Mocvd carrier disk disk drift compensation method, system, computer device and storage medium

Abstract

The application provides a disc drift compensation method and system of an MOCVD bearing disc, a computer device and a storage medium, and relates to the technical field of semiconductor detection. The application obtains original measurement signals of multiple continuous circles of the bearing disc, constructs a joint optimization objective function based on a two-dimensional space mapping model to be optimized, an ideal sampling track and two-dimensional space translation parameters corresponding to each circle, characterizes disc drift as a two-dimensional space offset of an in-situ measurement probe relative to a sampling position on the surface of the bearing disc, synchronously solves the corresponding relationship between the space position of the surface of the bearing disc and the measured physical quantity, and the disc drift offset corresponding to each circle in the joint optimization process, and accurately determines the space offset compensation amount of the actual sampling track relative to the ideal sampling track. The original measurement signals are corrected based on the space offset compensation amount, which can simultaneously reduce the phase misplacement, amplitude error and waveform distortion caused by disc drift, and improve the accuracy and reliability of the in-situ measurement signals.

Description

Technical Field

[0001] This application relates to the field of semiconductor testing technology, and in particular to a method, system, computer equipment, and storage medium for compensating for disk drift of an MOCVD carrier disk. Background Technology

[0002] Metal-organic chemical vapor deposition (MOCVD) is a crucial process in the epitaxial growth of compound semiconductors, widely used in the fabrication of devices such as LEDs, laser diodes, and high-electron-mobility transistors. MOCVD typically involves introducing a metal-organic source and reactive gases into a high-temperature reaction chamber, where they react and deposit on the substrate surface to form an epitaxial thin film. Since the thickness, composition, stress state, and crystal quality of the epitaxial film directly affect device performance, precise monitoring and control of process parameters during the growth process are essential.

[0003] In MOCVD equipment, multiple substrates are typically supported by a carrier disk, which rotates during the growth process to improve the uniformity of the thermal distribution of the reaction gas, thereby enhancing the growth uniformity of the epitaxial film on the substrate surface. To obtain real-time growth status, existing equipment usually sets up in-situ measurement probes at preset positions in the reaction chamber. By fixing the probes, the substrate surface passing through the measurement area as the carrier disk rotates is continuously sampled to obtain raw measurement signals such as temperature, reflectivity, film thickness, and curvature.

[0004] However, in actual operation, due to factors such as wear of the transmission mechanism, assembly errors, uneven thermal expansion, and airflow disturbances, the carrier disk may experience susceptor drift. Susceptor drift causes the actual sampling position of the in-situ measurement probe to deviate from the ideal sampling position, resulting in misalignment of measurement signals in different revolutions on the time or angle axis; it can even cause amplitude errors and waveform distortion in the measurement signal due to uneven spatial distribution of the measured physical quantity on the substrate surface or carrier disk surface.

[0005] Traditional techniques typically employ peak alignment, cross-correlation between adjacent loops, or phase compensation to correct measurement signals. These methods primarily assume that only phase shift exists between signals from different loops, achieving a certain degree of alignment in time or angle. However, these methods struggle to correct amplitude errors and waveform distortions caused by spatial position shifts due to disk drift. Furthermore, compensation methods based on recursion between adjacent loops are prone to error accumulation, and the compensation parameters often fail to directly reflect the actual spatial drift state of the bearing disk, affecting the accuracy of in-situ measurement signals.

[0006] Therefore, there is an urgent need for a method that can identify the spatial offset of the bearing disk and compensate for the in-situ measurement signal to improve the accuracy of the in-situ measurement signal. Summary of the Invention

[0007] The purpose of this application is to provide a method, system, computer equipment, and storage medium for compensating for disk drift in MOCVD, so as to overcome the shortcomings of traditional technologies that only rely on phase shift or adjacent ring signal alignment for compensation, which leads to the inaccuracy of in-situ measurement signals.

[0008] Firstly, this application proposes a method for compensating for disk drift in MOCVD carrier disks, the method comprising: The original measurement signals of the MOCVD carrier disk during continuous rotation are acquired; the original measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed. The two-dimensional spatial mapping model characterizes the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under disk-drift-free conditions. The two-dimensional spatial translation parameters characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space for each loop. Based on the alternating optimization strategy, the joint optimization objective function is iteratively optimized to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; Based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory is determined, and the original measurement signal is corrected based on the spatial offset compensation amount to obtain the target measurement signal.

[0009] In one embodiment, the joint optimization objective function includes: The data fidelity term is used to characterize the error between the original measurement signal and the corresponding model prediction signal for each lap; wherein, the model prediction signal is obtained by spatially correcting the ideal sampling trajectory based on the two-dimensional spatial translation parameters of the corresponding lap, obtaining the actual sampling trajectory of the corresponding lap, and then sampling along the actual sampling trajectory from the two-dimensional spatial mapping model; The regularization term is used to characterize the difference between the two-dimensional spatial translation parameters corresponding to adjacent cycles, and to suppress the difference during the iterative optimization process of the joint optimization objective function, so as to constrain the physical continuity of the two-dimensional spatial translation parameters of each cycle. In one embodiment, the construction of a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop includes: Based on the data fidelity term and the regularization term, a joint optimization objective function is constructed, the expression of which is:

[0010] in, The sequence number of the circle; This represents the total number of cycles of the continuously acquired raw measurement signal; The sampling position in each lap; For the first Circle at the sampling location The raw measurement signal collected at the location; The two-dimensional spatial mapping model to be optimized; Sampling position in the ideal sampling trajectory The corresponding ideal sampling position; For the first The two-dimensional spatial translation parameters corresponding to the circle, and Indicates the first The disk drift offset in the first spatial direction. Indicates the first The disk drift offset in the second spatial direction; To obtain from the two-dimensional spatial mapping model along the first... The model prediction signal obtained by sampling the actual sampling trajectory of the circle; The regularization coefficient is used. This is a regularization term.

[0011] In one embodiment, the regularization term includes at least one of a first-order smoothing constraint, a second-order smoothing constraint, and a periodic consistency constraint; The first-order smoothing constraint is used to constrain the difference between the two-dimensional spatial translation parameters of adjacent cycles, the second-order smoothing constraint is used to constrain the difference between the changes in the two-dimensional spatial translation parameters of adjacent cycles, and the period consistency constraint is used to constrain the difference between the two-dimensional spatial translation parameters separated by a preset number of cycles.

[0012] In one embodiment, the step of iteratively optimizing the joint optimization objective function based on an alternating optimization strategy to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters includes: Initialize the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; In each iteration, the two-dimensional spatial translation parameters corresponding to each current circle are fixed, and the two-dimensional spatial mapping model is updated based on the original measurement signal; The updated two-dimensional spatial mapping model is fixed, and the two-dimensional spatial translation parameters corresponding to each circle are updated based on the error between the original measurement signal and the corresponding model prediction signal; wherein, the model prediction signal is obtained by sampling from the updated two-dimensional spatial mapping model along the actual sampling position of the corresponding circle; Repeat the update process of the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters until the joint optimization objective function satisfies the preset convergence condition, and obtain the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters.

[0013] In one embodiment, the method for initializing the two-dimensional spatial mapping model includes: Based on the original measurement signal of the first loop, the initial values ​​of the two-dimensional spatial mapping model are generated according to the ideal sampling trajectory; Alternatively, the initial values ​​of the two-dimensional spatial mapping model are generated based on the average signal of the original measurement signals from the first cycle for a consecutive preset number of cycles, according to the ideal sampling trajectory; Alternatively, with the two-dimensional spatial translation parameters corresponding to each circle initialized to zero, the original measurement signals of multiple circles are mapped to the surface spatial coordinate system of the bearing disk according to the ideal sampling trajectory, and the measurement values ​​mapped to the same spatial neighborhood are fused to generate the initial value of the two-dimensional spatial mapping model. In one embodiment, fixing the two-dimensional spatial translation parameters corresponding to each current circle and updating the two-dimensional spatial mapping model based on the original measurement signal includes: Based on the two-dimensional spatial translation parameters corresponding to each circle, the sampling positions corresponding to the original measurement signals of each circle are corrected to the actual sampling positions; The measured values ​​of the original measurement signals of each circle are mapped to the surface space coordinate system corresponding to the actual sampling position; The two-dimensional spatial mapping model is updated based on the measured values ​​mapped to the surface spatial coordinate system. In one embodiment, updating the two-dimensional spatial mapping model based on measurements mapped to the surface spatial coordinate system includes: Multiple measurements mapped to the same spatial neighborhood are fused to obtain an updated two-dimensional spatial mapping model; wherein, the fusion process includes weighted averaging, kernel function weighting, inverse distance weighting, interpolation fitting, or regression fitting.

[0014] In one embodiment, the fixed and updated two-dimensional spatial mapping model, and based on the error between the original measurement signal and the corresponding model prediction signal, updates the two-dimensional spatial translation parameters corresponding to each circle, including: For each original measurement signal, the ideal sampling trajectory is spatially corrected based on the candidate two-dimensional spatial translation parameters corresponding to that lap to obtain the candidate actual sampling trajectory; Sample along the candidate actual sampling trajectory from the updated two-dimensional spatial mapping model to obtain the corresponding model prediction signal; Based on the error between the original measurement signal and the corresponding model prediction signal, a two-dimensional spatial translation parameter is determined that minimizes the error.

[0015] In one embodiment, determining the two-dimensional spatial translation parameter that makes the error satisfy a preset minimization condition includes: The error between the original measurement signal of the circle and the corresponding model prediction signal is constructed as a nonlinear least squares problem, and the two-dimensional spatial translation parameters corresponding to the circle are solved using a preset algorithm; wherein, the preset algorithm includes at least one of the following: Levenberg-Marquardt algorithm, Gauss-Newton algorithm, gradient descent algorithm, grid search algorithm and coarse search combined with fine search algorithm.

[0016] In one embodiment, the method further includes: During the iterative optimization of the joint optimization objective function, the most recent preset number of original measurement signals are selected from the continuously acquired original measurement signals as the current optimization window according to the preset window length; Upon acquiring a new set of original measurement signals, the current optimization window is updated, and the joint optimization objective function is iteratively optimized based on the original measurement signals within the updated current optimization window.

[0017] In one embodiment, when there are multiple in-situ measurement probes, each in-situ measurement probe is set at a different preset position; the ideal sampling trajectory includes multiple ideal sampling sub-trajectories corresponding to each of the in-situ measurement probes. Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed, including: Based on the original measurement signals acquired by each of the in-situ measurement probes, the ideal sampling sub-trajectories corresponding to each of the in-situ measurement probes, the two-dimensional spatial translation parameters to be optimized, and the two-dimensional spatial mapping model to be optimized, the joint optimization objective function is constructed.

[0018] In one embodiment, the determination of the spatial offset compensation amount of the actual sampling trajectory of each loop relative to the ideal sampling trajectory based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, and the correction of the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal, includes: Based on the converged two-dimensional spatial translation parameters, the actual sampling position corresponding to each sampling point in each loop is determined. Based on the converged two-dimensional spatial mapping model, the model values ​​corresponding to the actual sampling positions and the model values ​​corresponding to the ideal sampling positions in the ideal sampling trajectory are determined respectively. The spatial offset compensation amount is determined based on the difference between the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position. The spatial offset compensation is superimposed on the corresponding original measurement signal to obtain the target measurement signal.

[0019] Secondly, this application proposes a disk drift compensation system for MOCVD carrier disks, the system comprising: The acquisition module is used to acquire the raw measurement signals of the MOCVD carrier disk during multiple rotations; the raw measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. A construction module is used to construct a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop. The two-dimensional spatial mapping model characterizes the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under disk drift-free conditions. The two-dimensional spatial translation parameters characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space for each loop. An optimization module is used to iteratively optimize the joint optimization objective function based on an alternating optimization strategy, so as to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters. The compensation module is used to determine the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, and to correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

[0020] Thirdly, this application proposes an in-situ measurement device, the device comprising: At least one in-situ measurement probe is set at a preset position in the reaction chamber to perform in-situ measurements on the substrate surface as it rotates through the measurement area, so as to obtain the original measurement signal during the epitaxial film growth process on the substrate surface; the type of the measured physical quantity corresponding to the original measurement signal includes at least one of temperature, reflectivity, film thickness, curvature, spectral intensity, ellipticity parameter or X-ray diffraction intensity; The MOCVD carrier disk drift compensation system described in the second aspect is used to generate a spatial offset compensation amount based on the original measurement signal, and to correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

[0021] Fourthly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method steps of the first aspect.

[0022] Fifthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the method steps of the first aspect.

[0023] The above-mentioned MOCVD disk drift compensation method, system, computer equipment, and storage medium have at least the following advantages: This application acquires the raw measurement signals from multiple consecutive turns of the MOCVD carrier disk and constructs a joint optimization objective function based on the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters corresponding to each turn. This transforms the carrier disk drift from a mere equivalent phase shift of the measurement signal on the time or angle axis into a two-dimensional spatial shift of the in-situ measurement probe relative to the sampling position on the carrier disk surface. Based on this, the application can simultaneously solve for the correspondence between the spatial position on the carrier disk surface and the measured physical quantity, as well as the disk drift offset corresponding to each turn, during the joint optimization process. This allows for accurate determination of the spatial offset compensation amount between the actual sampling trajectory and the ideal sampling trajectory. Correcting the raw measurement signal based on this spatial offset compensation amount can simultaneously reduce phase misalignment, amplitude error, and waveform distortion caused by disk drift, improving the accuracy and reliability of the MOCVD in-situ measurement signal. Furthermore, the two-dimensional spatial translation parameters obtained after convergence have clear spatial physical meaning and can be further used to characterize the carrier disk drift trajectory, providing data for equipment condition monitoring, fault early warning, and process backtracking. Attached Figure Description

[0024] Figure 1 This is a structural block diagram of an in-situ measurement device in one embodiment; Figure 2This is a flowchart illustrating a method for compensating for disk drift in an MOCVD carrier disk in one embodiment. Figure 3 This is a structural block diagram of a disk drift compensation system for an MOCVD carrier disk in one embodiment. Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0025] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0026] Some exemplary embodiments of this application have been described for illustrative purposes. It should be understood that this application may be implemented in other ways not specifically shown in the accompanying drawings.

[0027] Please see Figure 1 In one exemplary embodiment, this application provides an in-situ measurement device, including: a disk drift compensation system for an MOCVD carrier disk and at least one in-situ measurement probe.

[0028] At least one in-situ measurement probe is set at a preset position in the reaction chamber to perform in-situ measurements on the substrate surface that passes through the measurement area as the carrier disk rotates, so as to obtain the original measurement signal during the epitaxial film growth process on the substrate surface; the type of the measured physical quantity corresponding to the original measurement signal includes at least one of temperature, reflectivity, film thickness, curvature, spectral intensity, ellipsoid parameter or X-ray diffraction intensity.

[0029] Specifically, the MOCVD equipment includes a reaction chamber, a support plate disposed within the reaction chamber, and an in-situ measurement probe disposed at a preset position within the reaction chamber. The support plate is used to support multiple substrates. During epitaxial growth, the support plate rotates at a preset speed, causing each substrate to sequentially pass through the measurement area corresponding to the in-situ measurement probe.

[0030] Furthermore, the in-situ measurement probe can be positioned at the top of the reaction chamber corresponding to the measurement window, with the measurement optical path of the in-situ measurement probe facing the substrate surface on the carrier disk. When the carrier disk rotates, each substrate located on the carrier disk moves with the carrier disk and periodically passes through the measurement area. The in-situ measurement probe performs continuous in-situ measurements on the substrate surface passing through the measurement area to obtain the original measurement signals during the epitaxial film growth process on the substrate surface.

[0031] For example, the in-situ measurement probe in this application embodiment can be at least one of an optical pyrometer, a reflectivity measurement probe, a spectral measurement probe, an ellipsometry measurement probe, a curvature measurement probe, or an X-ray diffraction measurement probe. In response, the type of the measured physical quantity corresponding to the original measurement signal includes at least one of temperature, reflectivity, film thickness, curvature, spectral intensity, ellipsometry parameter, or X-ray diffraction intensity.

[0032] Furthermore, there can be multiple in-situ measurement probes, each of which is set at a different preset position in the reaction chamber. Each preset position can be arranged along the radial direction, circumferential direction, or different measurement window positions of the carrier plate.

[0033] Furthermore, in this embodiment, the original measurement signal is recorded according to the number of rotations of the carrier disk and the sampling position within each rotation. For example, the original measurement signal collected in the i-th rotation can be represented as a sequence of measurement values ​​corresponding to different sampling angles or different sampling times within the i-th rotation.

[0034] The MOCVD carrier disk drift compensation system is used to generate spatial offset compensation based on the original measurement signal, and to correct the original measurement signal based on the spatial offset compensation to obtain the target measurement signal. Specifically, the MOCVD carrier disk drift compensation system acquires the original measurement signals of the MOCVD carrier disk during multiple consecutive rotations. These original measurement signals are obtained through in-situ measurement probes positioned at preset locations within the reaction chamber. Based on the original measurement signals, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each rotation, a joint optimization objective function is constructed. The two-dimensional spatial mapping model characterizes the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk in a drift-free state. The two-dimensional spatial translation parameters characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space for each rotation. Based on an alternating optimization strategy, the joint optimization objective function is iteratively optimized to jointly solve for the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters. Based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, the spatial offset compensation amount of the actual sampling trajectory relative to the ideal sampling trajectory for each rotation is determined. The original measurement signals are then corrected based on this spatial offset compensation amount to obtain the target measurement signal. The aforementioned in-situ measurement device uses at least one in-situ measurement probe positioned at a predetermined location within the reaction chamber to perform in-situ measurements on the substrate surface as the carrier disk rotates through the measurement area. This acquires the original measurement signal during the epitaxial film growth process on the substrate surface. A joint optimization objective function is constructed based on the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters corresponding to each loop. This transforms the carrier disk drift from a mere equivalent phase shift of the measurement signal on the time or angular axis into a two-dimensional spatial offset of the in-situ measurement probe relative to the sampling position on the carrier disk surface. Based on this, the present application can simultaneously solve for the correspondence between the spatial position on the carrier disk surface and the measured physical quantity, as well as the disk drift offset corresponding to each loop, during the joint optimization process. This allows for accurate determination of the spatial offset compensation amount between the actual sampling trajectory and the ideal sampling trajectory. Correcting the original measurement signal based on this spatial offset compensation amount can simultaneously reduce phase misalignment, amplitude error, and waveform distortion caused by disk drift, thereby improving the accuracy and reliability of the MOCVD in-situ measurement signal. Furthermore, the two-dimensional spatial translation parameters obtained after convergence have clear spatial physical meanings and can be further used to characterize the disk drift trajectory of the bearing disk, providing data basis for equipment condition monitoring, fault early warning and process backtracking. Please see Figure 2 In one exemplary embodiment, this application provides a method for compensating for disk drift in MOCVD, specifically including the following steps: Step 202: Obtain the original measurement signals of the MOCVD carrier disk during multiple rotations. The original measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. Specifically, during the epitaxial growth process in an MOCVD (Multi-Layer Chemical Vapor Deposition) device, a carrier disk carries multiple substrates and rotates continuously at a preset speed. An in-situ measurement probe is positioned at a preset location within the reaction chamber, with its measurement optical path facing the substrate surface on the carrier disk. As the carrier disk rotates, the substrate periodically passes through the measurement area corresponding to the in-situ measurement probe. The probe continuously samples the substrate surface passing through this measurement area, thereby acquiring the original measurement signals during the epitaxial film growth process on the substrate surface.

[0035] For example, to facilitate subsequent labeling of disk drift parameters and signal compensation, this embodiment divides the acquired raw measurement signals according to the number of rotations of the carrier disk. For the first... The raw measurement signal obtained from the acquisition can be expressed as:

[0036] in, Indicates the number of revolutions of the carrier disk. This indicates the total number of consecutive data collection cycles. Indicates the first A sampling angle position or the sampling time corresponding to that sampling angle position. This indicates the number of sampling points in each lap. Indicates the first Circle at the sampling location The original measurement signal obtained at the location.

[0037] For example, the starting position of each revolution can be determined based on the synchronization pulse signal output by the rotary encoder of the carrier disk, and sampling can be performed at equal angular intervals within each revolution. For instance, each revolution can be divided into 3600 sampling points, corresponding to an angular resolution of 0.1°. Thus, the raw measurement signals acquired from multiple consecutive revolutions can form a signal sequence arranged by revolution number and sampling position, providing a data foundation for subsequent establishment of a two-dimensional spatial mapping model, solving for two-dimensional spatial translation parameters, and determining the spatial offset compensation amount.

[0038] Step 204: Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each circle, construct a joint optimization objective function. Specifically, after acquiring multiple consecutive rounds of raw measurement signals, a two-dimensional spatial mapping model to be optimized can be established. This two-dimensional spatial mapping model characterizes the correspondence between the spatial location on the surface of the support disk and the measured physical quantity corresponding to the raw measurement signal. For example, for any spatial location on the surface of the support disk, the two-dimensional spatial mapping model can provide the corresponding temperature value, reflectivity value, film thickness-related signal, or other measured physical quantity value at that location.

[0039] Alternatively, the two-dimensional spatial mapping model can be represented in different ways depending on the measurement accuracy requirements and computational resources.

[0040] In one embodiment, the two-dimensional spatial mapping model can be represented using a discrete mesh model. Specifically, the surface area of ​​the support disk can be discretized into multiple regular meshes, with each mesh point storing the measured physical quantity value at the corresponding spatial location. When it is necessary to obtain the model value at any spatial location, the model value at that spatial location can be determined by using bilinear interpolation, spline interpolation, or other interpolation methods based on the physical quantity values ​​of the mesh points surrounding that spatial location. This approach is simple to implement, has low computational cost, and is suitable for general accuracy requirements or online processing scenarios.

[0041] In another embodiment, the two-dimensional space mapping model can also be represented using a parameterized basis function model. Specifically, the two-dimensional space mapping model can be represented as a linear combination of multiple basis functions:

[0042] in, Indicates the number of basis functions. Indicates the first The coefficients to be optimized for each basis function. Indicates the first There are several basis functions. These basis functions can be at least one of Zernike polynomials, radial basis functions, or B-spline basis functions. By employing this scheme and using a parameterized basis function model, the number of variables to be optimized in the two-dimensional spatial mapping model can be reduced, thus improving solution efficiency.

[0043] In another embodiment, the two-dimensional spatial mapping model can also be represented by a neural network model. Specifically, the spatial coordinates of the bearing disk surface can be used as the input to the neural network, and the measured physical quantity value at the corresponding spatial location can be used as the output of the neural network. A neural network structure such as a multilayer perceptron is then used to fit the two-dimensional spatial mapping model. This approach can adapt to more complex spatial distribution relationships, and the network parameters can be optimized using a backpropagation algorithm.

[0044] The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under the condition of no disk drift. That is, in the absence of disk drift, the carrier disk maintains an ideal fixed-axis rotation, and the theoretical sampling path formed by the in-situ measurement probe relative to the surface of the carrier disk is the ideal sampling trajectory that the in-situ measurement probe should sample the substrate surface or the surface of the carrier disk in each revolution.

[0045] Two-dimensional spatial translation parameters are used to characterize the disk drift offset relative to the in-situ measurement probe in two-dimensional space during the corresponding revolution. For example, the two-dimensional spatial translation parameter for the i-th revolution can be expressed as: ,in This represents the offset of the i-th ring in the first spatial direction. This represents the offset of the i-th lap in the second spatial direction. Using this two-dimensional spatial translation parameter, the ideal sampling trajectory can be corrected to the actual sampling trajectory for the corresponding lap. Furthermore, based on the above, this application generates model prediction signals for each loop according to the two-dimensional spatial mapping model, the ideal sampling trajectory, and the two-dimensional spatial translation parameters for each loop. These model prediction signals characterize the signals that the in-situ measurement probe should theoretically acquire under the condition that the current two-dimensional spatial mapping model and the current two-dimensional spatial translation parameters are valid. Subsequently, the model prediction signals are compared with the actual acquired raw measurement signals to construct a joint optimization objective function. This joint optimization objective function is used to evaluate whether the current two-dimensional spatial mapping model and the two-dimensional spatial translation parameters can adequately explain the raw measurement signals across multiple loops.

[0046] Step 206: Based on the alternating optimization strategy, iteratively optimize the joint optimization objective function to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters. Specifically, since both the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters of each circle are unknowns, directly solving them simultaneously would result in high computational complexity. Therefore, this embodiment employs an alternating optimization strategy for iterative solving. In one iteration, one type of unknown is temporarily fixed, such as the two-dimensional spatial translation parameters corresponding to each circle, and the two-dimensional spatial mapping model is updated. Then, the updated result is fixed again, and the two-dimensional spatial translation parameters corresponding to each circle are updated in reverse order. This process is repeated continuously, gradually reducing the joint optimization objective function. When the changes in the two-dimensional spatial mapping model, the changes in the two-dimensional spatial translation parameters, or the number of iterations meet the preset convergence conditions, the iteration stops, yielding the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters.

[0047] For example, if the model change in two adjacent iterations is less than a first preset threshold, the root mean square of the change in the drift parameter in two adjacent iterations is less than a second preset threshold, or the preset maximum number of iterations is reached, then the preset convergence condition is considered to be met.

[0048] Step 208: Based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, determine the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory, and correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

[0049] Specifically, after completing the alternating optimization, the converged two-dimensional spatial translation parameter can represent the actual disk drift offset of each ring of the bearing disk relative to the in-situ measurement probe. Based on this two-dimensional spatial translation parameter, the spatial offset of the actual sampling trajectory relative to the ideal sampling trajectory in each ring can be determined.

[0050] Furthermore, based on the converged two-dimensional spatial mapping model, the measured physical quantity values ​​corresponding to each sampling position on the ideal sampling trajectory and the measured physical quantity values ​​corresponding to each sampling position on the actual sampling trajectory can be determined separately. The difference between the two can be used as a spatial offset compensation amount. This spatial offset compensation amount is the signal deviation that should be corrected in the original measurement signal. In this way, this application not only aligns the signals of different revolutions on the time axis or angle axis, but also converts disk drift into spatial sampling position offset for processing. Therefore, this application can compensate for the phase misalignment caused by disk drift, and also reduce the amplitude error and waveform distortion caused by the change of the actual sampling position, thereby improving the accuracy and reliability of MOCVD in-situ measurement signals.

[0051] Optionally, the joint optimization objective function includes: The data fidelity term is used to characterize the error between the original measurement signal and the corresponding model prediction signal for each lap. The model prediction signal is obtained by spatially correcting the ideal sampling trajectory based on the two-dimensional spatial translation parameters of the corresponding lap, and then sampling along the actual sampling trajectory from the two-dimensional spatial mapping model.

[0052] The regularization term is used to characterize the differences between the two-dimensional spatial translation parameters of adjacent cycles and to suppress these differences during the iterative optimization of the joint objective function, thereby constraining the physical continuity of the two-dimensional spatial translation parameters of each cycle.

[0053] Specifically, this application elevates the problem of MOCVD disk drift from time alignment to the level of spatial coordinate-physical quantity mapping. That is, this application no longer understands disk drift merely as the phase shift of the measurement signal on the time axis or angle axis, but rather as the spatial offset of the in-situ measurement probe relative to the actual sampling position on the surface of the disk.

[0054] For example, embodiments of this application establish a two-dimensional spatial mapping model of the physical quantities to be measured on the surface of the bearing disk. Used to characterize the spatial position on the surface of the bearing disk The correspondence between the measured physical quantity and the measured physical quantity, wherein the measured physical quantity includes at least one of temperature, reflectivity, film thickness, curvature, spectral intensity, ellipsoid parameter or X-ray diffraction intensity.

[0055] Regarding the first The ideal sampling trajectory of the in-situ measurement probe, assuming no disk drift, is represented by the original measurement signal as follows: ;in, The sampling position in each revolution can be an angular position, a rotation phase, or a time sampling position corresponding to the angular position.

[0056] In the event of disc drift, the first... The two-dimensional spatial translation parameters corresponding to the circle can be expressed as: ;in, Indicates the first The disk drift offset in the first spatial direction. Indicates the first The disk drift offset in the second spatial direction.

[0057] Based on this two-dimensional spatial translation parameter, the first... The actual sampling position of the circle is represented as follows: .

[0058] Therefore, the first Circle at the sampling location The original measurement signal at that location can be described as:

[0059] in, For the first Circle at the sampling location The raw measurement signal collected at the location; To obtain from the two-dimensional spatial mapping model along the first... The model prediction signal obtained by sampling the actual sampling trajectory; This is the noise term.

[0060] Optionally, based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed, including: The joint optimization objective function is constructed based on the data fidelity term and the regularization term, and its expression is as follows:

[0061] in, The sequence number of the circle; This represents the total number of cycles of the continuously acquired raw measurement signal; The sampling position in each lap; For the first Circle at the sampling location The raw measurement signal collected at the location; The two-dimensional spatial mapping model to be optimized; Sampling position in the ideal sampling trajectory The corresponding ideal sampling position; For the first The two-dimensional spatial translation parameters corresponding to the circle, and Indicates the first The disk drift offset in the first spatial direction. Indicates the first The disk drift offset in the second spatial direction; To obtain from the two-dimensional spatial mapping model along the first... The model prediction signal obtained by sampling the actual sampling trajectory of the circle; This is the regularization coefficient, used to balance the degree of data fit and the requirements for smoothness. It can be set through cross-validation or empirically. For example, ; This is a regularization term.

[0062] Optionally, the regularization term includes at least one of a first-order smoothing constraint, a second-order smoothing constraint, and a periodic consistency constraint. The first-order smoothing constraint is used to constrain the differences between the two-dimensional spatial translation parameters of adjacent cycles, the second-order smoothing constraint is used to constrain the differences between the changes in the two-dimensional spatial translation parameters of adjacent cycles, and the periodic consistency constraint is used to constrain the differences between the two-dimensional spatial translation parameters separated by a preset number of cycles.

[0063] Specifically, the regularization term in the embodiments of this application can be set according to the physical change characteristics of the disk drift. Since the disk drift generated during continuous rotation of the disk usually has mechanical continuity and will not undergo irregular large jumps between adjacent rotations, smoothness or periodic consistency constraints can be introduced into the joint optimization objective function to improve the stability of the solution results for the two-dimensional spatial translation parameters.

[0064] For example, the regularization term can employ a first-order smoothing constraint to constrain the difference between the two-dimensional spatial translation parameters of adjacent loops, and its expression is:

[0065] in, This represents the total number of revolutions of the continuously acquired raw measurement signal. This constraint is used to suppress abrupt changes in the disk drift offset between adjacent revolutions, ensuring that the solved disk drift trajectory remains continuous in the revolution direction.

[0066] For example, the regularization term can also employ a second-order smoothing constraint to constrain the differences between the variations in the two-dimensional spatial translation parameters of adjacent loops, and its expression is:

[0067] This constraint is used to suppress abrupt changes in the rate of change of the disk drift, making the change trend of the two-dimensional spatial translation parameters smoother. It is suitable for scenarios where the disk drift changes slowly over time.

[0068] For example, when the bearing disk drift has periodic variation characteristics, the regularization term can also employ a periodic consistency constraint to constrain the difference between two-dimensional spatial translation parameters spaced apart by a preset number of periods, and its expression is:

[0069] in, This indicates the preset number of cycle rotations. This constraint is used to ensure that the disk drift parameters are consistent or nearly consistent at intervals of the preset number of cycle rotations. It is applicable to scenarios where disk drift is related to the mechanical structure of the bearing disk, spindle eccentricity, or periodic disturbances.

[0070] It should be noted that the first-order smoothing constraint, second-order smoothing constraint, and periodic consistency constraint mentioned above can be used individually or in combination. For example, the regularization term can be expressed as:

[0071] in, To jointly optimize the regularization term in the objective function, , , These are the weighting coefficients corresponding to the first-order smoothing constraint, the second-order smoothing constraint, and the periodic consistency constraint, respectively. By adjusting these weighting coefficients, the continuity, smoothness, and periodic consistency of the two-dimensional spatial translation parameters can be constrained according to the disk drift characteristics of different MOCVD equipment.

[0072] By employing the above scheme, this application constructs a joint optimization objective function including a data fidelity term and a regularization term. This function can uniformly correlate the original measurement signals of each loop, the two-dimensional spatial mapping model, the ideal sampling trajectory, and the two-dimensional spatial translation parameters. The data fidelity term characterizes the error between the original measurement signals of each loop and the corresponding model prediction signals, enabling the joint optimization objective function to constrain the model prediction signals to be as close as possible to the actual acquired signals. This establishes a quantitative relationship between disk drift spatial offset and the original measurement signals. Compared to methods based solely on peak alignment or phase alignment of adjacent loops, this application can describe the impact of disk drift on the measurement signals from the perspective of changes in spatial sampling positions, providing a basis for subsequent correction of phase misalignment, amplitude error, and waveform distortion caused by disk drift.

[0073] Furthermore, this application introduces a regularization term into the joint optimization objective function, and makes the regularization term characterize the difference between the two-dimensional spatial translation parameters of adjacent rings. This leverages the characteristic that the bearing disk drift typically exhibits mechanical continuity during continuous rotation to impose physical constraints on the two-dimensional spatial translation parameters of each ring. This suppresses abrupt changes in the two-dimensional spatial translation parameters caused by noise, abnormal sampling points, or local signal fluctuations, preventing the disk drift parameters from deviating from the actual mechanical motion law. It improves the constraint stability and noise resistance of the joint optimization objective function on the disk drift spatial offset, thereby enhancing the accuracy and reliability of the in-situ measurement signal compensation results.

[0074] Optionally, based on an alternating optimization strategy, the joint optimization objective function is iteratively optimized to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters, including: Initialize the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; In each iteration, the two-dimensional spatial translation parameters corresponding to each current circle are fixed, and the two-dimensional spatial mapping model is updated based on the original measurement signal; The updated two-dimensional spatial mapping model is fixed, and the two-dimensional spatial translation parameters corresponding to each circle are updated based on the error between the original measurement signal and the corresponding model prediction signal; wherein, the model prediction signal is obtained by sampling from the updated two-dimensional spatial mapping model along the actual sampling position of the corresponding circle; Repeat the update process of the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters until the joint optimization objective function satisfies the preset convergence condition, and obtain the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters.

[0075] Specifically, since the two-dimensional spatial translation parameters corresponding to each circle have not been accurately solved in the initial stage, it is impossible to directly determine the actual sampling position of the original measurement signal of each circle in the spatial coordinate system on the surface of the bearing disk. Therefore, this application first generates the initial value of the two-dimensional spatial mapping model based on the original measurement signal and the ideal sampling trajectory.

[0076] Optionally, the methods for initializing the two-dimensional spatial mapping model include: Based on the original measurement signal from the first loop, the initial values ​​of the two-dimensional spatial mapping model are generated according to the ideal sampling trajectory; Alternatively, based on the average signal of the original measurement signals from a continuous preset number of laps starting from the first lap, the initial values ​​of the two-dimensional spatial mapping model are generated according to the ideal sampling trajectory; Alternatively, with the two-dimensional spatial translation parameters corresponding to each circle initialized to zero, the original measurement signals from multiple circles are mapped to the surface spatial coordinate system of the bearing disk according to the ideal sampling trajectory, and the measurement values ​​mapped to the same spatial neighborhood are fused to generate the initial values ​​of the two-dimensional spatial mapping model.

[0077] For example, in one embodiment, this application generates initial values ​​for a two-dimensional spatial mapping model based on the first round of original measurement signals, following an ideal sampling trajectory. That is, for the original measurement signals at the sampling positions in the first round, these are used as the initial physical quantity values ​​at the ideal sampling positions, and the initial values ​​for the two-dimensional spatial mapping model are generated accordingly. For spatial locations not covered by the first round of original measurement signals, they can be filled in by interpolation, copying and expansion, or preset fill values. Using the above scheme, the calculation process is simple and can quickly establish the initial correspondence between the spatial positions on the surface of the bearing plate and the measured physical quantities.

[0078] In another embodiment, the average signal of the original measurement signals from a predetermined number of consecutive revolutions starting from the first revolution can be mapped onto the surface space coordinate system of the bearing disk according to an ideal sampling trajectory to generate the initial values ​​of the two-dimensional spatial mapping model. Using this approach, the influence of single-revolution noise can be reduced by averaging signals from multiple revolutions, thereby improving the stability of the initial model.

[0079] In another embodiment, with the two-dimensional spatial translation parameters corresponding to each circle initialized to zero, the original measurement signals from multiple circles can be mapped onto the surface spatial coordinate system of the bearing disk according to an ideal sampling trajectory. Measurements mapped to the same spatial neighborhood are then fused to generate the initial values ​​of the two-dimensional spatial mapping model. That is, assuming no disk drift in each circle, the initial value of the two-dimensional spatial translation parameters for each circle is set to zero, and the sampling points in the original measurement signals of each circle are mapped onto the surface spatial coordinate system of the bearing disk according to an ideal sampling trajectory. Further, for multiple measurement values ​​mapped to the same spatial location or the same spatial neighborhood, methods such as average fusion, weighted average fusion, kernel function weighted fusion, or interpolation fusion can be used to process them, thereby obtaining the initial values ​​of the two-dimensional spatial mapping model. Using the above scheme, with the two-dimensional spatial translation parameters initialized to zero, mapping and fusing the original measurement signals from multiple circles to generate initial values ​​can fully utilize multi-circle data and improve the representativeness of the initial two-dimensional spatial mapping model for the overall measurement data.

[0080] Furthermore, when iteratively optimizing the joint optimization objective function based on the alternating optimization strategy, this application splits the joint optimization problem into two sub-problems: one sub-problem is used to update the two-dimensional spatial mapping model with fixed two-dimensional spatial translation parameters, and the other sub-problem is used to update the two-dimensional spatial translation parameters with fixed two-dimensional spatial mapping model.

[0081] Optionally, the two-dimensional spatial translation parameters corresponding to each current lap are fixed, and the two-dimensional spatial mapping model is updated based on the original measurement signal, including: Based on the two-dimensional spatial translation parameters corresponding to each circle, the sampling positions corresponding to the original measurement signals of each circle are corrected to the actual sampling positions; The measured values ​​of the original measurement signals of each circle are mapped to the surface space coordinate system corresponding to the actual sampling position; The two-dimensional spatial mapping model is updated based on the measurements mapped to the surface spatial coordinate system.

[0082] Specifically, in the first In the next iteration, the two-dimensional spatial translation parameters corresponding to each current circle are fixed first, and the two-dimensional spatial mapping model is updated based on the original measurement signal.

[0083] For the Sampling position in the original measurement signal In the absence of disk drift, the ideal sampling position is: .

[0084] In the current iteration, the first The expression for the two-dimensional spatial translation parameter corresponding to the circle is:

[0085] Based on the current two-dimensional spatial translation parameters, the sampling position is corrected to the actual sampling position, and its expression is:

[0086] in, Indicates the first Circle at the sampling location The corresponding actual sampling location.

[0087] Subsequently, the first Circle at the sampling location The original measurement signal at the location As the actual sampling location The observed values ​​at each location are mapped to the surface space coordinate system of the bearing disk, and the original measurement signals of each circle are mapped to the surface space coordinate system of the bearing disk.

[0088] Optionally, based on measurements mapped to the surface space coordinate system, the two-dimensional spatial mapping model is updated, including: Multiple measurements mapped to the same spatial neighborhood are fused to obtain an updated two-dimensional spatial mapping model; the fusion process includes weighted averaging, kernel function weighting, inverse distance weighting, interpolation fitting, or regression fitting.

[0089] Specifically, for any spatial position in the spatial coordinate system of the bearing disk surface This allows for the collection of multiple observations mapped to the neighborhood of a given spatial location, followed by fusion of these observations to obtain the model value at that location. For example, a kernel-weighted approach can be used to update the two-dimensional spatial mapping model, with the following expression:

[0090] in, Indicates the first The updated two-dimensional spatial mapping model in spatial location Model value at; This is a weighting function or kernel function used to determine the weights based on the distance between the observed location and the spatial location to be updated. The closer the distance, the larger the weight can be. For example, the weighting function can be a Gaussian kernel function, an inverse distance weighting function, or another kernel function.

[0091] In other embodiments, weighted average, inverse distance weighting, interpolation fitting, regression fitting, regular grid interpolation, or kriging interpolation can be used to fuse multiple measurements mapped to the same spatial neighborhood to obtain an updated two-dimensional spatial mapping model.

[0092] Optionally, the updated two-dimensional spatial mapping model is fixed, and the two-dimensional spatial translation parameters corresponding to each loop are updated based on the error between the original measured signal and the corresponding model prediction signal, including: For each loop of original measurement signal, the ideal sampling trajectory is spatially corrected based on the candidate two-dimensional spatial translation parameters corresponding to that loop to obtain the candidate actual sampling trajectory; sampling is performed along the candidate actual sampling trajectory from the updated two-dimensional spatial mapping model to obtain the corresponding model prediction signal; based on the error between the original measurement signal and the corresponding model prediction signal for that loop, the two-dimensional spatial translation parameters that minimize the error are determined.

[0093] Specifically, for the first The original measurement signal of the circle is used as the two-dimensional spatial translation parameter corresponding to the circle, which is denoted as the variable to be solved: .

[0094] Based on candidate two-dimensional spatial translation parameters Spatial correction is performed on the ideal sampling trajectory to obtain the candidate actual sampling trajectory, which is expressed as: .

[0095] Furthermore, from the updated two-dimensional spatial mapping model Sampling is performed along the candidate actual sampling trajectory to obtain the first... The model prediction signal corresponding to the circle is expressed as follows: 。

[0096] Optionally, the two-dimensional spatial translation parameters that minimize the error are determined, including: The error between the original measurement signal and the corresponding model prediction signal of the circle is constructed as a nonlinear least squares problem, and the two-dimensional spatial translation parameters corresponding to the circle are solved by a preset algorithm; wherein, the preset algorithm includes at least one of the following: Levenberg-Marquardt algorithm, Gauss-Newton algorithm, gradient descent algorithm, grid search algorithm and coarse search combined with fine search algorithm.

[0097] Specifically, the first The error between the original measured signal and the corresponding model predicted signal is constructed as a nonlinear least squares problem, and the two-dimensional spatial translation parameter that minimizes this error is solved. Its expression is:

[0098] in, Indicates the first Circle at the sampling location The original measurement signal at the location; This represents the updated two-dimensional spatial mapping model; Represents the weighting coefficients of the local smoothing term; This represents a local smoothing term, used to constrain the difference between the current loop's 2D spatial translation parameters and those of adjacent loops or loops before and after it, as needed. It should be noted that if the joint optimization objective function already includes a global regularization term, a separate local smoothing term may not be necessary.

[0099] By adopting the above scheme, this application decomposes the complex problem of simultaneously solving two unknown parameters into two relatively easy sub-problems, reducing the overall solution difficulty and improving the algorithm's feasibility and convergence stability.

[0100] Furthermore, by employing the aforementioned iterative optimization algorithm, when updating the two-dimensional spatial mapping model, this application can map the original measurement signals from multiple loops to the surface spatial coordinate system of the bearing disk based on the current two-dimensional spatial translation parameters, and reconstruct the spatial distribution of the measured physical quantity through fusion processing. When updating the two-dimensional spatial translation parameters, the updated two-dimensional spatial mapping model can also be used to inversely constrain the disk drift offset of each loop, gradually reducing the error between the original measurement signals of each loop and the model's predicted signals. Through the aforementioned iterative process, the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters can mutually correct each other and gradually approximate the true state.

[0101] Therefore, this application can fully utilize the spatial correlation information and temporal continuity information in the continuous multi-loop original measurement signals, avoiding the problem of error propagation loop by loop in traditional adjacent loop recursive compensation, and improving the accuracy of disk drift parameter identification. At the same time, since the two-dimensional spatial mapping model is updated jointly by multi-loop data during the iteration process, it can reduce the impact of single-loop noise, abnormal sampling points or local fluctuations on the compensation results, thereby improving the accuracy, stability and reliability of the target measurement signal.

[0102] Optionally, the above-mentioned MOCVD carrier disk drift compensation method further includes: During the iterative optimization of the joint optimization objective function, the most recent preset number of original measurement signals are selected from the continuously acquired original measurement signals as the current optimization window according to the preset window length. When a new round of original measurement signals is acquired, the current optimization window is updated, and the joint optimization objective function is iteratively optimized based on the original measurement signals in the updated current optimization window.

[0103] Specifically, during the continuous operation of the MOCVD equipment, the in-situ measurement probe continuously acquires the raw measurement signals during the rotation of the carrier disk, and selects the raw measurement signal from the most recent preset number of revolutions as the current optimization window. For example, in this embodiment, the preset window length is set to 100 revolutions. In practical applications, its specific value can be determined based on the carrier disk rotation speed, sampling frequency, computing resources, and real-time compensation requirements.

[0104] In the initial stage, once the preset window length is reached, the original measurement signals within that window length are used as the current optimization window, and a joint optimization objective function is constructed or updated based on the original measurement signals within the current optimization window. Subsequently, the data within the current optimization window is iteratively optimized using an alternating optimization strategy to solve for the corresponding two-dimensional spatial mapping model and the two-dimensional spatial translation parameters of each cycle within the current optimization window.

[0105] Upon acquiring a new set of raw measurement signals, the new signal is added to the current optimization window, while the oldest signal from that cycle is removed. This ensures that the current optimization window always includes raw measurement signals from the most recent preset number of cycles. Then, based on the updated raw measurement signals within the current optimization window, the joint objective function is iteratively optimized.

[0106] By adopting the above scheme, since the current optimization window always contains the most recently acquired multi-cycle original measurement signals, it can retain the temporal continuity and spatial correlation information required for multi-cycle joint optimization, thus giving the solution of the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters better stability. Furthermore, using the solution results of the previous optimization window as initial values ​​can reduce the convergence time of iterative optimization and improve online processing efficiency.

[0107] Optionally, when there are multiple in-situ measurement probes, each in-situ measurement probe is set at a different preset position; the ideal sampling trajectory includes multiple ideal sampling sub-trajectories corresponding to each in-situ measurement probe. Based on the original measurement signals, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed, including: based on the original measurement signals acquired by each in-situ measurement probe, the ideal sampling sub-trajectory corresponding to each in-situ measurement probe, the two-dimensional spatial translation parameters to be optimized, and the two-dimensional spatial mapping model to be optimized, a joint optimization objective function is constructed.

[0108] By adopting the above scheme, the original measurement signals collected at different preset locations are incorporated into the same joint optimization objective function, so that the two-dimensional spatial mapping model is constructed by integrating measurement data on multiple ideal sampling sub-trajectories. This can expand the coverage of spatial information on the surface of the bearing disk and improve the ability of the two-dimensional spatial mapping model to represent the spatial distribution of the measured physical quantity.

[0109] In addition, different in-situ measurement probes can be used to collect different measured physical quantities, thereby realizing the synergistic use of multi-source in-situ measurement data, improving the reliability of disk drift compensation results, and facilitating the acquisition of more accurate target measurement signals.

[0110] Optionally, based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, the spatial offset compensation amount of the actual sampling trajectory in each loop relative to the ideal sampling trajectory is determined, and the original measurement signal is corrected based on the spatial offset compensation amount to obtain the target measurement signal, including: Based on the converged two-dimensional spatial translation parameters, the actual sampling position corresponding to each sampling point in each loop is determined; based on the converged two-dimensional spatial mapping model, the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position in the ideal sampling trajectory are determined respectively; the spatial offset compensation amount is determined according to the difference between the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position; the spatial offset compensation amount is superimposed on the corresponding original measurement signal to obtain the target measurement signal.

[0111] Specifically, the spatial offset compensation amount can reflect the signal deviation caused by the sampling position shift due to disk drift. Correcting the original measurement signal based on this spatial offset compensation amount can reduce the amplitude error and waveform distortion caused by the actual sampling position deviating from the ideal sampling position, making the target measurement signal closer to the measurement result under the state of no disk drift.

[0112] Optionally, after signal correction, the converged two-dimensional spatial translation parameters for each circle can be output, and a disk drift parameter trajectory for the bearing disk can be generated based on these parameters. Specifically, the disk drift offset in the first and second spatial directions for each circle can be output sequentially, and the change of the disk drift parameter over time or circle number can be displayed as a curve. This disk drift parameter trajectory can be directly used for mechanical condition monitoring, fault early warning, and process backtracking. For example, during mechanical condition monitoring, it can be observed whether the disk drift parameter shows a trend change over time or circle number; if the disk drift offset in a certain direction continues to increase, it may indicate problems such as thermal deformation of the bearing disk, spindle misalignment, bearing wear, or abnormal transmission mechanism. During fault early warning, the disk drift amount can be compared with a preset drift threshold. When the disk drift amount exceeds the preset drift threshold, the system outputs a warning message to prompt maintenance personnel to inspect and maintain the bearing disk, bearings, transmission mechanism, or related installation structure. During process backtracking, signal anomalies in a specific cycle can be correlated with the disk drift parameters of the corresponding cycle to determine whether the original measurement signal anomaly in that cycle was caused by disk drift, thereby improving the accuracy of anomaly cause location.

[0113] The aforementioned MOCVD carrier disk drift compensation method establishes a two-dimensional spatial mapping model to characterize the correspondence between the spatial position of the carrier disk surface and the measured physical quantity, and characterizes the disk drift of each revolution as a two-dimensional spatial translation parameter. This transforms disk drift compensation from traditional phase alignment of the time and angle axes into sampling position correction in a spatial coordinate system. Therefore, this application can determine the difference in physical quantity caused by disk drift based on the spatial offset between the actual sampling trajectory and the ideal sampling trajectory, and compensate for the original measurement signal. This not only corrects the phase misalignment between signals from different revolutions but also reduces amplitude errors and waveform distortion caused by sampling position offsets, improving the accuracy and reliability of in-situ measurement signals.

[0114] Furthermore, this application constructs a joint optimization objective function based on continuous multi-cycle raw measurement signals and jointly solves the two-dimensional spatial mapping model and the corresponding two-dimensional spatial translation parameters for each cycle. This ensures that the data for each cycle are constrained relative to the same spatial mapping model, avoiding the problem of error propagation and accumulation in traditional recursive compensation between adjacent cycles. In addition, by fusing multi-cycle measurement data in the spatial coordinate system, the influence of single-cycle noise, local anomalies, or signal fluctuations on the compensation results can be reduced, improving the stability of disk drift parameter identification and target measurement signal generation.

[0115] Furthermore, this application introduces a regularization term into the joint optimization objective function to constrain the differences between the two-dimensional spatial translation parameters of adjacent circles. This can take advantage of the mechanical continuity of the bearing disk drift to suppress parameter abrupt changes that do not conform to the actual mechanical motion law, making the obtained disk drift trajectory smoother and more reasonable, thereby improving the noise resistance and physical reliability of the compensation process.

[0116] Furthermore, the two-dimensional spatial translation parameters output by this application can characterize the actual disk drift offset in the first and second spatial directions, and have a clear physical meaning. The disk drift parameter trajectory formed based on the two-dimensional spatial translation parameters of each circle can be used for various applications such as mechanical condition monitoring, fault early warning, and process backtracking.

[0117] Furthermore, this application does not rely on whether the original measurement signal has obvious peak characteristics, but rather on compensation based on spatial mapping relationships. Therefore, it is applicable to various in-situ measurement signals such as temperature, reflectivity, film thickness, curvature, spectral intensity, ellipticity parameters, and X-ray diffraction intensity, exhibiting good versatility. The converged two-dimensional spatial mapping model can also serve as a spatial distribution characterization of the measured physical quantities on the surface of the support disk or substrate, used for subsequent process diagnosis, process optimization, and equipment status analysis, providing a data foundation for the accurate monitoring and stable control of the MOCVD epitaxial growth process. It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0118] Based on the same inventive concept, this application also provides a disk drift compensation system for MOCVD carrier disks. This system is applicable to the above-mentioned disk drift compensation method for MOCVD carrier disks. The solution provided by this system is similar to the solution described in the above-mentioned method. Therefore, the specific limitations of one or more system embodiments provided below can be found in the limitations of the method above, and will not be repeated here.

[0119] Please see Figure 3 In one embodiment, the MOCVD carrier disk drift compensation system includes: an acquisition module, a construction module, an optimization module, and a compensation module.

[0120] The acquisition module is used to acquire the raw measurement signals of the MOCVD carrier disk during continuous rotation. The raw measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. The construction module is used to construct a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop. The two-dimensional spatial mapping model is used to characterize the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under the state of no disk drift. The two-dimensional spatial translation parameters are used to characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space in the corresponding loop. The optimization module is used to iteratively optimize the joint optimization objective function based on an alternating optimization strategy, so as to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters.

[0121] The compensation module is used to determine the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, and to correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

[0122] Optionally, the joint optimization objective function includes: a data fidelity term, used to characterize the error between the original measurement signal and the corresponding model prediction signal for each cycle; wherein the model prediction signal is obtained by spatially correcting the ideal sampling trajectory based on the two-dimensional spatial translation parameters of the corresponding cycle to obtain the actual sampling trajectory of the corresponding cycle, and then sampling along the actual sampling trajectory from the two-dimensional spatial mapping model. A regularization term, used to characterize the difference between the two-dimensional spatial translation parameters of adjacent cycles, and suppressing the difference during the iterative optimization process of the joint optimization objective function to constrain the physical continuity of the two-dimensional spatial translation parameters of each cycle.

[0123] Optionally, the construction module constructs a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, including: The joint optimization objective function is constructed based on the data fidelity term and the regularization term, and its expression is as follows:

[0124] in, The sequence number of the circle; This represents the total number of cycles of the continuously acquired raw measurement signal; The sampling position in each lap; For the first Circle at the sampling location The raw measurement signal collected at the location; The two-dimensional spatial mapping model to be optimized; Sampling position in the ideal sampling trajectory The corresponding ideal sampling position; For the first The two-dimensional spatial translation parameters corresponding to the circle, and Indicates the first The disk drift offset in the first spatial direction. Indicates the first The disk drift offset in the second spatial direction; To obtain from the two-dimensional spatial mapping model along the first... The model prediction signal obtained by sampling the actual sampling trajectory of the circle; This is the regularization coefficient, used to balance the degree of data fit and the requirements for smoothness. It can be set through cross-validation or empirically. For example, ; This is a regularization term.

[0125] Optionally, the optimization module iteratively optimizes the joint optimization objective function based on an alternating optimization strategy to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters. This includes: initializing the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; fixing the two-dimensional spatial translation parameters corresponding to each current cycle during each iteration and updating the two-dimensional spatial mapping model based on the original measurement signal; fixing the updated two-dimensional spatial mapping model and updating the two-dimensional spatial translation parameters corresponding to each cycle based on the error between the original measurement signal and the corresponding model prediction signal; wherein, the model prediction signal is obtained by sampling from the updated two-dimensional spatial mapping model along the actual sampling position of the corresponding cycle; and repeating the update process of the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters until the joint optimization objective function satisfies the preset convergence condition, thereby obtaining the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters.

[0126] Optionally, the optimization module initializes the two-dimensional spatial mapping model in the following ways: generating initial values ​​of the two-dimensional spatial mapping model based on the first round of original measurement signals according to an ideal sampling trajectory; or, generating initial values ​​of the two-dimensional spatial mapping model based on the average signal of the original measurement signals of a consecutive preset number of rounds starting from the first round according to an ideal sampling trajectory; or, when the two-dimensional spatial translation parameters corresponding to each round are initialized to zero, mapping the original measurement signals of multiple rounds to the surface spatial coordinate system of the bearing disk according to the ideal sampling trajectory, and fusing the measurement values ​​mapped to the same spatial neighborhood to generate initial values ​​of the two-dimensional spatial mapping model.

[0127] Optionally, the optimization module fixes the two-dimensional spatial translation parameters corresponding to each current circle and updates the two-dimensional spatial mapping model based on the original measurement signals. This includes: correcting the sampling positions corresponding to the original measurement signals of each circle to the actual sampling positions based on the two-dimensional spatial translation parameters corresponding to each current circle; mapping the measured values ​​of the original measurement signals of each circle to the surface spatial coordinate system corresponding to the actual sampling positions; and updating the two-dimensional spatial mapping model based on the measured values ​​mapped to the surface spatial coordinate system. The updating of the two-dimensional spatial mapping model based on the measured values ​​mapped to the surface spatial coordinate system includes: fusing multiple measured values ​​mapped to the same spatial neighborhood to obtain the updated two-dimensional spatial mapping model; wherein the fusing process includes weighted averaging, kernel function weighting, inverse distance weighting, interpolation fitting, or regression fitting.

[0128] Optionally, the optimization module fixes the updated two-dimensional spatial mapping model and updates the two-dimensional spatial translation parameters corresponding to each loop based on the error between the original measurement signal and the corresponding model prediction signal. This includes: for each loop's original measurement signal, spatially correcting the ideal sampling trajectory based on the candidate two-dimensional spatial translation parameters corresponding to that loop to obtain a candidate actual sampling trajectory; sampling along the candidate actual sampling trajectory from the updated two-dimensional spatial mapping model to obtain the corresponding model prediction signal; and determining the two-dimensional spatial translation parameters that satisfy the preset minimization condition based on the error between the original measurement signal and the corresponding model prediction signal for that loop. Determining the two-dimensional spatial translation parameters that satisfy the preset minimization condition includes: constructing the error between the original measurement signal and the corresponding model prediction signal for that loop as a nonlinear least squares problem, and solving for the two-dimensional spatial translation parameters corresponding to that loop using a preset algorithm; wherein the preset algorithm includes at least one of the following: Levenberg-Marquardt algorithm, Gauss-Newton algorithm, gradient descent algorithm, grid search algorithm, and a combination of coarse and fine search algorithms.

[0129] Optionally, the optimization module is also used to select the most recent preset number of original measurement signals from the continuously acquired original measurement signals as the current optimization window according to the preset window length during the iterative optimization of the joint optimization objective function; when a new round of original measurement signals is acquired, the current optimization window is updated, and the joint optimization objective function is iteratively optimized based on the original measurement signals in the updated current optimization window.

[0130] Optionally, when there are multiple in-situ measurement probes, each in-situ measurement probe is set at a different preset position; the ideal sampling trajectory includes multiple ideal sampling sub-trajectories corresponding to each in-situ measurement probe. The construction module constructs a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop. This includes: constructing a joint optimization objective function based on the original measurement signal acquired by each in-situ measurement probe, the ideal sampling sub-trajector corresponding to each in-situ measurement probe, the two-dimensional spatial translation parameters to be optimized, and the two-dimensional spatial mapping model to be optimized.

[0131] Optionally, the compensation module determines the spatial offset compensation amount of the actual sampling trajectory relative to the ideal sampling trajectory for each loop based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, and corrects the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal. This includes: determining the actual sampling position corresponding to each sampling point in each loop based on the converged two-dimensional spatial translation parameters; determining the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position in the ideal sampling trajectory based on the converged two-dimensional spatial mapping model; determining the spatial offset compensation amount based on the difference between the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position; and superimposing the spatial offset compensation amount onto the corresponding original measurement signal to obtain the target measurement signal.

[0132] The aforementioned MOCVD carrier disk drift compensation system establishes a two-dimensional spatial mapping model to characterize the correspondence between the spatial position of the carrier disk surface and the measured physical quantity. It also characterizes the drift of each revolution as a two-dimensional spatial translation parameter, transforming drift compensation from traditional time and angle axis phase alignment to sampling position correction in a spatial coordinate system. Therefore, this application can determine the difference in physical quantity caused by disk drift based on the spatial offset between the actual and ideal sampling trajectories, and compensate for the original measurement signal. This not only corrects phase misalignment between signals from different revolutions but also reduces amplitude errors and waveform distortion caused by sampling position offsets, improving the accuracy and reliability of in-situ measurement signals.

[0133] Furthermore, this application constructs a joint optimization objective function based on continuous multi-cycle raw measurement signals and jointly solves the two-dimensional spatial mapping model and the corresponding two-dimensional spatial translation parameters for each cycle. This ensures that the data for each cycle are constrained relative to the same spatial mapping model, avoiding the problem of error propagation and accumulation in traditional recursive compensation between adjacent cycles. In addition, by fusing multi-cycle measurement data in the spatial coordinate system, the influence of single-cycle noise, local anomalies, or signal fluctuations on the compensation results can be reduced, improving the stability of disk drift parameter identification and target measurement signal generation.

[0134] Furthermore, this application introduces a regularization term into the joint optimization objective function to constrain the differences between the two-dimensional spatial translation parameters of adjacent circles. This can take advantage of the mechanical continuity of the bearing disk drift to suppress parameter abrupt changes that do not conform to the actual mechanical motion law, making the obtained disk drift trajectory smoother and more reasonable, thereby improving the noise resistance and physical reliability of the compensation process.

[0135] Furthermore, the two-dimensional spatial translation parameters output by this application can characterize the actual disk drift offset in the first and second spatial directions, and have a clear physical meaning. The disk drift parameter trajectory formed based on the two-dimensional spatial translation parameters of each circle can be used for various applications such as mechanical condition monitoring, fault early warning, and process backtracking.

[0136] Furthermore, this application does not rely on whether the original measurement signal has obvious peak characteristics, but rather on compensation based on spatial mapping relationships. Therefore, it is applicable to various in-situ measurement signals such as temperature, reflectivity, film thickness, curvature, spectral intensity, ellipticity parameters, and X-ray diffraction intensity, exhibiting good versatility. The converged two-dimensional spatial mapping model can also serve as a spatial distribution characterization of the measured physical quantities on the surface of the support disk or substrate, used for subsequent process diagnosis, process optimization, and equipment status analysis, providing a data foundation for the accurate monitoring and stable control of the MOCVD epitaxial growth process. The modules in the aforementioned MOCVD disk drift compensation system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0137] In one feasible embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements the aforementioned MOCVD disk drift compensation method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0138] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0139] In one feasible embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method steps in the above-described method for compensating for disk drift of MOCVD carrier disk.

[0140] In one feasible embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When executed by a processor, the computer program implements the method steps in the above-described method for compensating for disk drift of an MOCVD carrier disk. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0141] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for compensating for disk drift in MOCVD, characterized in that, The method includes: The original measurement signals of the MOCVD carrier disk during continuous rotation are acquired; the original measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed. The two-dimensional spatial mapping model characterizes the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under disk-drift-free conditions. The two-dimensional spatial translation parameters characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space for each loop. Based on the alternating optimization strategy, the joint optimization objective function is iteratively optimized to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; Based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory is determined, and the original measurement signal is corrected based on the spatial offset compensation amount to obtain the target measurement signal.

2. The method according to claim 1, characterized in that, The joint optimization objective function includes: The data fidelity term is used to characterize the error between the original measurement signal and the corresponding model prediction signal for each lap; wherein, the model prediction signal is obtained by spatially correcting the ideal sampling trajectory based on the two-dimensional spatial translation parameters of the corresponding lap, obtaining the actual sampling trajectory of the corresponding lap, and then sampling along the actual sampling trajectory from the two-dimensional spatial mapping model; The regularization term is used to characterize the difference between the two-dimensional spatial translation parameters corresponding to adjacent cycles, and to suppress the difference during the iterative optimization process of the joint optimization objective function, so as to constrain the physical continuity of the two-dimensional spatial translation parameters of each cycle.

3. The method according to claim 2, characterized in that, Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed, including: Based on the data fidelity term and the regularization term, a joint optimization objective function is constructed, the expression of which is: in, The sequence number of the circle; This represents the total number of cycles of the continuously acquired raw measurement signal; The sampling position in each lap; For the first Circle at the sampling location The raw measurement signal collected at the location; The two-dimensional spatial mapping model to be optimized; Sampling position in the ideal sampling trajectory The corresponding ideal sampling position; For the first The two-dimensional spatial translation parameters corresponding to the circle, and Indicates the first The disk drift offset in the first spatial direction. Indicates the first The disk drift offset in the second spatial direction; To obtain from the two-dimensional spatial mapping model along the first... The model prediction signal obtained by sampling the actual sampling trajectory of the circle; The regularization coefficient is used. This is a regularization term.

4. The method according to claim 2, characterized in that: The regularization term includes at least one of first-order smoothing constraint, second-order smoothing constraint, and periodic consistency constraint. The first-order smoothing constraint is used to constrain the difference between the two-dimensional spatial translation parameters of adjacent cycles, the second-order smoothing constraint is used to constrain the difference between the changes in the two-dimensional spatial translation parameters of adjacent cycles, and the period consistency constraint is used to constrain the difference between the two-dimensional spatial translation parameters separated by a preset number of cycles.

5. The method according to claim 1, characterized in that, The method of iteratively optimizing the joint optimization objective function based on the alternating optimization strategy to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters includes: Initialize the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters; In each iteration, the two-dimensional spatial translation parameters corresponding to each current circle are fixed, and the two-dimensional spatial mapping model is updated based on the original measurement signal; The updated two-dimensional spatial mapping model is fixed, and the two-dimensional spatial translation parameters corresponding to each circle are updated based on the error between the original measurement signal and the corresponding model prediction signal; wherein, the model prediction signal is obtained by sampling from the updated two-dimensional spatial mapping model along the actual sampling position of the corresponding circle; Repeat the update process of the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters until the joint optimization objective function satisfies the preset convergence condition, and obtain the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters.

6. The method according to claim 5, characterized in that, The methods for initializing the two-dimensional spatial mapping model include: Based on the original measurement signal of the first loop, the initial values ​​of the two-dimensional spatial mapping model are generated according to the ideal sampling trajectory; Alternatively, the initial values ​​of the two-dimensional spatial mapping model are generated based on the average signal of the original measurement signals from the first cycle for a consecutive preset number of cycles, according to the ideal sampling trajectory; Alternatively, with the two-dimensional spatial translation parameters corresponding to each circle initialized to zero, the original measurement signals of multiple circles are mapped to the surface spatial coordinate system of the bearing disk according to the ideal sampling trajectory, and the measurement values ​​mapped to the same spatial neighborhood are fused to generate the initial value of the two-dimensional spatial mapping model.

7. The method according to claim 5, characterized in that, The step of fixing the two-dimensional spatial translation parameters corresponding to each current circle and updating the two-dimensional spatial mapping model based on the original measurement signal includes: Based on the two-dimensional spatial translation parameters corresponding to each circle, the sampling positions corresponding to the original measurement signals of each circle are corrected to the actual sampling positions; The measured values ​​of the original measurement signals of each circle are mapped to the surface space coordinate system corresponding to the actual sampling position; The two-dimensional spatial mapping model is updated based on the measured values ​​mapped to the surface spatial coordinate system.

8. The method according to claim 7, characterized in that, The step of updating the two-dimensional spatial mapping model based on the measured values ​​mapped to the surface spatial coordinate system includes: Multiple measurements mapped to the same spatial neighborhood are fused to obtain an updated two-dimensional spatial mapping model; wherein, the fusion process includes weighted averaging, kernel function weighting, inverse distance weighting, interpolation fitting, or regression fitting.

9. The method according to claim 5, characterized in that, The fixed and updated two-dimensional spatial mapping model, based on the error between the original measurement signal and the corresponding model prediction signal, updates the two-dimensional spatial translation parameters corresponding to each circle, including: For each original measurement signal, the ideal sampling trajectory is spatially corrected based on the candidate two-dimensional spatial translation parameters corresponding to that lap to obtain the candidate actual sampling trajectory; Sample along the candidate actual sampling trajectory from the updated two-dimensional spatial mapping model to obtain the corresponding model prediction signal; Based on the error between the original measurement signal and the corresponding model prediction signal, a two-dimensional spatial translation parameter is determined that minimizes the error.

10. The method according to claim 9, characterized in that, The determination of the two-dimensional spatial translation parameters that make the error satisfy the preset minimization condition includes: The error between the original measurement signal of the circle and the corresponding model prediction signal is constructed as a nonlinear least squares problem, and the two-dimensional spatial translation parameters corresponding to the circle are solved using a preset algorithm; wherein, the preset algorithm includes at least one of the following: Levenberg-Marquardt algorithm, Gauss-Newton algorithm, gradient descent algorithm, grid search algorithm and coarse search combined with fine search algorithm.

11. The method according to claim 5, characterized in that, The method further includes: During the iterative optimization of the joint optimization objective function, the most recent preset number of original measurement signals are selected from the continuously acquired original measurement signals as the current optimization window according to the preset window length; Upon acquiring a new set of original measurement signals, the current optimization window is updated, and the joint optimization objective function is iteratively optimized based on the original measurement signals within the updated current optimization window.

12. The method according to claim 1, characterized in that, When there are multiple in-situ measurement probes, each in-situ measurement probe is set at a different preset position; the ideal sampling trajectory includes multiple ideal sampling sub-trajectories corresponding to each of the in-situ measurement probes. Based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop, a joint optimization objective function is constructed, including: Based on the original measurement signals acquired by each of the in-situ measurement probes, the ideal sampling sub-trajectories corresponding to each of the in-situ measurement probes, the two-dimensional spatial translation parameters to be optimized, and the two-dimensional spatial mapping model to be optimized, the joint optimization objective function is constructed.

13. The method according to claim 1, characterized in that, Based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, the spatial offset compensation amount of each circle's actual sampling trajectory relative to the ideal sampling trajectory is determined, and the original measurement signal is corrected based on the spatial offset compensation amount to obtain the target measurement signal, including: Based on the converged two-dimensional spatial translation parameters, the actual sampling position corresponding to each sampling point in each loop is determined. Based on the converged two-dimensional spatial mapping model, the model values ​​corresponding to the actual sampling positions and the model values ​​corresponding to the ideal sampling positions in the ideal sampling trajectory are determined respectively. The spatial offset compensation amount is determined based on the difference between the model value corresponding to the actual sampling position and the model value corresponding to the ideal sampling position. The spatial offset compensation is superimposed on the corresponding original measurement signal to obtain the target measurement signal.

14. A disk drift compensation system for MOCVD carrier disks, characterized in that, The system includes: The acquisition module is used to acquire the raw measurement signals of the MOCVD carrier disk during multiple rotations; the raw measurement signals are acquired by an in-situ measurement probe set at a preset position in the reaction chamber. A construction module is used to construct a joint optimization objective function based on the original measurement signal, the two-dimensional spatial mapping model to be optimized, the ideal sampling trajectory, and the two-dimensional spatial translation parameters to be optimized for each loop. The two-dimensional spatial mapping model characterizes the correspondence between the surface spatial position of the carrier disk and the measured physical quantity corresponding to the original measurement signal. The ideal sampling trajectory is the spatial sampling trajectory of the in-situ measurement probe relative to the carrier disk under disk drift-free conditions. The two-dimensional spatial translation parameters characterize the disk drift offset of the carrier disk relative to the in-situ measurement probe in two-dimensional space for each loop. An optimization module is used to iteratively optimize the joint optimization objective function based on an alternating optimization strategy, so as to jointly solve the two-dimensional spatial mapping model and the two-dimensional spatial translation parameters. The compensation module is used to determine the spatial offset compensation amount of the actual sampling trajectory of each circle relative to the ideal sampling trajectory based on the converged two-dimensional spatial mapping model and the converged two-dimensional spatial translation parameters, and to correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

15. An in-situ measuring device, characterized in that, The device includes: At least one in-situ measurement probe is set at a preset position in the reaction chamber to perform in-situ measurements on the substrate surface as it rotates through the measurement area, so as to obtain the original measurement signal during the epitaxial film growth process on the substrate surface; the type of the measured physical quantity corresponding to the original measurement signal includes at least one of temperature, reflectivity, film thickness, curvature, spectral intensity, ellipticity parameter or X-ray diffraction intensity; The MOCVD carrier disk drift compensation system as described in claim 14 is used to generate a spatial offset compensation amount based on the original measurement signal, and to correct the original measurement signal based on the spatial offset compensation amount to obtain the target measurement signal.

16. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-13.

17. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-13.

Images