An intelligent processing monitoring system for integrated circuit chips

Abstract

The present invention relates to the technical field of integrated circuit chip processing monitoring, and more specifically, to an intelligent processing monitoring system for integrated circuit chips. The system comprises: acquiring integrated circuit chip image data within a target area and processing data collected by sensors on a processing line; calculating a defect feature index for each sub-image; segmenting the defective areas of the integrated circuit chips in the sub-images based on the defect feature index; extracting features of the defective areas of the integrated circuit chips; identifying and analyzing the features of the defective areas to generate appearance monitoring analysis results; analyzing the processing data to generate non-appearance monitoring analysis results; and marking the integrated circuit chips within the target area based on the appearance monitoring analysis results and the non-appearance monitoring analysis results. The system is capable of rapidly identifying surface defects of the integrated circuit chips and promptly detecting abnormal changes in processing parameters, thereby improving the accuracy and efficiency of processing monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated circuit chip processing monitoring, and in particular to an intelligent processing monitoring system for integrated circuit chips. Background Art

[0002] The chip manufacturing process involves hundreds of complex steps, including wafer preparation, photolithography, etching, thin film deposition, doping, packaging, and testing. Each step has extremely strict requirements for process parameters. Any slight deviation in these parameters can lead to reduced chip performance, lower yield, or even the scrapping of an entire batch of chips, resulting in huge economic losses.

[0003] Currently, traditional chip processing monitoring relies primarily on manual labor, with operators regularly inspecting production lines and recording various equipment operating parameters, such as temperature, pressure, and speed. This approach is not only inefficient but also susceptible to human factors, such as operator fatigue and negligence, leading to inaccurate and untimely data recording and inability to detect abnormal changes in process parameters. Furthermore, existing technologies lack a method for monitoring chip processing quality using image data from integrated circuit chips, resulting in low monitoring efficiency.

[0004] Therefore, there is an urgent need for an intelligent processing monitoring system for integrated circuit chips to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent processing monitoring system for integrated circuit chips: it aims to solve the technical problems that traditional chip processing monitoring mainly relies on manual work, resulting in inaccurate and untimely monitoring, and inability to timely detect abnormal changes in process parameters and chip surface defects.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] An intelligent processing monitoring system for integrated circuit chips, the system comprising:

[0008] A monitoring data acquisition unit is used to acquire integrated circuit chip image data within a target area and processing data collected by sensors on a processing line, and to send the integrated circuit chip image data to an appearance analysis unit and the processing data to a non-appearance analysis unit;

[0009] an appearance analysis unit, configured to receive integrated circuit chip image data, divide the integrated circuit chip image into M sub-images, calculate a defect feature index for each sub-image, segment defective parts of the integrated circuit chip in the sub-image based on the defect feature index, extract features of the defective parts of the integrated circuit chip, identify and analyze the features of the defective parts, generate appearance monitoring analysis results, and send the appearance monitoring analysis results to the monitoring and early warning unit;

[0010] The non-appearance analysis unit is used to receive the processed data, analyze the processed data, generate non-appearance monitoring analysis results, and send the non-appearance monitoring analysis results to the monitoring and early warning unit;

[0011] The monitoring and early warning unit is used to receive the appearance monitoring analysis results and the non-appearance monitoring analysis results to mark the integrated circuit chips in the target area.

[0012] Furthermore, dividing the integrated circuit chip image into M sub-images specifically includes the following process:

[0013] Step 1: Initialize the population: set the initial population size m and the fitness threshold , maximum number of iterations , search range , randomly generate an n-dimensional vector with each component between 0 and 1 ,by is the initial value, based on the regression equation Iterate to get N vectors ,in, ;

[0014] Step 2: Divide the initial population into NP group and C group: Calculate the fitness value of each particle in the initial population and sort them, select the first m particles as NP group particles, and the remaining m particles as C group particles; and record the individual extreme values ​​of the NP group and NP group global extremum , record the global extreme value of group C ;

[0015] Step 3: Determine whether the maximum number of iterations has been reached. If so, execute step 5; otherwise, execute steps 31 to 34.

[0016] Step 4: Cross the NP group and the C group: select G particles with the best fitness values ​​from the C group to replace the G particles with the worst fitness values ​​in the NP group, and return to step 3 to continue iterating until the number of iterations is reached ;

[0017] Step 5: The iteration ends, and the target optimization parameter M is output. Based on the target optimization parameter M, the integrated circuit chip image is divided into M sub-images.

[0018] Furthermore, executing steps 31 to 34 specifically includes the following process:

[0019] Step 3.1: Update particles: Update the speed and position of NP group particles using the following formula

[0020] ;

[0021] ;

[0022] Among them, the learning factor and Both are 0.01, is the number of iterations; is the inertia weight, For the The best historical position experienced by a particle, is the best position experienced by all particles in the group, and is a random number between 0 and 1 that follows a uniform distribution. For the The particle velocity of the iteration, For the The particle position at iteration , For the The particle position at iteration , For the The particle velocity of the iteration;

[0023] Update the position of group C with the following formula:

[0024] ;

[0025] in, For the The particle position at iteration , For the The particle position at iteration , The value of is 3.57;

[0026] Step 32: Update the individual extreme value of the NP group and NP group global extremum , evaluate the particle fitness value after the NP group is updated, if the individual fitness value of the particle is better than , then Replace with the current particle position; compare and ,like Greater than , then Replace with current ;

[0027] Step 33: Update the global extreme value of group C :Evaluate the fitness value of group C particles. If the fitness value is better than , then Replace with the current particle position;

[0028] Step 3 and 4: Calculate the population fitness variance :

[0029] ; ;

[0030] in, is the total number of particles in the NP group, is the individual fitness value of the i-th particle, is the overall average fitness value of the NP group, and max represents the maximum value operation;

[0031] when When it is less than the preset threshold, it is determined to be in a premature state and step 4 is executed. If it is greater than or equal to the preset threshold, return to step 3.

[0032] Furthermore, calculating the defect feature index of each sub-image specifically includes the following process:

[0033] Calculate the contrast saliency value between the sub-image and the preset defect-free sub-image :

[0034] ;

[0035] in, is the number of superpixel regions in the sub-image, is the number of pixels in each superpixel region, Superpixel area and the superpixel area of ​​the preset defect-free sub-image The Euclidean distance between Superpixel area and preset defect-free sub-image superpixel area The distance from the center of mass, is the median distance between each pair of superpixel centroids;

[0036] Calculate the sub-image color distribution eigenvalues :

[0037] ;

[0038] in, Pick and the maximum value of 0, and They are super pixel areas in RGB color space The R, B, and G values ​​of the j-th pixel in ;

[0039] Calculate the defect feature index of each sub-image :

[0040] ;

[0041] in, 、 、 are weight coefficients, is the preset contrast significance threshold, is the preset color distribution feature threshold, is the defect coefficient of the sub-image in the hyperspectral space.

[0042] Furthermore, obtaining the defect coefficient of the sub-image in the hyperspectral space specifically includes the following process:

[0043] Obtain the characteristic wavelength reflectance map of the sub-image in the hyperspectral space, add the preset band to the characteristic wavelength reflectance map, and extract the three eigenvectors of the characteristic wavelength reflectance map in the preset band 、 、 , the wavelengths corresponding to the vectors are 、 、 , calculate the defect coefficient of the sub-image in the hyperspectral space ; Among them, the three eigenvectors are The ratio of the reflectivity difference in the axial projection direction is equal.

[0044] Furthermore, segmenting the defective parts of the integrated circuit chip in the sub-image based on the defect feature index and extracting the features of the defective parts of the integrated circuit chip specifically includes the following processes:

[0045] Divide the sub-image into several parts and determine the Does it exceed the preset threshold? If so, the part is regarded as the defective part of the integrated circuit chip, and the feature extraction of the defective part of the integrated circuit chip is performed to obtain the defective part feature of the integrated circuit chip. :

[0046] ;

[0047] in, represents the number of channels in the image, Indicates the Pixels on the The color value of each channel, Indicates the number of pixels. Indicates the The color mean of each channel.

[0048] Furthermore, the identification and analysis of defective part features and the generation of appearance monitoring analysis results specifically include the following processes:

[0049] Comparing the defect part features with the defect part dataset, a learning representation of the node is obtained, and determining whether the learning representation of the node contains a component missing representation. If so, it is determined that the integrated circuit chip has a component missing;

[0050] Determine whether the learning representation of the node has a component position welding error representation, and if so, determine that the integrated circuit chip has a component position welding error;

[0051] Determine whether the learning representation of the node has component redundancy representation, and if so, determine that the integrated circuit chip has component redundancy;

[0052] The judgment result is recorded as the appearance monitoring analysis result.

[0053] Furthermore, the processing data is analyzed to generate non-appearance monitoring analysis results, which specifically includes the following process:

[0054] Based on the processing data, equipment operation data of the integrated circuit chip processing line is obtained, where the equipment includes a wafer cleaning machine, a photolithography machine, an etching machine, a chemical mechanical polishing machine, an electroplating machine, and a wafer bonding machine. The equipment operation data is compared and analyzed with the equipment operation standard data in the processing index to determine whether the equipment operation data is the same as the equipment operation standard data in the processing index. If so, the processing line equipment parameter coefficient JG is 1; otherwise, the processing line equipment parameter coefficient JG is 0;

[0055] Acquire a temperature characterization value WD within a processing period based on processing data;

[0056] Obtaining a wafer cutting speed, a photolithography speed, an etching speed, and a metal filling speed within a processing period based on the processing data, respectively calculating a difference between the wafer cutting speed and a preset wafer cutting speed threshold, a difference between the photolithography speed and a preset photolithography speed threshold, a difference between the etching speed and a preset etching speed threshold, and a difference between the metal filling speed and a preset metal filling speed threshold, taking absolute values ​​of all the differences and summing them to obtain a sum value SD;

[0057] Based on the processing data, the internal pressure of the processing equipment, the pressure of the gas pipeline used for etching or cleaning, and the liquid pressure in the liquid processing process during the processing period are obtained, wherein the liquid processing process includes the coating of the photoresist and the spraying of the developer, and it is determined whether all the pressures are within a preset range. If so, the pressure parameter YL is 1; if not, the pressure parameter YL is 0;

[0058] Substitute the processing line equipment parameter coefficient JG, temperature characterization values ​​WD, SD and pressure parameter YL into the processing qualification index calculation formula to calculate the processing qualification index LMS. The calculation formula is as follows:

[0059] ;

[0060] in, The value of is 2.72, 、 are weight coefficients, with values ​​of 0.6 and 0.4 respectively;

[0061] Determine whether the processing qualification index exceeds the preset processing qualification index threshold. If so, the processing of the integrated circuit chip is determined to be qualified. If not, the processing of the integrated circuit chip is determined to be unqualified, and the judgment result is recorded as the generated non-appearance monitoring analysis result.

[0062] Furthermore, obtaining the temperature characterization value WD within the processing period includes: dividing the processing period into several sub-periods, detecting the temperature of the wafer manufacturing process, the temperature of the lithography and etching process, and the temperature of the ion implantation and metal filling process in each sub-period, establishing a rectangular coordinate system with the execution time of each sub-period as the X-axis and the temperature value as the Y-axis, connecting the temperatures in each manufacturing process, generating a temperature curve of the wafer manufacturing process, a temperature curve of the lithography and etching process, and a temperature curve of the ion implantation and metal filling process, calculating the total length of all temperature curves above the preset temperature curve, and recording the total length of the curves as the temperature characterization value WD within the processing period.

[0063] Compared with the existing solutions, the present invention achieves the following beneficial effects:

[0064] The present invention obtains integrated circuit chip image data within a target area and processing data collected by sensors on a processing production line, divides the integrated circuit chip image into M sub-images, calculates a defect feature index for each sub-image, segments defective parts of the integrated circuit chip in the sub-image based on the defect feature index, extracts features of the defective parts of the integrated circuit chip, identifies and analyzes the features of the defective parts, generates appearance monitoring analysis results, analyzes the processing data, generates non-appearance monitoring analysis results, and marks the integrated circuit chips within the target area based on the appearance monitoring analysis results and the non-appearance monitoring analysis results. The present invention can quickly identify surface defects of the integrated circuit chips and promptly discover abnormal changes in processing parameters, thereby improving the accuracy and efficiency of processing monitoring.

[0065] Furthermore, the intelligent processing monitoring system of the integrated circuit chip realizes all-round monitoring and quality control of the chip manufacturing process through comprehensive and accurate data acquisition and integration, refined appearance defect analysis and identification, in-depth non-appearance parameter analysis and evaluation, comprehensive early warning and precise marking, etc., effectively improving production efficiency, reducing costs, and improving chip quality and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0067] Figure 1 This is a system block diagram of an intelligent processing monitoring system for integrated circuit chips according to an embodiment of the present invention;

[0068] Figure 2 This is a workflow diagram of the first integrated circuit chip intelligent processing monitoring system according to an embodiment of the present invention;

[0069] Figure 3 This is a workflow diagram of the second intelligent processing monitoring system for integrated circuit chips according to an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0071] In addition, the described features, structures or characteristics can be combined in any suitable manner in one or more example embodiments. In the following description, many specific details are provided to provide a full understanding of the example embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. can be adopted. In other cases, well-known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0072] This embodiment provides an intelligent processing monitoring system for integrated circuit chips. Figure 1 This is a system block diagram of an intelligent processing monitoring system for integrated circuit chips according to an embodiment of the present invention. Figure 1As shown, the system includes:

[0073] A monitoring data acquisition unit is used to acquire integrated circuit chip image data within a target area and processing data collected by sensors on a processing line, and to send the integrated circuit chip image data to an appearance analysis unit and the processing data to a non-appearance analysis unit;

[0074] an appearance analysis unit, configured to receive integrated circuit chip image data, divide the integrated circuit chip image into M sub-images, calculate a defect feature index for each sub-image, segment defective parts of the integrated circuit chip in the sub-image based on the defect feature index, extract features of the defective parts of the integrated circuit chip, identify and analyze the features of the defective parts, generate appearance monitoring analysis results, and send the appearance monitoring analysis results to the monitoring and early warning unit;

[0075] The non-appearance analysis unit is used to receive the processed data, analyze the processed data, generate non-appearance monitoring analysis results, and send the non-appearance monitoring analysis results to the monitoring and early warning unit;

[0076] The monitoring and early warning unit is used to receive the appearance monitoring analysis results and the non-appearance monitoring analysis results to mark the integrated circuit chips in the target area.

[0077] It is worth noting that the monitoring and early warning unit marks unqualified integrated circuit chips and sends out early warning signals.

[0078] In summary, the present invention obtains the image data of the integrated circuit chip in the target area and the processing data collected by the sensors on the processing production line, divides the integrated circuit chip image into M sub-images, calculates the defect feature index of each sub-image, segments the defective parts of the integrated circuit chip in the sub-image based on the defect feature index, extracts the features of the defective parts of the integrated circuit chip, identifies and analyzes the features of the defective parts, generates appearance monitoring analysis results, analyzes the processing data, generates non-appearance monitoring analysis results, marks the integrated circuit chips in the target area based on the appearance monitoring analysis results and the non-appearance monitoring analysis results, can quickly identify the surface defects of the integrated circuit chips and timely discover abnormal changes in the processing parameters, thereby improving the accuracy and efficiency of processing monitoring.

[0079] In some embodiments, dividing the integrated circuit chip image into M sub-images specifically includes the following process:

[0080] Step 1: Initialize the population: set the initial population size m and the fitness threshold , maximum number of iterations , search range , randomly generate an n-dimensional vector with each component between 0 and 1 ,by is the initial value, based on the regression equation Iterate to get N vectors ,in, ;

[0081] Step 2: Divide the initial population into NP group and C group: Calculate the fitness value of each particle in the initial population and sort them, select the first m particles as NP group particles, and the remaining m particles as C group particles; and record the individual extreme values ​​of the NP group and NP group global extremum , record the global extreme value of group C ;

[0082] Step 3: Determine whether the maximum number of iterations has been reached. If so, execute step 5; otherwise, execute steps 31 to 34.

[0083] Step 3.1: Update particles: Update the speed and position of NP group particles using the following formula

[0084] ;

[0085] ;

[0086] Among them, the learning factor and Both are 0.01, is the number of iterations; is the inertia weight, For the The best historical position experienced by a particle, is the best position experienced by all particles in the group, and is a random number between 0 and 1 that follows a uniform distribution. For the The particle velocity of the iteration, For the The particle position at iteration , For the The particle position at iteration , For the The particle velocity of the iteration;

[0087] Update the position of group C with the following formula:

[0088] ;

[0089] in, For the The particle position at iteration , For the The particle position at iteration , The value of is 3.57;

[0090] Step 32: Update the individual extreme value of the NP group and NP group global extremum , evaluate the particle fitness value after the NP group is updated, if the individual fitness value of the particle is better than , then Replace with the current particle position; compare and ,like Greater than , then Replace with current ;

[0091] Step 33: Update the global extreme value of group C :Evaluate the fitness value of group C particles. If the fitness value is better than , then Replace with the current particle position;

[0092] Step 3 and 4: Calculate the population fitness variance :

[0093] ; ;

[0094] in, is the total number of particles in the NP group, is the individual fitness value of the i-th particle, is the overall average fitness value of the NP group, and max represents the maximum value operation;

[0095] when When it is less than the preset threshold, it is determined to be in a premature state and step 4 is executed. If it is greater than or equal to the preset threshold, return to step 3.

[0096] Step 4: Cross the NP group and the C group: select G particles with the best fitness values ​​from the C group to replace the G particles with the worst fitness values ​​in the NP group, and return to step 3 to continue iterating until the number of iterations is reached ;

[0097] Step 5: The iteration ends, and the target optimization parameter M is output. Based on the target optimization parameter M, the integrated circuit chip image is divided into M sub-images.

[0098] It's worth noting that segmenting an integrated circuit chip image into M sub-images based on the target optimization parameter M enables refined processing of the chip image. Each sub-image is relatively small, allowing the system to focus on feature analysis in a localized area. For example, during the chip manufacturing process, tiny defects such as pinholes and scratches may not be easily noticeable in the entire image. However, when the image is segmented into sub-images, these defects become more prominent in the local sub-image, thereby improving the accuracy of defect detection. Chip images may contain some background noise or irrelevant areas, which can affect defect detection accuracy. By segmenting the image into sub-images, interfering factors can be confined to a smaller area, reducing their impact on defect detection. For example, stains or light and shadow variations on the chip edge may interfere with overall defect detection. However, in a sub-image, these interfering factors only affect a portion of the sub-image, allowing more accurate defect detection in the remaining sub-images.

[0099] In some embodiments, calculating the defect feature index of each sub-image specifically includes the following process:

[0100] Calculate the contrast saliency value between the sub-image and the preset defect-free sub-image :

[0101] ;

[0102] in, is the number of superpixel regions in the sub-image, is the number of pixels in each superpixel region, Superpixel area and the superpixel area of ​​the preset defect-free sub-image The Euclidean distance between Superpixel area and preset defect-free sub-image superpixel area The distance from the center of mass, is the median distance between each pair of superpixel centroids;

[0103] Calculate the sub-image color distribution eigenvalues :

[0104] ;

[0105] in, Pick and the maximum value of 0, and They are super pixel areas in RGB color space The R, B, and G values ​​of the j-th pixel in ;

[0106] Calculate the defect feature index of each sub-image :

[0107] ;

[0108] in, 、 、 are weight coefficients, is the preset contrast significance threshold, is the preset color distribution feature threshold, is the defect coefficient of the sub-image in the hyperspectral space.

[0109] It is worth noting that obtaining the defect coefficient of the sub-image in the hyperspectral space specifically includes the following process:

[0110] Obtain the characteristic wavelength reflectance map of the sub-image in the hyperspectral space, add the preset band to the characteristic wavelength reflectance map, and extract the three eigenvectors of the characteristic wavelength reflectance map in the preset band 、 、 , the wavelengths corresponding to the vectors are 、 、 , calculate the defect coefficient of the sub-image in the hyperspectral space ; Among them, the three eigenvectors are The ratio of the reflectivity difference in the axial projection direction is equal.

[0111] In some embodiments, segmenting the integrated circuit chip defect portion in the sub-image based on the defect feature index and extracting the integrated circuit chip defect portion feature specifically includes the following process:

[0112] Divide the sub-image into several parts and determine the Does it exceed the preset threshold? If so, the part is regarded as the defective part of the integrated circuit chip, and the feature extraction of the defective part of the integrated circuit chip is performed to obtain the defective part feature of the integrated circuit chip. :

[0113] ;

[0114] in, represents the number of channels in the image, Indicates the Pixels on the The color value of each channel, Indicates the number of pixels. Indicates the The color mean of each channel.

[0115] In some embodiments, identifying and analyzing the characteristics of defective parts and generating appearance monitoring analysis results specifically include the following process:

[0116] Compare the defect part features with the defect part dataset to obtain the learning representation of the node:

[0117] Data preparation:

[0118] Collect a large amount of integrated circuit chip image data containing various defects and annotate them to clarify the characteristics of the defective parts in each image and the corresponding defect types, such as missing components, incorrect component position welding, redundant components, etc.

[0119] Construct a defect part dataset that covers various possible defect situations and ensures the diversity and representativeness of the data to improve the accuracy of subsequent comparative learning.

[0120] Feature extraction

[0121] Extract features of defective areas from the IC chip image to be analyzed. These features can include visual features such as the defect's shape, size, color, and texture, or advanced features such as edge contours and corner points extracted through image processing techniques.

[0122] Use appropriate feature extraction algorithms, such as the convolution layer and pooling layer in the convolutional neural network (CNN), to convert the defect area image into a feature vector that can be processed by the computer.

[0123] Contrastive Learning

[0124] The extracted defect features are compared with the features in the defect dataset. Similarity metrics such as Euclidean distance and cosine similarity can be used to calculate the similarity between the feature to be analyzed and each feature in the dataset.

[0125] Through comparative learning, we find the features in the dataset that are most similar to the features to be analyzed and obtain the node learning representations corresponding to these features. The node learning representations contain the feature information and category information of similar defects in the dataset.

[0126] Determine whether the learning representation of the node has a component missing representation, if so, determine that the integrated circuit chip has a component missing representation; determine whether the learning representation of the node has a component position welding error representation, if so, determine that the integrated circuit chip has a component position welding error; determine whether the learning representation of the node has a component redundancy representation, if so, determine that the integrated circuit chip has component redundancy;

[0127] The judgment result is recorded as the appearance monitoring analysis result.

[0128] In some embodiments, Figure 2 This is a workflow diagram of the first integrated circuit chip intelligent processing monitoring system according to an embodiment of the present invention. Figure 2 As shown, analyzing the processed data and generating non-appearance monitoring analysis results specifically includes the following processes:

[0129] Step S201: Acquire equipment operation data of an integrated circuit chip processing line based on processing data, wherein the equipment includes a wafer cleaning machine, a photolithography machine, an etching machine, a chemical mechanical polishing machine, an electroplating machine, and a wafer bonding machine, compare and analyze the equipment operation data with the equipment operation standard data in the processing index, and determine whether the equipment operation data is the same as the equipment operation standard data in the processing index. If so, the processing line equipment parameter coefficient JG is 1; otherwise, the processing line equipment parameter coefficient JG is 0;

[0130] Step S202: obtaining a temperature characterization value WD within a processing period based on processing data;

[0131] Step S203: obtaining the wafer cutting speed, photolithography speed, etching speed, and metal filling speed within the processing period based on the processing data, respectively calculating the difference between the wafer cutting speed and a preset wafer cutting speed threshold, the difference between the photolithography speed and a preset photolithography speed threshold, the difference between the etching speed and a preset etching speed threshold, and the difference between the metal filling speed and a preset metal filling speed threshold, taking the absolute values ​​of all the differences and summing them to obtain a sum value SD;

[0132] Step S204: obtaining the internal pressure of the processing equipment, the pressure of the gas pipeline used for etching or cleaning, and the liquid pressure in the liquid treatment process during the processing period based on the processing data, wherein the liquid treatment process includes the coating of the photoresist and the spraying of the developer, and determining whether all the pressures are within a preset range. If so, the pressure parameter YL is 1; otherwise, the pressure parameter YL is 0;

[0133] Step S205: Substitute the processing line equipment parameter coefficient JG, the temperature characterization values ​​WD and SD, and the pressure parameter YL into the processing qualification index calculation formula to calculate the processing qualification index LMS. The calculation formula is as follows:

[0134] ;

[0135] in, The value of is 2.72, 、 are weight coefficients, with values ​​of 0.6 and 0.4 respectively;

[0136] Step S206: Determine whether the processing qualification index exceeds the preset processing qualification index threshold. If so, the processing of the integrated circuit chip is determined to be qualified. If not, the processing of the integrated circuit chip is determined to be unqualified, and the judgment result is recorded as generating a non-appearance monitoring analysis result.

[0137] In some embodiments, Figure 3 This is a workflow diagram of the second integrated circuit chip intelligent processing monitoring system according to an embodiment of the present invention. Figure 3 As shown, obtaining the temperature characterization value WD during the processing period includes the following process:

[0138] Step S301: Divide the processing period into several sub-periods, and detect the temperature during wafer manufacturing, the temperature during photolithography and etching, and the temperature during ion implantation and metal filling in each sub-period;

[0139] Step S302: A rectangular coordinate system is established with the execution time of each sub-period as the X-axis and the temperature value as the Y-axis. The temperatures of each manufacturing process are connected to generate a temperature curve for the wafer manufacturing process, a temperature curve for the photolithography and etching process, and a temperature curve for the ion implantation and metal filling process.

[0140] Step S303: Calculate the total length of all temperature curves above the preset temperature curve, and record the total length of the curves as the temperature characterization value WD during the processing period.

[0141] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0142] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0143] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0144] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only for some logical functions. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0145] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0146] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An intelligent processing monitoring system for integrated circuit chips, characterized in that: The system includes: A monitoring data acquisition unit is used to acquire integrated circuit chip image data within a target area and processing data collected by sensors on a processing line, and to send the integrated circuit chip image data to an appearance analysis unit and the processing data to a non-appearance analysis unit; The appearance analysis unit is configured to receive integrated circuit chip image data, divide the integrated circuit chip image into M sub-images, calculate a defect feature index for each sub-image, segment defective parts of the integrated circuit chip in the sub-image based on the defect feature index, extract features of the defective parts of the integrated circuit chip, identify and analyze the features of the defective parts, generate appearance monitoring analysis results, and send the appearance monitoring analysis results to the monitoring and early warning unit. Calculating the defect feature index for each sub-image specifically includes the following process: Calculate the contrast saliency value between the sub-image and the preset defect-free sub-image : ； in, is the number of superpixel regions in the sub-image, is the number of pixels in each superpixel region, Superpixel area and the superpixel area of ​​the preset defect-free sub-image The Euclidean distance between Superpixel area and preset defect-free sub-image superpixel area The distance from the center of mass, is the median distance between each pair of superpixel centroids; Calculate the sub-image color distribution eigenvalues : ； in, Pick and the maximum value of 0, and They are super pixel areas in RGB color space The R, B, and G values ​​of the j-th pixel in ; Calculate the defect feature index of each sub-image : ； in, 、 、 are weight coefficients, is the preset contrast significance threshold, is the preset color distribution feature threshold, is the defect coefficient of the sub-image in the hyperspectral space; The process of obtaining the defect coefficient of the sub-image in the hyperspectral space specifically includes the following steps: Obtain the characteristic wavelength reflectance map of the sub-image in the hyperspectral space, add the preset band to the characteristic wavelength reflectance map, and extract the three eigenvectors of the characteristic wavelength reflectance map in the preset band 、 、 , the wavelengths corresponding to the vectors are 、 、 , calculate the defect coefficient of the sub-image in the hyperspectral space ; Among them, the three eigenvectors are The ratio of the reflectivity differences in the axial projection direction is equal; Segmenting the defective parts of the integrated circuit chip in the sub-image based on the defect feature index and extracting the features of the defective parts of the integrated circuit chip specifically includes the following processes: Divide the sub-image into several parts and determine the Does it exceed the preset threshold? If so, the part is regarded as the defective part of the integrated circuit chip, and the feature extraction of the defective part of the integrated circuit chip is performed to obtain the defective part feature of the integrated circuit chip. : ； in, represents the number of channels in the image, Indicates the Pixels on the The color value of each channel, Indicates the number of pixels. Indicates the The color mean of each channel; The non-appearance analysis unit is used to receive the processed data, analyze the processed data, generate non-appearance monitoring analysis results, and send the non-appearance monitoring analysis results to the monitoring and early warning unit; The monitoring and early warning unit is used to receive the appearance monitoring analysis results and the non-appearance monitoring analysis results to mark the integrated circuit chips in the target area.

2. The intelligent processing monitoring system for integrated circuit chips according to claim 1, characterized in that: The process of dividing the integrated circuit chip image into M sub-images specifically includes the following steps: Step 1: Initialize the population: set the initial population size m and the fitness threshold , maximum number of iterations , search range , randomly generate an n-dimensional vector with each component between 0 and 1 ,by is the initial value, based on the regression equation Iterate to get N vectors ,in, ; Step 2: Divide the initial population into NP group and C group: Calculate the fitness value of each particle in the initial population and sort them, select the first m particles as NP group particles, and the remaining m particles as C group particles; and record the individual extreme values ​​of the NP group and NP group global extremum , record the global extreme value of group C ; Step 3: Determine whether the maximum number of iterations has been reached. If so, proceed to step 5. Otherwise, proceed to steps 31 to 34: Step 3.1: Update particles: Update the speed and position of NP group particles using the following formula ； ； Among them, the learning factor and Both are 0.01, is the number of iterations; is the inertia weight, For the The best historical position experienced by a particle, is the best position experienced by all particles in the group, and is a random number between 0 and 1 that follows a uniform distribution. For the The particle velocity of the iteration, For the The particle position at iteration , For the The particle position at the iteration, For the The particle velocity of the iteration; Update the position of group C with the following formula: ； in, For the The particle position at the iteration, For the The particle position at the iteration, The value of is 3.57; Step 32: Update the individual extreme value of the NP group and NP group global extremum , evaluate the particle fitness value after the NP group is updated, if the individual fitness value of the particle is better than , then Replace with the current particle position; compare and ,like Greater than , then Replace with current ; Step 33: Update the global extreme value of group C :Evaluate the fitness value of group C particles. If the fitness value is better than , then Replace with the current particle position; Step 3 and 4: Calculate the population fitness variance : ； ； in, is the total number of particles in the NP group, is the individual fitness value of the i-th particle, is the overall average fitness value of the NP group, and max represents the maximum value operation; when When it is less than the preset threshold, it is determined to be in a premature state and step 4 is executed. If it is greater than or equal to the preset threshold, return to step 3; Step 4: Cross the NP group and the C group: select G particles with the best fitness values ​​from the C group to replace the G particles with the worst fitness values ​​in the NP group, and return to step 3 to continue iterating until the number of iterations is reached ; Step 5: The iteration ends, and the target optimization parameter M is output. Based on the target optimization parameter M, the integrated circuit chip image is divided into M sub-images.

3. The intelligent processing monitoring system for integrated circuit chips according to claim 1, characterized in that: Identifying and analyzing the characteristics of defective parts and generating appearance monitoring analysis results specifically include the following processes: Comparing the defect part features with the defect part dataset, a learning representation of the node is obtained, and determining whether the learning representation of the node contains a component missing representation. If so, it is determined that the integrated circuit chip has a component missing; Determine whether the learning representation of the node has a component position welding error representation, and if so, determine that the integrated circuit chip has a component position welding error; Determine whether the learning representation of the node has component redundancy representation, and if so, determine that the integrated circuit chip has component redundancy; The judgment result is recorded as the appearance monitoring analysis result.

4. The intelligent processing monitoring system for integrated circuit chips according to claim 1, characterized in that: Analyzing the processed data and generating non-appearance monitoring analysis results specifically includes the following processes: Based on the processing data, equipment operation data of the integrated circuit chip processing line is obtained, where the equipment includes a wafer cleaning machine, a photolithography machine, an etching machine, a chemical mechanical polishing machine, an electroplating machine, and a wafer bonding machine. The equipment operation data is compared and analyzed with the equipment operation standard data in the processing index to determine whether the equipment operation data is the same as the equipment operation standard data in the processing index. If so, the processing line equipment parameter coefficient JG is 1; otherwise, the processing line equipment parameter coefficient JG is 0; Acquire a temperature characterization value WD within a processing period based on processing data; Obtaining a wafer cutting speed, a photolithography speed, an etching speed, and a metal filling speed within a processing period based on the processing data, respectively calculating a difference between the wafer cutting speed and a preset wafer cutting speed threshold, a difference between the photolithography speed and a preset photolithography speed threshold, a difference between the etching speed and a preset etching speed threshold, and a difference between the metal filling speed and a preset metal filling speed threshold, taking absolute values ​​of all the differences and summing them to obtain a sum value SD; Based on the processing data, the internal pressure of the processing equipment, the pressure of the gas pipeline used for etching or cleaning, and the liquid pressure in the liquid processing process during the processing period are obtained, wherein the liquid processing process includes the coating of the photoresist and the spraying of the developer, and it is determined whether all the pressures are within a preset range. If so, the pressure parameter YL is 1; if not, the pressure parameter YL is 0; Substitute the processing line equipment parameter coefficient JG, temperature characterization values ​​WD, SD and pressure parameter YL into the processing qualification index calculation formula to calculate the processing qualification index LMS. The calculation formula is as follows: ； in, The value of is 2.72, 、 are weight coefficients, with values ​​of 0.6 and 0.4 respectively; Determine whether the processing qualification index exceeds the preset processing qualification index threshold. If so, the processing of the integrated circuit chip is determined to be qualified. If not, the processing of the integrated circuit chip is determined to be unqualified, and the judgment result is recorded as the generated non-appearance monitoring analysis result.

5. The intelligent processing monitoring system for integrated circuit chips according to claim 4, characterized in that: Obtaining the temperature characterization value WD within the processing period includes: dividing the processing period into several sub-periods, detecting the temperature of the wafer manufacturing process, the temperature of the photolithography and etching process, and the temperature of the ion implantation and metal filling process in each sub-period, establishing a rectangular coordinate system with the execution time of each sub-period as the X-axis and the temperature value as the Y-axis, connecting the temperatures in each manufacturing process to generate a temperature curve of the wafer manufacturing process, a temperature curve of the photolithography and etching process, and a temperature curve of the ion implantation and metal filling process, calculating the total length of all temperature curves above the preset temperature curve, and recording the total curve length as the temperature characterization value WD within the processing period.

Images