Uniformity control method and device for large-area laser annealing

Abstract

The invention relates to the technical field of automatic control, and discloses a uniformity control method and device for large-area laser annealing, and the method comprises the steps: obtaining a three-dimensional height distribution diagram of a substrate, dividing sub-regions through a self-adaptive segmentation algorithm, and obtaining a height mean value and boundary coordinates; and calculating a focal length adjustment amount according to the height mean value to generate a focus parameter table. And through laser test comparison, dynamic focusing is carried out until the focus offset is smaller than a threshold value, and corrected focus data are obtained. Energy transition features are extracted, and a scanning path is generated with minimum energy gradient constraints. And controlling the laser beam according to the path, collecting the energy distribution signal, detecting the initial annealing depth, adjusting the sub-region laser parameters which do not reach the consistency, smoothing the annealing characteristics of the edge region, and generating a control instruction which meets the uniform tolerance. According to the method, the height distribution of the whole area of the substrate is measured, the substrate is divided into different sub-areas, energy transition between the sub-areas is smoothed, and uniform control over laser annealing is achieved.

Description

Technical Field

[0001] The present invention relates to the field of automation control technology, and particularly to a method and device for controlling the uniformity of large-area laser annealing. Background Art

[0002] In the field of semiconductor manufacturing, the laser annealing technology is a key process for improving the crystallization quality and electrical properties of materials. However, in practical applications, there are often many complex situations such as highly uneven surfaces and material property differences on the semiconductor substrate surface, which pose great challenges to achieving a uniform annealing effect. Currently, the laser annealing technology commonly used in the industry is a technology based on a fixed focal length and a preset scanning path. This technology attempts to achieve a relatively uniform annealing effect in different regions by adjusting the laser power and scanning speed.

[0003] However, due to the height differences on the substrate surface, the laser system with a fixed focal length cannot dynamically adapt to different heights, resulting in uneven laser energy distribution. Some regions may overheat due to excessive energy, while other regions may not be annealed sufficiently due to insufficient energy, seriously affecting the consistency of the annealing depth. In addition, the material properties of different regions of the substrate are different, and their absorption and response to laser energy are also different. The traditional fixed parameter settings are difficult to take into account the annealing requirements of all regions and cannot achieve an accurate annealing effect. More intractably, between adjacent regions, the transition of laser energy is often not smooth enough, easily causing edge effects. Such edge effects not only lead to non-uniform local annealing but may also introduce additional stress and defects, further reducing the performance and reliability of semiconductor devices.

[0004] In summary, the existing technology cannot achieve precise and uniform control of the laser annealing process when dealing with complex substrate surfaces and is difficult to meet the annealing requirements of all regions at the same time. Summary of the Invention

[0005] The present invention provides a method and device for controlling the uniformity of large-area laser annealing. By measuring the height distribution of the entire substrate area, different sub-regions are divided on the substrate and the energy transition between regions is smoothed to achieve uniform control of laser annealing.

[0006] In the first aspect, to solve the above technical problems, the present invention provides a method for controlling the uniformity of large-area laser annealing, including: Obtaining a three-dimensional height distribution map of the entire substrate area; Based on the three-dimensional height distribution map, dividing the substrate into multiple sub-regions by using an adaptive segmentation algorithm, and obtaining the height mean value and boundary coordinates of each sub-region; Calculating the focal length adjustment amount of the laser focus in each sub-region according to the height mean value, and generating a focus parameter table; Determine the actual focal position through laser testing and compare it with the focal parameter table to obtain the focal shift of each sub-region; Adjust the position of the laser lens through the dynamic focusing method until the focal shift is less than the preset position tolerance threshold, and obtain the corrected focal distribution data; Extract the energy transition characteristics between sub-regions from the corrected focal distribution data, and use the path optimization algorithm to generate the optical scanning path trajectory with the minimum energy gradient as the constraint; Control the movement of the laser beam based on the optical scanning path trajectory data and collect the real-time energy distribution signal; Detect the initial annealing depth from the real-time energy distribution signal, adjust the laser power and scanning speed of the sub-regions where the initial annealing depth does not meet the preset consistency tolerance threshold, and collect the updated annealing depth; Extract the annealing characteristics of the edge region from the updated annealing depth, smooth the annealing characteristics of the edge region between sub-regions, and generate a control instruction with the annealing depth uniformity of the whole region meeting the preset uniformity tolerance.

[0007] In an alternative embodiment, the obtaining of the three-dimensional height distribution map of the entire substrate area includes: Scan the entire area with a measuring device to obtain an initial height set; Process the initial height set through a smoothing algorithm to obtain smoothed height data; Apply an interpolation algorithm to the smoothed height data to generate a continuous three-dimensional height distribution map.

[0008] In an alternative embodiment, the dividing of the substrate into multiple sub-regions based on the three-dimensional height distribution map and obtaining the height mean value and boundary coordinates of each sub-region includes: Use an adaptive segmentation algorithm to process the three-dimensional height distribution map according to a preset segmentation threshold to obtain multiple initial sub-regions; Calculate the height mean value within the initial sub-region according to the three-dimensional height distribution map to obtain a preliminary height mean value set; Process the preliminary height mean value set through a mean filtering algorithm to obtain a smoothed height mean value set; Determine the height difference between the initial sub-regions according to the smoothed height mean value set to obtain a height difference distribution; Adjust the segmentation threshold according to the height difference distribution, and use the adaptive segmentation algorithm to re-divide the sub-regions to obtain optimized sub-regions; Extract the boundary coordinates and height mean value of the optimized sub-region according to the three-dimensional height distribution map.

[0009] In an alternative embodiment, calculating the focal length adjustment amount of the laser focus in each sub-region according to the height mean value and generating a focus parameter table includes: Calculating the difference between the height corresponding to the reference focal length according to the height mean value to obtain an initial focal length adjustment value, where the reference focal length is a preset laser focal length; Processing the initial focal length adjustment value by using a mean filtering algorithm to obtain a filtered focal length adjustment value; Repairing the filtered focal length adjustment value by using an interpolation method to obtain an optimized focal length adjustment set; Determining the adjustment direction of the laser focus in each sub-region according to the optimized focal length adjustment set, obtaining a focus change distribution, and generating a corresponding focus parameter table.

[0010] In an alternative embodiment, extracting the energy transition characteristics between sub-regions from the corrected focus distribution data and generating an optical scanning path trajectory with the minimum energy gradient as a constraint by using a path optimization algorithm includes: Calculating the mean pixel intensity of each sub-region from the corrected focus distribution data to obtain an energy value; Determining the change trend of the energy transition by using the gradient descent method according to the energy value; Generating an initial path trajectory by using a path optimization algorithm according to the change trend; Extracting the position with the maximum pixel intensity change rate as the key point coordinates of the optical scanning according to the initial path trajectory and determining the scanning order; Generating a final optical scanning path trajectory according to the scanning order and the key point coordinates.

[0011] In an alternative embodiment, detecting the initial annealing depth from the real-time energy distribution signal, adjusting the laser power and scanning speed of the sub-regions where the initial annealing depth does not meet the preset consistency tolerance threshold, and collecting the updated annealing depth includes: Real-time collecting the energy distribution signal through a temperature sensor to detect the initial annealing depth; If the initial annealing depth does not meet the preset consistency tolerance threshold, identifying the corresponding inconsistent sub-regions; Calculating the laser power adjustment amount and the scanning speed adjustment amount according to the inconsistent sub-regions to obtain adjustment parameters; Updating the laser power and scanning speed by using the adjustment parameters and recording the updated energy distribution to obtain the updated annealing depth.

[0012] In an alternative embodiment, extracting the annealing characteristics of the edge region from the updated annealing depth, smoothing the annealing characteristics of the edge region between sub-regions, and generating a control instruction with the annealing depth uniformity of the whole region meeting the preset uniformity tolerance includes: Extract the edge region features from the updated annealing depth using the PCA feature extraction method; Smoothly adjust the edge region features to obtain a smooth annealing feature distribution; Determine whether the smooth annealing feature distribution is greater than a preset uniform tolerance. If so, adjust the calculation deviation amount for the sub-region. If not, do nothing and determine the adjusted feature parameters; Generate an annealing depth distribution according to the adjusted feature parameters, and obtain the constrained annealing features under the uniformity constraint through inter-region smoothing processing; Extract the full-region control parameters from the constrained annealing features to generate control instructions.

[0013] In a second aspect, the present invention provides a uniformity control device for large-area laser annealing, including: A three-dimensional height distribution acquisition module for acquiring a three-dimensional height distribution map of the entire region of the substrate; An adaptive segmentation module for dividing the substrate into multiple sub-regions based on the three-dimensional height distribution map using an adaptive segmentation algorithm, and obtaining the height mean and boundary coordinates of each sub-region; A focal length adjustment amount calculation module for calculating the focal length adjustment amount of the laser focus in each sub-region according to the height mean, and generating a focus parameter table; A focus offset comparison module for determining the actual focus position through laser testing and comparing it with the focus parameter table to obtain the focus offset of each sub-region; A dynamic focusing module for adjusting the position of the laser lens through a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and obtaining corrected focus distribution data; A path optimization module for extracting the energy transition features between sub-regions from the corrected focus distribution data, and generating an optical scanning path trajectory with the minimum energy gradient as the constraint using a path optimization algorithm; A laser beam control module for controlling the movement of the laser beam based on the optical scanning path trajectory data and collecting real-time energy distribution signals; An annealing depth adjustment module for detecting the initial annealing depth from the real-time energy distribution signal, adjusting the laser power and scanning speed of the sub-regions where the initial annealing depth does not meet the preset consistency tolerance threshold, and collecting the updated annealing depth; An annealing feature smoothing module for extracting the edge region annealing features from the updated annealing depth, smoothing the edge region annealing features between sub-regions, and generating control instructions with the full-region annealing depth uniformity meeting the preset uniform tolerance.

[0014] In a third aspect, the present invention further provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the uniformity control method of large-area laser annealing described in any one of the above is implemented.

[0015] In a fourth aspect, the present invention further provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program. Wherein, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the uniformity control method of large-area laser annealing described in any one of the above.

[0016] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: The present invention discloses a method for controlling the uniformity of laser annealing depth, including: By obtaining the three-dimensional height distribution map of the substrate and dividing sub-regions, the laser focus is accurately adjusted according to the height mean value of each sub-region, effectively solving the problem of uneven annealing caused by uneven surface height of the substrate; The path optimization algorithm generates a scanning path with the minimum energy gradient as a constraint, ensuring smooth energy transition between adjacent sub-regions and avoiding edge effects caused by energy mutation, thereby significantly improving the overall annealing uniformity; The dynamic focusing method combines real-time energy distribution signal acquisition, can adjust the laser focus position and power parameters in real time, ensure that the annealing depth of each sub-region meets the preset requirements, significantly improve the processing accuracy, and effectively solve the problem of uneven local annealing caused by edge effects in the traditional method through the extraction and smoothing processing of the annealing characteristics in the edge region, further improving the processing quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the uniformity control method of large-area laser annealing of the present invention.

[0018] Figure 2 It is a structural schematic diagram of the uniformity control method and device of large-area laser annealing of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] Refer to Figure 1, the method and device for controlling the uniformity of large-area laser annealing in this embodiment may specifically include: S101. Obtain a three-dimensional height distribution map of the entire area of the substrate; S102. Based on the three-dimensional height distribution map, divide the substrate into multiple sub-regions by using an adaptive segmentation algorithm, and obtain the height mean value and boundary coordinates of each sub-region; S103. Calculate the focal length adjustment amount of the laser focus in each sub-region according to the height mean value, and generate a focus parameter table; S104. Determine the actual focus position through laser testing and compare it with the focus parameter table to obtain the focus offset of each sub-region; S105. Adjust the position of the laser lens through a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and obtain corrected focus distribution data; S106. Extract the energy transition characteristics between sub-regions from the corrected focus distribution data, and use a path optimization algorithm to generate an optical scanning path trajectory with the minimum energy gradient as the constraint; S107. Control the movement of the laser beam based on the optical scanning path trajectory data, and collect real-time energy distribution signals; S108. Detect the initial annealing depth from the real-time energy distribution signal, adjust the laser power and scanning speed of the sub-regions where the initial annealing depth does not meet the preset consistency tolerance threshold, and collect the updated annealing depth; S109. Extract the annealing characteristics of the edge region from the updated annealing depth, smooth the annealing characteristics of the edge region between sub-regions, and generate a control instruction for the uniformity of the annealing depth of the entire region to meet the preset uniformity tolerance.

[0021] In step S101, obtaining a three-dimensional height distribution map of the entire area of the substrate includes: Scan the entire area with a measuring device to obtain an initial height set; Process the initial height set through a smoothing algorithm to obtain smoothed height data; For the smoothed height data, apply an interpolation algorithm to generate a continuous three-dimensional height distribution map.

[0022] It should be noted that the initial height set is a set of the heights of each point of the substrate measured by a scanning measurement device.

[0023] In one implementation, when processing the initial set of heights by scanning, a smoothing algorithm can be applied to optimize the data quality. The core of the smoothing algorithm lies in reducing noise interference while preserving the true surface topography features. The mean filter method is adopted, and the average value within a 5×5 region around each data point is taken as the new value for processing. Suppose the original height of a certain point is 10μm, and there are noise points as high as 15μm in its neighborhood. After mean filtering, it may be smoothed to 9.8μm. This method is simple and efficient, and can effectively filter out the mutation points caused by equipment jitter or environmental interference.

[0024] Preferably, the smoothed data is more suitable for subsequent analysis because it reduces the interference of local anomalies on the overall trend.

[0025] It should be noted that for the smoothed height data, an interpolation algorithm is used to generate a continuous three-dimensional height distribution map, and this process is the key to converting discrete points into a smooth surface.

[0026] In the embodiment of the present invention, the Kriging interpolation method is used, which predicts the unknown region based on spatial correlation.

[0027] Suppose the distribution of smoothed height data points in a certain area of the substrate is: (0,0,5μm), (1,0,6μm), (0,1,5.5μm). Through Kriging interpolation, the height of the middle point (0.5,0.5) can be estimated to be approximately 5.7μm. This method takes into account the spatial weights of the data points and can generate a continuous surface that conforms more to physical laws. It can be understood that compared with simple linear interpolation, Kriging interpolation is more accurate when dealing with complex surfaces, especially suitable for the tiny bumps and depressions that may exist on the substrate.

[0028] In step S102, based on the three-dimensional height distribution map, an adaptive segmentation algorithm is used to divide the substrate into multiple sub-regions, and the height mean value and boundary coordinates of each sub-region are obtained, including: Using the adaptive segmentation algorithm to process the three-dimensional height distribution map according to a preset segmentation threshold to obtain multiple initial sub-regions; According to the three-dimensional height distribution map, calculate the height mean value within the initial sub-region to obtain a preliminary height mean value set. The preliminary height mean value set is processed by a mean filter algorithm to obtain a smoothed height mean value set; According to the smoothed height mean value set, determine the height difference between the initial sub-regions to obtain a height difference distribution. Adjust the segmentation threshold according to the height difference distribution, and use the adaptive segmentation algorithm to re-divide the sub-regions to obtain optimized sub-regions; According to the three-dimensional height distribution map, extract the boundary coordinates and height mean value of the optimized sub-region.

[0029] Exemplarily, an adaptive segmentation algorithm is used to process the three-dimensional height distribution map according to a preset segmentation threshold to obtain multiple initial sub-regions. The core of this process lies in decomposing complex three-dimensional data into manageable parts.

[0030] According to the three-dimensional height distribution map, calculate the height mean value within each initial sub-region to obtain a preliminary height mean value set. This step evaluates the average height of each sub-region by statistically analyzing the height data within the sub-region.

[0031] Process the preliminary height mean value set through a mean filtering algorithm to obtain a smoothed height mean value set. Mean filtering can effectively reduce the influence of noise on the height mean value and make the data smoother.

[0032] According to the smoothed height mean value set, calculate the height difference between adjacent sub-regions to obtain a height difference distribution. The height difference distribution is a composite data structure containing three data: two comparison regions, whether they are adjacent, and the height difference. This step is used to evaluate the height change situation between sub-regions.

[0033] Adjust the preset segmentation threshold according to the height difference distribution and reapply the adaptive segmentation algorithm to obtain optimized sub-regions. This step ensures that the sub-region division is more in line with the actual height change of the substrate surface and improves the accuracy of segmentation.

[0034] Extract the boundary coordinates and height mean value of each sub-region from the optimized sub-regions. These data will be used for subsequent laser annealing processing to ensure the accuracy of the processing path and parameter settings.

[0035] Specifically, assume that the initial threshold is set to 20um. According to the initial threshold, an adaptive segmentation algorithm is used to obtain multiple initial sub-regions. After smoothing and calculating the height difference distribution, it may be found that the height differences between many adjacent sub-regions are less than the preset minimum height difference. Then, modify the threshold to 21um and reapply the adaptive segmentation algorithm to obtain multiple sub-regions, and so on, until the height difference between adjacent sub-regions is greater than the preset minimum height difference.

[0036] It should be noted that the minimum height difference is determined according to the maximum acceptable processing error. In the embodiment of the present invention, the preset minimum height difference is 2um.

[0037] In step S103, calculating the focal length adjustment amount of the laser focus in each sub-region according to the height mean value and generating a focus parameter table includes: According to the height mean value, calculate the difference from the height corresponding to the reference focal length to obtain an initial focal length adjustment value, where the reference focal length is a preset laser focal length; Process the initial focal length adjustment value through a mean filtering algorithm to obtain a filtered focal length adjustment value. Repair the filtered focal length adjustment value through interpolation to obtain an optimized focal length adjustment set; According to the optimized focal length adjustment set, determine the adjustment direction of the laser focus in each sub-region, obtain the focus change distribution, and generate the corresponding focus parameter table.

[0038] Exemplarily, calculating the difference from the reference focal length based on the height mean is to preliminarily determine the need for laser focus adjustment.

[0039] In an embodiment of the present invention, the reference focal length is preset to 50um, and the height mean of a certain sub-region is 20um, and another is 43um. Since the height mean reflects the substrate surface characteristics and is associated with the focal length adjustment, the initial value of the focal length adjustment can be obtained through a simple difference.

[0040] For example, the height mean of 20um corresponds to an initial adjustment value of 50um - 20um = 30um, and the height mean of 43um corresponds to 7um.

[0041] In one implementation, the mean filtering algorithm is used to process the initial value of the focal length adjustment, aiming to smooth the data fluctuations.

[0042] Specifically, assume that the initial values of the focal length adjustment for three adjacent sub-regions are 30um, 7um, and 10mm. The mean filtering takes the average of the surrounding values and adjusts them to 15um, 8um, and 5um. This smoothing process can reduce the adjustment error caused by local terrain mutations and ensure that the focus change is more in line with the overall trend. The smoothed focal length adjustment set is repaired by the linear interpolation method to further optimize the data continuity.

[0043] Preferably, if the optimized value is positive, such as 15um, the focus needs to be adjusted upward; if it is negative, such as -7um, it needs to be adjusted downward.

[0044] In one embodiment, the generation of the focus parameter table is a process of structuring and outputting the above results.

[0045] It should be noted that the accuracy of the parameter table depends on the continuous optimization of the foregoing steps to ensure that the focus adjustment highly matches the terrain features.

[0046] It can be understood that the output of the focus parameter table provides a reliable basis for subsequent automated operations, especially in dynamic terrain measurement with remarkable effects.

[0047] In step S104, determine the actual focus position through laser testing and compare it with the focus parameter table to obtain the focus offset of each sub-region.

[0048] It should be noted that the focus parameter table contains the theoretically laser focus position, and the actual focus position refers to the coordinate position of the laser focus under laser testing. By calculating the difference between the actual focus position and the position of the corresponding laser focus in the focus parameter table, the focus offset of each sub-region can be obtained.

[0049] In step S105, the position of the laser lens is adjusted by a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and corrected focus distribution data is obtained.

[0050] It should be noted that the dynamic focusing method is a technology that realizes precise focusing by real-time feedback and adjustment of the lens position. Its core lies in continuously optimizing the focusing state of the optical system according to the focus offset detected in real time. The offset between the focus and the target position is monitored in real time through sensors (such as phase detection sensors, contrast sensors) or algorithms (such as contrast detection methods); the lens is driven to move by a motor or piezoelectric element to form a closed-loop feedback system of "detection - adjustment - re-detection". For example, the phase detection method divides the light into two beams through a beam splitter, calculates the offset by comparing the phase differences of the two beams of light, and then controls the lens to move to compensate for the error; when the focus offset is less than the preset position tolerance threshold (such as micron-level accuracy), the adjustment is stopped to ensure that the focus reaches a stable state.

[0051] It should be noted that the corrected focus distribution data is a set of calibration parameters generated through the dynamic focusing process, which is used to describe the compensation result of the system focusing error and the optimized focus spatial distribution characteristics. By recording the actual focus coordinates after the lens position is adjusted, the deviation caused by initial installation or environmental factors is eliminated; statistical information such as the mean and variance of the focus offset during multiple focusing processes is stored to reflect the system stability; according to the lens position adjustment parameter tables corresponding to different preset thresholds (such as 0.1μm, 0.5μm), it is used to quickly respond to similar error scenarios.

[0052] In step S106, extracting the energy transition characteristics between sub-regions from the corrected focus distribution data and generating an optical scanning path trajectory with the minimum energy gradient as a constraint by using a path optimization algorithm includes: Calculating the mean pixel intensity of each sub-region from the corrected focus distribution data to obtain an energy value; According to the energy value, using the gradient descent method to determine the change trend of the energy transition; According to the change trend, generating an initial path trajectory through a path optimization algorithm; according to the initial path trajectory, extracting the position with the largest pixel intensity change rate as the key point coordinates of the optical scan and determining the scan order; Generating the final optical scanning path trajectory according to the scan order and the key point coordinates.

[0053] It should be noted that calculating the mean pixel intensity of each sub-region from the corrected focus distribution data to obtain an energy value is essentially a process of dividing the optical scanning area into multiple sub-regions and quantifying the light intensity characteristics of each sub-region through statistical methods.

[0054] For example, in optical microscope scanning, assume that a field of view is divided into 16 sub-regions, and each sub-region contains a number of pixel points. The energy of the region can be characterized by calculating the average gray value of these pixel points.

[0055] Exemplarily, if the pixel gray value range of a certain sub-region is between 0 and 255 and the average value is 180, the energy value of this region can be regarded as relatively high, reflecting a strong light intensity distribution. The advantage of this method is that it can quickly locate the non-uniformity of the light intensity distribution and provide a data basis for subsequent optimization. Using the gradient descent method to determine the change trend of energy transition according to the energy value, this step aims to analyze the continuous change law of the energy value in space.

[0056] In one implementation, the energy value of the sub-region can be regarded as a discrete function on a two-dimensional grid. The gradient descent method finds the local extreme direction of energy change through iterative adjustment.

[0057] For example, assume that the energy values of adjacent sub-regions are 180, 150, and 120 respectively. The gradient descent method will identify the decreasing trend of energy from high to low, indicating that the light intensity gradually weakens. This trend analysis helps to understand the transmission characteristics of light energy in the optical system and lays a foundation for path planning.

[0058] Preferably, this method can also reduce misjudgment caused by noise interference and improve the robustness of trend judgment. Generate an initial path trajectory through a path optimization algorithm according to the change trend. This process is to transform the law of energy change into an executable scanning path.

[0059] In the embodiment of the present invention, the classical greedy algorithm is adopted, and the region with a large energy change is preferentially selected as the starting point of the path.

[0060] For example, if the energy value of a certain region drops suddenly from 180 to 120, the path optimization algorithm will tend to start from this region and extend along the direction with the largest gradient to form an initial trajectory. This way of path generation can ensure that the scanning process covers the regions with significant optical property changes, thereby improving the scanning efficiency and accuracy. Extract the position with the largest pixel intensity change rate as the key point coordinates of optical scanning according to the initial path trajectory. This technical topic focuses on the positioning of key feature points.

[0061] In one embodiment, the point with the largest change rate can be found by calculating the gray difference between adjacent pixels.

[0062] For example, if the pixel gray value jumps from 150 to 200 within a certain path segment and the change rate is significant, this point can be marked as a key point. These key points usually correspond to the edge or focus regions in the optical system, and extracting them can effectively characterize the target features.

[0063] It should be noted that the selection of key points directly affects the rationality of the subsequent scanning order, and its beneficial effect is to improve the coverage rate of important details in the scanning. Determining the scanning order is based on the logical arrangement of the coordinates of the key points.

[0064] In the embodiment of the present invention, the key points are sorted according to the Euclidean distance between them, and the key points with closer distances are scanned preferentially.

[0065] For example, if the distance between key points A and B is 5 pixel units and the distance between B and C is 10 pixel units, the scanning order can be arranged as A - B - C. This order design reduces the ineffective movement of the scanning head and improves the time efficiency.

[0066] It can be understood that a reasonable order can also reduce mechanical wear and extend the service life of the equipment. Generating the final optical scanning path trajectory based on the scanning order and the key point coordinates is a process of integrating the foregoing steps into a complete solution.

[0067] For example, starting from key point A, extending along the energy gradient direction to B, and then adjusting to C, finally forming a continuous scanning trajectory. This path trajectory not only covers the key feature area but also optimizes the smoothness of the scanning path.

[0068] Specifically, this method can significantly improve the image quality in high - resolution imaging, and at the same time reduce the time waste caused by path redundancy.

[0069] Preferably, by repeatedly verifying the coverage rate and efficiency of the path, the final solution can achieve a balance between precision and speed, providing reliable technical support for optical scanning.

[0070] In step S107, based on the optical scanning path trajectory data, control the movement of the laser beam and collect the real - time energy distribution signal.

[0071] It should be noted that the real - time energy distribution refers to the energy intensity distribution of the laser beam in space (such as a two - dimensional plane or a three - dimensional region) measured and recorded instantaneously by a sensor during the laser processing or detection process. Its core role is to monitor the uniformity, stability of the laser energy output and its matching degree with the target area, so as to optimize the process parameters and ensure the processing quality.

[0072] In step S108, detecting the initial annealing depth from the real - time energy distribution signal, adjusting the laser power and scanning speed of the sub - regions where the initial annealing depth does not meet the preset consistency tolerance threshold, and collecting the updated annealing depth, including: Collect the energy distribution signal in real - time through a temperature sensor to detect the initial annealing depth; If the initial annealing depth does not meet the preset consistency tolerance threshold, identify the corresponding inconsistent sub - regions; Calculate the laser power adjustment amount and the scanning speed adjustment amount according to the inconsistent sub-region to obtain adjustment parameters; Update the laser power and the scanning speed using the adjustment parameters, and record the updated energy distribution to obtain the updated annealing depth.

[0073] It should be noted that the real-time energy distribution signal is the distribution of the laser energy on the substrate surface collected in real time by the sensor, which reflects the intensity and distribution characteristics of the laser energy; the initial annealing depth is the depth reached by the current annealing process detected according to the real-time energy distribution signal; the consistency tolerance threshold is a preset standard for judging whether the annealing depth meets the requirements, that is, the allowable error range of the annealing depth; the inconsistent sub-region refers to the substrate region where the annealing depth does not reach the preset consistency tolerance threshold; the laser power adjustment amount and the scanning speed adjustment amount are the values for adjusting the laser power and the scanning speed to make the annealing depth meet the requirements; the adjustment parameter is the parameter for updating the laser power and the scanning speed according to the calculated adjustment amount.

[0074] Exemplarily, collecting the energy distribution signal in real time by the temperature sensor is to dynamically monitor the thermal effect distribution during the laser annealing process.

[0075] Exemplarily, in the semiconductor wafer annealing scenario, the temperature sensor can be arranged at the key points on the wafer surface to record the temperature changes in each region in real time and generate an energy distribution map. If the detected initial annealing depth is 50um while the preset target depth is 60um, it indicates that there is a deviation.

[0076] It should be noted that the initial annealing depth reflects the effect of the laser energy on the material. If it does not reach the consistency tolerance threshold, such as ±5um, it indicates that the energy distribution is uneven, which may lead to unstable device performance.

[0077] Specifically, the adjustment amount can be calculated according to the preset formula, and the calculation formula is as follows: ; ; where is the adjustment amount of the laser power, is the adjustment amount of the scanning speed, is the target annealing depth, is the currently detected annealing depth, and are the proportionality coefficients for converting the error of the annealing depth into the adjustment amounts of the power and the speed.

[0078] In one embodiment, after updating the laser system using the adjustment parameters, the annealing process is restarted.

[0079] For example, the adjusted laser device scans the wafer with new parameters, and the temperature sensor records the updated energy distribution. The temperature in the edge area rises to 450°C, and the temperature in the central area slightly drops to 480°C, and the overall trend is towards equilibrium.

[0080] It can be understood that such adjustment can effectively reduce the thermal gradient and improve the annealing depth consistency. After obtaining the updated annealing depth, the measurement shows that the edge reaches 58um and the center is 61um, both within the tolerance range.

[0081] In step S109, the extracting the annealing characteristics of the edge area from the updated annealing depth, smoothing the annealing characteristics of the edge area between sub-regions, and generating a control instruction with the annealing depth uniformity of the entire region meeting the preset uniformity tolerance includes: Extracting the edge area characteristics from the updated annealing depth by using the PCA feature extraction method; Smoothing and adjusting the edge area characteristics to obtain a smooth annealing feature distribution; Judging whether the smooth annealing feature distribution is greater than the preset uniformity tolerance. If so, calculating the deviation amount for adjustment of the sub-region. If not, no processing is performed, and the adjustment feature parameters are determined; Generating an annealing depth distribution according to the adjustment feature parameters, and obtaining a constrained annealing feature under uniformity constraints through inter-region smoothing processing; Extracting the overall region control parameters from the constrained annealing feature and generating a control instruction.

[0082] Exemplarily, in the semiconductor wafer annealing scenario, when extracting the edge area characteristics from the updated annealing depth by using the PCA feature extraction method, it is realized by analyzing the spatial distribution characteristics of the depth.

[0083] Exemplarily, the annealing depth in the edge area may show a different change trend from that in the central area. For example, the edge depth is 58 nanometers and the center is 61 nanometers. This difference reflects the non-uniformity of the thermal effect.

[0084] Specifically, the gradient change of the depth curve can be used as a feature to extract the uniqueness of the edge area. For example, the edge depth change is relatively gentle, while the central area is relatively steep. This feature extraction helps to identify the area that needs to be adjusted. When smoothing and adjusting the edge area characteristics, the mean filtering method can be used.

[0085] In one implementation, if the smoothed edge depth is 58 nanometers and the center is 61 nanometers, with a difference of exactly 3 nanometers, which is within the tolerance, no further adjustment is required and the characteristic parameters can be directly determined. Conversely, if the difference reaches 4 nanometers, the deviation amount needs to be calculated. For example, the depth in the edge region needs to be increased by 2 nanometers. This judgment method ensures the accuracy of the process. When generating the annealing depth distribution according to the adjusted characteristic parameters, it can be extended to the entire region through interpolation methods.

[0086] In one embodiment, between the 58 nanometers after edge adjustment and the 61 nanometers at the center, a gradient depth distribution is generated through smoothing processing. For example, from the edge to the center, it is 58, 59, 60, 61 nanometers in sequence. This smoothing processing between regions forms a constrained annealing feature under uniformity constraints, ensuring the continuity of the depth.

[0087] Specifically, this distribution can prevent the device performance from being damaged due to depth jumps. When extracting the full-region control parameters from the constrained annealing feature, the overall trend of the depth distribution can be analyzed.

[0088] For example, if the depth transition from the edge to the center is smooth and within the tolerance range, control parameters can be extracted, such as the laser power needs to be maintained at about 1200W and the scanning speed is stable at 8.5mm / s.

[0089] It should be noted that this basis provides a clear direction for subsequent processes. When generating control instructions, these parameters can be directly converted into signals recognizable by the device, such as "power setting 1200W, speed 8.5mm / s".

[0090] Exemplarily, if the feature extraction in the edge region shows a low depth, after smoothing processing and deviation adjustment, not only is the depth distribution more uniform, but also the working stability of the devices on the wafer can be improved.

[0091] In one embodiment, this method can also provide a reference for subsequent batches, avoiding the cost of repeated adjustments.

[0092] It can be understood that the generated control instructions can optimize the device operation in real time through closed-loop feedback, significantly improving production efficiency.

[0093] Preferably, this technology can also adapt to different wafer sizes, enhancing the flexibility of the process.

[0094] Refer to Figure 2 , the present invention provides a uniformity control device for large-area laser annealing, including: A three-dimensional height distribution acquisition module for acquiring a three-dimensional height distribution map of the entire region of the substrate; An adaptive segmentation module, which is used to divide the substrate into multiple sub-regions based on the three-dimensional height distribution map by using an adaptive segmentation algorithm, and obtain the height mean value and boundary coordinates of each sub-region; A focal length adjustment amount calculation module, which is used to calculate the focal length adjustment amount of the laser focus in each sub-region according to the height mean value, and generate a focus parameter table; A focus offset comparison module, which is used to determine the actual focus position through laser testing and compare it with the focus parameter table to obtain the focus offset of each sub-region; A dynamic focusing module, which is used to adjust the position of the laser lens through a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and obtain corrected focus distribution data; A path optimization module, which is used to extract the energy transition characteristics between sub-regions from the corrected focus distribution data, and generate an optical scanning path trajectory with the minimum energy gradient as a constraint by using a path optimization algorithm; A laser beam control module, which is used to control the movement of the laser beam based on the optical scanning path trajectory data and collect real-time energy distribution signals; An annealing depth adjustment module, which is used to detect the initial annealing depth from the real-time energy distribution signal, adjust the laser power and scanning speed of the sub-regions whose initial annealing depth does not meet the preset consistency tolerance threshold, and collect the updated annealing depth; An annealing feature smoothing module, which is used to extract the annealing features of the edge regions from the updated annealing depth, smooth the annealing features of the edge regions between sub-regions, and generate a control instruction with the annealing depth uniformity of the entire region meeting the preset uniformity tolerance.

[0095] As mentioned above, the above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

[0096] It should be noted that a uniformity control device for large-area laser annealing provided by an embodiment of the present invention is used to execute all the process steps of the uniformity control method for large-area laser annealing in the above embodiment. The working principles and beneficial effects of the two correspond one by one, and thus will not be elaborated here.

[0097] An embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a uniformity control program for large-area laser annealing. When the processor executes the computer program, the steps in the above-mentioned embodiments of the uniformity control method for large-area laser annealing are implemented, such as Figure 1The step S101 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module in the above device embodiments, such as the annealing feature smoothing module.

[0098] Exemplarily, the computer program may be divided into one or more modules, and the one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0099] The electronic device may be a computing device such as a desktop computer, a notebook, a palm computer, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation on the electronic device. It may include more or fewer components than the above, or combine some components, or different components. For example, the electronic device may further include input / output devices, network access devices, a bus, etc.

[0100] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the electronic device and connects various parts of the entire electronic device through various interfaces and lines.

[0101] The memory can be used to store the computer programs and modules. By running or executing the computer programs and modules stored in the memory, and invoking the data stored in the memory, the processor can implement various functions of the electronic device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash device, or other volatile solid-state storage devices.

[0102] Among them, if the modules integrated in the electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0103] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationships between the modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0104] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for controlling uniformity of large-area laser annealing, characterized in that: The method comprises: Obtain a three-dimensional height distribution map of the entire substrate area; Based on the three-dimensional height distribution map, an adaptive segmentation algorithm is used to divide the substrate into a plurality of sub-regions, and a height mean and boundary coordinates of each sub-region are obtained; Calculate the focal length adjustment amount of the laser focus in each sub-area according to the height average value, and generate a focus parameter table; Determine the actual focus position by laser testing and compare it with the focus parameter table to obtain the focus offset of each sub-area; Adjust the position of the laser lens by a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and obtain corrected focus distribution data; Extracting energy transition features between sub-regions from the modified focus distribution data, and generating an optical scanning path trajectory using a path optimization algorithm with a minimum energy gradient as a constraint; Control the movement of the laser beam based on the optical scanning path trajectory data and collect real-time energy distribution signals; Detecting an initial annealing depth from the real-time energy distribution signal, adjusting the laser power and scanning speed of a sub-region where the initial annealing depth does not meet a preset consistency tolerance threshold, and collecting and updating the annealing depth; The edge region annealing feature is extracted from the updated annealing depth, the edge region annealing feature between sub-regions is smoothed, and a control instruction is generated for the uniformity of the annealing depth of the entire region to meet a preset uniformity tolerance.

2. The method according to claim 1, characterized in that The step of obtaining a three-dimensional height distribution map of the entire substrate area includes: Use measurement equipment to scan the entire area and obtain an initial set of heights; Processing the initial set of heights by a smoothing algorithm to obtain smoothed height data; An interpolation algorithm is applied to the smoothed height data to generate a continuous three-dimensional height distribution map.

3. The method according to claim 1, characterized in that Based on the three-dimensional height distribution map, the substrate is divided into a plurality of sub-regions by using an adaptive segmentation algorithm, and the height mean and boundary coordinates of each sub-region are obtained, including: Using an adaptive segmentation algorithm to process the three-dimensional height distribution map according to a preset segmentation threshold to obtain a plurality of initial sub-regions; Calculating the mean height within the initial sub-area according to the three-dimensional height distribution map to obtain a preliminary mean height set; Processing the preliminary height mean value set by a mean filtering algorithm to obtain a smoothed height mean value set; Determine the height difference between the initial sub-areas according to the smoothed height mean value set to obtain a height difference distribution; Adjust the segmentation threshold according to the height difference distribution, and use an adaptive segmentation algorithm to re-divide the sub-region to obtain an optimized sub-region; According to the three-dimensional height distribution map, the boundary coordinates and the height mean of the optimized sub-region are extracted.

4. The method according to claim 1, characterized in that: The step of calculating the focal length adjustment amount of the laser focus in each sub-area according to the height average value and generating a focus parameter table includes: According to the height average, the difference between the height and the height corresponding to the reference focal length is calculated to obtain an initial value of focal length adjustment, wherein the reference focal length is a preset laser focal length; The focal length adjustment initial value is processed by a mean filtering algorithm to obtain a filtered focal length adjustment value; Repairing the filtered focal length adjustment value by interpolation to obtain an optimized focal length adjustment set; According to the optimized focal length adjustment set, the adjustment direction of the laser focus in each sub-area is determined, the focus change distribution is obtained, and a corresponding focus parameter table is generated.

5. The method according to claim 1, characterized in that The step of extracting energy transition characteristics between sub-regions from the modified focus distribution data and generating an optical scanning path trajectory using a path optimization algorithm with a minimum energy gradient as a constraint includes: Calculating the average pixel intensity of each sub-region from the corrected focus distribution data to obtain an energy value; According to the energy value, a gradient descent method is used to determine a change trend of energy transition; According to the change trend, an initial path trajectory is generated by a path optimization algorithm; According to the initial path trajectory, the position with the largest pixel intensity change rate is extracted as the key point coordinate of the optical scanning, and the scanning order is determined; A final optical scanning path trajectory is generated according to the scanning sequence and the key point coordinates.

6. The method according to claim 1, characterized in that The detecting the initial annealing depth from the real-time energy distribution signal, adjusting the laser power and scanning speed of the sub-region where the initial annealing depth does not meet the preset consistency tolerance threshold, and collecting and updating the annealing depth includes: The energy distribution signal is collected in real time by the temperature sensor to detect the initial annealing depth; If the initial annealing depth does not meet the preset consistency tolerance threshold, identifying the corresponding inconsistent sub-region; According to the inconsistent sub-regions, a laser power adjustment amount and a scanning speed adjustment amount are calculated to obtain adjustment parameters; The laser power and scanning speed are updated by adjusting the parameters, and the updated energy distribution is recorded to obtain the updated annealing depth.

7. The method according to claim 1, characterized in that The step of extracting edge region annealing features from the updated annealing depth, smoothing the edge region annealing features between sub-regions, and generating a control instruction for the uniformity of the annealing depth of the entire region to meet a preset uniformity tolerance includes: Extracting edge region features from the updated annealing depth using a PCA feature extraction method; Smoothing out the edge region features to obtain a smoothed annealing feature distribution; Determine whether the smoothed annealing characteristic distribution is greater than a preset uniform tolerance, if so, adjust the calculated deviation for the sub-region, if not, do not do anything, and determine the adjustment characteristic parameter; Generating an annealing depth distribution according to the adjustment characteristic parameters, and obtaining a constrained annealing feature under a uniformity constraint by smoothing between regions; Full-area control parameters are extracted from the constrained annealing features to generate control instructions.

8. A uniformity control device for large-area laser annealing, characterized in that: The device comprises: A three-dimensional height distribution acquisition module is used to obtain a three-dimensional height distribution map of the entire substrate area; An adaptive segmentation module, for dividing the substrate into a plurality of sub-regions based on the three-dimensional height distribution map by using an adaptive segmentation algorithm, and obtaining a height mean and boundary coordinates of each sub-region; A focal length adjustment amount calculation module, used to calculate the focal length adjustment amount of the laser focus in each sub-area according to the height average value, and generate a focus parameter table; A focus offset comparison module, used to determine the actual focus position through laser testing and compare it with the focus parameter table to obtain the focus offset of each sub-area; A dynamic focusing module, used for adjusting the position of the laser lens by a dynamic focusing method until the focus offset is less than a preset position tolerance threshold, and obtaining corrected focus distribution data; A path optimization module, used to extract energy transition characteristics between sub-regions from the modified focus distribution data, and generate an optical scanning path trajectory using a path optimization algorithm with a minimum energy gradient as a constraint; A laser beam control module, used to control the movement of the laser beam based on the optical scanning path trajectory data and collect real-time energy distribution signals; An annealing depth adjustment module is used to detect the initial annealing depth from the real-time energy distribution signal, adjust the laser power and scanning speed of the sub-area where the initial annealing depth does not meet the preset consistency tolerance threshold, and collect and update the annealing depth; The annealing feature smoothing module is used to extract the edge area annealing features from the updated annealing depth, smooth the edge area annealing features between sub-areas, and generate a control instruction for the uniformity of the annealing depth of the entire area to meet a preset uniformity tolerance.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for controlling uniformity of large-area laser annealing as claimed in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the uniformity control method for large-area laser annealing according to any one of claims 1 to 7.

Images