Wafer surface pollution analysis method, device and equipment

Abstract

The invention relates to the technical field of wafer pollution analysis, and discloses a wafer surface pollution analysis method, device and equipment, and the method comprises the steps: carrying out the multi-source detection data collection of the surface of a wafer through an ultraviolet module, a terahertz module and a microwave module, carrying out the information mapping of the multi-source data, constructing a pollution feature space, applying environment disturbance, and collecting disturbance data; according to the method, wafer surface pollution is identified, pollution distribution is verified, cleaning adaptability is analyzed, multi-dimensional cleaning strategy analysis is carried out based on pollution characteristic distribution, an optimal cleaning scheme is generated, and the method is beneficial for achieving accurate identification and dynamic evaluation of wafer surface pollution, improving adaptability and scientificity of the pollution cleaning scheme, optimizing the cleaning effect and reducing cost. The manufacturing yield and stability are enhanced, and the problem that the pollution distribution condition of the wafer surface is difficult to detect in a single identification mode in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of wafer contamination analysis, and in particular to a method, device and equipment for analyzing wafer surface contamination. Background Art

[0002] Detecting and cleaning wafer surface contamination is a crucial step in the semiconductor manufacturing process, especially in modern integrated circuit (IC) production, where control requirements for wafer surface contamination are becoming increasingly stringent. Surface contamination not only affects wafer quality but can also cause defects during production, impacting product yield and performance. Traditional wafer contamination detection methods generally rely on single physical or chemical sensing methods, making it difficult to fully and accurately reflect the multi-dimensional characteristics of wafer surface contamination. This is especially true in complex, multi-source contamination environments. Without accurate identification of contamination distribution on the wafer surface, it's difficult to determine the most appropriate cleaning solution, leading to issues such as over-cleaning and under-cleaning. Summary of the Invention

[0003] The purpose of the present invention is to provide a wafer surface contamination analysis method, device and equipment, aiming to solve the problem in the prior art that a single identification method is difficult to detect the contamination distribution status of the wafer surface.

[0004] The present invention is implemented as follows: In a first aspect, the present invention provides a method for analyzing wafer surface contamination, comprising: The wafer surface is sensed and data is collected through ultraviolet, terahertz, and microwave detection modules to obtain multi-source detection data. Performing internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; Applying environmental disturbance to the wafer surface and collecting multi-source detection data during the environmental disturbance to substitute the data into the surface contamination feature space for contamination distribution verification and contamination cleaning adaptability analysis to generate a surface contamination feature distribution; A multi-dimensional analysis of the surface contamination cleaning strategy of the wafer is performed according to the surface contamination characteristic distribution to generate a surface contamination cleaning plan for the wafer.

[0005] In a second aspect, the present invention provides a wafer surface contamination analysis device, which is used to implement a wafer surface contamination analysis method according to any one of the first aspects, comprising: The data acquisition module is used to collect sensor data from the wafer surface through ultraviolet, terahertz and microwave detection modules to obtain multi-source detection data; An information mapping module, configured to perform internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; A contamination analysis module, configured to apply environmental disturbances to the wafer surface and collect multi-source detection data obtained during the environmental disturbances, and substitute the data into the surface contamination feature space to perform contamination distribution verification and contamination cleaning adaptability analysis, thereby generating a surface contamination feature distribution; The solution analysis module is used to perform a multi-dimensional analysis of the surface contamination cleaning strategy of the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer.

[0006] In a third aspect, the present invention provides a wafer surface contamination analysis device, comprising a memory and a processor, wherein the memory stores a wafer surface contamination analysis program that can be run on the processor, and when the processor executes the wafer surface contamination analysis program, it implements a wafer surface contamination analysis method described in any one of the first aspects.

[0007] The present invention provides a wafer surface contamination analysis method, which has the following beneficial effects: The present invention uses ultraviolet, terahertz and microwave modules to collect multi-source detection data on the wafer surface, performs information mapping on the multi-source data, constructs a pollution feature space, applies environmental disturbance and collects disturbance data, verifies pollution distribution and analyzes cleaning adaptability, performs multi-dimensional cleaning strategy analysis based on pollution feature distribution, and generates an optimal cleaning solution. This method is beneficial for achieving accurate identification and dynamic evaluation of wafer surface contamination, improving the adaptability and scientific nature of pollution cleaning solutions, optimizing cleaning effects, reducing costs, and enhancing manufacturing yield and stability, solving the problem in the prior art that a single identification method is difficult to detect the pollution distribution status of the wafer surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is a schematic diagram of the steps of a wafer surface contamination analysis method provided by an embodiment of the present invention; Figure 2 It is a structural schematic diagram of a wafer surface contamination analysis device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0009] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0010] The implementation of the present invention is described in detail below with reference to specific embodiments.

[0011] Reference Figure 1 、 Figure 2 As shown, a preferred embodiment of the present invention is provided.

[0012] In a first aspect, the present invention provides a method for analyzing wafer surface contamination, comprising: S1: Collect sensor data from the wafer surface through ultraviolet, terahertz, and microwave detection modules to obtain multi-source detection data; S2: performing internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; S3: applying environmental disturbance to the wafer surface and collecting multi-source detection data during the environmental disturbance, and substituting the data into the surface contamination feature space to perform contamination distribution verification and contamination cleaning adaptability analysis to generate a surface contamination feature distribution; S4: Performing a multi-dimensional analysis of the surface contamination cleaning strategy for the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer.

[0013] Specifically, in step S1 of the embodiment provided by the present invention, the ultraviolet (UV) module captures organic pollutants by performing molecular electron transition detection on the wafer surface. The high energy of ultraviolet rays can stimulate electron transitions within the molecules, causing them to transition from the ground state to the excited state, and is particularly sensitive to the absorption characteristics of organic pollutant molecules.

[0014] More specifically, ultraviolet light is irradiated onto the wafer surface. After ultraviolet light irradiation, the organic molecules will absorb ultraviolet energy and undergo transitions, releasing specific fluorescent signals or reflected light. The ultraviolet module obtains information about organic pollutants by monitoring these fluorescent or reflected signals.

[0015] More specifically, ultraviolet light can excite organic pollutant molecules, can efficiently identify organic contamination on the wafer surface, can distinguish different types of organic pollutants, provide qualitative and quantitative data, and the detection process is fast, making it suitable for real-time monitoring on the production line.

[0016] More specifically, the terahertz (THz) module uses molecular vibration spectroscopy to identify nano-metal particles on the wafer surface. The frequency of electromagnetic waves in the THz band is between microwaves and infrared rays, and has good penetrability and material sensitivity. It can detect tiny metal particles on the wafer surface. The THz module emits THz waves toward the wafer surface. The THz waves interact with the nano-metal particles on the wafer surface, generating a resonance effect and reflecting back to the sensor. By analyzing the reflected THz signal, the distribution and properties of the metal particles on the wafer surface are obtained.

[0017] More specifically, terahertz waves can accurately identify tiny metal particles, especially nanometer-level particles. Terahertz technology can detect surface contaminants without contact, avoiding damage to the wafer surface, and can effectively distinguish different metal particles and their distribution on the wafer surface.

[0018] More specifically, the microwave module detects surface insulation layer contamination by identifying changes in the dielectric constant of the insulation layer on the wafer surface. The high penetration of microwaves enables it to penetrate deep into the insulation layer of the wafer and detect the physical and chemical state of the layer. The microwave module transmits microwave signals to the surface of the wafer. The microwave signals interact with the material of the insulation layer on the surface of the wafer and reflect back signals based on the dielectric constant of the material (such as changes caused by contaminants). By analyzing the reflected microwave signals, information about the contamination of the insulation layer on the surface of the wafer is obtained.

[0019] More specifically, it can detect tiny changes in the insulation layer and identify surface insulation contamination. Microwave signals have strong penetrating power and can detect deeper contamination layers. Microwave detection is less sensitive to changes in the external environment and is suitable for stable real-time monitoring.

[0020] It is understandable that through the combined use of these three modules, comprehensive detection of various types of contamination on the wafer surface can be carried out. Ultraviolet detection focuses on organic pollutants, terahertz detection focuses on nano-metal particles, and microwave detection focuses on insulation layer contamination. These detection methods can provide detailed and highly sensitive pollutant data, help generate a more accurate pollution feature space, and provide a basis for the formulation of subsequent pollution cleaning strategies. This multi-source data fusion detection method greatly improves the comprehensiveness and accuracy of detection, and also provides a scientific basis for the cleaning and optimization of wafer surface contamination.

[0021] Specifically, in step S2 of the embodiment provided by the present invention, the multi-source data collected from the ultraviolet, terahertz and microwave detection modules are complete and can be used for subsequent analysis. These data include: ultraviolet detection data: fluorescence intensity, wavelength information, etc. of organic pollutants; terahertz detection data: size, distribution density and spatial position of metal particles on the wafer surface; microwave detection data: changes in the dielectric constant of the insulating layer, density and spatial distribution of pollutants. These data are usually collected based on different sensing principles and physical measurements, so their forms may be different. For example, ultraviolet detection data is spectral data, and terahertz and microwave data are physical quantities (such as reflection intensity, penetration depth, etc.).

[0022] More specifically, since the detection data of each module are in different forms and dimensions, the data first needs to be standardized and normalized. The data should be standardized with a mean of 0 and a variance of 1 to eliminate the influence of the dimensions of various types of data. Data from different sources should be scaled according to a unified scale to ensure that the data are compared at the same level. This process ensures the comparability between different data and provides a basis for subsequent data fusion and information mapping.

[0023] More specifically, internal interactive information fusion is performed to map information from different data sources into a unified surface pollution feature space, and representative features are extracted for each data source (ultraviolet, terahertz, microwave), such as: organic matter concentration and pollutant type from ultraviolet data; particle size and particle density from terahertz data; and insulating pollutant concentration and thickness from microwave data.

[0024] More specifically, the extracted features are weighted averaged or multi-source data are fused into a low-dimensional feature space through techniques such as principal component analysis (PCA). Through these methods, redundant information between different data sources can be eliminated, and the most representative features can be extracted. Coordinate system mapping or finite element meshing methods are used to map the processed features to the spatial position on the wafer surface. Each contamination feature has a corresponding coordinate point in space, which describes the distribution of contaminants on the wafer surface.

[0025] More specifically, to ensure the accuracy and reliability of the mapping results, the generated pollution feature space needs to be verified. By comparing the detection results of different modules, it is necessary to check whether there are false detections (incorrectly marking non-contaminated areas as contaminated areas) or missed detections (failure to detect real contaminated areas). For example, ultraviolet detection may not be able to detect all metal particle contamination. Terahertz and microwave modules can supplement this information. By comparing the spatial positions of the detection data of different modules, it is checked whether the data of each module are spatially consistent to ensure that the spatial mapping of pollution features is accurate.

[0026] More specifically, after interactive verification, errors or inaccurate features may be found in some areas. In this case, rasterization technology is used to subdivide the surface pollution feature space into multiple small units, and each small unit is corrected individually. A threshold mechanism or optimization algorithm (such as cluster analysis or machine learning algorithm) is used to adjust the pollution feature space and eliminate inconsistencies.

[0027] More specifically, through the calculation and correction of the aforementioned process, a wafer surface contamination feature space containing all contamination information is finally generated, including the contamination characteristics of different areas on the wafer surface, describing the spatial distribution of organic contamination, metal particle contamination and insulation layer contamination. Due to multiple verifications and corrections, the accuracy of the feature space has been greatly improved, which can provide an accurate basis for subsequent contamination cleaning decisions.

[0028] It can be understood that through the above steps, a high-precision wafer surface contamination feature space can be generated, which comprehensively considers multi-source data of ultraviolet, terahertz and microwaves to fully reflect the contamination status of the wafer surface. Through multiple rounds of verification and correction, the high accuracy of the contamination feature space is ensured, and false detection and missed detection are reduced. The feature space can be dynamically updated with the addition of new detection data, adapting to the ever-changing contamination situation, and mapping the contamination information to spatial coordinates, providing intuitive data support for the subsequent design of contamination cleaning solutions.

[0029] Specifically, in step S3 of the embodiment provided by the present invention, according to the actual process requirements, an appropriate form of environmental disturbance is selected, such as temperature change (heating or cooling), humidity adjustment (drying or humidification), airflow / pressure disturbance (simulating wind speed change or vacuum change), electromagnetic interference (low-intensity microwave or terahertz stimulation), etc. More specifically, a controllable environmental chamber or experimental platform is used to apply preset disturbances to the wafer surface, and sensors are used to accurately record parameters during the disturbance process (such as temperature curve, humidity level, electric field strength, etc.), and the amplitude, duration and application method of the disturbance are controlled to maintain repeatability.

[0030] More specifically, during the disturbance process, three types of detection modules, namely ultraviolet, terahertz, and microwave, are used simultaneously to monitor the wafer surface in real time to obtain multi-source data under disturbance conditions: the ultraviolet module collects the spectral change response of the pollutants; the terahertz module monitors the physical position and response behavior of the metal particles; and the microwave module analyzes the slight changes in dielectric properties under disturbance. The data is collected in the form of a time series to capture the dynamic behavior and response characteristics of the pollutants under disturbance.

[0031] More specifically, feature extraction is performed on the collected multi-source disturbance response data (such as the contaminant migration rate, the response sensitivity of a specific area, etc.), the spatial movement trend and de-attachment tendency of the contaminants under disturbance are analyzed, the detection results under disturbance conditions are mapped back to the existing wafer contamination feature space, the original feature points are verified (to verify whether there are mislabeling or errors), and the stability, mobility or removability of the contamination distribution are judged.

[0032] More specifically, through disturbance-enhanced response, previously imperceptible or hidden contamination areas (such as weakly adhered particles and interface-type contamination) can be identified, and it can be determined whether the contaminants are easy to desorb under disturbance. The response trends of different contaminants to cleaning methods (such as thermal cleaning, plasma cleaning, etc.) are evaluated, and the contaminated areas are divided into priority cleaning areas based on the response characteristics; and zoned cleaning process parameters are formulated for different types of contamination.

[0033] More specifically, the detection data before and after the disturbance are integrated and combined with the cleaning adaptability analysis results to output a "pollution characteristic distribution map" with more dynamic information. This map includes: the spatial distribution of pollutants, the response type and desorption difficulty of each contaminated area, and the adaptability score or suggestions for different cleaning strategies. The map visualizes the contamination status of the wafer surface and provides a scientific basis for subsequent precise cleaning and process control.

[0034] It is understandable that the dynamic response under the action of pollution identification enhancement disturbance improves the contrast and resolution of pollution detection, especially the identification of weak pollution and hidden pollution is significantly improved, the accuracy of pollution map is improved, the multi-source disturbance data mapping corrects the error points in the original pollution space, and builds a more realistic and detailed pollution distribution model. The precise matching of cleaning strategies can evaluate the desorption response and residual trend of various pollutants, which is helpful to select the optimal cleaning method and parameters. Decision support is enhanced to build a pollution-response-cleaning three-dimensional coupling model, which provides strong support for pollution control and quality management in the production process. The core value of this process lies in: transforming static detection into a dynamic verification mechanism, through the closed-loop system of "disturbance-response-mapping-evaluation", to achieve comprehensive identification, adaptability evaluation and visual modeling of wafer surface contamination at the micro scale, laying a solid foundation for subsequent intelligent cleaning control.

[0035] Specifically, in step S4 of the embodiment provided by the present invention, based on the pollution characteristic distribution map after disturbance and combined with multi-source detection data (ultraviolet, terahertz, microwave, etc.), the pollution characteristic space of the wafer surface is updated. The characteristic information of each contaminated area includes the type of pollutant, the number of pollutants, the pollution distribution density, the adhesion of the pollutant to the surface, the difficulty of removing the pollutant, etc. The contaminated area is divided into multiple categories: such as high-adhesion pollution area, easy-to-remove pollution area, particulate pollution area, thick film pollution area, etc. Each type of pollutant has different properties, distribution characteristics and reaction modes, ensuring that the subsequent cleaning strategy is targeted.

[0036] More specifically, based on the type and adhesion of the pollutants, the difficulty of removing the contaminated area is assessed. Specifically, for easily desorbed pollutants (such as light organic matter and soluble pollutants), the minimum energy required for cleaning (such as temperature, humidity, and plasma cleaning intensity) is evaluated; for stubborn pollutants (such as metal particles and difficult-to-remove insulating pollution), the high-intensity cleaning methods required (such as laser cleaning, wet chemical cleaning, etc.) are evaluated.

[0037] More specifically, the adaptability of different cleaning technologies is evaluated based on the characteristics of different contaminated areas. The technical means include: thermal cleaning: suitable for organic contamination, plasma cleaning: suitable for removing low-adhesion particles or thin film contamination, wet cleaning: for soluble contaminants, laser cleaning: used to remove stubborn metal particles and other difficult-to-clean contaminants, airflow cleaning: suitable for removing particulate contaminants. The evaluation content includes the decontamination effect of each technology, the risk of wafer damage during cleaning, and the cost of technology implementation.

[0038] More specifically, based on the results of the pollution feature space and cleaning adaptability analysis, a multi-stage cleaning plan is designed: pre-cleaning stage: use airflow or low-intensity cleaning technology to remove most easily desorbed pollutants; main cleaning stage: select appropriate cleaning technology (such as plasma cleaning, thermal cleaning, laser cleaning, etc.) according to the type and location of the pollutants for deep cleaning; post-processing stage: perform final inspection and fine-tuning on the wafer surface after cleaning to ensure that there is no residual contamination on the surface, and detailed cleaning can be performed through light cleaning.

[0039] More specifically, based on the pollution characteristics and the adaptability assessment results of the cleaning technology, the cleaning parameters of each stage are optimized, including: cleaning time, temperature, humidity, plasma power, laser intensity, etc., and the selection of cleaning media (such as solution, gas type, etc.).

[0040] More specifically, simulation software is used to conduct virtual testing of the designed cleaning scheme to simulate the pollution removal effect and wafer surface damage risk during the cleaning process. During the simulation process, factors such as the force on the wafer surface, temperature, and the interaction between the cleaning medium and contaminants are considered to further verify the effectiveness of the cleaning scheme. In the actual production environment, some wafer samples are selected for cleaning tests, and the cleaned wafers are subjected to contamination detection and surface quality assessment to check whether the contamination has been effectively removed and to confirm whether there is any surface damage caused by the cleaning process.

[0041] More specifically, for problems found in simulation and verification (such as incomplete removal of certain contamination and unstable cleaning effects), the cleaning plan is fine-tuned, cleaning parameters and process sequence are adjusted, or new cleaning technologies are added to meet the removal requirements of specific contaminated areas. After sufficient verification and optimization, the final wafer surface contamination cleaning plan is formulated to ensure its efficiency, stability and low damage.

[0042] It is understandable that the accuracy of pollution identification and classification is based on precise pollution feature space analysis, which can accurately identify the type of pollution area (such as highly attached pollutants, particulate pollutants, etc.) and provide cleaning priority division; through adaptability analysis and multi-stage cleaning design, the best cleaning method is selected for different types of pollutants to ensure the maximum pollution removal rate without damaging the wafer surface; according to different pollution distributions and cleaning adaptability, flexible cleaning solutions are designed to adapt to different production needs and pollution environments, and parameters can be adjusted to deal with uncertain factors; through simulation and verification experiments, the feasibility and efficiency of the cleaning solution are ensured, problems encountered in the test are reduced, and unnecessary waste of resources is avoided.

[0043] More specifically, during the cleaning process, the risk of damage to the wafer surface is minimized, ensuring the final surface quality and performance to meet the requirements of high-precision semiconductor production. Based on the adaptability assessment of the cleaning technology, the appropriate cleaning method is selectively used, reducing energy consumption and material costs during the cleaning process while improving the efficiency of contaminant removal. This process achieves a comprehensive analysis from contamination feature identification, cleaning adaptability assessment, to cleaning process optimization. The resulting surface contamination cleaning solution can select the most appropriate cleaning technology and process for different contamination types, contamination areas, and their difficulty, effectively improving cleaning results, reducing unnecessary waste and damage, and ensuring that wafer surface quality meets high standards. This provides strong technical support for contamination control and cleaning processes in the semiconductor industry.

[0044] The present invention provides a wafer surface contamination analysis method, which has the following beneficial effects: The present invention uses ultraviolet, terahertz and microwave modules to collect multi-source detection data on the wafer surface, performs information mapping on the multi-source data, constructs a pollution feature space, applies environmental disturbance and collects disturbance data, verifies pollution distribution and analyzes cleaning adaptability, performs multi-dimensional cleaning strategy analysis based on pollution feature distribution, and generates an optimal cleaning solution. This method is beneficial for achieving accurate identification and dynamic evaluation of wafer surface contamination, improving the adaptability and scientific nature of pollution cleaning solutions, optimizing cleaning effects, reducing costs, and enhancing manufacturing yield and stability, solving the problem in the prior art that a single identification method is difficult to detect the pollution distribution status of the wafer surface.

[0045] Preferably, the step of collecting sensing data on the wafer surface by ultraviolet, terahertz, and microwave detection modules to obtain multi-source detection data includes: S11: Performing molecular electron transition detection on the wafer surface through an ultraviolet detection module to obtain sensing data of organic pollutants on the wafer surface; S12: Performing molecular vibration spectrum detection on the wafer surface through a terahertz detection module to obtain nano-metal particle sensing data on the wafer surface; S13: Identifying a change in the dielectric constant of the wafer surface through a microwave detection module to obtain insulating layer contamination sensing data on the wafer surface; S14: Integrate the organic pollutant sensing data collected by the ultraviolet detection module, the nano-metal particle sensing data collected by the terahertz detection module, and the insulation layer pollution sensing data collected by the microwave detection module to generate multi-source detection data.

[0046] Specifically, an ultraviolet light source is used to irradiate the wafer surface to stimulate the electronic transition of the surface molecules. Depending on the different pollutants, the electronic transition behavior varies. Changes in ultraviolet absorption or scattering can reflect whether there are organic pollutants on the wafer surface. The ultraviolet module detects the reflected or transmitted spectrum of the wafer surface after excitation by the ultraviolet light source to identify the organic pollutants that may exist on the surface. By analyzing the specific wavelength information in the reflected spectrum, the type and quantity of organic matter are identified, and the relevant sensing data is recorded. The acquired spectral data of organic pollutants is converted into digital signals, and pre-processed such as noise removal and smoothing are performed to improve data quality.

[0047] More specifically, terahertz radiation is used to scan the wafer surface. Terahertz radiation can produce plasmon resonance reactions with the surface of metal or nanoparticles. This resonance phenomenon can significantly enhance the interaction between metal particles and radiation. The terahertz module detects the interaction between metal particles on the wafer surface and terahertz waves, records information such as the size, distribution and concentration of the metal particles, and determines the presence of nano-metal particles and their distribution on the wafer surface by identifying changes in the resonance frequency. The data obtained from the terahertz detection module is denoised and normalized for subsequent integration and analysis.

[0048] More specifically, the microwave module measures changes in the dielectric constant of the surface material by sending microwave signals to the wafer surface. Insulation layer contaminants (such as grease, dust, etc.) will cause changes in the dielectric constant of the wafer surface. When the microwave signal interacts with the insulation layer material, the change in the dielectric constant reflects the type and thickness of the contaminant. By monitoring the reflection changes of the microwave signal, data on insulation layer contamination is obtained. After denoising and standardization of the microwave detection data, the characteristic data of insulation layer contamination is extracted and stored.

[0049] More specifically, the organic pollutant, nano-metal particle and insulation layer pollution data collected by the ultraviolet, terahertz and microwave modules respectively are integrated into a unified data set through a data fusion algorithm. Data fusion techniques (such as weighted average method and principal component analysis) are used to integrate data from different sources to identify the type and distribution of pollution and its impact on wafer performance. The integrated data can be further analyzed through machine learning, pattern recognition and other methods to extract key information such as the distribution characteristics, pollution concentration and location of pollution, providing data support for subsequent cleaning plans. By classifying and analyzing the integrated multi-source data, a complete data set of wafer surface pollution characteristics is generated, forming a full picture of the pollutants.

[0050] It is understandable that the pollution identification accuracy is achieved through the mutual complementation of three detection methods: ultraviolet light, terahertz, and microwave, to achieve multi-dimensional and high-precision identification of wafer surface pollution, and can accurately identify organic pollutants, nano-metal particles, and insulating layer pollutants. The multi-source data integration method can efficiently integrate data from different detection modules to obtain comprehensive and reliable pollution characteristics, providing a scientific basis for subsequent cleaning. The detection solution realizes real-time data collection and processing, and can monitor the pollution situation of the wafer surface in real time during the production process, quickly identify potential problems and issue early warnings.

[0051] More specifically, the integrated contamination data can provide a basis for formulating accurate surface contamination cleaning plans, ensuring the targetedness and effectiveness of the cleaning process. Combined with ultraviolet, terahertz and microwave detection modules, it can simultaneously identify multiple types of contaminants (such as organic matter, metal particles, insulation layer contamination, etc.), greatly improving the comprehensiveness of detection. This multi-source detection technology is not only suitable for the semiconductor industry, but can also be widely used in other precision manufacturing fields, such as surface quality inspection of optical devices and electronic components. Through the multi-source data collection and integration of ultraviolet, terahertz and microwave detection modules, this method can achieve accurate and multi-dimensional analysis of wafer surface contamination, provide an efficient and comprehensive pollution monitoring method, and help to achieve more accurate pollution management and cleaning process optimization in the semiconductor manufacturing process.

[0052] Preferably, the step of performing internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer includes: S21: Acquire structural information of the wafer, and construct a characteristic space coordinate system of the wafer surface based on the structural information using a finite element meshing method; S22: performing data mapping on the feature space coordinate system according to the multi-source detection data, so as to map the feedback content of the multi-source detection data to a corresponding position on the feature space coordinate system in the form of feature coding; S23: assigning module labels to the feature codes according to the detection modules corresponding to the feature codes, and interactively verifying the feature codes of the various module labels to detect false detections and missed detections in the feature space coordinate system; S24: Based on the interactive verification result, a summary description of the pollutant distribution status under multiple scale ranges is performed on the feature coding on the feature space coordinate system, and corresponding distribution feature information is generated and assigned to the feature space coordinate system to generate the surface contamination feature space of the wafer.

[0053] Specifically, the geometric structure data of the wafer is collected to understand the morphology, microstructure, material properties, etc. of the wafer surface. Based on the acquired wafer surface structure data, the finite element method is used to grid the wafer surface and construct a three-dimensional feature space coordinate system of the wafer surface. In the feature space coordinate system, each grid point represents the position of a tiny area on the wafer surface, which can correspond to different contamination characteristics.

[0054] More specifically, the system uses inspection data from ultraviolet, terahertz, and microwave modules as input and maps it to corresponding grid locations based on a feature space coordinate system. Each piece of inspection data is correlated with the wafer surface geometry and contaminant type, and then fed back to a specific location in the feature space coordinate system to form a signature code. This code is then generated for each type of contaminant (e.g., organic contaminants, metal particles, and insulating layer contaminants). Each code represents the inspection result for a single contaminant. Through feature mapping, these signature codes are mapped to corresponding locations in the feature space coordinate system, accurately reflecting the contaminant distribution on the wafer surface.

[0055] More specifically, according to the detection module corresponding to the feature code (such as ultraviolet module, terahertz module, microwave module), a corresponding module label is assigned to each feature code. For example, the ultraviolet module can be labeled "UV" and the terahertz module can be labeled "THz". The feature codes marked by different detection modules are cross-verified to check whether there are false detections or missed detections at the same location. The data in the feature space is verified through cross-validation to ensure that the feature code of each polluted area is accurate. If the pollution information of a certain area is not detected, the detection method needs to be further optimized.

[0056] More specifically, based on the interactive verification results, the areas of false detection and missed detection are processed. If missed detection or false detection occurs in some areas, multiple data feedback, increased sensitivity or replacement of detection methods are used for correction. Based on the data after interactive verification, the distribution of pollutants is summarized and described within multiple scales. These descriptions may include information such as the type, concentration, and distribution range of pollutants. The description process may adopt a hierarchical approach to ensure that the pollution characteristics of the wafer surface can be fully presented at different scales. The summarized pollutant distribution information is embedded in the feature space coordinate system to form a complete wafer surface pollution feature space. This feature space will reflect the pollution distribution on the wafer surface and provide a basis for subsequent cleaning and processing.

[0057] It can be understood that the pollution feature space accuracy is improved through high-precision grid division and data mapping. The pollution feature space generated can accurately reflect the pollution situation of each tiny area on the wafer surface. The interactive verification is used to avoid false detection and missed detection, and the accuracy and credibility of the feature space are improved. By assigning module labels, the source of pollution can be clearly distinguished. The multi-scale analysis method based on multiple scales can accurately describe the distribution of pollutants at different levels, provide richer information, and help accurately judge the distribution characteristics of contaminated areas.

[0058] More specifically, through precise contamination feature space, scientific support can be provided for surface cleaning strategies, ensuring targeted cleaning of contamination sources in actual operations, reducing waste and operational errors. The modular data processing method can flexibly add or replace detection modules to improve the adaptability and scalability of the system. The data mapping and interactive verification process is efficient, and potential problems can be quickly identified and corrected. The contamination feature space is updated in real time to ensure the real-time and accuracy of the detection process. Through this series of precise steps and technical means, the wafer surface contamination feature space is finally generated, which can not only accurately reflect the type, location and concentration of surface contamination, but also provide an important basis for subsequent pollution control and cleaning strategies, thereby improving the effectiveness and efficiency of pollution monitoring in the semiconductor manufacturing process.

[0059] Preferably, the step of applying environmental disturbance to the wafer surface and collecting multi-source detection data during the environmental disturbance to substitute into the surface contamination feature space for contamination distribution verification and contamination cleaning adaptability analysis to generate a surface contamination feature distribution includes: S31: constructing a multidimensional environmental disturbance space constructed by thermodynamic dimensions, electrochemical dimensions, and mechanical vibration dimensions, and simulating the environmental disturbance effect on the surface contamination feature space based on the multidimensional environmental disturbance space, so as to apply environmental disturbance to the wafer surface according to the simulation results; S32: collecting multi-source detection data on the wafer surface under environmental disturbance, and substituting the multi-source detection data under environmental disturbance into the surface contamination feature space to perform feature analysis in time domain, frequency domain, and spatial evolution; S33: performing multi-dimensional feature analysis of the thermal stability, mechanical adhesion, chemical affinity, and electrical response of the contamination distribution on the surface contamination feature space according to the result of the feature analysis, so as to obtain multi-dimensional reference properties of the surface contamination feature space; S34: verifying the pollution distribution of the surface pollution feature space according to the multi-dimensional reference property to correct the information fed back by the surface pollution feature space; S35: performing adaptability analysis of the pollution cleaning strategy on the surface pollution feature space after information correction according to the multi-dimensional reference property, so as to obtain adaptability characteristics of the surface pollution feature space corresponding to various pollution cleaning strategies; S36: performing spatial unitization and clean adaptability expression on the surface contamination feature space according to the adaptability characteristics to generate a surface contamination feature distribution.

[0060] Specifically, a multidimensional environmental disturbance space is constructed through multiple dimensions such as thermodynamics, electrochemistry, and mechanical vibration. Each dimension represents an environmental disturbance factor, such as temperature fluctuations, electric field changes, or vibration frequencies. Thermodynamic dimension: the impact of temperature changes on the contamination characteristics of the wafer surface; electrochemical dimension: the impact of electric fields, currents, and other factors on the electrical response of the wafer surface; mechanical vibration dimension: the impact of vibration intensity, frequency, and other factors on the contamination of the wafer surface.

[0061] More specifically, based on the multi-dimensional environmental disturbance space constructed above, environmental disturbance simulation is performed. Through simulation analysis, the contamination changes on the wafer surface under different environmental disturbance conditions are predicted. Through numerical calculation or simulation technology, the impact of environmental disturbance on the surface contamination characteristic space is simulated, so as to apply appropriate environmental disturbance for subsequent experiments.

[0062] More specifically, multi-source detection data is collected on the wafer surface after environmental disturbances are applied. Multi-source detection can include data from thermal, electrical, and vibration sensors. The collected data includes, but is not limited to, time domain, frequency domain, and spatial distribution, to fully reflect the evolution of contamination. The collected multi-source detection data is then inserted into the surface contamination feature space. Through feature mapping and encoding, the data under environmental disturbances is embedded into corresponding positions in the feature space coordinate system. Through this process, the impact of environmental disturbances is converted into contamination feature data that can be quantified and analyzed.

[0063] More specifically, the data substituted into the feature space is subjected to multi-dimensional feature analysis, including time domain, frequency domain and spatial evolution analysis. The time domain analysis focuses on the changes in pollution characteristics over time, such as the diffusion rate of pollutants on the surface, the change in pollutant concentration, etc. The frequency domain analysis studies the response characteristics of pollution characteristics at different frequencies, especially the frequency influence in mechanical vibration and electrochemical dimensions. The spatial evolution analysis analyzes the distribution changes of pollutants in the surface space and explores the spatial expansion and aggregation phenomena of pollutants.

[0064] More specifically, based on the analysis results, the contamination feature space is analyzed in multiple dimensions, including thermal stability, mechanical adhesion, chemical affinity, and electrical response. Thermal stability: analyzes the stability and changes of contaminants at different temperatures; mechanical adhesion: evaluates the mechanical adhesion between contaminants and the wafer surface; chemical affinity: analyzes the chemical reactivity between contaminants and wafer surface materials; and electrical response characteristics: studies the response and impact of contaminants on the electrochemical environment.

[0065] More specifically, the contamination distribution in the contamination signature space is verified based on the multi-dimensional reference properties. By comparing it with actual test data, the accuracy of the contamination distribution is checked, potential errors or deviations are identified, and the information in the contamination signature space is corrected to ensure that the simulation results are consistent with the actual situation. The corrected contamination signature space will more accurately reflect the distribution and changes of contamination on the wafer surface.

[0066] More specifically, based on the corrected pollution feature space, an adaptability analysis is conducted on different pollution cleaning strategies. The adaptability of various cleaning methods (such as chemical cleaning, physical cleaning, etc.) to the pollution distribution is analyzed, and adaptability characteristics are generated for each cleaning strategy. The effects of different strategies under specific pollution characteristics are evaluated, and the pollution feature space is spatially decomposed into units. Each unit is analyzed in detail based on the characteristics of the pollution distribution. Through the adaptability analysis results, the adaptability characteristics of the cleaning strategy are mapped to each unit to form a cleaning adaptability expression.

[0067] More specifically, based on the results of adaptability analysis and spatial unitization decomposition, a surface contamination feature distribution is finally generated, which reflects the location, concentration and adaptability of cleaning strategies of different contaminants on the wafer surface.

[0068] It can be understood that the accuracy of multi-dimensional environmental disturbance simulation can accurately predict the distribution and changes of pollutants under different environmental conditions by constructing a multi-dimensional environmental disturbance space and conducting simulations, providing a scientific basis for the subsequent actual environmental disturbance application, and verifying the pollution distribution based on multi-dimensional reference properties. It can identify errors in the pollution distribution and ensure that the final pollution characteristic space is more accurate. Through adaptive analysis, the most suitable cleaning strategy can be formulated according to the characteristics and distribution of different pollutants to improve the efficiency and effectiveness of the pollution cleaning process.

[0069] More specifically, by decomposing the pollution feature space into spatial units, the pollution situation in each area can be accurately analyzed, and a cleaning plan can be tailored for each area to optimize the cleaning process. Through multi-dimensional analysis of time domain, frequency domain and spatial evolution, the change process of pollutants can be fully understood, and the timeliness and accuracy of pollution detection can be further improved. Through cleaning adaptability analysis and combined with the characteristics of pollutants, the optimal cleaning method can be effectively selected to improve the cleaning effect and reduce the occurrence of incorrect cleaning or missed cleaning. Finally, through the above steps and technical means, the surface pollution feature distribution generated can provide efficient support for wafer pollution detection and cleaning strategy optimization, which not only helps to improve the yield of wafer production, but also provides an accurate basis for quality control in semiconductor manufacturing.

[0070] Preferably, the step of verifying the pollution distribution of the surface pollution feature space according to the multi-dimensional reference property to correct the information fed back by the surface pollution feature space includes: S341: performing rasterization processing on the surface contamination feature space to divide the surface contamination feature space into a plurality of verification units; S342: Based on the theoretical property information of various contaminants adhering to the wafer surface, simulate the theoretical properties of the surface contaminants under environmental disturbances for each verification unit, and expand the reasonable fluctuation range of the simulation results based on the theoretical property information to obtain a baseline simulation feature for each verification unit; S343: performing property matching on the benchmark simulation characteristics of each verification unit according to the multi-dimensional reference property to generate deviation parameters of each verification unit; S344: Evaluating the deviation parameters of each verification unit using a preset multi-level threshold, and constructing a rationality feature matrix of the surface contamination feature space based on the evaluation results; S345: Perform feature clustering on the rationality feature matrix, and perform information adjustment on the surface contamination feature space corresponding to the feature clustering location. At the same time, supervise and correct the information adjustment based on the difference in the successive changes of the rationality feature matrix, so as to realize the correction of the information fed back by the surface contamination feature space.

[0071] Specifically, the surface pollution feature space is gridded, that is, divided into multiple small units, to facilitate separate verification and analysis of the pollution characteristics of each area. Each grid unit represents a verification unit in the surface pollution feature space. Each unit can have different pollutant distribution and environmental conditions. According to the complexity and needs of the space, the appropriate grid size is selected to divide the pollution feature space into several small units. These units not only consider the geometric shape of the space, but also the possible distribution characteristics of the pollutants.

[0072] More specifically, a simulation analysis is performed based on the theoretical attachment property information between different contaminants and the wafer surface. These theoretical properties include the thermodynamic stability, chemical affinity, mechanical adhesion, etc. of the contaminants. Under environmental disturbance conditions, the theoretical properties of the contaminants in each verification unit are simulated to display the behavioral characteristics of the contaminants in the unit. Based on the simulation results, the behavioral characteristics of the contaminants are expanded to a reasonable fluctuation range. For example, considering the uncertainty of environmental disturbances, a certain range of errors is allowed to ensure the robustness of the results. The fluctuation range is set to obtain the baseline simulation characteristics of each verification unit, that is, the predicted behavior value of the contaminants in each area.

[0073] More specifically, the benchmark simulation characteristics are compared and matched with multi-dimensional reference properties. The multi-dimensional reference properties may include reference standards for factors such as temperature, electric field, and vibration intensity. Through the matching process, the difference between the performance of each verification unit under actual environmental disturbance conditions and the theoretical prediction is evaluated. Based on the matching results, a deviation parameter is generated for each verification unit. The deviation parameter reflects the difference between the simulation results and the reference properties, helping to further adjust the pollution feature space. The purpose of this step is to accurately locate the unreasonable area or error range in the pollution feature space by quantifying the deviation.

[0074] More specifically, the deviation parameters of each verification unit are evaluated using a preset multi-level threshold. The multi-level threshold is a standard set according to the size of the deviation, the characteristics of the pollutants, and the intensity of the environmental disturbance. The purpose of the evaluation is to determine whether the simulation results of each verification unit are within a reasonable range and whether they conform to the behavior patterns of actual pollutants. Based on the evaluation results, a rationality feature matrix is ​​constructed, which records the rationality evaluation results of each verification unit and reflects the rationality and deviation distribution of the pollution feature space as a whole.

[0075] More specifically, feature clustering is performed on the data in the rationality feature matrix to group similar pollution features together. Through clustering, similar or related areas in the pollution feature space can be identified. The clustering results help to further optimize the structure of the pollution feature space, discover and correct potential error areas, and adjust the information in the pollution feature space based on the results of feature clustering. The goal of the adjustment is to make the pollution feature space more accurately reflect the distribution and behavioral characteristics of actual pollutants. At the same time, by analyzing the differences in the changes of the rationality feature matrix, the adjustment process is supervised and the adjustment strategy is corrected as needed.

[0076] More specifically, the information fed back by the surface contamination feature space is corrected through the above steps to ensure the accuracy and consistency of the contamination distribution. The corrected feature space can more accurately guide the optimization and practical application of the contamination cleaning strategy.

[0077] It can be understood that the surface pollution feature space is divided into multiple verification units through rasterization processing, which facilitates independent verification and adjustment of pollution characteristics in different areas. Simulation based on the theoretical properties of pollutants and expansion within a reasonable fluctuation range can adapt to the uncertainty of environmental disturbances and improve the robustness and accuracy of simulation results. The deviation of each verification unit is evaluated through multi-level thresholds to quantify the error and accurately locate the deviation area in the pollution feature space. Through cluster analysis of the rationality feature matrix, the correlation and similarity in the pollution feature space can be discovered, thereby further optimizing the pollution distribution structure and correcting the unreasonable part.

[0078] More specifically, information adjustment is performed on the pollution feature space to ensure that the feedback information of the pollution feature space is more accurate, and the effectiveness of the adjustment process is guaranteed by supervising the correction process. Finally, the corrected surface pollution feature space is more consistent with the actual pollutant behavior, and can provide more accurate pollution cleaning strategies and detection basis, thereby improving the efficiency and accuracy of the cleaning process. Through the above steps, the pollution characteristics of the wafer surface can be accurately simulated, verified and corrected under multi-dimensional environmental disturbances, so that the pollution distribution is more consistent with the actual situation. This process provides strong support for the optimization of pollution cleaning strategies, thereby improving production efficiency and the reliability of quality control.

[0079] Preferably, the step of performing a multi-dimensional analysis of the surface contamination cleaning strategy for the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer includes: S41: performing feature space mapping processing on the surface pollution feature distribution to divide the surface pollution feature distribution into a plurality of space units with consistent adaptive features; S42: constructing a multi-dimensional analysis framework based on each of the spatial units, which includes cleaning effect, cleaning cost, and interference with other spatial units; S43: With the maximization of the consistency of the cleaning strategies adopted by each spatial unit and the optimal balance of the cleaning effect cost as constraints, the multi-dimensional analysis framework is optimally solved by the quantum annealing algorithm to generate a surface contamination cleaning solution composed of the surface contamination cleaning strategies of each space.

[0080] Specifically, feature space mapping is performed on the contamination characteristic distribution on the wafer surface. First, it is necessary to identify information such as the type, distribution density, and physical and chemical properties of the pollutants, and then convert the contamination characteristic information into a feature space model through spatial mapping technology.

[0081] More specifically, based on the spatial distribution of pollution characteristics, the entire surface is divided into several spatial units with consistent adaptive characteristics. Each unit represents a pollution characteristic area, where each unit may contain different pollutants and their different cleaning requirements. These units can be consistent in dimensions such as pollution degree, pollutant type, and surface structure to facilitate subsequent analysis and cleaning strategy design.

[0082] More specifically, a multi-dimensional analysis framework was constructed based on the aforementioned spatial unit divisions. Key elements of the framework include: cleaning effectiveness, which evaluates the pollutant removal efficiency of each spatial unit after a specific cleaning strategy is implemented; cleaning cost, which calculates the cost of implementing a specific cleaning strategy for each spatial unit, including the time, resources, and equipment consumed; and interference with other spatial units, which analyzes the potential interference of each cleaning strategy with adjacent or other spatial units to ensure the effectiveness of the overall cleaning process.

[0083] More specifically, the cleaning effect and cleaning cost of each spatial unit need to be quantified through experimental data or simulation models. At the same time, the possible mutual influences during the pollutant removal process (such as cross-contamination, equipment limitations, etc.) need to be considered.

[0084] More specifically, in the selection of cleaning strategies for each spatial unit, the consistency of the strategies is required to be maximized, that is, the cleaning methods for the same type of pollutants should be as unified as possible to improve the overall cleaning efficiency. Through the optimization algorithm, the best balance is required between the cleaning effect and cost of each spatial unit, that is, while ensuring efficient cleaning, the cost required for cleaning should be reduced as much as possible.

[0085] More specifically, the quantum annealing algorithm is used to optimize and solve the entire multi-dimensional analysis framework. The quantum annealing algorithm can search for the optimal solution in high-dimensional space, avoiding the dilemma of local optimality that traditional algorithms are prone to. In the quantum annealing algorithm, an energy function (objective function) is constructed to represent the relationship between cleaning effect, cost and interference, and the optimal solution is gradually found through a simulated annealing process. In the process of finding the global optimal solution, the quantum annealing algorithm can quickly and efficiently process a large number of variables and constraints, and is particularly suitable for solving large-scale, multi-dimensional optimization problems.

[0086] More specifically, a final surface contamination cleaning plan is generated based on the solution optimized by the quantum annealing algorithm. This plan specifies the most appropriate cleaning strategy for each spatial unit. The cleaning plan includes the cleaning method, equipment used, cleaning time and cost required, and its impact on other units. The feasibility and effectiveness of the cleaning plan are evaluated to ensure that the cleaning tasks of all spatial units can be successfully completed within the budget and time requirements. The plan is then converted into practical cleaning operation instructions for implementation.

[0087] It can be understood that through feature space mapping, the pollution feature distribution is converted into an operational spatial unit, and the cleaning requirements of each unit are consistent, which facilitates subsequent analysis and strategy selection. An analysis framework with multiple dimensions such as cleaning effect, cost, and interference is constructed to provide a comprehensive and detailed evaluation perspective to ensure the comprehensive performance of the cleaning strategy. Through the global search characteristics of the quantum annealing algorithm, it can handle complex multi-dimensional optimization problems, avoid the local optimal dilemma that traditional algorithms may encounter, and can efficiently find the optimal cleaning solution.

[0088] More specifically, the cleaning plan generated based on the optimization results can reduce costs and avoid cross-pollution impact while ensuring efficient cleaning, providing a scientific and accurate surface contamination cleaning strategy with the optimal balance between cleaning effect and cost, so that the cleaning strategy of each spatial unit achieves a good balance between effect and economy, achieving the best cleaning efficiency and resource utilization. The interference analysis of the cleaning strategy ensures that there will be no unnecessary mutual interference between adjacent units, thereby ensuring the overall smooth progress of the cleaning work and improving the comprehensiveness and accuracy of cleaning.

[0089] More specifically, through the above steps, combined with the powerful optimization capabilities and multi-dimensional analysis framework of the quantum annealing algorithm, an optimal solution for cleaning wafer surface contamination can be efficiently designed. This solution not only maximizes the cleaning effect, but also balances the cleaning cost and interference with other units, ensuring the efficiency, economy and reliability of the entire cleaning process.

[0090] Reference Figure 2As shown, in a second aspect, the present invention provides a wafer surface contamination analysis device, which is used to implement a wafer surface contamination analysis method according to any one of the first aspects, comprising: The data acquisition module is used to collect sensor data from the wafer surface through ultraviolet, terahertz and microwave detection modules to obtain multi-source detection data; An information mapping module, configured to perform internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; A contamination analysis module, configured to apply environmental disturbances to the wafer surface and collect multi-source detection data obtained during the environmental disturbances, and substitute the data into the surface contamination feature space to perform contamination distribution verification and contamination cleaning adaptability analysis, thereby generating a surface contamination feature distribution; The solution analysis module is used to perform a multi-dimensional analysis of the surface contamination cleaning strategy of the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer.

[0091] In this embodiment, for the specific implementation of each module in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0092] In a third aspect, the present invention provides a wafer surface contamination analysis device, comprising a memory and a processor, wherein the memory stores a wafer surface contamination analysis program that can be run on the processor, and when the processor executes the wafer surface contamination analysis program, it implements a wafer surface contamination analysis method described in any one of the first aspects.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A wafer surface contamination analysis method, characterized in that: include: The wafer surface is sensed and data is collected through ultraviolet, terahertz, and microwave detection modules to obtain multi-source detection data. Performing internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; Applying environmental disturbance to the wafer surface and collecting multi-source detection data during the environmental disturbance to substitute the data into the surface contamination feature space for contamination distribution verification and contamination cleaning adaptability analysis to generate a surface contamination feature distribution; A multi-dimensional analysis of the surface contamination cleaning strategy of the wafer is performed according to the surface contamination characteristic distribution to generate a surface contamination cleaning plan for the wafer.

2. The wafer surface contamination analysis method according to claim 1, wherein: The steps of collecting sensor data from the wafer surface using ultraviolet, terahertz, and microwave detection modules to obtain multi-source detection data include: The wafer surface is tested for molecular electron transitions using an ultraviolet detection module to obtain sensing data on organic pollutants on the wafer surface. The terahertz detection module is used to perform molecular vibration spectrum detection on the wafer surface to obtain nano-metal particle sensing data on the wafer surface; The microwave detection module is used to identify the dielectric constant change on the wafer surface to obtain the contamination sensing data of the insulating layer on the wafer surface; The organic pollutant sensing data collected by the ultraviolet detection module, the nano-metal particle sensing data collected by the terahertz detection module, and the insulation layer pollution sensing data collected by the microwave detection module are integrated to generate multi-source detection data.

3. The wafer surface contamination analysis method according to claim 1, wherein: The steps of performing internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer include: Acquire structural information of the wafer, and construct a characteristic space coordinate system of the wafer surface based on the structural information using a finite element meshing method; Performing data mapping on the feature space coordinate system according to the multi-source detection data, so as to map feedback content of the multi-source detection data to corresponding positions on the feature space coordinate system in the form of feature codes; Assign module labels to feature codes according to the detection modules they correspond to, and interactively verify the feature codes of various module labels to detect false detections and missed detections in the feature space coordinate system. Based on the interactive verification results, the feature coding on the feature space coordinate system is used to summarize the distribution of pollutants in multiple scale ranges, and the corresponding distribution feature information is generated and assigned to the feature space coordinate system to generate the surface contamination feature space of the wafer.

4. The wafer surface contamination analysis method according to claim 1, wherein: The steps of applying environmental disturbance to the wafer surface and collecting multi-source detection data during the environmental disturbance to substitute the data into the surface contamination feature space for contamination distribution verification and contamination cleaning adaptability analysis to generate a surface contamination feature distribution include: Constructing a multidimensional environmental disturbance space constructed by thermodynamic dimensions, electrochemical dimensions, and mechanical vibration dimensions, and simulating the environmental disturbance effect on the surface contamination feature space based on the multidimensional environmental disturbance space, so as to apply environmental disturbance to the wafer surface according to the simulation results; Collecting multi-source detection data on the wafer surface under environmental disturbance, and substituting the multi-source detection data under environmental disturbance into the surface contamination feature space to perform feature analysis in time domain, frequency domain, and spatial evolution; Performing multi-dimensional feature analysis of the surface contamination feature space based on the results of the feature analysis, including thermal stability, mechanical adhesion, chemical affinity, and electrical response of the contamination distribution, to obtain multi-dimensional reference properties of the surface contamination feature space; Verifying the pollution distribution of the surface pollution feature space according to the multi-dimensional reference property to correct the information fed back by the surface pollution feature space; Performing an adaptability analysis of the pollution cleaning strategy on the surface pollution feature space after information correction according to the multi-dimensional reference property to obtain adaptability characteristics of the surface pollution feature space corresponding to various pollution cleaning strategies; The surface contamination feature space is spatially unitized and expressed in a clean adaptability manner according to the adaptability characteristics to generate a surface contamination feature distribution.

5. The wafer surface contamination analysis method according to claim 4, wherein: The step of verifying the pollution distribution of the surface pollution feature space according to the multi-dimensional reference property to correct the information fed back by the surface pollution feature space includes: Performing rasterization processing on the surface contamination feature space to divide the surface contamination feature space into a plurality of verification units; Based on the theoretical properties of various contaminants attached to the wafer surface, the theoretical properties of surface contaminants under environmental disturbances are simulated for each verification unit. The simulation results are then expanded within a reasonable fluctuation range based on the theoretical properties to obtain the baseline simulation characteristics of each verification unit. matching the properties of the benchmark simulation characteristics of each verification unit according to the multi-dimensional reference properties to generate deviation parameters for each verification unit; Evaluating the deviation parameters of each verification unit through a preset multi-level threshold, and constructing a rationality feature matrix of the surface contamination feature space based on the evaluation results; The rationality feature matrix is ​​clustered, and information adjustment is performed on the surface contamination feature space corresponding to the feature clustering location. At the same time, the information adjustment is supervised and corrected according to the difference in the successive changes of the rationality feature matrix to achieve the correction of the information fed back by the surface contamination feature space.

6. The wafer surface contamination analysis method according to claim 1, wherein: The steps of performing a multi-dimensional analysis of the surface contamination cleaning strategy of the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer include: Performing feature space mapping processing on the surface contamination feature distribution to divide the surface contamination feature distribution into a plurality of space units with consistent adaptive features; Building a multi-dimensional analysis framework based on each of the spatial units, consisting of cleaning effect, cleaning cost, and interference with other spatial units; With the maximization of the consistency of the cleaning strategies adopted by each spatial unit and the optimal balance of the cleaning effect cost as constraints, the multidimensional analysis framework is optimally solved by the quantum annealing algorithm to generate a surface contamination cleaning plan consisting of the surface contamination cleaning strategies of each space.

7. A wafer surface contamination analysis device, characterized in that: A wafer surface contamination analysis method for implementing any one of claims 1 to 6, comprising: The data acquisition module is used to collect sensor data from the wafer surface through ultraviolet, terahertz and microwave detection modules to obtain multi-source detection data; An information mapping module, configured to perform internal interactive information mapping on the multi-source detection data to generate a surface contamination feature space of the wafer; A contamination analysis module, configured to apply environmental disturbances to the wafer surface and collect multi-source detection data obtained during the environmental disturbances, and substitute the data into the surface contamination feature space to perform contamination distribution verification and contamination cleaning adaptability analysis, thereby generating a surface contamination feature distribution; The solution analysis module is used to perform a multi-dimensional analysis of the surface contamination cleaning strategy of the wafer according to the surface contamination characteristic distribution to generate a surface contamination cleaning solution for the wafer.

8. A wafer surface contamination analysis device, comprising a memory and a processor, wherein the memory stores a wafer surface contamination analysis program that can be run on the processor, characterized in that: When the processor executes the wafer surface contamination analysis program, a wafer surface contamination analysis method according to any one of claims 1 to 6 is implemented.

Images