Dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback

Abstract

The present application provides a dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback, which relates to the field of semiconductor manufacturing technology. The epitaxial layer of the wafer epitaxy process is determined by a surface density filling evaluation unit, and the information of the epitaxial layer is evaluated to determine the surface density filling of the epitaxial layer; the surface material information detection unit detects the surface material information of the wafer to be epitaxy; the surface treatment granularity optimization unit combines the surface density filling, and maps and matches the surface material information according to the surface treatment matching module to determine the surface treatment granularity of the wafer to be epitaxy; the surface treatment execution unit regulates the wafer surface treatment device to treat the surface of the epitaxial wafer according to the optimized surface treatment granularity. The present application solves the technical problem that the existing surface treatment process usually adopts static parameter settings, which is difficult to adapt to small changes in material properties, resulting in unstable treatment results, and achieves the technical effect of improving the quality and efficiency of wafer surface treatment.

Description

Technical Field

[0001] The present application relates to the field of semiconductor manufacturing technology, and in particular to a dynamic optimization system for pre-epitaxial wafer surface treatment based on multi-dimensional detection feedback. Background Art

[0002] Crystal epitaxy is one of the core technologies of modern semiconductor manufacturing. It changes the electronic properties of the wafer by growing a new material layer (epitaxial layer) on the surface of the wafer. Wafer surface treatment, as a pre-step of the epitaxy process, has a direct impact on the quality and performance of the epitaxial layer. The result of surface treatment will directly affect the growth quality and electronic performance of the epitaxial layer.

[0003] Most of the existing wafer surface treatment processes are based on static parameter control and lack the ability to detect and adjust the wafer surface status in real time. This static control mode is difficult to adapt to slight changes in wafer material properties, such as slight differences in wafer surfaces, fluctuations in environmental conditions, etc., resulting in insufficient precision in the control of surface flatness and roughness, and unable to meet the high requirements of advanced processes for surface quality. This limits the further improvement of epitaxial layer quality and becomes a bottleneck in the manufacture of high-performance devices. Summary of the invention

[0004] The present application provides a dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback, which solves the technical problem that the existing wafer surface treatment process usually adopts static parameter settings, which is difficult to adapt to slight changes in wafer material properties, resulting in unstable wafer surface treatment results, and achieves the technical effect of improving the quality and efficiency of wafer surface treatment.

[0005] In view of the above problems, the present application provides a dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback, the system comprising: a surface density filling degree evaluation unit, the surface density filling degree evaluation unit is used to determine the epitaxial layer of the wafer epitaxy process, and to evaluate the surface density filling degree of the epitaxial layer based on the information of the epitaxial layer, wherein the information of the epitaxial layer includes the thickness of the epitaxial layer, the doping concentration and the conductivity type; a surface material information detection unit, the surface material information detection unit is used to detect the surface material information of the wafer to be epitaxially treated, including the material particle structure, the material coating properties and the material flow adsorption; a surface treatment granularity optimization ... The surface treatment granularity optimization unit is used to combine the surface density filling degree, map and match the surface material information according to the surface treatment matching module, and determine the surface treatment granularity of the wafer to be epitaxial, wherein the surface treatment matching module includes a feedback network layer, and the feedback network layer performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and outputs the optimized surface treatment granularity, and the surface treatment granularity includes surface roughness and surface flatness; a surface treatment execution unit, and the surface treatment execution unit is used to control the wafer surface treatment device to process the surface of the wafer to be epitaxial according to the optimized surface treatment granularity.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The surface density filling degree evaluation unit determines the epitaxial layer of the wafer epitaxy process, and uses the information of the epitaxial layer for evaluation to determine the surface density filling degree of the epitaxial layer, wherein the information of the epitaxial layer includes the thickness of the epitaxial layer, the doping concentration and the conductivity type; by evaluating the surface density filling degree of the epitaxial layer, this unit can judge the quality and uniformity of the epitaxial layer, and provide a reference for subsequent surface treatment. The surface material information detection unit detects the surface material information of the wafer to be epitaxially grown, including the material particle structure, the material coating properties and the material flow adsorption; by detecting the surface material information of the wafer to be epitaxially grown, this unit determines the physical and chemical properties of the wafer surface, and provides data support for optimizing the surface treatment strategy. The surface treatment granularity optimization unit combines the surface density filling degree, maps and matches the surface material information according to the surface treatment matching module, and determines the surface treatment granularity of the wafer to be epitaxially grown, wherein the surface treatment matching module includes a feedback network layer, and the feedback network layer performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and outputs the optimized surface treatment granularity; this unit optimizes the surface treatment granularity in real time according to the multi-dimensional control parameters through the integrated feedback network layer, and determines the granularity of the wafer surface treatment to achieve the best treatment effect. The surface treatment execution unit controls the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity.

[0008] To summarize, this application uses multi-dimensional detection to comprehensively evaluate multiple parameters of the wafer surface, and integrates a feedback network layer to perform feedback optimization on the surface treatment granularity, dynamically adjust the wafer surface treatment process, reduce uncertainties and errors in the production process, and improve the quality and efficiency of wafer surface treatment.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of the structure of a dynamic optimization system for pre-epitaxial wafer surface treatment based on multi-dimensional detection feedback provided in an embodiment of the present application;

[0011] Figure 2 A schematic diagram of a process for constructing a surface treatment matching module in a dynamic optimization system for surface treatment of a wafer before epitaxy based on multi-dimensional detection feedback provided in an embodiment of the present application;

[0012] Figure 3 A schematic diagram of a flow chart for feedback optimization of surface treatment granularity in a dynamic optimization system for pre-epitaxial wafer surface treatment based on multi-dimensional detection feedback provided in an embodiment of the present application.

[0013] Description of reference numerals: surface density filling degree evaluation unit 10 , surface material information detection unit 20 , surface treatment particle size optimization unit 30 , surface treatment execution unit 40 . DETAILED DESCRIPTION

[0014] The embodiment of the present application solves the technical problem that the existing wafer surface treatment process usually adopts static parameter settings, which is difficult to adapt to slight changes in wafer material properties and leads to unstable wafer surface treatment results by providing a dynamic optimization system for pre-epitaxial wafer surface treatment based on multi-dimensional detection feedback, thereby achieving the technical effect of improving the quality and efficiency of wafer surface treatment.

[0015] like Figure 1 As shown, the embodiment of the present application provides a dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback, the system comprising:

[0016] The surface density filling degree evaluation unit 10 is used to determine the epitaxial layer of the wafer epitaxial process, and to evaluate the surface density filling degree of the epitaxial layer based on the information of the epitaxial layer, wherein the information of the epitaxial layer includes the epitaxial layer thickness, doping concentration and conductivity type.

[0017] Specifically, first determine the epitaxial layer of the wafer epitaxial process, collect relevant information of the epitaxial layer, including the thickness, doping concentration and conductivity type of the epitaxial layer, and use this information to determine the surface density filling degree of the epitaxial layer, and guide the optimization and adjustment of the target and direction for the surface treatment of the epitaxial wafer to be processed. Among them, the thickness of the epitaxial layer is the vertical thickness of the epitaxial layer on the surface of the wafer, which can be measured using a non-contact thickness meter. The doping concentration is the concentration of the dopant in the epitaxial layer, which can be measured by secondary ion mass spectrometry (SIMS) or photoluminescence (PL) technology. The conductivity type refers to the conductive properties of the epitaxial layer, such as n-type or p-type, which can be measured using a four-probe method or Hall effect. The surface density filling degree refers to the density of atoms or molecules in the epitaxial layer and their uniformity of distribution on the surface. The surface density filling degree evaluation unit 10 analyzes the relationship between the epitaxial layer information and the surface density filling degree through a pre-trained machine learning model based on the acquired epitaxial layer information, evaluates the surface density filling degree of the epitaxial layer, and then determines the physical properties of the epitaxial layer.

[0018] The surface material information detection unit 20 is used to detect the surface material information of the wafer to be epitaxially grown, including the material particle structure, the material coating properties and the material flow adsorption properties.

[0019] Specifically, the surface material information of the wafer to be epitaxially grown includes the material particle structure, material coating properties, and material flow adsorption. The material particle structure refers to the microstructure of the material constituting the surface of the wafer, including the size, shape, distribution of the particles, and the defects of the crystal. The material coating property refers to whether there is a covering layer on the surface of the surface material, and the properties of these covering layers, which affect the growth quality of the epitaxial layer. The material flow adsorption refers to the ability of the wafer surface material to adsorb external gases or chemicals, which affects the uniformity and quality of epitaxial growth.

[0020] The surface material information detection unit 20 uses a series of high-precision detection tools and methods to comprehensively detect and analyze the surface material characteristics of the wafer to be epitaxially grown, collect the surface material information of the wafer to be epitaxially grown, and provide data reference for subsequent surface treatment process optimization. For example, scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), etc. can be used to detect the material particle structure; X-ray photoelectron spectroscopy (XPS), secondary ion mass spectrometry (SIMS), etc. can be used to detect the material coating properties; dynamic adsorption isotherms, thermogravimetric analysis (TGA), contact angle measurement, etc. can be used to detect the material flow adsorption.

[0021] Through the precise analysis of the surface material information detection unit 20 , the surface material characteristics of the wafer before entering the epitaxial growth process can be determined, thereby facilitating targeted surface treatment process adjustments.

[0022] A surface treatment granularity optimization unit 30 is used to map and match the surface material information in combination with the surface density filling degree according to a surface treatment matching module to determine the surface treatment granularity of the wafer to be epitaxially grown, wherein the surface treatment matching module includes a feedback network layer, and the feedback network layer performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and outputs the optimized surface treatment granularity, wherein the surface treatment granularity includes surface roughness and surface flatness.

[0023] Specifically, surface treatment granularity refers to the degree of fineness of surface treatment, including surface roughness and surface flatness. The surface treatment matching module is a software component responsible for mapping and matching surface material information with the required surface treatment granularity. Its core is a feedback network layer, which is a network based on a mapping relationship. The feedback network layer receives multi-dimensional control parameters of the wafer surface treatment device, such as cleaning time, etching intensity, polishing pressure, etc. Based on these data, the surface treatment granularity is optimized and the optimized surface treatment granularity is output. Among them, the optimized surface treatment granularity refers to the optimal parameter setting for wafer surface treatment generated by the feedback network layer, including the specific values ​​of surface roughness and flatness.

[0024] The surface treatment granularity optimization unit 30 maps and matches the detected surface material information with the known surface treatment parameters through the surface treatment matching module to determine the treatment granularity that best suits the current wafer surface state. Through continuous feedback and optimization, the surface treatment granularity is ensured to match the characteristics of the wafer, thereby ensuring that the wafer surface reaches the best treatment state before epitaxial growth.

[0025] The surface treatment execution unit 40 is used to control the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity.

[0026] Specifically, the wafer surface treatment device is a device used to process the wafer surface and improve its roughness, flatness and other surface properties, such as chemical mechanical polishing (CMP) equipment, plasma etching equipment, etc. The surface treatment execution unit 40 receives the processing parameters from the surface treatment granularity optimization unit 30, converts these optimization parameters into control signals and sends them to the wafer surface treatment device, and the wafer surface treatment device starts to process the wafer surface according to the received control signals. For example, if the optimization parameters require reducing the surface roughness, the execution unit will adjust the pressure, grinding disc speed or grinding liquid flow of the chemical mechanical polishing (CMP) equipment.

[0027] Through the above steps, the surface treatment execution unit 40 ensures that the wafer surface reaches the preset optimization target, provides high-quality surface conditions for subsequent epitaxial growth, and thus improves the performance and manufacturing yield of semiconductor devices.

[0028] Further, such as Figure 2 As shown, the system of the embodiment of the present application constructs the surface treatment matching module, and the execution steps include:

[0029] A plurality of wafer samples are set, wherein each wafer sample is a different wafer sample with the same surface material information but different surface treatment particle size; a filling quality analysis is performed on the plurality of wafer samples according to the surface density filling degree to obtain a plurality of filling qualities, and a mapping training data set is generated according to a mapping relationship between the plurality of wafer samples and the plurality of filling qualities; a surface treatment matching model is trained according to the mapping training data set, and the surface treatment matching model is loaded into the surface treatment matching module.

[0030] Specifically, first, prepare multiple wafer samples with the same surface material information but different surface treatment particle sizes, and each sample represents a different surface treatment condition. Then, use the surface density filling degree to perform filling quality analysis on these samples to obtain the filling quality data of each sample. Map each wafer sample with its corresponding filling quality to generate a mapping training data set. Using the generated mapping training data set, train the surface treatment matching model through a machine learning algorithm (such as a neural network). The model predicts the optimal surface treatment particle size under given wafer surface material information by learning the correlation between surface treatment particle size and filling quality. The trained surface treatment matching model is loaded into the surface treatment matching module for real-time prediction and optimization of the surface treatment particle size of the wafer to be processed to ensure the optimal growth conditions of the epitaxial layer.

[0031] Through the above steps, the system builds a highly precise surface treatment matching module that can provide customized surface treatment solutions for different wafers, significantly improving the uniformity of epitaxial layer growth and device performance.

[0032] Furthermore, the system of the embodiment of the present application performs filling quality analysis on the plurality of wafer samples according to the surface density filling degree to obtain a plurality of filling qualities, and the execution steps include:

[0033] Among them, the filling quality analysis includes the deposition epitaxial rate, deposition surface uniformity, deposition surface defect rate and conductive performance of the epitaxial layer on the surface of the wafer to be epitaxially grown; training is performed based on the deposition epitaxial rate, the deposition surface uniformity, the conductive performance and identification information indicating the filling quality size, and a filling quality analysis model is output; based on the filling quality analysis model, multiple filling qualities are obtained.

[0034] Specifically, filling quality is a comprehensive quality indicator for measuring the growth of epitaxial layers on the surface of wafers, including deposition epitaxy rate, surface uniformity, surface defect rate, and electrical conductivity. The epitaxial layer growth process of multiple wafer samples is monitored to collect data such as deposition epitaxy rate, surface uniformity, surface defect rate, and electrical conductivity. Among them, the deposition epitaxy rate is the amount of material deposited on the epitaxial layer per unit time, reflecting the growth rate; the deposition surface uniformity is the uniformity of the thickness and composition of the epitaxial layer surface, which is used to measure the quality of epitaxial layer growth; the deposition surface defect rate is the distribution density of defects (such as voids and cracks) in the epitaxial layer, which affects the device performance; the electrical conductivity refers to the electrical properties of the epitaxial layer, such as resistivity and carrier mobility.

[0035] Generate identification information for the filling quality of each wafer sample. The identification information is calculated based on the experimental data and can intuitively reflect the filling quality of the wafer sample. For example, based on the comprehensive score calculated based on the above data, a filling quality grade can be assigned to each wafer sample, such as A (above 90 points), B (80-90 points), C (70-80 points), etc. Use the collected data and the corresponding identification information to train a filling quality analysis model, which can be built based on a support vector machine (SVM), a neural network, etc., with the growth parameters of the wafer sample as input and the filling quality grade as output. The model is trained to identify the complex relationship between the epitaxial layer growth parameters and the filling quality. The accuracy and generalization ability of the model are evaluated through cross-validation, and the algorithm parameters or feature selection are adjusted as needed to improve the model performance.

[0036] The trained filling quality analysis model is used to predict the filling quality of new wafer samples and guide the parameter optimization of the epitaxial layer growth process. By inputting the growth parameters of the new wafer sample, the filling quality analysis model will predict its possible filling quality level, thereby providing guidance for optimizing the epitaxial layer growth conditions, that is, the wafer surface treatment process, and improving the epitaxial layer quality.

[0037] Furthermore, the surface treatment particle size optimization unit 30 in the embodiment of the present application combines the surface density filling degree and maps and matches the surface material information according to the surface treatment matching module, and the execution steps include:

[0038] Obtain a surface treatment particle size threshold of the wafer surface treatment device; call a surface treatment matching model in the surface treatment matching module, and obtain, according to the surface material information, a surface treatment particle size that matches the surface density filling degree under the surface treatment particle size threshold, wherein the matched surface treatment particle size is a surface treatment particle size whose filling quality is greater than a preset threshold.

[0039] Specifically, according to the growth requirements of the epitaxial layer, the surface treatment particle size threshold of the wafer surface treatment device is determined, that is, the upper and lower limits of the surface treatment particle size, including the minimum or maximum surface roughness and flatness standards set to ensure the quality of epitaxial layer growth. The surface material information of the wafer to be epitaxially grown is input into the surface treatment matching model. Based on previous training, the model can predict the surface treatment particle size that meets the surface treatment particle size threshold under the surface material information. The surface treatment particle size with a filling quality greater than or equal to the preset filling quality threshold is screened out from the model results as the output of the optimization unit.

[0040] Exemplarily, for a specific wafer material (such as silicon), the upper limit of the surface roughness is set to 0.6nm, the lower limit is 0.2nm, and the preset filling quality threshold is 90%. The surface treatment matching model predicts that when the surface roughness is 0.5nm, the surface density filling degree is 88%, which does not reach the preset 90% threshold. When the surface roughness drops to 0.4nm, the filling degree increases to 92%, meeting the conditions and outputting the surface roughness of 0.4nm. When the requirements for epitaxial layer growth increase, the preset surface treatment granularity threshold is adjusted to an upper limit of 0.5nm and a lower limit of 0.3nm. At this point, the surface treatment matching model needs to be re-evaluated to find a surface treatment granularity whose filling quality is still greater than 90% under the new threshold.

[0041] Through the above steps, the surface treatment granularity optimization unit 30 can ensure that the wafer surface treatment not only meets the surface roughness and flatness requirements of the epitaxial layer growth, but also achieves an ideal surface density filling degree, creating conditions for high-quality epitaxial layer growth.

[0042] Further, such as Figure 3 As shown, the feedback network layer in the surface treatment granularity optimization unit 30 of the embodiment of the present application performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and the execution steps include:

[0043] Acquire the processing components of the wafer surface processing device; collect multidimensional control parameter samples of each component in the processing components, perform surface processing granularity influence association according to the abnormality of the multidimensional control parameter samples, and obtain the influence coefficient of each processing component; construct a feedback network layer based on the mapping relationship between the processing components and the influence coefficient; when the abnormality of the control parameter of the processing component is greater than the preset abnormality, activate the feedback network layer, and obtain the influence index corresponding to the abnormal processing component based on the mapping relationship; perform feedback optimization on the surface processing granularity according to the influence index corresponding to the abnormal processing component, and output the optimized surface processing granularity.

[0044] Specifically, first, identify the processing components in the wafer surface processing device, that is, the physical components in the wafer surface processing device that are responsible for specific processing tasks, such as polishing heads, grinding discs, etc. Collect multi-dimensional control parameter samples of each processing component regularly or in real time, including the rotation speed, pressure, chemical flow rate, temperature, etc. of the processing component. For the collected multi-dimensional control parameter samples, analyze their deviations from the standard values ​​and calculate the degree of abnormality. Then use statistical correlation analysis, such as the Pearson correlation coefficient, to evaluate the relationship between the degree of abnormality and the surface treatment particle size, and obtain the influence coefficient of each processing component. Among them, the degree of abnormality is an indicator to measure the deviation of the processing parameters from the standard or expected value, and the influence coefficient is a quantitative indicator of the influence of the control parameters of the processing component on the surface treatment particle size.

[0045] Based on the mapping relationship between the processing components and the influence coefficients, a neural network is used to construct a feedback network layer, which can learn and predict the influence of different control parameters on the surface treatment particle size. When it is detected that the abnormality of the control parameters of a processing component is greater than the preset threshold, the feedback network layer is activated. According to the influence index corresponding to the abnormal processing component, the feedback network layer adjusts the relevant control parameters to optimize the surface treatment particle size. For example, it is detected that the polishing head speed is 10% lower than the standard value, causing the surface treatment particle size to deviate from the target. The feedback network layer analysis found that increasing the grinding time by 10% can compensate for this influence and output the optimized surface treatment particle size parameters.

[0046] Through the above steps, the feedback network layer can ensure that the control parameters in the wafer surface treatment process are always in the best state, and can be quickly adjusted even when abnormalities occur, ensuring the stability and optimization of the surface treatment particle size, laying the foundation for high-quality epitaxial layer growth.

[0047] Furthermore, the feedback network layer in the embodiment of the present application is also connected to a wafer surface cleaning device, and the execution steps of the surface treatment granularity optimization unit 30 also include:

[0048] Acquire a cleaning fluid medium of the wafer surface cleaning device; perform an influence analysis on the material coating properties and material flow adsorption of the wafer to be epitaxially grown based on the influence of the cleaning fluid medium to determine a coating property-influence index and an adsorption property-influence index; update the surface material information according to the coating property-influence index and adsorption property-influence index; and redetermine the surface treatment particle size of the wafer to be epitaxially grown based on the updated surface material information.

[0049] Specifically, the wafer surface cleaning device is a device used to clean the wafer surface, remove impurities, and prepare for the growth of the epitaxial layer. The feedback network layer is connected to the wafer surface cleaning device to obtain the cleaning liquid medium of the wafer surface cleaning device, that is, a chemical solution used to clean the wafer surface, such as deionized water, acidic or alkaline solution. Use chemical analysis methods to evaluate the coating effect of the cleaning liquid medium on the wafer surface material, such as forming a protective layer or a chemical reaction layer, and test the adsorption of the cleaning liquid medium on the wafer surface through experiments, including adsorption speed and residual amount. Based on the above analysis, determine the degree of influence of the cleaning liquid medium on the material coating properties and flow adsorption, construct a mathematical model, associate the material coating properties, flow adsorption and surface treatment particle size, use the model to calculate the specific values ​​of the cleaning liquid medium on the material coating properties and flow adsorption, and obtain the coating property-influence index and adsorption-influence index. Integrate the coating property-influence index and adsorption-influence index into the surface material information, update the surface material information, and reflect the material state after the cleaning liquid medium acts. Using the updated surface material information, the most suitable surface treatment particle size is recalculated through the feedback network layer to ensure the quality of epitaxial layer growth.

[0050] Through the above steps, the feedback network layer can dynamically adjust the surface treatment parameters according to the influence of the wafer surface cleaning process, optimize the epitaxial layer growth conditions, and ensure the quality of the wafer surface treatment.

[0051] Furthermore, the surface treatment execution unit 40 in the embodiment of the present application controls the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity, and the execution steps include:

[0052] Among them, the wafer surface processing device includes a roughness detection unit and a flatness detection unit; when the wafer surface processing device processes the surface of the wafer to be epitaxially grown, the real-time roughness of the roughness detection unit and the real-time flatness of the flatness detection unit are obtained; the optimized surface processing granularity is obtained, and the optimized surface processing granularity includes the optimized surface roughness and the optimized surface flatness; when the real-time roughness meets the optimized surface roughness, and the real-time flatness meets the optimized surface flatness, a processing completion reminder message is generated.

[0053] Specifically, the wafer surface processing device is built with a roughness detection unit for real-time monitoring of the wafer surface roughness and a flatness detection unit for real-time detection of the wafer surface flatness. When the surface of the epitaxial wafer is processed, the surface roughness and surface flatness of the epitaxial wafer are obtained in real time through the roughness detection unit and the flatness detection unit to obtain the real-time roughness and real-time flatness.

[0054] The surface treatment execution unit 40 receives the optimized surface roughness and surface flatness target values ​​output by the feedback network layer. During the surface treatment process, the real-time roughness and real-time flatness monitored in real time are continuously compared with the optimized surface roughness and surface flatness target values. When the real-time roughness and real-time flatness meet the optimized target values ​​at the same time, the surface treatment execution unit 40 generates a processing completion reminder message, indicating that the wafer surface treatment has reached the expected standard.

[0055] Through the above steps, the surface treatment execution unit 40 can monitor the surface treatment process of the wafer to be epitaxially grown in real time, ensure that the optimization target is achieved accurately and efficiently, and provide high-quality surface conditions for subsequent epitaxial layer growth.

[0056] Furthermore, the material coating properties of the surface material information described in the embodiment of the present application are detected by X-ray photoelectron spectroscopy on the surface of the wafer to be epitaxially grown, and the signal energy is analyzed and obtained; the material flow adsorption properties of the surface material information are obtained by performing a solution test on the surface of the wafer to be epitaxially grown, and measuring the contact angle between the droplet and the surface.

[0057] Specifically, in a specific real-time method, X-ray photoelectron spectroscopy (XPS) is used to detect the coating properties of the material of the epitaxial wafer. The specific process includes: using an X-ray source to excite electrons on the surface of the wafer to generate photoelectrons, and measuring the kinetic energy of the photoelectrons through an energy analyzer. The kinetic energy of the photoelectrons is related to the binding energy, which depends on the chemical environment of the atoms. The type and chemical state of the surface elements are determined based on the binding energy of the photoelectrons, and whether there is a coating layer or a contamination layer on the surface is analyzed.

[0058] The flow adsorption of the material of the epitaxial wafer is determined by the contact angle measurement method. The specific process includes: dropping a certain volume of test solution on the surface of the epitaxial wafer, using a contact angle meter, and measuring the contact angle between the droplet and the surface through image recognition technology. The size of the contact angle reflects the force between the droplet and the solid surface, thereby indirectly evaluating the flow adsorption of the material. A small contact angle (acute angle) indicates that the droplet has a strong interaction with the surface, indicating that the material has good flow adsorption; on the contrary, a larger contact angle (obtuse angle) indicates a weak interaction and poor adsorption.

[0059] Through XPS and contact angle measurement, the chemical state and flow adsorption of the wafer surface can be accurately evaluated, providing optimization direction and goals for the wafer surface treatment process. When X-ray photoelectron spectroscopy (XPS) analysis shows the presence of a slight oxide layer, it may affect the adhesion and growth quality of the epitaxial material. In the contact angle measurement, if it is found that the contact angle between the droplet and the wafer surface is large (such as greater than 90 degrees), it indicates that the flow adsorption of the wafer surface is weak, and surface modification or pretreatment is required to improve the uniformity and quality of epitaxial growth.

[0060] In summary, the dynamic optimization system for pre-epitaxial wafer surface treatment based on multi-dimensional detection feedback provided by the embodiment of the present application has the following technical effects:

[0061] The surface density filling degree evaluation unit 10 determines the epitaxial layer of the wafer epitaxy process, and uses the information of the epitaxial layer for evaluation to determine the surface density filling degree of the epitaxial layer, wherein the information of the epitaxial layer includes the thickness of the epitaxial layer, the doping concentration and the conductivity type; this unit can judge the quality and uniformity of the epitaxial layer by evaluating the surface density filling degree of the epitaxial layer, and provide a reference for subsequent surface treatment. The surface material information detection unit 20 detects the surface material information of the wafer to be epitaxially grown, including the material particle structure, the material coating properties and the material flow adsorption; this unit determines the physical and chemical properties of the wafer surface by detecting the surface material information of the wafer to be epitaxially grown, and provides data support for optimizing the surface treatment strategy.

[0062] By performing filling quality analysis on multiple wafer samples, training the surface treatment matching model, and constructing the surface treatment matching module, the surface treatment granularity optimization unit 30 combines the surface density filling degree, maps and matches the surface material information according to the surface treatment matching module, and determines the surface treatment granularity of the wafer to be epitaxially grown. The surface treatment matching module includes a feedback network layer, which performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and outputs the optimized surface treatment granularity. The feedback network layer is also connected to the wafer surface cleaning device, and according to the type of the cleaning liquid medium of the wafer surface cleaning device, the material's coating properties and flow adsorption are quantitatively analyzed, and the surface material information is dynamically adjusted. This unit integrates the feedback network layer, optimizes the surface treatment granularity in real time according to the multi-dimensional control parameters, and determines the granularity of the wafer surface treatment to achieve the best treatment effect. The surface treatment execution unit 40 regulates the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity.

[0063] In general, the embodiments of the present application comprehensively evaluate multiple parameters of the wafer surface through multi-dimensional detection, and integrate the feedback network layer to perform feedback optimization on the surface treatment granularity, dynamically adjust the wafer surface treatment process, reduce the uncertainty and error in the production process, and improve the quality and efficiency of wafer surface treatment. The above description of the disclosed embodiments enables professionals and technicians in this field to implement or use the present application. Various modifications to these embodiments will be obvious to professionals and technicians in this field, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features disclosed herein.

Claims

1. A dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback, characterized in that: The system comprises: A surface density filling degree evaluation unit, wherein the surface density filling degree evaluation unit is used to determine the epitaxial layer of the wafer epitaxial process, and to evaluate the surface density filling degree of the epitaxial layer based on the information of the epitaxial layer, wherein the information of the epitaxial layer includes the thickness, doping concentration and conductivity type of the epitaxial layer; A surface material information detection unit, which is used to detect surface material information of the wafer to be epitaxially grown, including material particle structure, material coating properties, and material flow adsorption; A surface treatment granularity optimization unit, the surface treatment granularity optimization unit is used to map and match the surface material information according to the surface treatment matching module in combination with the surface density filling degree, and determine the surface treatment granularity of the wafer to be epitaxially grown, wherein the surface treatment matching module includes a feedback network layer, and the feedback network layer performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and outputs the optimized surface treatment granularity, and the surface treatment granularity includes surface roughness and surface flatness; A surface treatment execution unit is used to control the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity.

2. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 1, characterized in that: The system constructs the surface treatment matching module, and the execution steps include: Setting a plurality of wafer samples, wherein each wafer sample is a different wafer sample having the same surface material information but different surface treatment granularity; Performing filling quality analysis on the plurality of wafer samples according to the surface density filling degree to obtain a plurality of filling qualities, and generating a mapping training data set according to a mapping relationship between the plurality of wafer samples and the plurality of filling qualities; A surface treatment matching model is trained according to the mapping training data set, and the surface treatment matching model is loaded into the surface treatment matching module.

3. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 2, characterized in that: The system performs filling quality analysis on the plurality of wafer samples according to the surface density filling degree, obtains a plurality of filling qualities, and executes the steps include: The filling quality analysis includes the deposition epitaxial rate, deposition surface uniformity, deposition surface defect rate and conductive performance of the epitaxial layer on the surface of the wafer to be epitaxially grown; Training is performed according to the deposition epitaxy rate, the deposition surface uniformity, the conductive property, and the identification information identifying the filling quality size, and outputting a filling quality analysis model; According to the filling quality analysis model, a plurality of filling qualities are obtained.

4. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 2, characterized in that: The surface treatment particle size optimization unit combines the surface density filling degree and maps and matches the surface material information according to the surface treatment matching module, and the execution steps include: Obtaining a surface treatment granularity threshold of the wafer surface treatment device; The surface treatment matching model in the surface treatment matching module is called to obtain, according to the surface material information, a surface treatment particle size that matches the surface density filling degree under the surface treatment particle size threshold, wherein the matched surface treatment particle size is a surface treatment particle size whose filling mass is greater than a preset threshold.

5. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 1, characterized in that: The feedback network layer in the surface treatment granularity optimization unit performs feedback optimization on the surface treatment granularity according to the multi-dimensional control parameters of the wafer surface treatment device, and the execution steps include: Acquire a processing component of the wafer surface processing device; Collecting multidimensional control parameter samples of each component in the processing component, performing surface treatment particle size influence correlation according to the abnormality of the multidimensional control parameter samples, and obtaining the influence coefficient of each processing component; Based on the mapping relationship between the processing components and the influence coefficients, construct a feedback network layer; When the abnormality degree of the control parameter of the processing component is greater than a preset abnormality degree, the feedback network layer is activated, and the influence index corresponding to the abnormal processing component is obtained based on the mapping relationship; Feedback optimization is performed on the surface treatment particle size according to the influencing index corresponding to the abnormal processing component, and the optimized surface treatment particle size is output.

6. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 5, characterized in that: The feedback network layer is also connected to the wafer surface cleaning device, and the execution steps of the surface treatment granularity optimization unit also include: Obtaining a cleaning liquid medium of the wafer surface cleaning device; Analyze the influence of the cleaning liquid medium on the coating properties of the material of the epitaxial wafer and the material flow adsorption, and determine the coating property-influence index and the adsorption-influence index; Update the surface material information according to the coating property-influence index and the adsorption-influence index; The surface treatment granularity of the wafer to be epitaxially grown is re-determined based on the updated surface material information.

7. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 1, characterized in that: The surface treatment execution unit controls the wafer surface treatment device to process the surface of the wafer to be epitaxially grown according to the optimized surface treatment granularity, and executes the steps include: Wherein, the wafer surface processing device includes a roughness detection unit and a flatness detection unit; When the wafer surface processing device processes the surface of the wafer to be epitaxially grown, the real-time roughness of the roughness detection unit and the real-time flatness of the flatness detection unit are acquired; Obtaining an optimized surface treatment particle size, where the optimized surface treatment particle size includes an optimized surface roughness and an optimized surface flatness; When the real-time roughness satisfies the optimized surface roughness, and the real-time flatness satisfies the optimized surface flatness, a processing completion reminder message is generated.

8. The dynamic optimization system for wafer surface treatment before epitaxy based on multi-dimensional detection feedback as claimed in claim 1, characterized in that: The material coating property of the surface material information is detected by X-ray photoelectron spectroscopy on the surface of the wafer to be epitaxially grown, and the signal energy is obtained by analysis; The material flow adsorption of the surface material information is obtained by performing a solution test on the surface of the wafer to be epitaxially grown and measuring the contact angle between the droplet and the surface.

Images