A domestically produced chip-based process control station online monitoring system and method

Abstract

This invention relates to the field of chip manufacturing monitoring, and more particularly to an online monitoring system and method for a domestically produced chip process control station. The system includes: real-time acquisition of parameters during the chip manufacturing process via sensors and data acquisition cards; establishment of a chip defect identification neural network model based on pre-processed chip production process data; outputting analysis and prediction results; calculation of etching rate correction parameters based on chip manufacturing processes, raw materials, chip production equipment, and basic material factors; adjustment of the etching rate based on a comparison of the etching rate correction parameters with preset etching rate correction parameter coefficients; refinement of the etching rate based on a comparison of the actual dynamic change in etching rate with the dynamic change in standard etching rate; and monitoring of etching residues based on the refined etching rate through an automated detection and feedback mechanism.

Description

Technical Field

[0001] This invention relates to the field of chip manufacturing monitoring, and in particular to an online monitoring system and method for a domestically produced chip process control station. Background Technology

[0002] With the rapid development of China's domestic chip industry, the requirements for monitoring the chip manufacturing process are becoming increasingly stringent. Traditional offline monitoring methods cannot reflect changes in the production process in real time, leading to low production efficiency and unstable product quality. In existing technologies, advancements in big data analytics have solved problems such as insufficient data processing capabilities, poor real-time performance, and inability to effectively support multiple monitoring parameters in online monitoring systems, thus improving chip manufacturing efficiency and quality to some extent.

[0003] However, existing technologies still have the following problems: they lack effective prediction and analysis of chip defects, making it impossible to perform timely maintenance through prediction. Furthermore, they fail to assess defects based on prediction results, resulting in the system's inability to quickly respond to dangerous situations, rationally allocate detection and repair resources, and accurately control the etching rate. They also lack the ability to calculate precise etching rate correction parameters by combining multiple factors, which cannot guarantee the reduction of waste products caused by non-compliant production processes. Moreover, they lack dynamic optimization of the production process and dynamic monitoring of chip etching rate deviations, which cannot improve the automation level of the etching process and leads to process instability. At the same time, they lack real-time monitoring of etching residues, making it impossible to achieve automatic adjustment of chip manufacturing process parameters.

[0004] Therefore, there is a need for an online monitoring system and method for process control stations using domestically produced chips. Summary of the Invention

[0005] This invention provides an online monitoring system and method for a domestically produced chip process control station. The purpose is to identify potential defects in the chip manufacturing process in advance through real-time data analysis and predictive models, formulate corresponding maintenance plans, and reduce equipment failures and downtime; to conduct defect assessments based on prediction results, quickly determine the type and severity of defective products, thereby optimizing resource allocation; to calculate precise etching rate correction parameters by combining multiple factors, ensuring the etching process meets production requirements and improving product yield; to dynamically optimize the production process by real-time monitoring and adjustment of the etching rate, ensuring process stability and product consistency; and to establish a real-time monitoring mechanism for etching residues, ensuring that residues are within a controllable range, thereby improving product quality.

[0006] The specific technical solution of this invention is as follows:

[0007] A method for online monitoring of a domestically produced chip process control station includes the following steps:

[0008] Step S1. Collect parameters in real time during the chip manufacturing process using sensors and data acquisition cards. Based on the pre-processed chip production process data, establish a neural network model for chip defect identification and output analysis and prediction results.

[0009] Step S2. Calculate etching rate correction parameters based on chip manufacturing process, raw materials, chip production equipment, and basic material factors. Adjust the etching rate based on the comparison results of the etching rate correction parameters and the preset etching rate correction parameter comparison coefficients. Refine the etching rate based on the comparison results of the actual etching rate dynamic change and the standard etching rate dynamic change.

[0010] Step S3. Based on the etching rate obtained after refinement, monitor the etching residue through an automated detection and feedback mechanism.

[0011] Furthermore, step S1 specifically includes:

[0012] The chip defect identification neural network model includes an input layer, an identification layer, an analysis layer, and an output layer. In the identification layer, features are extracted from the input data to identify potential chip defect patterns. In the analysis layer, the output of the identification layer is connected to the connection layer to integrate information and perform feature learning, resulting in a comprehensive analysis. The output layer outputs the results using a Softmax function. Defect evaluation parameters are set. It determines whether the emergency warning mode of the production line is triggered and classifies the chips into quality levels.

[0013] Furthermore, a time decay coefficient is added in step S1. The final chip defect estimation and evaluation parameters are obtained. The process is as follows:

[0014] in, Indicates the weighting coefficient of historical data; Indicates the size of the historical data window; Indicates the first One historical defect value; define the threshold for defect assessment parameters. , , and the final defect assessment parameters respectively with , To make a comparison, if If it is a superior product, then the etching rate is defined as follows: , ;if If it is of medium quality, then the etching rate is defined as... , ;if If the result is poor, then the etching rate is defined as follows: , ;in, This indicates the normal production etching rate; , , These represent the first, second, and third etching rate adjustment coefficients, respectively.

[0015] Furthermore, step S1 specifically includes:

[0016] if If this occurs, the emergency warning system will be triggered, immediately shutting down the system for maintenance and testing while simultaneously initiating the root cause analysis process; among which, Indicates a warning factor.

[0017] Furthermore, in step S2, the etching rate correction parameters are calculated. The specific process is as follows:

[0018] in, Indicates the influence coefficient of ion bombardment; Indicates the influence coefficient of gaseous chemical reactions; Indicates the influence coefficient of temperature effect; Mask material influence coefficient; Indicates device status; This represents the random noise term.

[0019] Furthermore, step S2 specifically includes:

[0020] Define the first preset etching rate correction reference parameter as follows: The second etching rate correction reference parameter is: , The etching rate correction parameters are now... respectively with , Compare and then, based on the comparison results, Adjust to The optimal etching rate for the current stage can be obtained by setting an objective function. .

[0021] Furthermore, step S2 specifically includes: calculating the dynamic change in the actual etching rate. Dynamic variation with standard etching rate relative deviation , Set a threshold to determine the threshold. ,if Then the etching rate needs to be refined; if Therefore, there is no need to refine the etching rate.

[0022] Furthermore, in step S2, the first preset relative deviation is defined as... The second preset relative deviation is , ,Will respectively with and A comparison was made, and the etching rate was refined based on the comparison results. The specific process is as follows:

[0023] if The refined etching rate ;if The refined etching rate ;if The refined etching rate ;in, This represents the first refinement factor. This represents the second refinement coefficient.

[0024] Furthermore, step S3 specifically includes:

[0025] Define the thickness of the residue as During real-time sampling, the state of the residue is updated. ;in, Indicates the sampling time step. Indicates the sampling interval; if , This indicates a threshold value that triggers endpoint detection or an alarm.

[0026] A domestically produced chip-based process control station online monitoring system includes the following:

[0027] Chip-related data acquisition module, chip defect identification module, defect assessment and quality grading module, etching rate control module, etching residue monitoring module;

[0028] The chip-related data acquisition module is used to collect data in real time during the chip manufacturing process, providing basic data support for subsequent analysis and decision-making.

[0029] The chip defect identification module establishes a neural network model for chip defect identification based on preprocessed chip manufacturing process data and outputs analysis and prediction results.

[0030] The defect assessment and quality grading module calculates defect assessment parameters based on the output prediction results of the chip defect identification neural network model and compares them with the defect assessment parameter thresholds to determine the chip quality grade.

[0031] The etching rate control module is used to automatically adjust the etching rate based on real-time monitoring data and preset etching rate correction parameters.

[0032] The etching residue monitoring module monitors etching residues based on the refined etching rate through an automated detection and feedback mechanism.

[0033] The etching rate control module includes a calibration parameter calculation and comparison unit, a dynamic change comparison and judgment unit, and an etching rate refinement unit.

[0034] Beneficial effects: 1. This invention establishes a chip defect identification neural network model by collecting manufacturing parameters in real time to identify anomalies in the production process. When potential defects are predicted, equipment parameters can be adjusted in advance or the machine can be shut down for maintenance to avoid batch defects. Furthermore, predictive and timely maintenance can avoid high repair costs for sudden failures. By timely identifying and classifying defects, measures can be taken quickly to prevent defective products from entering the market, thereby reducing the cost of subsequent rework and returns. Based on set thresholds, high-risk batches can be quickly isolated to prevent the spread of defects. At the same time, an emergency warning mode can be triggered to quickly respond to problems on the production line, reduce potential losses and impacts, and maintain the stable operation of the production line. By classifying chip quality into levels, detection and repair resources can be allocated more rationally, concentrating efforts on solving high-risk or high-impact defects.

[0035] 2. This invention calculates precise correction parameters based on various factors such as chip manufacturing processes, raw materials, chip production equipment, and basic materials. This eliminates systematic deviations, ensuring high precision in the etching process, reducing defects caused by uneven etching, and allowing for timely adjustment of the etching rate. This effectively reduces waste products generated due to non-compliance in the production process, dynamically optimizes the production flow, and improves overall production efficiency. Dynamic monitoring and intelligent adjustment of chip etching rate deviations reduce human intervention, increase the automation level of the etching process, ensure consistency between different batches of products, improve yield, and achieve precise control of the etching rate. It has wide applicability, avoiding process instability caused by over-adjustment and quickly correcting significant deviations, thereby improving the reliability, consistency, and production efficiency of the etching process.

[0036] 3. This invention significantly improves process control by real-time monitoring of etching residues. Through dynamic prediction of residue thickness and real-time measurement data, closed-loop control of etching depth can be achieved, reducing the risk of over-etching or under-etching and preventing device failure due to residue thickness deviations. By optimizing the etching endpoint time, invalid process waiting or rework is avoided, allowing for precise process termination, saving equipment time and energy consumption, reducing wafer scrap due to process failures, and directly lowering production costs. Integration with a MES system enables automatic feedback adjustment of process parameters, providing a foundation for the digitalization and intelligentization of etching processes, and further optimizing the model through machine learning. Attached Figure Description

[0037] Figure 1This is a flowchart of an online monitoring method for a domestically produced chip process control station according to the present invention;

[0038] Figure 2 This is a block diagram of an online monitoring system for a domestically produced chip process control station according to the present invention;

[0039] Figure 3 This is a schematic diagram of the etching rate control module described in this invention. Detailed Implementation

[0040] To better understand the above technical solutions, a detailed description of the solutions will be provided below in conjunction with the accompanying drawings and specific embodiments. It should also be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of the invention.

[0041] See attached document Figure 1 This embodiment provides an online monitoring method for a domestically produced chip process control station, including the following steps:

[0042] S1. Collect parameters in real time during the chip manufacturing process using sensors and data acquisition cards. Based on the pre-processed chip production process data, establish a neural network model for chip defect identification and output analysis and prediction results.

[0043] By using sensors and data acquisition cards, key parameters in the chip manufacturing process are collected in real time, such as temperature, pressure, flow rate, pressure valve opening, equipment status, process parameters, and environmental conditions, to ensure stable production. The collected data is also preprocessed, including filtering, noise reduction, and data normalization.

[0044] Based on the preprocessed chip manufacturing process data, a chip defect identification neural network model is established. The preprocessed historical chip manufacturing process data is used as training samples and as input data for the chip defect identification neural network model. The model is trained based on deep learning. The chip defect identification neural network model includes an input layer, an identification layer, an analysis layer, and an output layer.

[0045] Input data is represented as , , This represents chip production-related data obtained at any timestamp. , This represents the different timestamps at which the data was obtained. Each timestamp is then used to obtain... This type of data information is represented as ,in, As the input to the neural network model for chip defect identification, there are a total of One neuron, will The data is then passed to the recognition layer, and the specific process is as follows:

[0046] in, This represents the input to the recognition layer; This represents the connection weights between the input layer and the recognition layer; This represents the output value of the input layer; This indicates the bias of the recognition layer.

[0047] In the recognition layer, features are extracted from the input data to identify potential chip defect patterns. The specific process is as follows:

[0048] in, This represents the output of the recognition layer; Indicates the total number of production batches; Indicates the first Feature weight coefficients in each batch; Indicates in Chip-related feature values ​​in the time step; Indicates bias; Represents the stability coefficient; This represents the threshold. If the quality of the finished chip is determined to be greater than or equal to the threshold in the recognition layer, the recognition result is passed to the analysis layer.

[0049] In the analysis layer, the output of the recognition layer is connected to the connection layer to integrate information and learn features, making a comprehensive analysis. The specific process is as follows:

[0050] , in, This represents the input to the analysis layer; This represents the connection weight values ​​between the recognition layer and the analysis layer; Indicates the bias of the second hidden layer; This represents the output of the second hidden layer; This represents the balance coefficient, used to identify the impact of key parameters during the balance control identification process; Indicates the total number of identified features; Indicates the first One identification variable parameter; Indicates the first Expected bias of each identification variable; Indicates the first Standard parameters for each identification variable; This represents the constant coefficient.

[0051] Finally, the analysis and prediction results are output in the output layer using the Softmax function. , ,in, This indicates the output result of the output layer; This represents the connection weight values ​​between the lifting layer and the output layer; This indicates the bias of the lifting layer.

[0052] In actual production, we continue to monitor product quality, collect new data and update the model regularly, and optimize model parameters and structure based on the new data to improve the accuracy of chip defect prediction and chip quality.

[0053] Based on the output analysis and prediction results of the chip defect identification neural network model, defect assessment parameters are set. Determine whether the production line emergency warning mode has been triggered; classify chips into quality levels and set etching rates. .

[0054] Based on the output prediction results of the chip defect identification neural network model, the defect assessment parameters are calculated. Set the output prediction results , Indicates the probability of a good product, 0 , Indicates the chip number The probability of class defects, Define the process sensitivity coefficient. This reflects the severity of various defects; the specific process is as follows:

[0055] in, This represents the yield sensitivity coefficient, which is recommended to be 0.6-0.9 in this embodiment of the invention; This represents the defect weighting coefficient, which is recommended to be 0.3-0.5 in this embodiment of the invention; This represents the magnification factor for the most severe defect; a value of 0.2-0.4 is recommended. Increase the time decay factor. For continuous chip production, the final and accurate chip defect estimate can be obtained. The process is as follows:

[0056] in, Indicates the final defect assessment parameters; Indicates the attenuation coefficient; This represents the weighting coefficient for historical data, balancing the importance of the current value with the historical average. ; This indicates the size of the historical data window, i.e., the number of samples. Indicates the first A historical defect value, from the past N detections. The record is stored in the buffer.

[0057] Define defect assessment parameter thresholds , , and the final defect assessment parameters respectively with , A comparison is made to determine the chip quality grade and, based on the comparison results, to determine whether to trigger the emergency warning mode, while simultaneously setting the etching rate. The specific process is as follows:

[0058] Simultaneously define , The threshold value for defect assessment parameters, and ;if If it is a superior product, then the etching rate is defined as follows: , ;

[0059] if If it is of medium quality, then the etching rate is defined as... , ;

[0060] if If the result is poor, then the etching rate is defined as follows: , ;

[0061] if If this occurs, the emergency warning system will be triggered, the machine will be immediately shut down for maintenance and testing, and the root cause analysis process will be initiated.

[0062] in, The warning factor is set based on historical experimental and production experience and expert advice. This indicates the normal production etching rate; , , These represent the first, second, and third etching rate adjustment coefficients, respectively.

[0063] This invention uses real-time acquisition of manufacturing parameters to establish a neural network model for chip defect identification, recognizing anomalies in the production process, reducing the flow of defective products into subsequent stages, and predicting potential defects to adjust equipment parameters or halt maintenance in advance, avoiding batch defects. Furthermore, predictive and timely maintenance can avoid high repair costs associated with sudden failures. By promptly identifying and classifying defects, measures can be taken quickly to prevent defective products from entering the market, thereby reducing subsequent rework and return costs. Based on set thresholds, high-risk batches can be quickly isolated to prevent defect spread, while an emergency warning mode is triggered to rapidly respond to production line issues, reducing potential losses and impacts, and maintaining stable production line operation. By classifying chip quality into levels, testing and repair resources can be allocated more rationally, concentrating efforts on resolving high-risk or high-impact defects.

[0064] The output of neural network models can provide data support for production decisions, help management make more scientific judgments, improve product quality and production efficiency, systematically assess and classify defects, help companies continuously improve production processes, improve the overall quality and reliability of products, and thus enhance the company's market competitiveness.

[0065] S2. Calculate etching rate correction parameters based on various factors such as chip manufacturing process, raw materials, chip production equipment, and base materials. Correct parameters based on etching rate The etching rate is adjusted based on the comparison results with the preset etching rate correction parameter coefficient, and the dynamic changes in the etching rate are also considered. The etching rate was refined by comparing the results with the dynamic changes of the standard etching rate.

[0066] Etching rate correction parameters are calculated based on various factors such as chip manufacturing process, chip production equipment, and basic materials. The specific process is as follows:

[0067] in, Indicates the influence coefficient of ion bombardment; Indicates the influence coefficient of gaseous chemical reactions; Indicates the influence coefficient of temperature effect; Mask material influence coefficient; Indicates device status; Represents random noise;

[0068] Specifically ,in, This represents the ion flux density. Indicates sputtering output. Indicates ion energy; ,in, Indicates the first The reaction rate of the gas Indicates the first Partial pressure of the gas Indicates the activation energy of the reaction. Indicates substrate temperature. Represents the Boltzmann constant; Among them, data obtained through Arrhenius fitting of experimental data, This indicates the pre-activation factor of the thermal activation process. Representing the thermal activation energy, after separating the Arrhenius term, the relationship between the residual rate and temperature is fitted. Represents the power-law coefficient. Indicates the temperature power exponent; ,in, Indicates the mask shadow angle. Indicates mask thickness. Indicates the mean free path of ions / radicals; ,in, Indicates the initial calibration coefficient of the equipment. Indicates the continuous operating time of the equipment. Indicates the maintenance cycle. Indicates the current radio frequency power. The rated radio frequency power is indicated. In this embodiment of the invention, the above parameters can be set by existing publicly available methods such as direct reading from measuring devices, experimental calibration, literature reference, or simulation tools.

[0069] Correction parameters based on etching rate The etching rate is adjusted based on the comparison results with the preset etching rate correction parameters, and the dynamic changes in the etching rate are taken into account. The etching rate was refined by comparing the results with the dynamic changes of the standard etching rate.

[0070] Define the first preset etching rate correction reference parameter as follows: The second etching rate correction reference parameter is: , The etching rate correction parameters are now... respectively with , The comparison is conducted, and adjustments are made based on the comparison results. The specific process is as follows:

[0071] if Then the etching rate will be adjusted to , ;

[0072] if Then the etching rate will be adjusted to , ;

[0073] if Then the etching rate will be adjusted to , ;

[0074] in, This indicates a reduction in the coefficient; This indicates an increase in the coefficient.

[0075] Define the target etching rate as Through adjustable parameters For example: partial pressure of gases Substrate temperature Current radio frequency power Parameters, etc., make Approaching At the same time, minimize noise That is, to establish the objective function: ;in, This represents the regularization coefficient, used to balance etching rate accuracy and stability; This represents the noise standard deviation. The optimal etching rate for the current stage is obtained through iterative gradient descent. .

[0076] Dynamic change of etching rate Dynamic variation with standard etching rate For comparison, the dynamic change of the standard etching rate The etching rate is determined based on process requirements or historical data, and the decision to refine the etching rate is made based on the comparison results. The specific process is as follows:

[0077] in, This represents the proportional term, which quickly responds to the current error and adjusts immediately. ; This represents the integral term, eliminating historical accumulated errors; It represents the differential term, predicts error trends, and reduces oscillations; This indicates an error signal.

[0078] Calculate the dynamic change of the actual etching rate Dynamic variation of the defined standard etching rate relative deviation , And set a threshold to judge the threshold. ,if Then the etching rate needs to be refined; if Therefore, there is no need to refine the etching rate.

[0079] Define the first preset relative deviation as The second preset relative deviation is , ,Will respectively with and A comparison was made, and the etching rate was refined based on the comparison results. The specific process is as follows:

[0080] if This indicates that the deviation between the actual change and the standard change is small, suggesting that the chip etching rate is relatively high. The etching rate is already precise enough and requires no adjustment; therefore, the refined etching rate... ;

[0081] if This indicates that the deviation between the actual change and the standard change is within a moderate range, requiring a slight adjustment to the etching rate. Therefore, the refined etching rate... ;

[0082] if This indicates a significant deviation between the actual change and the standard change, necessitating a substantial adjustment to the etching rate. Therefore, the refined etching rate... ;

[0083] in, This represents the first refinement factor. This represents the second refinement coefficient. .

[0084] This invention calculates precise correction parameters based on various factors such as chip manufacturing processes, raw materials, chip production equipment, and basic materials. This eliminates systematic deviations, ensuring high precision in the etching process, reducing defects caused by uneven etching, and allowing for timely adjustment of the etching rate. This effectively reduces waste products generated due to non-compliance in the production process, dynamically optimizes the production flow, and improves overall production efficiency. Dynamic monitoring and intelligent adjustment of chip etching rate deviations reduce human intervention, increase the automation level of the etching process, ensure consistency between different batches of products, improve yield, and achieve precise control of the etching rate. It has wide applicability, avoiding process instability caused by over-adjustment and quickly correcting significant deviations, thereby improving the reliability, consistency, and production efficiency of the etching process.

[0085] S3. Based on the etching rate obtained after refinement By using automated detection and feedback mechanisms to monitor etching residues, the reliability of the chip manufacturing process is monitored, ensuring that each production step meets quality standards and improving chip quality.

[0086] Based on the etching rate obtained after refinement The etching residue was monitored, and the specific process is as follows:

[0087] Define the thickness of the residue as The dynamic change in residue thickness can be expressed as The integral is expressed as:

[0088] in, Indicates the initial thickness; This indicates a perturbation term representing the redeposition of residues or incomplete etching.

[0089] During real-time sampling, the state of the residue is updated. ;in, Indicates the sampling time step. Indicates the sampling interval;

[0090] if , This indicates a threshold value that triggers endpoint detection or an alarm.

[0091] The residue thickness is corrected through a feedback mechanism, as shown in the following process:

[0092] in, Indicates the first The thickness of the residue after step correction; Indicates the filter gain; Indicates the first Predicting the thickness of the residue in the next step; Indicates the first The actual etching depth and the amount of etching removal directly measured by equipment such as ellipsometry and SEM; Indicates the first The predicted etching depth can be determined based on... Calculate. If Prioritize trusting actual measured data; It depends on the predicted value.

[0093] When the thickness of the residue is close to the target value The criteria for determining when to terminate etching are:

[0094] in, This indicates the allowable error. At this point, the following must be satisfied: , indicating from the initial time Time to finish Etching rate The integral over time equals the difference between the initial residue thickness and the target thickness, where, Indicates the end time. It is the target variable in chip manufacturing process control.

[0095] This invention significantly improves process control by real-time monitoring of etching residues. Through dynamic prediction of residue thickness and real-time measurement data, closed-loop control of etching depth can be achieved, reducing the risk of over-etching or under-etching and preventing device failure due to residue thickness deviations. By optimizing the etching endpoint time, invalid process waiting or rework is avoided, allowing for precise process termination, saving equipment time and energy, reducing wafer scrap due to process failures, and directly lowering production costs. Integration with a MES system enables automatic feedback adjustment of process parameters, providing a foundation for the digitalization and intelligentization of etching processes, and further optimizing the model through machine learning.

[0096] See attached document Figure 2This embodiment provides an online monitoring system for a domestically produced chip process control station, including the following:

[0097] Chip-related data acquisition module, chip defect identification module, defect assessment and quality grading module, etching rate control module, etching residue monitoring module;

[0098] The chip-related data acquisition module is used to collect various data in the chip production process in real time, providing basic data support for subsequent analysis and decision-making, and ensuring the accuracy and real-time performance of the monitoring system.

[0099] The chip defect identification module establishes a neural network model for chip defect identification based on preprocessed chip manufacturing process data and outputs analysis and prediction results.

[0100] The defect assessment and quality grading module calculates defect assessment parameters based on the output prediction results of the chip defect identification neural network model and compares them with the defect assessment parameter thresholds to determine the chip quality grade.

[0101] The etching rate control module is used to automatically adjust the etching rate based on real-time monitoring data and preset etching rate correction parameters.

[0102] The etching residue monitoring module monitors the etching rate obtained after refinement. The reliability of the chip manufacturing process is monitored by detecting etching residues through automated detection and feedback mechanisms.

[0103] See attached document Figure 3 The etching rate control module includes a calibration parameter calculation and comparison unit, a dynamic change comparison and judgment unit, and an etching rate refinement unit.

[0104] The calibration parameter calculation and comparison unit is used to calculate the etching rate calibration parameters based on various factors such as chip manufacturing process, chip production equipment, and base materials, compare them with the calibration comparison parameters, and make adjustments based on the comparison results.

[0105] The dynamic change comparison and judgment unit is used to calculate the relative deviation between the dynamic change of the etching rate and the dynamic change of the standard etching rate, and to determine whether the etching rate should be refined based on the relative deviation.

[0106] The etching rate refinement unit is used to compare the relative deviation with the preset relative deviation and refine the etching rate based on the comparison results.

[0107] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0108] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0109] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0110] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0111] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. An online monitoring method for a domestically produced chip process control station, characterized in that, Includes the following steps: Step S1. Collect parameters in real time during the chip manufacturing process using sensors and data acquisition cards. Based on the pre-processed chip production process data, establish a neural network model for chip defect identification and output analysis and prediction results. Increase time decay coefficient The final chip defect estimation and evaluation parameters are obtained. The process is as follows: in, Indicates the weighting coefficient of historical data; Indicates the size of the historical data window; Indicates the first One historical defect value; define the threshold for defect assessment parameters. , The final chip defect estimation and evaluation parameters respectively with , To make a comparison, if If it is a superior product, then the etching rate is defined as follows: , ;if If it is of medium quality, then the etching rate is defined as... , ;if If the result is poor, then the etching rate is defined as follows: , ;in, This indicates the normal production etching rate; , , These represent the first, second, and third etching rate adjustment coefficients, respectively. if If this occurs, the emergency warning system will be triggered, immediately shutting down the system for maintenance and testing while simultaneously initiating the root cause analysis process; among which, Indicates the warning factor; Step S2. Calculate etching rate correction parameters based on chip manufacturing process, raw materials, chip production equipment, and basic material factors. Adjust the etching rate based on the comparison results of the etching rate correction parameters and the preset etching rate correction parameter comparison coefficients. Refine the etching rate based on the comparison results of the actual etching rate dynamic change and the standard etching rate dynamic change. Step S3. Monitor etching residues based on the refined etching rate through an automated detection and feedback mechanism.

2. The online monitoring method for a domestically produced chip process control station according to claim 1, characterized in that, Step S1 specifically includes: the chip defect identification neural network model includes an input layer, an identification layer, an analysis layer, and an output layer; in the identification layer, features are extracted from the input data to identify potential chip defect patterns; in the analysis layer, the output of the identification layer is connected to the connection layer to integrate information and perform feature learning, making a comprehensive analysis; in the output layer, the results are output through the Softmax function; and defect evaluation parameters are set. It determines whether the emergency warning mode of the production line is triggered and classifies the chips into quality levels.

3. The online monitoring method for a domestically produced chip process control station according to claim 1, characterized in that, In step S2, the etching rate correction parameter is calculated. The specific process is as follows: in, Indicates the influence coefficient of ion bombardment; Indicates the influence coefficient of gaseous chemical reactions; Indicates the influence coefficient of temperature effect; Mask material influence coefficient; Indicates device status; This represents the random noise term.

4. The online monitoring method for a domestically produced chip process control station according to claim 1, characterized in that, Step S2 specifically includes: defining the first preset etching rate correction reference parameter as follows: The second etching rate correction reference parameter is: , The etching rate correction parameters are now... respectively with , Compare and then, based on the comparison results, Adjust to The optimal etching rate for the current stage can be obtained by setting an objective function. ; if Then the etching rate will be adjusted to , ;if Then the etching rate will be adjusted to , ;if Then the etching rate will be adjusted to , ;in, This indicates a reduction in the coefficient; This indicates an increase in the coefficient.

5. The online monitoring method for a domestically produced chip process control station according to claim 1, characterized in that, Step S2 specifically includes: calculating the dynamic change of the actual etching rate. Dynamic variation with standard etching rate relative deviation , Set a threshold to determine the threshold. ,if Then the etching rate needs to be refined; if Therefore, there is no need to refine the etching rate.

6. The online monitoring method for a domestically produced chip process control station according to claim 5, characterized in that, In step S2, the first preset relative deviation is defined as... The second preset relative deviation is , ,Will respectively with and A comparison was made, and the etching rate was refined based on the comparison results. The specific process is as follows: If The refined etching rate ;if The refined etching rate ;if The refined etching rate ;in, This represents the first refinement factor. This represents the second refinement coefficient.

7. The online monitoring method for a domestically produced chip process control station according to claim 1, characterized in that, Step S3 specifically includes: defining the thickness of the residue as... During real-time sampling, the state of the residue is updated. ;in, Indicates the sampling time step. Indicates the sampling interval; if , This indicates a threshold value that triggers endpoint detection or an alarm.

8. An online monitoring system for a domestically produced chip process control station, applied to the online monitoring method for a domestically produced chip process control station as described in claim 1, characterized in that, Includes the following: Chip-related data acquisition module, chip defect identification module, defect assessment and quality grading module, etching rate control module, etching residue monitoring module; The chip-related data acquisition module is used to collect data in real time during the chip production process, providing basic data support for subsequent analysis and decision-making. The chip defect identification module establishes a chip defect identification neural network model based on preprocessed chip manufacturing process data and outputs analysis and prediction results. The defect assessment and quality grading module calculates defect assessment parameters based on the output prediction results of the chip defect identification neural network model and compares them with the defect assessment parameter thresholds to determine the chip quality grading. The etching rate control module is used to automatically adjust the etching rate based on real-time monitored data and preset etching rate correction parameters. The etching residue monitoring module monitors etching residues based on the refined etching rate through an automated detection and feedback mechanism. The etching rate control module includes a calibration parameter calculation and comparison unit, a dynamic change comparison and judgment unit, and an etching rate refinement unit.

Images