Wafer polishing method and system based on multi-parameter optimization

Abstract

The invention relates to the technical field of wafer polishing, and particularly discloses a wafer polishing method and system based on multi-parameter optimization, and the method comprises the steps: obtaining a historical wafer defect image, building a historical wafer defect grid chart according to the historical wafer defect image, and determining a polishing pressure set according to the historical wafer defect grid chart; the target wafer is polished according to the polishing pressure set, the real-time removal amount of the target wafer in the polishing process is obtained, and the polishing rotating speed is determined according to the real-time removal amount of the target wafer; the real-time polishing temperature of the target wafer in the polishing process is obtained, and the polishing temperature is determined according to the real-time polishing temperature change curve of the target wafer; and performing polishing control on the target wafer according to the polishing rotating speed and the polishing temperature. The polishing pressure of the wafer can be optimized, meanwhile, the polishing rotating speed and the polishing temperature are adjusted in real time, overall uniformity and planarization of the wafer are achieved, and the polishing quality of the wafer is improved.

Description

Technical Field

[0001] The present application relates to the field of wafer polishing technology, and more specifically, to a wafer polishing method and system based on multi-parameter optimization. Background Art

[0002] Polishing involves removing various materials from the wafer surface at the micron and nanometer levels through a combination of surface chemistry and mechanical abrasion, thereby flattening the wafer surface at the nanometer level. The wafer undergoes a series of polishing steps to create a smooth, mirror-like surface. This is crucial because any surface defects or contaminants can adversely affect the performance and reliability of the final semiconductor device. The polishing process typically combines chemical and mechanical techniques, such as chemical mechanical planarization (CMP). This process uses a slurry containing abrasive particles and chemical reagents to remove material from the wafer surface in a controlled manner.

[0003] In the existing technology, during the operation of chemical mechanical polishing equipment, due to the complex types of operating products, wafers are prone to defects such as scratches and warping during the processing. At this time, using fixed parameters to polish defective wafers will cause the flatness of the wafers to deteriorate, affecting the polishing quality of the wafers. Summary of the Invention

[0004] The present invention provides a wafer polishing method and system based on multi-parameter optimization, which is used to solve the problem in the prior art that the polishing quality of wafers cannot be guaranteed due to processing defects, including: Acquire historical wafer defect images, establish a historical wafer defect grid map based on the historical wafer defect images, and determine a polishing pressure set based on the historical wafer defect grid map; Polishing the target wafer according to the polishing pressure set, obtaining a real-time removal amount of the target wafer during the polishing process, and determining a polishing speed according to the real-time removal amount of the target wafer; Obtaining the real-time polishing temperature of the target wafer during the polishing process, and determining the polishing temperature according to the real-time polishing temperature change curve of the target wafer; The target wafer is polished according to the polishing speed and polishing temperature.

[0005] Furthermore, establishing a historical wafer defect grid map based on the historical wafer defect images and determining a polishing pressure set based on the historical wafer defect grid map includes: Determining film thickness data of the historical wafer defect image according to the historical wafer defect image, and segmenting the defect area according to the film thickness data of the historical wafer defect image; Obtain a preset grid, segment the historical wafer defect image according to the preset grid, and obtain a historical wafer defect grid map; Determining a defect area in the historical wafer defect grid map according to the historical wafer defect grid map, and obtaining a defect grid area composed of a plurality of defect grids; A defect grid feature set of the defect grid area is counted, and a polishing pressure set of the target wafer is determined according to the defect grid feature set of the defect grid area.

[0006] Furthermore, segmenting the defective area according to the film thickness data of the historical wafer defect image includes: Clustering the film thickness data of historical wafer defect images, and using the cluster center corresponding to the film thickness data as the seed point of the historical wafer defect image; Obtain film thickness data of points adjacent to the seed point in the historical wafer defect image, and calculate the difference between the film thickness data of the seed point and the adjacent points; Determine whether the difference between the film thickness data of the seed point and the adjacent point is less than a first preset threshold value, and if the difference between the film thickness data of the seed point and the adjacent point is less than the first preset threshold value, set the corresponding adjacent point of the seed point as a new growth seed point; Taking the new growth seed point as the center, continue to detect the adjacent points of the new seed point until no new growth seed point can be obtained, and obtain the defect area of ​​the historical wafer defect image.

[0007] Furthermore, the defective grid feature set of the statistical defective grid area includes: Obtain all defect grids in the defect grid area, calculate the average distance between all defect grids and the central grid of the historical wafer defect grid map, and obtain a first grid eigenvalue; Draw the minimum bounding rectangle of the defective mesh area, calculate the proportion of the defective mesh in the minimum bounding rectangle to all meshes, and obtain the second mesh eigenvalue; Extract the central grid of the defective grid area, establish rays from the central grid to the remaining grids in the defective grid area, and use the average length of the rays as the third grid eigenvalue; A defective grid feature set is established according to the first grid feature value, the second grid feature value, and the third grid feature value.

[0008] Furthermore, determining the polishing pressure set of the target wafer according to the defect grid feature set of the defect grid area includes: Obtain all defect network feature sets and corresponding target removal amounts in historical wafer defect images, and preprocess the defect network feature sets and corresponding target removal amounts; Establishing a training sample set based on the preprocessed defect network feature set and the corresponding target removal amount, establishing a polishing evaluation model based on the training sample set, and training the polishing evaluation model to obtain a trained polishing evaluation model; Obtain a defect grid feature set of a target wafer, input the defect grid feature set of the target wafer into a trained polishing evaluation model, obtain a target removal amount of the target wafer, and determine a polishing pressure set according to the target removal amount of the target wafer.

[0009] Furthermore, determining the polishing rotation speed according to the real-time removal amount of the target wafer includes: Obtaining a correlation coefficient between the removal amount and the polishing speed during a historical wafer polishing process, and determining a speed control weight based on the correlation coefficient between the removal amount and the polishing speed during the historical wafer polishing process; Obtaining a real-time removal amount of a target wafer, and calculating a difference between a target removal amount and a real-time removal amount of the target wafer to obtain a removal amount difference; The dynamic polishing speed is determined according to the speed control weight and the removal amount difference, and the dynamic polishing speed is determined as the polishing speed.

[0010] Furthermore, determining the dynamic polishing rotation speed according to the speed control weight and the removal amount difference includes: The dynamic polishing speed is determined according to the dynamic polishing speed calculation formula, and the dynamic polishing speed calculation formula is specifically: , in, is the dynamic polishing speed, is the speed control weight, is the removal amount difference of the target wafer at the current moment, is the removal amount difference of the target wafer at a certain moment, is the preset standard removal difference, To preset the standard polishing speed, is the natural exponential function.

[0011] Furthermore, determining the polishing temperature according to the real-time polishing temperature variation curve of the target wafer includes: The polishing temperature variation curve is fitted based on the least square method to obtain the polishing temperature prediction curve; Predicting the time required for the polishing temperature to reach a second preset threshold value according to the polishing temperature prediction curve, and determining a polishing temperature control coefficient according to the time required for the polishing temperature to reach the second preset threshold value; The preset standard polishing temperature is corrected according to the polishing temperature control coefficient, and the corrected preset standard polishing temperature is determined as the polishing temperature.

[0012] Furthermore, determining the polishing temperature control coefficient according to the time required for the polishing temperature to reach the second preset threshold value includes: Calculating the difference between the time required for the polishing temperature to reach a second preset threshold and the preset allowable time to obtain a polishing temperature risk coefficient; determining whether the polishing temperature risk coefficient is greater than a third preset threshold, and if the polishing temperature risk coefficient is greater than the third preset threshold, setting the first control coefficient as the polishing temperature control coefficient; If the polishing temperature risk coefficient is less than or equal to the third preset threshold, determining whether the polishing temperature risk coefficient is greater than a fourth preset threshold; If the polishing temperature risk coefficient is greater than a fourth preset threshold, setting the second control coefficient as the polishing temperature control coefficient; If the polishing temperature risk coefficient is less than or equal to the fourth preset threshold, the third control coefficient is set as the polishing temperature control coefficient.

[0013] In order to achieve the above object, the present invention further provides a wafer polishing system based on multi-parameter optimization, characterized in that it includes: a pressure optimization module, configured to obtain historical wafer defect images, establish a historical wafer defect grid map based on the historical wafer defect images, and determine a polishing pressure set based on the historical wafer defect grid map; A rotation speed optimization module is used to polish the target wafer according to the polishing pressure set, obtain the real-time removal amount of the target wafer during the polishing process, and determine the polishing speed according to the real-time removal amount of the target wafer; A temperature optimization module is used to obtain the real-time polishing temperature of the target wafer during the polishing process and determine the polishing temperature according to the real-time polishing temperature change curve of the target wafer; The polishing control module is used to control the polishing of the target wafer according to the polishing speed and polishing temperature.

[0014] The beneficial effects of the present invention are: By applying the above technical solutions, the present invention can optimize the polishing pressure of the wafer according to the defects of the wafer, and at the same time adjust the polishing speed and temperature in real time during the polishing process, so as to achieve global uniform flatness of the wafer and improve the polishing quality of the wafer. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 The figure shows an overall flow chart of a wafer polishing method based on multi-parameter optimization proposed in an embodiment of the present invention; Figure 2 A schematic structural diagram of a wafer polishing system based on multi-parameter optimization proposed in an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] The present application embodiment provides a wafer polishing method based on multi-parameter optimization, such as Figure 1 Shown, including: S101 , acquiring a historical wafer defect image, establishing a historical wafer defect grid map according to the historical wafer defect image, and determining a polishing pressure set according to the historical wafer defect grid map.

[0019] In some embodiments of the present application, the method of establishing a historical wafer defect grid map based on historical wafer defect images and determining a polishing pressure set based on the historical wafer defect grid map includes: determining film thickness data of the historical wafer defect images based on the historical wafer defect images, and segmenting defect areas based on the film thickness data of the historical wafer defect images; obtaining a preset grid, segmenting the historical wafer defect images based on the preset grid, and obtaining a historical wafer defect grid map; determining defect areas in the historical wafer defect grid map based on the historical wafer defect grid map, and obtaining a defect grid area composed of several defect grids; and statistically calculating a defect grid feature set of the defect grid area, and determining a polishing pressure set of the target wafer based on the defect grid feature set of the defect grid area.

[0020] In this embodiment, historical wafer defect images of wafer defects existing in historical wafer processing are statistically collected, and the historical wafer defect images include normal areas and defect areas that differ from normal film thickness data. The defect grid areas corresponding to the defect areas are obtained by establishing a historical wafer defect grid map. The wafer defect area types have a certain regularity, and therefore the defect grid feature set is obtained by calculating the regularity characteristics of the defect grid areas, thereby obtaining the polishing pressure set of the target wafer by matching the defect grid area of ​​the target wafer with the defect grid area of ​​the historical wafer defect grid map. The polishing pressure set is a polishing pressure time series, which includes the polishing pressure values ​​of the target wafer at each time point from the start to the end of polishing.

[0021] In some embodiments of the present application, the defect area is segmented according to the film thickness data of the historical wafer defect image, including: clustering the film thickness data of the historical wafer defect image, and using the cluster center corresponding to the film thickness data as the seed point of the historical wafer defect image; obtaining the film thickness data of the adjacent points of the seed point in the historical wafer defect image, and calculating the difference between the film thickness data of the seed point and the adjacent points; judging whether the difference between the film thickness data of the seed point and the adjacent points is less than a first preset threshold value, if the difference between the film thickness data of the seed point and the adjacent points is less than the first preset threshold value, setting the corresponding adjacent points of the seed point as new growth seed points; with the new growth seed point as the center, continue to detect the adjacent points of the new seed point until no new growth seed point can be obtained, and obtain the defect area of ​​the historical wafer defect image.

[0022] In this embodiment, the points in the historical wafer defect image are clustered based on the film thickness data based on the k-means clustering algorithm, and region growth is performed based on the cluster center points after clustering. The region with the largest area is set as the normal region, and the remaining regions are defect regions.

[0023] In some embodiments of the present application, the defect grid feature set of the statistical defect grid area includes: obtaining all defect grids in the defect grid area, calculating the average distance value of all defect grids from the central grid of the historical wafer defect grid map, and obtaining a first grid feature value; drawing the minimum circumscribed rectangle of the defect grid area, calculating the proportion of the defect grid in all grids in the minimum circumscribed rectangle, and obtaining a second grid feature value; extracting the central grid of the defect grid area, establishing a ray from the central grid to the remaining grids in the defect grid area, and taking the average value of the ray length as the third grid feature value; and establishing a defect grid feature set based on the first grid feature value, the second grid feature value, and the third grid feature value.

[0024] In this embodiment, the position characteristics of the defect grid area are obtained by the average distance value of all defect grids from the central grid of the historical wafer defect grid map, the density characteristics of the defect grid area are obtained by calculating the proportion of the defect grid in all grids in the minimum circumscribed rectangle, and the morphological characteristics of the defect grid area are obtained by the average value of the ray length, thereby comprehensively establishing a defect grid feature set based on the position, density, and morphology of the defect area.

[0025] In some embodiments of the present application, the method of determining the polishing pressure set of the target wafer based on the defect grid feature set of the defect grid area includes: obtaining all defect network feature sets and corresponding target removal amounts in historical wafer defect images, and preprocessing the defect network feature sets and the corresponding target removal amounts; establishing a training sample set based on the preprocessed defect network feature set and the corresponding target removal amount, establishing a polishing evaluation model based on the training sample set and training the polishing evaluation model to obtain a trained polishing evaluation model; obtaining the defect grid feature set of the target wafer, inputting the defect grid feature set of the target wafer into the trained polishing evaluation model to obtain the target removal amount of the target wafer, and determining the polishing pressure set based on the target removal amount of the target wafer.

[0026] In this embodiment, the target removal amount of the wafer is the wafer volume planned to be removed during the polishing process. During the historical polishing process, the defect grid feature set of the wafer defect grid area in the historical wafer defect image and the corresponding target removal amount are used as a set of sample data. All the defect grid feature sets and their corresponding target removal amounts in the historical polishing process are counted and preprocessed. The preprocessing includes at least one of the following: data cleaning, normalization, feature construction and feature extraction. A training sample set is established through the preprocessed data, and the defect grid feature set in the training sample set is used as an input variable, and the target removal amount is used as an output variable. A deep learning neural network model is established through the training sample set to obtain a polishing evaluation model, whose network structure is ResNet50. 80% of the training sample set is used as a training set and 20% is used as a test set. The mean square error loss function is used, and the optimizer uses Adam. After training, a trained polishing evaluation model is finally obtained.

[0027] In this embodiment, the target wafer is a wafer to be polished, and a wafer defect image of the target wafer is obtained. The defect grid feature set of the target wafer is detected by the above-mentioned historical wafer defect image detection method. The defect grid feature set of the target wafer is input into the trained polishing evaluation model to output the target removal amount of the target wafer. The target removal amount has a linear relationship with the polishing pressure set, and the polishing pressure set corresponding to the target removal amount of the target wafer can be obtained by establishing a target removal amount-polishing pressure set mapping table.

[0028] S102, polishing the target wafer according to the polishing pressure set, obtaining a real-time removal amount of the target wafer during the polishing process, and determining a polishing speed according to the real-time removal amount of the target wafer; In some embodiments of the present application, determining the polishing speed based on the real-time removal amount of the target wafer includes: obtaining the correlation coefficient between the removal amount and the polishing speed during the historical wafer polishing process, and determining the speed control weight based on the correlation coefficient between the removal amount and the polishing speed during the historical wafer polishing process; obtaining the real-time removal amount of the target wafer, and statistically calculating the difference between the target removal amount and the real-time removal amount of the target wafer to obtain the removal amount difference; determining the dynamic polishing speed based on the speed control weight and the removal amount difference, and determining the dynamic polishing speed as the polishing speed.

[0029] In some embodiments of the present application, determining the dynamic polishing speed according to the speed control weight and the removal amount difference includes: determining the dynamic polishing speed according to a dynamic polishing speed calculation formula, wherein the dynamic polishing speed calculation formula is specifically: , in, is the dynamic polishing speed, is the speed control weight, is the removal amount difference of the target wafer at the current moment, is the removal amount difference of the target wafer at a certain moment, is the preset standard removal difference, To preset the standard polishing speed, is the natural exponential function.

[0030] In this embodiment, the dynamic polishing speed is determined by the speed control weight and the removal amount difference, which can timely reduce the polishing speed when the removal amount of the target wafer fluctuates, making the polishing process more stable and ensuring polishing flatness.

[0031] S103, obtaining the real-time polishing temperature of the target wafer during the polishing process, and determining the polishing temperature according to the real-time polishing temperature variation curve of the target wafer; In some embodiments of the present application, determining the polishing temperature based on the real-time polishing temperature change curve of the target wafer includes: performing curve fitting on the polishing temperature change curve based on the least squares method to obtain a polishing temperature prediction curve; predicting the time required for the polishing temperature to reach a second preset threshold value based on the polishing temperature prediction curve, and determining a polishing temperature control coefficient based on the time required for the polishing temperature to reach the second preset threshold value; correcting the preset standard polishing temperature based on the polishing temperature control coefficient, and determining the corrected preset standard polishing temperature as the polishing temperature.

[0032] In some embodiments of the present application, the polishing temperature control coefficient is determined based on the time required for the polishing temperature to reach a second preset threshold, including: calculating the difference between the time required for the polishing temperature to reach the second preset threshold and the preset allowable time to obtain a polishing temperature risk coefficient; judging whether the polishing temperature risk coefficient is greater than a third preset threshold, if the polishing temperature risk coefficient is greater than the third preset threshold, setting the first control coefficient as the polishing temperature control coefficient; if the polishing temperature risk coefficient is less than or equal to the third preset threshold, judging whether the polishing temperature risk coefficient is greater than a fourth preset threshold; if the polishing temperature risk coefficient is greater than the fourth preset threshold, setting the second control coefficient as the polishing temperature control coefficient; if the polishing temperature risk coefficient is less than or equal to the fourth preset threshold, setting the third control coefficient as the polishing temperature control coefficient.

[0033] In this embodiment, the polishing temperature during the polishing process is monitored in real time by establishing a polishing temperature prediction curve. At the same time, a polishing temperature control coefficient is set based on the real-time monitoring of the temperature. The shorter the time required for the polishing temperature to reach the second preset threshold, the lower the corresponding polishing temperature control coefficient, thereby preventing the polishing temperature from exceeding the limit and affecting the process stability of the wafer.

[0034] S104, controlling the polishing of the target wafer according to the polishing speed and the polishing temperature.

[0035] Based on the same technical concept, such as Figure 2 As shown, the present invention also provides a wafer polishing system based on multi-parameter optimization, including: a pressure optimization module, used to obtain historical wafer defect images, establish a historical wafer defect grid map based on the historical wafer defect images, and determine a polishing pressure set based on the historical wafer defect grid map; a speed optimization module, used to polish the target wafer according to the polishing pressure set, obtain the real-time removal amount of the target wafer during the polishing process, and determine the polishing speed according to the real-time removal amount of the target wafer; a temperature optimization module, used to obtain the real-time polishing temperature of the target wafer during the polishing process, and determine the polishing temperature according to the real-time polishing temperature change curve of the target wafer; and a polishing control module, used to control the polishing of the target wafer according to the polishing speed and polishing temperature.

[0036] By applying the above technical solution, the present invention obtains historical wafer defect images, establishes a historical wafer defect grid map based on the historical wafer defect images, and determines a polishing pressure set based on the historical wafer defect grid map; polishes a target wafer based on the polishing pressure set, obtains the real-time removal amount of the target wafer during the polishing process, and determines the polishing speed based on the real-time removal amount of the target wafer; obtains the real-time polishing temperature of the target wafer during the polishing process, and determines the polishing temperature based on the real-time polishing temperature change curve of the target wafer; and controls the polishing of the target wafer based on the polishing speed and polishing temperature. The present invention can optimize the polishing pressure of the wafer and simultaneously adjust the polishing speed and temperature in real time to achieve global uniform flattening of the wafer, thereby improving the polishing quality of the wafer.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical feature values ​​therein. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A wafer polishing method based on multi-parameter optimization, characterized in that: include: Acquire historical wafer defect images, establish a historical wafer defect grid map based on the historical wafer defect images, and determine a polishing pressure set based on the historical wafer defect grid map; Polishing the target wafer according to the polishing pressure set, obtaining a real-time removal amount of the target wafer during the polishing process, and determining a polishing speed according to the real-time removal amount of the target wafer; Obtaining the real-time polishing temperature of the target wafer during the polishing process, and determining the polishing temperature according to the real-time polishing temperature change curve of the target wafer; The target wafer is polished according to the polishing speed and polishing temperature.

2. The wafer polishing method based on multi-parameter optimization according to claim 1, characterized in that: The step of establishing a historical wafer defect grid map based on historical wafer defect images and determining a polishing pressure set based on the historical wafer defect grid map includes: Determining film thickness data of the historical wafer defect image according to the historical wafer defect image, and segmenting the defect area according to the film thickness data of the historical wafer defect image; Obtain a preset grid, segment the historical wafer defect image according to the preset grid, and obtain a historical wafer defect grid map; Determining a defect area in the historical wafer defect grid map according to the historical wafer defect grid map, and obtaining a defect grid area composed of a plurality of defect grids; A defect grid feature set of the defect grid area is counted, and a polishing pressure set of the target wafer is determined according to the defect grid feature set of the defect grid area.

3. The wafer polishing method based on multi-parameter optimization according to claim 2, characterized in that: The method of segmenting the defective area according to the film thickness data of the historical wafer defect image includes: Clustering the film thickness data of historical wafer defect images, and using the cluster center corresponding to the film thickness data as the seed point of the historical wafer defect image; Obtain film thickness data of points adjacent to the seed point in the historical wafer defect image, and calculate the difference between the film thickness data of the seed point and the adjacent points; Determine whether the difference between the film thickness data of the seed point and the adjacent point is less than a first preset threshold value, and if the difference between the film thickness data of the seed point and the adjacent point is less than the first preset threshold value, set the corresponding adjacent point of the seed point as a new growth seed point; Taking the new growth seed point as the center, continue to detect the adjacent points of the new seed point until no new growth seed point can be obtained, and obtain the defect area of ​​the historical wafer defect image.

4. The wafer polishing method based on multi-parameter optimization according to claim 2, characterized in that: The defective grid feature set of the statistical defective grid area includes: Obtain all defect grids in the defect grid area, calculate the average distance between all defect grids and the central grid of the historical wafer defect grid map, and obtain a first grid eigenvalue; Draw the minimum bounding rectangle of the defective mesh area, calculate the proportion of the defective mesh in the minimum bounding rectangle to all meshes, and obtain the second mesh eigenvalue; Extract the central grid of the defective grid area, establish rays from the central grid to the remaining grids in the defective grid area, and use the average length of the rays as the third grid eigenvalue; A defective grid feature set is established according to the first grid feature value, the second grid feature value, and the third grid feature value.

5. The wafer polishing method based on multi-parameter optimization according to claim 2, characterized in that: The step of determining the polishing pressure set of the target wafer according to the defect grid feature set of the defect grid area includes: Obtain all defect network feature sets and corresponding target removal amounts in historical wafer defect images, and preprocess the defect network feature sets and corresponding target removal amounts; A training sample set is established based on the preprocessed defect network feature set and the corresponding target removal amount, a polishing evaluation model is established based on the training sample set, and the polishing evaluation model is trained to obtain a trained polishing evaluation model; Obtain a defect grid feature set of a target wafer, input the defect grid feature set of the target wafer into a trained polishing evaluation model, obtain a target removal amount of the target wafer, and determine a polishing pressure set according to the target removal amount of the target wafer.

6. The wafer polishing method based on multi-parameter optimization according to claim 1, characterized in that: Determining the polishing rotation speed according to the real-time removal amount of the target wafer includes: Obtaining a correlation coefficient between the removal amount and the polishing speed during a historical wafer polishing process, and determining a speed control weight based on the correlation coefficient between the removal amount and the polishing speed during the historical wafer polishing process; Obtaining a real-time removal amount of a target wafer, and calculating a difference between a target removal amount and a real-time removal amount of the target wafer to obtain a removal amount difference; The dynamic polishing speed is determined according to the speed control weight and the removal amount difference, and the dynamic polishing speed is determined as the polishing speed.

7. The wafer polishing method based on multi-parameter optimization according to claim 6, characterized in that: Determining the dynamic polishing rotation speed according to the speed control weight and the removal amount difference includes: The dynamic polishing speed is determined according to the dynamic polishing speed calculation formula, and the dynamic polishing speed calculation formula is specifically: ， in, is the dynamic polishing speed, is the speed control weight, is the removal amount difference of the target wafer at the current moment, is the removal amount difference of the target wafer at a certain moment, is the preset standard removal difference, To preset the standard polishing speed, is the natural exponential function.

8. The wafer polishing method based on multi-parameter optimization according to claim 1, characterized in that: Determining the polishing temperature according to the real-time polishing temperature variation curve of the target wafer includes: The polishing temperature variation curve is fitted based on the least square method to obtain the polishing temperature prediction curve; Predicting the time required for the polishing temperature to reach a second preset threshold value according to the polishing temperature prediction curve, and determining a polishing temperature control coefficient according to the time required for the polishing temperature to reach the second preset threshold value; The preset standard polishing temperature is corrected according to the polishing temperature control coefficient, and the corrected preset standard polishing temperature is determined as the polishing temperature.

9. The wafer polishing method based on multi-parameter optimization according to claim 8, characterized in that: Determining the polishing temperature control coefficient according to the time required for the polishing temperature to reach the second preset threshold value includes: Calculating the difference between the time required for the polishing temperature to reach a second preset threshold and the preset allowable time to obtain a polishing temperature risk coefficient; determining whether the polishing temperature risk coefficient is greater than a third preset threshold, and if the polishing temperature risk coefficient is greater than the third preset threshold, setting the first control coefficient as the polishing temperature control coefficient; If the polishing temperature risk coefficient is less than or equal to the third preset threshold, determining whether the polishing temperature risk coefficient is greater than a fourth preset threshold; If the polishing temperature risk coefficient is greater than a fourth preset threshold, setting the second control coefficient as the polishing temperature control coefficient; If the polishing temperature risk coefficient is less than or equal to the fourth preset threshold, the third control coefficient is set as the polishing temperature control coefficient.

10. A wafer polishing system based on multi-parameter optimization, characterized in that: include: a pressure optimization module, configured to obtain historical wafer defect images, establish a historical wafer defect grid map based on the historical wafer defect images, and determine a polishing pressure set based on the historical wafer defect grid map; A rotation speed optimization module is used to polish the target wafer according to the polishing pressure set, obtain the real-time removal amount of the target wafer during the polishing process, and determine the polishing speed according to the real-time removal amount of the target wafer; A temperature optimization module is used to obtain the real-time polishing temperature of the target wafer during the polishing process and determine the polishing temperature according to the real-time polishing temperature change curve of the target wafer; The polishing control module is used to control the polishing of the target wafer according to the polishing speed and polishing temperature.

Images