Coating control method and system based on data processing

Abstract

The application provides a coating control method and system based on data processing, and relates to the technical field of coating control.The method comprises the following steps: obtaining coating information of an infrared high-reflection film without stress elimination; determining a plurality of first stress elimination coating schemes based on the coating information of the infrared high-reflection film without stress elimination, wherein the first stress elimination coating scheme comprises doping atomic concentration distribution information of the infrared high-reflection film and stress compensation layer thickness information; coating a plurality of original single crystal silicon substrates based on the plurality of first stress elimination coating schemes respectively, and obtaining single crystal silicon substrate information of each coated substrate; determining a target stress elimination coating scheme based on the single crystal silicon substrate information of each coated substrate; and coating the original single crystal silicon substrate with an infrared high-reflection film based on the target stress elimination coating scheme.The method can efficiently and accurately analyze and eliminate the stress of the infrared high-reflection film.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coating control, in particular to a coating control method and system based on data processing. BACKGROUND

[0002] As the core substrate material of precision optical devices such as high-power lasers and infrared optical sensors, the performance of the infrared high-reflection film deposited on the surface of the single crystal silicon substrate directly affects the optical stability and service life of the device. In the field of semiconductor optoelectronics, stress relief of the infrared high-reflection film is a key link to ensure the optical uniformity and long-term reliability of the single crystal silicon substrate. After the infrared high-reflection film is deposited, the existence of a prominent stress region in the film layer will cause stress imbalance, which not only leads to excessive reflectivity fluctuations, but also may cause fatal defects such as film cracking and peeling. Traditional stress relief mainly relies on empirical formulas or single parameter optimization, which has significant technical bottlenecks and defects. The micro-defect differences of different batches of single crystal silicon substrates and the drift of the coating parameters make the effect of the same stress relief scheme fluctuate significantly, which will significantly increase the yield difference between batches. In addition, the traditional method also relies too much on manual adjustment of process parameters, which has a long adjustment period and is easily affected by subjective factors, making it difficult to meet the requirements of efficiency and consistency in large-scale production.

[0003] Therefore, how to efficiently and accurately analyze and eliminate the stress of the infrared high-reflection film is a problem to be solved. SUMMARY

[0004] The technical problem solved by the present application is how to efficiently and accurately analyze and eliminate the stress of the infrared high-reflection film.

[0005] According to a first aspect, the present application provides a coating control method based on data processing, comprising: obtaining coating information of an infrared high-reflection film without stress relief; determining a plurality of first stress relief coating schemes based on the coating information of the infrared high-reflection film without stress relief, the first stress relief coating scheme including doping atomic concentration distribution information of the infrared high-reflection film and stress compensation layer thickness information; coating a plurality of original single crystal silicon substrates based on the plurality of first stress relief coating schemes respectively, and obtaining single crystal silicon substrate information after each coating, the single crystal silicon substrate information after each coating including three-dimensional point cloud data of the single crystal silicon substrate and infrared reflection information of the single crystal silicon substrate after coating under the first stress relief scheme; determining a target stress relief coating scheme based on the single crystal silicon substrate information after each coating; and coating the original single crystal silicon substrate with the infrared high-reflection film based on the target stress relief coating scheme.

[0006] In one possible implementation, the determining of a plurality of first stress-relieving coating schemes based on the coating information of the infrared high-reflective film that has not relieved stress includes: determining a plurality of protruding stress point information based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating; determining a plurality of protruding stress areas based on the plurality of protruding stress point information; determining a preferred range of doping atom concentration and a preferred range of stress compensation layer thickness for each protruding stress area based on the plurality of protruding stress areas; and determining a plurality of first stress-relieving coating schemes based on the preferred range of doping atom concentration and a preferred range of stress compensation layer thickness for each protruding stress area.

[0007] In one possible implementation, determining a target stress relief coating scheme based on the information of each single crystal silicon substrate after coating includes: constructing a coating map, the coating map including multiple nodes and multiple edges between the multiple nodes, the node characteristics of each node including a stress relief coating scheme, a single crystal silicon substrate information after coating, and the edges between the nodes representing the difference information of the doping atom concentration distribution information and the difference information of the stress compensation layer thickness between the stress relief coating schemes; and processing the coating map based on a neural network model to determine the target stress relief coating scheme.

[0008] In a possible implementation, the coating information of the infrared high-reflection film without stress relief includes the coating parameters of the infrared high-reflection film, the three-dimensional point cloud data of the single crystal silicon substrate after the original coating, and the infrared reflection information of the single crystal silicon substrate after the original coating.

[0009] According to a second aspect, the present invention provides a coating control system based on data processing, comprising: an acquisition module for acquiring coating information of an infrared high-reflection film that has not eliminated stress; a generation module for determining a plurality of first stress-relieving coating schemes based on the coating information of the infrared high-reflection film that has not eliminated stress, the first stress-relieving coating schemes including doping atom concentration distribution information and stress compensation layer thickness information of the infrared high-reflection film; a stress-relieving module for coating a plurality of original single-crystal silicon substrates respectively based on the plurality of first stress-relieving coating schemes, and acquiring information of each single-crystal silicon substrate after coating, the information of each single-crystal silicon substrate after coating including three-dimensional point cloud data of the single-crystal silicon substrate and infrared reflection information of the single-crystal silicon substrate after coating with the first stress-relieving scheme; a scheme determination module for determining a target stress-relieving coating scheme based on the information of each single-crystal silicon substrate after coating; and a coating control module for coating the original single-crystal silicon substrate with an infrared high-reflection film based on the target stress-relieving coating scheme.

[0010] In one possible implementation, the generation module is also used to: determine multiple protruding stress point information based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating; determine multiple protruding stress areas based on the multiple protruding stress point information; determine the preferred range of doping atom concentration and the preferred range of stress compensation layer thickness for each protruding stress area based on the multiple protruding stress areas; determine multiple first stress relief coating schemes based on the preferred range of doping atom concentration and the preferred range of stress compensation layer thickness for each protruding stress area.

[0011] In one possible implementation, the scheme determination module is also used to: construct a coating map, the coating map includes multiple nodes and multiple edges between the multiple nodes, the node characteristics of each node include a stress relief coating scheme, a single crystal silicon substrate information after coating, and the edges between the nodes represent the difference information of the doping atom concentration distribution information and the difference information of the stress compensation layer thickness between the stress relief coating schemes; the coating map is processed based on the neural network model to determine the target stress relief coating scheme.

[0012] In a possible implementation, the coating information of the infrared high-reflection film without stress relief includes the coating parameters of the infrared high-reflection film, the three-dimensional point cloud data of the single crystal silicon substrate after the original coating, and the infrared reflection information of the single crystal silicon substrate after the original coating.

[0013] According to a third aspect, an embodiment of the present invention provides an electronic device, comprising: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method as described above, the method comprising: obtaining coating information of an infrared high-reflection film that has not relieved stress; determining multiple first stress-relieving coating schemes based on the coating information of the infrared high-reflection film that has not relieved stress, the first stress-relieving coating schemes including doping atom concentration distribution information and stress compensation layer thickness information of the infrared high-reflection film; coating multiple original single-crystal silicon substrates based on the multiple first stress-relieving coating schemes, and obtaining information of each single-crystal silicon substrate after coating, the information of each single-crystal silicon substrate after coating including three-dimensional point cloud data of the single-crystal silicon substrate and infrared reflection information of the single-crystal silicon substrate after coating with the first stress-relieving scheme; determining a target stress-relieving coating scheme based on the information of each single-crystal silicon substrate after coating; and coating the original single-crystal silicon substrate with an infrared high-reflection film based on the target stress-relieving coating scheme.

[0014] According to the fourth aspect, this embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned data processing-based coating control method, the method comprising: obtaining coating information of an infrared high-reflection film without stress relief; determining a plurality of first stress relief coating schemes based on the coating information of the infrared high-reflection film without stress relief, the first stress relief coating schemes including doping atom concentration distribution information and stress compensation layer thickness information of the infrared high-reflection film; coating a plurality of original single-crystal silicon substrates respectively based on the plurality of first stress relief coating schemes, and obtaining information of each single-crystal silicon substrate after coating, the information of each single-crystal silicon substrate after coating including three-dimensional point cloud data of the single-crystal silicon substrate and infrared reflection information of the single-crystal silicon substrate after coating with the first stress relief scheme; determining a target stress relief coating scheme based on the information of each single-crystal silicon substrate after coating; and coating the original single-crystal silicon substrate with an infrared high-reflection film based on the target stress relief coating scheme.

[0015] The present invention provides a coating control method and system based on data processing, which includes: obtaining coating information of an infrared high-reflection film that has not eliminated stress; determining multiple first stress-eliminating coating schemes based on the coating information of the infrared high-reflection film that has not eliminated stress, the first stress-eliminating coating schemes including doping atom concentration distribution information and stress compensation layer thickness information of the infrared high-reflection film; coating multiple original single-crystal silicon substrates based on the multiple first stress-eliminating coating schemes, and obtaining information of each single-crystal silicon substrate after coating, the information of each single-crystal silicon substrate after coating including three-dimensional point cloud data of the single-crystal silicon substrate and infrared reflection information of the single-crystal silicon substrate after coating with the first stress-eliminating scheme; determining a target stress-eliminating coating scheme based on the information of each single-crystal silicon substrate after coating; and coating the original single-crystal silicon substrate with an infrared high-reflection film based on the target stress-eliminating coating scheme. This method can efficiently and accurately analyze and eliminate stress of the infrared high-reflection film. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of a coating control method based on data processing provided by an embodiment of the present invention;

[0017] Figure 2 is a schematic diagram of a single crystal silicon substrate according to an embodiment of the present invention;

[0018] Figure 3 Schematic diagram of an infrared high-reflective film according to an embodiment of the present invention;

[0019] Figure 4 A schematic diagram of a process for determining multiple first stress relief coating solutions provided by an embodiment of the present invention;

[0020] Figure 5A schematic diagram of a process for determining a target stress relief coating solution provided by an embodiment of the present invention;

[0021] Figure 6 A schematic diagram of a coating map constructed according to an embodiment of the present invention;

[0022] Figure 7 A schematic diagram of a coating control system based on data processing provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0023] The present invention will be further described in detail below by means of specific embodiments in conjunction with the accompanying drawings. Similar elements in different embodiments are numbered with associated similar elements. In the following embodiments, many detailed descriptions are provided to enable the present invention to be better understood. However, those skilled in the art will readily appreciate that some of the features may be omitted under different circumstances, or may be replaced by other elements, materials, or methods. In some cases, some operations related to the present invention are not shown or described in the specification. This is to avoid the core of the present invention being overwhelmed by excessive descriptions, and for those skilled in the art, it is not necessary to describe these related operations in detail. They can fully understand the related operations based on the description in the specification and the general technical knowledge in the art.

[0024] In an embodiment of the present invention, there is provided Figure 1 A coating control method based on data processing is shown, and the coating control method based on data processing includes steps S1 to S5:

[0025] Step S1, obtaining the coating information of the infrared high-reflection film that has not been stress-relieved.

[0026] In some embodiments, the coating is an infrared high-reflective coating on a single crystal silicon substrate.

[0027] Single crystal silicon substrate is the basic carrier in the semiconductor field. It is made of high-purity single crystal silicon as raw material, and then made through cutting, grinding and other processes, and has an ordered crystal structure. Figure 2 Schematic diagram of a single crystal silicon substrate in an embodiment of the present invention.

[0028] Infrared high-reflective film coating refers to the preparation of a multi-layer film system on the surface of a single-crystal silicon substrate through a physical or chemical deposition process. This process uses the interference effect of light to achieve high reflectivity for infrared light. Figure 3 Schematic diagram of an infrared high-reflection film in an embodiment of the present invention.

[0029] Unstressed infrared high-reflection coating information refers to the state and information of a monocrystalline silicon substrate obtained after the coating is applied to the original monocrystalline silicon substrate using a coating process that does not include stress relief. Unstressed infrared high-reflection coating information includes the coating parameters, the 3D point cloud data of the original monocrystalline silicon substrate after coating, and the infrared reflectance information of the original monocrystalline silicon substrate after coating.

[0030] The coating parameters of infrared high-reflection film are the process parameters set during the coating process. The coating parameters include deposition temperature, film thickness, and the air pressure inside the coating equipment.

[0031] The three-dimensional point cloud data of the original coated single-crystal silicon substrate is a set of discrete point clouds of spatial coordinates used to characterize the morphology of the surface of the single-crystal silicon substrate obtained through three-dimensional scanning technology.

[0032] The infrared reflection data of a monocrystalline silicon substrate after coating is obtained by real-time acquisition of the optical properties of different regions of the film using sensors and optical detection devices. This information reflects the infrared reflection behavior of each location on the film. This infrared reflection information includes data such as the reflectivity and reflection spectrum of the substrate in the infrared band after coating.

[0033] Step S2: determining a plurality of first stress relief coating schemes based on the coating information of the infrared high reflective film without stress relief, wherein the first stress relief coating schemes include doping atom concentration distribution information and stress compensation layer thickness information of the infrared high reflective film.

[0034] In some embodiments, Figure 4 A schematic diagram of a process for determining a plurality of first stress relief coating solutions provided by an embodiment of the present invention, wherein the process of determining a plurality of first stress relief coating solutions includes steps S21 to S24:

[0035] Step S21 : determining information of a plurality of protruding stress points based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating.

[0036] In some embodiments, a stress point analysis model can be used to determine multiple protruding stress point information based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating. The stress point analysis model is a convolutional neural network model. The input of the stress point analysis model is the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating, and the output of the stress point analysis model is multiple protruding stress point information.

[0037] Convolutional neural network models include convolutional neural networks (CNNs), which are deep learning models that can process data with grid-like topologies. Convolutional neural networks consist of convolutional layers, pooling layers, and fully connected layers. Convolutional layers interact with local regions of the input data through convolution kernels, extracting features such as edges, shapes, and textures. Pooling layers reduce the dimensionality of data while retaining important features, and fully connected layers integrate the extracted features.

[0038] Prominent stress point information is generated by the stress point analysis model and is the output of discrete points on the surface of a single-crystal silicon substrate where the stress values ​​differ significantly from those of the surrounding areas. This information includes the spatial location and stress numerical characteristics of the prominent stress point.

[0039] The spatial position is the specific coordinate position (x, y, z) of the physical location of the stress anomaly point on the substrate surface determined by three-dimensional point cloud data.

[0040] The stress numerical characteristics include the stress magnitude and stress type of the prominent stress point.

[0041] Stress types include tensile stress and compressive stress.

[0042] For example, 3D point cloud data from a certain area of ​​a coated monocrystalline silicon substrate reveals a localized surface protrusion (e.g., at coordinates (10, 20, 0.5) μm). Simultaneously, the infrared reflectance spectrum at this location exhibits a 15% drop in reflectivity in the 8-12 μm band (with reflectivity fluctuations in the surrounding area ≤5%). Using a stress point analysis model, the calculated stress value at this point is determined to be 200 MPa, while the average stress value of the five surrounding points is 50 MPa. This difference in stress values ​​exceeds the normal fluctuation range of the material's stress distribution, thus determining this point as a prominent stress point.

[0043] The three-dimensional point cloud data of the coated monocrystalline silicon substrate records the geometric deformations of the convex or concave surfaces of the monocrystalline silicon substrate after coating, and these deformations are directly related to the stress of the film layer. The infrared reflection information of the coated monocrystalline silicon substrate can also reflect the degree of lattice distortion of the material, and the infrared absorption rate of the stress concentration area will show a characteristic shift. The combination of the two can accurately locate the location of stress anomalies. Convolutional neural networks have powerful spatial feature extraction capabilities. Convolutional neural networks can capture local features and structural changes in the surface morphology from the three-dimensional point cloud data of the coated monocrystalline silicon substrate. At the same time, they can also extract optical feature data related to stress from the spectral data of the infrared reflection information. By learning and analyzing this data, the convolutional neural network can identify points where the stress values ​​are significantly different from those of the surrounding areas, thereby determining the information of multiple prominent stress points.

[0044] Step S22: determining a plurality of protruding stress areas based on the plurality of protruding stress point information.

[0045] In some embodiments, determining a plurality of protruding stress areas based on the plurality of protruding stress point information includes steps S31 to S33:

[0046] Step S31, determining the difference significance feature level of each prominent stress point, the position key index of each prominent stress point, and the degree of association between each prominent stress point and adjacent prominent stress points based on the information of the plurality of prominent stress points;

[0047] In some embodiments, a deep neural network can be used to determine the difference significance feature level of each protruding stress point, the position critical index of each protruding stress point, and the degree of association between each protruding stress point and adjacent protruding stress points based on the information of the multiple protruding stress points.

[0048] Deep neural network models include deep neural networks, which are composed of multiple layers of neurons. Through multi-layered nonlinear transformations, deep neural networks can learn complex input-output mappings. Deep neural networks can process high-dimensional data and possess powerful feature representation and learning capabilities. They can extract deep features from complex input data.

[0049] The difference significance level of a prominent stress point is an indicator that uses a deep neural network to quantify the difference in stress values ​​between a prominent stress point and the surrounding area and classify its significance. The difference significance level can reflect the degree of stress anomaly.

[0050] The location criticality index (PCI) of a prominent stress point is output by a deep neural network and measures the importance of the location of a prominent stress point on the substrate surface to the overall stress distribution. For example, a prominent stress point in the center of the substrate may have a higher PCI than a prominent stress point in the edge of the substrate, as it has a greater impact on overall flatness.

[0051] The degree of correlation between a prominent stress point and its adjacent protruding stress points is output by a deep neural network and represents the correlation between the spatial distribution and stress characteristics of these two protruding stress points. For example, if the distance between two adjacent protruding stress points is less than 50 μm and the stress difference is less than 10%, these two protruding stress points are likely to belong to the same stress concentration area, and the degree of correlation between the two protruding stress points is high.

[0052] Deep neural networks automatically extract deep features from high-dimensional input data through multiple layers of nonlinear transformations. They process the relationships between adjacent prominent stress points and calculate spatial correlations. Convolutional layers extract the spatial distribution of stress points, thereby identifying key locations. Attention mechanisms weight stress value differences and quantify their significance.

[0053] Step S32: determining a plurality of high-importance prominent stress points, a plurality of medium-importance prominent stress points, and a plurality of general-importance prominent stress points based on the difference significance feature level of each prominent stress point, the position criticality index of each prominent stress point, and the degree of association between each prominent stress point and adjacent prominent stress points;

[0054] In some embodiments, a deep neural network can be used to determine multiple high-importance protruding stress points, multiple medium-importance protruding stress points, and multiple general-importance protruding stress points based on the difference significance feature level of each protruding stress point, the position criticality index of each protruding stress point, and the degree of association between each protruding stress point and adjacent protruding stress points.

[0055] High-importance prominent stress points are stress anomaly points determined by deep neural networks that have a significant and decisive influence on the stress distribution and optical properties of the infrared high-reflection film on the single-crystal silicon substrate. High-importance prominent stress points can be located in key functional areas of the substrate (such as the optical center and mechanical support boundary). The stress anomalies caused by high-importance prominent stress points may lead to failure of the film structure or a significant decrease in infrared reflectivity.

[0056] Moderately important prominent stress points are stress anomaly points that have a moderate impact on the stress distribution and optical properties of the infrared high-reflection film on the single-crystal silicon substrate, as determined by deep neural networks. Moderately important prominent stress points are located in the secondary critical functional areas of the substrate, such as the edge of the optical area and the non-main support area.

[0057] Prominent stress points of general importance are stress anomalies that have a slight impact on the stress distribution and optical properties of the infrared high-reflection film of the single-crystal silicon substrate, as determined by deep neural networks. Prominent stress points of general importance are located in non-critical functional areas of the substrate.

[0058] Deep neural networks can automatically learn the complex weight relationships between the difference significance feature levels, position key indexes, and correlation degrees, and then transform these index combinations into accurate salient stress point importance grading through nonlinear calculations.

[0059] Step S33, determining a plurality of prominent stress regions based on the plurality of prominent stress point information, the plurality of high-importance prominent stress points, the plurality of medium-importance prominent stress points, and the plurality of general-importance prominent stress points.

[0060] In some embodiments, the plurality of prominent stress regions can be determined based on the plurality of prominent stress point information, the plurality of high-importance prominent stress points, the plurality of medium-importance prominent stress points, and the plurality of general-importance prominent stress points by a deep neural network.

[0061] A prominent stress region is an abnormal region with obvious stress abnormal features and consistent influence determined by a deep neural network. The prominent stress region is a specific physical unit that causes stress imbalance of the infrared high-reflection film of the single crystal silicon substrate.

[0062] The deep neural network can learn the spatial distribution regularity and importance level correlation mode of the prominent stress points. The deep neural network can take the high-importance prominent stress points as the core of the region, and then automatically cluster the position proximity and stress feature similarity of the medium-importance prominent stress points and the general-importance prominent stress points. The model can identify the physical distribution mode of the high-importance prominent stress points and the surrounding associated stress points, such as when a plurality of medium-importance prominent stress points exist around the high-importance prominent stress points and the spatial distance is less than a preset threshold, the high-importance prominent stress points and the surrounding associated stress points can be divided into the same prominent stress region.

[0063] Step S23, determining the preferred interval of the doping atom concentration and the preferred interval of the stress compensation layer thickness of each prominent stress region based on the plurality of prominent stress regions.

[0064] In some embodiments, the preferred interval of the doping atom concentration and the preferred interval of the stress compensation layer thickness of each prominent stress region can be determined based on the plurality of prominent stress regions using a preferred model, the preferred model being a convolutional neural network model, the input of the preferred model being the plurality of prominent stress regions, and the output of the preferred model being the preferred interval of the doping atom concentration and the preferred interval of the stress compensation layer thickness of each prominent stress region.

[0065] The doping atom refers to introducing controllable concentration of impurity atoms (such as boron, phosphorus, etc.) into the prominent stress region during the process of coating the infrared high-reflection film on the single crystal silicon substrate, and using the atomic radius difference to generate local stress to offset the inherent stress generated during the deposition of the high-reflection film. For example, when the prominent stress region exhibits tensile stress concentration, a larger radius of the doping atom is needed to generate compressive stress for balance.

[0066] The unit of the concentration of the doping atoms is atoms per cubic centimeter (atoms / cm³). The concentration of the doping atoms is too low to effectively compensate for the stress, and is too high to cause lattice distortion.

[0067] The preferred range of the concentration of the doping atoms is a reasonable range of the concentration of the doping atoms that can effectively improve the stress state for each of the prominent stress regions output by the optimization model.

[0068] The stress compensation layer is an additional thin film layer deposited on the surface of the substrate. By designing the material properties and thickness thereof, a stress field opposite to the original stress direction can be generated. The thickness of the stress compensation layer is in nanometers (nm).

[0069] The preferred range of the thickness of the stress compensation layer is a reasonable range of the thickness of the stress compensation layer that can effectively compensate for the stress for each of the prominent stress regions output by the optimization model.

[0070] The plurality of prominent stress regions includes information such as the spatial position, range, and stress distribution characteristics of each region. Different stress distribution characteristics require different concentrations of doping atoms and thicknesses of stress compensation layers for adjustment and compensation. The stress characteristics of these prominent stress regions provide the basis for determining the preferred range. The model can extract features related to the concentration of the doping atoms and the thickness of the stress compensation layer from the spatial characteristics and the stress distribution characteristics of the plurality of prominent stress regions, and then determine the appropriate concentration of the doping atoms and the preferred range of the thickness of the stress compensation layer for each prominent stress region by learning the mapping relationship between these features and effective stress adjustment.

[0071] In step S24, a plurality of first stress elimination coating schemes are determined based on the preferred range of the concentration of the doping atoms and the preferred range of the thickness of the stress compensation layer for each of the prominent stress regions.

[0072] In some embodiments, the plurality of first stress elimination coating schemes can be determined based on the preferred range of the concentration of the doping atoms and the preferred range of the thickness of the stress compensation layer for each of the prominent stress regions using a scheme adaptation model, the scheme adaptation model being a deep neural network model, the input of the scheme adaptation model being the preferred range of the concentration of the doping atoms and the preferred range of the thickness of the stress compensation layer for each of the prominent stress regions, and the output of the scheme adaptation model being the plurality of first stress elimination coating schemes.

[0073] The plurality of first stress elimination coating schemes is a set of differentiated feasible schemes output by the scheme adaptation model. Each first stress elimination coating scheme includes information about the concentration distribution of the doping atoms of the infrared high-reflection film and the thickness of the stress compensation layer.

[0074] The preferred interval of the doping atom concentration of each prominent stress region and the preferred interval of the stress compensation layer thickness provide a range constraint of key parameters for the design of the film coating scheme. These preferred intervals reflect the requirements of different regions for stress adjustment parameters. By synthesizing these information, a plurality of feasible and effective film coating schemes can be constructed. The deep neural network model can analyze and determine a plurality of first stress elimination film coating schemes that meet the requirements by learning the mapping relationship between the preferred interval of the doping atom concentration of each prominent stress region, the preferred interval of the stress compensation layer thickness and the effective stress elimination film coating scheme. The multi-layer structure of the deep neural network enables it to capture complex patterns and nonlinear relationships in the data, thereby accurately determining suitable film coating schemes.

[0075] Step S3, based on the plurality of first stress elimination film coating schemes, a plurality of original single crystal silicon substrates are respectively coated, and information of each coated single crystal silicon substrate is obtained, each coated single crystal silicon substrate information includes three-dimensional point cloud data of the single crystal silicon substrate, and infrared reflection information of the single crystal silicon substrate after being coated according to the first stress elimination scheme.

[0076] The three-dimensional point cloud data in the information of the coated single crystal silicon substrate is a set of discrete point clouds of spatial coordinates of the surface of the coated single crystal silicon substrate obtained by three-dimensional scanning technology after the original single crystal silicon substrate is coated with infrared high-reflection film based on each first stress elimination film coating scheme.

[0077] The infrared reflection information of the single crystal silicon substrate after being coated according to the first stress elimination scheme is the reflection behavior information of the infrared band obtained by real-time collection of the optical properties of each position of the coated film layer through sensors and optical detection devices. The infrared reflection information includes reflectivity, reflectance spectrum and other data.

[0078] Step S4, determining a target stress elimination film coating scheme based on the information of each coated single crystal silicon substrate.

[0079] In some embodiments, Figure 5 A flowchart for determining a target stress elimination film coating scheme is provided for the embodiments of the present application, and the determination of the target stress elimination film coating scheme includes steps S41-S42:

[0080] Step S41, constructing a film coating map, the film coating map includes a plurality of nodes and a plurality of edges between the nodes, the node characteristics of each node include a stress elimination film coating scheme and a coated single crystal silicon substrate information, and the edges between the nodes represent the difference information of the doping atom concentration distribution information and the difference information of the stress compensation layer thickness between the stress elimination film coating schemes.

[0081] A coating map is a structure consisting of multiple nodes and multiple edges between them. Each node features a stress-relief coating solution and the corresponding single-crystal silicon substrate information after coating. The edges between nodes represent the difference in dopant concentration distribution and stress-compensation layer thickness between two stress-relief coating solutions. The coating map can be used to characterize the relationship between different stress-relief coating solutions and their corresponding coating effects.

[0082] The difference information of the doping atom concentration distribution information refers to the information of the difference distribution of the doping atom concentrations at all spatial positions on the surface of the single crystal silicon substrate between the two stress relief coating solutions.

[0083] The difference information of the stress compensation layer thickness is the numerical difference between the stress compensation layer thickness parameters in two stress relief coating solutions.

[0084] In some embodiments, a deep neural network may be used to determine difference information of doping atom concentration distribution information and difference information of stress compensation layer thickness between stress relief coating solutions.

[0085] Figure 6 Schematic diagram of a coating map constructed according to an embodiment of the present invention. Figure 6 As shown, Figure 6 Including stress relief coating scheme A, stress relief coating scheme B, stress relief coating scheme C, stress relief coating scheme D, the edge between two stress relief coating scheme nodes is the difference information of the doping atom concentration distribution information and the difference information of the stress compensation layer thickness between the two stress relief coating schemes.

[0086] The information of each single-crystal silicon substrate after coating reflects the coating effect of the corresponding stress relief coating scheme. By constructing a coating map, the relationship between different schemes and their effects can be intuitively represented in the form of a map structure, and structured data representation can be provided.

[0087] Step S42: Processing the coating map based on a neural network model to determine a target stress relief coating solution.

[0088] In some embodiments, the neural network model is a graph convolutional network, the input of the neural network model is the coating map, and the output of the neural network model is a target stress relief coating solution.

[0089] A graph convolutional network (GCN) is a deep learning model that can be used to process graph data. By defining convolution operations on the nodes of a coating graph, a GCN effectively captures the spatial dependencies and structural features between nodes in the graph. GCNs can combine node features with the structural information of the graph to learn, enabling tasks such as classification, clustering, and prediction of nodes in the graph.

[0090] The target stress-relief coating scheme is determined by processing the coating map using a graph convolutional network. It optimally eliminates stress in infrared high-reflection films and ensures that the coated single-crystal silicon substrate has good stress state and optical properties. The target stress-relief coating scheme includes the target dopant concentration distribution information for the infrared high-reflection film and the target stress compensation layer thickness information.

[0091] In the coating atlas, each node represents a stress relief coating scheme and the corresponding information about the single-crystal silicon substrate after coating. The node features include the stress relief coating scheme and the information about the single-crystal silicon substrate after coating with the stress relief coating scheme. These features can directly reflect the implementation effect of the stress relief coating scheme and the substrate response. The edges between nodes represent the difference between the doping atom concentration distribution and the stress compensation layer thickness between different schemes. The edges can reflect the degree of difference in the stress relief coating schemes and the correlation between their impact on substrate performance. For example, schemes with large concentration differences may correspond to significant changes in the stress state of the substrate. By processing the coating atlas data through a graph convolutional network, the potential pattern of the optimal scheme can be mined from the association between node features and edges, thereby assisting in determining the target stress relief coating scheme. This helps to quickly locate the optimal parameter combination that can effectively eliminate the stress of the infrared high-reflection film and ensure the optical performance of the substrate in a complex combination of schemes, and avoid blind trial and error. Through the convolution operation of graph nodes, the graph convolution network can capture the spatial dependencies and structural characteristics between solutions. The graph convolution network can combine node features with the graph topology to generate low-dimensional embedding representations. These embeddings contain the mapping relationship between stress relief coating solution parameters and stress relief effects, thereby accurately predicting and screening target solutions.

[0092] Step S5 , coating the original single crystal silicon substrate with an infrared high reflective film based on the target stress relief coating solution.

[0093] When the target stress relief coating solution is determined, the target stress relief coating solution is used to coat the original single crystal silicon substrate with an infrared high-reflective film.

[0094] Based on the same inventive concept, Figure 7 A schematic diagram of a coating control system based on data processing provided by an embodiment of the present invention, wherein the coating control system based on data processing includes:

[0095] The acquisition module 71 is configured to acquire film information of the infrared high-reflection film without stress elimination.

[0096] The generation module 72 is configured to determine a plurality of first stress-elimination film coating schemes based on the film information of the infrared high-reflection film without stress elimination, wherein the first stress-elimination film coating scheme comprises doping atomic concentration distribution information of the infrared high-reflection film and stress compensation layer thickness information.

[0097] The stress elimination module 73 is configured to perform film coating on a plurality of original single-crystal silicon substrates based on the plurality of first stress-elimination film coating schemes respectively, and acquire single-crystal silicon substrate information after film coating, wherein the single-crystal silicon substrate information after film coating comprises three-dimensional point cloud data of the single-crystal silicon substrate and infrared reflection information of the single-crystal silicon substrate after film coating under the first stress-elimination film coating scheme.

[0098] The scheme determination module 74 is configured to determine a target stress-elimination film coating scheme based on the single-crystal silicon substrate information after film coating.

[0099] The film coating control module 75 is configured to perform infrared high-reflection film coating on the original single-crystal silicon substrate based on the target stress-elimination film coating scheme.

[0100] In addition, unless the claim explicitly states otherwise, the order of the processing elements and sequences described in this specification, the use of the lettered or numbered alphabets, or the use of other names, is not intended to limit the order of the processes and methods of this specification. Although some currently considered useful embodiments are discussed in the above disclosure through various examples, it should be understood that such details are only for the purpose of illustration, and the additional claims are not limited to the disclosed embodiments, but rather, the claims are intended to cover all modifications and equivalent combinations within the spirit and scope of the embodiments. For example, although the system components described above can be implemented by hardware devices, they can also be implemented by software solutions, such as installing the described system on existing servers or mobile devices.

[0101] Similarly, it should be noted that, in order to simplify the expression of this disclosure and to help understand one or more embodiments, the description of the embodiments of this specification sometimes combines various features into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the object of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0102] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. A coating control method based on data processing, characterized in that: include: Obtain coating information of infrared high-reflective films that have not relieved stress; Determining a plurality of first stress relief coating schemes based on the coating information of the infrared high reflective film where stress has not been relieved, wherein the first stress relief coating schemes include doping atom concentration distribution information and stress compensation layer thickness information of the infrared high reflective film; Coating a plurality of original single crystal silicon substrates based on the plurality of first stress relief coating schemes respectively, and obtaining information of each single crystal silicon substrate after coating, wherein the information of each single crystal silicon substrate after coating includes three-dimensional point cloud data of the single crystal silicon substrate and infrared reflection information of the single crystal silicon substrate after coating with the first stress relief scheme; Determining a target stress relief coating solution based on the information of each coated single crystal silicon substrate; Based on the target stress relief coating scheme, an infrared high-reflective film is coated on the original single crystal silicon substrate.

2. The coating control method based on data processing according to claim 1, wherein determining a plurality of first stress relief coating schemes based on the coating information of the infrared high-reflective film that has not relieved stress comprises: Determine multiple prominent stress point information based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating; determining a plurality of protruding stress areas based on the plurality of protruding stress point information; Determining a preferred range of doping atom concentration and a preferred range of stress compensation layer thickness for each protrusive stress region based on the plurality of protrusive stress regions; A plurality of first stress relief coating schemes are determined based on the preferred range of doping atom concentration and the preferred range of stress compensation layer thickness of each protruding stress region.

3. The coating control method based on data processing according to claim 1, wherein determining a target stress relief coating solution based on the information of each single crystal silicon substrate after coating comprises: Constructing a coating map, wherein the coating map includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein node features of each node include a stress relief coating scheme and information about a single crystal silicon substrate after coating, and edges between the nodes represent difference information of dopant atom concentration distribution information and difference information of stress compensation layer thickness between the stress relief coating schemes; The coating map is processed based on a neural network model to determine a target stress relief coating solution.

4. The coating control method based on data processing as described in claim 1, wherein the coating information of the infrared high-reflection film without stress relief includes the coating parameters of the infrared high-reflection film, the three-dimensional point cloud data of the single crystal silicon substrate after the original coating, and the infrared reflection information of the single crystal silicon substrate after the original coating.

5. A coating control system based on data processing, used to execute the coating control method based on data processing according to any one of claims 1 to 4, characterized in that: include: An acquisition module is used to obtain the coating information of the infrared high-reflection film that has not been stress-relieved; a generating module, configured to determine a plurality of first stress-relieving coating schemes based on the coating information of the infrared high-reflective film without stress relief, wherein the first stress-relieving coating schemes include doping atom concentration distribution information and stress compensation layer thickness information of the infrared high-reflective film; a stress relief module, configured to coat a plurality of original single crystal silicon substrates based on the plurality of first stress relief coating schemes, and obtain information of each single crystal silicon substrate after coating, wherein the information of each single crystal silicon substrate after coating includes three-dimensional point cloud data of the single crystal silicon substrate and infrared reflection information of the single crystal silicon substrate after coating with the first stress relief scheme; a scheme determination module, configured to determine a target stress relief coating scheme based on information of each single crystal silicon substrate after coating; The coating control module is used to coat the original single crystal silicon substrate with an infrared high-reflective film based on the target stress relief coating solution.

6. The coating control system based on data processing according to claim 5, characterized in that: The generation module is further configured to: Determine multiple prominent stress point information based on the three-dimensional point cloud data of the single crystal silicon substrate after coating and the infrared reflection information of the single crystal silicon substrate after coating; determining a plurality of protruding stress areas based on the plurality of protruding stress point information; Determining a preferred range of doping atom concentration and a preferred range of stress compensation layer thickness for each protrusive stress region based on the plurality of protrusive stress regions; A plurality of first stress relief coating schemes are determined based on the preferred range of doping atom concentration and the preferred range of stress compensation layer thickness of each protruding stress region.

7. The coating control system based on data processing according to claim 5, characterized in that: The solution determination module is also used for: Constructing a coating map, wherein the coating map includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein node features of each node include a stress relief coating scheme and information about a single crystal silicon substrate after coating, and edges between the nodes represent difference information of dopant atom concentration distribution information and difference information of stress compensation layer thickness between the stress relief coating schemes; The coating map is processed based on a neural network model to determine a target stress relief coating solution.

8. The coating control system based on data processing according to claim 5, characterized in that: The coating information of the infrared high-reflection film without stress relief includes the coating parameters of the infrared high-reflection film, the three-dimensional point cloud data of the single crystal silicon substrate after the original coating, and the infrared reflection information of the single crystal silicon substrate after the original coating.

9. An electronic device, characterized in that: include: processor; Memory; and a computer program; wherein the computer program is stored in the memory and is configured to be executed by the processor to implement the coating control method based on data processing as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the coating control method based on data processing as described in any one of claims 1 to 4 is implemented.

Images