spatial distance between the contamination particle and the target TSV; obtaining a single particle contribution value of the contamination particle to the target TSV based on the basic influence value and the distance attenuation factor; and accumulating the single particle contribution values ​​of all contamination particles to obtain the TSV contamination score of the target TSV.

[0009] In one possible implementation of the first aspect, the distance attenuation factor is obtained through a preset attenuation function, wherein the standard deviation of the attenuation function is positively correlated with the size information of the contaminant particles.

[0010] In one possible implementation of the first aspect, after generating the TSV contamination score for the target TSV, the method further includes: obtaining the optical characteristic parameters of each contamination particle in the particle distribution map; matching a preset material risk coefficient to each contamination particle based on its optical characteristic parameters; and correcting the TSV contamination score based on the material risk coefficient to obtain an adjusted TSV contamination score.

[0011] In one possible implementation of the first aspect, the optical characteristic parameters include at least one of the light scattering intensity and reflectivity of the contamination particles.

[0012] In one possible implementation of the first aspect, the correction of the TSV pollution score based on the material risk coefficient includes: when calculating the contribution value of a single particle, using the material risk coefficient corresponding to the pollution particle as a weighting factor, and combining it with the basic impact value and the distance attenuation factor.

[0013] Secondly, this application also proposes a method for detecting and analyzing contamination particles on the surface of a TSV wafer after CMP cleaning. The system includes: a data acquisition module for acquiring particle distribution map and TSV layout data of the wafer to be inspected; a data alignment module for aligning the particle distribution map and the TSV layout data based on a preset reference mark to generate fused data in a unified coordinate system; a mesh generation module for dividing the surface of the wafer to be inspected into an analysis mesh composed of multiple analysis units; and a score generation module for generating a TSV contamination score characterizing its contamination risk for each target TSV based on the fused data, according to the size information of each contamination particle around the target TSV and the spatial distance between each contamination particle and the geometric center of the target TSV.

[0014] In one possible implementation of the second aspect, the score generation module is specifically used to: for any contamination particle around the target TSV, determine its basic influence value based on the size information of the contamination particle; obtain a distance attenuation factor based on the spatial distance between the contamination particle and the target TSV; combine the basic influence value and the distance attenuation factor to obtain the single particle contribution value of the contamination particle to the target TSV; and accumulate the single particle contribution values ​​of all contamination particles to obtain the TSV contamination score of the target TSV.

[0015] In one possible implementation of the second aspect, the system further includes: a risk correction module, used to obtain optical characteristic parameters of each contaminant particle after the score generation module generates the TSV contamination score, match a preset material risk coefficient for each contaminant particle according to the optical characteristic parameters, and correct the TSV contamination score based on the material risk coefficient to obtain an adjusted TSV contamination score. Attached Figure Description

[0016] Figure 1 A schematic flowchart of a pollution particle detection and analysis method provided for some embodiments of this application; Figure 2A schematic diagram of the structure of a pollution particle detection and analysis system provided for some embodiments of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0018] In the following description, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0019] Furthermore, in this application, directional terms such as "upper," "lower," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and may change accordingly depending on the orientation of the components in the accompanying drawings.

[0020] In this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "electrical connection" can refer to the manner in which an electrical connection is used to achieve signal transmission.

[0021] As used herein, “about,” “approximately,” or “approximately” includes the stated value and a reference value within an acceptable range of deviation from the given value, characterized in that the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the given quantity (i.e., the limitations of the measurement method).

[0022] This application provides a method and system for detecting and analyzing contamination particles on the surface of TSV wafers after CMP cleaning. The core of this approach lies in abandoning the traditional binary judgment logic based on the total number of particles or a simple "kill zone." Instead, it constructs a risk assessment model that comprehensively considers particle size, spatial location, and material properties by precisely integrating the physical particle scanning data of the wafer with electronic design automation (EDA) layout data. This model can generate a unique contamination score with clear physical meaning for each TSV structure on the wafer, thereby enabling precise location, quantitative classification, and root cause analysis of potential manufacturing defects.

[0023] like Figure 1As shown, the method includes: S101. Obtain the particle distribution map and TSV layout data of the wafer to be tested.

[0024] For example, a particle distribution map of the wafer to be inspected is obtained. In the semiconductor manufacturing process, after the wafer to be inspected completes the CMP (Chemical-Mechanical Polishing) planarization process and a series of subsequent complex cleaning processes (e.g., brushing and spray cleaning using megasonic energy and chemical cleaning solutions), it is sent to a high-precision surface scanning inspection device for full wafer inspection to ensure that the surface cleanliness meets the standards. Such devices, such as KLA-Tencor's Surfscan SPx series or Applied Materials' Enlight series, typically utilize dark-field or bright-field laser scattering technology. In dark-field mode, a laser beam is incident on the rotating wafer surface at an angle, and the detector is placed in a position that does not receive specular reflection light; when the beam encounters surface foreign objects (i.e., contaminant particles), it scatters in various directions, and some of the scattered light is captured by the detector. The device analyzes information such as the intensity and angular distribution of these scattered light signals, and through complex signal processing and image recognition algorithms, finally generates a particle distribution map.

[0025] This particle distribution map is typically stored in a standardized text file (such as CSV or TSV format) or a vendor-specific binary format. Its core information records the detailed properties of each contamination particle detected on the wafer. For example, the data structure of a particle distribution map could be a file containing... A table of row records, This represents the total number of particles detected on this wafer. Each row represents one contamination particle. (in (and must include at least the following key fields: an ID to uniquely identify the particle during data processing; the center position of the particle in the physical scan coordinate system, denoted as...) The origin and orientation of this coordinate system are determined by the scanning equipment during initial calibration; along with particle size information. Particle size is typically expressed as the "Latex Sphere Equivalent" (LSE) diameter, denoted as [missing information - likely a typology]. The physical meaning of LSE size is that the intensity of scattered light produced by an unknown particle is equivalent to the intensity of scattered light produced by a standard latex sphere particle of that diameter under the same conditions. Therefore, for any particle... Its core information can be abstracted into a triple containing position and size: .

[0026] Simultaneously, it is also necessary to obtain the TSV (Through Silicon Via) layout data for this wafer product. This data originates from the electronic design automation (EDA) stage of the chip. During production design, chip design engineers use EDA tools (such as Cadence Virtuoso or Synopsys Custom Compiler) to draw a layout containing all circuit elements and interconnect structures, and finally output it in an industry-standard format (such as GDSII or OASIS). For the analysis in this application, it is necessary to extract specific layers related to the TSV structure. This extraction process can be automated by a preprocessing script, and its steps generally include: reading the GDSII / OASIS file; identifying the layer number that defines the main body of the TSV structure (e.g., the metal fill layer of the TSV or the isolation layer around it) according to the design rules or technical documents (PDK) of the specific product; traversing all geometries (Polygon or Path, etc.) on the layer; for each geometry representing the TSV, using a geometric algorithm (such as the centroid algorithm) to calculate the position coordinates of its geometric center; and finally, outputting the geometric center coordinates of all TSVs into a simplified, easy-to-process data file. The resulting TSV layout data is a collection of... TSVs (of which A list of geometric center coordinates. Each TSV (denoted as...) The location information can be represented as It is important to emphasize that this coordinate system is located in the design layout coordinate system, and its origin and direction are defined by the design rules of the layout. It is independent of the aforementioned physical scanning coordinate system, so the two cannot be directly compared.

[0027] S102. Based on the preset reference markers, align the particle distribution map and TSV layout data to generate fused data in a unified coordinate system.

[0028] Because the physical coordinate system of the particle distribution map and the design coordinate system of the TSV layout data typically differ in origin, orientation, and scale due to translation, rotation, and (minor) scaling, coordinate alignment must be performed before spatial correlation analysis. The alignment process relies on pre-defined, physically and design-specific fiducial marks on the wafer. These marks are specially designed graphics with unique and easily identifiable geometric features (such as L-shapes, crosses, or more complex patterns). They exist on the photomask and are precisely transferred to specific locations on the wafer (typically at the wafer edge or within the scribe line) through a photolithography process.

[0029] For example, the corresponding coordinates of these reference markers are identified in two data sources. In the particle distribution map, the physical scan coordinates of at least two (in practice, three or more are often used to improve accuracy and perform error checking) first reference markers are identified by pattern recognition algorithms or directly by reports from the surface scanning device. For example, the coordinates of marker A are... The coordinates of marker B are Accordingly, during the TSV layout data processing stage, it is also necessary to extract the design coordinates of the corresponding second datum marker from the layout file. For example, the design coordinates of marker A are... The design coordinates of marker B are .

[0030] Based on these pairs of coordinate points (i.e.) Then, the affine transformation matrix that maps the physical coordinate system to the design coordinate system can be calculated. Affine transformations can handle translation, rotation, and scaling simultaneously. A two-dimensional affine transformation can be represented as:

[0031] In this expression, These are the original physical coordinates. These are the desired design coordinates after the transformation. Using known reference point pairs, an overdetermined linear equation system containing multiple equations can be established. Through numerical optimization methods such as the least squares method, the six transformation parameters that best fit all reference points can be solved. By using multiple reference points, measurement errors at individual marker points can be effectively smoothed out, improving the robustness of the overall registration.

[0032] The calculated coordinate transformation matrix Systematically applied to all particle distribution maps The original position coordinates of each pollution particle This allows us to obtain their new coordinates in a unified coordinate system (usually the design coordinate system as the reference). After this step, all contamination particles and all TSVs have position information in the same coordinate system.

[0033] S103. Divide the surface of the wafer to be tested into an analysis grid consisting of multiple analysis units.

[0034] For example, the effective analysis area of ​​the wafer is logically overlaid with a two-dimensional Cartesian grid. This grid consists of a large number of uniformly sized rectangular or square analysis cells. The size of the analysis cell is a configurable key parameter, and its selection requires a trade-off between query efficiency and memory consumption. If the cell is too large, each cell will contain too many particles, leading to decreased query efficiency; if the cell is too small, it will generate a large number of empty cells, wasting memory. Preferably, the side length of the cell is slightly larger than the particle influence radius to be defined in subsequent steps.

[0035] After meshing, each entity (i.e., each contamination particle) in the fused data needs to be mapped to its corresponding analysis cell. This mapping process can be accomplished using simple coordinate division and rounding operations. For example, the index of an analysis cell can be determined by its row and column numbers in the mesh. To represent, its calculation method is as follows and ,in These are the aligned coordinates of the particles, and These are the width and height of the cell, respectively. In terms of data structure, a hash table or a two-dimensional array can be used, where each entry corresponds to a cell and stores a list of particle IDs that fall into that cell. This spatial indexing structure allows for subsequent calculations where, when it's necessary to find particles surrounding a specific TSV, it's not necessary to traverse tens of thousands of particles across the entire wafer (complexity O(n log n)). Instead of directly determining the cell containing the TSV, we only need to first identify that cell and then check its immediate neighboring cells (e.g., the eight neighbors in the Moore neighborhood). This significantly reduces the time complexity of the search algorithm, bringing it close to constant time. This ensures the scalability of the entire analysis process for production-level data.

[0036] S104. For each target TSV, generate a TSV contamination score characterizing its contamination risk. This step is used to calculate a quantitative, comprehensive risk score, i.e., a TSV contamination score, for each TSV on the wafer. This score is not a simple "present / absent" judgment, but a continuous numerical value that reflects the level of risk.

[0037] For example, on the wafer Any one of the TSVs selected as the target TSV for analysis is denoted as . .for Define a calculation radius of influence And relevant particles are screened out. This radius is an empirical value determined based on process sensitivity; its physical meaning lies in defining the maximum range within which a particle might have a significant impact on TSV (such as stress, electric field, or chemical contamination diffusion). For example, it can be set to... .by geometric center With the center as the center, Using a radius of 1, define a circular potential influence region in the fused data space. Using the analysis grid established in the previous step, identify all contamination particles whose center points fall within this influence region. Assume a total of 100 particles are identified. Let the related particles be denoted as set. .

[0038] for Each of the related particles All of them need to be calculated for their effect on the target. The contribution value of a single particle ( ), denoted as This contribution value is determined by both the particle's inherent harmfulness (determined by its size) and its external spatial influence (determined by its distance from the TSV).

[0039] For example, quantifying the potential harm of the particles themselves (baseline impact value) Since the harmfulness of a particle is directly proportional to the area it can cover, a larger coverage area means a higher probability of short circuits or open circuits. Therefore, the base impact value... It can be expressed as the particle's equivalent diameter The square of:

[0040] In this model, larger particles are more likely to physically bridge TSVs with other structures, or break into more smaller particles in subsequent high-temperature processing steps, causing wider contamination.

[0041] After determining the fundamental influence of a particle, its spatial location must also be considered. The farther a particle is from the TSV, the less of a threat it poses. This attenuation relationship can be expressed by a distance attenuation factor. To indicate:

[0042] In this function, It is a particle center and center The Euclidean distance between them. This refers to the standard deviation of the function, which controls the rate of decay of the influencing field, i.e., the width of the influence range. This application uses the standard deviation... Set to the size of the particle itself It exhibits a positive correlation, rather than a fixed value that is the same for all particles. Larger particles have a wider potential range of influence. For example, a linear relationship can be established as follows to determine... :

[0043] in, It is a dimensionless proportionality constant, which can be understood as a "scaling factor for the range of influence"; while It is a basic standard deviation used to provide a minimum influence halo for extremely small particles, preventing their influence range from becoming zero. and These are all empirical parameters calibrated and optimized by analyzing a large number of known failure cases (i.e., TSV failures known to be caused by a certain particle) and good product data, using statistical regression or machine learning methods.

[0044] For example, in order to determine the parameters and The following calibration process can be performed: First, select several known particles of a specific size. At a specific distance The TSV samples that caused the failure are identified, and their risk values ​​are assigned a high normalized value (e.g., 1.0). Simultaneously, a large number of good TSV samples are selected, and their risk values ​​are set to 0. Then, optimization algorithms such as least squares or gradient descent are used to solve for the parameters that best fit these known risk values. and For example, by performing linear regression analysis on a set of data containing 10 failed samples and 1000 good samples, a set of empirical values, such as a dimensionless proportionality constant, may be obtained. and the basic standard deviation These calibrated parameters are then embedded into the model of the analysis system for risk assessment of unknown wafers.

[0045] When the base influence value and distance decay factor Once all the particles are identified, they can be combined to obtain particles. right The final :

[0046] It will affect all areas within the region. The related particles Perform linear accumulation to obtain the final target TSV. This accumulation process reflects the superposition effect of risks, that is, if there are multiple particles around a TSV, the total risk it faces is the sum of the individual risks of these particles.

[0047]

[0048] By analyzing all of the wafer Each TSV repeatedly executes the above S104 process, eventually generating a This is a result set that corresponds one-to-one with the TSV layout data, where each TSV is assigned a unique, quantified attribute. Value. Based on this map, various analyses can be performed, such as: analyzing all TSVs by... The values ​​are sorted in descending order to directly identify the top-N TSVs with the highest risk, guiding subsequent costly physical failure analysis; alternatively, all TSVs can be mapped onto the wafer map in the form of a heatmap. The values ​​are visualized and rendered to intuitively identify process areas (Hot Spots) with systemic contamination problems.

[0049] In real semiconductor manufacturing environments, a 0.5 μm copper (Cu) particle, due to its high conductivity and easy diffusion, poses a far greater threat to the electrical performance of TSVs than a similarly sized, chemically stable silicon dioxide (SiO2) particle or an insulating organic polymer residue. Therefore, the consideration of particle material properties needs to be incorporated into risk assessment models to improve their accuracy and practicality.

[0050] For example, in the data acquisition phase, in addition to the particle's position and size information, it is also necessary to acquire the optical characteristic parameters of the contaminant particles. These optical characteristic parameters may include light scattering intensity and / or surface reflectivity. These parameters can serve as an effective indirect "fingerprint" for inferring the particle's material composition.

[0051] After obtaining the optical property parameters, a preset material risk coefficient can be matched to each particle based on these parameters. For example, a mapping relationship can be established from the optical property parameter space to the material category and then to the specific risk coefficient value.

[0052] To enable those skilled in the art to implement this, a specific example of a mapping rule is provided below. It is assumed that the surface scanning device can provide the normalized reflectance for each particle. and normalized scattering intensity By calibrating the particles of known materials, a look-up table can be established to determine the material risk factor. :

[0053] During the analysis, the system obtained the first... Optical parameters of individual particles and Then, a unique material risk coefficient can be matched to it based on this lookup table. .

[0054] It's important to note that this mapping relationship (i.e., the lookup table) can be embedded in a pre-defined database or "recipe." The creation of this database typically requires controlled experiments (e.g., deliberately depositing known types of particles on a clean wafer) and physical verification using high-order analytical tools (such as scanning electron microscopy / energy-dispersive X-ray spectroscopy, i.e., SEM / EDX) to build a reliable training dataset. Then, using this dataset, classification rules are established and optimized through statistical methods or machine learning models (such as support vector machines, decision trees, etc.).

[0055] Ultimately, the obtained material risk coefficient Its value reflects the potential hazard level of the material and will serve as a key weighting factor for correcting the base TCS value.

[0056] Therefore, the revised , recorded as Its calculation formula has been updated to:

[0057] Accordingly, the final TSV pollution score, adjusted for material properties, was obtained. Its calculation formula is the sum of the corrected contributions of all relevant particles:

[0058] By introducing a correction mechanism, in the risk assessment model, the contribution value of a small particle made of a high-risk metal will be significantly amplified, while the risk contribution of a larger benign particle will be moderately reduced, thus making the final risk assessment result more correlated with the probability of electrical failure.

[0059] This application also provides a TSV wafer surface contamination particle detection and analysis system, which can be a standalone software package deployed on a server or a functional module integrated into a wafer fab manufacturing execution system (MES) or yield management system (YMS).

[0060] like Figure 2As shown, the system includes: a data acquisition module for acquiring particle distribution map and TSV layout data of the wafer to be inspected; a data alignment module for aligning the particle distribution map and the TSV layout data based on a preset reference mark to generate fused data in a unified coordinate system; a mesh generation module for dividing the surface of the wafer to be inspected into an analysis mesh composed of multiple analysis units; and a score generation module for generating a TSV contamination score characterizing the contamination risk of each target TSV based on the fused data, according to the size information of each contamination particle around the target TSV and the spatial distance between each contamination particle and the geometric center of the target TSV.

[0061] For example, the score generation module is specifically used to: for any contamination particle around the target TSV, determine its basic influence value based on the size information of the contamination particle; obtain a distance attenuation factor based on the spatial distance between the contamination particle and the target TSV; combine the basic influence value and the distance attenuation factor to obtain the single particle contribution value of the contamination particle to the target TSV; and accumulate the single particle contribution values ​​of all contamination particles to obtain the TSV contamination score of the target TSV.

[0062] For example, the system may further include a risk correction module, which is used to obtain the optical characteristic parameters of each pollutant particle after the score generation module generates the TSV pollution score, match a preset material risk coefficient for each pollutant particle according to the optical characteristic parameters, and correct the TSV pollution score based on the material risk coefficient to obtain an adjusted TSV pollution score.

[0063] 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.

[0064] 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.

[0065] 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.

[0066] 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.

[0067] 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 contaminant particles on the surface of a TSV wafer after CMP cleaning, characterized in that, The method includes: Obtain a particle distribution map of the wafer to be inspected, the particle distribution map including the position coordinates and size information of each contaminant particle on the wafer surface; The TSV layout data of the wafer to be tested is obtained, and the TSV layout data includes the geometric center coordinates of each TSV on the wafer surface; Based on preset reference markers, the particle distribution map and the TSV layout data are aligned to generate fused data in a unified coordinate system; The surface of the wafer to be tested is divided into an analysis grid consisting of multiple analysis units; For each target TSV, based on the fused data, a TSV pollution score characterizing its pollution risk is generated for the target TSV according to the size information of each contaminant particle around the target TSV and the spatial distance between each contaminant particle and the geometric center of the target TSV. Specifically, for any contaminant particle around the target TSV, its basic influence value is determined based on the particle's size information; a distance attenuation factor is obtained based on the spatial distance between the contaminant particle and the target TSV; a single particle contribution value of the contaminant particle to the target TSV is obtained based on the basic influence value and the distance attenuation factor; and the single particle contribution values ​​of all contaminant particles are summed to obtain the TSV pollution score of the target TSV.

2. The method according to claim 1, characterized in that, Aligning the particle distribution map with the TSV layout data includes: Identify the first reference marker in the particle distribution map and the second reference marker in the TSV layout data; Based on the coordinates of the first reference mark and the second reference mark, obtain the coordinate transformation matrix; The coordinate transformation matrix is ​​applied to the position coordinates of the particle distribution map to align the coordinate system of the particle distribution map with the coordinate system of the TSV layout data.

3. The method according to claim 1, characterized in that, The distance attenuation factor is obtained through a preset attenuation function, and the standard deviation of the attenuation function is positively correlated with the size information of the pollutant particles.

4. The method according to claim 1, characterized in that, After generating the TSV contamination score for the target TSV, the method further includes: Obtain the optical characteristic parameters of each pollutant particle in the particle distribution map; Based on the optical properties of each pollutant particle, a preset material risk coefficient is matched to it; Based on the material risk coefficient, the TSV pollution score is corrected to obtain the adjusted TSV pollution score.

5. The method according to claim 4, characterized in that, The optical characteristic parameters include at least one of the light scattering intensity and reflectivity of the pollutant particles.

6. The method according to claim 4, characterized in that, The correction of the TSV pollution score based on the material risk coefficient includes: When calculating the contribution value of a single particle, the material risk coefficient corresponding to the contamination particle is used as a weighting factor and combined with the basic impact value and the distance attenuation factor.

7. A system for detecting and analyzing contaminant particles on the surface of a TSV wafer after CMP cleaning, characterized in that, The system includes: The data acquisition module is used to acquire the particle distribution map and TSV layout data of the wafer to be inspected; The data alignment module is used to align the particle distribution map and the TSV layout data based on a preset reference mark, and generate fused data in a unified coordinate system. The mesh generation module is used to divide the surface of the wafer to be inspected into an analysis mesh composed of multiple analysis units; The score generation module is used to generate a TSV pollution score for each target TSV based on the fused data, according to the size information of each contaminant particle around the target TSV and the spatial distance between each contaminant particle and the geometric center of the target TSV. Specifically, for any contaminant particle around the target TSV, its basic influence value is determined based on the particle's size information; a distance attenuation factor is obtained based on the spatial distance between the contaminant particle and the target TSV; a single particle contribution value of the contaminant particle to the target TSV is obtained based on the basic influence value and the distance attenuation factor; and the single particle contribution values ​​of all contaminant particles are summed to obtain the TSV pollution score of the target TSV.

8. The system according to claim 7, characterized in that, The system also includes: The risk correction module is used to obtain the optical characteristic parameters of each pollutant particle after the score generation module generates the TSV pollution score, match a preset material risk coefficient for each pollutant particle according to the optical characteristic parameters, and correct the TSV pollution score based on the material risk coefficient to obtain an adjusted TSV pollution score.

Images