A method for detecting and analyzing trace etching effect of Si wafer based on IWC cleaning

Abstract

This application provides a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning, relating to the field of integrated circuit technology. It aims to improve the technical problem of poor performance in detecting and analyzing the micro-etching effect of Si wafers in related technologies. The method includes: obtaining a first optical characteristic distribution map of the Si wafer before IWC cleaning; obtaining a second optical characteristic distribution map of the Si wafer after IWC cleaning; generating an initial micro-etching depth matrix based on the first and second optical characteristic distribution maps; dividing the initial micro-etching depth matrix into multiple spatial sub-regions according to a preset geometric partitioning rule; obtaining all etching depth values ​​within each of the multiple spatial sub-regions and determining a representative etching depth value for that region; and generating a first etching effect analysis index characterizing the global radial uniformity of micro-etching on the Si wafer based on the representative etching depth values ​​of all spatial sub-regions.

Description

Technical Field

[0001] This application relates to the field of integrated circuit technology, and in particular to a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning. Background Technology

[0002] In the field of semiconductor integrated circuit manufacturing, silicon (Si) wafers are the most basic substrate material, and the cleanliness and integrity of their surfaces have a decisive impact on the performance and yield of subsequent processes and final devices. As device feature sizes continue to shrink to the nanometer level, the requirements for controlling contaminants on wafer surfaces are becoming increasingly stringent.

[0003] In-situ wet cleaning (IWC) is a key step in modern semiconductor manufacturing processes. It typically involves cleaning the wafer within an integrated process chamber using specific chemical solutions (such as SC-1, SC-2, etc.) to remove contaminants such as particles, metal ions, and organic matter from the surface. In many applications, IWC not only performs cleaning but is also designed for precise "micro-etching." This micro-etching aims to remove an extremely thin layer of material (typically a few to tens of angstroms, or 0.1 nanometers to several nanometers) from the wafer surface. Its purposes may include removing surface damage layers caused by processes such as ion implantation, adjusting surface morphology, or achieving precise thickness control of extremely thin functional layers.

[0004] Because the etching amount is extremely small and the uniformity requirement across the entire wafer surface is extremely high, how to quickly, accurately, and non-destructively detect and analyze the micro-etching effect after IWC cleaning has become a significant technical challenge. Traditional detection methods have many limitations. For example, measuring the mass change of the entire wafer before and after etching by weighing only yields the average etching amount across the entire wafer, failing to provide information on the uniformity of the etching. While using atomic force microscopy (AFM) or a profilometer to measure a few points on the wafer offers high precision, it is a destructive test (requiring the creation of steps), and the limited number of sampling points cannot provide a macroscopic distribution map of the entire wafer, making it difficult to effectively assess global uniformity. Summary of the Invention

[0005] This application provides a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning, which improves the technical problem of poor detection and analysis of micro-etching effect of Si wafers in related technologies.

[0006] To achieve the above objectives, the embodiments of this application adopt the following technical solutions:

[0007] This application provides a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning, comprising: acquiring a first optical property distribution map of the Si wafer under test before IWC cleaning; acquiring a second optical property distribution map of the Si wafer under test after IWC cleaning; generating an initial micro-etching depth matrix based on the first and second optical property distribution maps; dividing the initial micro-etching depth matrix into multiple spatial sub-regions according to a preset geometric partitioning rule; acquiring all etching depth values ​​within each of the multiple spatial sub-regions and determining a representative etching depth value for the region; and generating a first etching effect analysis index characterizing the global radial uniformity of micro-etching on the Si wafer based on the representative etching depth values ​​of all the spatial sub-regions.

[0008] In one possible implementation, generating an initial micro-etching depth matrix based on the first optical property distribution map and the second optical property distribution map includes: determining a spatial alignment transformation relationship based on wafer reference features in the first and second optical property distribution maps; applying the spatial alignment transformation relationship to spatially align the second optical property distribution map with the first optical property distribution map to generate an aligned optical property distribution map; and comparing the first optical property distribution map with the aligned optical property distribution map point by point to generate the initial micro-etching depth matrix.

[0009] In one possible implementation, the preset geometric partitioning rule is to divide the effective data area of ​​the initial micro-etching depth matrix into multiple concentric annular regions centered on the wafer center, and the multiple concentric annular regions are the multiple spatial sub-regions.

[0010] In one possible implementation, determining a representative etching depth value for a region includes: obtaining all etching depth values ​​within the spatial sub-region, calculating the median of all etching depth values, and using the median as the representative etching depth value for the region.

[0011] In one possible implementation, generating a first etching effect analysis index characterizing the global radial uniformity of the micro-etching of the Si wafer includes: obtaining the maximum and minimum values ​​of the representative etching depth values ​​of all the regions; combining the maximum and minimum values ​​to obtain a range value; and combining the range value with a benchmark value characterizing the overall etching depth to obtain the first etching effect analysis index.

[0012] In one possible implementation, the method further includes: determining a globally representative etching depth contour line based on the initial micro-etching depth matrix; obtaining the geometric perimeter of the globally representative etching depth contour line and the geometric area it encloses; and generating a second etching effect analysis index characterizing the anisotropy of micro-etching on the Si wafer based on the geometric perimeter and the geometric area.

[0013] In one possible implementation, determining a globally representative etching depth contour line includes: calculating the overall median of all etching depth values ​​in the initial micro-etching depth matrix; and extracting contour lines on the initial micro-etching depth matrix whose etching depth is equal to the overall median, as the globally representative etching depth contour line.

[0014] In one possible implementation, the method further includes: performing spatial smoothing filtering on the initial micro-etching depth matrix to generate a smoothed etching depth matrix; subtracting the initial micro-etching depth matrix from the smoothed etching depth matrix to obtain a deviation depth matrix; and identifying local etching anomalies in the deviation depth matrix based on a preset deviation threshold.

[0015] In one possible implementation, the method further includes: spatially clustering the local etching anomaly points to form at least one local etching anomaly region; and obtaining the geometric dimensions and location information of the at least one local etching anomaly region as a corrective etching effect analysis report. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning, 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] To make the objectives, technical solutions, and advantages of this invention clearer, the specific embodiments of this invention will be described in detail below. It should be understood that the specific embodiments described herein are one of the preferred implementations of this invention and are not intended to limit the entire scope of protection of this invention.

[0023] This invention provides a method for detecting and analyzing the micro-etching effect of Si wafers after IWC cleaning. This method aims to achieve precise, multi-dimensional quantitative characterization of the micro-etching effect on wafers through non-destructive optical measurements combined with a data processing workflow that emphasizes programmed and structured analysis. The core of this method lies in analyzing the uniformity and anomalies of etching from different physical dimensions through geometric partitioning, contour morphology analysis, and local deviation detection.

[0024] like Figure 1 As shown, the method includes:

[0025] S100: Obtain the first optical property distribution spectrum of the Si wafer under test before IWC cleaning.

[0026] In this preferred embodiment, a spectral ellipsometer (SE) is used as the measurement tool. Spectroscopic ellipsometer is a highly sensitive non-contact optical measurement technique capable of accurately characterizing thin film thickness and material optical constants. For bare silicon wafers about to undergo IWC processing, their surfaces are typically covered with a layer of natural silicon oxide (SiO2) with a thickness of 1-2 nanometers. The spectral ellipsometer measures the change in polarization state of polarized light after reflection from the wafer surface (determined by the amplitude-to-angle ratio). and phase difference angle (Description), which can accurately reflect the combined optical information of this natural oxide layer and the silicon substrate beneath it.

[0027] For example, the Si wafer to be tested is placed on the automated scanning platform of a spectroscopic ellipsometry. The system scans the wafer surface point by point according to a pre-set high-density two-dimensional grid. For example, for a wafer with a diameter of... The wafer can be set with a Cartesian coordinate grid covering the effective area of ​​the wafer, containing hundreds of thousands of measurement points. At each grid point, the spectroscopic ellipsometer measures a wide spectral range (e.g., ...). to Measurements were performed within the area to obtain the ellipsometric spectral data for that point. and .

[0028] After data acquisition, the spectral data at each measurement point is analyzed. For example, the measured data can be analyzed using an optical model (e.g., a model of monolayer SiO2 on a Si substrate). and The spectrum is converted into a single numerical value, such as the equivalent thickness of the surface oxide layer at that point.

[0029] For example, this analytical process can be achieved by establishing a three-layer optical model consisting of "environmental medium / silica layer / silicon substrate" and using nonlinear least squares regression analysis based on the Levenburg-Marquardt algorithm to fit and solve for the silica layer thickness, which is the only unknown variable in the model. It should be noted that extracting film thickness from ellipsometric spectral data using the aforementioned optical model and fitting algorithm is a conventional technique used by those skilled in the art and is an inherent function of standard analysis software in commercial ellipsometric measurement equipment; this application does not impose any limitations on this.

[0030] Organizing the values ​​obtained from all measurement points according to their two-dimensional spatial positions on the wafer constitutes the first optical property distribution map, denoted as . In terms of data structure, It can be viewed as a two-dimensional matrix, where the matrix elements That is, the first The equivalent surface thickness at each location point before cleaning.

[0031] S200: Obtain the second optical property distribution spectrum of the Si wafer under test after IWC cleaning.

[0032] After the baseline measurements are completed, the Si wafer is fed into an IWC device to perform a micro-etching process. This process uses a specific chemical solution (such as SC-1 solution) under precisely controlled conditions to treat the wafer surface and etch away an extremely thin layer of silicon material.

[0033] After the process is completed, the wafer is immediately or under controlled conditions transferred back to the same spectroellipsometry (or a cross-calibrated device of the same model) for a second full-wafer scan. To ensure the comparability of the two measurement results, all parameters of the second scan, including the definition of the scanning grid, spectral range, incident angle, etc., must be consistent with the first measurement in S100, thereby eliminating systematic errors introduced by changes in measurement conditions.

[0034] Similar to S100, after the second scan, the spectral data at each measurement point undergoes the same analytical processing to obtain a second optical property distribution map describing the surface state of the wafer after IWC treatment, denoted as... Similarly, It is also a two-dimensional matrix, whose elements Represents the first The equivalent thickness value of the surface after cleaning at each location point. Theoretically, due to the effect of micro-etching, [the following is likely a separate, unrelated statement:] The values ​​in will be systematically smaller than The value at the corresponding position in the text.

[0035] S300: Based on the first optical characteristic distribution map and the second optical characteristic distribution map, an initial micro-etching depth matrix is ​​generated. This step aims to correct for map misalignment caused by physical displacement and rotation of the wafer under test between two measurements, and then obtain a depth map reflecting the true etching amount through precise point-by-point comparison. Exemplarily, S300 includes:

[0036] S310. Based on the wafer reference features in the first optical property distribution map and the second optical property distribution map, determine a spatial alignment transformation relationship.

[0037] Although the same scanning procedure was used for both measurements, slight deviations in the wafer's position and angle on the measurement platform were unavoidable during the two loading and unloading processes. Without correction, directly loading... and Subtraction would misinterpret this geometric misalignment as severe etching inhomogeneity. Therefore, precise spatial alignment is essential. The alignment process is achieved by identifying and matching stable, invariant physical features present in both patterns. For example, for a standard Si wafer, its circular outline and positioning notch are ideal reference features.

[0038] For example, image processing algorithms (such as edge detection and Hough circle transform) are used to respectively... and The circular boundary of a wafer is detected. By performing a least-squares circular fitting on the boundary points, the circular boundary can be calculated. The center coordinates of the middle wafer and radius ,as well as The center coordinates of the middle wafer and radius .

[0039] Search for and locate wafer gaps on the boundary between the two maps. Gaps are typically represented as abrupt changes in boundary curvature, and their locations can be precisely identified.

[0040] By matching at least three non-collinear feature points (e.g., two circle centers, two notch vertices, etc.), an affine transformation matrix capable of describing translation, rotation, and minute scaling can be solved. This matrix represents the desired spatial alignment transformation relationship.

[0041] S320. Apply the spatial alignment transformation relationship to spatially align the second optical characteristic distribution map with the first optical characteristic distribution map to generate an aligned optical characteristic distribution map.

[0042] After obtaining the affine transformation matrix Afterwards, a coordinate-transformed and aligned object needs to be generated. The new, cleaned map that is perfectly aligned in space is denoted as... .

[0043] For example, traversing the target graph Each pixel coordinate Using the inverse matrix of an affine transformation Calculate the coordinates Original cleaned image The source coordinates corresponding to the middle .because These coordinates are typically not integers, and their values ​​need to be calculated using an interpolation algorithm. To strike a balance between preserving detail and computational efficiency, this embodiment preferably employs bilinear interpolation. Bilinear interpolation utilizes... The values ​​of the four nearest neighbor pixels are weighted and averaged to obtain a smooth and accurate interpolation result.

[0044] By performing an inverse transformation and interpolation operation on all target coordinates, a result was ultimately generated that is consistent with... Optical property distribution maps after alignment with the same size and coordinate system .

[0045] S330. Compare the first optical characteristic distribution map with the aligned optical characteristic distribution map point by point to generate the initial micro-etching depth matrix.

[0046] Obtaining spatially precisely aligned pre-cleaning maps and the image after cleaning Then, the actual etching depth at each point can be calculated.

[0047] Create a new two-dimensional matrix, namely the initial micro-etch depth matrix, denoted as . Its dimensions are the same as those of the aforementioned map. any position in the middle element value The following is obtained by performing a difference operation on the element values ​​at the same position in the two spectra:

[0048]

[0049] because and The element value represents the equivalent thickness, and this difference directly corresponds to the thickness of material removed at that point by the IWC process.

[0050] Understandably, what was ultimately obtained The matrix is ​​the foundation of all subsequent analyses, containing the most original and comprehensive spatial distribution information about the etching effect, but it is also mixed with the random noise inherent in the measurement process.

[0051] S400. According to a preset geometric partitioning rule, the initial micro-etching depth matrix is ​​divided into multiple spatial sub-regions.

[0052] Unlike methods that use global mathematical models for fitting, this application employs a geometric decomposition approach, breaking down the etching depth map of the entire wafer into several local regions with clear physical meaning to facilitate subsequent regional statistical analysis. Furthermore, this method does not presuppose any functional form of the etching distribution, but rather allows the data to "express itself" within a predefined geometric framework.

[0053] In this preferred embodiment, the geometric partitioning rule divides the effective data area of ​​the wafer into a series of concentric rings centered on the wafer center. This partitioning method closely matches the physical phenomenon of radial symmetry or radial variation that naturally forms due to wafer rotation in many wet processes.

[0054] For example, in In the matrix, determine the center pixel coordinates of the wafer. This coordinate was obtained during the alignment process of S310. A set of radius values ​​was preset. This is used to define the boundary of a series of concentric rings. It is usually 0. This corresponds to the maximum radius measured on the wafer. The selection of these radius values ​​can be based on process experience or analytical requirements. For example, for a... (radius For a wafer, the following zone boundaries can be defined: This divides the wafer surface into six spatial sub-regions (ring-shaped regions), each covering... , Equal radial range.

[0055] Traversal Each valid data point in the matrix (i.e., a point located inside the wafer). Calculate the distance from this point to the wafer center. radial distance .according to The value of determines which annular region the point belongs to. For example, if Then point and its corresponding etching depth value Classified as the first Each spatial sub-region.

[0056] Through this step, the original The matrix was reorganized into a series of data sets, each set corresponding to all etch depth data within a ring-shaped region.

[0057] For example, suppose the wafer is divided into 3 regions: the central region ( ), Central District ( ) and edge areas ( After the above steps, you will get three lists or arrays, each containing the etching depth values ​​of tens of thousands of points in these three regions.

[0058] S500: For each of the plurality of spatial sub-regions, obtain all etching depth values ​​within it, and determine a representative etching depth value for a region. After organizing the data into regions in S400, this step calculates a single value for each spatial sub-region that can represent the overall etching level of that region.

[0059] To ensure that the representative value is insensitive to local outliers within the region (which may be caused by measurement noise or minor defects), i.e., to be robust, this embodiment preferably uses the "median" as the representative etching depth value for the region. Compared to the "mean," the median is not affected by extreme maximum or minimum values ​​and can better reflect the central tendency of most data within a region.

[0060] For example, for each spatial sub-region obtained in S400 (e.g., the first... (A ring-shaped region) is defined, and all the etch depth values ​​contained within it are extracted to form a numerical sequence. This numerical sequence is then sorted, and the value at the middle position is found. This value is the median of the region, denoted as . .

[0061] Repeat this process to calculate a median for each spatial sub-region. Ultimately, this will result in a sequence of representative etching depth values ​​for each region: ,in This represents the total number of regions divided. This sequence clearly and quantitatively describes the macroscopic trend of etching depth as a function of radial distance.

[0062] For example, following the example of S400, the following results can be obtained by calculating the data of the three regions:

[0063] Median etching depth in the central area (Region 1) .

[0064] Median etching depth in the central area (region 2) .

[0065] Median etching depth of the edge region (region 3) .

[0066] This sequence Visually, the etching of this wafer exhibits a radial distribution characteristic of being "shallower in the center, deepest in the middle, and then shallower at the edges."

[0067] S600. Based on the representative etching depth values ​​of all the spatial sub-regions, generate a first etching effect analysis index characterizing the global radial uniformity of the micro-etching of the Si wafer.

[0068] After obtaining a representative value sequence describing the radial trend Next, this step condenses the information from this sequence into a single global uniformity index. This index is primarily used to evaluate the variation in etching depth at different radial locations.

[0069] This embodiment uses a normalization index based on range. The calculation process of this index is as follows:

[0070] In sequence Find the maximum value in the middle. and minimum value The difference between the two values ​​is the range of the representative values ​​for the region. This range directly reflects the maximum difference in macroscopic etching depth between different annular regions across the entire wafer.

[0071] To give the metrics relative meaning, a baseline value is needed to normalize the range. This baseline value should represent the overall etching level of the entire wafer. For example, the entire... The overall median of all effective etch depth values ​​in the matrix is ​​denoted as . .

[0072] range Compared with the benchmark value The final index is obtained by combining these parameters, such as defining it as the "radial uniformity index," denoted as... .

[0073]

[0074] this A smaller value indicates a smaller difference in etching depth between different annular regions, meaning better radial uniformity. Because its calculation is based entirely on the median, which is insensitive to outliers, this index exhibits high stability and reliability.

[0075] For example, following the example of S500, we have a sequence .

[0076] Maximum value .

[0077] Minimum value .

[0078] Range .

[0079] Assuming the overall median etch depth of the entire wafer is calculated .

[0080] Therefore, the first etching effect analysis index (radial uniformity index) is:

[0081]

[0082] This 8.22% is a final quantitative evaluation of the radial uniformity of the IWC process.

[0083] The processes described above, S100 to S600, primarily focus on the radial (or region-specific) uniformity of the etching depth. However, etching non-uniformity can also manifest as non-radial symmetry; for example, one side of the wafer may be etched faster than the other, a phenomenon known as anisotropy. To capture this non-uniformity, the method of this application further includes:

[0084] S700: Based on the initial micro-etching depth matrix, generate a second etching effect analysis index characterizing the anisotropy of micro-etching on the Si wafer. S700 includes:

[0085] S710. Based on the initial micro-etching depth matrix, determine a globally representative etching depth contour line.

[0086] Contour lines, also known as contour lines, are lines connecting all points on a map with equal numerical values. In this scenario, they connect all points on the wafer with the same etch depth. The shape of the contour lines visually reflects the pattern of the etch distribution. If the etching is perfectly radially symmetrical, then contour lines of any depth should be perfect circles centered on the wafer.

[0087] The goal of this step is to extract the most representative contour line, such as the contour line corresponding to the "median" etch depth of the entire wafer.

[0088] For example, the initial micro-etch depth matrix is ​​calculated. The overall median of all valid data points in the dataset is denoted as . This value has also been calculated in S600. The wafer surface is roughly divided into "over-etched" (depth > 1) based on the number of data points. ) and "under-etched" (depth < (Two parts)

[0089] Apply standard contour line generation algorithms (such as Marching Squares algorithm) to Extracting values ​​from the matrix that equal to... The contour lines. These are called the globally representative etching depth contour lines, denoted as . In practice, multiple closed or open curve segments may be generated, but usually only the longest closed curve is considered, as it represents the main macroscopic distribution boundary.

[0090] S720, Obtain the geometric perimeter and the geometric area enclosed by the globally representative etch depth contour lines. After obtaining the contour lines... After (usually represented as a series of connected vertex coordinates), this step is used to calculate its key geometric properties.

[0091] For example, the geometric perimeter of contour lines It is obtained by summing the lengths of all its segments. If the contour lines are composed of a vertex sequence... If the structure is such that the perimeter is:

[0092]

[0093] Geometric area enclosed by contour lines The area can be accurately calculated using polygon area calculation formulas (such as the shoelace formula).

[0094]

[0095] in, .

[0096] S730. Based on the geometric perimeter and the geometric area, generate a second etching effect analysis index that characterizes the anisotropy of micro-etching on the Si wafer.

[0097] Using the calculated perimeter and area This allows us to construct an index that quantifies how much a shape deviates from a circle. In geometry and image processing, a classic index is "circularity" or "shape factor".

[0098] This embodiment uses the following circularity index. As a second metric for etching effect analysis, it is used to characterize anisotropy:

[0099]

[0100] The indicators The numerical value of this index characterizes the degree of consistency between the shape of the contour lines and a perfect circle. The numerical behavior of this index follows the following relationship:

[0101] When the contour lines are perfect circles and The indicator The value is 1.

[0102] When the contour lines are of any non-circular shape, their perimeter will be "longer" relative to their area, resulting in The value is less than 1. The more irregular and elongated the shape, the lower its value. The closer the value is to 0.

[0103] therefore, The value directly quantifies the symmetry of the etching distribution. The closer the value is to 1, the closer the etching distribution is to the ideal radial symmetry, and the smaller the anisotropy. If the value is significantly less than 1, it indicates the presence of significant asymmetry, such as a significantly faster etching rate in one direction than in other directions.

[0104] For example, suppose we obtain the area of ​​a median contour line through S710 and S720. ,perimeter .

[0105] Therefore, the second etching effect analysis index (circularity index) is:

[0106]

[0107] This value of 0.83 indicates that there is a moderate degree of anisotropy in the etching distribution.

[0108] S800. Based on the initial micro-etching depth matrix, identify local etching anomalies on the Si wafer.

[0109] The above process primarily analyzes macroscopic, smooth uniformity. However, another key evaluation dimension of the IWC process is the ability to effectively control localized, sudden etching anomalies (such as "under-etched" points caused by particle masking, or "over-etched" points caused by surface defects). For example, S800 includes:

[0110] S810. Perform spatial smoothing filtering on the initial micro-etching depth matrix to generate a smoothed etching depth matrix. This step is used to generate a matrix that can represent... A smoothed graph representing the "background" or "local trend." This smoothed graph will serve as a benchmark for comparison, highlighting points that behave abnormally relative to their surroundings.

[0111] This embodiment uses a "median filter" for spatial smoothing. A median filter is a nonlinear filter that works by operating within a sliding window (e.g., ...). The median filter (a window of pixels) replaces the center pixel value with the median of all pixel values ​​within the window. Median filters excel at removing salt-and-pepper noise (i.e., isolated bright / dark spots) while preserving edge information well, making them ideal for generating a smooth yet not overly blurred background map.

[0112] Apply the median filter to the entire The matrix is ​​used to obtain the smoothed etch depth matrix, denoted as . . It can be understood as a point Typical etching depth in the local neighborhood.

[0113] S820. Subtract the initial micro-etching depth matrix from the smoothed etching depth matrix to obtain a deviation depth matrix.

[0114] By subtracting the original, noisy depth matrix from the smoothed background matrix, rapidly changing, high-frequency signals can be separated, which mainly include random noise and real local anomalies.

[0115] Create a new matrix, namely the deviation depth matrix, denoted as... :

[0116]

[0117] If the value in the value is close to zero, it indicates that the etching depth at that point is consistent with the surrounding environment. If it is a large positive value, it indicates that the point is an "over-etching" anomaly; if it is a large negative value, it indicates that the point is an "under-etching" anomaly.

[0118] S830. Based on a preset deviation threshold, identify local etching anomalies in the deviation depth matrix.

[0119] In order to To identify statistically significant outliers, rather than ordinary background noise fluctuations, a threshold needs to be set. A robust threshold setting method is based on... The statistical distribution characteristics of the data, especially the use of statistics that are insensitive to outliers. This embodiment uses a method based on Median Absolute Deviation (MAD).

[0120] For example, calculation The median of the matrix is ​​the sum of all its elements, which is theoretically close to 0. Then, the absolute value of the difference between each element value and this median is calculated, and the median of these absolute differences is the MAD.

[0121]

[0122] MAD is a very robust indicator for measuring the degree of dispersion of data.

[0123] Deviation threshold Set to a multiple of MAD.

[0124]

[0125] in It is a preset coefficient, usually taking a value between 3 and 7. The larger the value, the more "extreme" the outliers will be.

[0126] Traversal Matrix, if a point value If so, then mark the point as a local etching anomaly.

[0127] S840. Spatial clustering is performed on the local etching anomaly points to form local etching anomaly regions.

[0128] A single anomaly may still be noise, while a real physical defect typically affects a small area, forming a "cluster" of anomalies. Therefore, a spatial clustering analysis (such as the DBSCAN algorithm or simple connected component analysis) can be performed on all anomalies identified in the S830. Spatially adjacent anomalies are grouped into a "local etching anomaly region," and information such as the area, center location, and average anomaly deviation of this region is calculated. This information can form a corrective etching effect analysis report, providing process engineers with direct defect location and diagnostic basis.

[0129] Thus, the method provided by the embodiments of the present invention provides a macroscopic uniformity evaluation through radial partitioning analysis, reveals the anisotropy of the process through contour morphology analysis, and accurately locates local anomalies through filter-based deviation detection.

[0130] 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.

[0131] 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.

[0132] 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.

[0133] 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 the micro-etching effect of Si wafers after IWC cleaning, characterized in that, include: Obtain the first optical property distribution spectrum of the Si wafer under test before IWC cleaning; Obtain the second optical property distribution spectrum of the Si wafer under test after IWC cleaning; Based on the first optical property distribution map and the second optical property distribution map, an initial micro-etching depth matrix is ​​generated; According to the preset geometric partitioning rules, the effective data area of ​​the initial micro-etching depth matrix is ​​divided into multiple concentric annular regions with the wafer center as the center. These multiple concentric annular regions are multiple spatial sub-regions. For each of the plurality of spatial sub-regions, obtain all etching depth values ​​within it, calculate the median of all etching depth values, and use the median as the representative etching depth value of the region. Based on the representative etching depth values ​​of all the aforementioned spatial sub-regions, a first etching effect analysis index characterizing the global radial uniformity of the micro-etching on the Si wafer is generated, including: Obtain the maximum and minimum values ​​of the representative etching depth for all the aforementioned regions; The maximum value and the minimum value are combined to obtain a range value; The range value is combined with a benchmark value that characterizes the overall etching depth to obtain the first etching effect analysis index. The benchmark value is the overall median of all effective etching depth values ​​in the initial micro-etching depth matrix. The method further includes: Based on the initial micro-etching depth matrix, a globally representative etching depth contour line is determined, including: calculating the overall median of all etching depth values ​​in the initial micro-etching depth matrix; and extracting contour lines on the initial micro-etching depth matrix whose etching depth is equal to the overall median, as the globally representative etching depth contour line. Obtain the geometric perimeter of the global representative etching depth contour line and the geometric area it encloses; Based on the geometric perimeter and the geometric area, a second etching effect analysis index is generated to characterize the anisotropy of micro-etching on the Si wafer.

2. The method according to claim 1, characterized in that, The step of generating an initial micro-etching depth matrix based on the first optical characteristic distribution map and the second optical characteristic distribution map includes: Based on the wafer reference features in the first and second optical property distribution maps, a spatial alignment transformation relationship is determined. Applying the spatial alignment transformation relationship, the second optical characteristic distribution map is spatially aligned with the first optical characteristic distribution map to generate an aligned optical characteristic distribution map. The initial micro-etching depth matrix is ​​generated by comparing the first optical characteristic distribution map with the aligned optical characteristic distribution map point by point.

3. The method according to claim 1, characterized in that, The method further includes: The initial micro-etch depth matrix is ​​subjected to spatial smoothing filtering to generate a smoothed etch depth matrix. Subtract the initial micro-etching depth matrix from the smoothed etching depth matrix to obtain a deviation depth matrix; Based on a preset deviation threshold, local etching anomalies in the deviation depth matrix are identified.

4. The method according to claim 3, characterized in that, The method further includes: Spatial clustering is performed on the local etching anomalies to form at least one local etching anomaly region. Obtain the geometric dimensions and location information of the at least one local etch abnormality area as a corrective etching effect analysis report.

Images