Wafer pre-alignment method for a transport process

Abstract

The application relates to the technical field of wafer pre-alignment, and discloses a wafer pre-alignment method for a transmission process, which comprises the following steps: scanning the edge of a wafer on a rotating platform by a line laser sensor, obtaining a continuous data set of two-dimensional coordinate points of the wafer edge, filtering and removing abnormal points, and generating a discrete data set. The discrete data set is subjected to circle fitting based on a least square method, and the coordinates of the center of the wafer are accurately calculated; meanwhile, a Fourier transform is used to extract a cutting edge feature, and a Fourier inverse transform is used to determine a cutting edge angle. In this way, the speed and accuracy of pre-alignment are significantly improved. The method is particularly suitable for high-precision requirement scenes in semiconductor manufacturing, and provides a reliable positioning basis for subsequent processing steps.

Description

Technical Field

[0001] This application relates to the field of wafer pre-alignment technology, and more specifically, to a wafer pre-alignment method for a transfer process. Background Technology

[0002] During wafer transfer, pre-alignment is often required to provide a higher position for subsequent processing. Pre-alignment is a core issue that needs to be addressed in wafer transfer.

[0003] Pre-alignment involves finding the center and the cutting edge angle. To solve these problems, CCD, point laser, or line laser sensors are generally used to detect the wafer edges, and then algorithms are used to determine the center and cutting edge angle. Current technologies have drawbacks: 1. Point laser measurement requires multiple tests, which is time-consuming and has been gradually phased out; 2. CCD and line laser sensors are the industry's preferred choice due to their high accuracy. However, even with the same hardware system, accuracy and efficiency may vary. The software algorithm ultimately determines the accuracy and speed of pre-alignment. Currently, many machines on the market either require multiple rotations for pre-alignment, which is too time-consuming, or lack sufficient accuracy.

[0004] Based on this, this application provides a wafer pre-alignment method for the transfer process. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a wafer pre-alignment method for a transfer process, which significantly improves the speed and accuracy of pre-alignment. This method is particularly suitable for high-precision scenarios in semiconductor manufacturing, providing a reliable positioning basis for subsequent processing steps.

[0006] According to one aspect of this application, a wafer pre-alignment method for a transmission process is provided, comprising: controlling a line laser sensor to scan the wafer edge on a rotating platform to obtain a continuous dataset of two-dimensional coordinate points of the wafer edge; preprocessing the continuous dataset of two-dimensional coordinate points of the wafer edge to obtain a discrete dataset of two-dimensional coordinate points of the wafer edge; performing circle fitting on the discrete dataset of two-dimensional coordinate points of the wafer edge based on the least squares method to obtain the wafer center coordinates; and processing the discrete dataset of two-dimensional coordinate points of the wafer edge based on Fourier transform and inverse Fourier transform to obtain the cutting angle.

[0007] In the above-described wafer pre-alignment method for the transmission process, the control line laser sensor scans the wafer edge on the rotating platform to obtain a continuous dataset of two-dimensional coordinate points of the wafer edge, including: placing the wafer on the rotating platform; driving the rotating platform to move the wafer, and during the rotation of the wafer, the line laser sensor acquires a continuous dataset of two-dimensional coordinate points of the wafer edge.

[0008] In the above-described wafer pre-alignment method for the transmission process, preprocessing the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge includes: filtering the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a filtered dataset of two-dimensional coordinate points at the wafer edge; and removing outliers from the filtered dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge.

[0009] In the aforementioned wafer pre-alignment method for the transmission process, data filtering is performed on the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a filtered dataset of two-dimensional coordinate points at the wafer edge. This includes: determining a neighborhood window for each data point in the continuous dataset of two-dimensional coordinate points at the wafer edge; and updating the two-dimensional coordinates of the wafer edge for each data point based on the sample distribution of all data points within the neighborhood window of each data point, as expressed as:

[0010]

[0011]

[0012] in, The radius of the neighborhood window, The original two-dimensional coordinates of the wafer edge in the neighborhood window. This is an index variable used to iterate through all data points within the neighborhood window. The index of the current data point. These are the updated two-dimensional coordinates of the wafer edge.

[0013] The wafer pre-alignment method described above for the transmission process further includes: when updating the wafer edge two-dimensional coordinate points of each data point based on the sample distribution of all data points within the neighborhood window of each data point, introducing a neighborhood-based circular tangential fitting mechanism to optimize the updated wafer edge two-dimensional coordinate points.

[0014] In the aforementioned wafer pre-alignment method for the transmission process, a neighborhood-based circular tangential fitting mechanism is introduced to optimize the updated two-dimensional coordinate points of the wafer edge, including:

[0015] For each pair of coordinates obtained by weighting through the neighborhood window ( )and( First, let:

[0016]

[0017] To obtain the coordinates of each coordinate ( The corresponding tangential circle fitting parameter pair () );

[0018] Based on the tangential circle fitting parameter pair ( ), to analyze the data distribution within the neighborhood window ( Optimize:

[0019]

[0020]

[0021] in, and It is a scaling factor used to optimize fine-tuning. These are the optimized two-dimensional coordinate points at the edge of the wafer.

[0022] In the above-described wafer pre-alignment method for the transmission process, outlier removal is performed on the filtered dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge. This includes: determining the initial fitting center; calculating the radial distance between each data point in the filtered dataset of two-dimensional coordinate points at the wafer edge and the initial fitting center to obtain a set of radial distances; calculating the mean and standard deviation of the set of radial distances; calculating the standard score of each radial distance in the set of radial distances based on the mean and standard deviation; and determining whether to remove the two-dimensional coordinate points at the wafer edge corresponding to the radial distance based on a comparison between the standard score and a preset condition.

[0023] In the wafer pre-alignment method described above for the transmission process, calculating the standard fraction of each radial distance in the set of radial distances based on the mean and standard deviation includes: calculating the standard fraction of each radial distance using the following formula, expressed as:

[0024]

[0025] in, Radial distance, The mean, Standard deviation, Standard fraction representing radial distance.

[0026] In the wafer pre-alignment method described above for the transmission process, the preset condition is that the absolute value of the standard fraction is greater than 3.

[0027] In the aforementioned wafer pre-alignment method for the transmission process, the discrete dataset of two-dimensional coordinate points at the wafer edge is processed based on Fourier transform and inverse Fourier transform to obtain the slicing angle. This includes: arranging the discrete dataset of two-dimensional coordinate points at the wafer edge in the order of the wafer edge to obtain a wafer edge contour signal; performing a Fourier transform on the wafer edge contour signal to obtain a complex spectrum; extracting slicing features from the complex spectrum; mapping the slicing features back to the spatial domain through inverse Fourier transform to obtain the boundary point coordinates of the slicing edge; and determining the slicing angle based on the boundary point coordinates of the slicing edge.

[0028] Compared to existing technologies, the wafer pre-alignment method provided in this application for the transfer process scans the wafer edge on a rotating platform using a control line laser sensor to acquire a continuous two-dimensional coordinate point dataset. This dataset is then filtered and outlier-removed to generate a discrete dataset. A circle fit is performed on the discrete dataset using the least squares method to accurately calculate the wafer center coordinates. Simultaneously, Fourier transform is used to extract edge features, and inverse Fourier transform is used to determine the edge angle. This significantly improves the speed and accuracy of pre-alignment. This method is particularly suitable for high-precision scenarios in semiconductor manufacturing, providing a reliable positioning basis for subsequent processing steps. Attached Figure Description

[0029] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0030] Figure 1 This is a schematic flowchart of a wafer pre-alignment method for a transmission process according to an embodiment of this application.

[0031] Figure 2 This is a schematic flowchart of S1 in a wafer pre-alignment method for a transmission process according to an embodiment of this application.

[0032] Figure 3 This is a schematic diagram of a structural portion according to an embodiment of this application.

[0033] Figure 4 This is a schematic flowchart of S2 in the wafer pre-alignment method for the transmission process according to an embodiment of this application.

[0034] Figure 5 This is a schematic flowchart of S21 in a wafer pre-alignment method for a transmission process according to an embodiment of this application.

[0035] Figure 6 This is a schematic flowchart of S22 in the wafer pre-alignment method for the transmission process according to an embodiment of this application.

[0036] Figure 7 This is a graphical illustration of data obtained through preprocessing according to an embodiment of this application.

[0037] Figure 8 This is a schematic diagram of least squares method for fitting a circle according to an embodiment of this application.

[0038] Figure 9 This is a schematic flowchart of S4 in a wafer pre-alignment method for a transmission process according to an embodiment of this application.

[0039] Figure 10 This is a schematic flowchart of a specific embodiment of this application. Detailed Implementation

[0040] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0041] Based on this, this application provides a wafer pre-alignment method for the transmission process. Figure 1 This is a schematic flowchart of a wafer pre-alignment method for a transfer process according to an embodiment of this application. Figure 1 As shown, the wafer pre-alignment method for the transmission process includes: S1, controlling a line laser sensor to scan the wafer edge on a rotating platform to obtain a continuous dataset of two-dimensional coordinate points of the wafer edge; S2, preprocessing the continuous dataset of two-dimensional coordinate points of the wafer edge to obtain a discrete dataset of two-dimensional coordinate points of the wafer edge; S3, performing circle fitting on the discrete dataset of two-dimensional coordinate points of the wafer edge based on the least squares method to obtain the wafer center coordinates; S4, processing the discrete dataset of two-dimensional coordinate points of the wafer edge based on Fourier transform and inverse Fourier transform to obtain the cutting angle.

[0042] In one embodiment, such as Figure 2 As shown, in step S1, the line laser sensor scans the wafer edge on the rotating platform to obtain a continuous dataset of two-dimensional coordinate points of the wafer edge, including: S11, placing the wafer on the rotating platform; S12, driving the rotating platform to move the wafer, and during the rotation of the wafer, the line laser sensor acquires a continuous dataset of two-dimensional coordinate points of the wafer edge. Here, it can be seen that the structural part of the wafer pre-alignment method for the transmission process in this application uses a line laser plus a rotating platform, as shown... Figure 3As shown.

[0043] In step S11, the wafer is first correctly placed on the rotating platform. In this application, a robotic arm or manual operation is used to carefully place the wafer in the designated position, ensuring that its contact surface with the rotating platform is flat and without offset. This is because any deviation in the initial position will directly affect the accuracy of subsequent data acquisition.

[0044] In step S12, once the wafer is correctly placed, the next step is to start the rotating platform. At this point, the platform's rotation speed and stability are crucial. Ideally, the rotation speed should remain constant to ensure the line laser sensor can acquire data points at the wafer edge within uniform time intervals. This stability and consistency are fundamental to ensuring data quality. The line laser sensor is one of the core components of the entire system. Its function is to capture information from the wafer edge in real time and convert it into two-dimensional coordinate points. When the rotating platform starts rotating, the line laser sensor emits a narrow beam of light along the radial direction of the wafer. This beam intersects with the wafer edge, forming a series of reflected light points. The position information of these reflected light points is the two-dimensional coordinate point of the wafer edge. It is worth noting that the resolution and response time of the line laser sensor directly affect the accuracy and efficiency of data acquisition. Therefore, when selecting a sensor, a sensor that meets the requirements of high precision and high speed can be chosen based on actual production conditions. As the rotating platform continues to rotate, the line laser sensor continuously acquires data points from the wafer edge, forming a continuous data stream, that is, a continuous dataset of the two-dimensional coordinate points of the wafer edge.

[0045] In one embodiment, such as Figure 4 As shown, in step S2, the continuous dataset of the two-dimensional coordinate points at the wafer edge is preprocessed to obtain a discrete dataset of the two-dimensional coordinate points at the wafer edge, including: S21, data filtering of the continuous dataset of the two-dimensional coordinate points at the wafer edge is performed to obtain a filtered dataset of the two-dimensional coordinate points at the wafer edge; S22, outlier removal of the filtered dataset of the two-dimensional coordinate points at the wafer edge is performed to obtain a discrete dataset of the two-dimensional coordinate points at the wafer edge.

[0046] In step S21, after acquiring the continuous dataset of two-dimensional coordinate points at the wafer edge, the first step is data filtering. Since the data acquired by the sensor is inevitably affected by environmental noise, the raw data must be filtered to reduce noise interference and improve data quality. Specifically, low-pass filters or other types of filtering algorithms, such as Kalman filtering or median filtering, can be used. These filtering algorithms can effectively smooth data curves, remove high-frequency noise components, and make the data more stable and reliable.

[0047] In a specific embodiment, such as Figure 5As shown, S21, data filtering is performed on the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a filtered dataset of two-dimensional coordinate points at the wafer edge, including: S211, determining the neighborhood window of each data point in the continuous dataset of two-dimensional coordinate points at the wafer edge; S212, updating the two-dimensional coordinates of the wafer edge of each data point based on the sample distribution of all data points within the neighborhood window of each data point, as expressed as:

[0048]

[0049]

[0050] in, The radius of the neighborhood window, The original two-dimensional coordinates of the wafer edge in the neighborhood window. This is an index variable used to iterate through all data points within the neighborhood window. The index of the current data point. The updated two-dimensional coordinates of the wafer edge are shown below. Here, considering that the original acquired data often contains random noise caused by factors such as sensor accuracy limitations and environmental interference, this noise, if left unprocessed, will directly affect subsequent data analysis and fitting results. By averaging each data point with its neighboring data points, the randomly distributed noise components can be effectively "cancelled," resulting in a more stable and reliable dataset. Although this method is simple, it can significantly reduce local fluctuations without changing the overall trend of the data, laying a good foundation for subsequent processing steps. Furthermore, such smoothing not only helps reduce noise but also improves data consistency and continuity. In practice, due to measurement errors or other unforeseen factors, some data points may deviate from the overall trend. By integrating information from neighboring data points, these "abrupt" data points can be effectively corrected, making them more consistent with the overall trend.

[0051] In another specific embodiment, the continuous dataset of wafer edge two-dimensional coordinate points is filtered to obtain a filtered dataset of wafer edge two-dimensional coordinate points. The method further includes: when updating the wafer edge two-dimensional coordinate points of each data point based on the sample distribution of all data points within the neighborhood window of each data point, introducing a neighborhood-based circular tangential fitting mechanism to optimize the updated wafer edge two-dimensional coordinate points.

[0052] Here, when removing outliers from the filtered dataset of wafer edge 2D coordinate points using a standard score based on radial distance, considering that when updating the wafer edge 2D coordinates of each data point based on the sample distribution of all data points within the neighborhood window of each data point, the neighborhood window of each data point essentially includes both radial and tangential neighborhood directions, a neighborhood-based circular tangential fitting mechanism can be introduced to further improve the accuracy of outlier removal. That is, a neighborhood-based circular tangential fitting mechanism is introduced to optimize the updated wafer edge 2D coordinates, including: for each pair of coordinates obtained by weighting through the neighborhood window (… )and( First, let:

[0053]

[0054] To obtain the coordinates of each coordinate ( The corresponding tangential circle fitting parameter pair () ).

[0055] In this way, the tangential circle fitting parameters are used ( Based on the minimum enclosing circle, the data distribution within the neighborhood window is used to analyze (...). Optimize:

[0056]

[0057]

[0058] and It is a scaling factor used to optimize fine-tuning. These are the optimized two-dimensional coordinate points at the edge of the wafer.

[0059] In other words, by making the geometric properties of the fitting parameters of the tangent direction of the circle consistent with the data distribution trend in the local neighborhood, when removing outliers based on the radial distance of the fitted circle from the filtered dataset of the two-dimensional coordinate points of the wafer edge obtained by neighborhood calculation, the natural connection between the adjacent neighborhoods of the outliers can be achieved by smoothing the data through the radial fitting accuracy of the circle fitting, so as to avoid the overfitting of outlier removal affecting the accuracy of the tangent angle calculation.

[0060] In step S22, it is considered that although the filtered dataset is relatively clean, there may still be some outliers. These outliers may be caused by sensor malfunctions, measurement errors, or other unforeseen factors. If not handled, these outliers will seriously affect the subsequent fitting and calculation results. Therefore, this application then removes outliers from the filtered dataset.

[0061] In a specific embodiment, such as Figure 6 As shown, S22, outlier removal is performed on the filtered dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge, including: S221, determining the initial fitting center; S222, calculating the radial distance between each data point in the filtered dataset of two-dimensional coordinate points at the wafer edge and the initial fitting center to obtain a set of radial distances; S223, calculating the mean and standard deviation of the set of radial distances; S224, calculating the standard score of each radial distance in the set of radial distances based on the mean and standard deviation; S225, determining whether to remove the two-dimensional coordinate points at the wafer edge corresponding to the radial distance based on a comparison between the standard score and a preset condition.

[0062] Specifically, in the outlier removal process, a preliminary fitting center is first determined. This center can be estimated using simple geometric methods or the least squares method based on existing data points. Then, the radial distance from each data point to this preliminary fitting center is calculated, forming a set of radial distances. Next, the mean and standard deviation of this set are calculated, and these statistics are used to evaluate the reasonableness of each data point.

[0063] In one specific embodiment, a set of filtered wafer edge data points has been obtained, and a preliminary fitting center has been determined. Next, the radial distance from each data point to the center is calculated, and the mean and standard deviation of these distances are obtained. Then, according to the standard score formula (i.e., Z-score), a standard score for each radial distance is calculated. In this embodiment, the preset condition is that the absolute value of the standard score is greater than 3. If the absolute value of the standard score of a data point is greater than 3, the data point is considered an outlier and should be removed. This outlier removal method is based on statistical principles and can effectively identify and remove data points that deviate from the normal range. This not only improves the overall consistency of the data but also avoids fitting errors caused by individual outliers. Furthermore, setting the standard score threshold to 3 is also an empirical choice, because under a normal distribution, data points exceeding 3 standard deviations are considered extremely low probability events and are usually outliers.

[0064] In one specific embodiment, calculating the standard score of each radial distance in the set of radial distances based on the mean and standard deviation includes: calculating the standard score of each radial distance using the following formula, expressed as:

[0065]

[0066] in, Radial distance, The mean, Standard deviation, Standard fraction representing radial distance.

[0067] After data filtering and outlier removal, this application obtains a relatively clean discrete dataset of two-dimensional coordinate points at the wafer edge. However, this does not mean that the work of this application is finished. To further improve the fitting accuracy, the initial fitting results can be optimized. Specifically, an iterative method can be used to continuously adjust the fitting parameters until the optimal fitting effect is achieved. The advantage of this method is that it not only considers the initial fitting results but also further improves the fitting accuracy through iterative optimization. Especially when dealing with complex wafer edge data, this method can better cope with various possible sources of error and ensure the reliability of the final result. In a specific embodiment, preprocessing is used to obtain data such as... Figure 7 The data graph shown.

[0068] Specifically, in step S3, a circle fit is performed on the discrete dataset of two-dimensional coordinate points at the wafer edge using the least squares method to obtain the wafer center coordinates. The least squares method is a mathematical method widely used in curve fitting; its core idea is to find the optimal fitting parameters by minimizing the sum of squared errors. During wafer pre-alignment, the least squares method is used to fit the data points at the wafer edge, thereby determining the wafer center coordinates and radius. The least squares method fits the circle as shown below. Figure 8 As shown. The reason for using the least squares method for circle fitting is its good mathematical properties and wide applicability. The least squares method finds the best fitting parameters by minimizing the sum of squared errors. This method is not only simple and easy to implement, but also performs well when dealing with noisy data. Since wafer edge data usually contains a certain amount of noise and measurement error, the least squares method can effectively smooth out this noise and provide more accurate fitting results.

[0069] Specifically, there is a set of two-dimensional coordinate points at the edge of the wafer. ,in The goal is to find the equation of a circle. This allows the circle to best fit these data points. Here, and These represent the x and y coordinates of the center of the circle, respectively. This represents the radius of the circle.

[0070] The core of the least squares method lies in minimizing the following objective function:

[0071]

[0072] To solve this optimization problem, we can work by considering the objective function with respect to... , and Taking the partial derivatives and setting them to zero yields a set of nonlinear equations. However, in practical applications, directly solving this set of nonlinear equations can be quite complex. Therefore, iterative optimization algorithms, such as gradient descent or the Levenberg-Marquardt algorithm, are typically used to gradually approximate the optimal solution.

[0073] Specifically, first, select an initial estimate. , , These initial estimates can be obtained using simple geometric methods, such as calculating the average coordinates of all data points as an initial estimate of the circle's center, and then estimating the initial radius based on the distances from these points to the center. Next, the Levenberg-Marquardt algorithm is applied for iterative optimization. This algorithm iteratively adjusts the fitting parameters... , and This minimizes the sum of squared residuals. Specifically, each iteration calculates the residual vector for the current parameters and updates the fitted parameters based on the residual vector. After multiple iterations, the algorithm eventually converges to the optimal solution.

[0074] In one embodiment, in step S4, it can be seen from the collected data that the cut edge is more prominent compared to other points, but it may be located at any position on the wafer and may be surrounded by some interference signals, making it difficult to process in the spatiotemporal domain. The data is converted to the frequency domain using Fourier transform, then the cut edge is extracted through filtering, and finally, its spatial coordinates are obtained through inverse Fourier transform.

[0075] In a specific embodiment, such as Figure 9 As shown, S4, processing the discrete dataset of two-dimensional coordinate points at the wafer edge based on Fourier transform and inverse Fourier transform to obtain the tangent angle, includes: S41, arranging the discrete dataset of two-dimensional coordinate points at the wafer edge according to the order of the wafer edge to obtain the wafer edge contour signal; S42, performing a Fourier transform on the wafer edge contour signal to obtain a complex spectrum; S43, extracting tangent features from the complex spectrum; S44, mapping the tangent features back to the spatial domain through inverse Fourier transform to obtain the boundary point coordinates of the tangent; S45, determining the tangent angle based on the boundary point coordinates of the tangent.

[0076] Specifically, firstly, the discrete dataset of two-dimensional coordinate points at the wafer edge is arranged in the order of the wafer edges to form a continuous wafer edge profile signal. This wafer edge profile signal is actually one-dimensional time-series data, where each data point represents the coordinate value of a certain position on the wafer edge. The angular information of these points (e.g., the angle relative to the wafer center) can be used as a time-domain signal, or the index order of the points can be directly used as the "time" axis in the time domain. The key is to form a one-dimensional signal representing the profile change along the wafer edge.

[0077] Then, a Fourier transform is performed on the wafer edge contour signal to obtain a complex spectrum. The Fast Fourier Transform (FFT) algorithm can be used to efficiently compute the Discrete Fourier Transform. Here, the Fourier transform is a powerful mathematical tool that can convert time-domain signals into frequency-domain signals. During wafer pre-alignment, the Fourier transform is used to analyze the frequency components in the wafer edge contour signal, thereby extracting the edge trimming features. The Fourier transform decomposes the time-domain signal into a superposition of sine and cosine waves of different frequencies. The result is a complex spectrum containing amplitude and phase spectra, which describes the intensity and phase information of the signal at different frequencies. Specifically, the contour signal formed by arranging the two-dimensional coordinate points of the wafer edge in sequence is converted into a complex spectrum in the frequency domain using a Fourier transform. Here, the Fourier transform formula is:

[0078]

[0079] in, The result of the Fourier transform is represented by the function. The representation in the frequency domain. It describes the distribution of the signal across different frequency components. It represents the original signal or function, usually a signal in the spatial or time domain. This represents a frequency variable used to describe the frequency components of a signal. This represents a complex exponential function, used to transform a signal from the spatial or time domain to the frequency domain. Here... It is the imaginary unit.

[0080] Next, the slicing features are extracted from the complex spectrum. Since slicing features typically manifest as abrupt changes or discontinuities in the contour signal, they often appear as significant frequency components in the frequency domain. By analyzing these frequency components, the parts corresponding to the slicing features can be identified. Specifically, frequency components with larger amplitudes can be selected from the complex spectrum as potential slicing features. The phase information corresponding to these frequency components can help determine the specific location of the slicing. To further improve accuracy, filtering techniques can be applied to remove noise and other irrelevant frequency components. For example, a threshold can be set to retain only frequency components with amplitudes exceeding that threshold. After extracting the slicing features, the slice features are mapped back to the spatial domain using an inverse Fourier transform to obtain the coordinates of the boundary points of the slicing. The inverse Fourier transform formula is:

[0081]

[0082] in, It represents the original signal or function, which is recovered from the frequency domain to the spatial or time domain through inverse Fourier transform. This represents the result of the Fourier transform, indicating the signal's representation in the frequency domain. This represents the frequency variable, which is the same as the frequency variable in the Fourier transform. This represents a complex exponential function, used to convert a signal from the frequency domain back to the spatial or time domain. This represents the normalization factor, ensuring energy conservation between the Fourier transform and the inverse transform.

[0083] Finally, the cutting edge angle is determined based on the boundary point coordinates of the cutting edge. In the time-domain signal obtained by the inverse transform, the cutting edge position will manifest as a significant change in the signal (e.g., a peak or valley). The location of this significant change in the time-domain signal is found, and this location corresponds to the cutting edge position on the wafer edge profile. Based on the index of the cutting edge in the data point sequence, the original wafer edge coordinate points can be reversed, and the center coordinates or boundary point coordinates of the cutting edge can be calculated. Finally, the angle of the cutting edge relative to the wafer reference direction (e.g., the horizontal axis or the vertical axis) is determined. This can be done by calculating the angle between the cutting edge center point and the wafer center, or by analyzing the coordinates of the cutting edge boundary points to determine the direction of the cutting edge.

[0084] In a specific example, a 32-array is used to simulate the wafer edge profile signal. The array index represents the angular position (0 to 31), and the value represents a certain characteristic value of the edge. Assume the cut edge is near index 8. Below is a signal example for illustration only: signal = [1, 1, 1, 1, 1, 1, 1, 1, 0.5, 0.5, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], with smoothing at indices 0-7, decreasing values ​​at indices 8-9 to simulate the cut edge, and smoothing again at indices 10-31. First, a Fourier transform is performed on the signal using an FFT. This yields the frequency domain signal (a complex array). Then, the amplitude spectrum of the frequency domain signal is analyzed. It may be found that in addition to low-frequency components, there are also some significant amplitudes in the mid-to-high frequency region. Frequency domain filtering is then performed. To simplify the example, it is assumed that the frequency components with the largest amplitudes (excluding the DC component) are directly identified. Next, an inverse Fourier transform is performed to retain only the main frequency components, followed by an inverse FFT. Then, edge localization is performed; the signal after the inverse transform may show a more pronounced decrease in value near indices 8 and 9. By finding the minimum value of the signal or the location of the largest gradient change, the approximate location of the edge (indices 8-9) can be located. Finally, angle calculation is performed. Based on the positions of indices 8 and 9 in the entire data sequence, the angle range corresponding to the edge can be calculated. For example, if 32 indices correspond to 360 degrees, then index 8 corresponds to (8 / 32)*360 = 90 degrees, and index 9 corresponds to (9 / 32)*360 = 101.25 degrees. The angle range of the edge is approximately 90-101.25 degrees.

[0085] like Figure 10 As shown, in one specific embodiment, the process begins with the following steps: First, the robotic arm picks up the wafer: it removes the wafer from its storage location and places it on a pre-alignment platform. Next, the pre-alignment platform rotates 360 degrees to collect data: while rotating 360 degrees, a line laser sensor or other detection device scans the wafer edge, collecting a continuous dataset of two-dimensional coordinate points. This step ensures the comprehensiveness and accuracy of the data. Then, the center and angle are calculated: based on the collected data, the wafer's center coordinates are calculated using mathematical algorithms (such as least squares), and Fourier transform is used to extract the dicing features and determine the dicing angle. This step is the core of the entire pre-alignment process, ensuring the accuracy of wafer positioning. Next, the center is aligned: based on the calculated center coordinates, the wafer's position is adjusted to align with the preset processing position. Angle rotation: based on the calculated dicing angle, the wafer's rotation angle is adjusted to ensure the wafer is in the correct orientation during subsequent processing. Finally, the process ends: after all adjustments are completed, the wafer is ready for the next processing step.

[0086] In summary, this application provides a wafer pre-alignment method for the transfer process. It uses a control line laser sensor to scan the wafer edge on a rotating platform, acquiring a continuous dataset of two-dimensional coordinate points. This dataset is then filtered and outlier-removed to generate a discrete dataset. The discrete dataset is then fitted with a circle using the least squares method to accurately calculate the wafer center coordinates. Simultaneously, Fourier transform is used to extract edge features, and inverse Fourier transform is used to determine the edge angle. This significantly improves the speed and accuracy of pre-alignment. This method is particularly suitable for high-precision applications in semiconductor manufacturing, providing a reliable positioning basis for subsequent processing steps.

[0087] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0088] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0089] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0090] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0091] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A wafer pre-alignment method for a transfer process, characterized in that, include: The control line laser sensor scans the wafer edge on the rotating platform to obtain a continuous dataset of two-dimensional coordinate points of the wafer edge; Preprocessing the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge includes: performing data filtering on the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a filtered dataset of two-dimensional coordinate points at the wafer edge, wherein the data filtering includes: determining a neighborhood window for each data point in the continuous dataset of two-dimensional coordinate points at the wafer edge; updating the two-dimensional coordinates of each data point at the wafer edge based on the sample distribution of all data points within the neighborhood window of each data point; and introducing a neighborhood-based circular tangential fitting mechanism to optimize the updated two-dimensional coordinates of the wafer edge. The wafer center coordinates are obtained by performing a circle fitting on the discrete dataset of two-dimensional coordinate points at the wafer edge using the least squares method. The discrete dataset of two-dimensional coordinate points at the edge of the wafer is processed based on Fourier transform and inverse Fourier transform to obtain the cutting angle. Specifically, a neighborhood-based circular tangential fitting mechanism is introduced to optimize the updated two-dimensional coordinate points at the wafer edge, including: For each pair of coordinates obtained by weighting through the neighborhood window ( )and( First, let: To obtain the coordinates of each coordinate ( The corresponding tangential circle fitting parameter pair () ); Based on the tangential circle fitting parameter pair ( ), to analyze the data distribution within the neighborhood window ( Optimize: in, and It is a scaling factor used to optimize fine-tuning. These are the optimized two-dimensional coordinate points at the edge of the wafer.

2. The wafer pre-alignment method for the transmission process according to claim 1, characterized in that, The control line laser sensor scans the wafer edge on the rotating platform to obtain a continuous dataset of two-dimensional coordinate points at the wafer edge, including: The wafer is placed on the rotating platform; The rotating platform is driven to move the wafer, and during the rotation of the wafer, the line laser sensor collects a continuous dataset of two-dimensional coordinate points at the edge of the wafer.

3. The wafer pre-alignment method for the transmission process according to claim 1, characterized in that, Preprocessing the continuous dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge further includes: Outlier removal is performed on the filtered dataset of two-dimensional coordinate points at the wafer edge to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge.

4. The wafer pre-alignment method for the transmission process according to claim 3, characterized in that, Based on the sample distribution of all data points within the neighborhood window of each data point, the two-dimensional coordinates of the wafer edge of each data point are updated, as follows: in, The radius of the neighborhood window, The original two-dimensional coordinates of the wafer edge in the neighborhood window. This is an index variable used to iterate through all data points within the neighborhood window. The index of the current data point. These are the updated two-dimensional coordinates of the wafer edge.

5. The wafer pre-alignment method for the transmission process according to claim 3, characterized in that, The filtered dataset of two-dimensional coordinate points at the wafer edge is subjected to outlier removal to obtain a discrete dataset of two-dimensional coordinate points at the wafer edge, including: Determine the initial center of the fitted circle; Calculate the radial distance between each data point in the filtered dataset of the two-dimensional coordinate points at the edge of the wafer and the center of the initially fitted circle to obtain a set of radial distances; Calculate the mean and standard deviation of the set of radial distances; Based on the mean and standard deviation, calculate the standard score of each radial distance in the set of radial distances; Based on the comparison between the standard score and the preset conditions, it is determined whether to remove the two-dimensional coordinate points of the wafer edge corresponding to the radial distance.

6. The wafer pre-alignment method for the transmission process according to claim 5, characterized in that, Based on the mean and standard deviation, calculate the standard score of each radial distance in the set of radial distances, including: calculating the standard score of each radial distance using the following formula, expressed as: in, Radial distance, The mean, Standard deviation, Standard fraction representing radial distance.

7. The wafer pre-alignment method for the transmission process according to claim 5, characterized in that, The preset condition is that the absolute value of the standard score is greater than 3.

8. The wafer pre-alignment method for the transmission process according to claim 1, characterized in that, The discrete dataset of two-dimensional coordinate points at the wafer edge is processed based on Fourier transform and inverse Fourier transform to obtain the tangent angle, including: The discrete dataset of two-dimensional coordinate points at the wafer edge is arranged in the order of the wafer edge to obtain the wafer edge contour signal; Perform a Fourier transform on the wafer edge contour signal to obtain a complex spectrum; Extract the tangent features from the complex spectrum; The slice features are mapped back to the spatial domain using an inverse Fourier transform to obtain the coordinates of the boundary points of the slice edges; The angle of the tangent is determined based on the coordinates of the boundary point of the tangent.

Images