A VCSEL growth process real-time monitoring method based on image processing

Abstract

The present application belongs to the technical field of image processing, and particularly relates to a VCSEL growth process real-time monitoring method based on image processing, which comprises the following steps: collecting an initial reference image and growth images at each time; calculating a deposition rate and a cumulative abruptness at each time; obtaining an incremental displacement field and a correlation coefficient by a digital speckle correlation method; determining a cumulative displacement field based on the incremental displacement field and an anchor point displacement field of the current round; in response to the cumulative abruptness being greater than or equal to a cumulative abruptness threshold, taking the cumulative displacement field at the time as a new anchor point displacement field and taking the growth image at the time as a new reference image; repeating iteration to obtain a cumulative displacement field in a whole growth period and calculate a cumulative stress. The present application significantly improves the accuracy of cumulative stress calculation.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method for real-time monitoring of the VCSEL growth process based on image processing. Background Technology

[0002] As a core light source device for cutting-edge technologies such as 3D sensing, optical communication, and lidar, the performance of VCSELs is highly dependent on the precise control of the thickness and stress state of multilayer thin film materials during epitaxial growth. During the growth process, due to lattice mismatch and differences in thermal expansion coefficients between different materials, internal stress is inevitably introduced, causing the crystal to warp. Excessive warping not only affects subsequent processes but may also cause material cracking, severely reducing the yield and reliability of the device. Therefore, real-time monitoring of the stress state during the growth process is crucial.

[0003] Currently, most methods for monitoring optoelectronic semiconductor devices use optical measurement techniques. For example, the digital speckle correlation method compares speckle images acquired at different times and uses image correlation algorithms to calculate the displacement vector of each point on the surface, thereby obtaining deformation and curvature information.

[0004] However, the basic assumption of the Digital Speckle Correlation Method (DSCM) is that the speckle pattern on the surface of the object being measured remains stable during deformation, with only geometrical shifts. But during VCSEL growth, new atomic layers are continuously deposited on the crystal surface at a rate of angstroms per second. This change in surface characteristics caused by material growth makes the reference speckle image acquired in the early stages of growth no longer match the actual speckle pattern on the current crystal surface in a short period of time. This is known as growth decorrelation, which undermines the physical basis of correlation calculation, leading to the interruption of displacement tracking and making it impossible to accurately monitor the cumulative stress throughout the entire growth cycle. Summary of the Invention

[0005] To address the technical problem of growth decorrelation, which prevents existing methods from accurately monitoring stress accumulation throughout the entire growth cycle, this invention provides a real-time monitoring method for VCSEL growth process based on image processing. The method includes: acquiring an initial reference image and growth images at various time points during the growth process; calculating the rate of change of the average grayscale of the growth image at each time point relative to the previous time point, as the deposition rate; performing polynomial fitting on the deposition rates of all historical time points prior to each time point to obtain a deposition rate curve; calculating the time integral of the deposition rate curve from the initial time point to each time point, as the cumulative abrupt change; and obtaining the growth image at each time point using digital speckle correlation. The process involves calculating the correlation coefficient and incremental displacement field relative to the reference image of the current cycle; summing the incremental displacement field at each time step with the anchor point displacement field of the current cycle to obtain the cumulative displacement field at each time step; setting the initial anchor point displacement field as a zero vector; in response to a cumulative abrupt change greater than or equal to a cumulative abrupt change threshold, using the cumulative displacement field at that time step as the new anchor point displacement field and the growth image at that time step as the new reference image; repeating the iteration until the cumulative displacement field of the entire growth cycle is obtained; extracting the vertical component of the cumulative displacement field of the entire growth cycle for quadratic surface fitting to determine the curvature; and calculating the cumulative stress using the Stony formula, issuing an early warning in response to the stress exceeding the maximum stress of growth.

[0006] The innovation of this invention lies in its proposal of a dynamic reference image update mechanism, which no longer relies on a fixed initial reference image. By calculating the rate of change of the average gray level of the image, a cumulative mutation index is constructed to assess the degree of surface change during growth. By testing and calibrating the cumulative mutation at the moment when the correlation coefficient first decreases as the cumulative mutation threshold, an objective criterion for updating the reference image is obtained. When the monitored cumulative mutation reaches this threshold, the system not only dynamically updates the reference image, but also links the displacement fields of the old and new cycles through an anchor displacement field mechanism. This solves the growth decorrelation phenomenon and realizes continuous tracking of the displacement field throughout the entire growth cycle, thereby ensuring the accuracy and reliability of the final cumulative stress calculation.

[0007] Preferably, the average gray level of the grown image is equal to the average gray level of all pixels in the grown image.

[0008] Preferably, the deposition rate satisfies the expression: In the formula, For the first Deposition rate at each time point; For the first The average gray level of the grown image at each time step; For the first The average gray level of the grown image at each time step; The time interval for image acquisition; The sign for absolute value.

[0009] This invention evaluates the deposition rate by calculating the time change rate of the average gray value of the growth image. This method utilizes the physical principle that the thin film interference effect in epitaxial growth leads to changes in surface reflectivity, and can reflect the update speed of the crystal physical surface in real time. This provides key data basis for subsequent calculation of surface cumulative changes and judgment of decorrelation risk.

[0010] Preferably, the cumulative mutationality satisfies the expression: In the formula, For the first The cumulative mutability at each moment; For the first Deposition rate curves at each time point; For the initial time and the first A timestamp at a specific moment; For time differentiation; The symbol is for integrals.

[0011] The present invention uses a cumulative mutation index to reflect the cumulative change in the physical and optical properties of the crystal surface caused by the continuous growth of the material from the initial moment to the current moment. This index can accurately reflect the difference between the current image and the reference image.

[0012] Preferably, the cumulative mutation threshold is obtained by: performing a complete VCSEL growth test, using the growth image at the initial moment of the growth test as the reference image, calculating the correlation coefficient between the growth images at all moments and the reference image, iterating through and calculating the difference between the correlation coefficient at each moment and the previous moment, and using the moment when the difference is less than 0 for the first time as the cumulative mutation threshold.

[0013] This invention defines the cumulative abrupt change corresponding to the moment when the correlation coefficient first decreases as the threshold through a complete VCSEL growth test. This method calibrates the maximum amount of surface change that the digital speckle correlation method can tolerate, ensuring that the reference image can be updated in a timely manner across material stages.

[0014] Preferably, the repeated iterations until the cumulative displacement field of the entire growth cycle is obtained include: at the beginning of a new round, the cumulative abrupt change is cleared to zero, the incremental displacement field at the new round time is calculated based on the new reference image and superimposed on the new anchor point displacement field, until the cumulative displacement field at the last moment is obtained, the iteration is stopped, and the cumulative displacement field of the entire growth cycle is obtained.

[0015] In the new iteration, the cumulative abrupt change is cleared to zero, and the incremental displacement is calculated based on the new reference image. This displacement is then superimposed on the new anchor point displacement field. This anchor point linking mechanism solves the problem of displacement field calculation interruption caused by reference image updates, realizes seamless linking of displacement increment results, and ensures that even if the initial image and the final image are completely unrelated, a continuous cumulative displacement field for the entire growth cycle can be obtained, thereby improving the accuracy of the final cumulative stress calculation.

[0016] Preferably, the step of extracting the vertical component of the cumulative displacement field throughout the entire growth cycle and performing quadratic surface fitting to determine the curvature includes: extracting the component of the cumulative displacement field perpendicular to the crystal surface, including... and The perpendicular components in two directions; perform quadratic surface fitting on all perpendicular components to obtain the fitting equation, and then select the corresponding components from the fitting equation. and The coefficients of the surface are used as the surface coefficients. and Curvature of direction.

[0017] Preferably, the calculation of cumulative stress using the Stony formula includes: substituting process parameters and the radius of curvature of the substrate growth into the Stony formula to calculate the cumulative stress, wherein the process parameters include the Young's modulus of the substrate material, the Poisson's ratio of the substrate, the thickness of the substrate, and the thickness of the film; the radius of curvature of the substrate growth is equal to the surface curvature at... and The reciprocal of the average curvature of the direction.

[0018] Preferably, the maximum stress during growth is obtained from a parameter table of the crystal material.

[0019] Preferably, the method further includes: performing grayscale conversion and mean filtering on the acquired growth image.

[0020] The beneficial effects of this invention are as follows:

[0021] (1) By introducing deposition rate and cumulative abruptness index, this invention directly links the changes in surface optical properties caused by material growth with the decorrelation risk at the image analysis level, providing a reliable physical criterion for whether the algorithm needs to update the reference image;

[0022] (2) The present invention obtains an objective criterion for updating the reference image by pre-testing the cumulative abruptness threshold of the correlation decrease. When the detected cumulative abruptness reaches the threshold, the system immediately updates the reference image and links the newly calculated incremental displacement to the cumulative displacement of the previous stage through the anchor displacement field mechanism. This method solves the growth decorrelation problem and enables seamless linking of displacement tracking throughout the entire growth cycle, thereby ensuring accurate monitoring of cumulative stress. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a real-time monitoring method for the growth process of VCSEL based on image processing according to the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0026] This invention discloses a method for real-time monitoring of the VCSEL growth process based on image processing, referring to... Figure 1 This includes steps S1 to S5:

[0027] S1. Acquire the initial reference image and growth images at each time point during the VCSEL growth process.

[0028] It should be noted that the technical basis of digital speckle correlation is to calculate deformation by tracking the displacement of speckles on the surface of an object. Therefore, a stable speckle pattern must first be constructed on the monitored object, i.e., the surface of the crystal substrate, as a reference for subsequent displacement tracking. At the same time, it is necessary to continuously acquire image sequences during the crystal growth process to record the dynamic changes of speckles as the crystal warps, providing raw data for subsequent correlation calculation and displacement analysis. Considering that actual speckle images are sensitive to noise, the acquired growth images must first be denoised to avoid noise data interfering with the monitoring process.

[0029] Specifically, before VCSEL growth begins, the surface of the crystal substrate to be grown is illuminated by an external coherent light source, and the optical system is adjusted. A CMOS image acquisition device is used to clearly capture the real-time speckle field. The growth image acquired at the initial moment is set as the initial reference image. Throughout the growth process, images are captured at preset time intervals. Multiple growth images are continuously acquired; the preset time interval can be 0.2 seconds, which can be adjusted according to the actual situation; at the same time, the acquired growth images are subjected to grayscale conversion and mean filtering to obtain the grayscale value of each pixel.

[0030] Thus, the initial reference image and growth images at each time point in the VCSEL growth process have been obtained.

[0031] S2. For each time step: Calculate the rate of change of the average gray level of the growth image at that time step relative to the previous time step, and use it as the deposition rate at that time step.

[0032] It should be noted that, considering that changes in surface optical properties caused by the growth of new materials can lead to the failure of reference images and thus affect the accuracy of deformation assessment, it is necessary to construct an index that can reflect the current physical growth rate. Since the increase in film thickness during epitaxial growth will cause thin film interference effect, resulting in periodic or trend changes in the overall reflectivity of the crystal surface, this change will be directly reflected in the average gray value of the acquired speckle image. Therefore, by calculating the time change rate of the average gray value, the speed of change of surface optical properties can be reflected.

[0033] Specifically, the average gray level of the grown image at each time step is obtained. The average gray level of the grown image at each time step is equal to the average gray level of all pixels in the grown image at that time step.

[0034] The deposition rate at each time step is determined based on the change in the average gray level of the growth image relative to the previous time step over time; the deposition rate satisfies the expression:

[0035]

[0036] In the formula, For the first Deposition rate at each time point; For the first The average gray level of the grown image at each time step; For the first The average gray level of the grown image at each time step; The time interval for image acquisition; The sign for absolute value.

[0037] in, Reflects the first The rate at which the optical properties of the crystal surface change due to material deposition per unit time at a given moment is considered to be positively correlated. A larger value indicates a more rapid change in the optical properties of the crystal surface at that moment. The more drastic the change in surface reflectivity at a given moment per unit time, the greater the thin film interference effect caused by the increase in film thickness during epitaxial growth. This means that the renewal rate of the physical surface changes during growth, which may be due to the switching of new materials, resulting in a higher deposition rate.

[0038] Thus, the deposition rate at each time point was obtained.

[0039] S3. Determine the cumulative abrupt change based on the product of the deposition rate of all historical moments prior to this moment and the image acquisition interval.

[0040] It should be noted that the instantaneous deposition rate only reflects the instantaneous changes of the surface, while decorrelation is a cumulative effect that depends on the total amount of changes that have occurred on the surface since the reference image was set. Therefore, it is necessary to construct a cumulative abrupt change index, which describes the total amount of changes in the physical and optical properties of the crystal surface during the growth process from the reference time to the current time by integrating the surface change rate over time. The higher the cumulative abrupt change index value, the greater the difference between the current surface and the reference surface, and the higher the risk of decorrelation.

[0041] Specifically, for each time point, the optical deposition rate of all historical time points prior to that time point is obtained and polynomial fitting is performed to obtain the deposition rate curve; for example, the th... At that moment , No. All historical moments before a given moment refer to All times within the interval.

[0042] Calculate the time integral of the deposition rate curve from the initial time to each time step to determine the cumulative abrupt change at each time step; the cumulative abrupt change satisfies the expression:

[0043]

[0044] In the formula, For the first The cumulative mutability at each moment; For the first Deposition rate curves at each time point; For the initial time and the first A timestamp at a specific moment; For time differentiation; The symbol is for integrals.

[0045] in, Reflects from the initial moment to the... The cumulative total change in the physical and optical properties of the crystal surface caused by the continuous growth of the material during the nth time period. The larger this value, the more significant the change in the physical and optical properties during the nth time period. The greater the difference between the surface speckle pattern at a given time and the reference image, the higher the risk of growth-related decorrelation; conversely, the smaller the value, the lower the risk of growth-related decorrelation.

[0046] Thus, the cumulative mutability at each time point is obtained.

[0047] S4. Obtain the incremental displacement field and correlation coefficient using the digital speckle correlation method; determine the cumulative displacement field based on the incremental displacement field and the anchor point displacement field of the current cycle; in response to the cumulative abruptness being greater than or equal to the cumulative abruptness threshold, use the cumulative displacement field at that moment as the new anchor point displacement field and the growth image at that moment as the new reference image; repeat the iteration to obtain the cumulative displacement field of the entire growth cycle.

[0048] It should be noted that, considering the standard VCSEL epitaxial growth process, when using the initial growth image as a fixed reference image for digital speckle correlation analysis of the image signal, the basic assumption of digital speckle correlation is that the speckle pattern on the surface of the measured object remains stable during deformation, only undergoing geometrical movement. That is, when there is no change in the growth material, the correlation coefficient with the initial speckle image is equal. However, once the surface growth material undergoes a sudden change, the cumulative mutability at the corresponding moment will be greater, and the correlation coefficient relative to the initial speckle image will decrease due to the growth decorrelation effect, indicating that the reference image may have become invalid. Therefore, a complete VCSEL growth process can be performed as a test to analyze the change in the real-time correlation coefficient over time, extracting the cumulative mutability corresponding to the initial decrease moment as the cumulative mutability threshold. This threshold represents the maximum surface change the algorithm can tolerate. When the cumulative mutability exceeds the threshold, it indicates that the current reference image is about to become invalid and must be forcibly updated to improve the accuracy of subsequent displacement field calculations. To ensure the continuity of monitoring, a displacement increment linking mechanism is used during updates, and these incremental results are seamlessly linked through the anchor point displacement field, thereby achieving continuous displacement tracking throughout the entire growth cycle.

[0049] Specifically, the correlation coefficient and incremental displacement field of the current growth image relative to the reference image of the current cycle are calculated using the digital speckle correlation method.

[0050] The cumulative mutation threshold is obtained as follows: A complete VCSEL growth test is performed, using the initial growth image as the reference image. The correlation coefficients of the growth images relative to the reference image at all time points are calculated. The differences between the correlation coefficients at each time point and the correlation coefficients at the previous time point are calculated iteratively. The cumulative mutation threshold is defined as the moment when the difference first becomes less than 0. It should be noted that the calculation of the correlation coefficient is a well-known aspect of digital speckle correlation analysis and will not be elaborated upon here.

[0051] If the cumulative mutability at the current moment Less than the cumulative mutation threshold Explanation of the first At this moment, the crystal surface growth change is within the tolerance range, and the reference image for the current round is valid; at this time, the... The cumulative displacement field at time n is equal to the first time n. The sum of the incremental displacement field at each moment and the anchor displacement field in the current round; where the initial anchor displacement field is the zero vector.

[0052] If the cumulative mutability at the current moment Greater than or equal to the cumulative mutability threshold Explanation of the first At time point 1, the crystal surface growth has reached a critical point, and the reference image for the current cycle must be updated. The cumulative displacement field at time n is used as the new anchor point displacement field, and the cumulative displacement field at time n is used as the anchor point displacement field. The growth image at each time step is used as a new reference image. At the start of a new round, the cumulative abrupt change is cleared to zero. The incremental displacement field at subsequent time steps is calculated based on the new reference image and superimposed on the new anchor displacement field to achieve the link. This process continues until the cumulative displacement field at the last time step is obtained, at which point the iteration stops, and the cumulative displacement field of the entire growth cycle is obtained.

[0053] For example, Initial anchor point displacement field Zero vector, initial reference image ; for the first At that moment, ,because ,illustrate Valid, at this time, obtain relatively Incremental displacement field , No. The cumulative displacement field at each moment ; for the first At that moment, ,because This indicates that the reference image needs to be updated. At this point, obtain... relatively Incremental displacement field , No. The cumulative displacement field at each moment Meanwhile, the new anchor point displacement field The new reference image is This is used for calculations at subsequent time points, and the cumulative mutation count is reset to zero, starting from the [number]th [time]. Accumulation begins again at the [time point]; for the [time point]... At that moment, ,because ,illustrate Valid, at this time, obtain relatively Incremental displacement field , No. The cumulative displacement field at each moment This process avoids inaccurate displacement tracking caused by growth decorrelation and ensures that even when... and They are completely unrelated, and the algorithm can still pass. This anchor point, By linking to the initial state, continuous tracking throughout the entire lifecycle is achieved.

[0054] Thus, the cumulative displacement field for the entire growth cycle is obtained.

[0055] S5. Extract the vertical component of the cumulative displacement field throughout the entire growth cycle, perform quadratic surface fitting, determine the curvature, and calculate the cumulative stress using the Stony formula. If the stress exceeds the maximum stress during growth, issue an early warning.

[0056] It should be noted that obtaining the total displacement field is not the ultimate goal. The ultimate goal is to monitor whether the stress during VCSEL growth is normal. According to the Stony formula, the film stress is directly related to the change in crystal curvature. Abnormal stress mainly causes the crystal to warp, and this warping is perpendicular to the crystal surface, i.e., in the total displacement field. The displacement along the axial direction; therefore, the component perpendicular to the crystal surface in the cumulative displacement field can be extracted and fitted with a quadratic surface. The fitting equation for the surface corresponds to... and The coefficients represent the surface in and The curvature in the direction is used to calculate the magnitude of the accumulated stress in the epitaxial film by means of biaxial average curvature.

[0057] Specifically, based on the cumulative displacement field throughout the entire growth cycle, the component perpendicular to the crystal surface in the cumulative displacement field is extracted, including... and The perpendicular components in two directions; perform quadratic surface fitting on all perpendicular components to obtain the fitting equation, and then select the corresponding components from the fitting equation. and The coefficients of the surface are used as the surface coefficients. and Curvature of direction.

[0058] The curved surface and The reciprocal of the average curvature of the direction is used as the radius of curvature for substrate growth.

[0059] Obtain process parameters, including: Young's modulus of the substrate material, Poisson's ratio of the substrate, thickness of the substrate, and thickness of the film; substitute the process parameters and the radius of curvature of the substrate growth into the Stony formula to calculate the cumulative stress.

[0060] If the accumulated stress exceeds the maximum tensile stress during growth, indicating a problem in the growth process and potential issues such as warping, an immediate warning is issued, sending a VCSEL growth failure signal. The maximum tensile stress during growth is obtained from the crystal material's parameter table.

Claims

1. A method for real-time monitoring of a VCSEL growth process based on image processing, characterized in that, The method comprises the following steps: collecting an initial reference image and growth images at each time during growth; calculating the change rate of the average gray level of the growth image at each time relative to the growth image at the previous time as the deposition rate; performing polynomial fitting on the deposition rates of all historical times before each time to obtain a deposition rate curve; and calculating the time integral of the deposition rate curve from the initial time to each time as the cumulative abruptness; obtaining the correlation coefficient and the incremental displacement field of the growth image at each time relative to the reference image of the current round by means of digital speckle correlation; calculating the sum of the incremental displacement field at each time and the anchor point displacement field of the current round as the cumulative displacement field at each time; and setting the initial anchor point displacement field as a zero vector; in response to the cumulative abruptness being greater than or equal to a cumulative abruptness threshold, taking the cumulative displacement field at the time as a new anchor point displacement field and taking the growth image at the time as a new reference image; repeating the iteration until the cumulative displacement field of the whole growth period is obtained; extracting the vertical component of the cumulative displacement field of the whole growth period to perform quadratic surface fitting and determining the curvature; and calculating the cumulative stress in combination with the Stoney formula, and in response to the cumulative stress being greater than the maximum stress of growth, performing a warning; The cumulative mutability satisfies an expression: ; in which, is the cumulative mutability at the th time; is the deposition rate curve at the th time; is the time stamp at the initial time and the th time; is the time differential; is the integral sign; the cumulative abruptness threshold is obtained in the following manner: performing a complete VCSEL growth test, taking the growth image at the initial time in the growth test as a reference image, calculating the correlation coefficient of the growth image at each time relative to the reference image, and traversing to calculate the difference between the correlation coefficient at each time and the correlation coefficient at the previous time, and taking the time corresponding to the first time when the difference is less than 0 as the cumulative abruptness threshold; The cumulative stress is calculated by the Stonely formula, which includes: substituting the process parameters and the curvature radius of the substrate growth into the Stonely formula to calculate the cumulative stress, wherein the process parameters include the Young's modulus of the substrate material, the Poisson's ratio of the substrate, the thickness of the substrate and the thickness of the film; the curvature radius of the substrate growth is equal to the inverse of the average value of the curvature of the curved surface in the direction of the substrate growth. and the substrate growth.

2. The method of claim 1, wherein the method is characterized by, the average gray level of the growth image is equal to the average value of the gray levels of all pixel points in the growth image.

3. The method of claim 1, wherein the method is characterized by: the deposition rate satisfies the expression: ； wherein is the deposition rate at the time instant; is the average gray level of the growth image at the time instant; is the average gray level of the growth image at the time instant; is the image acquisition time interval; is the absolute value symbol.

4. The method of claim 1, wherein the method is characterized by: the repeated iteration until the cumulative displacement field of the whole growth period is obtained comprises the following steps: at the beginning of a new round, clearing the cumulative abruptness, calculating the incremental displacement field at the time of the new round based on the new reference image and superimposing the incremental displacement field on the new anchor point displacement field until the cumulative displacement field at the last time is obtained, stopping the iteration, and obtaining the cumulative displacement field of the whole growth period.

5. The method of claim 1, wherein the method is characterized by: The vertical component of the extracted cumulative displacement field of the whole growth cycle is subjected to quadratic surface fitting to determine the curvature, comprising: extracting the component perpendicular to the crystal surface in the cumulative displacement field, including and the vertical components of the two directions; performing quadratic surface fitting on all the vertical components to obtain a fitting formula, and taking the coefficients corresponding to and in the fitting formula as the curvatures of the surface in the and directions.

6. The method of claim 1, wherein the method is characterized by: the maximum stress of growth is obtained from a parameter table of a crystal material.

7. The method of claim 1, wherein the method is characterized by: The method further comprises: performing gray scale processing and mean filtering processing on the collected growth images.

Images