Methods, systems, equipment, and storage media for calculating the rotation parameters of semiconductor carrier disks.

Abstract

This application provides a method, system, device, and storage medium for calculating the rotation parameters of a semiconductor carrier disk, relating to the field of semiconductor testing technology. This application employs a single photoelectric sensor, avoiding high costs and maintenance burdens. The feature model does not rely on precise geometric coordinate drawings, and the matching results have a stronger tolerance to dimensional drift and slight deformation, adapting to various carrier disks and significantly improving versatility and long-term stability. The core calculations focus on feature extraction, multi-period statistics, and topology matching of one-dimensional time-domain sequences, enabling rapid output under low computing power conditions, meeting the response speed requirements of online monitoring and closed-loop control. Through a two-level strategy combining initial positioning and model fine matching, a globally consistent absolute phase relationship can be established relying solely on the disk features and their measured distribution, thereby stably obtaining the target rotation parameters and ensuring accurate and reliable measurement and positioning results even under complex conditions such as sparse feature points and non-full disk operation.

Description

Technical Field

[0001] This application relates to the field of semiconductor testing technology, and in particular to a method, system, device and storage medium for calculating the rotation parameters of a semiconductor carrier disk. Background Technology

[0002] In the semiconductor manufacturing field, metal-organic chemical vapor deposition (MOCVD) equipment is one of the key pieces of equipment for producing optoelectronic chips such as light-emitting diodes, lasers, and radio frequency devices. During MOCVD epitaxial growth, the high-speed rotation of the carrier disk is crucial for ensuring the uniformity of thin film growth. Therefore, real-time and accurate monitoring of the carrier disk's rotation speed and determining its absolute phase or absolute starting point are essential for process stability control, automated formulation execution, and equipment fault diagnosis.

[0003] In traditional MOCVD equipment, the monitoring of graphite disk rotation speed and phase typically relies on the synchronous trigger signal output by an optical encoder or Hall sensor mounted on the rotating shaft as a calculation reference. However, in cases of retrofitting older equipment, machine structure limitations, or sensor failure, this trigger signal may become unavailable. To address this, the industry has proposed non-contact software solutions that do not require external trigger signals. For example, multiple optical sensors are arranged around the reaction chamber to collect reflectivity signals and synthesize periodic images. These images are then combined with a pre-input absolute angular coordinate model of the disk for global matching to obtain the phase and rotation speed. While such solutions can achieve positioning without triggers, they generally suffer from problems such as multiple hardware channels, complex synchronization and installation calibration, strong reliance on accurate mechanical drawings, large computational load leading to limited real-time performance, and susceptibility to failure under conditions of incomplete disk operation or insufficient features, resulting in low detection accuracy. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, device and storage medium for calculating the rotation parameters of a semiconductor carrier disk, so as to overcome the shortcomings of traditional technology in scenarios where the trigger signal is missing, which requires complex hardware and accurate models, and is difficult to balance universality, real-time performance and non-full disk adaptability, resulting in low detection accuracy.

[0005] In a first aspect, this application proposes a method for calculating the rotation parameters of a semiconductor carrier disk, applicable to a semiconductor detection device including a photoelectric sensor, wherein the photoelectric sensor is disposed on the sidewall of a semiconductor reaction chamber, and the carrier disk is provided with multiple trays for placing wafer objects; the method includes:

[0006] Acquire the reflected signal output by the photoelectric sensor during the rotation of the carrier disk;

[0007] Feature extraction is performed on the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence;

[0008] Multi-period analysis was performed on the gap sequence to obtain preliminary rotation parameters;

[0009] Based on the preliminary rotation parameters, the time-series feature information is mapped to angular coordinates to construct a list of measured slide object positions.

[0010] The measured list of carrier object positions and the preset feature model are subjected to circumferential phase matching to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance.

[0011] The target rotation parameters are obtained by correcting the initial rotation parameters based on the optimal phase offset.

[0012] In one embodiment, the step of extracting features from the reflected signal to obtain temporal feature information includes:

[0013] The reflected signal is filtered to obtain a preprocessed signal;

[0014] The preprocessed signal is binarized, and edge detection is performed based on the processing result to obtain temporal feature information.

[0015] In one embodiment, the step of binarizing the preprocessed signal and performing edge detection based on the processing result to obtain temporal feature information includes:

[0016] The preprocessed signal is adaptively and dynamically thresholded based on a sliding window, and binarized using hysteresis comparison to obtain a binarized sequence.

[0017] Edge detection is performed on the binarized sequence to obtain the rising edge sequence representing the leading edge of the slide object and the falling edge sequence representing the trailing edge of the slide object;

[0018] For each wafer object, the difference in the number of points between its corresponding rising edge and falling edge is calculated as the pulse width of that wafer object;

[0019] For two adjacent slide objects, the difference in the number of points between the rising edge of the latter slide object and the falling edge of the former slide object is calculated as the gap to form an alternating pulse width sequence and gap sequence.

[0020] In one embodiment, the step of adaptively calculating a dynamic threshold for the preprocessed signal based on a sliding window and binarizing the preprocessed signal using hysteresis comparison to obtain a binarized sequence includes:

[0021] For each sampling time, a window signal is extracted from the preprocessed signal by a sliding window with a preset number of points before and after that sampling time;

[0022] The window signals are clustered to obtain a first cluster corresponding to the high-level state and a second cluster corresponding to the low-level state;

[0023] A dynamic threshold is determined based on the centers of the first cluster and the centers of the second cluster;

[0024] The upper hysteresis threshold and the lower hysteresis threshold are determined based on the dynamic threshold.

[0025] When the preprocessed signal crosses the upper hysteresis threshold from bottom to top, the binarized sequence is set to the first state; when the preprocessed signal crosses the lower hysteresis threshold from top to bottom, the binarized sequence is set to the second state to output a stable binarized sequence.

[0026] In one embodiment, the step of performing edge detection on the binarized sequence to obtain a rising edge sequence representing the leading edge of the slide object and a falling edge sequence representing the trailing edge of the slide object includes:

[0027] In the binarized sequence, the binary states of adjacent sampling points are traversed;

[0028] When an adjacent sampling point is detected to transition from the second state to the first state, the index of that sampling point is recorded as the rising edge index, and the rising edge indices are arranged in chronological order to form the rising edge sequence.

[0029] When an adjacent sampling point is detected to transition from the first state to the second state, the index of that sampling point is recorded as the falling edge index, and the falling edge indices are arranged in chronological order to form the falling edge sequence.

[0030] In one embodiment, the initial rotation parameters include an initial absolute starting point and an initial rotational speed;

[0031] The process of performing multi-period analysis on the gap sequence to obtain preliminary rotation parameters includes:

[0032] The gap sequence is aligned in multiple loops and a statistical matrix is ​​constructed based on the data of each loop after alignment. The characteristic gaps of the bearing disk are identified based on the statistical matrix.

[0033] The timing position corresponding to the end time of the aforementioned feature gap is taken as the initial absolute starting point;

[0034] The average rotation period is calculated based on the aligned data of each revolution, and the initial rotation speed is calculated based on the average rotation period.

[0035] In one embodiment, the step of aligning the gap sequence through multiple loops and constructing a statistical matrix based on the aligned loop data, and identifying the characteristic gaps of the bearing disk based on the statistical matrix, includes:

[0036] Based on the estimated rotational speed of the bearing plate, the gap sequence is segmented according to the theoretical rotational period to obtain a gap subsequence corresponding to multiple revolutions;

[0037] Perform initial phase alignment on each of the aforementioned gap subsequences so that gaps with the same sequence number in different rings correspond to the same physical position on the bearing disk;

[0038] Arrange the aligned inter-loop subsequences by loop number as rows and by circumferential position number as columns to construct a statistical matrix;

[0039] For each column of the statistical matrix, calculate the median and interquartile range of the data in that column;

[0040] Based on the median and interquartile range of each column, the target column that meets the preset conditions is determined, and the circumferential position of the target column is determined as the location of the characteristic gap of the bearing plate.

[0041] The preset conditions are determined by the following method: if the median of a column is greater than the first threshold and the interquartile range is less than the second threshold, then the preset conditions are considered to be met.

[0042] In one embodiment, the step of mapping the temporal feature information to angular coordinates based on the preliminary rotation parameters to construct a list of measured slide object positions includes:

[0043] The rotation period is determined based on the initial rotation speed;

[0044] Using the time position corresponding to the initial absolute starting point as the zero point of angle, the sampling time sequence is converted into an angle coordinate system based on the rotation period;

[0045] For each slice object in the pulse width sequence and each gap in the gap sequence, they are converted into angle information in the angle coordinate system; wherein, the timing feature information includes the start timing position and the end timing position, and the angle information includes the start angle and the end angle;

[0046] The angle information corresponding to each slide object is used as a position entry and written into the measured slide object position list that characterizes the circumferential distribution of the bearing disk.

[0047] In one embodiment, the feature model includes the number of model trays and the relative angle tolerance of each interval;

[0048] The step of performing circumferential phase matching between the measured slide object position list and the preset feature model to determine the optimal phase offset includes:

[0049] The number of currently detected wafer objects is determined based on the measured wafer object location list;

[0050] Based on the difference between the number of the slide objects and the number of the model trays, a matching strategy is determined; based on the matching strategy, the matching degree between the measured slide object position list and the feature model is calculated during the circumferential phase search process, and the phase offset corresponding to the optimal matching degree is determined as the optimal phase offset.

[0051] In one embodiment, if the quantity difference is less than or equal to a third threshold, the full-board condition is considered met, and the matching strategy includes:

[0052] Align the list of measured slide object positions with the feature model under multiple candidate phase offsets;

[0053] For each candidate phase offset, the number of wafer objects falling within the relative angle tolerance of the tray corresponding to the feature model is counted.

[0054] The candidate phase offset that maximizes the number of the wafer objects is taken as the optimal phase offset.

[0055] In one embodiment, if the quantity difference is greater than a third threshold, the non-full condition is considered met, and the matching strategy includes:

[0056] Align the list of measured slide object positions with the feature model under multiple candidate phase offsets;

[0057] For any candidate phase offset, if all the wafer objects can be mapped to the relative angle tolerance of the corresponding tray in the feature model, the type of the occupied tray is compatible with the feature of the measured wafer object, and the circumferential interval corresponding to the unoccupied tray corresponds to the gap interval in the position list of the measured wafer object, then the candidate phase offset is determined to be a valid candidate phase.

[0058] When multiple valid candidate phases exist, the valid candidate phase with the smallest matching error is selected as the optimal phase offset.

[0059] In one embodiment, the target rotation parameters include the target absolute starting point and the target rotation speed;

[0060] The process of correcting the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters includes:

[0061] Based on the optimal phase offset, the preliminary absolute starting point is corrected to obtain the target absolute starting point;

[0062] The timing reference point is the time position corresponding to the absolute starting point of the target.

[0063] After the timing reference point, obtain the timing positions of the wafer objects corresponding to at least two reference trays with known circumferential angular intervals in the feature model;

[0064] The angular velocity of the carrier disk is determined based on the circumferential angular interval of the reference tray and the time difference between the timing positions of the corresponding carrier objects on the reference tray, and the target rotational speed is calculated based on the angular velocity.

[0065] Secondly, this application proposes a system for calculating the rotation parameters of a semiconductor carrier disk, which is installed within a semiconductor detection device. The semiconductor detection device includes a photoelectric sensor disposed on the sidewall of a semiconductor reaction chamber, and the carrier disk has multiple trays for placing wafer objects. The system includes:

[0066] The acquisition module is used to acquire the reflected signal output by the photoelectric sensor during the rotation of the carrier disk;

[0067] An extraction module is used to extract features from the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence;

[0068] The analysis module is used to perform multi-period analysis on the gap sequence to obtain preliminary rotation parameters;

[0069] The list construction module is used to map the temporal feature information to angular coordinates based on the preliminary rotation parameters, thereby constructing a list of measured slide object positions;

[0070] The matching module is used to perform circumferential phase matching between the measured list of carrier object positions and the preset feature model to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance.

[0071] The calculation module is used to correct the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters.

[0072] Thirdly, 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 in the first aspect.

[0073] Fourthly, 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.

[0074] The above-mentioned method, system, equipment, and storage medium for calculating the rotation parameters of the semiconductor carrier disk have at least the following advantages:

[0075] This application achieves a significant simplification of hardware configuration by using a single photoelectric sensor on the sidewall of the semiconductor reaction cavity to acquire the reflected signal during the rotation of the carrier disk. This avoids the high cost and maintenance burden associated with complex multi-channel acquisition and calibration. By extracting features from the reflected signal, calculable temporal feature information is obtained, including pulse width and gap sequences. Multi-cycle analysis of the gap sequence yields more stable preliminary rotation parameters. The temporal feature information is mapped to angular coordinates to construct a list of measured carrier object positions, providing a directly comparable benchmark for subsequent positioning. Finally, the list of measured carrier object positions is matched with the feature model to obtain the optimal phase offset. Based on the optimal phase offset, the preliminary rotation parameters are corrected to obtain the target rotation parameters. The feature model of this application does not rely on precise geometric coordinate drawings, making the algorithm more tolerant to manufacturing errors of the carrier plate, dimensional drift caused by thermal cycling, and slight deformation. It can be adapted to different models, batches, and even slightly deformed carrier plates, significantly improving versatility and long-term stability. At the same time, the core calculation of this application focuses on feature extraction of one-dimensional time-domain sequences, multi-period statistics, and topology matching. Compared with existing image synthesis and global graphic matching schemes, the calculation link is shorter, the implementation is simpler, and the real-time performance is stronger. It can achieve fast output under low computing power conditions, meeting the response speed requirements of online monitoring and closed-loop control. In addition, this application adopts a two-level strategy of initial positioning based on the measured carrier object position list combined with fine matching of the feature model. Without requiring the detection of all tray states, it can establish a globally consistent absolute phase relationship based only on the detected tray features and their measured distribution, thereby stably obtaining the target rotation parameters. This ensures that accurate and reliable measurement and positioning results can still be output under complex working conditions such as sparse feature points and non-full tray. Attached Figure Description

[0076] Figure 1 This is a flowchart illustrating a method for calculating the rotation parameters of a semiconductor carrier disk in one embodiment;

[0077] Figure 2 This is a flowchart illustrating the steps for extracting temporal feature information in one embodiment;

[0078] Figure 3 This is a flowchart illustrating the steps for obtaining preliminary rotation parameters in one embodiment;

[0079] Figure 4 This is a flowchart illustrating the steps of constructing a list of measured slide locations in one embodiment;

[0080] Figure 5 This is a flowchart illustrating the step of determining the optimal phase offset in one embodiment;

[0081] Figure 6This is a flowchart illustrating the steps for obtaining the target rotation parameters in one embodiment;

[0082] Figure 7 This is a block diagram of a system for calculating the rotation parameters of a semiconductor carrier disk in one embodiment;

[0083] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0084] 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.

[0085] 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.

[0086] Please see Figure 1 In one exemplary embodiment, this application provides a method for calculating the rotation parameters of a semiconductor carrier disk. This method is applicable to a semiconductor detection device, which includes a photoelectric sensor disposed on the sidewall of a semiconductor reaction chamber. A carrier disk made of high-purity graphite is disposed within the semiconductor reaction chamber, and the carrier disk has multiple trays for placing wafer objects. Supported by a rotating shaft driven by a motor, the carrier disk can rotate continuously during semiconductor epitaxial growth.

[0087] The calculation method for the aforementioned semiconductor carrier disk rotation parameters specifically includes the following steps:

[0088] Step 102: Obtain the reflected signal output by the photoelectric sensor during the rotation of the carrier disk.

[0089] Specifically, this application fixes a photoelectric sensor to the sidewall of the semiconductor reaction cavity. For example, the photoelectric sensor can be a laser reflection type, and its optical path is aligned with the edge of the carrier disk or a specific measurement track. The position of the photoelectric sensor spot is adjusted so that it periodically sweeps across the tray area and the gap area as the carrier disk rotates. When the measurement spot irradiates the tray area where the substrate is placed, due to the different surface reflectivity and geometry, the echo intensity received by the photoelectric sensor is distinguishable from that when irradiating the gap area, thereby generating a voltage signal at the output of the photoelectric sensor that varies with the rotation period. Further, after amplifying the voltage signal, analog-to-digital conversion is performed at a preset sampling frequency and continuous sampling is performed to obtain a sampling sequence arranged in chronological order as the reflected signal.

[0090] Step 104: Extract features from the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence.

[0091] Specifically, the purpose of feature extraction is to convert time-varying reflection signals into structured temporal features that can directly characterize the passing of the slide object and vacancy through the probe point. This results in alternating...

[0092] Pulse width refers to the duration of a signal segment corresponding to a single wafer object on the time axis; that is, the continuous duration during which the reflected signal remains at the corresponding level when the probe spot sweeps across the tray area where the wafer object is located. Typically, this duration can be represented by the number of sampling points or the time length; exemplarily, this embodiment uses the number of sampling points. Correspondingly, the pulse width sequence is a sequence obtained by arranging multiple pulse widths formed when multiple wafer objects sequentially pass through the probe points in chronological order.

[0093] The gap refers to the duration of the signal segment corresponding to the empty space between two adjacent wafer objects on the time axis, that is, the time interval between the end of the signal segment corresponding to the previous wafer object and the beginning of the signal segment corresponding to the next wafer object; similarly, the duration of the gap can also be expressed by the number of sampling points or the time length. The gap sequence is a sequence obtained by arranging multiple gaps between adjacent wafer objects within one loop in chronological order.

[0094] Step 106: Perform multi-period analysis on the gap sequence to obtain preliminary rotation parameters.

[0095] Specifically, since the gap sequence is closer to the inherent geometric characteristics of the bearing disk than the pulse width sequence, this embodiment uses the gap sequence for analysis. Furthermore, the preliminary rotation parameters include a preliminary absolute starting point and a preliminary rotational speed. By calculating multiple rotations, the recurring structural features in each rotation can be identified. The temporal position corresponding to this structural feature is used as the preliminary absolute starting point, ensuring a unified zero-degree reference for subsequent angle mapping.

[0096] Furthermore, by taking the time interval between two consecutive occurrences of the initial absolute starting point as an estimate of the rotation period, the initial rotational speed can be calculated.

[0097] Step 108: Based on the initial rotation parameters, the temporal feature information is mapped to angular coordinates to construct a list of measured slide object positions.

[0098] Specifically, angular coordinates refer to the way angles are used to represent the circumferential position of the bearing disk surface. In this embodiment, a preliminary absolute starting point is used as the zero-point reference for angles, and the rotation period corresponding to the preliminary rotation speed is used as the conversion scale from time to angle, so that the various characteristic positions of the reflected signal on the time axis are converted into angular positions on the circumference of the bearing disk.

[0099] The list of measured slide object positions is a structured representation of the time-to-angle mapping mentioned above. It can be a list arranged in circumferential order, and each record in the list describes the measured position information of a detected slide object on the carrier disk.

[0100] Step 110: Perform circumferential phase matching between the measured list of carrier object positions and the preset feature model to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance.

[0101] Specifically, the feature model in this embodiment is pre-read from the equipment control system or configuration file. Unlike traditional precise mechanical drawings, this feature model is not a precise geometric drawing, but rather a set describing the layout topology and relative angular tolerances of each pallet on the carrier plate in the circumferential direction. For example, the feature model is organized into multiple model entries, each corresponding to one pallet, and displayed as [P1, Angle θ0, Type A], [P2, Angle θ1, Type A], ... [Pn, Angle θn-1, Type B]. Type A and Type B are type information, which can be used to distinguish normal pallets, pallets located after feature gaps, and / or specially marked pallets for alignment, etc. Furthermore, the feature model also includes the allowable relative angular tolerances for each interval, used to describe the acceptable deviation range under manufacturing errors, assembly deviations, or thermal cycling deformation.

[0102] When performing circumferential phase matching, the list of measured object positions is used as the measured input, and a phase offset is made in the circumferential direction. Then, it is compared with the feature model until a phase offset with the best matching degree is found, which is taken as the optimal phase offset.

[0103] Step 112: Correct the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters.

[0104] Specifically, the target rotation parameters include the target absolute starting point and the target rotational speed. The optimal phase offset is used as a compensation amount to compensate for the initial absolute starting point, thus obtaining the target absolute starting point. Then, the time interval between two consecutive occurrences of the target absolute starting point is taken as the target rotational period, from which the target rotational speed can be calculated.

[0105] The aforementioned method for calculating the rotation parameters of the semiconductor carrier disk simplifies the hardware configuration by using a single photoelectric sensor on the sidewall of the semiconductor reaction cavity to acquire the reflected signal during the disk's rotation. This avoids the high costs and maintenance burden associated with complex multi-channel acquisition and calibration. Feature extraction of the reflected signal yields calculable temporal feature information, including pulse width and gap sequences. Multi-cycle analysis of the gap sequence provides more stable preliminary rotation parameters. Mapping the temporal feature information to angular coordinates constructs a list of measured carrier object positions, providing a directly comparable benchmark for subsequent positioning. Finally, matching the measured carrier object position list with the feature model yields the optimal phase offset. The preliminary rotation parameters are then corrected based on this optimal phase offset to obtain the target rotation parameters. The feature model of this application does not rely on precise geometric coordinate drawings, making the algorithm more tolerant to manufacturing errors of the carrier plate, dimensional drift caused by thermal cycling, and slight deformation. It can be adapted to different models, batches, and even slightly deformed carrier plates, significantly improving versatility and long-term stability. At the same time, the core calculation of this application focuses on feature extraction of one-dimensional time-domain sequences, multi-period statistics, and topology matching. Compared with existing image synthesis and global graphic matching schemes, the calculation link is shorter, the implementation is simpler, and the real-time performance is stronger. It can achieve fast output under low computing power conditions, meeting the response speed requirements of online monitoring and closed-loop control. In addition, this application adopts a two-level strategy of initial positioning based on the measured carrier object position list combined with fine matching of the feature model. Without requiring the detection of all tray states, it can establish a globally consistent absolute phase relationship based only on the detected tray features and their measured distribution, thereby stably obtaining the target rotation parameters. This ensures that accurate and reliable measurement and positioning results can still be output under complex working conditions such as sparse feature points and non-full tray.

[0106] Please see Figure 2 Optionally, feature extraction is performed on the reflected signal to obtain temporal feature information, including:

[0107] Step 202: Filter the reflected signal to obtain a preprocessed signal.

[0108] Step 204: Binarize the preprocessed signal and perform edge detection based on the processing result to obtain temporal feature information.

[0109] Specifically, the original reflected signal is bandpass filtered to suppress power frequency interference and high-frequency noise, resulting in a preprocessed signal.

[0110] Furthermore, binarizing the preprocessed signal can convert the continuous analog signal into discrete tray and gap states.

[0111] Optionally, the preprocessed signal is binarized, and edge detection is performed based on the processing result to obtain temporal feature information, including:

[0112] Adaptive dynamic threshold calculation is performed on the preprocessed signal based on a sliding window, and the preprocessed signal is binarized using hysteresis comparison to obtain a binarized sequence.

[0113] Edge detection is performed on the binarized sequence to obtain the rising edge sequence representing the leading edge of the slide object and the falling edge sequence representing the trailing edge of the slide object.

[0114] For each slice object, the difference in the number of points between its corresponding rising edge and falling edge is calculated as the pulse width of that slice object; for two adjacent slice objects, the difference in the number of points between the rising edge of the latter slice object and the falling edge of the former slice object is calculated as the gap, so as to form an alternating pulse width sequence and gap sequence.

[0115] Optionally, an adaptive dynamic threshold is calculated on the preprocessed signal based on a sliding window, and the preprocessed signal is binarized using hysteresis comparison to obtain a binarized sequence, including:

[0116] For each sampling time, a window signal is extracted from the preprocessed signal by a sliding window of a preset number of points before and after the sampling time; the window signal is clustered to obtain a first cluster corresponding to the high-level state and a second cluster corresponding to the low-level state; the dynamic threshold is determined based on the center of the first cluster and the center of the second cluster.

[0117] The upper and lower hysteresis thresholds are determined based on dynamic thresholds. When the preprocessed signal crosses the upper hysteresis threshold from bottom to top, the binarized sequence is set to the first state; when the preprocessed signal crosses the lower hysteresis threshold from top to bottom, the binarized sequence is set to the second state, so as to output a stable binarized sequence.

[0118] Specifically, this embodiment employs a sliding window method. For the nth sampling point, a window of data V_window is taken from W points before and after it. For example, W is set to 2000, corresponding to 0.2s. Further, a K-Means clustering algorithm (k=2) is used to quickly cluster the data within the window into a first cluster corresponding to a high-level state and a second cluster corresponding to a low-level state. The center C_high of the first cluster and the center C_low of the second cluster are calculated, and a dynamic threshold is determined based on this. For example, this embodiment sets the dynamic threshold to the median value between the two cluster centers, expressed as: Thr(n) = (C_high + C_low) / 2. Using the above scheme, it exhibits excellent adaptability to slow changes in signal amplitude and bimodal distributions.

[0119] Furthermore, to avoid noise causing oscillations near the threshold, this embodiment of the application also employs a hysteresis comparator for hysteresis comparison. For example, a small positive number δ is set, such as 5% of the threshold range, and the upper hysteresis threshold Thr_high(n) = Thr(n) + δ and the lower hysteresis threshold Thr_low(n) = Thr(n) - δ are determined based on δ. When the preprocessed signal V(n) crosses Thr_high(n) from bottom to top, the binarized sequence is set to the first state; for example, B(n) = 1 at this time. When V(n) crosses Thr_low(n) from top to bottom, the binarized sequence is set to the second state; for example, B(n) = 0 at this time, thus obtaining a stable binarized sequence B(n).

[0120] Optionally, edge detection is performed on the binarized sequence to obtain a rising edge sequence representing the leading edge of the slide object and a falling edge sequence representing the trailing edge of the slide object, including:

[0121] In the binarized sequence, the binary states of adjacent sampling points are traversed.

[0122] When an adjacent sampling point is detected to transition from the second state to the first state, the index of that sampling point is recorded as the rising edge index, and the rising edge indices are arranged into a rising edge sequence in chronological order.

[0123] When an adjacent sampling point is detected to transition from the first state to the second state, the index of that sampling point is recorded as the falling edge index, and the falling edge indices are arranged into a falling edge sequence in chronological order.

[0124] Specifically, the binarized sequence B(n) is traversed. If a transition is detected between two adjacent sampling points, the index of that sampling point is recorded. For example, a rising edge transition is a transition of B(n) from 0 to 1, i.e., from the second state to the first state, representing the leading edge of the substrate; a falling edge transition is a transition of B(n) from 1 to 0, i.e., from the first state to the second state, representing the trailing edge of the substrate. All transition actions are recorded and arranged in chronological order to obtain the rising edge sequence Rise_Index = [r1, r2, ...] and the falling edge sequence Fall_Index = [f1, f2, ...].

[0125] Furthermore, the pulse width W_i = f_i - r_i for each carrier object is calculated, and the width G_i = r_{i+1} - f_i for each gap is calculated, thereby obtaining the alternating pulse width sequence and gap sequence.

[0126] Please see Figure 3 Optionally, multi-period analysis is performed on the gap sequence to obtain preliminary rotation parameters, including:

[0127] Step 302: Align the gap sequence with multiple loops and construct a statistical matrix based on the aligned data of each loop. Identify the characteristic gaps of the bearing disk based on the statistical matrix and use the time sequence position corresponding to the end time of the characteristic gap as the initial absolute starting point.

[0128] Step 304: Calculate the average rotation period based on the aligned data of each revolution, and calculate the initial rotation speed based on the average rotation period.

[0129] Optionally, the gap sequence is aligned across multiple rings, and a statistical matrix is ​​constructed based on the aligned ring data. The characteristic gaps of the bearing plate are then identified based on the statistical matrix, including:

[0130] Based on the estimated rotational speed of the bearing plate, the gap sequence is segmented according to the theoretical rotational period to obtain the corresponding gap subsequence for multiple revolutions;

[0131] Perform initial phase alignment on each gap subsequence so that gaps with the same sequence number in different rings correspond to the same physical position on the bearing disk;

[0132] Arrange the aligned inter-loop subsequences by loop number as rows and by circumferential position number as columns to construct a statistical matrix;

[0133] For each column of the statistical matrix, calculate the median and interquartile range of the data in that column;

[0134] Based on the median and interquartile range of each column, the target column that meets the preset conditions is determined, and the circumferential position of the target column is determined as the location of the characteristic gap of the bearing plate.

[0135] The preset conditions are judged as follows: if the median of a column is greater than the first threshold and the interquartile range is less than the second threshold, then the preset conditions are considered to be met.

[0136] Specifically, the gap sequence is first segmented and aligned. Based on the estimated rotational speed of the bearing disk, the long sequence of gap data G_i is segmented according to the theoretical rotation period, forming multi-cycle data. Through sequence cross-correlation analysis or a gap pattern algorithm based on finding similar gap pairs, the starting phase of each cycle of data is aligned to ensure that gaps with the same sequence number in each cycle correspond to the same physical position.

[0137] Furthermore, a statistical matrix is ​​constructed. The aligned N-lap gap data are arranged into a statistical matrix M_stat with N rows and P columns, where P is the theoretical maximum number of gaps per lap.

[0138] Furthermore, robust statistics are performed on the statistical matrix. The median Med(j) and interquartile range IQR(j) (j=1,2,...,P) of each column in the statistical matrix M_stat are calculated.

[0139] Finally, the characteristic gaps in the statistical matrix are identified. The statistical matrix is ​​traversed to find the column index j_feature that satisfies Med(j) > mean(Med) + α × std(Med) and IQR(j) < β × mean(IQR). The circumferential position corresponding to this column index is determined as the location of the characteristic gap on the bearing plate. Here, α and β are empirical coefficients, exemplarily taken as α = 1.5 and β = 1.2; mean(Med) is the average of the medians of each column in the statistical matrix; std(Med) is the standard deviation of the medians of each column; and mean(IQR) is the average of the interquartile ranges of each column. If the above preset conditions are met, it means that the gap at this position is significantly larger and more stable, and can be determined as the inherent maximum gap on the bearing plate.

[0140] The physical gap corresponding to the identified column index j_feature is used as the feature gap. The end time of this feature gap in any revolution, i.e., the rising edge of the next wafer object, is defined as the initial absolute starting point S_initial of the rotation. Based on the aligned data of each revolution, the average rotation period T_avg is calculated, and thus the initial rotational speed RPM_initial = 60 / T_avg is obtained.

[0141] Please see Figure 4 Optionally, based on preliminary rotation parameters, the temporal feature information is mapped to angular coordinates to construct a list of measured slide object locations, including:

[0142] Step 402: Determine the rotation period based on the initial rotation speed.

[0143] Step 404: Using the timing position corresponding to the initial absolute starting point as the zero point of the angle, the sampling timing is converted into an angle coordinate system based on the rotation period.

[0144] Step 406: For each slice object in the pulse width sequence and each gap in the gap sequence, convert them into angle information in the angle coordinate system; wherein, the timing feature information includes the start timing position and the end timing position, and the angle information includes the start angle and the end angle.

[0145] Step 408: The angle information corresponding to each slide object is used as a position entry and written into the measured slide object position list that characterizes the circumferential distribution of the bearing disk.

[0146] Specifically, based on the initial absolute starting point S_initial and the initial rotational speed RPM_initial, the pulse width sequence and gap sequence are mapped to angular coordinates from 0 to 360 degrees, forming a list of measured slide object positions. This list includes the starting and ending angles of each detected slide object, as well as the starting and ending angles of each gap. It should be noted that this list of measured slide object positions is a measured position map.

[0147] Please see Figure 5 Optionally, the measured list of object positions on the slide and a preset feature model are subjected to circular phase matching to determine the optimal phase offset, including:

[0148] Step 502: Determine the number of currently detected wafer objects based on the measured wafer object location list.

[0149] Step 504: Determine the matching strategy based on the difference between the number of wafer objects and the number of model trays; based on the matching strategy, calculate the matching degree between the measured wafer object position list and the feature model during the circumferential phase search process, and determine the phase offset corresponding to the best matching degree as the optimal phase offset.

[0150] Specifically, the actual list of locations of the wafer objects may not be full, and there may be insufficient feature points, resulting in matching failures. Based on this, this application implements different matching strategies for different numbers of wafer objects to automatically find the most reasonable global layout assumptions.

[0151] Optionally, if the quantity difference is less than or equal to the third threshold, the full-board condition is considered met, and the matching strategy includes:

[0152] Align the list of measured slide object locations with the feature model under multiple candidate phase offsets;

[0153] For each candidate phase offset, count the number of wafer objects that fall within the relative angle tolerance of the tray corresponding to the feature model;

[0154] The candidate phase offset that maximizes the number of wafer objects is taken as the optimal phase offset.

[0155] Specifically, in a full-disc scenario, the number of measured slide objects and the number of model trays are close. By rotating and aligning the measured position map with the feature model on the circumference, the number of matches where each measured slide object falls into the model tray within the relative angular tolerance under different phase offsets φ is calculated. The optimal φ_optimal that maximizes the number of matches is found and taken as the optimal phase offset, and the overall matching error is calculated.

[0156] Optionally, if the difference in quantity is greater than the third threshold, the non-full board condition is considered met, and the matching strategy in this case includes:

[0157] Align the list of measured slide object locations with the feature model under multiple candidate phase offsets;

[0158] For any candidate phase offset, if all the slide objects can be mapped to the relative angle tolerance of the corresponding tray in the feature model, the type of the occupied tray is compatible with the feature of the measured slide object, and the circumferential interval of the unoccupied tray corresponds to the gap interval in the position list of the measured slide object, then the candidate phase offset is determined to be a valid candidate phase.

[0159] When multiple valid candidate phases exist, the valid candidate phase with the smallest matching error is selected as the optimal phase offset.

[0160] Specifically, when the disk is not full, the number of measured tile objects is less than the number of model trays, requiring sparse matching. This involves searching for an optimal phase φ_optimal, such that under this optimal phase, all measured tile objects uniquely fall within the relative angular tolerance of a certain model, and the tray types are compatible with the characteristics of the measured tile objects. Simultaneously, the positions of unoccupied model trays should correspond to the measured gap areas. Here, compatibility between tray type and measured tile object characteristics means that the tray and the tile objects placed on it match; for example, the relative width of the tile objects corresponds to the tray. Furthermore, based on the above matching results, even with sparse measured data, a precise mapping relationship between the measured position map and the disk feature model is obtained, enabling further inference of the actual state of all trays and thus completing the complete disk state map.

[0161] By employing the above solution, even during the process development phase where only a few wafers are placed, or during maintenance cycles when some trays are empty, this application can still accurately deduce the rotation start point, identify the absolute number of the placed wafers, and report the current total number of wafers placed. This greatly expands the applicable scenarios of the technology, achieving a leap from dedicated use for full trays to universal applicability across all operating conditions.

[0162] Furthermore, this application utilizes multi-ring interval statistics to automatically identify the inherent physical characteristics of the pallet. The feature model only describes the topological relationships and relative angular tolerances between pallets, rather than absolute coordinate drawings. This results in the matching results of this application having extremely high tolerance to pallet manufacturing errors, deformation caused by long-term thermal cycling, and non-standard designs. As long as the basic topological characteristics of the pallet remain unchanged, it can work stably. This means that the same algorithm can be applied to different batches, different manufacturers, and even slightly deformed graphite pallets, achieving high versatility.

[0163] Furthermore, the core of this application lies in the edge detection, interval statistics, and fast topology matching search of one-dimensional time-domain signals. The algorithm time is greatly reduced, making high-speed, real-time closed-loop control possible. It can provide speed feedback and status monitoring with an update rate of up to milliseconds, and can respond to rotational anomalies in a timely manner, meeting the stringent real-time requirements of advanced MOCVD processes for process control.

[0164] Please see Figure 6 Optionally, the target rotation parameters are obtained by correcting the initial rotation parameters based on the optimal phase offset, including:

[0165] Step 602: Correct the initial absolute starting point based on the optimal phase offset to obtain the target absolute starting point.

[0166] Step 604: Use the timing position corresponding to the absolute starting point of the target as the timing reference point.

[0167] Step 606: After the timing reference point, obtain the timing positions of the wafer objects corresponding to at least two reference trays with known circumferential angular intervals in the feature model.

[0168] Step 608: Determine the angular velocity of the carrier disk based on the circumferential angle interval of the reference tray and the time difference between the timing positions of the corresponding carrier objects on the reference tray, and calculate the target rotational speed based on the angular velocity.

[0169] Specifically, the initial absolute starting point is corrected based on the optimal phase offset to obtain the target absolute starting point, i.e., S_final = S_initial + φ_optimal. Then, using the known absolute angles of multiple trays after matching, the target rotational speed RPM_final is calculated with higher precision through more accurate timing differences.

[0170] Optionally, the method for calculating the rotation parameters of the semiconductor carrier disk in this application further includes: assigning a corresponding tray number to each detected wafer object and outputting a chart indicating the status of each tray. The tray status includes whether it contains wafers or is empty.

[0171] Optionally, the method for calculating the rotation parameters of the semiconductor carrier disk in this application further includes: calculating the confidence level of the current positioning result based on measurement data, wherein the measurement data includes the number of matches, matching error, and signal quality. Further, the high-confidence result is used to update the parameters of the dynamic threshold algorithm or enrich the details of the large disk feature model to achieve algorithm self-optimization. The high-confidence result includes a determined starting point and a wafer width pattern. Further, if the confidence level is too low, an alarm message is output, and the current calculation result is discarded.

[0172] 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.

[0173] Based on the same inventive concept, this application also provides a system for calculating the rotation parameters of a semiconductor carrier disk. This system is applicable to the above-mentioned method for calculating the rotation parameters of a semiconductor carrier disk. The solution provided by this system is similar to the solution described in the above-mentioned method. Therefore, the specific limitations in one or more device embodiments provided below can be found in the limitations of the method above, and will not be repeated here.

[0174] Please see Figure 7 In one embodiment, this application provides a system for calculating the rotation parameters of a semiconductor carrier disk. The system is disposed within a semiconductor detection device, which includes a photoelectric sensor disposed on the sidewall of a semiconductor reaction chamber. The carrier disk is provided with multiple trays for placing wafer objects.

[0175] The aforementioned semiconductor carrier disk rotation parameter calculation system includes: an acquisition module, an extraction module, an analysis module, a list construction module, a matching module, and a calculation module.

[0176] The acquisition module is used to acquire the reflected signal output by the photoelectric sensor during the rotation of the carrier disk.

[0177] The extraction module is used to extract features from the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence.

[0178] The analysis module is used to perform multi-period analysis on the gap sequence to obtain preliminary rotation parameters.

[0179] The list building module is used to map temporal feature information to angular coordinates based on preliminary rotation parameters, thereby constructing a list of measured slide object positions.

[0180] The matching module is used to perform circumferential phase matching between the measured list of carrier object positions and the preset feature model to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance.

[0181] The calculation module is used to correct the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters.

[0182] Optionally, the extraction module performs feature extraction on the reflected signal to obtain temporal feature information, including: filtering the reflected signal to obtain a preprocessed signal; binarizing the preprocessed signal and performing edge detection based on the processing result to obtain temporal feature information.

[0183] Optionally, the extraction module binarizes the preprocessed signal and performs edge detection based on the processing result to obtain temporal feature information, including: performing adaptive dynamic threshold calculation on the preprocessed signal based on a sliding window, and binarizing the preprocessed signal using hysteresis comparison to obtain a binarized sequence; performing edge detection on the binarized sequence to obtain a rising edge sequence representing the leading edge of the slide object and a falling edge sequence representing the trailing edge of the slide object; for each slide object, calculating the difference in the number of points between its corresponding rising edge and falling edge as the pulse width of the slide object; for two adjacent slide objects, calculating the difference in the number of points between the rising edge of the latter slide object and the falling edge of the former slide object as the gap, so as to form an alternating pulse width sequence and gap sequence.

[0184] Optionally, the extraction module performs adaptive dynamic threshold calculation on the preprocessed signal based on a sliding window and binarizes the preprocessed signal using hysteresis comparison to obtain a binarized sequence. This includes: for each sampling time, extracting a window signal from the preprocessed signal using a sliding window with a preset number of points before and after that sampling time; performing clustering processing on the window signal to obtain a first cluster corresponding to a high-level state and a second cluster corresponding to a low-level state; determining a dynamic threshold based on the center of the first cluster and the center of the second cluster; determining an upper hysteresis threshold and a lower hysteresis threshold based on the dynamic threshold; setting the binarized sequence to a first state when the preprocessed signal crosses the upper hysteresis threshold from bottom to top; and setting the binarized sequence to a second state when the preprocessed signal crosses the lower hysteresis threshold from top to bottom, so as to output a stable binarized sequence.

[0185] Optionally, the extraction module performs edge detection on the binarized sequence to obtain a rising edge sequence representing the leading edge of the slide object and a falling edge sequence representing the trailing edge of the slide object. This includes: traversing the binary states of adjacent sampling points in the binarized sequence; when an adjacent sampling point is detected to transition from the second state to the first state, recording the index of that sampling point as a rising edge index, and assembling the rising edge indices into a rising edge sequence in chronological order; when an adjacent sampling point is detected to transition from the first state to the second state, recording the index of that sampling point as a falling edge index, and assembling the falling edge indices into a falling edge sequence in chronological order.

[0186] Optionally, the analysis module performs multi-cycle analysis on the gap sequence to obtain preliminary rotation parameters, including: aligning the gap sequence into multiple cycles and constructing a statistical matrix based on the aligned data of each cycle; identifying the characteristic gaps of the bearing disk based on the statistical matrix; using the time sequence position corresponding to the end time of the characteristic gap as the preliminary absolute starting point; calculating the average rotation period based on the aligned data of each cycle; and calculating the preliminary rotation speed based on the average rotation period.

[0187] Optionally, the analysis module aligns the gap sequence across multiple rings and constructs a statistical matrix based on the aligned data of each ring. It then identifies the characteristic gaps of the bearing plate based on the statistical matrix, including: segmenting the gap sequence according to the theoretical rotation period based on the estimated rotation speed of the bearing plate to obtain corresponding multi-ring gap sub-sequences; performing initial phase alignment on each gap sub-sequence so that gaps with the same sequence number in different rings correspond to the same physical position on the bearing plate; arranging the aligned gap sub-sequences by ring number as rows and by circumferential position number as columns to construct a statistical matrix; calculating the median and interquartile range of each column in the statistical matrix; determining the target column that meets preset conditions based on the median and interquartile range of each column, and determining the circumferential position corresponding to the target column as the location of the characteristic gap of the bearing plate; wherein the preset conditions are determined by: if the median of a column is greater than a first threshold and the interquartile range is less than a second threshold, then the preset conditions are considered met.

[0188] Optionally, the list construction module maps the temporal feature information to angular coordinates based on the initial rotation parameters, thereby constructing a list of measured slide object positions, including: determining the rotation period based on the initial rotation speed; using the temporal position corresponding to the initial absolute starting point as the angular zero point, converting the sampling time sequence into an angular coordinate system based on the rotation period; converting the pulse width of each slide object in the pulse width sequence and each gap in the gap sequence into angle information in the angular coordinate system; wherein, the temporal feature information includes the starting time sequence position and the ending time sequence position, and the angle information includes the starting angle and the ending angle; and writing the angle information corresponding to each slide object as a position entry into the list of measured slide object positions representing the circumferential distribution of the bearing disk.

[0189] Optionally, the matching module performs circumferential phase matching between the measured list of wafer object locations and the preset feature model to determine the optimal phase offset, including: determining the number of currently detected wafer objects based on the measured list of wafer object locations; determining a matching strategy based on the difference between the number of wafer objects and the number of model trays; and calculating the matching degree between the measured list of wafer object locations and the feature model during the circumferential phase search process based on the matching strategy, and determining the phase offset corresponding to the optimal matching degree as the optimal phase offset.

[0190] Optionally, if the difference in quantity is less than or equal to the third threshold, the full disk condition is considered to be met. In this case, the matching strategy includes: aligning the list of measured wafer object positions with the feature model under multiple candidate phase offsets; for each candidate phase offset, counting the number of wafer objects falling within the relative angle tolerance of the corresponding tray of the feature model; and taking the candidate phase offset that maximizes the number of wafer objects as the optimal phase offset.

[0191] Optionally, if the difference in quantity is greater than the third threshold, it is considered to meet the non-full disk condition. In this case, the matching strategy includes: aligning the list of measured wafer objects with the feature model under multiple candidate phase offsets; for any candidate phase offset, if all wafer objects can be mapped separately and uniquely to the relative angle tolerance of the corresponding tray in the feature model, the type of the occupied tray is compatible with the feature of the measured wafer object, and the circumferential interval corresponding to the unoccupied tray corresponds to the gap interval in the list of measured wafer object positions, then the candidate phase offset is determined to be a valid candidate phase; if there are multiple valid candidate phases, the valid candidate phase with the smallest matching error is selected as the optimal phase offset.

[0192] Optionally, the calculation module corrects the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters, including: correcting the initial absolute starting point based on the optimal phase offset to obtain the target absolute starting point; using the timing position corresponding to the target absolute starting point as the timing reference point; after the timing reference point, obtaining the timing positions of the carrier objects corresponding to at least two reference trays with known circumferential angular intervals in the feature model; determining the angular velocity of the carrier disk based on the circumferential angular interval of the reference trays and the time difference between the timing positions of the carrier objects corresponding to the reference trays, and calculating the target rotational speed based on the angular velocity.

[0193] The aforementioned system for calculating the rotation parameters of the semiconductor carrier disk can acquire the reflected signal during the disk's rotation process by setting a single photoelectric sensor on the sidewall of the semiconductor reaction cavity. This simplifies the hardware configuration from the outset, avoiding the high costs and maintenance burdens associated with complex multi-channel acquisition and calibration. By extracting features from the reflected signal, calculable temporal feature information is obtained, including pulse width sequences and gap sequences. Multi-cycle analysis of the gap sequences yields more stable preliminary rotation parameters. The temporal feature information is mapped to angular coordinates to construct a list of measured carrier object positions, providing a directly comparable benchmark for subsequent positioning. Finally, the list of measured carrier object positions is matched with the feature model to obtain the optimal phase offset. Based on this optimal phase offset, the preliminary rotation parameters are corrected to obtain the target rotation parameters. The feature model of this application does not rely on precise geometric coordinate drawings, making the algorithm more tolerant to manufacturing errors of the carrier plate, dimensional drift caused by thermal cycling, and slight deformation. It can be adapted to different models, batches, and even slightly deformed carrier plates, significantly improving versatility and long-term stability. At the same time, the core calculation of this application focuses on feature extraction of one-dimensional time-domain sequences, multi-period statistics, and topology matching. Compared with existing image synthesis and global graphic matching schemes, the calculation link is shorter, the implementation is simpler, and the real-time performance is stronger. It can achieve fast output under low computing power conditions, meeting the response speed requirements of online monitoring and closed-loop control. In addition, this application adopts a two-level strategy of initial positioning based on the measured carrier object position list combined with fine matching of the feature model. Without requiring the detection of all tray states, it can establish a globally consistent absolute phase relationship based only on the detected tray features and their measured distribution, thereby stably obtaining the target rotation parameters. This ensures that accurate and reliable measurement and positioning results can still be output under complex working conditions such as sparse feature points and non-full tray.

[0194] Each module in the aforementioned semiconductor carrier disk rotation parameter calculation 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 the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0195] In one feasible embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface 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 interface. 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 interface is 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 executed by the processor, the computer program implements the calculation method for the rotation parameters of the semiconductor carrier disk. 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.

[0196] Those skilled in the art will understand that Figure 8 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.

[0197] 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 calculating the rotation parameters of the semiconductor carrier disk.

[0198] In one feasible embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the method steps in the above-described method for calculating the rotation parameters of the semiconductor carrier disk.

[0199] In one feasible embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the method steps in the above-described method for calculating the rotation parameters of the semiconductor carrier disk.

[0200] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0201] 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 calculating the rotation parameters of a semiconductor carrier disk, characterized in that, A semiconductor detection apparatus including a photoelectric sensor is applicable, wherein the photoelectric sensor is disposed on the sidewall of a semiconductor reaction chamber, and a carrier disk is provided with multiple trays for placing wafer objects; the method includes: Acquire the reflected signal output by the photoelectric sensor during the rotation of the carrier disk; Feature extraction is performed on the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence; Multi-period analysis was performed on the gap sequence to obtain preliminary rotation parameters; Based on the preliminary rotation parameters, the time-series feature information is mapped to angular coordinates to construct a list of measured slide object positions. The measured list of carrier object positions and the preset feature model are subjected to circumferential phase matching to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance. The target rotation parameters are obtained by correcting the initial rotation parameters based on the optimal phase offset.

2. The method according to claim 1, characterized in that, The step of extracting features from the reflected signal to obtain temporal feature information includes: The reflected signal is filtered to obtain a preprocessed signal; The preprocessed signal is binarized, and edge detection is performed based on the processing result to obtain temporal feature information.

3. The method according to claim 2, characterized in that, The step of binarizing the preprocessed signal and performing edge detection based on the processing result to obtain temporal feature information includes: The preprocessed signal is adaptively and dynamically thresholded based on a sliding window, and binarized using hysteresis comparison to obtain a binarized sequence. Edge detection is performed on the binarized sequence to obtain the rising edge sequence representing the leading edge of the slide object and the falling edge sequence representing the trailing edge of the slide object; For each wafer object, the difference in the number of points between its corresponding rising edge and falling edge is calculated as the pulse width of that wafer object; For two adjacent slide objects, the difference in the number of points between the rising edge of the latter slide object and the falling edge of the former slide object is calculated as the gap to form an alternating pulse width sequence and gap sequence.

4. The method according to claim 3, characterized in that, The step of adaptively calculating a dynamic threshold for the preprocessed signal based on a sliding window and binarizing the preprocessed signal using hysteresis comparison to obtain a binarized sequence includes: For each sampling time, a window signal is extracted from the preprocessed signal by a sliding window with a preset number of points before and after that sampling time; The window signals are clustered to obtain a first cluster corresponding to the high-level state and a second cluster corresponding to the low-level state; A dynamic threshold is determined based on the centers of the first cluster and the centers of the second cluster; The upper hysteresis threshold and the lower hysteresis threshold are determined based on the dynamic threshold. When the preprocessed signal crosses the upper hysteresis threshold from bottom to top, the binarized sequence is set to the first state; when the preprocessed signal crosses the lower hysteresis threshold from top to bottom, the binarized sequence is set to the second state to output a stable binarized sequence.

5. The method according to claim 4, characterized in that, The step of performing edge detection on the binarized sequence to obtain a rising edge sequence representing the leading edge of the slide object and a falling edge sequence representing the trailing edge of the slide object includes: In the binarized sequence, the binary states of adjacent sampling points are traversed; When an adjacent sampling point is detected to transition from the second state to the first state, the index of that sampling point is recorded as the rising edge index, and the rising edge indices are arranged in chronological order to form the rising edge sequence. When an adjacent sampling point is detected to transition from the first state to the second state, the index of that sampling point is recorded as the falling edge index, and the falling edge indices are arranged in chronological order to form the falling edge sequence.

6. The method according to claim 1, characterized in that, The preliminary rotation parameters include the preliminary absolute starting point and the preliminary rotational speed; The process of performing multi-period analysis on the gap sequence to obtain preliminary rotation parameters includes: The gap sequence is aligned in multiple loops and a statistical matrix is ​​constructed based on the data of each loop after alignment. The characteristic gaps of the bearing disk are identified based on the statistical matrix. The timing position corresponding to the end time of the aforementioned feature gap is taken as the initial absolute starting point; The average rotation period is calculated based on the aligned data of each revolution, and the initial rotation speed is calculated based on the average rotation period.

7. The method according to claim 6, characterized in that, The step of aligning the gap sequence through multiple loops and constructing a statistical matrix based on the aligned loop data, and identifying the characteristic gaps of the bearing disk based on the statistical matrix, includes: Based on the estimated rotational speed of the bearing plate, the gap sequence is segmented according to the theoretical rotational period to obtain a gap subsequence corresponding to multiple revolutions; Perform initial phase alignment on each of the aforementioned gap subsequences so that gaps with the same sequence number in different rings correspond to the same physical position on the bearing disk; Arrange the aligned inter-loop subsequences by loop number as rows and by circumferential position number as columns to construct a statistical matrix; For each column of the statistical matrix, calculate the median and interquartile range of the data in that column; Based on the median and interquartile range of each column, the target column that meets the preset conditions is determined, and the circumferential position of the target column is determined as the location of the characteristic gap of the bearing plate. The preset conditions are determined by the following method: if the median of a column is greater than the first threshold and the interquartile range is less than the second threshold, then the preset conditions are considered to be met.

8. The method according to claim 6, characterized in that, The process of mapping the temporal feature information to angular coordinates based on the preliminary rotation parameters to construct a list of measured slide object positions includes: The rotation period is determined based on the initial rotation speed; Using the time position corresponding to the initial absolute starting point as the zero point of angle, the sampling time sequence is converted into an angle coordinate system based on the rotation period; For each slice object in the pulse width sequence and each gap in the gap sequence, they are converted into angle information in the angle coordinate system; wherein, the timing feature information includes the start timing position and the end timing position, and the angle information includes the start angle and the end angle; The angle information corresponding to each slide object is used as a position entry and written into the measured slide object position list that characterizes the circumferential distribution of the bearing disk.

9. The method according to claim 1, characterized in that, The feature model includes the number of model trays and the relative angle tolerance of each interval; The step of performing circumferential phase matching between the measured slide object position list and the preset feature model to determine the optimal phase offset includes: The number of currently detected wafer objects is determined based on the measured wafer object location list; Based on the difference between the number of the slide objects and the number of the model trays, a matching strategy is determined; based on the matching strategy, the matching degree between the measured slide object position list and the feature model is calculated during the circumferential phase search process, and the phase offset corresponding to the optimal matching degree is determined as the optimal phase offset.

10. The method according to claim 9, characterized in that, If the difference in quantity is less than or equal to the third threshold, the full-board condition is considered met, and the matching strategy includes: Align the list of measured slide object positions with the feature model under multiple candidate phase offsets; For each candidate phase offset, the number of wafer objects falling within the relative angle tolerance of the tray corresponding to the feature model is counted. The candidate phase offset that maximizes the number of the wafer objects is taken as the optimal phase offset.

11. The method according to claim 9, characterized in that, If the difference in quantity is greater than the third threshold, the non-full board condition is considered met, and the matching strategy includes: Align the list of measured slide object positions with the feature model under multiple candidate phase offsets; For any candidate phase offset, if all the wafer objects can be mapped to the relative angle tolerance of the corresponding tray in the feature model, the type of the occupied tray is compatible with the feature of the measured wafer object, and the circumferential interval corresponding to the unoccupied tray corresponds to the gap interval in the position list of the measured wafer object, then the candidate phase offset is determined to be a valid candidate phase. When multiple valid candidate phases exist, the valid candidate phase with the smallest matching error is selected as the optimal phase offset.

12. The method according to claim 6, characterized in that, The target rotation parameters include the target absolute starting point and the target rotation speed; The process of correcting the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters includes: Based on the optimal phase offset, the preliminary absolute starting point is corrected to obtain the target absolute starting point; The timing reference point is the time position corresponding to the absolute starting point of the target. After the timing reference point, obtain the timing positions of the wafer objects corresponding to at least two reference trays with known circumferential angular intervals in the feature model; The angular velocity of the carrier disk is determined based on the circumferential angular interval of the reference tray and the time difference between the timing positions of the corresponding carrier objects on the reference tray, and the target rotational speed is calculated based on the angular velocity.

13. A system for calculating the rotation parameters of a semiconductor carrier disk, characterized in that, The semiconductor detection device includes a photoelectric sensor disposed on the sidewall of the semiconductor reaction chamber, and a support disk has multiple trays for placing wafers. The system includes: The acquisition module is used to acquire the reflected signal output by the photoelectric sensor during the rotation of the carrier disk; An extraction module is used to extract features from the reflected signal to obtain temporal feature information, which includes pulse width sequence and gap sequence; The analysis module is used to perform multi-period analysis on the gap sequence to obtain preliminary rotation parameters; The list construction module is used to map the temporal feature information to angular coordinates based on the preliminary rotation parameters, thereby constructing a list of measured slide object positions; The matching module is used to perform circumferential phase matching between the measured list of carrier object positions and the preset feature model to determine the optimal phase offset; the feature model is a set describing the tray layout topology and relative angle tolerance. The calculation module is used to correct the initial rotation parameters based on the optimal phase offset to obtain the target rotation parameters.

14. 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 1-12.

15. 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 1-12.

Images