End point detection method, device and system, semiconductor manufacturing equipment and medium

Abstract

The invention discloses an end point detection method, device and system, semiconductor manufacturing equipment and a medium, relates to the technical field of semiconductor manufacturing, and is applied to an end point detection system, the end point detection system comprises a spectrometer and an optical switch, the input end of the optical switch is connected with multiple sampling channels corresponding to a wafer to be detected, and the output end of the optical switch is connected with the spectrometer. Comprising the following steps: acquiring a signal processing result from a spectrograph; the signal processing result is a result obtained by processing a spectral signal received from the optical switch by the spectrometer based on the first parameter, and the spectral signal is obtained by the optical switch from a target sampling channel corresponding to the second parameter; wherein the first parameter and the second parameter belong to end point detection parameters; and detecting a target end point based on the signal processing result and the sampling point position of the target sampling channel to obtain an end point detection result. Therefore, the optical switch is compatible with spectral signals of different sampling channels, the terminal point state can be cooperatively judged, and the terminal point detection accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of semiconductor manufacturing technology, and in particular to endpoint detection methods, apparatus, systems, semiconductor manufacturing equipment and media. Background Technology

[0002] In the high-precision etching / deposition process of semiconductor manufacturing, the accuracy of endpoint detection directly determines the critical dimension control and yield of devices. Commonly used endpoint detection technologies are mainly based on two principles: Optical Emission Spectrometer (OES) and Interferometer Endpoint (IEP). OES technology acquires real-time chemical composition information during the etching / deposition process by monitoring the intensity of specific wavelengths in the plasma emission spectrum, thereby inferring changes in the etched / deposited material. IEP technology, on the other hand, dynamically monitors the thickness of the etched / deposited film by analyzing the intensity signal of the interference light. Each of these technologies acquires a signal from a single sampling point via an optical fiber, which is then transmitted to two spectrometers, OES and IEP, for integration. This approach can lead to insufficient accuracy in endpoint determination due to feature extraction from a single point during the etching / deposition process, and the use of multiple spectrometers imposes cost and space constraints. Summary of the Invention

[0003] This application discloses endpoint detection methods, apparatus, systems, semiconductor manufacturing equipment, and media, which address the cost pressures of multiple spectrometers and the problem of insufficient endpoint detection accuracy.

[0004] In a first aspect, this application provides an endpoint detection method applied to an endpoint detection system, the endpoint detection system including a spectrometer and an optical switch, wherein the input end of the optical switch is connected to multiple sampling channels corresponding to the wafer to be inspected, and the output end of the optical switch is connected to the spectrometer, the method comprising: acquiring a signal processing result from the spectrometer; the signal processing result being the result obtained by the spectrometer processing a spectral signal received from the optical switch based on a first parameter, the spectral signal being obtained by the optical switch from a target sampling channel corresponding to a second parameter; wherein the first parameter and the second parameter are endpoint detection parameters, the endpoint detection parameters including pre-set parameters related to performing endpoint detection operations on the wafer to be inspected; detecting a target endpoint based on the signal processing result and the sampling point of the target sampling channel to obtain an endpoint detection result corresponding to the wafer to be inspected; the target endpoint being an etching endpoint or a deposition endpoint. As can be seen, optical switches can guide spectral signals from multiple sampling channels to the same spectrometer for processing, enabling flexible control of multiple sampling channels. This allows for compatibility with spectral signals from different sampling channels, ensuring consistency in spectral responses between different channels and saving hardware resources. Furthermore, by combining the signal processing results corresponding to the spectral signals with the spatial location information of the sampling points in the sampling channels, the endpoint status of different sampling points can be collaboratively determined, thereby improving the accuracy of endpoint detection.

[0005] In one possible implementation, the endpoint detection parameters further include a third parameter characterizing spectral sampling using optical emission spectroscopy. The method then further includes: if no light source is configured in the endpoint detection system, performing a target configuration operation; if a light source is configured in the endpoint detection system, configuring the light source to a closed state and performing the target configuration operation. The target configuration operation includes configuring the spectrometer's integration parameters based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter. The target sampling channel is one or more sampling channels of the optical emission spectroscopy method that corresponds to the second parameter. Therefore, this solution is compatible with optical emission spectroscopy for spectral sampling and adapts to diverse sampling methods.

[0006] In one possible implementation, the endpoint detection parameters further include a fourth parameter characterizing spectral sampling using interferometry. The method then further includes: if a light source is configured in the endpoint detection system, configuring the light source to an on state and performing a target configuration operation; wherein the target configuration operation includes configuring the spectrometer's integration parameters based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter; the target sampling channel is one or more sampling channels of the interferometry method that corresponds to the second parameter. Therefore, this solution is compatible with interferometry for spectral sampling and adapts to diverse sampling methods.

[0007] In one possible implementation, the signal processing result is the result obtained by the spectrometer integrating the spectral signal received from the optical switch using the first parameter. Therefore, in this solution, the spectrometer can receive spectral signals from the optical switch and integrate them using configured parameters. This allows a single spectrometer to acquire and process spectral signals from different sampling channels from the optical switch, saving hardware resources.

[0008] In one possible implementation, the step of detecting the target endpoint based on the signal processing result and the sampling points of the target sampling channel to obtain the endpoint detection result corresponding to the wafer under test includes: extracting features from the signal processing result to obtain spectral feature data corresponding to each sampling point of the target sampling channel; the spectral feature data is used to reflect the spectral characteristics of the film layer of the wafer under test; and detecting the film layer thickness of the wafer under test based on the spectral feature data and the spatial location information of each sampling point to obtain the endpoint detection result corresponding to the wafer under test. It can be seen that this solution can combine the spectral feature data corresponding to the sampling points of the target sampling channel and the spatial location information of the sampling points to perform endpoint detection on the thickness of the corresponding film layer of the wafer under test. This allows for coordinated judgment of the endpoint state of the film layer at different sampling points, improving the accuracy of endpoint detection.

[0009] In one possible implementation, the step of detecting the target endpoint of the film thickness of the wafer to be tested based on the spectral feature data and the spatial location information of the sampling points includes: fitting thickness variation curves to the film thickness characterized by each of the spectral feature data to obtain thickness variation fitting curves corresponding to each of the spectral feature data; determining the target time corresponding to the target thickness in each of the thickness variation fitting curves; the target thickness being the film thickness corresponding to the target endpoint; and generating the endpoint detection result corresponding to the wafer to be tested based on the proportion of each target time within a preset time window and the spatial location information of each sampling point. It can be seen that in this scheme, the endpoint detection can be performed using thickness variation curve fitting to predict the target time when the film thickness matches the target endpoint; then, based on the time window in which the target time is located, the corresponding endpoint detection result is generated. This is more adaptable to asynchronous thickness changes and has robustness to local anomalies.

[0010] In one possible implementation, the step of detecting the film thickness of the wafer under test based on the spectral feature data and the spatial location information of each sampling point includes: assigning weighting factors to each spectral feature data based on target signal parameters of the spectral signal; the target signal parameters include signal-to-noise ratio and / or signal rate of change; performing weighted calculation on the film thickness variation characterized by the spectral feature data using the weighting factors; and generating the endpoint detection result corresponding to the wafer under test based on the matching relationship between the weighted film thickness and the film thickness corresponding to the target endpoint, as well as the spatial location information of each sampling point. It can be seen that this scheme can assign dynamic weighting factors to each spectral feature data based on the signal parameters of the spectral signal, reflecting its importance to endpoint detection. This allows for adaptive adjustment of the sampling channels of interest during nonlinear etching processes, improving the effectiveness of the acquired channel data.

[0011] In one possible implementation, the step of detecting the target endpoint of the film thickness of the wafer to be tested based on the spectral feature data and the spatial location information of the sampling points includes: processing the spectral feature data and the spatial location information of the sampling points based on Bayesian fusion to obtain a target probability result; the target probability result is the probability that the film thickness corresponding to the wafer to be tested matches the film thickness corresponding to the target endpoint; and generating the endpoint detection result corresponding to the wafer to be tested based on the relationship between the target probability result and a preset probability threshold. It can be seen that this solution can effectively handle the inconsistency of channel signals caused by uneven spatial distribution of film thickness by calculating the posterior probability of the endpoint through Bayesian inference. By statistical inference of different positions, it achieves consistency in the judgment of the global endpoint state and improves the consistency response capability to spatial etching endpoints.

[0012] In one possible implementation, the method further includes: determining the endpoint detection parameters corresponding to the current detection cycle; wherein the endpoint detection parameters include a synchronization timestamp, so as to synchronize the spectrometer and the optical switch based on the synchronization timestamp. Therefore, this solution can synchronize the spectrometer, optical switch, and light source using a synchronization timestamp, ensuring the accuracy of the acquisition timing and the reliability of feature extraction.

[0013] In one possible implementation, determining the endpoint detection parameters corresponding to the current detection cycle includes: if the endpoint detection result of the previous detection cycle indicates that the film thickness of the wafer to be inspected does not match the film thickness corresponding to the target endpoint, then at the beginning of the current detection cycle, adjusting the endpoint detection parameters corresponding to the previous detection cycle based on the endpoint detection result, and determining the endpoint detection parameters corresponding to the current detection cycle based on the adjusted parameters and the synchronization timestamp. Therefore, this solution generates preset sampling parameters for the current endpoint detection cycle by combining the endpoint detection results of previous detection cycles in different detection cycles, adapting to the dynamic changes in film thickness and improving endpoint detection accuracy.

[0014] In one possible implementation, the method further includes: updating the endpoint detection parameters based on the signal processing results within the current detection period, so as to trigger the next endpoint detection operation using the updated endpoint detection parameters. Therefore, within the current endpoint detection time period, the sampling parameters can also be dynamically refreshed based on the signal processing results to adapt to changes in film thickness in real time, thereby improving endpoint detection accuracy.

[0015] Secondly, this application provides an endpoint detection device applied to an endpoint detection system, the endpoint detection system including a spectrometer and an optical switch, wherein the input terminal of the optical switch is connected to multiple sampling channels corresponding to the wafer to be detected, and the output terminal of the optical switch is connected to the spectrometer, the device comprising:

[0016] The signal processing result acquisition module is used to acquire the signal processing result from the spectrometer; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operations on the wafer to be detected;

[0017] The endpoint detection module is used to detect the target endpoint based on the signal processing result and the sampling point of the target sampling channel, and obtain the endpoint detection result corresponding to the wafer to be tested; the target endpoint is the etching endpoint or the deposition endpoint.

[0018] Thirdly, this application provides an endpoint detection device, including a memory and a processor; wherein,

[0019] The memory is used to store computer programs;

[0020] The processor is used to execute the computer program to implement the endpoint detection method as described above.

[0021] Fourthly, this application provides an endpoint detection system, comprising:

[0022] An optical switch connected to multiple sampling channels corresponding to the wafer under test at its input end is used to acquire spectral signals from the target sampling channel corresponding to the second parameter.

[0023] A spectrometer connected to the output of the optical switch is used to process the spectral signal received from the optical switch based on a first parameter to obtain a signal processing result;

[0024] The endpoint detection device, as described above, connected to the spectrometer, is used to acquire the signal processing results from the spectrometer, and to detect the target endpoint based on the signal processing results and the sampling point of the target sampling channel, thereby obtaining the endpoint detection result corresponding to the wafer to be tested; the target endpoint is an etching endpoint or a deposition endpoint.

[0025] The first parameter and the second parameter are endpoint detection parameters, which include pre-set parameters related to performing endpoint detection operations on the wafer to be inspected.

[0026] Fifthly, this application provides a semiconductor manufacturing apparatus, including: a process chamber where a wafer to be inspected is located, and an endpoint detection system as described above connected to the process chamber via multiple sampling channels.

[0027] Sixthly, this application provides a computer-readable storage medium for storing a computer program that, when executed by a processor, implements the endpoint detection method described above.

[0028] Therefore, this method can be applied to an endpoint detection system, which includes a spectrometer and an optical switch. The input of the optical switch is connected to multiple sampling channels corresponding to the wafer under test, and the output of the optical switch is connected to the spectrometer. The method includes: acquiring a signal processing result from the spectrometer; the signal processing result is obtained by the spectrometer processing the spectral signal received from the optical switch based on a first parameter, where the spectral signal is obtained by the optical switch from a target sampling channel corresponding to a second parameter; wherein the first parameter and the second parameter are endpoint detection parameters, which include pre-set parameters related to performing endpoint detection operations on the wafer under test; detecting the target endpoint based on the signal processing result and the sampling point of the target sampling channel to obtain the endpoint detection result corresponding to the wafer under test; the target endpoint is an etching endpoint or a deposition endpoint. In this way, this solution can guide spectral signals from multiple sampling channels to the same spectrometer for processing via an optical switch, achieving flexible control of multiple sampling channels, compatibility with spectral signals from different sampling channels, saving hardware resources, and ensuring the consistency of spectral response between different channels. Furthermore, by combining the signal processing results corresponding to the spectral signals with the spatial location information of the sampling points of the sampling channels, the endpoint status of different sampling points can be collaboratively judged, which can improve the accuracy and consistency of target endpoint judgment. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a diagram illustrating the architecture of an endpoint detection system disclosed in an embodiment of this application.

[0031] Figure 2 This is a flowchart of an endpoint detection method disclosed in an embodiment of this application;

[0032] Figure 3 This is a flowchart of a specific endpoint detection method disclosed in an embodiment of this application;

[0033] Figure 4 This is a flowchart of another specific endpoint detection method disclosed in an embodiment of this application;

[0034] Figure 5 This is a flowchart of an endpoint determination process based on Bayesian fusion disclosed in an embodiment of this application;

[0035] Figure 6 This is a flowchart of another specific endpoint detection method disclosed in the embodiments of this application;

[0036] Figure 7 This is a schematic diagram illustrating a specific sampling point adjustment disclosed in an embodiment of this application;

[0037] Figure 8 This is a schematic diagram illustrating a specific sampling duration adjustment disclosed in an embodiment of this application;

[0038] Figure 9 This is a structural diagram of an endpoint detection device disclosed in an embodiment of this application;

[0039] Figure 10 This is a specific endpoint detection system architecture diagram disclosed in an embodiment of this application;

[0040] Figure 11 This is another specific endpoint detection system architecture diagram disclosed in the embodiments of this application;

[0041] Figure 12 This is a multi-channel sampling timing diagram disclosed in an embodiment of this application. Detailed Implementation

[0042] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] In the high-precision etching / deposition process of semiconductor manufacturing, OES and IEP technologies are involved. If these two technologies are used, when acquiring the spectral signal of the wafer, separate optical fibers are usually used to acquire the spectral signal of the sampling point of the wafer, and then the signal is transmitted to two spectrometers, OES and IEP, for integration processing. This will lead to insufficient accuracy in feature extraction and endpoint determination of a single sampling point during the etching / deposition process, and multiple spectrometers will bring cost and space pressure.

[0044] To address the aforementioned technical issues, this application proposes an endpoint detection method. By using an optical switch, spectral signals from multiple sampling channels can be guided to the same spectrometer for processing, enabling flexible control of multiple sampling channels. This method is compatible with spectral signals from different sampling channels, saves hardware resources, and ensures consistency in spectral responses between different channels. Furthermore, by combining the signal processing results corresponding to the spectral signals with the spatial location information of the sampling points in the sampling channels, the endpoint status of different sampling points can be collaboratively determined, improving the accuracy and consistency of target endpoint determination.

[0045] Furthermore, such as Figure 1 The diagram illustrates an endpoint detection system architecture. This system can connect to a semiconductor process chamber and perform endpoint detection during wafer etching / deposition process control. During the etching / deposition process control, endpoint detection parameters are sent to the processing unit. The processing unit of the endpoint detection system can use these parameters to configure the spectrometer, optical switch, and light source. The optical switch acquires spectral signals from the wafer film layer through sampling channels corresponding to different sampling points in the process chamber. It is understood that OES and IEP technologies have different requirements for the light source, corresponding to different sampling channels. The optical switch allows connection to different sampling channels at different times, ensuring compatibility with spectral signals from different sampling channels. The optical switch then uploads spectral signals from different sampling channels to the same spectrometer, which processes the spectral signals and uploads them to the processing unit. The processing unit analyzes the processed spectral signals to determine the etching / deposition endpoint or adjusts the endpoint detection parameters to continue acquiring spectral signals. It is understood that the endpoint detection system disclosed in this solution is applicable to semiconductor equipment requiring in-situ measurement of film layer states, such as plasma etching equipment, atomic layer deposition equipment, and physical / chemical vapor deposition equipment.

[0046] The endpoint detection method disclosed in this application can be applied to, for example... Figure 1 The endpoint detection system shown.

[0047] like Figure 2 As shown, this embodiment discloses an endpoint detection method. Exemplarily, this method is executed by the processing unit of the aforementioned endpoint detection system, and specifically includes:

[0048] Step S11: Obtain the signal processing result from the spectrometer; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operations on the wafer to be detected.

[0049] In this context, "endpoint" refers to the stopping point in semiconductor manufacturing that requires precise control, such as stopping the process when the wafer film reaches the ideal thickness; "endpoint detection" refers to determining whether the wafer film meets the corresponding process termination conditions in different semiconductor manufacturing processes, such as etching or deposition processes.

[0050] In one implementation, parameters related to the endpoint detection of film thickness on the wafer to be inspected can be preset and stored, i.e., endpoint detection parameters. These endpoint detection parameters include a first parameter corresponding to the spectrometer and a second parameter corresponding to the optical switch. It is understood that the first parameter specifically refers to the parameters involved in the calculation of the spectral signal, such as the integration time related to the integration calculation; the second parameter is used to characterize which sampling channels the optical switch is connected to.

[0051] In another embodiment, the endpoint detection parameters further include a third parameter characterizing spectral sampling using optical emission spectroscopy. Further, if no light source is configured in the endpoint detection system, a target configuration operation is performed; if a light source is configured in the endpoint detection system, the light source is configured to be off, and the target configuration operation is performed. The target configuration operation includes configuring the spectrometer's integration parameters based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter. The target sampling channel is one or more sampling channels corresponding to the optical emission spectroscopy method that correspond to the second parameter. It is understood that when performing endpoint detection, the sampling method needs to be clearly defined. The third parameter determines whether spectral sampling using optical emission spectroscopy is employed. Since optical emission spectroscopy utilizes the spontaneous emission of excited-state particles in plasma to analyze spectral intensities at different wavelengths, a light source is not required, and the light source can be configured to be off. Correspondingly, the sampling channel corresponding to this optical emission spectroscopy method is connected to an optical switch, so that the optical switch can send the spectral information transmitted by the corresponding sampling channel to the spectrometer; furthermore, parameters such as integration time can be set for the spectrometer to process the spectral information uploaded by the optical switch (the spectral signal obtained from the sampling channel connected to the optical switch).

[0052] In another embodiment, the endpoint detection parameters may further include a fourth parameter characterizing spectral sampling using interferometry. The method then further includes: if a light source is configured in the endpoint detection system, configuring the light source to an on state and performing a target configuration operation; wherein the target configuration operation includes configuring the spectrometer's integration parameters based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter; the target sampling channel is the sampling channel corresponding to the second parameter among one or more sampling channels corresponding to the interferometry method. It is understood that interferometry, by monitoring the change in interference intensity of laser / white light reflection signals to calculate the change in film thickness over time, requires a light source; when the fourth parameter characterizes spectral sampling using interferometry, the light source is configured to an on state; correspondingly, the optical switch also needs to be connected to the sampling channel related to the interferometry method, connecting the target sampling channel corresponding to the second parameter to the optical switch so that the spectral signals transmitted by each target sampling channel are sent to the spectrometer via the optical switch. Furthermore, the spectrometer also needs to be configured using the first parameter, such as the integration time, in order to process the spectral signal sent by the optical switch, such as performing preprocessing like filtering and normalization.

[0053] It should be noted that the signal processing result is obtained by the spectrometer integrating the spectral signal received from the optical switch using the first parameter. This means the spectrometer receives the spectral signal corresponding to the wafer under test from the relevant sampling channels of the optical switch, and then integrates these spectral signals using integration parameters pre-configured based on the first parameter, achieving unified processing of multispectral signals to obtain the signal processing result. Furthermore, the first parameter can specifically be the integration parameters related to the integration calculation of spectral signals from optical emission spectroscopy and interferometry, and the parameters to be used at which times for integration calculation.

[0054] Step S12: Based on the signal processing results and the sampling points of the target sampling channel, the target endpoint is detected to obtain the endpoint detection result corresponding to the wafer to be tested; the target endpoint is the etching endpoint or the deposition endpoint.

[0055] In this application, the above steps combine an optical switch and multiple sampling channels to perform spectral sampling on the wafer under test using the sampling method corresponding to the endpoint detection parameters. The signal processing result is then obtained by integrating the sampled spectral signal using a spectrometer. Next, by combining this signal processing result with the sampling points corresponding to the sampling channels (which locations on the wafer were sampled for spectral information), etching / deposition endpoint detection can be performed. The obtained endpoint detection result reflects whether the currently sampled spectral data indicates that the wafer film layer meets the corresponding process termination conditions. This combination of information from the spectral signal itself and the sampling point information of the sampling channels improves the accuracy of endpoint detection.

[0056] Therefore, this solution employs a multi-input single-output optical switch to achieve switching control of multiple optical sampling channels, including OES and IEP signals from multiple sampling positions. The optical switch guides signals from multiple positions to the same spectrometer for time-division acquisition, saving hardware resources and ensuring spectral response consistency between different channels, meeting the spatial resolution requirements for endpoint detection in large-area etched regions. Furthermore, it allows for simultaneous control of the light source, optical switch, and spectrometer, dynamically determining the next sampling channel and its integration time based on real-time sampling signal characteristics (such as signal-to-noise ratio, light intensity, and slope changes). This improves the signal quality of each channel and enables priority acquisition and accuracy optimization of critical areas, enhancing the overall system response efficiency and acquisition robustness.

[0057] Understandably, endpoint detection parameters need to be set when performing endpoint detection for the first time; such as Figure 3 As shown, this embodiment will specifically describe the initial setting of the endpoint detection parameters and the process of parameter adjustment and updating during the process, specifically including:

[0058] Step S21: Determine the endpoint detection parameters corresponding to the current detection cycle; wherein, the endpoint detection parameters include a synchronization timestamp, so as to synchronize the spectrometer and the optical switch based on the synchronization timestamp.

[0059] Understandably, at the start of a single endpoint detection cycle, specific endpoint detection parameters need to be set. This includes specifying the sampling method for the spectral signal, whether a light source is required, which sampling channels correspond to this sampling method, setting the connection status of the optical switch to these sampling channels, and setting the integration parameters of the spectrometer. In other words, the spectrometer and optical switch are configured before endpoint detection, and the light source is configured as needed, so that the configured endpoint detection system can perform endpoint detection on the wafer to be tested. To accurately control signal sampling and ensure the consistency of the light source, optical switch, and spectrometer, a synchronization timestamp can be set during the generation of endpoint detection parameters to synchronously trigger the configuration process of the light source, optical switch, and light source.

[0060] Step S22: If the endpoint detection result of the previous detection cycle indicates that the film thickness of the wafer to be tested does not match the film thickness corresponding to the target endpoint, then at the beginning of the current detection cycle, the endpoint detection parameters corresponding to the previous detection cycle are adjusted based on the endpoint detection result, and the endpoint detection parameters corresponding to the current detection period are determined based on the adjusted parameters and the synchronization timestamp.

[0061] Understandably, during an endpoint detection cycle, the endpoint detection results of the previous cycle can be considered, and the endpoint detection parameters can be periodically adjusted to update the configuration status of the light source, optical switch, and spectrometer. This allows for dynamic adjustment of the signal acquisition points and acquisition time corresponding to the endpoint detection, thereby gradually optimizing the endpoint detection effect.

[0062] Step S23: Within the current detection cycle, update the endpoint detection parameters according to the signal processing results, so as to trigger the next endpoint detection operation using the updated endpoint detection parameters.

[0063] Understandably, within a single endpoint detection cycle, the endpoint detection parameters corresponding to the current detection cycle can be dynamically refreshed by combining signal processing results (such as signal-to-noise ratio, slope changes, etc.); for example, the next sampling channel and its integration time can be adjusted based on the calculated signal-to-noise ratio and slope changes. This allows for real-time adjustment of endpoint detection parameters according to the actual spectral sampling conditions, improving the matching degree between endpoint detection-related parameters and the actual state of the wafer under test, thereby enhancing the accuracy of endpoint detection.

[0064] Therefore, this solution allows for pre-setting endpoint detection parameters including synchronization timestamps, precisely controlling signal sampling, and ensuring consistency between the light source, optical switch, and spectrometer. During endpoint detection, the endpoint detection parameters for the current detection cycle can be adjusted based on the results of the previous cycle. Furthermore, the endpoint detection parameters can be adjusted in real-time based on signal processing results within the current cycle. Through these steps of setting, adjusting, and updating endpoint detection parameters, the matching degree between the endpoint detection parameters and the actual wafer state can be dynamically optimized, improving the accuracy of the final acquired signal and further enhancing the accuracy of endpoint detection.

[0065] The foregoing embodiments focused on the spectral signal acquisition and processing of the endpoint detection method; based on this, such as Figure 4 As shown, this embodiment discloses an endpoint detection method, focusing on the process of obtaining endpoint detection results based on spectral feature data and spatial location information of sampling points; it can be applied to... Figure 1 The endpoint detection system shown; the method specifically includes:

[0066] Step S31: Obtain the signal processing result from the spectrometer; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operations on the wafer to be detected.

[0067] For details regarding step S31, please refer to the description in the foregoing embodiments; they will not be repeated here.

[0068] Step S32: Extract features from the signal processing results to obtain spectral feature data corresponding to the sampling points of each target sampling channel; the spectral feature data is used to reflect the spectral features of the film layer of the wafer to be tested.

[0069] Specifically, after obtaining the signal processing results transmitted by the spectrometer, feature data can be extracted, such as plasma excitation intensity, film thickness, film thickness change rate, and signal-to-noise ratio. For example, feature data can be extracted using methods such as principal component analysis and model fitting. It is understandable that the wafer to be tested can be sampled multiple times within a single time period, thus obtaining time-series corresponding feature data.

[0070] Step S33: Based on the spectral feature data and the spatial location information of each sampling point, the film thickness of the wafer to be tested is detected to the target endpoint, and the endpoint detection result corresponding to the wafer to be tested is obtained.

[0071] Specifically, by combining the extracted spectral feature data and the spatial location information of each sampling point, collaborative endpoint detection can be performed on the wafer. For example, different fusion strategies can be used for collaborative detection, such as the AND (simultaneous satisfaction) strategy where all points reach the endpoint (the spectral signals of the wafer represent that the film thickness of the relevant sampling points meets the corresponding process end adjustment), the center point priority strategy, the thickness-weighted fusion strategy, and other basic strategies, as well as the hybrid judgment strategy based on local fitting and global statistics, the time-weighted dynamic fusion strategy, the Bayesian fusion decision, etc. Any one of these strategies can be selected for endpoint detection, or multiple fusion strategies can be used in a weighted manner for endpoint detection. This application does not limit this approach.

[0072] In one embodiment, detecting the target endpoint of the film thickness of the wafer under test based on the spectral feature data and the spatial location information of each sampling point may include: fitting thickness variation curves to the film thickness characterized by each spectral feature data to obtain each thickness variation fitting curve corresponding to each spectral feature data; determining the target time corresponding to the target thickness in each thickness variation fitting curve (predicting when the wafer's film thickness will reach the ideal thickness based on the fitted curve); the target thickness is the film thickness corresponding to the target endpoint; and generating the endpoint detection result corresponding to the wafer under test based on the proportion of each target time within a preset time window and the spatial location information of each sampling point. It is understood that, based on the spectral feature data, curve fitting can be performed on the film thickness variation corresponding to each sampling channel, and based on the fitted thickness variation fitting curves, the future endpoint time (the time when the ideal thickness is reached) of the film thickness of the wafer under test can be predicted, i.e., the target time corresponding to the target thickness in the thickness variation fitting curve. Furthermore, the predicted target times are statistically analyzed, such as by calculating the median, standard deviation, or confidence interval. When the predicted endpoint times of most (e.g., over 80%) channels fall within a set time window, the overall endpoint can be considered near. Additionally, the spatial location information of the sampling points for different sampling channels needs to be considered. For example, sampling points at different locations such as the center, edge, or four corners of the wafer tray will have different predicted endpoint times due to spatial relationships. By comprehensively considering the spatial location information of different sampling points and the proportion of the corresponding target times within the preset time window, a hybrid judgment strategy of local curve fitting + global statistics is adopted. This approach can adapt to situations where thickness changes are asynchronous and is robust to local anomalies.

[0073] In another embodiment, detecting the film thickness of the wafer to be tested based on the spectral feature data and the spatial location information of the sampling points can include: assigning weighting factors to the spectral feature data based on target signal parameters of the spectral signals; the target signal parameters include signal-to-noise ratio and / or signal rate of change; using the weighting factors to perform weighted calculations on the film thickness changes characterized by the spectral feature data; and generating the endpoint detection result corresponding to the wafer to be tested based on the matching relationship between the weighted calculated film thickness and the film thickness corresponding to the target endpoint, as well as the spatial location information of the sampling points. It is understood that the target signal parameters can specifically refer to the signal-to-noise ratio, signal rate of change, etc., of the spectral signals (these signal parameters can be obtained when the spectrometer processes the spectral signals). Weighting factors can be assigned to the spectral feature data of the spectral signals corresponding to different sampling channels based on the target signal parameters; the weighting factors can reflect the importance of the spectral signals to the endpoint detection. Other factors can also be considered in the weighting factors, such as the film thickness corresponding to the spectral signals sampled by certain sampling channels at the previous moment. The closer to the predicted endpoint, the greater the weight that spectral signal corresponding to that channel can be set. Then, a weighted average of the real-time film thickness changes corresponding to each sampling channel can be calculated based on a weighting factor to determine whether it has reached the set endpoint film thickness threshold (i.e., the film thickness corresponding to the process termination condition mentioned in the previous embodiment). Furthermore, the spatial location information of the sampling points corresponding to different sampling channels can also be considered during the weighting process, such as sampling points at different locations like the center, edge, and corners of the wafer tray. The spatial location also affects the importance of the spectral signal for endpoint detection. By using this time-weighted dynamic fusion method for endpoint determination, the focus on certain sampling channels can be adaptively adjusted, improving the effectiveness of the spectral data.

[0074] In another embodiment, the detection of the film thickness of the wafer to be tested based on the spectral feature data and the spatial location information of the sampling points may include: processing the spectral feature data and the spatial location information of the sampling points based on Bayesian fusion to obtain a target probability result; the target probability result is the probability that the film thickness of the wafer to be tested, obtained from the sampled spectral data, matches the film thickness corresponding to the target endpoint (i.e., the process termination condition in the aforementioned embodiment); and generating an endpoint detection result corresponding to the wafer to be tested based on the relationship between the target probability result and a preset probability threshold. Understandably, if Bayesian fusion is used for endpoint detection, the endpoint judgment result of each sampling channel is treated as an independent piece of observational evidence (as a known fact). The posterior probability of the overall endpoint is calculated using a Bayesian inference framework (the probability of estimating a hypothesis based on known facts, here representing the probability that the overall film thickness of the wafer reaches the ideal thickness). First, for each sampling channel, the probability of "endpoint" and "non-endpoint" is calculated (based on spectral characteristic peak wavelength, light intensity, thickness change rate, etc.). Then, the observational evidence from multiple sampling channels is fused using the Bayesian formula to output a total endpoint probability (i.e., the probability of predicting the overall film thickness of the wafer reaching the ideal thickness based on the collected spectral data). When the posterior probability exceeds a certain set threshold (e.g., 95%), an endpoint signal is triggered, indicating that the process termination conditions have been met and the process can be stopped. Furthermore, in outputting the total endpoint probability, the spatial location information of the sampling points corresponding to different sampling channels can also be considered. For example, sampling points at different locations such as the center, edge, and four corners of the wafer tray will affect the endpoint probability due to spatial location. This effectively addresses the inconsistency of channel signals caused by uneven spatial distribution of film thickness. By statistically inferring the measured values ​​at different locations, a more robust and consistent determination of the global endpoint state is achieved, significantly improving the system's consistent response capability to spatial etching endpoints.

[0075] Furthermore, taking the Bayesian fusion decision-making strategy as an example, its corresponding endpoint determination process is as follows: Figure 5 As shown. It consists of three steps:

[0076] 1) Define the prior probability distribution of film thickness: , This represents the prior mean of the film thickness at the current time. This represents the prior uncertainty (variance) of the film thickness; the parameters can be dynamically updated based on historical processes or linear / exponential trends. Taking etching as an example, the final state of etching is determined by the film thickness:

[0077] ;

[0078] Where T represents the etching endpoint state, d represents the film thickness at a certain sampling point on the wafer, and dth represents the ideal film thickness set by the semiconductor process (i.e., the film thickness threshold). If the measured film thickness is lower than the film thickness threshold, the process endpoint is considered to have been reached.

[0079] The prior probability of the corresponding state is , which represents the probability of "the overall wafer film thickness reaching the endpoint" derived a priori from the film thickness.

[0080] 2) Observation modeling: For each sampling channel i, let its observation feature be S. i Assume the following likelihood model applies between "etching reaches its end point":

[0081] ;

[0082] in, This refers to the mean spectral characteristics of sampling channel i under film thickness T. The observation noise variance of sampling channel i

[0083] 3) Fusion determination of the target: Based on observations from all channels, estimate whether the endpoint has been reached, i.e., calculate the posterior probability:

[0084] ;

[0085] Where N is the number of sampling channels involved in the decision-making process.

[0086] Set the posterior decision threshold ,like: If the endpoint is reached, then the endpoint has been reached.

[0087] Therefore, to address the issue of spatial differences in film thickness caused by multiple IEP / OES spectral signals, various configurable endpoint determination strategies have been implemented. Thickness variation curve fitting can adapt to situations of asynchronous thickness changes and is robust to local anomalies. The weighting factor approach can adaptively adjust the channels of interest during nonlinear etching processes, improving the effectiveness of the sampled channel data. Bayesian fusion effectively handles the inconsistency of channel signals caused by uneven spatial distribution of film thickness. By statistically inferring spectral data from different sampling points, a more robust and consistent determination of the global endpoint state of the wafer is achieved, significantly improving the system's consistent response capability to spatial etching endpoints. This not only improves the spatial consistency of endpoint determination but also enhances the adaptability of endpoint detection to different processes.

[0088] Based on the solutions described in the foregoing embodiments, this application provides an endpoint detection method that can acquire spectral data of multiple sampling points through multiple sampling channels, perform feature extraction and fusion endpoint judgment, and effectively optimize the judgment error introduced by the feature differences of different sampling points.

[0089] like Figure 6 As shown, the method specifically includes:

[0090] Step S41: Configure pulse parameters and spectrometer integration parameters.

[0091] Specifically, the endpoint detection parameters first need to be pre-configured, including the spectrometer's integration parameters and pulse parameters; these pulse parameters include parameters related to the light source and optical switch. Understandably, the spectrometer's integration parameters can be directly sent to the spectrometer, which can then determine which integration parameters to use for calculation at a given time based on the pulse signal. The optical switch can determine, based on the pulse signal, which sampling channels' spectral data to acquire at which times; sampling channels correspond to sampling points, and different sampling points correspond to different spatial locations on the wafer. The light source can be configured to be turned on or off at specific times based on the pulse signal.

[0092] Step S42: Receive the spectral data of each sampling point and perform preprocessing.

[0093] The spectrometer can then receive the raw spectral data from the optical switch and perform preprocessing tasks such as filtering, baseline correction, background subtraction, and normalization.

[0094] Step S43: Extract features from the spectral data of multiple sampling points.

[0095] It should be noted that algorithms such as principal component analysis, fringe counting, and model fitting can be used to extract features from the preprocessed spectral data; specifically, these can include feature data such as plasma excitation intensity, film thickness, film thickness variation rate, and data signal-to-noise ratio. It can be seen that feature extraction can obtain spectral feature data reflecting the characteristics of the wafer film, such as plasma excitation intensity, as well as feature data reflecting the inherent characteristics of the spectral data itself, such as data signal-to-noise ratio.

[0096] Step S44: Based on the endpoint determination strategy, fuse multiple sampling points to determine the endpoint.

[0097] In this embodiment, after obtaining the feature data of the spectral data of multiple sampling points through the above steps, the feature data of multiple sampling points can be fused and used for endpoint determination according to a pre-set endpoint determination strategy. Fusion determination of the feature data of multiple sampling points can improve the accuracy of the final endpoint determination. Furthermore, the endpoint determination strategy here can be one or more combinations of the various fusion determination strategies mentioned in the above embodiments. It should be noted that this solution can also be used in processes related to nanoscale high-precision thin film processing, where signal acquisition and fusion determination at multiple sampling points can improve the accuracy of endpoint detection.

[0098] Step S45: Adjust the endpoint detection parameters based on feature differences and signal-to-noise ratio.

[0099] Furthermore, if the film thickness of the wafer determined by the fusion of multiple sampling points does not reach the film thickness corresponding to the endpoint, the endpoint detection parameters (such as adjusting the sampling points and sampling duration) can be adjusted according to the differences in characteristic data (such as signal-to-noise ratio differences) at different sampling points; then, spectral signal sampling is performed again. In some embodiments, sampling points can be sparsified in regions with small feature differences and denser in regions with large feature differences, specifically as follows: Figure 7 As shown, the main etching apparatus includes a wafer tray and acquires spectral data using interferometry (IEP). Four sampling points (A, B, C, and D) are set, which can be adjusted from ABD to ACD as needed. The sampling time is shortened for sampling points with high signal-to-noise ratio (SNR) and increased for sampling points with low SNR, as detailed below. Figure 8 As shown.

[0100] Step S46: Stop the etching / deposition process.

[0101] Correspondingly, if the film thickness of the wafer reaches the endpoint when the fusion of multiple sampling points determines the film thickness, it means that the wafer film thickness has reached the endpoint, the process meets the termination conditions, and the current etching / deposition process can be stopped.

[0102] Therefore, this solution, by precisely controlling the switching of multiple sampling channels, allows multiple channels of signals to be acquired using the same spectrometer at the physical level. This reduces hardware costs, simplifies system structure, avoids calibration differences between different spectrometers, and improves the consistency of signals across channels. Furthermore, it enables endpoint detection based on spectral data from multiple sampling points, fully utilizing spectral information from different spatial locations to collaboratively determine the endpoint status, thus improving the accuracy and consistency of endpoint determination. Moreover, it can dynamically adjust the sampling points and integration time based on the changing trends and signal-to-noise ratio characteristics of the sampled signals; this prioritizes the acquisition of key information and balances sampling time, improving information utilization efficiency.

[0103] like Figure 9 As shown, this application also discloses an endpoint detection device applied to an endpoint detection system. The endpoint detection system includes a spectrometer and an optical switch. The input terminal of the optical switch is connected to multiple sampling channels corresponding to the wafer to be inspected, and the output terminal of the optical switch is connected to the spectrometer. The device includes:

[0104] The signal processing result acquisition module 11 is used to acquire the signal processing result from the spectrometer; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operations on the wafer to be detected;

[0105] The endpoint detection module 12 is used to detect the target endpoint based on the signal processing result and the sampling point of the target sampling channel, and obtain the endpoint detection result corresponding to the wafer to be tested; the target endpoint is the etching endpoint or the deposition endpoint.

[0106] As can be seen, this scheme can guide spectral signals from multiple sampling channels to the same spectrometer for processing through optical switches, realizing flexible control of multiple sampling channels, being compatible with spectral signals from different sampling channels, saving hardware resources, and ensuring the consistency of spectral response between different channels. Furthermore, by combining the signal processing results corresponding to the spectral signals and the spatial location information of the sampling points of the sampling channels, the endpoint status of different sampling points can be collaboratively judged, which can improve the accuracy and consistency of target endpoint judgment.

[0107] For a detailed description of the functions of each module in the endpoint detection device, please refer to the description of the relevant steps in any of the aforementioned method embodiments, which will not be repeated here.

[0108] This application also discloses an endpoint detection device, including a memory and a processor; wherein,

[0109] The memory is used to store computer programs;

[0110] The processor is configured to execute the computer program to implement the endpoint detection method as described in any of the method embodiments.

[0111] This application also discloses an endpoint detection system, comprising:

[0112] An optical switch connected to multiple sampling channels corresponding to the wafer under test at its input end is used to acquire spectral signals from the target sampling channel corresponding to the second parameter.

[0113] A spectrometer connected to the output of the optical switch is used to process the spectral signal received from the optical switch based on a first parameter to obtain a signal processing result;

[0114] The endpoint detection device, as described above, connected to the spectrometer, is used to acquire the signal processing results from the spectrometer, and to detect the target endpoint based on the signal processing results and the sampling point of the target sampling channel, thereby obtaining the endpoint detection result corresponding to the wafer to be tested; the target endpoint is an etching endpoint or a deposition endpoint.

[0115] The first parameter and the second parameter are endpoint detection parameters, which include pre-set parameters related to performing endpoint detection operations on the wafer to be inspected.

[0116] Furthermore, such as Figure 10 As shown, this application also discloses an example of an endpoint detection system, which connects the main process unit (including a wafer tray) and an etching / deposition control unit, and is suitable for endpoint detection in etching or deposition processes. Specifically, the IEP light source is used to provide the white / monochrome light source required for IEP endpoint detection; the optical switch can connect different sampling channels (such as...) at different times according to the triggered signal (a signal corresponding to the endpoint detection parameters used to configure the spectrometer, optical switch, and light source). Figure 1The system comprises n OES sampling channels and n IEP sampling channels to transmit the spectral signals corresponding to the emitted light from the plasma or the reflected light from the IEP source to the spectrometer. The IEP source and IEP sampling channels are connected to an optical switch via couplers to acquire the IEP spectral signal. It is understood that the OES / IEP signal can be provided by one or multiple channels (i.e., one or more sampling channels) as needed. When multiple channels are deployed, they can be used to measure the film thickness at different locations in different spaces or on different parts of the film. Furthermore, the endpoint detection parameters for the spectrometer can specifically be integration parameters, specifying which integration parameters are used to process the spectral signal at which times. The endpoint detection parameters for the optical switch can specifically be specifying which sampling channels are connected at which times. The endpoint detection parameters for the source can specifically be specifying when the source is turned on or off. Furthermore, the spectrometer can receive spectral signals transmitted by the optical switch and send the spectral results obtained from processing the spectral signals to the feature extraction and judgment unit. This feature extraction and judgment unit can receive the spectral results (OES and IEP) sent by the spectrometer and extract thickness / composition features. Then, based on the feature judgment logic, it can issue new endpoint detection parameters in real time, adjusting the sampling point and integration duration to obtain the optimal signal-to-noise ratio data for the most suitable point. The pulse triggering unit supports dynamically receiving pulse parameters (i.e., endpoint detection parameters) with synchronization timestamps from the feature extraction and judgment unit and generating various high-precision synchronization trigger signals for timing control between the light source, optical switch (also known as an optical path selector), and spectrometer. In addition to the endpoint detection system, it also includes: a semiconductor main process device: mainly composed of a chamber, radio frequency source, matching network, and wafer tray. If it is an etching device, it selectively removes material from the surface of the etched object through high-energy plasma bombardment; an etching / deposition control unit: acquires feature information such as thickness fed back by the feature judgment unit, and precisely controls the temperature and plasma concentration in different areas.

[0117] Furthermore, such as Figure 11 As shown, this application also discloses a specific endpoint detection system; wherein, a pulse triggering unit controls an optical switch and a spectrometer to acquire OES or IEP spectral signals from multiple sampling points; different sampling points correspond to different spatial locations of the wafer to be detected, such as the wafer center, edge, four corners, or a certain position on the side of the main process unit where the wafer is located; all OES or IEP signals are fed into the optical switch through a sampling channel (specifically, an optical fiber); the optical switch aggregates and sends the spectral signals to a spectrometer, which integrates the received spectral signals according to pre-configured integration parameters, and finally outputs the integration result to a feature extraction and judgment unit (i.e., Figure 1The processing unit (in the process unit) pre-configures corresponding integration parameters for each sampling point. Furthermore, the pulse triggering unit receives pulse parameters from the feature extraction and judgment unit in real time. These pulse parameters are the endpoint detection parameters, which may include the endpoint detection parameters corresponding to a large polling cycle, the order of sampling points for each OES or IEP polled within a cycle, and the polling time for each sampling point. The pulse triggering unit periodically outputs pulse signals (IEP light source switch signal, optical switch channel switching trigger signal, spectrometer integration trigger signal) to the IEP light source, optical switch, and spectrometer according to the polling cycle time to control the acquisition of spectral signals at specific sampling points and the corresponding acquisition time. The pulse triggering unit can use ASIC (Application-Specific Integrated Circuit) or FPGA (Field Programmable Gate Array) hardware to implement precise timing trigger pulse signals, thereby polling and acquiring multiple OES and IEP signals in a set order and precisely controlling the acquisition time of each signal. Figure 12 The diagram shows the specific multi-channel sampling timing. The sampling channel (corresponding sampling point) is selected according to the spectral sampling method (OES or IEP), and the light source is configured to turn on and off. At the same time, the integration parameters of the spectrometer are triggered to achieve precise control of the multi-channel spectral signals.

[0118] Furthermore, this application also discloses a semiconductor manufacturing apparatus, including: a process chamber where a wafer to be inspected is located, and an endpoint detection system as described above connected to the process chamber via multiple sampling channels.

[0119] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the endpoint detection method described above.

[0120] Furthermore, this application also discloses a computer program product, including a computer program / instructions that, when executed by a processor, implement the endpoint detection method as described above.

[0121] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0122] Each of the above modules or units can be implemented through software, hardware, or a combination of both. "Implemented through software" means that the processor reads and executes program instructions stored in memory to achieve the functions corresponding to the above modules or units. Here, the processor refers to a processing circuit capable of executing program instructions, including but not limited to at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor, etc., and other processing circuits capable of running program instructions. In some embodiments, the processor may also include circuits with other processing functions (such as hardware circuits for hardware acceleration, bus and interface circuits, etc.). Processors can be presented as integrated chips, for example, as integrated chips whose processing functions only include executing software instructions, or they can be presented as SoCs (system on a chip), that is, on a single chip, in addition to the processing circuits (usually called "cores") that can run program instructions, there are also other hardware circuits for implementing specific functions (of course, these hardware circuits can also be implemented separately based on ASICs or FPGAs). Correspondingly, in addition to executing software instructions, the processing functions can also include various hardware acceleration functions (such as AI calculations, encoding and decoding, compression and decompression, etc.).

[0123] In this application, "implemented in hardware" means that the functions of the above-mentioned modules or units are implemented through hardware processing circuits that do not have program instruction processing capabilities. These hardware processing circuits can be composed of discrete hardware components or integrated circuits. To reduce power consumption and size, integrated circuits are typically used. Hardware processing circuits can include ASICs (application-specific integrated circuits) or PLDs (programmable logic devices); PLDs can include FPGAs (field-programmable gate arrays), CPLDs (complex programmable logic devices), and so on. These hardware processing circuits can be a single packaged semiconductor chip (e.g., packaged as an ASIC); or they can be integrated with other circuits (e.g., CPUs, DSPs) and packaged into a single semiconductor chip. For example, multiple hardware circuits and a CPU can be formed on a silicon substrate and packaged into a single chip; this type of chip is also called a SoC. Alternatively, circuits for implementing FPGA functions and a CPU can be formed on a silicon substrate and encapsulated into a single chip; this type of chip is also called a SoPC (system on a programmable chip).

[0124] It should be noted that when this application is implemented through software, hardware, or a combination of both, different software or hardware can be used, and it is not limited to using only one type of software or hardware. For example, one module or unit can be implemented using a CPU, while another module or unit can be implemented using a DSP. Similarly, when implemented using hardware, one module or unit can be implemented using an ASIC, while another module or unit can be implemented using an FPGA. Of course, it is not limited to using the same software (e.g., all through a CPU) or the same hardware (e.g., all through an ASIC) to implement some or all modules or units. Furthermore, those skilled in the art will understand that software is generally more flexible but less performant than hardware, while hardware is the opposite. Therefore, those skilled in the art can choose software, hardware, or a combination of both based on actual needs.

[0125] The foregoing preferred embodiments have further illustrated the objectives, technical solutions, and advantages of the present invention. It should be understood that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An endpoint detection method, characterized in that, An endpoint detection system is applied, comprising a spectrometer and an optical switch, wherein the input terminal of the optical switch is connected to multiple sampling channels corresponding to the wafer under test, and the output terminal of the optical switch is connected to the spectrometer; the method includes: The signal processing result from the spectrometer is obtained; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operation on the wafer to be detected; Based on the signal processing results and the sampling points of the target sampling channel, the target endpoint is detected to obtain the endpoint detection result corresponding to the wafer to be tested; the target endpoint is the etching endpoint or the deposition endpoint.

2. The endpoint detection method according to claim 1, characterized in that, The endpoint detection parameters also include a third parameter characterizing spectral sampling using optical emission spectroscopy; therefore, the method further includes: If the endpoint detection system is not equipped with a light source, then the target configuration operation is performed; If the endpoint detection system is equipped with a light source, then the light source is configured to be turned off, and the target configuration operation is performed. The target configuration operation includes performing an integration parameter configuration operation on the spectrometer based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter. The target sampling channel is the sampling channel that corresponds to the second parameter among one or more sampling channels of the optical emission spectroscopy method.

3. The endpoint detection method according to claim 1, characterized in that, The endpoint detection parameters also include a fourth parameter characterizing spectral sampling using interferometry; therefore, the method further includes: If the endpoint detection system is equipped with a light source, then the light source is configured to be turned on, and the target configuration operation is performed; The target configuration operation includes performing an integration parameter configuration operation on the spectrometer based on the first parameter, and configuring the optical switch to be connected to the target sampling channel based on the second parameter. The target sampling channel is the sampling channel corresponding to the second parameter among one or more sampling channels of the interferometry method.

4. The endpoint detection method according to claim 2 or 3, characterized in that, The signal processing result is obtained by the spectrometer integrating the spectral signal received from the optical switch using the first parameter.

5. The endpoint detection method according to any one of claims 1 to 4, characterized in that, The step of detecting the target endpoint based on the signal processing result and the sampling point position of the target sampling channel to obtain the endpoint detection result corresponding to the wafer to be detected includes: Feature extraction is performed on the signal processing results to obtain spectral feature data corresponding to the sampling points of each target sampling channel; the spectral feature data is used to reflect the spectral characteristics of the film layer of the wafer to be tested; Based on the spectral feature data and the spatial location information of each sampling point, the film thickness of the wafer to be tested is detected at the target endpoint, and the endpoint detection result corresponding to the wafer to be tested is obtained.

6. The endpoint detection method according to claim 5, characterized in that, The step of detecting the film thickness of the wafer to be inspected based on the spectral feature data and the spatial location information of each sampling point includes: The thickness variation curves of the film thickness characterized by each of the spectral feature data are fitted to obtain the thickness variation fitting curves corresponding to each of the spectral feature data. Determine the target time corresponding to the target thickness in each of the thickness change fitting curves; the target thickness is the film thickness corresponding to the target endpoint. Based on the proportion of each target time within a preset time window and the spatial location information of each sampling point, the endpoint detection result corresponding to the wafer to be detected is generated.

7. The endpoint detection method according to claim 5, characterized in that, The step of detecting the film thickness of the wafer to be inspected based on the spectral feature data and the spatial location information of each sampling point includes: Weighting factors are assigned to each of the spectral feature data based on the target signal parameters of the spectral signal; the target signal parameters include signal-to-noise ratio and / or signal rate of change. The weighting factor is used to perform a weighted calculation on the film thickness variation characterized by the spectral feature data; Based on the matching relationship between the weighted calculated film thickness and the film thickness corresponding to the target endpoint, as well as the spatial location information of each sampling point, the endpoint detection result corresponding to the wafer to be tested is generated.

8. The endpoint detection method according to claim 5, characterized in that, The step of detecting the film thickness of the wafer to be inspected based on the spectral feature data and the spatial location information of each sampling point includes: The spectral feature data and the spatial location information of each sampling point are processed based on the Bayesian fusion method to obtain the target probability result; the target probability result is the probability that the film thickness corresponding to the wafer to be detected matches the film thickness corresponding to the target endpoint. Based on the relationship between the target probability result and the preset probability threshold, the endpoint detection result corresponding to the wafer to be detected is generated.

9. The endpoint detection method according to any one of claims 1 to 8, characterized in that, Also includes: Determine the endpoint detection parameters corresponding to the current detection cycle; The endpoint detection parameters include a synchronization timestamp, which is used to synchronize the spectrometer and the optical switch.

10. The endpoint detection method according to claim 9, characterized in that, Determining the endpoint detection parameters corresponding to the current detection cycle includes: If the endpoint detection result of the previous detection cycle indicates that the film thickness of the wafer to be tested does not match the film thickness corresponding to the target endpoint, then at the beginning of the current detection cycle, the endpoint detection parameters corresponding to the previous detection cycle are adjusted based on the endpoint detection result, and the endpoint detection parameters corresponding to the current detection period are determined based on the adjusted parameters and the synchronization timestamp.

11. The endpoint detection method according to any one of claims 1 to 10, characterized in that, Also includes: Within the current detection cycle, the endpoint detection parameters are updated based on the signal processing results, so as to trigger the next endpoint detection operation using the updated endpoint detection parameters.

12. An endpoint detection device, characterized in that, An endpoint detection system is applied, comprising a spectrometer and an optical switch, wherein the input terminal of the optical switch is connected to multiple sampling channels corresponding to the wafer under test, and the output terminal of the optical switch is connected to the spectrometer. The device includes: The signal processing result acquisition module is used to acquire the signal processing result from the spectrometer; the signal processing result is the result obtained by the spectrometer processing the spectral signal received from the optical switch based on the first parameter, and the spectral signal is obtained by the optical switch from the target sampling channel corresponding to the second parameter; wherein, the first parameter and the second parameter are endpoint detection parameters, and the endpoint detection parameters include preset parameters related to performing endpoint detection operations on the wafer to be detected; The endpoint detection module is used to detect the target endpoint based on the signal processing result and the sampling point of the target sampling channel, and obtain the endpoint detection result corresponding to the wafer to be tested; the target endpoint is the etching endpoint or the deposition endpoint.

13. An endpoint detection device, characterized in that, Includes memory and processor; among which, The memory is used to store computer programs; The processor is configured to execute the computer program to implement the endpoint detection method as described in any one of claims 1 to 11.

14. An endpoint detection system, characterized in that, include: An optical switch connected to multiple sampling channels corresponding to the wafer under test at its input end is used to acquire spectral signals from the target sampling channel corresponding to the second parameter. A spectrometer connected to the output of the optical switch is used to process the spectral signal received from the optical switch based on a first parameter to obtain a signal processing result; The endpoint detection device as described in claim 13, connected to the spectrometer, is used to acquire signal processing results from the spectrometer, detect the target endpoint based on the signal processing results and the sampling point position of the target sampling channel, and obtain the endpoint detection result corresponding to the wafer to be detected; the target endpoint is an etching endpoint or a deposition endpoint. The first parameter and the second parameter are endpoint detection parameters, which include pre-set parameters related to performing endpoint detection operations on the wafer to be inspected.

15. A semiconductor manufacturing apparatus, characterized in that, include: The process chamber where the wafer to be tested is located, and the endpoint detection system as described in claim 14 connected to the process chamber via multiple sampling channels.

16. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the endpoint detection method as described in any one of claims 1 to 11.

Images