Method and apparatus for correcting thickness of film layer, and computer device

Abstract

A method and apparatus for correcting the thickness of a film layer, and a computer device. The method comprises: acquiring parameter values of controllable parameters and time information corresponding to the parameter values, and generating a first vector set and a second vector set; inputting the first vector set and the second vector set into a pre-trained prediction model to output a predicted value of a thickness correction value, the pre-trained prediction model comprising a convolutional neural network and a gated recurrent unit arranged in parallel, the convolutional neural network being configured to process the first vector set, and the gated recurrent unit being configured to process the second vector set; and adjusting the parameter values of the controllable parameters on the basis of the predicted value of the thickness correction value.

Description

Methods, apparatus, and computer equipment for correcting film thickness

[0001] Related applications

[0002] This application claims priority to Chinese patent application filed on June 17, 2024, with application number 202410773744.0 and entitled "Method, Apparatus, and Computer Equipment for Correcting Film Thickness", the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application relates to the field of vacuum evaporation technology, and in particular to a method, apparatus, and computer equipment for correcting film thickness. Background Technology

[0004] With the increasingly widespread application of display panels, vacuum evaporation technology has become an indispensable and important process. Vacuum evaporation technology refers to a process in which the coating material is evaporated into vapor by heating under vacuum conditions, causing the particles to fly to the surface of the substrate and condense into a film. For example, vacuum evaporation technology can be used to form the light-emitting layer in Organic Light-Emitting Diode (OLED) display panels.

[0005] Currently, in the process of forming films using vacuum evaporation technology, the evaporation rate is typically adjusted based on the difference between the current measured film thickness and the theoretically expected thickness to correct the final film thickness. However, this method is not very effective for correcting film thickness. Therefore, there is an urgent need for a film thickness correction method that can improve both the efficiency and accuracy of the correction. Summary of the Invention

[0006] According to various embodiments of this application, a method, apparatus, and computer device for correcting film thickness to improve correction efficiency and accuracy are provided.

[0007] In a first aspect, this application provides a method for correcting film thickness, comprising:

[0008] The parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values ​​are obtained to generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter;

[0009] The first vector set and the second vector set are input into a pre-trained prediction model, which outputs a predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set;

[0010] Based on the predicted value of the thickness correction value, the parameter value of the controllable parameter is adjusted until the film layer formed by the evaporation process reaches the target thickness.

[0011] In one embodiment, the convolutional neural network includes multiple parallel-connected convolutional layers, and the gated recurrent unit includes multiple hidden layers; the step of inputting the first vector set and the second vector set into a pre-trained prediction model and outputting a predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process includes:

[0012] The first vector set is input into a convolutional neural network so that multiple parallel convolutional layers process the first vector set simultaneously and output a first feature; the first feature is used to describe the coupling relationship between the controllable parameters.

[0013] The second vector set is input into a gated loop unit so that multiple hidden layers process the second vector set sequentially and output a second feature; the second feature is used to describe the time variation law of the controllable parameter.

[0014] Based on the first feature and the second feature, a predicted value for the thickness correction is output.

[0015] In one embodiment, the convolutional layer is a multi-channel convolutional layer, and the step of inputting the first vector set into the convolutional neural network so that multiple parallel-connected convolutional layers simultaneously process the first vector set and output a first feature includes:

[0016] Multiple sets of the first vectors are input into a convolutional neural network, so that each multi-channel convolutional layer uses multiple convolutional kernels to process multiple sets of the first vectors and outputs multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

[0017] In one embodiment, the prediction model includes a self-attention layer, wherein the predicted value for outputting the thickness correction value based on the first feature and the second feature is preceded by:

[0018] Weights are assigned to the first feature based on the self-attention layer connected to each convolutional layer.

[0019] Weights are assigned to the second feature based on the self-attention layer connected to the last hidden layer.

[0020] In one embodiment, the convolutional neural network includes three parallel-connected convolutional layers, and the gated recurrent unit includes three hidden layers.

[0021] In one embodiment, the step of acquiring the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values, and generating a first vector set and a second vector set, includes:

[0022] Obtain the parameter values ​​of controllable parameters during the vapor deposition process and the corresponding time information of the parameter values;

[0023] The parameter value is mapped to numerical data to obtain the numerical data corresponding to the controllable parameter and the time information corresponding to the numerical data.

[0024] Filter out the data of the numerical type corresponding to each of the time information to obtain the first vector corresponding to each of the time information;

[0025] Based on the numerical data corresponding to each of the controllable parameters, the statistical characteristics of each of the controllable parameters are calculated to form a first vector set;

[0026] The parameter values ​​of multiple controllable parameters are sorted according to the time information to obtain the corresponding second vector set of the multiple controllable parameters.

[0027] In one embodiment, the step of acquiring the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values, and generating a first vector set and a second vector set, includes:

[0028] The parameter values ​​of controllable parameters during the vapor deposition process, as well as the time and position information corresponding to the parameter values, are obtained, and a first vector set and a second vector set corresponding to each position information are generated.

[0029] The step of adjusting the parameter value of the controllable parameter based on the predicted value of the thickness correction value includes:

[0030] The target location information is determined from various location information.

[0031] Based on the predicted value of the thickness correction value corresponding to the target location information, adjust the parameter value of the controllable parameter at the location corresponding to the target location information.

[0032] In one embodiment, adjusting the parameter value of the controllable parameter based on the predicted value of the thickness correction value includes:

[0033] Adjust the parameter value of the controllable parameter based on the predicted value of the thickness correction value;

[0034] Obtain the current thickness of the film layer formed by the vapor deposition process. If the current thickness is within a preset first range, adjust the parameter value of the controllable parameter according to the difference between the current thickness and the target thickness, and then execute the step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value again.

[0035] Secondly, this application also provides a film thickness correction device, comprising:

[0036] The acquisition module is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values, and generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter;

[0037] The prediction module is used to input the first vector set and the second vector set into a pre-trained prediction model and output the predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set;

[0038] The correction module is used to adjust the parameter value of the controllable parameter according to the predicted value of the thickness correction value until the film layer formed by the evaporation process reaches the target thickness.

[0039] In one embodiment, the convolutional neural network includes multiple parallel-connected convolutional layers, and the gated recurrent unit includes multiple hidden layers; the prediction module includes:

[0040] The first processing submodule is used to input the first vector set into a convolutional neural network, so that multiple parallel-connected convolutional layers process the first vector set simultaneously and output a first feature; the first feature is used to describe the coupling relationship between the controllable parameters.

[0041] The second processing submodule is used to input the second vector set into the gated loop unit so that the multiple hidden layers process the second vector set in sequence and output the second feature; the second feature is used to describe the time variation law of the controllable parameter.

[0042] The output submodule is used to output a predicted value of the thickness correction value based on the first feature and the second feature.

[0043] In one embodiment, the convolutional layer is a multi-channel convolutional layer, and the first processing submodule includes:

[0044] The processing unit is used to input multiple sets of the first vectors into a convolutional neural network, so that each multi-channel convolutional layer uses multiple convolutional kernels to process the multiple sets of the first vectors and output multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

[0045] In one embodiment, the prediction model includes a self-attention layer, and the prediction module further includes:

[0046] The first allocation submodule is used to allocate weights to the first feature based on the self-attention layer connected to each convolutional layer;

[0047] The second allocation submodule is used to assign weights to the second feature based on the self-attention layer connected to the last hidden layer.

[0048] In one embodiment, the convolutional neural network includes three parallel-connected convolutional layers, and the gated recurrent unit includes three hidden layers.

[0049] In one embodiment, the acquisition module includes:

[0050] The first acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values;

[0051] The mapping submodule is used to map the parameter value to numeric data to obtain the numeric data corresponding to the controllable parameter and the time information corresponding to the numeric data.

[0052] The first generation submodule is used to filter out the data of the numerical type corresponding to each of the time information to obtain the first vector corresponding to each of the time information.

[0053] The second generation submodule is used to calculate the statistical characteristics of each controllable parameter based on the numerical data corresponding to each controllable parameter, and to form a first vector set.

[0054] The third generation submodule is used to sort the parameter values ​​of multiple controllable parameters according to the time information to obtain the corresponding second vector set of the multiple controllable parameters.

[0055] In one embodiment, the acquisition module includes:

[0056] The second acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process, as well as the time and position information corresponding to the parameter values, and generate a first vector set and a second vector set corresponding to each position information.

[0057] The correction module is also used for:

[0058] The target location information is determined from various location information.

[0059] Based on the predicted value of the thickness correction value corresponding to the target location information, adjust the parameter value of the controllable parameter at the location corresponding to the target location information.

[0060] In one embodiment, the correction module includes:

[0061] The first adjustment submodule is used to adjust the parameter value of the controllable parameter according to the predicted value of the thickness correction value;

[0062] The second adjustment submodule is used to obtain the current thickness of the film layer formed by the evaporation process, and when the current thickness is within a preset first range, adjust the parameter value of the controllable parameter according to the difference between the current thickness and the target thickness, and then execute the step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value again.

[0063] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described in any of the above-mentioned embodiments.

[0064] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0065] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the above claims.

[0066] Details of one or more embodiments of the present invention are set forth in the following drawings and description. Other features, objects, and advantages of the invention will become apparent from the specification, drawings, and claims. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the disclosed drawings without creative effort.

[0068] Figure 1 is a flowchart illustrating a method for correcting film thickness in one embodiment;

[0069] Figure 2 is a flowchart illustrating the process of inputting a first vector set and a second vector set into a pre-trained prediction model to output a predicted value of the thickness correction value corresponding to the film layer formed during the vapor deposition process in one embodiment.

[0070] Figure 3 is a flowchart illustrating how, in one embodiment, a first vector set is input into a convolutional neural network so that multiple parallel-connected convolutional layers process the first vector set simultaneously and output a first feature.

[0071] Figure 4 is a flowchart illustrating the process of inputting the first vector and the second vector into a pre-trained prediction model and outputting the predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process in another embodiment.

[0072] Figure 5 is a schematic diagram of the structure used for the prediction model in one embodiment;

[0073] Figure 6 is a flowchart of an embodiment for obtaining the parameter values ​​of controllable parameters and the time information corresponding to the parameter values ​​during the vapor deposition process, and generating a first vector set and a second vector set.

[0074] Figure 7 is a schematic diagram of the vapor deposition process in one embodiment;

[0075] Figure 8 is a flowchart illustrating the process of obtaining the parameter values ​​of controllable parameters and the corresponding time information during the vapor deposition process, and generating the first vector set and the second vector set in another embodiment.

[0076] Figure 9 is a flowchart illustrating the process of adjusting the parameter values ​​of controllable parameters based on the predicted values ​​of thickness correction values ​​in one embodiment.

[0077] Figure 10 is a structural block diagram of a film thickness correction device in one embodiment;

[0078] Figure 11 is an internal structure diagram of a computer device in one embodiment. Detailed Implementation

[0079] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0080] In one embodiment, as shown in Figure 1, a method for adjusting the film thickness is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0081] Step S102: Obtain the parameter values ​​of controllable parameters and the corresponding time information during the vapor deposition process, and generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter.

[0082] The controllable parameters can include parameters that can be directly adjusted during the vapor deposition process, or parameters that can be indirectly adjusted by changing other parameters or conditions. Controllable parameters include at least one or more of the following: evaporation source temperature, evaporation rate, evaporation source scanning rate, substrate temperature, substrate movement rate, process parameters, raw material characteristics, etc. For example, the parameter values ​​of the controllable parameters can be obtained through data acquisition devices such as monitoring equipment and sensors.

[0083] Because the vapor deposition process involves numerous controllable parameters, each deposition cycle yields a large number of parameter values. Therefore, preprocessing these parameter values ​​can reduce the difficulty of building and training the prediction model, while simultaneously improving its prediction efficiency and accuracy. For example, the parameter values ​​can first undergo preprocessing such as data cleaning and normalization; then, they can be stored in chronological order of data acquisition to form a time-series dataset; parameter values ​​at the same time point can be selected from the time-series dataset to form multiple first vectors; and finally, parameter values ​​for the same controllable parameter can be selected from the time-series dataset to form multiple second vectors.

[0084] Step S104: Input the first vector set and the second vector set into the pre-trained prediction model and output the predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set.

[0085] To reduce the defect rate of products such as display panels caused by the vapor deposition process, a high speed is required for correcting the film thickness. Therefore, the prediction model can be constructed by including parallel convolutional neural networks and gated recurrent units to improve the prediction efficiency of the model.

[0086] In one possible implementation, the training of the prediction model includes: collecting parameter values ​​of controllable parameters during multiple vapor deposition processes to form a dataset; training the prediction model based on this dataset so that the trained prediction model can predict the thickness correction value. Furthermore, the parameters of the prediction model can be iteratively optimized using a backpropagation algorithm to minimize the prediction error and improve the performance of the prediction model.

[0087] Step S106: Adjust the parameter values ​​of the controllable parameters according to the predicted value of the thickness correction value until the film layer formed by the evaporation process reaches the target thickness.

[0088] For example, throughout the evaporation process, the predicted thickness correction value can be monitored in real time. Based on the predicted thickness correction value, the values ​​of one or more controllable parameters (e.g., evaporation source temperature, evaporation source scan rate, etc.) can be dynamically adjusted until the formed film reaches the target thickness, at which point the current evaporation process ends. In one possible implementation, the parameter values ​​of the controllable parameters can be adjusted according to the sign of the predicted thickness correction value. For example, if the current predicted thickness correction value is positive, the evaporation source temperature can be increased.

[0089] Furthermore, before performing vapor deposition, controllable parameters such as process parameters and raw material characteristics can be initially adjusted based on the current target thickness.

[0090] In the aforementioned method for correcting film thickness, vectorizing the evaporation-related values ​​before processing them with the prediction model improves the training and prediction efficiency of the prediction model. Constructing the prediction model with a structure including parallel convolutional neural networks and gated recurrent units enables simultaneous feature mining in both the time and parameter dimensions, improving the prediction efficiency and accuracy. Dynamically adjusting the evaporation-related parameters using the predicted thickness correction values ​​output by the prediction model further enhances the efficiency and accuracy of film thickness correction, thus ensuring the quality of the evaporated film.

[0091] In one embodiment, the convolutional neural network may include multiple convolutional layers connected in parallel, and the gated recurrent unit may include multiple hidden layers, as shown in FIG2. Step S104 may include:

[0092] Step S402: Input the first vector set into the convolutional neural network so that multiple parallel convolutional layers process the first vector set simultaneously and output the first feature; the first feature is used to describe the coupling relationship between controllable parameters.

[0093] For example, for each convolutional layer, a one-dimensional convolution operation can be slid back and forth on the first vector to capture the local contextual information of each first vector, and this local contextual information can be compressed into a feature map to obtain the first feature. In one possible implementation, zero-padding convolution can be used.

[0094] For example, a first formula can be used to extract a first feature, which may include:

[0095] Among them, c i Let f be the i-th first feature, and f be a non-linear function; W be a one-dimensional convolution kernel, W∈R h×d h represents the window size for which the convolution kernel is applied, and d represents the dimension of the first vector; x i:i+j Represents the first vector x from the i-th to the (i+j-th)-th vector. i x i+1 ...x i+j b is a bias term, b∈R.

[0096] Step S404: Input the second vector set into the gated recurrent unit so that the multiple hidden layers process the second vector set in sequence and output the second feature; the second feature is used to describe the time variation law of the controllable parameter.

[0097] The gated loop unit can include an update gate and a reset gate. The update gate is used to retain and add information, while the reset gate is used to decide whether to discard information. For example, at time t, the update gate and reset gate can be calculated first; new memory content will use the reset gate to store relevant past information; finally, the hidden state is calculated, the information of the current unit is retained, and it is passed to the next unit.

[0098] For example, the parameters of a gated loop unit can be updated using the following formula: z t =σ(W z ·x t +U z ·h t-1 ); r t =σ(W r ·x t +U r ·h t-1 ); h t =(1-z) t )·h t-1 +zt ·h t ;

[0099] Among them, z t For the update gate, σ represents the sigmoid function, W z W r , U z U r , It is the weight matrix of the gated recurrent unit, x t Let h be the input vector at time t. t-1 r represents the state information at time t-1. t To reset the door, Indicates the candidate hidden state. h represents the element-wise multiplication operation. t Indicates a hidden state.

[0100] When the number of hidden units is s, the second vector H can be represented as: (h1, h2, ... h n ], and the second vector H satisfies H∈R n×s , where n represents the length of the second vector.

[0101] Step S406: Based on the first feature and the second feature, output the predicted value of the thickness correction value.

[0102] For example, the prediction model may also include a fully connected layer and an output layer. By processing the first feature and the second feature through the fully connected layer and the output layer, a predicted value of the thickness correction value can be obtained.

[0103] In this embodiment, by simultaneously processing the first vector through multiple parallel convolutional layers, local features of the first vector at different granularities can be quickly captured, helping the prediction model to comprehensively understand the coupling relationships between various controllable parameters during the vapor deposition process in terms of parameter dimensions. By processing the second vector through multiple hidden layers, more complex and deeper time-series information can be gradually extracted, helping the prediction model to accurately understand the temporal variation patterns of various controllable parameters during the vapor deposition process in terms of time dimensions. Therefore, the prediction model can quickly and accurately predict the required thickness correction value by comprehensively considering the local features and time-series information of parameter values ​​during the vapor deposition process.

[0104] In one embodiment, the convolutional layer can be a multi-channel convolutional layer, as shown in FIG3, and step S402 may include:

[0105] Step S4021: Input multiple sets of first vectors into the convolutional neural network so that each multi-channel convolutional layer uses multiple convolutional kernels to process multiple sets of first vectors and output multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

[0106] The multiple sets of first vectors are generated based on the parameter values ​​of multiple sets of controllable parameters. For example, the number of convolutional kernels can be determined based on the number of groups of the first vectors. The parameter values ​​of controllable parameters within the same group must share at least one identical additional attribute. For instance, the parameter values ​​of controllable parameters within the same group could be parameter values ​​obtained at the same location, or parameter values ​​obtained within the same time interval.

[0107] Furthermore, the number of channels in the convolution kernel can be determined based on the number of types of the first vector. Different types of the first vector represent different physical meanings.

[0108] In this embodiment, by constructing the convolutional neural network as a multi-channel convolutional layer with multiple parallel connections, it is possible to simultaneously extract local features of multiple sets of first vectors, thereby further improving the learning ability and prediction efficiency of the prediction model.

[0109] To further improve the accuracy of the predicted thickness correction value by the prediction model, in one embodiment, as shown in FIG4, step S104 may further include:

[0110] Step S4051: Assign weights to the first feature based on the self-attention layer connected to each convolutional layer.

[0111] Since convolution operations are used to obtain local feature information within a window, in order to extract the potential relationships between controllable parameters and avoid the loss of dependent information in pooling operations, a self-attention mechanism can be introduced into the convolutional neural network, forming a CNN-Att (Attention based Convolutional Neural Networks) structure. For example, the self-attention layer is used to extract the potential relationships between controllable parameters through matrix operations. k The matrix is ​​re-encoded into n×d v The matrix, where n represents the length of the input sequence, d k d represents the dimension of the key. v The dimension representing the value. The formula for calculating a self-attention layer can include:

[0112] Among them, f Att Represents a function. T is the transpose operator. This is used to prevent the inner product result from being too large.

[0113] Step S4052: Assign weights to the second feature based on the self-attention layer connected to the last hidden layer.

[0114] Similarly, in order to integrate global feature information more deeply while focusing on representative and highly relevant features, a self-attention mechanism can be introduced into the convolutional neural network to form the GRU-Att (Attention based Gated Recurrent Unit) structure.

[0115] In this embodiment, by introducing a self-attention mechanism into the prediction model, accurate weight allocation can be provided for the prediction of thickness correction values, further improving the prediction accuracy of the prediction model.

[0116] In one embodiment, please refer to Figure 5, which is a schematic diagram of the prediction model structure in one embodiment. The convolutional neural network includes three parallel convolutional layers, and the gated recurrent unit includes three hidden layers. Experimental tests show that the prediction model achieves high accuracy in predicting the film thickness correction value, and the computational cost of training the prediction model is low.

[0117] In one embodiment, as shown in FIG6, step S102 above may include:

[0118] Step A1: Obtain the parameter values ​​of the controllable parameters during the vapor deposition process and the corresponding time information.

[0119] The time information corresponding to the parameter value can be a timestamp. It can be understood that the parameter value of each controllable parameter includes the parameter value at different times, and each parameter can correspond to a timestamp.

[0120] Step A2: Map the parameter values ​​to numerical data to obtain the numerical data corresponding to the controllable parameters and the time information corresponding to the numerical data.

[0121] The controllable parameters can be continuous or discrete variables. For example, the parameter values ​​of continuous variables such as substrate temperature and evaporation source scanning rate can be converted into real numbers, and the parameter values ​​of discrete variables such as evaporation source location information can be converted into corresponding classification labels to obtain data of numerical type corresponding to each controllable parameter.

[0122] Step A3: Filter out the data of the numerical type corresponding to each time information to obtain the first vector corresponding to each time information.

[0123] Step A4: Based on the numerical data corresponding to each controllable parameter, calculate the statistical characteristics of each controllable parameter to form the first vector set.

[0124] The first vector can include the statistical characteristics of each controllable parameter (e.g., calculating the average evaporation rate, the rate of change of substrate temperature, etc.). In this way, processing the first vector through a convolutional neural network can help the prediction model learn the coupling relationship between the various controllable parameters more comprehensively.

[0125] Step A5: Sort the parameter values ​​of multiple controllable parameters according to the time information to obtain the second vector set corresponding to the multiple controllable parameters.

[0126] The parameter values ​​can be historical or current values. For example, for any controllable parameter, the values ​​can be sorted chronologically to obtain a second vector corresponding to that parameter. By processing this second vector through a gated recurrent unit, the prediction model can learn the temporal variation patterns of each controllable parameter.

[0127] Furthermore, the second vectors corresponding to each controllable parameter can be standardized to reduce the scale differences between the second vectors.

[0128] In this embodiment, by converting parameter values ​​into vectors according to the time dimension and the parameter dimension, and the time dimension vector can contain statistical features, and the parameter dimension vector can be a standardized time series, the prediction efficiency and accuracy of the prediction model can be further improved.

[0129] Please refer to Figure 7, which is a schematic diagram of the vapor deposition process in one embodiment. In practical applications, the length of the vapor deposition machine (e.g., a cluster-type vapor deposition machine) may be relatively long, resulting in significant differences in the vapor deposition effect in different areas of the machine. These different areas can be different vapor deposition units as shown in Figure 7, or different locations scanned by the evaporation source within a single cavity of the same evaporation unit. Therefore, data acquisition devices can be installed in each area of ​​the vapor deposition machine. In one embodiment, as shown in Figure 8, step S102 may include:

[0130] Step S1021: Obtain the parameter values ​​of controllable parameters during the vapor deposition process, as well as the time and position information corresponding to the parameter values, and generate a first vector set and a second vector set corresponding to each position information.

[0131] In one possible implementation, a location label can be added to each parameter value; in another possible implementation, the parameter values ​​of controllable parameters collected by different data acquisition devices can be stored separately.

[0132] Furthermore, step S106 above may include:

[0133] Step S1061: Determine the target location information from the various location information;

[0134] For example, the predicted value of the thickness correction value corresponding to the target location information may not be zero.

[0135] Step S1062: Adjust the parameter value of the controllable parameter at the location corresponding to the target location information based on the predicted value of the thickness correction value corresponding to the target location information.

[0136] For example, the first and second vectors corresponding to each location information can be input into a pre-trained prediction model to output the predicted thickness correction value corresponding to each location information; based on the predicted thickness correction value corresponding to each location information, the location that needs correction is determined; and the parameter values ​​of the controllable parameters at the location that needs correction are adjusted. The target thickness at each target location can be the same or different.

[0137] In this embodiment, by processing the controllable parameter values ​​of different areas of the vapor deposition machine separately, the vapor deposition effect of different areas can be effectively monitored. Furthermore, by adjusting the controllable parameters of the areas that need to be corrected separately, the problem of large differences in vapor deposition effect in different areas can be solved, improving the consistency and continuity of vapor deposition of the entire unit. At the same time, it can also meet the customized needs of different products.

[0138] Since errors and disturbances may occur during the vapor deposition process, in one embodiment, as shown in FIG9, step S106 above may include:

[0139] Step B1: Adjust the parameter values ​​of the controllable parameters based on the predicted values ​​of the thickness correction values.

[0140] Step B2: Obtain the current thickness of the film layer formed during the vapor deposition process. If the current thickness is within the preset first range, adjust the parameter value of the controllable parameter according to the difference between the current thickness and the target thickness. Then, execute the step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value again.

[0141] The preset first range can be greater than the preset minimum thickness and less than the preset maximum thickness. A current thickness not falling within the preset first range can be used to indicate an over-adjustment state.

[0142] For example, when the current thickness falls within a preset first range, the parameter value of the controllable parameter can be adjusted based on the difference between the current thickness and the target thickness, and the predicted value of the thickness correction value can be continuously monitored to ensure the stability of the film thickness correction; when the thickness does not fall within the preset first range (the current thickness is greater than the preset maximum thickness or less than the preset minimum thickness), the parameter value of the controllable parameter can be adjusted without feedback, but an alarm can be issued to relevant personnel to prevent over-adjustment.

[0143] Since the correction requirements may not be met by adjusting the controllable parameters during the evaporation process, alternatively, if the predicted thickness correction value does not fall within a preset second range, it can be determined that the film formed during the evaporation process needs to be corrected. The preset first range can be greater than a preset minimum correction value and less than a preset maximum correction value. The predicted thickness correction value falling within the preset second range can be used to indicate that the film is in a correctable state.

[0144] Understandably, when the predicted thickness correction value falls within the preset second range, it means that by adjusting the controllable parameters, the current thickness can be corrected to the target thickness, allowing for further monitoring of the film thickness. Conversely, when the predicted thickness correction value does not fall within the preset second range, it means that by adjusting the controllable parameters, the current thickness cannot be corrected to the target thickness. In this case, the film layer can undergo a correction process. For example, the settings of the vapor deposition machine can be changed to perform individual correction on the substrate containing the film layer, and the vapor deposition machine settings can be restored after the correction process is completed for the next vapor deposition.

[0145] In this embodiment, by monitoring the predicted values ​​of the current film thickness and film thickness correction value, and combining feedback control and correction processing for vapor deposition, the quality of the vapor-deposited film can be further guaranteed.

[0146] In summary, the above-mentioned methods for correcting film thickness improve the training and prediction efficiency of the prediction model by vectorizing the evaporation-related values ​​before processing them with the prediction model. Furthermore, by constructing the prediction model as a parallel convolutional neural network and gated recurrent units, simultaneous feature mining of both time and parameter dimensions is achieved, enhancing the prediction efficiency and accuracy. Finally, dynamically adjusting the evaporation-related parameters using the predicted thickness correction values ​​output by the prediction model further improves the efficiency and accuracy of film thickness correction, thus ensuring the quality of the evaporated film.

[0147] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0148] Based on the same inventive concept, this application also provides a film thickness correction device for implementing the film thickness correction method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the film thickness correction device provided below can be found in the limitations of the film thickness correction method described above, and will not be repeated here.

[0149] In an exemplary embodiment, as shown in FIG10, a film thickness correction device 300 is provided, comprising: an acquisition module 301, a prediction module 302, and a correction module 303, wherein:

[0150] The acquisition module 301 is used to acquire the parameter values ​​of controllable parameters and the time information corresponding to the parameter values ​​during the vapor deposition process, and generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter.

[0151] The prediction module 302 is used to input the first vector set and the second vector set into the pre-trained prediction model and output the predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set.

[0152] The correction module 303 is used to adjust the parameter values ​​of the controllable parameters according to the predicted value of the thickness correction value until the film layer formed by the evaporation process reaches the target thickness.

[0153] In an exemplary embodiment, the convolutional neural network may include multiple convolutional layers connected in parallel, and the gated recurrent unit may include multiple hidden layers; the prediction module 302 may include:

[0154] The first processing submodule is used to input the first vector set into the convolutional neural network so that multiple parallel convolutional layers process the first vector set simultaneously and output the first feature; the first feature is used to describe the coupling relationship between controllable parameters.

[0155] The second processing submodule is used to input the second vector set into the gated loop unit so that the multiple hidden layers process the second vector set in sequence and output the second feature; the second feature is used to describe the time variation law of the controllable parameter.

[0156] The output submodule is used to output the predicted value of the thickness correction value based on the first feature and the second feature.

[0157] In an exemplary embodiment, the convolutional layer may be a multi-channel convolutional layer, and the first processing submodule may include:

[0158] The processing unit is used to input multiple sets of first vectors into the convolutional neural network, so that each multi-channel convolutional layer uses multiple convolutional kernels to process multiple sets of first vectors and output multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

[0159] In an exemplary embodiment, the prediction model may include a self-attention layer, and the prediction module 302 may further include:

[0160] The first allocation submodule is used to assign weights to the first feature based on the self-attention layer connected to each convolutional layer.

[0161] The second allocation submodule is used to assign weights to the second feature based on the self-attention layer connected to the last hidden layer.

[0162] In an exemplary embodiment, the convolutional neural network described above may include three convolutional layers connected in parallel, and the gated recurrent unit may include three hidden layers.

[0163] In an exemplary embodiment, the acquisition module 301 described above may include:

[0164] The first acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process and the corresponding time information.

[0165] The mapping submodule is used to map parameter values ​​to numeric data, obtaining numeric data corresponding to controllable parameters and time information corresponding to the numeric data.

[0166] The first generation submodule is used to filter out the data of the numerical type corresponding to each time information and obtain the first vector corresponding to each time information.

[0167] The second generation submodule is used to calculate the statistical characteristics of each controllable parameter based on the numerical data corresponding to each controllable parameter, and form the first vector set.

[0168] The third generation submodule is used to sort the parameter values ​​of multiple controllable parameters according to time information to obtain the corresponding second vector set of multiple controllable parameters.

[0169] In an exemplary embodiment, the acquisition module 301 described above may include:

[0170] The second acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the evaporation process, as well as the time and position information corresponding to the parameter values, and to generate a first vector set and a second vector set c corresponding to each position information.

[0171] Furthermore, the parameter values ​​for adjusting the controllable parameters based on the predicted values ​​of the thickness correction values ​​can include:

[0172] The target location information is determined from various location information.

[0173] Based on the predicted value of the thickness correction value corresponding to the target location information, adjust the parameter values ​​of the controllable parameters at the corresponding location.

[0174] In one exemplary embodiment, the correction module 303 described above may include:

[0175] The first adjustment submodule is used to adjust the parameter values ​​of the controllable parameters based on the predicted values ​​of the thickness correction values.

[0176] The second adjustment submodule is used to obtain the current thickness of the film layer formed during the evaporation process. If the current thickness is within a preset first range, the parameter value of the controllable parameter is adjusted according to the difference between the current thickness and the target thickness. The step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value is executed again.

[0177] Each module in the aforementioned film thickness correction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0178] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram is shown in Figure 11. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a method for correcting film thickness.

[0179] Those skilled in the art will understand that the structure shown in Figure 11 is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0180] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0181] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0182] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0183] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0184] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0185] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for correcting film thickness, characterized in that, The method includes: The parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values ​​are obtained to generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter; The first vector set and the second vector set are input into a pre-trained prediction model, which outputs a predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set; Based on the predicted value of the thickness correction value, the parameter value of the controllable parameter is adjusted until the film layer formed by the evaporation process reaches the target thickness.

2. The method according to claim 1, characterized in that, The convolutional neural network includes multiple parallel-connected convolutional layers, and the gated recurrent unit includes multiple hidden layers; the step of inputting the first vector set and the second vector set into a pre-trained prediction model and outputting a predicted value of the thickness correction value corresponding to the film layer formed by the evaporation process includes: The first vector set is input into a convolutional neural network so that multiple parallel convolutional layers process the first vector set simultaneously and output a first feature; the first feature is used to describe the coupling relationship between the controllable parameters. The second vector set is input into a gated loop unit so that multiple hidden layers process the second vector set sequentially and output a second feature; the second feature is used to describe the time variation law of the controllable parameter. Based on the first feature and the second feature, a predicted value for the thickness correction is output.

3. The method according to claim 2, characterized in that, The convolutional layer is a multi-channel convolutional layer. The step of inputting the first vector set into the convolutional neural network, so that multiple parallel-connected convolutional layers simultaneously process the first vector set and output a first feature, includes: Multiple sets of the first vectors are input into a convolutional neural network, so that each multi-channel convolutional layer uses multiple convolutional kernels to process multiple sets of the first vectors and output multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

4. The method according to claim 2, characterized in that, The prediction model includes a self-attention layer, and the predicted value for outputting the thickness correction value based on the first feature and the second feature includes: Weights are assigned to the first feature based on the self-attention layer connected to each convolutional layer. Weights are assigned to the second feature based on the self-attention layer connected to the last hidden layer.

5. The method according to any one of claims 1 to 4, characterized in that, The convolutional neural network includes three parallel convolutional layers, and the gated recurrent unit includes three hidden layers.

6. The method according to claim 1, characterized in that, The process of acquiring the parameter values ​​of controllable parameters during the vapor deposition process and the corresponding time information, and generating a first vector set and a second vector set, includes: Obtain the parameter values ​​of controllable parameters during the vapor deposition process and the corresponding time information of the parameter values; The parameter value is mapped to numerical data to obtain the numerical data corresponding to the controllable parameter and the time information corresponding to the numerical data. Filter out the data of the numerical type corresponding to each of the time information to obtain the first vector corresponding to each of the time information; Based on the numerical data corresponding to each of the controllable parameters, the statistical characteristics of each of the controllable parameters are calculated to form a first vector set; The parameter values ​​of multiple controllable parameters are sorted according to the time information to obtain the corresponding second vector set of the multiple controllable parameters.

7. The method according to claim 1, characterized in that, The process of acquiring the parameter values ​​of controllable parameters during the vapor deposition process and the corresponding time information, and generating a first vector set and a second vector set, includes: The parameter values ​​of controllable parameters during the vapor deposition process, as well as the time and position information corresponding to the parameter values, are obtained, and a first vector set and a second vector set corresponding to each position information are generated. The step of adjusting the parameter value of the controllable parameter based on the predicted value of the thickness correction value includes: The target location information is determined from various location information. Based on the predicted value of the thickness correction value corresponding to the target location information, adjust the parameter value of the controllable parameter at the location corresponding to the target location information.

8. The method according to claim 1, characterized in that, The step of adjusting the parameter value of the controllable parameter based on the predicted value of the thickness correction value includes: Adjust the parameter value of the controllable parameter based on the predicted value of the thickness correction value; Obtain the current thickness of the film layer formed by the vapor deposition process. If the current thickness is within a preset first range, adjust the parameter value of the controllable parameter according to the difference between the current thickness and the target thickness, and then execute the step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value again.

9. A film thickness correction device, characterized in that, The device includes: The acquisition module is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values, and generate a first vector set and a second vector set; each first vector in the first vector set includes the parameter values ​​of multiple controllable parameters corresponding to the same time information, and each second vector in the second vector set is a time sequence composed of the parameter values ​​of the same controllable parameter; The prediction module is used to input the first vector set and the second vector set into a pre-trained prediction model and output the predicted value of the thickness correction value corresponding to the film layer formed by the vapor deposition process; wherein, the pre-trained prediction model includes a parallel convolutional neural network and a gated recurrent unit, the convolutional neural network is used to process the first vector set, and the gated recurrent unit is used to process the second vector set; The correction module is used to adjust the parameter value of the controllable parameter according to the predicted value of the thickness correction value until the film layer formed by the evaporation process reaches the target thickness.

10. The apparatus according to claim 9, characterized in that, The convolutional neural network includes multiple parallel-connected convolutional layers, and the gated recurrent unit includes multiple hidden layers; the prediction module includes: The first processing submodule is used to input the first vector set into a convolutional neural network, so that multiple parallel-connected convolutional layers process the first vector set simultaneously and output a first feature; the first feature is used to describe the coupling relationship between the controllable parameters. The second processing submodule is used to input the second vector set into the gated loop unit so that the multiple hidden layers process the second vector set in sequence and output the second feature; the second feature is used to describe the time variation law of the controllable parameter. The output submodule is used to output a predicted value of the thickness correction value based on the first feature and the second feature.

11. The apparatus according to claim 10, characterized in that, The convolutional layer is a multi-channel convolutional layer, and the first processing submodule includes: The processing unit is used to input multiple sets of the first vectors into a convolutional neural network, so that each multi-channel convolutional layer uses multiple convolutional kernels to process the multiple sets of the first vectors and output multiple first features; each first feature is used to describe the coupling relationship between controllable parameters within each group.

12. The apparatus according to claim 10, characterized in that, The prediction model includes a self-attention layer, and the prediction module further includes: The first allocation submodule is used to allocate weights to the first feature based on the self-attention layer connected to each convolutional layer; The second allocation submodule is used to assign weights to the second feature based on the self-attention layer connected to the last hidden layer.

13. The apparatus according to any one of claims 9 to 12, characterized in that, The convolutional neural network includes three parallel convolutional layers, and the gated recurrent unit includes three hidden layers.

14. The apparatus according to claim 9, characterized in that, The acquisition module includes: The first acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process and the time information corresponding to the parameter values; The mapping submodule is used to map the parameter value to numeric data to obtain the numeric data corresponding to the controllable parameter and the time information corresponding to the numeric data. The first generation submodule is used to filter out the data of the numerical type corresponding to each of the time information to obtain the first vector corresponding to each of the time information. The second generation submodule is used to calculate the statistical characteristics of each controllable parameter based on the numerical data corresponding to each controllable parameter, and to form a first vector set. The third generation submodule is used to sort the parameter values ​​of multiple controllable parameters according to the time information to obtain the corresponding second vector set of the multiple controllable parameters.

15. The apparatus according to claim 9, characterized in that, The acquisition module includes: The second acquisition submodule is used to acquire the parameter values ​​of controllable parameters during the vapor deposition process, as well as the time and position information corresponding to the parameter values, and generate a first vector set and a second vector set corresponding to each position information. The correction module is also used for: The target location information is determined from various location information. Based on the predicted value of the thickness correction value corresponding to the target location information, adjust the parameter value of the controllable parameter at the location corresponding to the target location information.

16. The apparatus according to claim 9, characterized in that, The correction module includes: The first adjustment submodule is used to adjust the parameter value of the controllable parameter according to the predicted value of the thickness correction value; The second adjustment submodule is used to obtain the current thickness of the film layer formed by the evaporation process, and when the current thickness is within a preset first range, adjust the parameter value of the controllable parameter according to the difference between the current thickness and the target thickness, and then execute the step of adjusting the parameter value of the controllable parameter according to the predicted value of the thickness correction value again.

17. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

19. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images