A method and system for monitoring a chip manufacturing production line

Abstract

This invention relates to the field of semiconductor-related technologies, specifically including a method and system for monitoring chip manufacturing production lines. The method includes: collecting photolithography process monitoring data; setting a feature matrix; determining the focus window boundary; configuring focal length offset compensation based on the feature matrix, combined with the focus window boundary and photolithography process monitoring data; and using an edge placement error model to determine a deviation amplitude score and provide graded alerts. This addresses the technical problem that the window boundary definition in chip manufacturing production line monitoring relies on experience-based fixed standards, which cannot effectively compensate for focal length offset and poses a risk of excessive edge placement errors. The method achieves this by performing coupled analysis of optical proximity effects and process windows in the mask layer, lens layer, and photoresist layer, constructing a feature matrix, dynamically matching compensation parameters with error trends, improving the accuracy of focal length offset compensation, and using an edge placement error model to determine a deviation amplitude score and provide graded alerts, thus enhancing the timeliness of anomaly handling.

Description

Technical Field

[0001] This invention relates to the field of semiconductor-related technologies, specifically to a method and system for monitoring chip manufacturing production lines. Background Technology

[0002] As the core step in chip pattern transfer, the photolithography process directly determines the consistency of key wafer dimensions through parameters such as exposure intensity, focal length shift, and optical proximity effect. The etching process, on the other hand, requires precise control of plasma state, temperature distribution, and other conditions to ensure that the pattern outline meets design requirements.

[0003] Conventional chip manufacturing production line monitoring methods only focus on independent parameters such as exposure dose and focal length, which cannot accurately define the focus window boundary and focal length offset compensation amount, posing a risk of excessive edge placement error. In addition, the focus window boundary definition relies on empirical data and does not consider the dynamic response relationship between exposure intensity and photoresist sensitivity under different focal length offsets, resulting in a discrepancy between the defined process safety zone and the actual operating conditions.

[0004] In summary, existing technologies have technical problems, such as relying on experience to set fixed standards for defining the window boundaries of chip processing production line monitoring, which cannot effectively compensate for focal length offsets and poses a risk of excessive edge placement errors. Summary of the Invention

[0005] This application provides a chip processing production line monitoring method and system, aiming to solve the technical problem in the prior art where the window boundary definition of chip processing production line monitoring relies on experience to set fixed standards, which cannot effectively compensate for focal length offset and poses a risk of excessive edge placement error.

[0006] In view of the above problems, the technical solution to achieve the present application is as follows:

[0007] In a first aspect, this application provides a method for monitoring a chip manufacturing production line. The method includes: collecting photolithography process monitoring data from the chip manufacturing production line, including a photolithography process segment; performing a coupling analysis of optical proximity effect and process window in the mask layer, lens layer, and photoresist layer, combining exposure intensity, and setting a feature matrix; determining the focus window boundary based on the intersection of the exposure intensity and the response points of the exposure sensitivity curve under different focal length offset conditions; configuring a focal length offset compensation amount under the photoresist refractive index based on the feature matrix, combined with the focus window boundary and the photolithography process monitoring data, wherein the focal length offset compensation amount is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect; and using the edge placement error model to determine a deviation amplitude score and provide graded alerts.

[0008] Preferably, the focus window boundary and the corresponding candidate exposure dose are used to define the process safety zone; wherein, the candidate exposure dose is determined by a combination of process parameters that minimizes the sum of the absolute values ​​of the critical dimension prediction deviations under the minimum process window.

[0009] Preferably, the optical proximity effect and the coupling coefficient of the process window are introduced as weight matrices; matrix multiplication is performed on the basic matrices corresponding to the mask layer, lens layer, and photoresist layer and the weight matrix to generate the feature matrix.

[0010] Preferably, multiple field units are divided by the wafer field corresponding to the mask layer, lens layer, and photoresist layer; based on the multiple field units, process feature vectors are arranged in order of field spatial location to configure the basic matrix.

[0011] Preferably, convolution operations are performed on the optical system point diffusion layer network and the photochemical reaction diffusion layer network to simulate the overlapping region of light intensity distribution and adhesive layer reaction threshold under different focal length offset conditions and exposure dose combinations; based on the simulation results, the sensitivity of key dimensions to changes in process parameters is quantified, and the process feature vector is configured.

[0012] Preferably, the optical system point diffusion layer network is set based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer; and the photochemical reaction diffusion layer network is set based on the thickness, refractive index, and exposure sensitivity curve of the photoresist layer.

[0013] Preferably, the initial focal length offset compensation amount is denoted as ; ,in, This represents the actual refractive index of the photoresist under the current temperature and humidity conditions. NA represents the thickness of the photoresist layer, and NA represents the numerical aperture of the projection lens module. The root mean square value of the prediction residual of the edge placement error model is given. The residual nonlinear response index is calibrated according to the photoresist type. The gradient coupling coefficient is the number of elements at the mask pattern density. The spatial gradient magnitude is given by the mask pattern density; within the wafer field, the initial focal length offset compensation is dynamically corrected, and the focal length offset compensation is... ,in, for The error between the exposure focal length at a given moment and the optimal focus position determined based on the focus window boundary. , The proportional-integral control coefficient, for The time integral term of the exposure focal length at a given moment and the error between the optimal focus position determined based on the focus window boundary.

[0014] Preferably, the chip processing production line includes a photolithography process section and an etching process section; based on the etching process section, etching process monitoring data including chamber plasma emission spectrum, radio frequency impedance matching status and wafer electrostatic chuck temperature distribution are collected; based on the etching process monitoring data and photolithography process monitoring data, a collaborative error propagation chain model between the photolithography process section and the etching process section is established.

[0015] Preferably, the influence gain coefficient of lithography residual error on etching profile offset is quantified through the collaborative error propagation chain model; the risk level of the profile offset of the etching process segment is judged by the influence gain coefficient and the process tolerance band; if it exceeds the process tolerance band, the affected wafer batch is marked and an anomaly alert is issued.

[0016] In a second aspect, this application provides a chip manufacturing production line monitoring system, comprising: a data acquisition module for acquiring photolithography process monitoring data of the chip manufacturing production line, including the photolithography process segment; a coupling analysis module for performing optical proximity effect and process window coupling analysis in the mask layer, lens layer, and photoresist layer, combined with exposure intensity, and setting a feature matrix; a focus window boundary determination module for determining the focus window boundary based on the exposure intensity and the intersection of the response of the exposure sensitivity curve under different focal length offset conditions; a focal length offset compensation configuration module for configuring the focal length offset compensation amount under the photoresist refractive index based on the feature matrix, combined with the focus window boundary and the photolithography process monitoring data, wherein the focal length offset compensation amount is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect; and a graded reminder module for determining the deviation magnitude score using the edge placement error model and providing graded reminders.

[0017] In summary, one or more technical solutions provided in this application achieve the following technical effects: by conducting coupled analysis of optical proximity effect and process window in the mask layer, lens layer, and photoresist layer and constructing a feature matrix, dynamically matching compensation parameters and error trends, improving the accuracy of focal length offset compensation, using an edge placement error model to determine the deviation magnitude score and provide graded alerts, and enhancing the timeliness of anomaly handling. Attached Figure Description

[0018] Figure 1 This application provides a flowchart illustrating a method for monitoring a chip manufacturing production line.

[0019] Figure 2 This application provides a schematic diagram of the structure of a chip processing production line monitoring system.

[0020] Explanation of reference numerals in the attached diagram: Data acquisition module M100, coupling analysis module M200, focus window boundary determination module M300, focal length offset compensation configuration module M400, and graded reminder module M500. Detailed Implementation

[0021] Example 1: The present application will be described in detail below with reference to the accompanying drawings, as follows... Figure 1 As shown, this application provides a method for monitoring a chip manufacturing production line, wherein the method includes:

[0022] S1: Collect photolithography process monitoring data from the chip manufacturing production line, including the photolithography process segment; S2: In the mask layer, lens layer, and photoresist layer, perform optical proximity effect and process window coupling analysis by combining exposure intensity, and set the feature matrix.

[0023] Specifically, lithography process monitoring data refers to the data related to the lithography process collected by the detection device, including parameters such as exposure intensity, focal length shift, and photoresist thickness, used to monitor the operating status of the lithography process in real time; optical proximity effect refers to the deviation of the light intensity distribution at the edge of the lithographic pattern from the ideal situation due to light diffraction and interference phenomena, thus affecting the accurate transfer of the pattern; process window refers to the range of process parameters that can ensure the pattern transfer accuracy and yield during the lithography process, which usually includes a combination of multiple parameters such as exposure dose and focal length shift; coupling analysis is used to evaluate the relationship between optical proximity effect and process window; feature matrix is ​​used to integrate relevant data of mask layer, lens layer, and photoresist layer for more accurate analysis and calculation.

[0024] Execution steps: Sensors installed in the lithography process section collect real-time monitoring data, such as exposure intensity, focal length shift, and photoresist thickness. In the mask layer, lens layer, and photoresist layer, a coupling analysis of the optical proximity effect and process window is performed, combining exposure intensity. Specifically, by introducing the coupling coefficient between the optical proximity effect and the process window as a weight matrix, matrix multiplication is performed on the corresponding fundamental and weight matrices of the mask layer, lens layer, and photoresist layer to generate a feature matrix. Through this coupling analysis and feature matrix setting in the mask layer, lens layer, and photoresist layer, a systematic integration of core influencing factors in lithography is achieved, enabling a more comprehensive evaluation of the lithography process's operational status and improving monitoring accuracy.

[0025] S3: Based on the exposure intensity and the intersection of the exposure sensitivity curve under different focal length shift conditions, determine the focus window boundary; S4: Based on the feature matrix, combined with the focus window boundary and the photolithography process monitoring data, configure the focal length shift compensation amount under the photoresist refractive index. The focal length shift compensation amount is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect; S5: Use the edge placement error model to determine the deviation amplitude score and provide graded reminders.

[0026] Specifically, the exposure sensitivity curve describes the response characteristics of photoresist under different exposure doses, reflecting the sensitivity of the photoresist to physicochemical changes during the exposure to development process. It is used to determine the optimal exposure dose to ensure accurate transfer of the lithographic pattern. The response intersection point is the intersection of the exposure intensity and the exposure sensitivity curve under different focal length offsets. It reflects the response of the photoresist to the exposure intensity under a specific focal length offset and is a key basis for determining the focus window boundary. The focus window boundary refers to the boundary of the focal length offset range that can ensure accurate transfer of the lithographic pattern during the lithography process. It defines the process safety zone, that is, within the process safety zone, the lithography process can be carried out stably, and the error of pattern transfer is within an acceptable range. The focal length offset compensation amount is the compensation amount set to correct the deviation between the actual focal length and the optimal focal length. The focal length offset compensation amount is configured based on the feature matrix, the focus window boundary, and the lithography process monitoring data to ensure the accuracy of the lithography process.

[0027] The edge placement error model is used to predict the error in the placement of lithographic patterns. It considers multiple factors such as optical proximity effect and focal length shift, and can provide predicted values ​​for the deviation of the pattern edge position. The prediction residual refers to the difference between the actual measured edge placement error and the predicted value of the edge placement error model, reflecting the accuracy of the model prediction and serving as the basis for adjusting the focal length shift compensation. The deviation amplitude score is a quantitative score based on the deviation amplitude determined by the edge placement error model. It is used to evaluate the stability and accuracy of the lithography process, serving as the basis for graded alerts to promptly identify and address potential process problems. Graded alerts refer to issuing different levels of alert signals based on different levels of deviation amplitude scores, helping operators to understand the process status in a timely manner, take corresponding measures, and prevent further deterioration of process problems.

[0028] Execution steps: Based on exposure intensity, analyze the response of the exposure sensitivity curve under different focal length shifts, identify the intersection points of the exposure intensity and the response of the exposure sensitivity curve. These intersection points determine the focus window boundary. Through these intersection points, determine the process safety zone, that is, within this zone, the lithography process can be carried out stably, and the pattern transfer error is within an acceptable range. Combining the feature matrix, focus window boundary, and lithography process monitoring data, calculate the focal length shift compensation amount. Furthermore, the focal length shift compensation amount has a strong correlation with the prediction residual of the edge placement error model under the optical proximity effect, ensuring that the compensation amount can dynamically match the error trend and improve the compensation accuracy. Using the edge placement error model, calculate the deviation amplitude based on the actual monitoring data, quantify and score the deviation amplitude, and provide graded alerts based on the scoring results to promptly identify and address potential process problems. In the above steps, the focus window boundary is determined by the intersection of the exposure intensity and exposure sensitivity curves. This allows for dynamic adjustment of the process safety zone based on actual operating conditions, making process control more precise and significantly reducing the risk of exceeding edge placement error limits. The focal length offset compensation amount configured based on the feature matrix, focus window boundary, and lithography process monitoring data is strongly correlated with the prediction residual of the edge placement error model. This dynamic matching method can significantly improve compensation accuracy and reduce pattern transfer errors caused by focal length offset. The edge placement error model is used to determine the deviation amplitude score and provide graded reminders. Different levels of reminder signals can be issued according to different levels of deviation amplitude, helping operators to understand the process status in a timely manner.

[0029] Furthermore, the method for determining the focus window boundaries in this application includes:

[0030] The focus window boundary and the corresponding candidate exposure dose are used to define the process safety zone; wherein, the candidate exposure dose is determined by the combination of process parameters that minimizes the sum of the absolute values ​​of the critical dimension prediction deviations under the minimum process window.

[0031] Specifically, the focus window boundary is defined as the focal length offset range that ensures accurate transfer of the lithographic pattern under specific exposure conditions. It is determined by analyzing the intersection of the exposure intensity and exposure sensitivity curves and is used to define the process safety zone. The candidate exposure dose refers to the exposure dose in the lithography process that minimizes the sum of the absolute values ​​of the critical dimension prediction deviations. The candidate exposure dose is determined by optimizing the combination of process parameters within the minimum process window and is used to ensure accurate transfer of the lithographic pattern. The process safety zone refers to the range of process parameters that ensure accurate transfer of the lithographic pattern in the lithography process. This area is jointly defined by the focus window boundary and the candidate exposure dose, ensuring that the lithography process can proceed stably within this area and that the pattern transfer error is within an acceptable range. The critical dimension prediction deviation refers to the deviation between the actual measured critical dimension and the design dimension. In the lithography process, accurate control of the critical dimension is a key factor in ensuring chip performance. The minimum process window refers to the minimum range of process parameters that ensures accurate transfer of the lithographic pattern in the lithography process. Within this range, even small changes in process parameters can lead to significant deviations in pattern transfer.

[0032] The specific process is as follows: By analyzing the intersection points of the exposure intensity and exposure sensitivity curves under different focal length offset conditions, the focus window boundaries are determined. These response intersection points define the focal length offset range that ensures accurate transfer of the lithographic pattern under specific exposure conditions. Within the minimum process window, by optimizing the combination of process parameters, the exposure dose that minimizes the sum of the absolute values ​​of the predicted deviations of critical dimensions is obtained. This exposure dose is called the candidate exposure dose and is used to define the process safety zone. Combining the focus window boundaries and the candidate exposure dose, the process safety zone is defined. Within the process safety zone, the lithography process can proceed stably, and the error in pattern transfer is within an acceptable range. In the above steps, the combination of the focus window boundaries and the candidate exposure dose can accurately define the process safety zone. Compared with the method of setting fixed boundaries based on experience, the definition of the process safety zone is more scientific and precise. By optimizing the combination of process parameters within the minimum process window, the exposure dose that minimizes the sum of the absolute values ​​of the predicted deviations of critical dimensions is determined, improving the accuracy and stability of the lithography process. Accurately defining the process safety zone and optimizing the exposure dose improves the stability of the lithography process.

[0033] Furthermore, in the mask layer, lens layer, and photoresist layer, optical proximity effect and process window coupling analysis are performed by combining exposure intensity, and a feature matrix is ​​set. The method of this application includes:

[0034] An optical proximity effect and a process window coupling coefficient are introduced as weight matrices; matrix multiplication is performed on the fundamental matrices corresponding to the mask layer, lens layer, and photoresist layer and the weight matrices to generate the feature matrix.

[0035] Specifically, the optical proximity effect refers to the deviation between the light intensity distribution at the edge of the lithographic pattern and the ideal situation caused by light diffraction and interference during photolithography, thus affecting the accurate transfer of the pattern; the process window refers to the range of process parameters that ensure accurate pattern transfer during photolithography, typically including a combination of multiple parameters such as exposure dose and focal length offset; the coupling coefficient is used to quantify the mutual influence between the optical proximity effect and the process window, and as part of the weight matrix, it is used to adjust and optimize the calculation of the feature matrix; the fundamental matrix represents the matrix of characteristics of the mask layer, lens layer, and photoresist layer, including their relationship with the photolithography process. The physical parameters related to the process, such as thickness and refractive index, are used. The weight matrix, which contains coupling coefficients, is used to weight the base matrix when calculating the characteristic matrix. Introducing the weight matrix allows the characteristic matrix to more accurately reflect the interaction between the optical proximity effect and the process window. Matrix multiplication is used to multiply two matrices to generate a new matrix. Furthermore, by multiplying the base matrix and the weight matrix, the characteristic matrix is ​​generated. The characteristic matrix is ​​the matrix generated by matrix multiplication and is used to comprehensively represent the characteristics of the mask layer, lens layer, photoresist layer, and the interaction between the optical proximity effect and the process window.

[0036] The specific process is as follows: Define the coupling coefficient between the optical proximity effect and the process window, and introduce it as a weight matrix. The coupling coefficient reflects the degree of mutual influence between the optical proximity effect and the process window. For the mask layer, lens layer, and photoresist layer, configure their corresponding basic matrices. The basic matrices contain physical parameters related to the photolithography process, such as thickness and refractive index. Perform matrix multiplication on the basic matrices and the weight matrices. Specifically, multiply each basic matrix by the weight matrix to generate a feature matrix. The feature matrix integrates the characteristics of the mask layer, lens layer, and photoresist layer, as well as the interaction between the optical proximity effect and the process window. In the above steps, by introducing a weight matrix and performing matrix multiplication, the generated feature matrix systematically integrates the characteristics of the mask layer, lens layer, and photoresist layer, as well as the interaction between optical proximity effect and process window. This allows for a more comprehensive evaluation of the lithography process's operating status and improves the monitoring accuracy of the lithography process. By introducing the weight matrix and generating the feature matrix, a foundation is provided for subsequent process parameter optimization. Furthermore, by analyzing the feature matrix, parameters such as exposure dose and focal length offset can be accurately adjusted, thereby optimizing the lithography process. Through systematic integration and optimization of process parameters, the stability of the lithography process can be significantly improved, enhancing production efficiency and product quality.

[0037] Furthermore, the method of this application involves performing matrix multiplication operations on the fundamental matrices corresponding to the mask layer, lens layer, and photoresist layer and the weight matrix.

[0038] Multiple field units are divided into wafer fields corresponding to the mask layer, lens layer, and photoresist layer; based on the multiple field units, process feature vectors are arranged in order of field spatial location, and the basic matrix is ​​configured.

[0039] Specifically, the wafer field refers to the area on the wafer exposed by the lithography machine during the lithography process. It is usually composed of multiple small exposure areas, which are exposed one by one during the lithography process. The field unit refers to the wafer field further subdivided into smaller unit regions, used to more accurately analyze and control the changes of various parameters during the lithography process. The process feature vector is a vector describing various characteristic parameters in the lithography process, including exposure intensity, focal length shift, photoresist thickness, etc., used to characterize the state of the lithography process. Arrangement by field spatial position means arranging the process feature vectors in a certain order according to the spatial position of the field units on the wafer. This arrangement helps to systematically represent and process these feature vectors in the matrix. The fundamental matrix is ​​a matrix used to represent and process various characteristic parameters in the lithography process. The fundamental matrix is ​​constructed by arranging the process feature vectors.

[0040] The specific process is as follows: Based on the wafer field corresponding to the mask layer, lens layer, and photoresist layer, the wafer field is subdivided into multiple field units. The division of these multiple field units can be adjusted according to the actual photolithography process requirements and precision requirements. For each field unit, a corresponding process feature vector is extracted. This process feature vector includes parameters such as exposure intensity, focal length shift, and photoresist thickness for that field unit. The process feature vectors are arranged into a basic matrix according to the spatial order of the field units on the wafer. In the above steps, by subdividing the wafer field into multiple field units and performing precise parameter analysis on each unit, fine-grained control of the photolithography process can be achieved, allowing for more accurate capture of minute changes during the photolithography process. This improves the precision and consistency of the photolithography process and reduces pattern transfer errors caused by changes in process parameters. By arranging the process feature vectors in spatial order to form a basic matrix, various parameters in the photolithography process are systematically integrated, enabling a more comprehensive evaluation of the photolithography process's operating status.

[0041] Furthermore, in arranging the process feature vectors according to their spatial location within the field, the method of this application also includes:

[0042] Convolution operations are performed on the optical system point diffusion layer network and the photochemical reaction diffusion layer network to simulate the overlapping region of light intensity distribution and adhesive layer reaction threshold under different focal length offset conditions and exposure dose combinations; based on the simulation results, the sensitivity of key dimensions to changes in process parameters is quantified, and the process feature vector is configured.

[0043] Specifically, the optical system point diffusion hierarchy network describes the function by which an optical system diffuses a point light source into a light spot. In photolithography, the PSF (Point Spread Filter) is used to simulate the propagation and distribution of light during the lithography process and is a tool for evaluating optical imaging quality. The photochemical reaction diffusion hierarchy network describes the photochemical reactions and diffusion processes of photoresist during exposure, considering the chemical properties and physical processes of the photoresist, and is used to predict the reaction of the photoresist under different exposure conditions. Convolution operations are used to simulate the interaction between the optical system and the photoresist, generating overlapping regions of light intensity distribution and photoresist reaction thresholds. Light intensity distribution describes the propagation and distribution of light in space. The intensity distribution during the exposure process determines the exposure of the photoresist and is a key factor affecting the quality of pattern transfer. The overlap region of the photoresist reaction threshold refers to the area where the photoresist undergoes a chemical reaction under specific exposure conditions. The size and shape of the overlap region directly affect the formation and quality of the photolithographic pattern. Critical dimension sensitivity refers to the sensitivity of critical dimensions to changes in process parameters. Furthermore, precise control of critical dimensions is a key factor in ensuring chip performance. The process feature vector is a vector used to describe various feature parameters in the photolithography process, including exposure intensity, focal length shift, and photoresist thickness, and is used to characterize the state of the photolithography process.

[0044] The specific process involves convolution operations on the optical system's point diffusion layer network and the photochemical reaction diffusion layer network. Specifically, the optical system's PSF is convolved with the photochemical reaction diffusion model of the photoresist to simulate the overlap region between light intensity distribution and photoresist reaction threshold under different focal length offset conditions and exposure dose combinations. Through convolution operations, light intensity distribution maps and photoresist reaction threshold maps are generated, visually displaying the exposure and reaction of the photoresist under different process parameters. Based on the simulation results of the convolution operations, the sensitivity of critical dimensions to changes in process parameters is quantified. Specifically, by analyzing the overlap region between light intensity distribution and photoresist reaction threshold, the changes in critical dimensions under different exposure doses and focal length offset conditions are evaluated. Based on the quantification results, a process feature vector is configured. This process feature vector contains important parameters such as critical dimension sensitivity, which are used for subsequent process optimization and control. In the above steps, convolution operations can accurately simulate the overlapping area of ​​light intensity distribution and resist reaction threshold under different focal length offset conditions and exposure dose combinations, more accurately predicting the exposure and reaction of photoresist, and improving the accuracy and consistency of photolithography. Simulating the overlapping area of ​​light intensity distribution and resist reaction threshold through convolution operations reduces pattern transfer errors caused by changes in process parameters, improving the accuracy of photolithography. By quantifying the sensitivity of key dimensions to changes in process parameters, the stability and reliability of photolithography can be accurately evaluated.

[0045] Furthermore, the method of this application includes performing convolution operations on the optical system point diffusion layer network and the photochemical reaction diffusion layer network, namely:

[0046] Based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer, a point diffusion layer network for the optical system is set; based on the thickness, refractive index, and exposure sensitivity curve of the photoresist layer, a photochemical reaction diffusion layer network is set.

[0047] Specifically, numerical aperture describes the ability of an optical system to collect light and is a crucial parameter for optical lenses. Furthermore, numerical aperture affects light focusing ability and resolution. Coherence coefficient describes the coherence of the light source and influences the interference effect during photolithography. Furthermore, coherence coefficient determines the coherence of the light source and the propagation characteristics of light. Aberration distribution describes the degree to which light deviates from the ideal path in an optical system. Aberrations affect the precise transfer of photolithographic patterns and therefore require precise control and correction. The optical system point spread hierarchy network describes the function by which an optical system diffuses a point light source into a light spot. Furthermore, PSF (point spread function) is a key parameter in optical systems. The point spread function (PSF) is used to simulate the propagation and distribution of light during photolithography and is a tool for evaluating optical imaging quality. The thickness of the photoresist layer affects the absorption and propagation of light during photolithography. The refractive index of the photoresist layer refers to its optical properties, affecting the propagation speed and refraction behavior of light within the photoresist. The exposure sensitivity curve describes the reaction characteristics of the photoresist under different exposure doses, reflecting the sensitivity of the photoresist to physicochemical changes during exposure and development. The photochemical reaction diffusion hierarchy network is a model describing the photochemical reactions and diffusion processes that occur in the photoresist during exposure, considering the chemical properties and physical processes of the photoresist, and is used to predict the reaction of the photoresist under different exposure conditions.

[0048] Execution steps: First, based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer, a point diffusion hierarchy network for the optical system is constructed. Specifically, the numerical aperture determines the optical system's ability to collect light, affecting light focusing and resolution; the coherence coefficient describes the coherence of the light source, affecting light interference; and the aberration distribution describes the degree to which light deviates from the ideal path in the optical system, requiring precise control and correction. These parameters are used to generate the optical system's PSF (Photoform Filter), which simulates the propagation and distribution of light during photolithography. Second, based on the thickness, refractive index, and exposure sensitivity curve of the photoresist layer, a photochemical reaction diffusion hierarchy network is constructed. Specifically, the photoresist layer thickness refers to the physical thickness of the photoresist layer, affecting light absorption and propagation; the refractive index of the photoresist layer describes its optical properties, affecting the propagation speed and refraction behavior of light within the photoresist; and the exposure sensitivity curve describes the photoresist's reaction characteristics under different exposure doses, reflecting the photoresist's sensitivity. Finally, based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer, a photochemical reaction diffusion model is generated to predict the photoresist's reaction under different exposure conditions. In the above steps, by precisely setting the optical system point diffusion layer network and the photochemical reaction diffusion layer network, the physical and chemical behavior in the photolithography process can be simulated more accurately, thereby improving the precision and consistency of the photolithography process. By precisely setting the PSF and the photochemical reaction diffusion model, the pattern transfer error caused by changes in process parameters is reduced, thereby improving the precision of the photolithography process.

[0049] Furthermore, based on the feature matrix, and combined with the focusing window boundary and the photolithography process monitoring data, the focal length offset compensation amount under the photoresist refractive index is configured. The method of this application also includes:

[0050] Let the initial focal length offset compensation amount be denoted as ; ,in, This represents the actual refractive index of the photoresist under the current temperature and humidity conditions. NA represents the thickness of the photoresist layer, and NA represents the numerical aperture of the projection lens module. The root mean square value of the prediction residual of the edge placement error model is given. The residual nonlinear response index is calibrated according to the photoresist type. The gradient coupling coefficient is the number of elements at the mask pattern density. The spatial gradient magnitude is given by the mask pattern density; within the wafer field, the initial focal length offset compensation is dynamically corrected, and the focal length offset compensation is... ,in, for The error between the exposure focal length at a given moment and the optimal focus position determined based on the focus window boundary. , The proportional-integral control coefficient, for The time integral term of the exposure focal length at a given moment and the error between the optimal focus position determined based on the focus window boundary.

[0051] Specifically, the initial focal length offset compensation amount refers to the initial compensation amount set during the photolithography process to correct the deviation between the actual focal length and the optimal focal length. This initial focal length offset compensation amount is calculated based on the physical properties of the photoresist and other process parameters. The actual refractive index of the photoresist refers to its actual refractive index under current temperature and humidity conditions, affecting the propagation speed and refraction behavior of light within the photoresist. The photoresist layer thickness is the physical thickness of the photoresist layer, affecting light absorption and propagation. The numerical aperture refers to the numerical aperture of the projection lens module, describing the optical system's ability to collect light and affecting light focusing and resolution. The root mean square value of the prediction residual refers to the root mean square value of the prediction residual of the edge placement error model, reflecting the accuracy of the edge placement error model's predictions. The residual nonlinear response index refers to the residual nonlinear response index calibrated according to the photoresist type, used to adjust the influence of residuals; the gradient coupling coefficient refers to the gradient coupling coefficient under mask pattern density, used to adjust the influence of mask pattern density; the spatial gradient magnitude refers to the spatial gradient magnitude under mask pattern density, reflecting the complexity of the mask pattern; dynamic correction refers to the dynamic adjustment of the initial focal length offset compensation amount based on real-time monitoring data within the wafer field to ensure the accuracy of the photolithography process; the proportional-integral control coefficient is the coefficient used for proportional-integral control, used to adjust the dynamic response of the compensation amount; the error time integral term refers to the time integral term of the error between the exposure focal length and the optimal focus position from the start time to the current time, used for integral control.

[0052] Execution steps: Based on the actual refractive index of the photoresist under the current temperature and humidity conditions. Photoresist layer thickness Numerical aperture (NA) of the projection lens module and root mean square value of the predicted residuals of the edge placement error model. Residual nonlinear response index Gradient coupling coefficient under mask pattern density Spatial gradient magnitude under mask pattern density Determine the initial focal length offset compensation amount , Within the wafer field, the initial focal length offset compensation is dynamically corrected based on real-time monitoring data. Specifically, a proportional-integral control method is used, taking into account the error between the current exposure focal length and the optimal focus position. The time integral term of the sum of errors Dynamically adjust the focal length shift compensation amount , In the above steps, determining the initial focal length offset compensation amount allows for dynamic correction of the focal length offset compensation amount based on the physical properties of the photoresist and process parameters. Through precise compensation and dynamic adjustment, the deviation between the actual focal length and the optimal focal length is corrected, reducing pattern transfer errors caused by changes in process parameters and improving the accuracy and consistency of the photolithography process. By dynamically correcting the focal length offset compensation amount, the compensation amount can be dynamically adjusted based on real-time monitoring data, ensuring the accuracy of the photolithography process. Using proportional-integral control, errors caused by changes in process parameters are effectively reduced, improving the stability of the photolithography process.

[0053] Furthermore, the method of this application also includes:

[0054] The chip processing production line includes a photolithography section and an etching section. Based on the etching section, etching process monitoring data, including chamber plasma emission spectrum, radio frequency impedance matching status, and wafer electrostatic chuck temperature distribution, are collected. Based on the etching process monitoring data and the photolithography process monitoring data, a collaborative error propagation chain model between the photolithography section and the etching section is established.

[0055] Specifically, the etching process refers to the part of the chip manufacturing line used to remove material to form specific patterns. The precision of the etching process directly affects the final structure and performance of the chip. The chamber plasma emission spectrum refers to the spectrum emitted by the plasma during the etching process, reflecting the chemical composition and energy state of the plasma. By analyzing the chamber plasma emission spectrum, the chemical reactions and plasma stability during the etching process can be monitored. The RF impedance matching state refers to the impedance matching between the RF power supply and the plasma during the etching process. Good impedance matching can improve energy transmission efficiency and ensure the stability and uniformity of the etching process. The wafer electrostatic chuck temperature distribution refers to the temperature distribution of the wafer on the electrostatic chuck. The uniformity of the temperature distribution affects the uniformity and repeatability of the etching process. The cooperative error propagation chain model is a mathematical model that describes the error propagation relationship between the lithography process and the etching process. It is used to quantify the impact of residual lithography errors on etching contour offset and to assess the contour offset risk of the etching process.

[0056] Execution steps: In the etching process segment, etching process monitoring data such as chamber plasma emission spectrum, RF impedance matching status, and wafer electrostatic chuck temperature distribution are collected for real-time monitoring of the etching process operation status. Based on the collected etching and lithography process monitoring data, a collaborative error propagation chain model between the lithography and etching processes is established. Specifically, the monitoring data of the lithography process segment is analyzed to determine residual lithography errors; the monitoring data of the etching process segment is analyzed to determine the contour offset during the etching process; a collaborative error propagation chain model is set up through mathematical modeling to quantify the gain coefficient of the impact of residual lithography errors on the etching contour offset; based on the gain coefficient, the contour offset risk of the etching process segment is assessed, and anomaly alerts are issued. In the above steps, by collecting monitoring data from the etching process segment, it is possible to comprehensively monitor the changes in various parameters during the etching process, ensuring the stability and uniformity of the etching process, and improving the accuracy and consistency of the etching process. By establishing a collaborative error transmission chain model, the influence of residual lithography errors on etching contour offset is quantified, realizing the collaborative optimization of the lithography and etching processes, reducing pattern transfer errors caused by changes in process parameters, and improving the overall accuracy and yield of chip processing.

[0057] Furthermore, this application establishes a collaborative error propagation chain model between the photolithography and etching processes. The method includes:

[0058] The collaborative error propagation chain model is used to quantify the gain coefficient of the influence of residual lithography error on etching profile offset. Based on the gain coefficient, the offset risk level of the profile offset of the etching process segment is judged according to the process tolerance band. If it exceeds the process tolerance band, the affected wafer batch is marked and an anomaly alert is issued.

[0059] Specifically, the collaborative error propagation chain model describes and quantifies the error propagation relationship between the lithography and etching processes, and can assess the impact of residual lithography errors on etching contour offset. The influence gain coefficient represents the degree of influence of residual lithography errors on etching contour offset, calculated through the collaborative error propagation chain model, and is used to assess the contour offset risk of the etching process. The process tolerance band represents the acceptable range of etching contour offset under normal process conditions. Offsets exceeding this range are considered abnormal and require handling. The offset risk level judgment is used to determine whether the contour offset of the etching process is within the acceptable range based on the influence gain coefficient and the process tolerance band. If it exceeds the process tolerance band, it is considered to have a high offset risk. The anomaly alert means that when the contour offset of the etching process is detected to exceed the process tolerance band, the system will issue an alert signal to notify the operator to handle it in a timely manner.

[0060] Execution steps: Analyze the impact of residual lithography errors on etching profile offset using a collaborative error propagation chain model. Specifically, input monitoring data from the lithography process segment, including residual lithography errors; input monitoring data from the etching process segment, including etching profile offset; use the collaborative error propagation chain model to determine the gain coefficient of the impact of residual lithography errors on etching profile offset; based on the calculated gain coefficient and the process tolerance band, assess the offset risk level of the etching profile offset, specifically determining the upper and lower limits of the process tolerance band; compare the etching profile offset with the upper and lower limits of the process tolerance band; if the etching profile offset exceeds the process tolerance band, mark the affected wafer batch and issue an anomaly alert. In the above steps, by quantifying the influence gain coefficient, the impact of residual lithography error on etching profile offset is accurately assessed, thereby improving the stability and yield of the etching process and reducing pattern transfer error caused by changes in process parameters. By accurately assessing risks and issuing timely anomaly alerts, the stability of the etching process can be significantly improved, and production efficiency and product quality can be significantly enhanced. When the etching profile offset is detected to exceed the process tolerance band, anomaly alerts are issued in a timely manner to ensure that process problems are dealt with promptly and to prevent further deterioration.

[0061] In summary, the beneficial effects of the embodiments of this application are:

[0062] This application provides a chip manufacturing production line monitoring method and system. It achieves improved focus shift compensation accuracy by conducting coupled analysis of optical proximity effects and process windows in the mask layer, lens layer, and photoresist layer, combining exposure intensity with the process window coupling analysis, and setting a feature matrix. Based on the exposure intensity and the intersection of the exposure sensitivity curve under different focal length shift conditions, the focus window boundary is determined. Based on the feature matrix, combined with the focus window boundary and the lithography process monitoring data, the focal length shift compensation amount under the photoresist refractive index is configured. The focal length shift compensation amount is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect. Using the edge placement error model, the deviation amplitude score is determined, and graded alerts are provided.

[0063] Example 2, based on the same inventive concept as the chip manufacturing production line monitoring method in the foregoing examples, such as... Figure 2 As shown in the figure, this application provides a chip processing production line monitoring system, wherein the system includes:

[0064] Data acquisition module M100: Collects photolithography process monitoring data from the chip manufacturing production line, including the photolithography process section.

[0065] Coupling Analysis Module M200: In the mask layer, lens layer, and photoresist layer, it performs coupling analysis of optical proximity effect and process window in combination with exposure intensity, and sets the feature matrix.

[0066] Focus window boundary determination module M300: Based on the exposure intensity and the intersection of the response of the exposure sensitivity curve under different focal length shift conditions, the focus window boundary is determined.

[0067] Focal length offset compensation configuration module M400: Based on the feature matrix, combined with the focus window boundary and the photolithography process monitoring data, configures the focal length offset compensation under the photoresist refractive index. The focal length offset compensation is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect.

[0068] The graded reminder module M500 uses the edge placement error model to determine the deviation magnitude score and provide graded reminders.

[0069] Furthermore, the focus window boundary determination module M300 is used to perform the following method:

[0070] The focus window boundary and the corresponding candidate exposure dose are used to define the process safety zone; wherein, the candidate exposure dose is determined by the combination of process parameters that minimizes the sum of the absolute values ​​of the critical dimension prediction deviations under the minimum process window.

[0071] Furthermore, the coupling analysis module M200 is used to perform the following methods:

[0072] An optical proximity effect and a process window coupling coefficient are introduced as weight matrices; matrix multiplication is performed on the fundamental matrices corresponding to the mask layer, lens layer, and photoresist layer and the weight matrices to generate the feature matrix.

[0073] Furthermore, the coupling analysis module M200 is also used to perform the following methods:

[0074] Multiple field units are divided into wafer fields corresponding to the mask layer, lens layer, and photoresist layer; based on the multiple field units, process feature vectors are arranged in order of field spatial location, and the basic matrix is ​​configured.

[0075] Furthermore, the coupling analysis module M200 is also used to perform the following methods:

[0076] Convolution operations are performed on the optical system point diffusion layer network and the photochemical reaction diffusion layer network to simulate the overlapping region of light intensity distribution and adhesive layer reaction threshold under different focal length offset conditions and exposure dose combinations; based on the simulation results, the sensitivity of key dimensions to changes in process parameters is quantified, and the process feature vector is configured.

[0077] Furthermore, the coupling analysis module M200 is also used to perform the following methods:

[0078] Based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer, a point diffusion layer network for the optical system is set; based on the thickness, refractive index, and exposure sensitivity curve of the photoresist layer, a photochemical reaction diffusion layer network is set.

[0079] Furthermore, the focal length offset compensation configuration module M400 is also used to perform the following method:

[0080] Let the initial focal length offset compensation amount be denoted as ; ,in, This represents the actual refractive index of the photoresist under the current temperature and humidity conditions. NA represents the thickness of the photoresist layer, and NA represents the numerical aperture of the projection lens module. The root mean square value of the prediction residual of the edge placement error model is given. The residual nonlinear response index is calibrated according to the photoresist type. The gradient coupling coefficient is the number of elements at the mask pattern density. The spatial gradient magnitude is given by the mask pattern density; within the wafer field, the initial focal length offset compensation is dynamically corrected, and the focal length offset compensation is... ,in, for The error between the exposure focal length at a given moment and the optimal focus position determined based on the focus window boundary. , The proportional-integral control coefficient, for The time integral term of the exposure focal length at a given moment and the error between the optimal focus position determined based on the focus window boundary.

[0081] Furthermore, the chip manufacturing production line monitoring system is also used to perform the following methods:

[0082] The chip processing production line includes a photolithography section and an etching section. Based on the etching section, etching process monitoring data, including chamber plasma emission spectrum, radio frequency impedance matching status, and wafer electrostatic chuck temperature distribution, are collected. Based on the etching process monitoring data and the photolithography process monitoring data, a collaborative error propagation chain model between the photolithography section and the etching section is established.

[0083] Furthermore, the chip manufacturing production line monitoring system is also used to perform the following methods:

[0084] The collaborative error propagation chain model is used to quantify the gain coefficient of the influence of residual lithography error on etching profile offset. Based on the gain coefficient, the offset risk level of the profile offset of the etching process segment is judged according to the process tolerance band. If it exceeds the process tolerance band, the affected wafer batch is marked and an anomaly alert is issued.

[0085] In summary, any step can be stored as a computer instruction or program in an unrestricted computer memory and can be called and recognized by an unrestricted computer processor; no further restrictions are imposed here.

[0086] Furthermore, the above technical solutions only embody the preferred technical solutions of the embodiments of this application. Any changes that those skilled in the art may make to certain parts of these solutions embody the novel principles of the embodiments of this application. Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application.

Claims

1. A method for monitoring a chip manufacturing production line, characterized in that, The method includes: Collect lithography process monitoring data from the chip manufacturing production line, including the lithography process section; In the mask layer, lens layer, and photoresist layer, the optical proximity effect and process window coupling analysis are performed by combining the exposure intensity, and a feature matrix is ​​set. Based on the exposure intensity and the intersection of the response of the exposure sensitivity curve under different focal length shifts, the focus window boundary is determined; Based on the feature matrix, combined with the focusing window boundary and the photolithography process monitoring data, the focal length offset compensation amount under the photoresist refractive index is configured. The focal length offset compensation amount is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect. Using the edge placement error model, the deviation magnitude score is determined and a graded alert is issued.

2. The chip manufacturing production line monitoring method as described in claim 1, characterized in that, The method for determining the focus window boundaries includes: The focus window boundary and the corresponding candidate exposure dose are used to define the process safety zone; The candidate exposure dose is determined by a combination of process parameters that minimizes the sum of the absolute values ​​of the critical dimension prediction deviations within the minimum process window.

3. The chip manufacturing production line monitoring method as described in claim 1, characterized in that, In the mask layer, lens layer, and photoresist layer, optical proximity effect coupled with process window analysis is performed by combining exposure intensity, and a feature matrix is ​​set. The method includes: The optical proximity effect and the coupling coefficient of the process window are introduced as weight matrices; The feature matrix is ​​generated by performing matrix multiplication on the fundamental matrix corresponding to the mask layer, lens layer, and photoresist layer and the weight matrix.

4. The chip manufacturing production line monitoring method as described in claim 3, characterized in that, The method involves performing matrix multiplication on the fundamental matrices corresponding to the mask layer, lens layer, and photoresist layer with the weight matrix. Multiple field units are divided based on the wafer field corresponding to the mask layer, lens layer, and photoresist layer; Based on the multiple field units, the process feature vectors are arranged in order of field spatial location, and the basic matrix is ​​configured.

5. The chip manufacturing production line monitoring method as described in claim 4, characterized in that, The method further includes arranging process feature vectors according to their spatial location within the field. Convolution operations were performed on the optical system point diffusion hierarchy network and the photochemical reaction diffusion hierarchy network to simulate the overlapping region of light intensity distribution and adhesive reaction threshold under different focal length offset conditions and exposure dose combinations. The sensitivity of key dimensions to changes in process parameters is quantified based on simulation results, and the process feature vector is configured accordingly.

6. The chip manufacturing production line monitoring method as described in claim 5, characterized in that, The method involves performing convolution operations on the point diffusion hierarchy network of an optical system and the photochemical reaction diffusion hierarchy network, including: Based on the numerical aperture, coherence coefficient, and aberration distribution of the lens layer, a point diffusion hierarchical network for the optical system is established. Based on the thickness, refractive index, and exposure sensitivity curves of the photoresist layer, the photochemical reaction diffusion layer network is set up.

7. The chip manufacturing production line monitoring method as described in claim 4, characterized in that, Based on the feature matrix, combined with the focusing window boundary and the photolithography process monitoring data, the focal length offset compensation amount under the photoresist refractive index is configured, and the method further includes: Let the initial focal length offset compensation amount be denoted as ; ,in, This represents the actual refractive index of the photoresist under the current temperature and humidity conditions. NA represents the thickness of the photoresist layer, and NA represents the numerical aperture of the projection lens module. The root mean square value of the prediction residual of the edge placement error model is given. The residual nonlinear response index is calibrated according to the photoresist type. The gradient coupling coefficient is the number of elements at the mask pattern density. The spatial gradient magnitude under the mask pattern density; Within the wafer field, the initial focal length offset compensation amount is dynamically corrected, and the focal length offset compensation amount... ,in, for The error between the exposure focal length at a given moment and the optimal focus position determined based on the focus window boundary. , The proportional-integral control coefficient, for The time integral term of the exposure focal length at a given moment and the error between the optimal focus position determined based on the focus window boundary.

8. The chip manufacturing production line monitoring method as described in claim 1, characterized in that, The method further includes: The chip processing production line includes a photolithography section and an etching section. Based on the etching process segment, etching process monitoring data including chamber plasma emission spectrum, radio frequency impedance matching status, and wafer electrostatic chuck temperature distribution are collected. Based on the etching process monitoring data and the photolithography process monitoring data, a collaborative error propagation chain model between the photolithography process segment and the etching process segment is established.

9. The chip manufacturing production line monitoring method as described in claim 8, characterized in that, A collaborative error propagation chain model between the photolithography and etching processes is established, the method comprising: The gain coefficient for quantifying the influence of residual lithography error on etching profile offset is obtained through the aforementioned collaborative error propagation chain model. The risk level of the contour offset of the etching process segment is determined by the influence gain coefficient and the process tolerance band. If the offset exceeds the process tolerance band, the affected wafer batch is marked and an anomaly alert is issued.

10. A monitoring system for a chip manufacturing production line, characterized in that, The system is used to implement the chip manufacturing production line monitoring method according to any one of claims 1-9, wherein the system comprises: Data acquisition module: Collects photolithography process monitoring data from the chip manufacturing production line, including the photolithography process segment; Coupling Analysis Module: In the mask layer, lens layer, and photoresist layer, optical proximity effect and process window coupling analysis are performed in combination with exposure intensity, and feature matrix is ​​set; Focus window boundary determination module: Based on the exposure intensity and the intersection of the response of the exposure sensitivity curve under different focal length shift conditions, the focus window boundary is determined; Focal length offset compensation configuration module: Based on the feature matrix, combined with the focus window boundary and the photolithography process monitoring data, configure the focal length offset compensation under the photoresist refractive index. The focal length offset compensation is strongly correlated with the prediction residual of the edge placement error model under the optical proximity effect. Graded Reminder Module: Using the edge placement error model, determine the deviation magnitude score and provide graded reminders.

Images