Micro-LED digital headlamp manufacturing method based on bonding after photoetching filling
Through the method of bonding after lithography filling and optimization of process parameters by optimizing process parameters by multi-layer perceptron model, the problem of device damage in Micro-LED digital headlight manufacturing is solved, the yield and production efficiency are improved, and the cost is reduced.
Patent Information
- Application Number
- CN202510477706.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-16
- Publication Date
- 2025-08-08
AI Technical Summary
In the manufacturing process of Micro-LED digital headlights, the traditional bonding and then peeling process causes damage to the device, affecting the reliability and yield of the finished product.
The bonding method after lithography filling is adopted, supporting materials are prefilled in the pixel voids, and all pixels are exposed through the lithography process, and then bonded according to the optimized process parameters. Finally, laser stripping is used to remove the substrate, and the process parameters are optimized in combination with the multi-layer perceptron model.
Significantly improves the reliability and yield of the device, simplifies production steps, improves production efficiency and reduces costs.
Smart Images
Figure CN120456692A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of light-emitting diode manufacturing processes, and in particular to a method for manufacturing a Micro-LED digital headlight based on photolithography-filled and then bonded. Background Art
[0002] Micro-LED-based digital headlights, with their high brightness, precise beam control, and excellent response speed, are becoming a key technology for next-generation smart lighting and in-vehicle displays. However, in the manufacturing process of Micro-LED digital headlights, the sapphire substrate of the Micro-LEDs must be peeled off to reduce optical crosstalk. This process typically uses laser lift-off technology, but the lack of structural support between pixels can lead to device damage during the lift-off process, compromising the reliability of the finished product. Summary of the Invention
[0003] Aiming at the low yield and stability of finished products in the existing Micro-LED digital headlight manufacturing process due to the traditional "bonding first and then peeling" process, a Micro-LED digital headlight manufacturing method based on photolithography filling and then bonding is proposed.
[0004] The technical solution of the present invention is: a method for manufacturing Micro-LED digital headlights based on photolithography filling and bonding, which pre-fills the pixel gaps with supporting materials, then exposes all pixels through a photolithography process, and then performs bonding operations according to optimized process parameters. Finally, laser stripping is used to remove one side of the substrate of the bonded Micro-LED chip to complete the manufacturing of the Micro-LED digital headlights.
[0005] Furthermore, the method for obtaining the optimized process parameters is as follows: S1. Obtain process parameters related to bonding quality parameters and the value range of each process parameter, randomly combine process parameters within the corresponding value range, collect bonding quality results under different process parameter conditions through experiments, and generate a data set; S2. Construct a multi-layer perceptron model, learn the rules between the bonding process parameters and the bonding quality in the data set of step S1, and convert the relationship between the bonding process parameters and the bonding quality results into a mathematical model to provide data-driven guidance for optimizing the bonding process; S2-1. The multi-layer perceptron consists of an input layer, a hidden layer and an output layer; S2-2. The process parameters input into the input layer include but are not limited to bonding temperature, bonding pressure, bonding time, and adhesive height parameters; S2-3. The hidden layer contains several nodes, each node represents a neuron, and is connected to the nodes of the input layer or the previous layer through weights and bias parameters. The weights and bias parameters are continuously optimized and adjusted through training; S2-4. The output layer includes but is not limited to bonding yield, brightness, and AOI result parameters; S3. Based on the data set in step S1, the multilayer perceptron model constructed in step S2 is trained, and the weights and biases of the model are optimized using a stochastic steepest descent method, an adaptive moment estimation method, or an impulse algorithm; through iterative optimization, the loss function value of the model meets the preset requirements, and the loss function is used to characterize the error between the predicted value and the actual value of the process parameter; S4. According to the multi-layer perceptron model trained in step S3, an initial process parameter set is set, where the initial process parameter set includes the selected process parameter type and its value range; the process parameter set is input into the multi-layer perceptron model to obtain a corresponding real-time prediction result, and the error between the predicted value and the target value is calculated; S5. Based on the prediction error in step S4, the Bayesian optimization algorithm is used for iterative search to obtain the optimal process parameter combination in the process parameter set of step S4, and the combination is sent to the multi-layer perceptron model for error verification until the optimal process parameter combination that meets the requirements is found.
[0006] The specific steps of the Micro-LED digital headlight manufacturing method based on photolithography filling and bonding are as follows: 1) Apply a layer of filler evenly on the driver substrate or Micro-LED chip with a single pixel fixed on it, so that the filler completely covers the pixel on it; 2) Use photolithography to expose the pixel, which is lower than the surrounding filler; 3) Bonding the unfilled driver substrate or Micro-LED chip on the other side to the Micro-LED chip or driver substrate processed in step 2); the bonding process parameters are the optimal bonding process parameter combination obtained; 4) Use laser stripping to remove the unfilled driver substrate or Micro-LED chip substrate.
[0007] Furthermore, the bonding process parameter combination includes a bonding pressure ranging from 1 to 4 kN, a bonding temperature ranging from 20 to 300° C., and a bonding time ranging from 1 to 60 min.
[0008] A method for obtaining optimal bonding process parameters, the specific steps are as follows: S1. Obtain process parameters related to bonding quality parameters and the value range of each process parameter, randomly combine process parameters within the corresponding value range, collect bonding quality results under different process parameter conditions through experiments, and generate a data set; S2. Construct a multi-layer perceptron model, learn the rules between the bonding process parameters and the bonding quality in the data set of step S1, and convert the relationship between the bonding process parameters and the bonding quality results into a mathematical model to provide data-driven guidance for optimizing the bonding process; S2-1. The multi-layer perceptron consists of an input layer, a hidden layer and an output layer; S2-2. The process parameters input into the input layer include but are not limited to bonding temperature, bonding pressure, bonding time, and adhesive height parameters; S2-3. The hidden layer contains several nodes, each node represents a neuron, and is connected to the nodes of the input layer or the previous layer through weights and bias parameters. The weights and bias parameters are continuously optimized and adjusted through training; S2-4. The output layer includes but is not limited to bonding yield, brightness, and AOI result parameters; The multilayer perceptron model is trained based on the data set in step S1, and the weights and biases of the model are optimized using the stochastic steepest descent method, the adaptive moment estimation method, or the impulse algorithm. Through iterative optimization, the loss function value of the model meets the preset requirements. The loss function is used to characterize the error between the predicted value and the actual value of the process parameter. S4. According to the multi-layer perceptron model trained in step S3, an initial process parameter set is set, where the initial process parameter set includes the selected process parameter type and its value range; the process parameter set is input into the multi-layer perceptron model to obtain a corresponding real-time prediction result, and the error between the predicted value and the target value is calculated; S5. Based on the prediction error in step S4, the Bayesian optimization algorithm is used for iterative search to obtain the optimal process parameter combination in the process parameter set of step S4, and the combination is sent to the multi-layer perceptron model for error verification until the optimal process parameter combination that meets the requirements is found.
[0009] A chip bonding method on a wafer, which pre-fills the chip gaps with support material, then uses a photolithography process to expose all chips, then performs bonding operations according to optimized process parameters, and finally uses laser stripping to remove one side of the bonded chip substrate to complete the chip bonding.
[0010] The beneficial effects of the present invention are as follows: the present invention is based on the method for manufacturing Micro-LED digital headlights based on photolithography filling and bonding. By using the photolithography process to fill the pixel gaps before laser stripping, the support structure of the epitaxial material is enhanced, the problem of device damage during the stripping process is effectively avoided, and the overall reliability of the device is significantly improved; this process scheme reduces the defective rate of finished products in the traditional "bonding first and then stripping" process by providing structural support before stripping, thereby improving the yield and consistency of the production process and ensuring higher production efficiency; filling the support material through the photolithography process and performing subsequent bonding simplifies the production steps, avoids complex post-processing processes, effectively improves production efficiency, and reduces production costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1 This is a diagram of the multi-layer perceptron deep learning model in the method of the present invention; Figure 2 This is a schematic diagram of the implementation steps of the method for manufacturing a micro-LED digital headlight based on photolithography filling and bonding according to the present invention. DETAILED DESCRIPTION
[0012] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0013] This invention proposes a "photolithography-first, filling-then-bonding" process. Using photolithography, support material is pre-filled in the pixel gaps. All pixels are then exposed using photolithography, followed by bonding. Finally, laser stripping is used to remove one side of the bonded Micro-LED chip's substrate, completing the fabrication of Micro-LED digital headlights. To enhance the optimization of the bonding process, this method utilizes a multilayer perceptron (MLP) model. By inputting core process parameters such as temperature, time, and pressure, this model predicts and dynamically adjusts the optimal bonding settings in real time, ensuring high precision, stability, and efficiency of the bonding operation. The addition of filler does not affect the final stripping step, and performing filling as the first step ensures the stability of all subsequent operations. The original process does not include filler bonding, requiring a glue potting step for the bonded sample. However, this invention prepares the filler directly before bonding, eliminating the glue potting step after bonding.
[0014] Step 1: Use a multilayer perceptron (MLP) to learn patterns from historical process data and transform the relationship between process parameters and bonding results into a mathematical model, providing data-driven guidance for process optimization. The input layer includes but is not limited to parameters such as bonding temperature, bonding pressure, bonding time, and adhesive height. The hidden layer contains several nodes, each representing a neuron, connected to nodes in the input layer or the previous layer through weights and bias parameters. The output layer includes but is not limited to result parameters such as bonding yield, brightness, and AOI. The specific implementation steps are as follows: S1. Obtain process parameters related to bonding quality parameters and the value range of each process parameter, randomly combine process parameters within the corresponding value range, collect bonding quality results under different process parameter conditions through experiments, and generate a data set; S2. Construct a multilayer perceptron model to learn the patterns between bonding process parameters and bonding quality in the data set of step S1, and convert the relationship between bonding process parameters and bonding quality results into a mathematical model to provide data-driven guidance for optimizing the bonding process; 1) The multilayer perceptron model consists of an input layer, a hidden layer, and an output layer; 2) The process parameters input into the input layer include but are not limited to bonding temperature, bonding pressure, bonding time, adhesive height, and other parameters; 3) The hidden layer contains several nodes, each node represents a neuron, and is connected to the nodes of the input layer or the previous layer through weights and bias parameters, and the weights and bias parameters are continuously optimized and adjusted through training; 4) The output layer includes but is not limited to result parameters such as bonding yield, brightness, and AOI; S3. Based on the data set in step S1, the multilayer perceptron model constructed in step S2 is trained, and the weights and biases of the model are optimized using a stochastic steepest descent method, an adaptive moment estimation method, or an impulse algorithm; through iterative optimization, the loss function value of the model meets the preset requirements, and the loss function is used to characterize the error between the predicted value and the actual value of the process parameter; S4. According to the multi-layer perceptron model trained in step S3, an initial process parameter set is set, where the initial process parameter set includes the selected process parameter type and its value range; the process parameter set is input into the multi-layer perceptron model to obtain a corresponding real-time prediction result, and the error between the predicted value and the target value is calculated; S5. Based on the prediction error in step S4, the Bayesian optimization algorithm is used for iterative search to obtain the optimal process parameter combination in the process parameter set of step S4, and the combination is sent to the multi-layer perceptron model for error verification until the optimal bonding process parameter combination that meets the requirements is searched; 1) the multiple errors corresponding to the multiple sets of initial process parameter sets obtained are substituted into the proxy function, and the updated prior distribution of the initial sample or the previous sample is calculated; 2) an acquisition function is set to optimize a new set of process parameter combinations according to the prior distribution, so that the acquisition function balances the proportion of exploration and utilization; 3) the new process parameter combination is input into the multi-layer perceptron model to predict the new quality parameters, and then the target value of the quality parameters is used to determine whether the design requirements are met; if so, the current process parameter combination is output as the optimal solution; if not, the process returns to step S4 to update the initial process parameter set, and steps S4 and S5 are repeated until the error between the real-time prediction value and the target value meets the design requirements, and the optimal solution process parameter combination is obtained.
[0015] Step 2: Apply a layer of filler evenly on the driver substrate or Micro-LED chip with a single pixel fixed on it, so that the filler completely covers the pixel on it.
[0016] Step 3: Use photolithography to expose the pixel points, which are now lower than the surrounding filler. (This process is only performed on one side of the sample, either the driver substrate or the Micro-LED chip.) Step 4: Bond the unfilled driver substrate or Micro-LED chip on the other side to the Micro-LED chip or driver substrate treated in Step 3. The bonding process parameters are the optimal combination of bonding process parameters obtained in Step 1. The bonding pressure range is 1-4 kN, the bonding temperature range is 20-300°C, and the bonding time range is 1-60 minutes. (Note: For the filled side, the filler will be squeezed into the gap of the unfilled sample on the other side due to heating and extrusion.) Step 5: Use laser lift-off to remove the unfilled driver substrate or Micro-LED chip substrate. (This step only removes the sapphire substrate of the LED chip) By pre-filling the spaces between pixels with support material before laser lift-off, this method not only effectively enhances the mechanical stability of the epitaxial structure after sapphire lift-off, but also significantly improves overall reliability. Compared to the traditional "bonding first, then lift-off" process, the present invention's "photolithography filling first, then bonding" approach improves the yield of finished products. To enhance the optimization level of the bonding process, this method utilizes a multilayer perceptron (MLP) model. By inputting core process parameters such as temperature, time, and pressure, it enables real-time prediction and dynamic adjustment of optimal bonding process parameter settings, thereby ensuring high precision, stability, and efficiency of the bonding operation. This method can improve the yield of subsequent processes, simplify the production process, and enhance overall production efficiency. It provides an innovative and feasible technical solution for the stable and efficient manufacturing of Micro-LED digital headlights.
[0017] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.
Claims
1. A method for manufacturing a Micro-LED digital headlight based on photolithography filling and bonding, characterized in that: The pixel gaps are pre-filled with supporting material, and then all pixels are exposed through a photolithography process. Bonding operations are then performed according to optimized process parameters. Finally, laser stripping is used to remove one side of the substrate of the bonded Micro-LED chip to complete the manufacturing of the Micro-LED digital headlights.
2. The method for manufacturing a Micro-LED digital headlight based on photolithography filling and bonding according to claim 1, characterized in that: The method for obtaining the optimized process parameters is as follows: S1. Obtain process parameters related to bonding quality parameters and the value range of each process parameter, randomly combine process parameters within the corresponding value range, collect bonding quality results under different process parameter conditions through experiments, and generate a data set; S2. Construct a multi-layer perceptron model, learn the rules between the bonding process parameters and the bonding quality in the data set of step S1, and convert the relationship between the bonding process parameters and the bonding quality results into a mathematical model to provide data-driven guidance for optimizing the bonding process; S2-1. The multi-layer perceptron consists of an input layer, a hidden layer and an output layer; S2-2. The process parameters input into the input layer include but are not limited to bonding temperature, bonding pressure, bonding time, and adhesive height parameters; S2-3. The hidden layer contains several nodes, each node represents a neuron, and is connected to the nodes of the input layer or the previous layer through weights and bias parameters. The weights and bias parameters are continuously optimized and adjusted through training; S2-4. The output layer includes but is not limited to bonding yield, brightness, and AOI result parameters; Combined with the data set in step S1, the multilayer perceptron model is trained, and the weights and biases of the model are optimized using the stochastic steepest descent method, the adaptive moment estimation method, or the impulse algorithm; Through iterative optimization, the loss function value of the model meets the preset requirements. The loss function is used to characterize the error between the predicted value and the actual value of the process parameter; S4. According to the multi-layer perceptron model trained in step S3, an initial process parameter set is set, where the initial process parameter set includes the selected process parameter type and its value range; the process parameter set is input into the multi-layer perceptron model to obtain a corresponding real-time prediction result, and the error between the predicted value and the target value is calculated; S5. Based on the prediction error in step S4, the Bayesian optimization algorithm is used for iterative search to obtain the optimal process parameter combination in the process parameter set of step S4, and the combination is sent to the multi-layer perceptron model for error verification until the optimal process parameter combination that meets the requirements is found.
3. The method for manufacturing a Micro-LED digital headlight based on photolithography filling and bonding according to claim 2, characterized in that: The specific steps are as follows: 1) Apply a layer of filler evenly on the driver substrate or Micro-LED chip with a single pixel fixed on it, so that the filler completely covers the pixel on it; 2) Use photolithography to expose the pixel, which is lower than the surrounding filler; 3) Bonding the unfilled driver substrate or Micro-LED chip on the other side to the Micro-LED chip or driver substrate processed in step 2); the bonding process parameters are the optimal bonding process parameter combination obtained; 4) Use laser stripping to remove the unfilled driver substrate or Micro-LED chip substrate.
4. The method for manufacturing a Micro-LED digital headlight based on photolithography filling and bonding according to claim 2 or 3, characterized in that: The bonding process parameter combination includes a bonding pressure ranging from 1 to 4 kN, a bonding temperature ranging from 20 to 300° C., and a bonding time ranging from 1 to 60 min.
5. A method for obtaining optimal bonding process parameters, characterized in that: The specific steps are as follows: S1. Obtain process parameters related to bonding quality parameters and the value range of each process parameter, randomly combine process parameters within the corresponding value range, collect bonding quality results under different process parameter conditions through experiments, and generate a data set; S2. Construct a multi-layer perceptron model, learn the rules between the bonding process parameters and the bonding quality in the data set of step S1, and convert the relationship between the bonding process parameters and the bonding quality results into a mathematical model to provide data-driven guidance for optimizing the bonding process; S2-1. The multi-layer perceptron consists of an input layer, a hidden layer and an output layer; S2-2. The process parameters input into the input layer include but are not limited to bonding temperature, bonding pressure, bonding time, and adhesive height parameters; S2-3. The hidden layer contains several nodes, each node represents a neuron, and is connected to the nodes of the input layer or the previous layer through weights and bias parameters. The weights and bias parameters are continuously optimized and adjusted through training; S2-4. The output layer includes but is not limited to bonding yield, brightness, and AOI result parameters; S3. Based on the data set in step S1, the multilayer perceptron model constructed in step S2 is trained, and the weights and biases of the model are optimized using the stochastic steepest descent method, the adaptive moment estimation method, or the impulse algorithm; Through iterative optimization, the loss function value of the model meets the preset requirements. The loss function is used to characterize the error between the predicted value and the actual value of the process parameter; S4. According to the multi-layer perceptron model trained in step S3, an initial process parameter set is set, where the initial process parameter set includes the selected process parameter type and its value range; the process parameter set is input into the multi-layer perceptron model to obtain a corresponding real-time prediction result, and the error between the predicted value and the target value is calculated; S5. Based on the prediction error in step S4, the Bayesian optimization algorithm is used for iterative search to obtain the optimal process parameter combination in the process parameter set of step S4, and the combination is sent to the multi-layer perceptron model for error verification until the optimal process parameter combination that meets the requirements is found.
6. A chip-on-wafer bonding method, characterized in that: The chip gaps are pre-filled with supporting materials, and then all chips are exposed through a photolithography process. The bonding operation is then performed according to the optimized process parameters. Finally, laser stripping is used to remove one side of the bonded chip substrate to complete the chip bonding.