A method and system for surface passivation against aging of a VCSEL epitaxial wafer

Abstract

The application belongs to the technical field of semiconductor devices, and particularly relates to a VCSEL epitaxial wafer anti-aging surface passivation treatment method and system, which comprises the following steps: obtaining an original gray image, suppressing periodic background through frequency domain filtering and constructing a standard reference unit, and obtaining an accurate defect distribution map through differential operation; extracting the volume equivalent and edge stress gradient of the defect area, calculating the stress and volume influence weight based on the wafter brittleness index determined by the ratio of the two, and constructing an aging sensitivity index containing a linear volume risk term and a nonlinear stress risk term; further combining the local distribution density of defects to calculate a gain coefficient, and adaptively determining a fixed-point passivation deposition time. The application can dynamically adjust the evaluation model for different aging inducements, and utilize a density negative feedback mechanism to avoid excessive repair in dense areas, improve the passivation accuracy of repair depth and process safety, and prolong the service life of the device.

Description

Technical Field

[0001] This invention relates to the technical field of semiconductor devices. More specifically, this invention relates to a method and system for anti-aging surface passivation treatment of VCSEL epitaxial wafers. Background Technology

[0002] As a core component of VCSEL devices, the surface quality of the epitaxial wafer directly affects the device's optoelectronic performance and long-term reliability. During production and distribution, defects such as particles and scratches inevitably occur on the surface of the epitaxial wafer. These defects can easily become the origin of aging failures during device operation. Therefore, using passivation materials to precisely cover and repair defective areas is a key process step in preventing environmental corrosion, inhibiting defect diffusion, and improving the device's anti-aging capabilities.

[0003] Currently, to achieve automated passivation repair, related technologies typically utilize high-resolution optical imaging systems to acquire images of the epitaxial wafer surface, identify defect locations through image processing algorithms, and determine the deposition time or amount of passivation material based on the defect's geometry. However, this conventional approach has certain problems in practical applications. VCSEL arrays have a strong periodic texture background, making it difficult for traditional image segmentation algorithms to accurately separate and extract the fine edge features of defects in complex backgrounds. Furthermore, existing control algorithms mainly focus on filling the missing material volume, often ignoring the differences in the physical mechanisms of aging caused by defects of different morphologies. For example, they cannot distinguish between gentle material wear and sharp stress microcracks, and stress concentration caused by stress microcracks is the main risk of brittle fracture of devices. In addition, existing technologies usually treat each defect as an independent entity when calculating repair parameters, lacking consideration of the local defect distribution density. In areas with dense defects, over-repair can easily lead to passivation layer stacking, thereby introducing new internal stress or heat dissipation problems, making it difficult to achieve the ideal anti-aging effect. Summary of the Invention

[0004] To address the technical problem of poor passivation effect on the surface of VCSEL epitaxial wafers, the present invention provides solutions in the following aspects.

[0005] In a first aspect, the present invention provides a method for anti-aging surface passivation treatment of VCSEL epitaxial wafers, comprising:

[0006] A raw grayscale image containing multiple VCSEL units is acquired; the raw grayscale image is processed to obtain a precise defect distribution map containing multiple defect regions; features of each defect region are extracted, including volume equivalent and edge stress gradient; the volume equivalent is positively correlated with the sum of grayscale values ​​of all pixels in the corresponding defect region in the defect distribution map; the edge stress gradient is positively correlated with the maximum grayscale gradient of pixels in the corresponding defect region in the defect distribution map; an aging sensitivity index is calculated for each defect region, which is positively correlated with both the volume equivalent and edge stress gradient of the corresponding defect region, and also positively correlated with stress influence weight and volume influence weight; the values ​​of stress influence weight and volume influence weight are related to the wafer brittleness index, which is the ratio of the average edge stress gradient to the average volume equivalent of all defect regions; based on the aging sensitivity index, a passivation device is controlled to perform point passivation processing on each defect region, the point passivation deposition time is positively correlated with the aging sensitivity index and also positively correlated with the gain coefficient; the gain coefficient is negatively correlated with the local defect density of the corresponding defect region.

[0007] This invention achieves matching of major failure modes for different batches of wafers by extracting the volume equivalent of defects and edge stress gradients, and adaptively adjusting their weights in aging assessment in conjunction with the wafer brittleness index. At the same time, it introduces a gain coefficient that is negatively correlated with local defect density, taking into account spatial distribution characteristics when controlling passivation repair intensity, avoiding process interference caused by over-repair in defect-dense areas, thereby improving the anti-aging repair effect and yield of VCSEL epitaxial wafers.

[0008] Preferably, 2. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, obtaining the accurate defect distribution map includes:

[0009] The original grayscale image is subjected to frequency domain filtering to suppress periodic background, resulting in a preliminary defect mask;

[0010] Based on the preliminary defect mask, multiple defect-free VCSEL units are selected from the original grayscale image as a set of clean units; the clean unit set is weighted and averaged to obtain a standard reference unit.

[0011] The standard reference units are tiled across the entire image area to create a standard background image.

[0012] The original grayscale image is compared with the standard background image pixel by pixel, and the absolute value is taken to obtain an accurate defect distribution map.

[0013] This invention employs a strategy that combines coarse localization using frequency domain filtering with fine extraction using standard reference cells. It utilizes frequency domain characteristics to suppress periodic background interference and performs full-image tiling difference by constructing noise-free standard reference cells, thereby achieving lossless background removal. This enables the high-precision acquisition of accurate defect distribution maps containing true depth and edge sharpness information.

[0014] Preferably, obtaining the weights of the weighted average includes:

[0015] Obtain the center coordinates of the cleaning unit and the distance between them and the center pixel coordinates of the original grayscale image;

[0016] The weight of any cleaning unit in the set of cleaning units is positively correlated with the negative exponent of the square of the distance.

[0017] Preferably, the acquisition of the stress influence weight and the volume influence weight includes:

[0018] The difference between the wafer brittleness index and a preset constant is mapped to the (0,1) interval using a Sigmoid function to obtain the stress influence weight. ;

[0019] The volume influence weight equal .

[0020] This invention determines the stress and volume impact weights by mapping the wafer brittleness index to the (0,1) interval using the Sigmoid function, thereby achieving a dynamic balance in the risk assessment model. It can automatically adjust the assessment system's emphasis on edge stress and volume defects based on the overall statistical characteristics of the current wafer, making the subsequently calculated aging sensitivity index more reflective of the true aging risk of this batch of wafers.

[0021] Preferably, the stress influence weight Satisfying the expression:

[0022] ;

[0023] In the formula, Indicates the stress influence weight; Indicates the wafer brittleness index; , These are the first and second adjustment constants of the mapping function.

[0024] Preferably, the aging sensitivity index satisfies the expression:

[0025] ;

[0026] In the formula, This represents the aging sensitivity index of the r-th defect region; This indicates the weighting of volume influence; This represents the volume equivalent of the r-th defect region; Indicates the reference volume equivalent; Indicates the weight of stress influence; This represents the edge stress gradient of the r-th defect region; Indicates the stress gradient at the reference edge; This represents the natural exponential function.

[0027] This invention treats volume equivalent as a linear risk term and edge stress gradient as an exponential nonlinear risk term. This distinction allows for a more accurate assessment of the device aging process, resulting in a more accurate aging sensitivity index.

[0028] Preferably, the gain coefficient satisfies the following expression:

[0029] ;

[0030] In the formula, Indicates the first Gain coefficient for each defective region; Represents the fundamental gain constant; Indicates density inhibition factor; Indicates the first Local defect density in a defect region.

[0031] Preferably, obtaining the local defect density includes:

[0032] A circular neighborhood is constructed with the geometric center of any defect region as the center and a preset neighborhood radius as the radius; the number of other defect regions falling into this circular neighborhood is counted and used as the local defect density of the defect region.

[0033] Preferably, the point passivation deposition time satisfies the expression:

[0034] ;

[0035] In the formula, Indicates the first The point passivation deposition time for each defect area; Indicates the basic process time; Indicates the maximum additional compensation time; Indicates the first Gain coefficient for each defective region; This represents the aging sensitivity index of the r-th defect region.

[0036] This invention transforms the theoretical aging sensitivity index into a time parameter that can be executed by actual equipment. By setting the basic process time and the maximum additional compensation time, a convergent time control law is constructed, which ensures that all defects can obtain the most basic coverage, limits the maximum deposition amount to prevent overflow, and uses a mapping function to achieve deep passivation of high-risk and sparse defects and moderate passivation of low-risk or dense defects, thereby improving the precision of point-to-point process control.

[0037] Secondly, the present invention provides a VCSEL epitaxial wafer anti-aging surface passivation treatment system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned VCSEL epitaxial wafer anti-aging surface passivation treatment method is implemented.

[0038] By adopting the above technical solution, a computer program is generated from the above-mentioned method for anti-aging surface passivation treatment of VCSEL epitaxial wafers and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0039] The beneficial effects of this invention are as follows:

[0040] (1) This invention utilizes frequency domain filtering and standard cell reconstruction techniques to improve the extraction accuracy of defects in a strongly periodic background;

[0041] (2) The present invention constructs a hybrid risk assessment model based on wafer brittleness index, which dynamically weights and fuses volume defects and edge stress, thereby improving the accuracy of defect extraction;

[0042] (3) The present invention designs a negative feedback gain control method based on local density, which avoids process interference in dense areas while ensuring repair depth, and improves the accuracy of aging risk assessment and the precision of point repair. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating an anti-aging surface passivation treatment method for VCSEL epitaxial wafers according to the present invention;

[0044] Figure 2 This is a schematic diagram illustrating the initial defect mask;

[0045] Figure 3 It schematically shows a precise defect distribution map. Detailed Implementation

[0046] This invention discloses a method for anti-aging surface passivation treatment of VCSEL epitaxial wafers, referring to... Figure 1 This includes steps S1-S4:

[0047] S1: Based on a high-resolution microscopic imaging system, acquire the original grayscale image of the VCSEL epitaxial wafer surface; based on frequency domain filtering, perform background suppression on the original grayscale image to obtain a preliminary defect mask.

[0048] It should be noted that during the surface quality inspection of VCSEL epitaxial wafers, defects such as particles and scratches are usually superimposed on the background of the VCSEL cell array with periodic texture. Therefore, the defect signal is weak and drowned out by the strong periodic background signal. If thresholding is performed directly in the spatial domain, it is difficult to effectively separate the defects from the background. Therefore, this invention uses Fourier transform to convert the VCSEL epitaxial wafer surface image to the frequency domain. By suppressing specific frequency components representing the periodic background, potential defect regions can be quickly and initially located in the spatial domain, providing guidance for subsequent precise analysis.

[0049] Specifically, based on a high-resolution microscopic imaging system, the original grayscale image of the VCSEL epitaxial wafer surface is acquired; based on frequency domain filtering, background suppression is performed on the original grayscale image to obtain a preliminary defect mask, including:

[0050] The original grayscale image of the VCSEL epitaxial wafer surface is acquired using a high-resolution microscopic imaging system; the original grayscale image is then converted to the frequency domain using a two-dimensional fast Fourier transform to obtain a spectrum.

[0051] It should be noted that periodic textures appear as discrete high-energy bright spots in the frequency domain, while random defects and background noise appear as broadband signals.

[0052] To identify high-energy frequency components, the amplitude spectrum of the spectrogram is first calculated, and then the average energy value of the amplitude spectrum is calculated. Frequency points in the amplitude spectrum whose energy values ​​exceed a preset multiple of the average energy value are identified as high-energy frequency components. For example, the preset multiple is 3 times.

[0053] It should be noted that, in order to preserve the overall illumination information of the image, the central DC component region needs to be excluded.

[0054] A notch filter is applied to zero out the identified high-energy frequency components and their neighborhoods, resulting in a processed spectrum. An inverse Fourier transform is then performed on the processed spectrum to convert it to the spatial domain, yielding a differential image with suppressed background. This differential image is then binarized with a preliminary screening threshold (e.g., a grayscale value of 10) to generate a preliminary defect mask. For example, the preliminary screening threshold is 10 grayscale values.

[0055] It should be noted that the two-dimensional fast Fourier transform and inverse Fourier transform are existing technologies and will not be elaborated upon here. Figure 2This is a preliminary defect mask diagram, which shows that only suspected defect areas with grayscale values ​​exceeding the preliminary screening threshold are retained.

[0056] Thus, a preliminary defect mask was obtained.

[0057] S2: Based on the preliminary defect mask, standard reference units are selected and constructed from the original grayscale image; based on the standard reference units and the array period, a standard background image is synthesized; the original grayscale image and the standard background image are subjected to differential operation to obtain an accurate defect distribution map.

[0058] It should be noted that frequency domain transformation and the initial screening threshold processing can lead to the loss of some defect details with frequencies close to the background, resulting in distorted defect morphology. In the precision scenarios of semiconductor passivation repair, the true depth and edge sharpness of the defect are key physical characteristics that determine the repair process parameters. Therefore, this invention restores defects by first using the clean background area information in the initial defect mask to construct noise-free and distortion-free standard reference cells through statistical averaging; then, periodic tiling of the standard reference cells is used to synthesize a standard background image; finally, by subtracting from the original grayscale image, the background is eliminated without loss, thereby restoring the complete defect.

[0059] Specifically, based on the initial defect mask, standard reference units are selected and constructed from the original grayscale image, including:

[0060] Based on the periodic prior knowledge of VCSEL units, a standard VCSEL unit image is pre-extracted as a template. A template matching algorithm is used to perform a global search in the original grayscale image to obtain the center coordinates of all candidate VCSEL units. For any candidate VCSEL unit, if the image region corresponding to the candidate VCSEL unit does not have any pixels marked as defects in the preliminary defect mask, then the corresponding candidate VCSEL unit is determined to be a clean unit and included in the clean unit set. The template matching algorithm is a normalized cross-correlation template matching algorithm.

[0061] A standard reference unit image is constructed by weighted averaging of all clean unit images. It should be noted that the weighting coefficients in this weighted averaging process depend on the reliability of each clean unit during the averaging process. The imaging quality of the image center region is better than that of the edge regions. Therefore, the weighting coefficients are constructed by assigning higher weights to areas closer to the image center and lower weights to areas farther away, prioritizing the use of high-quality data. Considering that the two-dimensional Gaussian function has the characteristic of having the highest value at the center and smoothly decaying towards the edges, it can be used to construct the weighting coefficients.

[0062] Obtain the coordinates of the center pixel of the original grayscale image.

[0063] The weighting coefficients of any cleaning cell satisfy the expression:

[0064] ;

[0065] In the formula, This represents the weight coefficient of the k-th cleaning unit; , Represents the horizontal and vertical center coordinates of the k-th cleaning unit; , The horizontal and vertical coordinates of the center pixel of the original grayscale image; Indicates the scale parameter; This represents the natural exponential function. For example, This is used to control the rate of weight decay.

[0066] In the formula, This represents the square of the distance from the k-th cleaning unit to the center of the image. The smaller this value, the closer the unit is to the center. This represents the mapping of the square of the distance to... The interval is used to generate a weight value that is 1 at the center and smoothly decays to 0 towards the edge.

[0067] It should be noted that by weighting and averaging multiple cleaning cells, random imaging noise and sensor noise can be suppressed, while preserving the inherent deterministic cell morphology of the batch of epitaxial wafers, thus obtaining the grayscale distribution of the standard reference cells.

[0068] The grayscale value of any pixel in the standard reference unit image satisfies the expression:

[0069] ;

[0070] In the formula, Represents the image coordinates of the standard reference unit. The grayscale value of the pixel; Indicates the number of cleaning units; Indicates the first Each cleaning cell in the original grayscale image has coordinates relative to the standard reference cell image. The grayscale value of the corresponding pixel; Indicates the first The weighting coefficient of each cleaning unit.

[0071] In the formula, Indicates the first The grayscale contribution of the cleaning unit, this value reflects the grayscale contribution of the first cleaning unit. The information strength of each cleaning unit after considering location credibility; This represents the sum of the weighted grayscale contributions of all cleaning units; this value reflects the coordinate point. In a statistical sense, the sum of gray-level energy, through accumulation, causes random noise to cancel each other out, while the inherent structural features are enhanced. This represents the sum of the weights of all cleaning units, used for normalization; This represents the normalized weighted average gray value. The larger the value, the stronger the coordinates of the standard reference cell image. High grayscale is a definite structural feature in most cleaning cells, thus enabling the construction of a standard reference cell after noise reduction.

[0072] The gray values ​​of all pixels at all coordinates in the standard reference unit image together constitute the standard reference unit.

[0073] Preferably, based on the standard reference cell and array period, a standard background image is synthesized, and a difference operation is performed on the original grayscale image and the standard background image to obtain an accurate defect distribution map, including:

[0074] Based on the inherent row and column spacing of the VCSEL array, standard reference cells are tiled across the entire image according to lattice coordinates to synthesize a standard background image. The original grayscale image is then subtracted pixel-by-pixel from the standard background image, and the absolute value is taken to obtain a precise defect distribution map. It should be noted that...

[0075] like Figure 3 To provide an accurate defect distribution map, it shows the defect intensity of each pixel through a heatmap.

[0076] At this point, a precise defect distribution map was obtained.

[0077] S3: Based on the accurate defect distribution map, extract the volume equivalent and edge stress gradient features of each defect region; based on the proportional relationship between the volume equivalent and the edge stress gradient, obtain the stress influence weight and volume influence weight; based on the volume equivalent, edge stress gradient features, stress influence weight and volume influence weight, obtain the aging sensitivity index of each defect region.

[0078] It should be noted that during the long-term operation of VCSEL devices, epitaxial wafers from different batches or regions may exhibit different primary aging causes. For example, some batches are mainly characterized by material defects, while others are mainly characterized by stress cracking. If a fixed risk assessment model is used, optimal passivation treatment cannot be achieved. Therefore, this invention analyzes the macroscopic statistical characteristics of all defects on the epitaxial wafer and adaptively adjusts the risk model parameters to match the specific condition of the epitaxial wafer.

[0079] Specifically, based on the accurate defect distribution map, the volume equivalent and edge stress gradient features of each defect region are extracted; based on the proportional relationship between the volume equivalent and the edge stress gradient, the stress influence weight and volume influence weight are obtained, including:

[0080] The precise defect distribution map is truncated for noise, setting pixels with grayscale values ​​less than the background noise tolerance to 0. An 8-connected component analysis is performed on the image, aggregating all adjacent non-zero pixels into a single pixel set, with each pixel set representing a defect region. For example, the background noise tolerance is a grayscale value of 5.

[0081] Extract the volume equivalent and edge stress gradient of all defect regions from the accurate defect distribution map. The volume equivalent is the sum of the values ​​of all pixels in the defect region in the accurate defect distribution map, and the edge stress gradient is the maximum value of the gradient of all pixels in the defect region in the accurate defect distribution map. Calculate the mean volume equivalent and the mean edge stress gradient of all defects. Use the ratio of the mean edge stress gradient to the mean volume equivalent as the wafer brittleness index.

[0082] It should be noted that the wafer brittleness index can reflect the overall morphological characteristics of defects on the epitaxial wafer. If the wafer brittleness index is large, it indicates that the defects generally have the characteristics of small volume but large edge gradient, which manifests as sharp microcracks. This indicates that the main aging risk mode of the wafer is stress-driven brittle cracking. If the wafer brittleness index is low, it indicates that the defects generally have the characteristics of large volume but gentle edge gradient, which manifests as gentle pits. This indicates that the main aging risk mode of the wafer is wear caused by material loss.

[0083] Furthermore, based on this wafer brittleness index, volume influence weights and stress influence weights are constructed to dynamically adjust the relative importance of the two risk modes in the final evaluation model.

[0084] The stress influence weight satisfies the following expression:

[0085] ;

[0086] ;

[0087] In the formula, Indicates the stress influence weight; This indicates the weight of volume influence; Indicates the wafer brittleness index; , These are the first and second adjustment constants of the mapping function. For example, , .

[0088] In the formula, This indicates the deviation of the wafer brittleness index from the second adjustment constant; This represents the exponential decay term based on the aforementioned deviation; This indicates that the unbounded brittleness exponent is mapped using the properties of the Sigmoid function. The stress influence weight of the interval is such that the larger the value, the higher the system's reference to the stress gradient, thus adjusting the focus of risk assessment.

[0089] It should be noted that stress effects and volume effects have different characteristics. The severity of aging risk caused by volume defects is linearly related to the amount of defect; however, the aging risk caused by edge stress concentration has non-linear characteristics. When the sharpness of the defect edge exceeds a certain value, the cracking risk increases exponentially. Therefore, this invention constructs a hybrid model, using linear and exponential terms to represent these two types of risks respectively, and uses stress effect weights and volume effect weights to perform a weighted summation of the linear and exponential terms to obtain the final aging sensitivity index.

[0090] Preferably, based on the volume equivalent, edge stress gradient characteristics, stress influence weight, and volume influence weight, the aging sensitivity index of each defect region is obtained, including:

[0091] The aging sensitivity index satisfies the following expression:

[0092] ;

[0093] In the formula, This represents the aging sensitivity index of the r-th defect region; This indicates the weight of volume influence; This represents the volume equivalent of the r-th defect region; Indicates the reference volume equivalent; Indicates the stress influence weight; This represents the edge stress gradient of the r-th defect region; Indicates the stress gradient at the reference edge; This represents the natural exponential function. It should be noted that... and These are preset values, representing the process-acceptable volume equivalent and edge stress gradient, used to normalize the volume equivalent and edge stress gradient of the defect region. For example... , .

[0094] In the formula, , The normalized volume equivalent and edge stress gradient of the r-th defect region reflect the direct aging risk caused by material defects. This represents the linear volumetric risk term determined by the volume equivalent, and is expressed by... Weighting; This represents the nonlinear stress risk term determined by the edge stress gradient, and is derived from... Weighting is applied and amplified using an exponential function to simulate the outbreak characteristics of stress risk; This indicates that the linear volume risk term and the nonlinear stress risk term are added together. The larger this value is, the higher the comprehensive aging risk of the defect under the dual effects of volume loss and stress concentration, and the more in-depth passivation treatment is required.

[0095] Thus, the aging sensitivity index of each defect area was obtained.

[0096] S4: Based on the accurate defect distribution map, calculate the local defect density of each defect region and obtain the gain coefficient; based on the response gain coefficient and aging sensitivity index, construct the fixed-point passivation deposition time; perform fixed-point passivation repair and post-processing verification.

[0097] It should be noted that when mapping the aging sensitivity index to a specific passivation deposition time, not only the severity of the risk of individual defects must be considered, but also the spatial distribution of defects. In semiconductor passivation processes, if the defect distribution is too dense, and a high-gain repair strategy (i.e., long deposition time) is used for each defect, the edges of the repair areas may overlap, leading to excessively thick local passivation layers, which can introduce new internal stress or affect device heat dissipation. Therefore, this invention introduces an adaptive response gain coefficient, following the control principle that the denser the defects, the lower the gain; and the sparser the defects, the higher the gain. An aggressive strategy is used for rapid repair in sparse defect areas, while a conservative strategy is used for fine repair in dense defect areas.

[0098] Specifically, based on the accurate defect distribution map, the local defect density of each defect region is calculated, and the gain coefficient is obtained, including:

[0099] A circular neighborhood is constructed with the geometric center of any defect region as its center and a preset neighborhood radius as its radius. The number of other defect regions falling within this circular neighborhood is counted and recorded as the local defect density of the defect region. It should be noted that the preset neighborhood radius should be set according to the effective working range of the passivation nozzle; for example, the preset neighborhood radius is 50 micrometers.

[0100] It should be noted that the construction of the gain coefficient is a negative feedback adjustment process. When the local defect density increases, the repair intensity of individual defects must be reduced to prevent process interference. Considering that the inverse proportional function can describe this suppression relationship where the higher the density, the smaller the coefficient, and to ensure that the gain has a maximum value when the density is zero, this invention introduces a constant term in the denominator to construct a density-based gain attenuation model.

[0101] The gain coefficient of any defect region satisfies the expression:

[0102] ;

[0103] In the formula, Indicates the first Gain coefficient for each defective region; Represents the fundamental gain constant; Indicates density inhibition factor; Indicates the first Local defect density in a defect region.

[0104] It should be noted that, This represents the standard response strength in the absence of neighboring interference, for example, ; Used to adjust the degree to which density affects gain, for example, .

[0105] In the formula, This represents the density suppression term. The larger the value, the higher the value. The more other defects exist around a defect region, the more crowded the spatial distribution, and the larger the density suppression term. This represents the gain coefficient after density correction, and its value varies with... The decrease in the value indicates that the system automatically reduces the sensitivity of the process response in dense areas to prevent process interference caused by the simultaneous high-intensity repair of multiple adjacent defects.

[0106] It should be noted that the calculation of passivation deposition time maps the theoretical comprehensive risk to the execution time. This mapping process needs to meet the following conditions: a minimum base process time must be guaranteed to form effective coverage, and a maximum deposition time must be set to prevent material overflow. Therefore, this invention constructs a time control law to ensure that high-risk defects are fully repaired while the process time always converges within a safe range.

[0107] Preferably, the point passivation deposition time is constructed based on the response gain coefficient and the aging sensitivity index, including:

[0108] The point passivation deposition time for any defect region satisfies the expression:

[0109] ;

[0110] In the formula, Indicates the first The point passivation deposition time for each defect area; Indicates the basic process time; Indicates the maximum additional compensation time; Indicates the first Gain coefficient for each defective region; This represents the aging sensitivity index of the r-th defect region. For example, , .

[0111] In the formula, This represents the overall risk score after environmental gain correction, which reflects the actual remediation needs after taking local density constraints into account. Indicates the normalized denominator; This represents the risk attenuation factor; the higher the overall risk score, the closer this value is to 0. This represents a saturation mapping function that maps an unbounded comprehensive risk score to... The interval ensures that the calculated time compensation amount will not increase indefinitely; Indicates the final deposition time, specifically for high-risk defects with sparse surrounding areas. It will approach the maximum value, achieving deep passivation; while for low-risk or extremely dense defects, There will be limitations; avoid over-processing.

[0112] Using the calculated point passivation deposition time, the passivation equipment is controlled to perform point spraying on each defect area, and optical scanning verification is performed after completion.

[0113] This completes the anti-aging surface passivation treatment of the VCSEL epitaxial wafer.

[0114] This invention also discloses a VCSEL epitaxial wafer anti-aging surface passivation treatment system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a VCSEL epitaxial wafer anti-aging surface passivation treatment method according to the present invention.

[0115] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0116] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for anti-aging surface passivation treatment of VCSEL epitaxial wafers, characterized in that, include: Obtain the original grayscale image containing multiple VCSEL units; The original grayscale image is processed to obtain a precise defect distribution map containing multiple defect regions, including: performing frequency domain filtering on the original grayscale image to suppress periodic background and obtain a preliminary defect mask; based on the preliminary defect mask, selecting multiple defect-free VCSEL units from the original grayscale image as a set of clean units; performing a weighted average on the set of clean units to obtain a standard reference unit; tiling the standard reference unit across the entire image to synthesize a standard background image; and performing pixel-by-pixel difference between the original grayscale image and the standard background image and taking the absolute value to obtain the precise defect distribution map. Features of each defect region are extracted, including volume equivalent and edge stress gradient; the volume equivalent is positively correlated with the sum of gray values ​​of all pixels in the corresponding defect region in the defect distribution map; the edge stress gradient is positively correlated with the maximum gray value gradient of pixels in the corresponding defect region in the defect distribution map. The aging sensitivity index of each defect region is calculated. The aging sensitivity index is positively correlated with the volume equivalent and edge stress gradient of the corresponding defect region, and is also positively correlated with the stress influence weight and volume influence weight. The values ​​of the stress influence weight and volume influence weight are related to the wafer brittleness index, which is the ratio of the average edge stress gradient to the average volume equivalent of all defect regions. Based on the aging sensitivity index, the passivation equipment is controlled to perform point passivation treatment on each defect area. The point passivation deposition time is positively correlated with the aging sensitivity index and also positively correlated with the gain coefficient. The gain coefficient is negatively correlated with the local defect density of the corresponding defect area.

2. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, The acquisition of the weights for the weighted average includes: Obtain the center coordinates of the cleaning unit and the distance between them and the center pixel coordinates of the original grayscale image; The weight of any cleaning unit in the set of cleaning units is positively correlated with the negative exponent of the square of the distance.

3. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, The acquisition of the stress influence weight and volume influence weight includes: The difference between the wafer brittleness index and a preset constant is mapped to the (0,1) interval using a Sigmoid function to obtain the stress influence weight. ; The volume influence weight equal .

4. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 3, characterized in that, The stress influence weight Satisfying the expression: ； In the formula, Indicates the stress influence weight; Indicates the wafer brittleness index; , These are the first and second adjustment constants of the mapping function.

5. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, The aging sensitivity index satisfies the expression: ； In the formula, This represents the aging sensitivity index of the r-th defect region; This indicates the weight of volume influence; This represents the volume equivalent of the r-th defect region; Indicates the reference volume equivalent; Indicates the stress influence weight; This represents the edge stress gradient of the r-th defect region; Indicates the stress gradient at the reference edge; This represents the natural exponential function.

6. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, The gain coefficient satisfies the following expression: ； In the formula, Indicates the first Gain coefficient for each defective region; Represents the fundamental gain constant; Indicates density inhibition factor; Indicates the first Local defect density in a defect region.

7. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 6, characterized in that, The acquisition of the local defect density includes: A circular neighborhood is constructed with the geometric center of any defect region as the center and a preset neighborhood radius as the radius; the number of other defect regions falling into this circular neighborhood is counted and used as the local defect density of the defect region.

8. The method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to claim 1, characterized in that, The fixed-point passivation deposition time satisfies the expression: ； In the formula, Indicates the first The point passivation deposition time for each defect area; Indicates the basic process time; Indicates the maximum additional compensation time; Indicates the first Gain coefficient for each defective region; This represents the aging sensitivity index of the r-th defect region.

9. A VCSEL epitaxial wafer anti-aging surface passivation treatment system, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a method for anti-aging surface passivation treatment of VCSEL epitaxial wafers according to any one of claims 1-8.

Images