Intelligent hydrocarbon cleaning machine and using method thereof

Abstract

The invention discloses an intelligent hydrocarbon cleaning machine and a use method thereof, particularly relates to the technical field of silicon wafer cleaning, and is used for solving the problem that the subsequent photoetching process quality is influenced due to the fact that the surface interface characteristics of a silicon wafer are changed due to solvent residues in an existing hydrocarbon cleaning process. The method comprises the following steps: acquiring residual concentration and contact angle data of a solvent on the surface of a silicon wafer through an online monitoring system, identifying a volatile and polarity characteristic collaborative abnormal state of residual components of the solvent, calling process parameters of a previous process and historical pollution data to carry out correlation analysis when abnormity is found, identifying a cross-process pollution transmission chain, and carrying out analysis. The competitive adsorption tendency of solvent residues and photoresist components is evaluated, the silicon wafer surface photoetching compatibility risk level is determined according to the evaluation result, and finally the rinsing flow and the drying temperature parameter are dynamically adjusted based on the risk level. The method realizes accurate prevention and control of molecular-level interface pollution, effectively eliminates secondary pollution caused by solvent residue, and ensures compatibility of surface characteristics of the silicon wafer and a photoetching process.

Description

Technical Field

[0001] This invention relates to the field of silicon wafer cleaning technology, and more specifically, to an intelligent hydrocarbon cleaning machine and its usage method. Background Technology

[0002] Cleaning is a crucial step in the manufacturing of monocrystalline silicon wafers. Its purpose is to remove contaminants such as particles, organic residues, and metal ions from the wafer surface to meet the stringent surface cleanliness requirements of subsequent advanced processes. Hydrocarbon solvent cleaning technology, due to its environmental friendliness, low toxicity, and excellent solubility for various contaminants, has become an important alternative to traditional strong acid and alkali cleaning methods. Existing hydrocarbon cleaning processes typically include basic steps such as cleaning, rinsing, and drying. Its core technology lies in achieving effective contaminant removal through a combination of physical and chemical processes.

[0003] However, existing hydrocarbon cleaning methods still face contradictions in application: after the cleaning medium completes its main decontamination function, it may introduce imperceptible secondary interface contamination. Specifically, the molecular-level residues of hydrocarbon solvents and their trace additives may still be physically adsorbed on the silicon wafer surface in the form of an invisible thin film even after macroscopic drying. This residual film will irreversibly change the interface characteristics of the silicon wafer surface, especially its free energy and wettability, thereby causing interface compatibility problems such as uneven photoresist adhesion and poor local wetting in subsequent photolithography processes, ultimately manifesting as pattern defects and directly affecting device yield. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent hydrocarbon cleaning machine and its usage method to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method of using an intelligent hydrocarbon cleaning machine includes: S1. Obtain data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; S2. Based on surface solvent residual concentration and surface contact angle data, identify the abnormal state of synergistic effect between volatility and polarity characteristics in the residual solvent components; S3. When an abnormal state of synergistic effect is identified, retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process, and perform correlation analysis with the surface solvent residual concentration to identify the existence of cross-process contamination transmission chain. S4. When a cross-process contamination transmission chain is identified, assess the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface. S5. Based on the competitive adsorption tendency, assess the compatibility risk level of the silicon wafer surface for the photolithography process; S6. Adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaning machine dynamically according to the compatibility risk level.

[0006] Furthermore, data on the residual solvent concentration and surface contact angle of the silicon wafer after hydrocarbon cleaning are obtained through an online monitoring system, including: After the hydrocarbon cleaning process is completed, Fourier transform infrared spectroscopy analysis is performed on the silicon wafer surface to obtain the residual concentration of surface solvent. Simultaneously, static contact angle measurements were performed on the silicon wafer surface by depositing ultrapure water droplets and recording their contours to obtain surface contact angle data. The Fourier transform infrared spectroscopy analysis and static contact angle measurement are completed within the same time window, ensuring the synchronization and consistency of data acquisition.

[0007] Furthermore, based on surface solvent residual concentration and surface contact angle data, the synergistic effect anomaly of volatility and polarity characteristics in the residual solvent components is identified, including: Volatility characteristic parameters were determined based on the residual concentration of surface solvent; Polarity characteristic parameters are determined based on surface contact angle data; Analyze the correlation between volatility characteristic parameters and polarity characteristic parameters; When the correlation is lower than the preset correlation threshold, it is determined to be an abnormal state of synergy effect.

[0008] Furthermore, determining the volatility characteristic parameters based on the residual concentration of surface solvents includes: querying a pre-stored solvent property database, matching the residual concentration of surface solvents with the volatility parameters of standard solvents in the database, thereby determining the volatility characteristic parameters; Determining polarity characteristic parameters based on surface contact angle data includes: processing the surface contact angle data using a surface free energy calculation model, and extracting polarity characteristic parameters from the polarity components of the calculation results.

[0009] Furthermore, when an abnormal state of synergistic effect is identified, the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process are retrieved and correlated with the surface solvent residual concentration to identify the existence of cross-process contamination transmission chains, including: When an abnormal state of synergy effect is identified, the process parameters and historical contamination data of the corresponding silicon wafer in the previous manufacturing process are retrieved from the manufacturing execution system. A time-series correlation analysis was performed between the residual concentration of surface solvent and the process parameters of the preceding manufacturing process. The surface solvent residual concentration was compared with historical pollution data to ensure consistency of composition. When both time-series correlation analysis and component consistency comparison show significant correlation, it is determined that a cross-process contamination transmission chain exists.

[0010] Furthermore, when cross-process contamination transmission chains are identified, the competitive adsorption tendencies of solvent residues and key photoresist components on the silicon wafer surface are evaluated, including: Key active components in solvent residues were identified based on cross-process contamination transmission chains. Obtain standard adsorption free energy data of the main components of photoresist on the silicon wafer surface; The competitive adsorption tendency is assessed by comparing the relative adsorption free energies of the key active components and the main components of the photoresist.

[0011] Furthermore, by comparing the relative adsorption free energies of the key active components and the main components of the photoresist, the competitive adsorption tendency is assessed by: identifying the key active components based on the cross-process contamination transport chain; obtaining standard adsorption free energy data of the key active components and the main components of the photoresist; and assessing the competitive adsorption tendency by directly comparing the adsorption free energy values.

[0012] Furthermore, based on competitive adsorption tendencies, the compatibility risk level of the silicon wafer surface with the photolithography process is assessed, including: The basic level is determined based on the strength of the competitive adsorption tendency; The corresponding basic level is corrected by combining the persistence impact coefficient of pollutants in the cross-process pollution transmission chain; The corrected base level is output as the compatibility risk level.

[0013] Furthermore, based on the compatibility risk level, the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaner are dynamically adjusted, including: Based on the compatibility risk level, match the corresponding rinsing flow rate baseline value and drying temperature baseline value; A rinse flow rate parameter adjustment instruction is generated based on the rinse flow rate baseline value; Generate drying temperature parameter adjustment instructions based on the drying temperature reference value; The rinsing flow rate parameter adjustment command and the drying temperature parameter adjustment command are executed to dynamically control the hydrocarbon cleaning machine.

[0014] On the other hand, the present invention provides an intelligent hydrocarbon cleaning machine, comprising the following modules: The data acquisition module is used to acquire data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; Anomaly detection module is used to identify abnormal states of synergistic effects between volatile and polar characteristics in residual solvent components based on surface solvent residual concentration and surface contact angle data. An identification module is used to retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process when an abnormal state of synergistic effect is identified, and to perform correlation analysis with the surface solvent residual concentration to identify the existence of a cross-process contamination transmission chain. The tendency assessment module is used to evaluate the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface when a cross-process contamination transmission chain is identified. The rating assessment module is used to assess the compatibility risk level of the silicon wafer surface for photolithography processes based on competitive adsorption tendencies. The parameter adjustment module is used to dynamically adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaner according to the compatibility risk level.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By integrating online monitoring and intelligent analysis, precise control of interface contamination during hydrocarbon cleaning is achieved. By simultaneously acquiring surface solvent residual concentration and contact angle data, the adsorption state of solvent molecules on the silicon wafer surface can be captured in real time, thereby identifying the synergistic anomalies of volatility and polarity characteristics. This can effectively identify molecular-level residual patterns that are difficult to detect by traditional methods, thus providing early warning of interface property degradation caused by solvent residue. Combined with the correlation analysis of upstream process parameters and historical contamination data, the cross-process transfer path of contaminants can be traced, identifying the root cause of interface compatibility problems from the source, and avoiding the limitations of single-process optimization.

[0016] 2. By assessing the competitive adsorption tendency of solvent residues and photoresist components, the interfacial behavior of silicon wafers during photolithography is accurately predicted, thereby establishing a compatibility risk level based on the actual interface state. Based on this risk level, the rinsing flow rate and drying temperature parameters are dynamically adjusted, enabling the cleaning process to adapt to different contamination conditions, effectively disrupting the adsorption balance of residual solvent molecules, and optimizing the drying process to suppress the formation of secondary contamination films. This not only improves the cleaning process's ability to eliminate micro-interfacial contamination but also ensures the matching degree between silicon wafer surface characteristics and subsequent photolithography processes, fundamentally solving the problem of pattern defects caused by cleaning residues. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the usage method of an intelligent hydrocarbon cleaning machine according to the present invention.

[0018] Figure 2 This is a schematic diagram of the structure of an intelligent hydrocarbon cleaning machine according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: Figure 1 The present invention provides a method for using a hydrocarbon cleaning machine, comprising: S1. Obtain data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; S2. Based on surface solvent residual concentration and surface contact angle data, identify the abnormal state of synergistic effect between volatility and polarity characteristics in the residual solvent components; S3. When an abnormal state of synergistic effect is identified, retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process, and perform correlation analysis with the surface solvent residual concentration to identify the existence of cross-process contamination transmission chain. S4. When a cross-process contamination transmission chain is identified, assess the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface. S5. Based on the competitive adsorption tendency, assess the compatibility risk level of the silicon wafer surface for the photolithography process; S6. Adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaning machine dynamically according to the compatibility risk level.

[0021] S1. Obtain data on the residual solvent concentration and surface contact angle of the silicon wafer after hydrocarbon cleaning through an online monitoring system. Specifically, the implementation is as follows: Immediately after the hydrocarbon cleaning process, the silicon wafers are transferred to the integrated measurement station for surface analysis. Fourier transform infrared spectroscopy analysis of the silicon wafer surface is performed using reflectance mode to acquire spectral data, with a spectral scan range of wavenumber 4000 cm⁻¹. -1 Up to 650 cm -1 The resolution is 4 cm. -1 A total of 64 scans were performed. The characteristic absorption peak of the carbon-hydrogen bond at 2930 cm⁻¹ was identified. -1 The peak intensity at the specified location was compared with a pre-established calibration curve, constructed by preparing standard solvent samples of known concentrations on a clean silicon wafer and measuring their infrared absorption spectra. This converted the characteristic peak intensity into the residual surface solvent concentration, expressed in ng / cm³. 2 Quantify the units.

[0022] Simultaneously, static contact angle measurements were performed on the silicon wafer surface. Using an automated droplet analyzer, 2 μL ultrapure water droplets were deposited at three different locations on the silicon wafer surface. The droplet deposition rate was controlled at 0.5 μL / s. A high-speed camera was used to record the droplet profile at a rate of 100 frames / second. The droplet profile curve was fitted using the Young-Laplace equation to calculate the static contact angle value. The average value of the three measurement locations was taken as the surface contact angle data of the silicon wafer, in degrees.

[0023] To ensure the synchronization and consistency of data acquisition, Fourier transform infrared spectroscopy analysis and static contact angle measurement are completed within the same time window, defined as 90 seconds from the arrival of the silicon wafer at the measurement station. Both measurements share the same silicon wafer positioning coordinate system, and the silicon wafer is maintained in a constant temperature and humidity environment of 23°C and 45% relative humidity during the measurement process. The measurement sequence is coordinated by a central controller, and the time difference between the start time of Fourier transform infrared spectroscopy analysis and the first droplet deposition time of the static contact angle measurement does not exceed 5 seconds.

[0024] The process of establishing the calibration curve for the residual concentration of surface solvent includes preparing a concentration gradient of 0 ng / cm³. 2 Up to 200 ng / cm 2 For standard solvent samples, five parallel samples were prepared for each concentration level. The relationship between characteristic peak intensity and solvent concentration was fitted using the least squares method, and the goodness-of-fit index R0 was used. 2 The value is not less than 0.995. The droplet profile analysis of static contact angle measurement adopts the tangent method. A tangent is drawn at the contact point between the droplet and the solid surface, and the angle between the tangent and the solid surface is the contact angle. Profile data of each droplet is collected and analyzed 3 seconds after deposition.

[0025] In Fourier transform infrared spectroscopy analysis, the identification of characteristic absorption peaks employs a peak detection algorithm. This algorithm sets the peak height threshold to three times the standard deviation of the baseline noise and the peak half-width at half-maximum threshold to be no less than 8 cm. -1 The resistivity of the ultrapure water used in the static contact angle measurement was 18.2 MΩ·cm, the surface tension was 72.0 mN / m, the distance between the droplet deposition needle and the silicon wafer surface was kept at 1 mm, and the needle diameter was 0.5 mm.

[0026] Data acquisition synchronization is achieved through a timestamp mechanism, with the time difference between the completion of Fourier transform infrared spectroscopy analysis and the completion of static contact angle measurement not exceeding 15 seconds. Environmental conditions are continuously monitored by an environmental monitoring system, with temperature fluctuations controlled within ±0.5℃ and relative humidity fluctuations within ±3%. The positioning accuracy of the silicon wafer during measurement is better than 0.1 mm in all three X, Y, and Z directions.

[0027] In the quantification of surface solvent residual concentration, the influence of substrate effect and molecular orientation on infrared absorption intensity was considered. Compensation was achieved by introducing a substrate correction factor, which was determined by comparing the differences in infrared absorption intensity on different substrates with the same solvent concentration. In static contact angle measurement, the droplet deposition positions were selected according to a uniform distribution principle, with three measurement points located at the center of the silicon wafer and two symmetrical points at one-third of the diameter from the center.

[0028] Fourier transform infrared spectroscopy data processing includes baseline correction, atmospheric compensation, and normalization steps. Baseline correction uses linear baseline fitting, atmospheric compensation mainly eliminates interference absorption bands from carbon dioxide and water vapor, and normalization is performed on a silicon substrate at 2100 cm⁻¹. -1 The reflection intensity at the point is used as the internal standard. In the droplet profile analysis based on static contact angle measurement, the droplet edge detection uses the Canny edge detection algorithm. The high threshold of this algorithm is set at 70% of the gray-level gradient histogram, and the low threshold is set to 0.5 times the high threshold.

[0029] The measurement system is calibrated at fixed intervals. The Fourier transform infrared spectrometer uses polystyrene standard sheets for wavelength calibration, and the static contact angle meter uses standard angle templates for angle calibration. The calibration cycle is once every 7 days. All measurement data is automatically recorded and linked to the corresponding silicon wafer's identification code, forming a complete traceable data chain.

[0030] S2. Based on surface solvent residual concentration and surface contact angle data, identify the abnormal state of synergistic effect between volatility and polarity characteristics in the solvent residual components. Specifically, this is implemented as follows: When determining volatile characteristic parameters based on surface solvent residual concentration, a pre-stored solvent property database is first retrieved from the data storage unit. This database contains volatile parameters for various standard solvents, such as evaporation rate and saturated vapor pressure, which are measured and stored using standard experimental methods. The surface solvent residual concentration is then matched with the standard solvent volatile parameters in the database. The matching process uses a nearest neighbor algorithm, calculating the Euclidean distance between the surface solvent residual concentration and each standard solvent concentration in the database, and selecting the volatile parameter corresponding to the standard solvent with the smallest distance as the volatile characteristic parameter. The volatile characteristic parameter is expressed as a dimensionless value, ranging from 0 to 1, where 0 represents no volatility and 1 represents the highest volatility. Database construction includes collecting property data for common hydrocarbon solvents, such as n-hexane and isopropanol, and validating their volatile parameters using gas chromatography to ensure data accuracy and reliability. During the matching process, if the surface solvent residual concentration is close to multiple standard solvent concentrations, the average value of the volatile parameters is taken as the final result. The determination of volatile characteristic parameters also takes into account environmental factors, such as ambient temperature and pressure, and is compensated by introducing an environmental correction factor, which is calculated based on real-time monitored environmental conditions.

[0031] When determining polar characteristic parameters based on surface contact angle data, a surface free energy calculation model is applied to process the surface contact angle data. This model, based on the Owens-Wendt method, calculates the surface free energy from the contact angle data and extracts the polar component as the polar characteristic parameter. The input parameters of the surface free energy calculation model include the surface contact angle data and the liquid surface tension parameter, which is taken as the standard value of 72.0 mN / m for ultrapure water. The calculation process first substitutes the contact angle into the Young equation to obtain the total surface free energy, and then decomposes it into polar and dispersive components using the Owens-Wendt formula. The polar component is the polar characteristic parameter. The polar characteristic parameter is expressed in mN / m, and its value range is typically 0 to 50 mN / m. In the model calculation, the contact angle data is taken as the average of three measurement points to ensure the representativeness of the results. The extraction of polar characteristic parameters also considers the influence of surface roughness, which is corrected by introducing a roughness correction factor. The roughness correction factor is obtained by measuring the surface morphology using atomic force microscopy and calculating it. The parameters of the surface free energy calculation model are set based on standard values ​​in the literature. For example, the reference value for the dispersive component is set to 21.8 mN / m, and the reference value for the polar component is set to 51.0 mN / m. These parameters are calibrated and verified using standard samples before calculation.

[0032] When analyzing the correlation between volatile and polar characteristic parameters, statistical correlation analysis is employed to calculate the Pearson correlation coefficient. The Pearson correlation coefficient is calculated based on at least 10 sets of paired data for volatile and polar characteristic parameters, derived from measurements of the same batch of silicon wafers. The formula for calculating the correlation coefficient is described as follows: first, calculate the covariance of the volatile and polar characteristic parameters; then, calculate the standard deviation of each parameter; finally, divide the covariance by the product of the two standard deviations to obtain the correlation coefficient. The correlation coefficient ranges from -1 to 1, where a positive value indicates a positive correlation and a negative value indicates a negative correlation. Data preprocessing in the correlation analysis includes outlier removal. Outlier identification is based on box plot methods; for example, data points exceeding 1.5 times the quartile range are considered outliers. The results of the correlation analysis are output numerically and recorded in the analysis report.

[0033] When the correlation is lower than a preset correlation threshold, it is determined to be an abnormal state of synergy effect. The preset correlation threshold is set based on historical data statistics. By analyzing the correlation data of the volatile characteristic parameters and polarity characteristic parameters of the past 100 normal silicon wafers, the average and standard deviation of their correlation coefficients are calculated. The average value minus twice the standard deviation is used as the preset correlation threshold. For example, if the historical average correlation coefficient is 0.8 and the standard deviation is 0.1, then the preset correlation threshold is set to 0.6. The judgment logic is: if the calculated correlation coefficient is lower than the preset correlation threshold, an abnormal state of synergy effect is triggered; otherwise, it is considered a normal state. The judgment of abnormal state of synergy effect also considers process fluctuation factors. For example, if the current process parameters deviate significantly from historical data, the preset correlation threshold is adjusted, and the adjustment range is calculated based on the degree of process deviation. The judgment result is automatically recorded and associated with the silicon wafer identification code for subsequent analysis.

[0034] The correlation analysis between volatile characteristic parameters and polar characteristic parameters also includes trend analysis. This involves observing the parameter distribution by plotting scatter plots. If the data points exhibit a significantly discrete or non-linear distribution, the correlation is considered insufficient. In the trend analysis, a moving average method is used to smooth the data, with the moving window size set to 5 data sets to eliminate the influence of random fluctuations. The output of the correlation analysis also includes a confidence level, obtained by calculating the p-value of the correlation coefficient. A p-value less than 0.05 indicates a significant correlation. The dynamic adjustment of the preset correlation threshold is based on real-time process data. For example, if the ambient temperature fluctuates by more than ±2℃, the preset correlation threshold is temporarily lowered by 0.1 to cope with changes in conditions.

[0035] During the determination of volatile characteristic parameters, the database is updated every 30 days. Updates include adding new solvent data or correcting existing data, based on the latest experimental reports. For polar characteristic parameter calculations, the surface free energy calculation model is validated using standard samples. For example, the polar component of polytetrafluoroethylene (PTFE) samples should be close to 0 mN / m, and the polar component of polyethylene samples should be approximately 0.5 mN / m, with the validation error controlled within ±5%. Correlation analysis is performed synchronously with each silicon wafer production batch, with at least one analysis per batch to ensure timely detection of anomalies. The results of synergistic effect anomaly detection are also used to trigger early warning signals, which are communicated to operators via audible and visual alarms.

[0036] S3. When an abnormal state of synergistic effect is identified, retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the previous manufacturing process, and perform correlation analysis with the surface solvent residual concentration to identify the existence of a cross-process contamination transmission chain. The specific implementation is as follows: When an abnormal state of synergy is identified, the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process are retrieved from the Manufacturing Execution System (MES). The retrieval process involves precise matching of the silicon wafer identification code with the data records in the MES. The front-end manufacturing process parameters include, for example, chemical mechanical polishing pressure values ​​of 0.5 psi to 5 psi, cleaning solution concentration values ​​of 10% to 30%, and drying temperatures of 80°C to 150°C. Historical contamination data includes, for example, particulate contaminant concentration values ​​of 0.1 μg / cm³. 2 Up to 5 μg / cm 2 The organic residues are either fatty acid esters or siloxanes. The data retrieval time window covers the entire processing cycle of the current silicon wafer in the preceding processes, such as all process nodes from wafer feeding to the current cleaning process, typically spanning 2 to 8 hours. Data in the Manufacturing Execution System (MES) is stored in a structured format. Process parameters are recorded as time-series data with sampling intervals of 1 to 5 minutes. Historical contamination data is recorded as test report files, containing contaminant concentration values ​​and chemical composition information. During retrieval, if data is missing or abnormal, a data completion procedure is initiated. Linear interpolation is performed using corresponding parameters from other silicon wafers in the same batch. The interpolation calculation is based on the assumption of time series continuity, and the value of a missing data point is determined by the arithmetic mean of two consecutive valid data points.

[0037] A time-series correlation analysis was performed between the residual surface solvent concentration and the process parameters of the preceding manufacturing process. The analysis employed a dynamic time warping algorithm to dynamically align and compare the similarity of the time-series data of the residual surface solvent concentration with that of the preceding manufacturing process parameters. During the analysis, the two time-series data were first standardized using z-score normalization to eliminate dimensional differences. Then, the dynamic time warping distance between the sequences was calculated; a smaller distance indicates greater similarity in the two time series. The significance of the time-series correlation analysis was determined using a preset correlation threshold, which was derived from historical normal production data. For example, by analyzing the time-series correlation data of 100 normal silicon wafers, the 95th percentile of the dynamic time warping distance was used as the correlation threshold, typically ranging from 0.1 to 0.5. If the calculated dynamic time warping distance was less than the correlation threshold, the time-series correlation was considered significant. The time-series correlation analysis also considers the interactive effects between process parameters. For example, when chemical mechanical polishing pressure and cleaning fluid concentration both fluctuate abnormally, the judgment criteria for the time-series correlation threshold will be adjusted accordingly, and the adjustment range is based on a weighted calculation of the degree of parameter deviation.

[0038] The composition consistency of surface solvent residue concentration is compared with historical pollution data. This comparison is achieved by comparing the chemical composition of the surface solvent residue with that of historical pollution data. Specifically, this includes comparing parameters such as molecular structure characteristics, functional group types, and elemental composition. For example, functional group types are compared using infrared spectral characteristic peaks, and molecular weight distribution is compared using mass spectrometry data. The comparison process employs similarity calculation algorithms, such as calculating the cosine similarity between the two sets of data to quantify composition consistency. The cosine similarity calculation is based on the dot product operation of eigenvectors. The significance of the composition consistency comparison is determined by a preset composition consistency threshold, which is statistically derived from a historical pollution database. For example, by analyzing data from known pollution events, the 90th percentile of the similarity value is used as the composition consistency threshold, typically ranging from 0.7 to 0.9. If the calculated similarity is greater than the composition consistency threshold, the composition consistency is considered significant. A minimum data requirement is also set for the composition consistency comparison; for example, at least five sets of historical pollution data are required for effective comparison.

[0039] When both time-series correlation analysis and component consistency comparison show significant correlations, a cross-process contamination transmission chain is determined to exist. The determination logic is that when the dynamic time warp distance in the time-series correlation analysis is less than the time-series correlation threshold, and the similarity in the component consistency comparison is greater than the component consistency threshold, a cross-process contamination transmission chain determination is triggered. The determination process also considers the influence of environmental factors. For example, when the ambient temperature fluctuates by more than ±2℃, the time-series correlation threshold and the component consistency threshold are dynamically adjusted. The adjustment range is calculated based on the product of the temperature change and the threshold sensitivity coefficient. The temperature change is obtained through real-time monitoring data, and the threshold sensitivity coefficient is experimentally determined to be 0.05 degrees Celsius. The determination results are stored in association with the silicon wafer identification code, and a contamination transmission chain report is generated. The report records information such as the involved process nodes, contamination types, and correlation strength. The correlation strength is calculated based on the weighted average of the dynamic time warp distance and the similarity.

[0040] Data preprocessing for time-series correlation analysis includes outlier removal and smoothing. Outlier removal uses box plots, marking data points exceeding 1.5 times the quartile range as outliers. Smoothing uses moving averages with a window size of 5 data points. For component consistency comparison, the historical pollution database is updated regularly every 30 days, including adding new pollutant types and revising existing data, based on the latest pollution detection reports. The determination of cross-process pollution transmission chains is also used to trigger an early warning mechanism. Early warning levels are divided into multiple grades based on correlation strength: high correlation corresponds to a correlation strength greater than 0.8, moderate correlation corresponds to a correlation strength of 0.5 to 0.8, and low correlation corresponds to a correlation strength less than 0.5. Different grades correspond to different processing procedures.

[0041] Data retrieval in the Manufacturing Execution System (MES) employs a hierarchical query strategy. First, basic process parameters are queried, then relevant auxiliary parameters are queried based on the type of abnormal state. In time-series correlation analysis, the dynamic time warping algorithm is constrained to a symmetrical step size mode, with the slope of the warped path limited to between 0.5 and 2 to ensure the rationality of time series alignment. In component consistency comparison, feature vectors include molecular weights ranging from 100 g / mol to 500 g / mol, polarity parameters from 0 to 1, and volatility parameters from 0 to 1. These parameters are obtained through experimental measurements or literature data. The determination of cross-process contamination transmission chains also considers the interactive effects of process parameters. For example, when multiple process parameters simultaneously deviate from the standard range, the judgment standard for time-series correlation thresholds is correspondingly increased, with the increase based on a weighted calculation of the parameter deviation degrees. In component consistency comparison, if a new contaminant is encountered, an expert review process is initiated. After determining its chemical characteristics through laboratory analysis, the comparison rules are updated. The execution frequency of the entire judgment process is synchronized with the production cycle, with each batch of silicon wafers undergoing at least one complete analysis to ensure timely identification of contamination transmission chains.

[0042] S4. When a cross-process contamination transmission chain is identified, assess the competitive adsorption tendency of solvent residues and main photoresist components on the silicon wafer surface. Specifically, this is implemented as follows: When identifying key active components in solvent residues based on cross-process contamination transport chains, the first step is to analyze the pollutant types and concentrations recorded in the cross-process contamination transport chain, and then screen organic compounds with specific functional groups as candidate components. Screening criteria include molecular weights ranging from 100 g / mol to 500 g / mol, containing polar functional groups such as hydroxyl or carboxyl groups, and compounds with an adsorption tendency on silicon wafer surfaces. The determination of key active components is achieved by calculating the adsorption potential energy of each component. Adsorption potential energy calculations are based on molecular dynamics simulations, considering factors such as molecular size, polarity, and surface interaction energy. During the calculation process, the three to five components with the highest adsorption potential energy values ​​are selected as key active components. These components typically have high surface activity and competitive adsorption capacity. The identification of key active components also references the frequency of component occurrence in historical contamination data; for example, components with a frequency greater than 10% are prioritized as key active components.

[0043] When obtaining standard adsorption free energy data for the main components of photoresist on the silicon wafer surface, the adsorption free energy values ​​of the main components are extracted from a pre-established thermodynamic database. The thermodynamic database contains standard adsorption free energy data for common photoresist components such as phenolic resin, photosensitizers, and additives, obtained through experimental measurements or literature reports. The unit of standard adsorption free energy data is kJ / mol, and the numerical range is typically -10 kJ / mol to -50 kJ / mol; negative values ​​indicate that the adsorption process is spontaneous. The database establishment process includes collecting the physical properties of different photoresist components, determining their adsorption free energies through surface adsorption experiments, and simulating actual process environments under experimental conditions, such as a temperature range of 20°C to 30°C and a relative humidity range of 40% to 60%. The database is updated regularly, every 6 months, with updates including the addition of new photoresist components and revisions to existing data.

[0044] Competitive adsorption tendency is assessed by comparing the relative adsorption free energies of the key active component and the main components of the photoresist. The comparison process first compares the absolute values ​​of the adsorption free energies of the key active component and the main components of the photoresist, calculating the absolute value of the difference. The assessment of competitive adsorption tendency is based on a preset adsorption free energy difference threshold, which is derived statistically from historical experimental data. For example, by analyzing 50 known cases of competitive adsorption, the 80th percentile of the adsorption free energy difference is taken as the adsorption free energy difference threshold, which is typically between 5 kJ / mol and 15 kJ / mol. If the calculated adsorption free energy difference is greater than the threshold, the key active component is considered to have a strong competitive adsorption tendency; if the adsorption free energy difference is less than or equal to the threshold, the competitive adsorption tendency is considered to be weak.

[0045] The determination of key active components also considers the synergistic effect of components. For example, when multiple key active components are present simultaneously, a weighted average method is used to calculate the comprehensive adsorption potential energy, with the weights determined based on the concentration ratio and polarity parameters of each component. When acquiring standard adsorption free energy data for the main components of the photoresist, if a new photoresist component is encountered, supplementary data are obtained through laboratory measurements. Measurement methods include gravimetric methods or quartz crystal microbalance methods, with the measurement temperature controlled at 25℃±0.5℃ and the relative humidity controlled at 50%±5%. The assessment of competitive adsorption tendency also includes dynamic factor analysis, such as considering the impact of process temperature changes on adsorption free energy. A temperature correction coefficient is introduced to adjust the adsorption free energy value, calculated based on the Arrhenius equation.

[0046] During the identification of key active components, if unknown components are included in the cross-process contamination transfer chain, a component identification process is initiated. Gas chromatography-mass spectrometry (GC-MS) is used to determine their chemical structure, and then their adsorption potential is calculated. The standard adsorption free energy database of the main components of the photoresist is validated using standard samples. For example, the standard adsorption free energy of polymethyl methacrylate (PMMA) should be -25 kJ / mol ± 2 kJ / mol, with the validation error controlled within ±5%. The evaluation results of competitive adsorption tendency are also used to generate an adsorption tendency report. The report records a list of key active components, comparative data on adsorption free energies, and a competitive adsorption tendency level, which is divided into three levels: strong, medium, and weak.

[0047] When identifying key active components in solvent residues based on cross-process contamination transport chains, the residual time of contaminants on the silicon wafer surface was also considered; for example, components with a residual time greater than 10 minutes were prioritized for investigation. When obtaining standard adsorption free energy data of the main photoresist components on the silicon wafer surface, a multi-condition matching strategy was employed in the database query, including query conditions such as photoresist type, component molecular weight, and functional group type. When assessing competitive adsorption tendency by comparing the relative magnitudes of the adsorption free energies of key active components and the main photoresist components, the influence of surface coverage was also considered; for example, when the surface coverage of the key active component was greater than 50%, the competitive adsorption tendency was automatically determined to be strong.

[0048] During the screening process for key active components, a minimum concentration threshold was set; for example, only components with a concentration greater than 0.1 μg / cm³ were considered. 2 The components are included in the candidate list. When acquiring standard adsorption free energy data for the main components of the photoresist, if multiple data sources exist for the same component, the average value from each data source is taken as the final value. During the evaluation of competitive adsorption tendency, if the difference in adsorption free energy between the key active component and the main components of the photoresist is in the critical region, for example, within ±1 kJ / mol of the adsorption free energy difference threshold, a verification procedure is initiated, and verification analysis is performed using surface analysis instruments such as X-ray photoelectron spectroscopy. The execution frequency of the entire evaluation process is synchronized with the identification of contamination transmission chains; a comprehensive evaluation is completed for each contamination transmission chain identified.

[0049] S5. Based on the competitive adsorption tendency, assess the compatibility risk level of the silicon wafer surface for the photolithography process. The specific implementation is as follows: When determining the basic level based on the strength of competitive adsorption tendency, the assessment results are first divided into three levels: strong competitive adsorption tendency, moderate competitive adsorption tendency, and weak competitive adsorption tendency. Strong competitive adsorption tendency corresponds to basic level 3, moderate competitive adsorption tendency to basic level 2, and weak competitive adsorption tendency to basic level 1. The strength of competitive adsorption tendency is classified based on the adsorption free energy difference threshold, which is obtained through statistical analysis of historical experimental data. For example, by analyzing 100 known competitive adsorption cases, the 70th percentile of the adsorption free energy difference is taken as the strong tendency threshold, and the 30th percentile as the weak tendency threshold. The determination of the basic level also considers the surface coverage parameter; for example, when the surface coverage is greater than 50%, the basic level is automatically upgraded by one level. The mapping relationship of the basic levels is implemented through a predefined rule table, which is constructed based on process experience data and experimental verification results. The rule table is updated every 6 months, based on newly accumulated process data.

[0050] When adjusting the baseline level based on the persistence impact coefficient of pollutants in the cross-process contamination transmission chain, the persistence impact coefficient is calculated based on historical data of the pollutant's residual time and concentration on the silicon wafer surface. The persistence impact coefficient ranges from 0.1 to 1.0, with higher values ​​indicating stronger persistent impact. The persistence impact coefficient is calculated by analyzing the pollutant's residual half-life and concentration decay curves on the silicon wafer surface; for example, pollutants with a residual half-life greater than 10 minutes are assigned a higher persistence impact coefficient value. When adjusting the baseline level, the baseline level is multiplied by the persistence impact coefficient to obtain the adjusted level value. The adjustment process also considers the influence of pollutant type; for example, organic pollutants have a higher persistence impact coefficient weight than inorganic pollutants. Environmental factor corrections are also incorporated into the calculation of the persistence impact coefficient; for example, when the ambient humidity is greater than 60%, the persistence impact coefficient is increased by 0.1.

[0051] When outputting the corrected base level as the compatibility risk level, the corrected level value is mapped to a specific risk level classification. Compatibility risk levels are divided into three levels: high risk, medium risk, and low risk. For example, a corrected level value greater than 2.5 corresponds to high risk, a corrected level value between 1.5 and 2.5 corresponds to medium risk, and a corrected level value less than 1.5 corresponds to low risk. The risk level output is transmitted to the Manufacturing Execution System (MES) via a data interface, generating a risk report document. The risk report document contains information such as the silicon wafer identification code, risk level value, risk classification, and determination criteria. The determination of the compatibility risk level also references historical risk case data; for example, the current risk level is compared and verified with the historical risk levels of similar silicon wafers.

[0052] In determining the baseline risk level, the assessment of the strength of competitive adsorption tendency also considers the influence of temperature. For example, when the process temperature exceeds 25℃, the criteria for judging competitive adsorption tendency are adjusted accordingly, with the strong tendency threshold increased by 5%. The calculation of the persistence coefficient is based on pollutant concentration time series data. The concentration change curve is fitted using an exponential decay model, and the decay constant is extracted as the main parameter of the persistence coefficient. When correcting the corresponding baseline level, if multiple pollutants exist simultaneously, the maximum value of the persistence coefficients of each pollutant is taken as the final correction coefficient. The output format of the compatibility risk level follows the plant data specifications, such as using JSON format to encapsulate risk data, including timestamps and equipment identification information.

[0053] The baseline risk level is based on process stability studies, such as establishing the correlation between competitive adsorption tendency and defect rate by analyzing photolithography defect data from 500 silicon wafer samples. The calculation of the persistence factor also incorporates contaminant toxicity parameters, obtained by consulting chemical safety databases; for example, a toxicity parameter of 1.0 is set for contaminants with an LD50 value less than 100 mg / kg. During the correction process, the multiplication of the baseline risk level and the persistence factor is rounded to the nearest integer to ensure the level value is an integer. The output of the compatibility risk level also includes a confidence index, which is derived by calculating the matching degree of historical data; for example, when the matching degree between the current data and historical patterns is greater than 90%, the confidence level is set to high.

[0054] The determination of the baseline risk level also considers surface energy parameters. For example, when the surface energy of the silicon wafer is less than 30 mN / m, the baseline level is automatically reduced by one level. The persistence coefficient is updated every 24 hours, based on pollutant concentration data from real-time monitoring. When correcting the corresponding baseline level, if the persistence coefficient is less than 0.3, the baseline level remains unchanged. The output interface for the compatibility risk level is linked to the alarm system; for example, when a high-risk level is output, an audible and visual alarm is automatically triggered. The entire evaluation process is performed once per silicon wafer, and the evaluation time is controlled within 5 minutes.

[0055] The mapping rule table for the basic risk levels is revised periodically, every three months, based on newly accumulated process data. In calculating the persistent impact coefficient, residual time data is obtained through surface analysis instruments, such as X-ray photoelectron spectroscopy, to measure the change in contaminant concentration over time. When revising the corresponding basic level, the result of the multiplication operation is limited to between 1 and 3 to avoid overflow of the level value. The output data for the compatibility risk level is stored in a dedicated database and retained for two years for subsequent quality traceability.

[0056] During the determination of the baseline level, if data on competitive adsorption tendency is missing, the default baseline level 2 is used. The calculation of the persistence factor considers pollutant interactions; for example, when multiple pollutants coexist, the persistence factor is the weighted average of the coefficients for each pollutant, with the weights determined based on the pollutant concentration ratio. When correcting the corresponding baseline level, the decimal part of the correction coefficient is rounded up; for example, a corrected level value of 1.6 is rounded to 2. Compatibility risk level reports are automatically distributed to relevant process engineers via email or instant messaging.

[0057] S6. Based on the compatibility risk level, dynamically adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaning machine, specifically as follows: When matching the baseline values ​​for rinsing flow rate and drying temperature based on the compatibility risk level, the compatibility risk level is first divided into three levels: high risk, medium risk, and low risk. High risk corresponds to a baseline rinsing flow rate of 2.5 to 3.5 liters per minute and a baseline drying temperature of 90°C to 110°C; medium risk corresponds to a baseline rinsing flow rate of 1.5 to 2.5 liters per minute and a baseline drying temperature of 70°C to 90°C; and low risk corresponds to a baseline rinsing flow rate of 0.5 to 1.5 liters per minute and a baseline drying temperature of 50°C to 70°C. The baseline values ​​are set based on historical process data statistics and experimental verification. For example, by analyzing 1000 sets of silicon wafer cleaning data, a relationship model between risk level and cleaning effect is established to determine the optimal parameter range for each risk level. The matching process is implemented through a predefined parameter mapping table, which is constructed based on process experience and experimental data. For example, a high-risk level is mapped to a high rinsing flow rate and a high drying temperature to enhance the cleaning effect and remove contaminants. The parameter mapping table is updated every 3 months, based on newly accumulated process data and cleaning effect feedback.

[0058] When generating a rinsing flow rate parameter adjustment instruction based on the rinsing flow rate baseline value, the current rinsing flow rate baseline value is first read, and then the difference between the target rinsing flow rate value and the current actual rinsing flow rate value is calculated. The generation of the rinsing flow rate parameter adjustment instruction is based on a proportional-integral-derivative (PID) control algorithm, which obtains the adjustment amount by calculating the sum of the proportional, integral, and derivative terms of the difference. The proportional coefficient of the PID control algorithm is set to 0.8, the integral time is set to 1 minute, and the derivative time is set to 0.1 minutes; these parameters are determined through process debugging. The rinsing flow rate parameter adjustment instruction is output in digital signal form, and the instruction content includes the target rinsing flow rate value, the adjustment range, and the execution time. For example, the instruction format is "Set rinsing flow rate to 2.8 liters per minute." Equipment limitations are also considered during the instruction generation process; for example, the rinsing flow rate adjustment range is limited to between 0.5 liters and 5.0 liters per minute to avoid equipment overload.

[0059] When generating a drying temperature parameter adjustment command based on a drying temperature reference value, the current drying temperature reference value is first read, and then the difference between the target drying temperature value and the current actual drying temperature value is calculated. The generation of the drying temperature parameter adjustment command employs a fuzzy logic control algorithm. This algorithm obtains the adjustment amount by fuzzifying the input difference, applying a fuzzy rule library, and defuzzifying the output. The fuzzy rule library is constructed based on the experience of process experts; for example, if the difference is large positive, the adjustment amount is large positive, and if the difference is small negative, the adjustment amount is small negative. The drying temperature parameter adjustment command is output in the form of an analog signal. The command content includes the target drying temperature value, the heating rate, and the holding time. For example, the command format is: set the drying temperature to 95℃ and heat at a rate of 5℃ per minute. Safety boundary conditions are also set during the command generation process, such as the upper limit of the drying temperature not exceeding 120℃, to prevent equipment damage.

[0060] When dynamically controlling the hydrocarbon cleaning machine by executing rinsing flow rate and drying temperature adjustment commands, the rinsing flow rate adjustment command is first sent to the cleaning machine's flow control valve. The flow control valve adjusts its opening according to the command to change the rinsing flow rate. The response time of the flow control valve is less than 1 second, and the control accuracy is ±2% of full scale. Simultaneously, the drying temperature adjustment command is sent to the cleaning machine's heating controller. The heating controller adjusts its power output according to the command to change the drying temperature. The response time of the heating controller is less than 3 seconds, and the control accuracy is ±1℃. During dynamic control, the actual values ​​of the rinsing flow rate and drying temperature are monitored in real time and compared with the target values. If the deviation exceeds the allowable range, the adjustment command is regenerated. After executing the command, the actual parameter values ​​and adjustment results are recorded for subsequent process optimization.

[0061] The matching process for the reference values ​​also considers the influence of silicon wafer type. For example, for 300mm silicon wafers, the rinsing flow rate reference value is increased by 10%. The generation of rinsing flow rate parameter adjustment commands also includes a data verification step, such as checking whether the target rinsing flow rate value is within the equipment's allowable range; if it exceeds the range, a boundary value is used. The generation of drying temperature parameter adjustment commands incorporates ambient temperature compensation; for example, when the ambient temperature is below 20°C, the drying temperature reference value is automatically increased by 5°C. An exception handling mechanism is set up during command execution; for example, if the flow control valve response times out, an alarm is triggered and the system switches to manual control mode.

[0062] The mapping relationship between compatibility risk levels and baseline values ​​is based on multi-factor decision-making, taking into account factors such as silicon wafer surface materials and contaminant types in addition to risk levels. The algorithm parameters for generating rinsing flow rate adjustment instructions are calibrated experimentally, for example, by testing control effectiveness under different flow conditions and optimizing proportional-integral-derivative coefficients. The fuzzy rule base for drying temperature parameter adjustment instructions contains 25 rules, covering all possible input-output combinations. After instruction execution, the system generates an execution report, which includes the instruction content, execution time, actual parameter values, and deviation data.

[0063] The determination of the baseline values ​​also incorporates real-time process feedback, such as dynamically adjusting the baseline value range based on the cleaning results of the most recent 10 silicon wafers. The transmission of rinsing flow rate parameter adjustment commands utilizes the industrial Ethernet protocol to ensure real-time and reliable data transmission. The execution of drying temperature parameter adjustment commands includes a preheating phase, such as maintaining the temperature for 60 seconds before heating to achieve thermal equilibrium. If a device malfunction is detected during dynamic control, command execution is automatically paused and maintenance personnel are notified.

[0064] The parameter mapping table is constructed based on big data analytics, such as using machine learning algorithms to analyze patterns in historical data and optimize mapping relationships. The generation of rinsing flow rate adjustment instructions takes into account fluid dynamics characteristics, such as correcting target flow rates based on pipeline pressure loss. The generation of drying temperature parameter adjustment instructions includes temperature gradient control, such as limiting the heating rate to no more than 10°C per minute to avoid thermal shock. The instruction execution system is integrated with the plant monitoring system to achieve remote monitoring and data analysis.

[0065] The baseline values ​​are adjusted based on continuous process improvements, such as quarterly reviews of baseline settings and updates based on new technological advancements. The generation of rinsing flow rate adjustment instructions utilizes digital twin technology, simulating the adjustment effect through a virtual model before sending the actual instructions. The execution of drying temperature parameter adjustment instructions involves multi-point temperature monitoring, such as measuring the temperature at three locations within the heating chamber and averaging the results. The dynamic control process is executed once per silicon wafer, ensuring personalized parameter adjustments.

[0066] If a new risk level is encountered during the baseline matching process, an expert review process is initiated to determine a temporary baseline value. The algorithm for generating rinsing flow rate parameter adjustment instructions also includes adaptive functions, such as automatically adjusting control parameters based on equipment wear. The transmission of drying temperature parameter adjustment instructions uses a redundant communication protocol to ensure reliable delivery of instructions. After the instructions are executed, the system performs an effect evaluation, such as measuring the cleanliness of the silicon wafer surface and comparing it with the expected target.

[0067] Example 2: Figure 2 A schematic diagram of the structure of an intelligent hydrocarbon cleaning machine according to the present invention is provided. The intelligent hydrocarbon cleaning machine includes the following modules: The data acquisition module is used to acquire data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; Anomaly detection module is used to identify abnormal states of synergistic effects between volatile and polar characteristics in residual solvent components based on surface solvent residual concentration and surface contact angle data. An identification module is used to retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process when an abnormal state of synergistic effect is identified, and to perform correlation analysis with the surface solvent residual concentration to identify the existence of a cross-process contamination transmission chain. The tendency assessment module is used to evaluate the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface when a cross-process contamination transmission chain is identified. The rating assessment module is used to assess the compatibility risk level of the silicon wafer surface for photolithography processes based on competitive adsorption tendencies. The parameter adjustment module is used to dynamically adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaner according to the compatibility risk level.

[0068] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0069] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0070] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0071] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0072] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0073] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0075] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0077] In conclusion, 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, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of using an intelligent hydrocarbon cleaning machine, characterized in that, include: S1. Obtain data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; S2. Based on surface solvent residual concentration and surface contact angle data, identify the abnormal state of synergistic effect between volatility and polarity characteristics in the residual solvent components; S3. When an abnormal state of synergistic effect is identified, retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process, and perform correlation analysis with the surface solvent residual concentration to identify the existence of cross-process contamination transmission chain. S4. When a cross-process contamination transmission chain is identified, assess the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface. S5. Based on the competitive adsorption tendency, assess the compatibility risk level of the silicon wafer surface for the photolithography process; S6. Adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaning machine dynamically according to the compatibility risk level.

2. The method of using the intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, Data on the residual solvent concentration and surface contact angle of silicon wafers after hydrocarbon cleaning were obtained through an online monitoring system, including: After the hydrocarbon cleaning process is completed, Fourier transform infrared spectroscopy analysis is performed on the silicon wafer surface to obtain the residual concentration of surface solvent. Simultaneously, static contact angle measurements were performed on the silicon wafer surface by depositing ultrapure water droplets and recording their contours to obtain surface contact angle data. The Fourier transform infrared spectroscopy analysis and static contact angle measurement are completed within the same time window, ensuring the synchronization and consistency of data acquisition.

3. The method of using the intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, Based on surface solvent residual concentration and surface contact angle data, the synergistic effect anomaly of volatility and polarity characteristics in residual solvent components is identified, including: Volatility characteristic parameters were determined based on the residual concentration of surface solvent; Polarity characteristic parameters are determined based on surface contact angle data; Analyze the correlation between volatility characteristic parameters and polarity characteristic parameters; When the correlation is lower than the preset correlation threshold, it is determined to be an abnormal state of synergy effect.

4. The method of using the intelligent hydrocarbon cleaning machine according to claim 3, characterized in that, Determining volatility characteristic parameters based on surface solvent residual concentration includes: querying a pre-stored solvent property database, matching the surface solvent residual concentration with the volatility parameters of standard solvents in the database, thereby determining the volatility characteristic parameters; Determining polarity characteristic parameters based on surface contact angle data includes: processing the surface contact angle data using a surface free energy calculation model, and extracting polarity characteristic parameters from the polarity components of the calculation results.

5. The method of using the intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, When an abnormal state of synergistic effect is identified, the process parameters and historical contamination data of the corresponding silicon wafer in the previous manufacturing process are retrieved and correlated with the surface solvent residual concentration to identify the existence of cross-process contamination transmission chains, including: When an abnormal state of synergy effect is identified, the process parameters and historical contamination data of the corresponding silicon wafer in the previous manufacturing process are retrieved from the manufacturing execution system. A time-series correlation analysis was performed between the residual concentration of surface solvent and the process parameters of the preceding manufacturing process. The surface solvent residual concentration was compared with historical pollution data to ensure consistency of composition. When both time-series correlation analysis and component consistency comparison show significant correlation, it is determined that a cross-process contamination transmission chain exists.

6. The method of using an intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, When cross-process contamination transmission chains are identified, assess the competitive adsorption tendencies of solvent residues and key photoresist components on the silicon wafer surface, including: Key active components in solvent residues were identified based on cross-process contamination transmission chains. Obtain standard adsorption free energy data of the main components of photoresist on the silicon wafer surface; The competitive adsorption tendency is assessed by comparing the relative adsorption free energies of the key active components and the main components of the photoresist.

7. The method of using an intelligent hydrocarbon cleaning machine according to claim 6, characterized in that, The competitive adsorption tendency is assessed by comparing the relative adsorption free energies of key active components and main photoresist components. This includes: identifying key active components based on cross-process contamination transmission chains; obtaining standard adsorption free energy data for key active components and main photoresist components; and assessing competitive adsorption tendency by directly comparing adsorption free energy values.

8. The method of using an intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, Based on competitive adsorption tendencies, assess the compatibility risk level of the silicon wafer surface for photolithography processes, including: The basic level is determined based on the strength of the competitive adsorption tendency; The corresponding basic level is corrected by combining the persistence impact coefficient of pollutants in the cross-process pollution transmission chain; The corrected base level is output as the compatibility risk level.

9. The method of using an intelligent hydrocarbon cleaning machine according to claim 1, characterized in that, Based on the compatibility risk level, the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaner are dynamically adjusted, including: Based on the compatibility risk level, match the corresponding rinsing flow rate baseline value and drying temperature baseline value; A rinse flow rate parameter adjustment instruction is generated based on the rinse flow rate baseline value; Generate drying temperature parameter adjustment instructions based on the drying temperature reference value; The rinsing flow rate parameter adjustment command and the drying temperature parameter adjustment command are executed to dynamically control the hydrocarbon cleaning machine.

10. An intelligent hydrocarbon cleaning machine, used to implement the method of using the intelligent hydrocarbon cleaning machine according to any one of claims 1-9, characterized in that, Includes the following modules: The data acquisition module is used to acquire data on the residual concentration of surface solvent and surface contact angle of silicon wafers after hydrocarbon cleaning through an online monitoring system; Anomaly detection module is used to identify abnormal states of synergistic effects between volatile and polar characteristics in residual solvent components based on surface solvent residual concentration and surface contact angle data. An identification module is used to retrieve the process parameters and historical contamination data of the corresponding silicon wafer in the front-end manufacturing process when an abnormal state of synergistic effect is identified, and to perform correlation analysis with the surface solvent residual concentration to identify the existence of a cross-process contamination transmission chain. The tendency assessment module is used to evaluate the competitive adsorption tendency of solvent residues and photoresist main components on the silicon wafer surface when a cross-process contamination transmission chain is identified. The rating assessment module is used to assess the compatibility risk level of the silicon wafer surface for photolithography processes based on competitive adsorption tendencies. The parameter adjustment module is used to dynamically adjust the rinsing flow rate and drying temperature parameters of the hydrocarbon cleaner according to the compatibility risk level.

Images