Plasma etching process detection method and device, storage medium and equipment

Abstract

The invention discloses a plasma etching process detection method and device, a storage medium and equipment. The method comprises the steps that in the plasma etching process of a first semiconductor structure, the light intensity of gas phase particles in an excited state in a plasma etching cavity is collected, and corresponding optical emission spectrum (OES) data are obtained; and based on the OES data, utilizing a preset machine learning model to obtain a prediction result of the current plasma etching process. By adopting the scheme, the detection efficiency of the plasma etching process can be improved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and particularly to a method, device, storage medium, and equipment for detecting a plasma etching process. Background Art

[0002] In the process of manufacturing semiconductor devices, liquid crystal displays (LCDs), light-emitting diodes (LEDs), and some photovoltaic devices (PVs), a plasma etching process is usually used in combination with photolithography.

[0003] Generally, before performing the plasma etching process, a layer of photoresist is coated on the surface of the substrate, and the pattern to be etched is transferred to the photoresist through a photolithography process. After that, through a development process, the unexposed photoresist is removed, leaving the required pattern. During the plasma etching process, the processed substrate is exposed to the plasma, and the plasma will react with the surface of the substrate, thereby selectively etching the pattern on the substrate. Finally, the photoresist is removed.

[0004] In the plasma etching process, in order to ensure the etching accuracy and quality, improve the process stability and product yield, it is necessary to detect the etching process to obtain the etching effect in real time.

[0005] However, the existing methods for detecting the plasma etching process have a slow detection efficiency. Summary of the Invention

[0006] The problem to be solved by the present invention is: how to improve the detection efficiency of the plasma etching process.

[0007] To solve the above problem, an embodiment of the present invention provides a method for detecting a plasma etching process, the method including:

[0008] During the plasma etching process of the first semiconductor structure, collecting the light intensity of the gas-phase particles in the excited state in the plasma etching chamber and obtaining the corresponding optical emission spectroscopy (OES) data;

[0009] Based on the OES data, using a preset machine learning model, obtaining a prediction result of the current plasma etching process.

[0010] In a possible embodiment, the obtaining a prediction result of the plasma etching process based on the OES data includes:

[0011] Selecting the OES data of the chemical components with a correlation coefficient greater than a correlation coefficient threshold between the OES data and the plasma etching process;

[0012] Input the OES data of the chemical components whose correlation coefficients are greater than the correlation coefficient threshold into a preset machine learning model to obtain the prediction result of the current plasma etching process.

[0013] In a possible embodiment, the preset machine learning model is obtained by training an initial machine learning model using the OES data of multiple etching samples and the etching process data corresponding to each of the etching samples.

[0014] In a possible embodiment, the preset machine learning model is trained by the following method:

[0015] Obtain the OES data of multiple etching samples and the etching process data corresponding to each of the etching samples;

[0016] Analyze the OES data of the obtained etching samples and the etching process data corresponding to each of the etching samples to obtain the OES data of the chemical components whose correlation coefficients with the etching process are greater than the preset correlation coefficient threshold;

[0017] Use the OES data of the chemical components whose correlation coefficients with the etching process are greater than the preset correlation coefficient threshold as training samples to train the initial machine learning model to obtain the preset machine learning model.

[0018] In a possible embodiment, the preset machine learning model corresponds one-to-one with the semiconductor structure of the plasma etching.

[0019] In a possible embodiment, the prediction result of the current plasma etching process includes at least one of the depth, width, and angle of the current plasma etching.

[0020] In a possible embodiment, the method further includes: outputting the current morphology information of the first semiconductor structure based on the prediction result of the current plasma etching process.

[0021] An embodiment of the present invention further provides a method for forming a semiconductor structure, the method including:

[0022] Provide a first semiconductor structure;

[0023] Perform a plasma etching operation on the first semiconductor structure;

[0024] Use any of the above plasma etching process detection methods to detect the plasma etching process of the first semiconductor structure to obtain the prediction result of the plasma etching process of the first semiconductor structure;

[0025] Control the plasma etching process based on the prediction result of the current plasma etching process to obtain a target semiconductor structure.

[0026] An embodiment of the present invention also provides a plasma etching process detection device, which includes:

[0027] A data acquisition unit, adapted to collect the light intensity of gas-phase particles in an excited state in a plasma etching chamber during the plasma etching process of a first semiconductor structure and obtain corresponding optical emission spectroscopy (OES) data;

[0028] A prediction unit, adapted to obtain a prediction result of the current plasma etching process based on the OES data by using a preset machine learning model.

[0029] An embodiment of the present invention also provides a semiconductor structure etching system, which includes:

[0030] An etching unit, adapted to perform a plasma etching operation on a first semiconductor structure;

[0031] The above-mentioned plasma etching process detection device is adapted to detect the plasma etching process of the first semiconductor structure and obtain a prediction result of the plasma etching process of the first semiconductor structure;

[0032] A control unit, adapted to control the plasma etching process based on the prediction result of the current plasma etching process to obtain a target semiconductor structure.

[0033] Compared with the prior art, the technical solution of the embodiment of the present invention has the following advantages:

[0034] Applying the solution of the present invention, after obtaining the optical emission spectroscopy (OES) data of the gas in the plasma etching chamber, based on the obtained OES data, a preset machine learning model is used to obtain a prediction result of the current plasma etching process. Without performing other calculation operations, the OES data is directly input into the trained machine learning model, and the prediction result of the etching process can be obtained. Thus, the detection efficiency can be effectively improved. Description of the Drawings

[0035] Figure 1 and Figure 2 is a cross-sectional schematic diagram during the formation process of a semiconductor structure;

[0036] Figure 3 is a flowchart of the plasma etching process detection method in the embodiment of the present invention;

[0037] Figure 4 is a schematic diagram of OES data;

[0038] Figure 5 is a flowchart of a training method for a preset machine learning model in the embodiment of the present invention;

[0039] Figure 6 is a schematic cross-sectional view of a semiconductor structure;

[0040] Figure 7 is a flowchart of a method for forming a semiconductor structure in an embodiment of the present invention;

[0041] Figure 8 is a schematic structural diagram of a plasma etching process detection device in an embodiment of the present invention;

[0042] Figure 9 is a schematic structural diagram of a semiconductor structure etching system in an embodiment of the present invention. Detailed implementation manners

[0043] Figure 1 is a schematic cross-sectional structure diagram of a semiconductor structure. Refer to Figure 1 , the semiconductor structure may include: a substrate 101 and a mask layer 102 on the substrate, and an opening 102a is formed in the mask layer 102. Place the substrate 101 and the mask layer 102 on the substrate 101 in a plasma chamber, and use the mask layer 102 as a mask to etch the substrate 101 by using a plasma etching process. As Figure 2 shown, a groove 101a is formed in the substrate 101, and the mask layer 102 is removed to obtain a target semiconductor structure.

[0044] Currently, it is mainly through analyzing the chemical composition of the gas in the plasma processing chamber to infer whether the plasma etching process has etched to the substrate 101. Specifically, when etching the substrate 101 by using a plasma etching process, the chemical composition of the gas in the plasma processing chamber will be changed, and different chemical compositions have different spectral characteristics, and thus have different optical emission spectrum (OES) data. By monitoring the OES data of the gas in the plasma processing chamber and calculating the etching trend variable based on the monitored OES data, the etching process can be obtained based on the value of the etching trend variable.

[0045] To solve this problem, the present invention provides a plasma etching process detection method. By using this method, after obtaining the OES data of the gas in the plasma etching chamber, directly inputting the OES data into a trained machine learning model can obtain the prediction result of the etching process, thereby effectively improving the detection efficiency.

[0046] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the accompanying drawings.

[0047] Refer to Figure 3, an embodiment of the present invention provides a method for detecting a plasma etching process, and the method may include the following steps:

[0048] Step 31, during the plasma etching process of the first semiconductor structure, collect the light intensity of the gas-phase particles in the plasma in the excited state and obtain the corresponding optical emission spectroscopy (OES) data.

[0049] During the plasma etching process, the plasma will react with the first semiconductor structure, thereby changing the chemical composition in the plasma etching chamber. An OES sensor can be set in the plasma etching system, and the OES sensor has a wide spectrum that quickly spans the ultraviolet (UV), visible light (VIS), and near-infrared (NIR) spectra. The OES sensor can collect the light intensity of the gas-phase particles in the plasma in the excited state, thereby obtaining the OES data of the atoms or molecules in the plasma in the excited state. The gas-phase particles in the excited state include, for example, argon (Ar) atoms, oxygen atoms (O), chlorine atoms (Cl), fluorine atoms (F), etc. The OES data of the atoms or molecules in the plasma in the excited state, that is, the values of the light intensity of the gas-phase particles in the excited state at different times, are as Figure 4 shown.

[0050] In a specific implementation, there may be multiple gas-phase particles in the plasma etching chamber in the excited state. The light intensity of all the gas-phase particles in the excited state can be collected for subsequent etching process detection. Moreover, the light intensity of the same gas-phase particle will change with the increase of time. When performing the etching process detection, the etching process detection can be performed based on all the OES data obtained for each gas-phase particle at the current moment and before the current moment. Thus, real-time etching process detection can be achieved without waiting for the end of the etching process or reaching a certain duration to perform the etching detection.

[0051] Step 32, based on the OES data, use a preset machine learning model to obtain a prediction result of the current plasma etching process.

[0052] In a specific implementation, the preset machine learning model is a model architecture established with the OES data as the independent variable and the etching process as the dependent variable. Specifically, machine learning algorithms such as random forest and neural network can be used to implement it, and there is no limitation here. That is, after inputting the OES data into the preset machine learning model, the preset machine learning model can directly output a prediction result indicating the etching process.

[0053] In a specific implementation, a plasma etching machine tool can be set with multiple preset machine learning models, and each preset machine learning model corresponds to a different semiconductor structure. The same preset machine learning model can be only used for predicting the etching process of the same semiconductor structure. At this time, the preset machine learning model corresponds one-to-one with the semiconductor structure of the plasma etching. After determining the semiconductor structure of the plasma etching, the corresponding preset machine learning model can be directly used for predicting the etching process.

[0054] In a specific implementation, when obtaining a prediction result of the plasma etching process based on the OES data by using a preset machine learning model, the OES data of chemical components with a correlation coefficient greater than a correlation coefficient threshold with the plasma etching process can be first selected from the OES data, and then the OES data of the chemical components with a correlation coefficient greater than the correlation coefficient threshold is input into the preset machine learning model to obtain a prediction result of the current plasma etching process.

[0055] That is to say, after obtaining the OES data of all excited gaseous particles in the plasma etching chamber, the OES data with less influence on the plasma etching process can be removed from all the OES data, and then the OES data with greater influence on the plasma etching process is used for predicting the etching process.

[0056] In a specific implementation, the influence degree of the OES data of various gaseous particles on the plasma etching process can be analyzed by independent component analysis (ICA) and principal component analysis (Kernel PCA) methods on the OES data of various gaseous particles in advance, so as to determine the correlation coefficient between the OES data of various gaseous particles and the plasma etching process. In this way, in practical applications, after obtaining the OES data of all excited gaseous particles in the plasma etching chamber, the OES data with less influence on the plasma etching process can be directly removed to obtain the OES data that can be input into the preset machine learning model, thereby further improving the detection efficiency.

[0057] In a specific implementation, the prediction results obtained by using the preset machine learning model include, but are not limited to, at least one of the depth, width and angle of the plasma etching. For example, by using the preset machine learning model, only the depth value of the current plasma etching can be obtained, or only the width value of the current plasma etching or the angle value of the current plasma etching can be obtained, or the depth value and width value, depth value and angle value, width value and angle value of the current plasma etching or the depth value, width value and angle value can be obtained.

[0058] Figure 5 This is a flowchart of the training method of the preset machine learning model in an embodiment of the present invention. Refer to Figure 5, the training method may include the following steps:

[0059] Step 51, obtain the OES data of multiple etching samples and the etching process data corresponding to each of the etching samples.

[0060] In a specific implementation, real data in an existing plasma etching process can be obtained, including the OES data and the etching process data of each etching sample in the existing plasma etching process. When the finally obtained preset machine learning model is used to predict the etching process of a semiconductor structure, the etching samples obtained at this time should be etching samples for etching the same semiconductor structure. For example, they are all etching samples for the first semiconductor structure. The etching process data at this time is the data directly reflecting the etching process, such as the etching depth, width, and angle of the semiconductor structure at different moments during the plasma etching process.

[0061] Step 52, analyze the OES data of the obtained etching samples and the etching process data corresponding to each of the etching samples to obtain the OES data of chemical components whose correlation coefficient with the etching process is greater than a preset correlation coefficient threshold.

[0062] In a specific implementation, methods such as independent component analysis (ICA) and principal component analysis (Kernel PCA) can be used to analyze the OES data of the obtained etching samples and the etching process data corresponding to each of the etching samples, determine the influence degree of the OES data of various gas-phase particles on the plasma etching process, and then obtain the correlation coefficient between the OES data of various gas-phase particles and the plasma etching process, so as to obtain the chemical components whose correlation coefficient with the etching process is greater than the preset correlation coefficient threshold and the corresponding OES data.

[0063] Step 53, use the OES data of chemical components whose correlation coefficient with the etching process is greater than a preset correlation coefficient threshold as training samples to train an initial machine learning model to obtain a preset machine learning model.

[0064] In a specific implementation, only the OES data of chemical components whose correlation coefficient with the etching process is greater than a preset correlation coefficient threshold needs to be input into the initial machine learning model and the initial machine learning model is trained to obtain a preset machine learning model, which can reduce the model training difficulty and improve the model training efficiency.

[0065] In a specific implementation, the preset machine learning model is a model architecture established with OES data as the independent variable and the etching process as the dependent variable. In an embodiment of the present invention, in order to improve the prediction accuracy and model robustness, a multi-level machine learning strategy can be adopted. Specifically, first, the Gradient Boosting algorithm is used to process the time series characteristics of OES data. The Gradient Boosting algorithm can better capture the time series characteristics of the spectrum in OES data through its unique ordinal encoding method. At the same time, the Light Gradient Boosting Machine (Light GBM) algorithm is introduced as a parallel predictor, and its gradient-based unilateral sampling technique is used to improve the sensitivity of the model to subtle spectral changes in OES data. The Gradient Boosting algorithm and the Light GBM algorithm are optimized and combined through an ensemble learning framework, which not only improves the prediction accuracy but also enhances the adaptability of the model to process fluctuations. Finally, the model can directly output the prediction result indicating the etching process, providing a reliable basis for real-time monitoring and control.

[0066] In some embodiments, referring to Figure 3 , the method may further include: outputting the current morphology information of the first semiconductor structure based on the prediction result of the current plasma etching process.

[0067] That is to say, subsequently, after obtaining the plasma etching process data, the plasma etching machine can also obtain the current morphology of the first semiconductor structure based on the obtained plasma etching process data.

[0068] For example, referring to Figure 6 , the first semiconductor structure may include a substrate 501, a first oxide layer 502 on the substrate, a silicon nitride layer 503 on the first oxide layer, and a second oxide layer 504 on the silicon nitride layer 503. Through the plasma etching process, it is desired to form a first opening 505 and a second opening 506 in the substrate 501, the first oxide layer 502, the silicon nitride layer 503, and the second oxide layer 504. At this time, as the plasma etching process progresses, using the preset machine learning model, the etching depth 1 of the first opening 505 and the second opening 506 in the silicon nitride layer 503, the spacing 2 between the first opening 505 and the second opening 506 at the top of the substrate 501, the spacing 3 between the first opening 505 and the second opening 506 in the middle of the substrate 501, the spacing 4 between the first opening 505 and the second opening 506 at the bottom of the substrate 501, and the depth of the first opening 505 and the second opening 506 in the substrate 501 can be obtained. When the morphology of the current semiconductor structure does not meet the expectations, or when forming the target semiconductor structure, the etching process can be stopped in time based on the morphology of the current semiconductor structure to prevent over-etching or reduce etching losses.

[0069] In some embodiments, the current topography information of the first semiconductor structure can also be output in a graphical manner. For example, it can directly output Figure 6 a cross-sectional view of the semiconductor structure in Figure 6 , and mark the dimensional information of each position in this cross-sectional view, so that users can more intuitively understand the current etching process and the current topography of the semiconductor structure, which is conducive to the automatic control of plasma etching.

[0070] By using the plasma etching process detection method in the embodiments of the present invention, a preset machine learning model can be utilized to more quickly detect the etching process, thereby facilitating the control of the etching progress. Moreover, based on the prediction result of the etching process, the topography data of the semiconductor structure can be output, thus facilitating the automatic control of plasma etching.

[0071] Referring to Figure 7 , the embodiments of the present invention further provide a method for forming a semiconductor structure, and the method may include the following steps:

[0072] Step 71, provide a first semiconductor structure.

[0073] In a specific implementation, the first semiconductor structure can be any semiconductor structure that needs to undergo a plasma etching process. A photolithography pattern is formed on the first semiconductor structure, and subsequent plasma etching operations can be performed on other parts of the first semiconductor structure using the photolithography pattern.

[0074] Step 72, detect the plasma etching process of the first semiconductor structure to obtain a prediction result of the plasma etching process of the first semiconductor structure.

[0075] In a specific implementation, the plasma etching process detection method in the embodiments of the present invention can be adopted. While performing the plasma etching operation, the OES data of relevant gas-phase particles in the plasma etching chamber is collected and input into a preset machine learning model. The output of the preset machine learning model is the prediction result of the plasma etching process.

[0076] Step 73, control the plasma etching process based on the prediction result of the current plasma etching process to obtain a target semiconductor structure.

[0077] In a specific implementation, based on the prediction result of the plasma etching process, the current depth, width, and angle of the plasma etching can be determined. When the etching depth and width reach the target depth and target width, the etching process is stopped in a timely manner. When the etching angle does not meet the expectation, the etching angle can be adjusted in a timely manner to obtain a target semiconductor structure that meets the expectation.

[0078] In some embodiments, based on the prediction result of the plasma etching process, the topography of the semiconductor structure can be obtained, so that other key dimension information of the semiconductor structure (such as Figure 6 the thickness of the silicon nitride layer 503 in

[0079] can be obtained. These other key dimension information can facilitate the user to judge whether the target semiconductor structure can be formed when continuing the etching. When the topography of the current semiconductor structure does not meet the expectation, the etching process can be stopped in time to prevent over-etching or reduce etching loss.

[0080] For better understanding and implementation of the present invention by those skilled in the art, the corresponding apparatus, test system, electronic device and computer-readable storage medium of the above method are described in detail below. Figure 8 Referring to

[0081] the data acquisition unit 81 is adapted to collect the light intensity of the gas-phase particles in the excited state in the plasma etching chamber during the plasma etching process of the first semiconductor structure and obtain the corresponding optical emission spectroscopy (OES) data;

[0082] the prediction unit 82 is adapted to obtain the prediction result of the current plasma etching process based on the OES data by using a preset machine learning model.

[0083] Regarding the data acquisition unit 81 and the prediction unit 82, the implementation can refer to the description of the plasma etching process detection method above, and will not be repeated here.

[0084] Referring to Figure 9 the present invention embodiment also provides a semiconductor structure etching system, which can include: an etching unit, a plasma etching process detection device 92 and a control unit 93. Among them:

[0085] the etching unit is adapted to perform a plasma etching operation on the first semiconductor structure;

[0086] the plasma etching process detection device 92 is adapted to detect the plasma etching process of the first semiconductor structure and obtain the prediction result of the plasma etching process of the first semiconductor structure;

[0087] the control unit 93 is adapted to control the plasma etching process based on the prediction result of the current plasma etching process to obtain a target semiconductor structure.

[0088] In a specific implementation, referring toFigure 9 , the semiconductor structure etching system may further include: a plasma etching chamber 90. A support 901 is provided in the plasma etching chamber 90, and the first semiconductor structure (as Figure 1 shown) can be placed on the support 901 for etching. The etching unit may include: a gas injection component 911b and a plasma source 911a. The gas injection component 911b is connected to the plasma chamber 90 and injects an ionizable gas into the plasma chamber 90. The plasma source 911a can ionize the injected gas by applying radio frequency power or the like to the injected gas, thereby forming a plasma. The formed plasma can be used to etch the first semiconductor structure 100.

[0089] The plasma etching process detection device 92 can measure the light intensity of the gas-phase particles in the excited state in the plasma chamber 90 during the plasma etching process in real time, thereby obtaining the OES data of the gas-phase particles in the excited state and inputting it into a preset machine learning model to obtain the prediction result of the current plasma etching process. The control unit 93 can control the plasma etching process based on the prediction result of the current plasma etching process, including stopping etching or continuing etching, etc.

[0090] It should be noted that by using the plasma etching process detection device 92 in the embodiments of the present invention, accurate prediction can be made for small-sized etching structures. Among them, small-sized etching structures refer to holes or trenches with an etching width or diameter of less than 100 nanometers. For example, in the existing process, the diameter of the contact hole can reach 80 nanometers to 120 nanometers, and the width of the isolation trench can even be made 30 nanometers to 50 nanometers. These small-sized etching structures pose great challenges to the manufacturing process. It is difficult to predict the etching process using traditional plasma etching process detection methods, while using the solution of the embodiments of the present invention, the etching process can be accurately predicted.

[0091] The embodiments of the present invention also provide a computer-readable storage medium, on which a computer program is stored. The computer program is executed by a processor to implement the steps of any of the above methods.

[0092] In specific implementations, the computer-readable storage medium may include: ROM, RAM, disk, or optical disc, etc.

[0093] The embodiments of the present invention also provide an electronic device. The electronic device includes a memory and a processor. A computer program capable of running on the processor is stored on the memory. When the processor runs the computer program, it executes the steps of any of the above methods.

[0094] Regarding each device and product described in the above embodiments, each module / unit included therein can be a software module / unit, a hardware module / unit, or can be partly a software module / unit and partly a hardware module / unit. For example, for each device and product applied to or integrated into a chip, each module / unit included therein can be implemented in the form of hardware such as circuits. Or, at least some of the modules / units can be implemented in the form of a software program that runs on a processor integrated inside the chip, and the remaining (if any) part of the modules / units can be implemented in the form of hardware such as circuits; for each device and product applied to or integrated into a chip module, each module / unit included therein can be implemented in the form of hardware such as circuits. Different modules / units can be located in the same component (such as a chip, a circuit module, etc.) or different components of the chip module. Or, at least some of the modules / units can be implemented in the form of a software program that runs on a processor integrated inside the chip module, and the remaining (if any) part of the modules / units can be implemented in the form of hardware such as circuits; for each device and product applied to or integrated into a terminal, each module / unit included therein can be implemented in the form of hardware such as circuits. Different modules / units can be located in the same component (such as a chip, a circuit module, etc.) or different components inside the terminal. Or, at least some of the modules / units can be implemented in the form of a software program that runs on a processor integrated inside the terminal, and the remaining (if any) part of the modules / units can be implemented in the form of hardware such as circuits.

[0095] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention should be subject to the scope defined by the claims.

Claims

1. A plasma etching process detection method, characterized in that: include: During the plasma etching process of the first semiconductor structure, light intensity of gas phase particles in an excited state in the plasma etching chamber is collected and corresponding optical emission spectrum (OES) data is obtained; Based on the OES data, a preset machine learning model is used to obtain a prediction result of the current plasma etching process.

2. The plasma etching process detection method according to claim 1, characterized in that: The method of obtaining a prediction result about the plasma etching process based on the OES data by using a preset machine learning model includes: Selecting OES data of chemical components having a correlation coefficient with the plasma etching process greater than a correlation coefficient threshold from the OES data; The OES data of the chemical components whose correlation coefficient is greater than the correlation coefficient threshold are input into a preset machine learning model to obtain a prediction result of the current plasma etching process.

3. The plasma etching process detection method according to claim 1, characterized in that: The preset machine learning model is obtained by training the initial machine learning model using the OES data of multiple etching samples and the etching process data corresponding to each of the etching samples.

4. The plasma etching process detection method according to claim 3, characterized in that: The preset machine learning model is trained using the following method: Obtaining OES data of a plurality of etching samples and etching process data corresponding to each of the etching samples; Analyze the acquired OES data of the etching samples and the etching process data corresponding to each of the etching samples to obtain OES data of chemical components having a correlation coefficient with the etching process greater than a preset correlation coefficient threshold; The OES data of chemical components whose correlation coefficient with the etching process is greater than a preset correlation coefficient threshold are used as training samples to train the initial machine learning model to obtain a preset machine learning model.

5. The plasma etching process detection method according to claim 4, characterized in that: The preset machine learning model corresponds one-to-one to the plasma-etched semiconductor structure.

6. The plasma etching process detection method according to claim 4, characterized in that: The predicted result of the current plasma etching process includes at least one of the depth, width and angle of the current plasma etching.

7. The plasma etching process detection method according to claim 1, characterized in that: Also includes: Based on the prediction result of the current plasma etching process, current morphology information of the first semiconductor structure is output.

8. A method for forming a semiconductor structure, characterized in that: include: providing a first semiconductor structure; performing a plasma etching operation on the first semiconductor structure; Using the plasma etching process detection method according to any one of claims 1 to 7 to detect the plasma etching process of the first semiconductor structure, and obtain a prediction result of the plasma etching process of the first semiconductor structure; Based on the predicted result of the current plasma etching process, the plasma etching process is controlled to obtain a target semiconductor structure.

9. A plasma etching process detection device, characterized in that: include: A data acquisition unit, adapted to collect the light intensity of gas phase particles in an excited state in a plasma etching chamber and obtain corresponding optical emission spectrum OES data during the plasma etching process of the first semiconductor structure; The prediction unit is adapted to obtain a prediction result of the current plasma etching process based on the OES data and using a preset machine learning model.

10. A semiconductor structure etching system, characterized in that: include: An etching unit adapted to perform a plasma etching operation on the first semiconductor structure ； The plasma etching process detection device of claim 9 is suitable for detecting the plasma etching process of the first semiconductor structure to obtain a prediction result of the plasma etching process of the first semiconductor structure; The control unit is adapted to control the plasma etching process based on the predicted result of the current plasma etching process to obtain a target semiconductor structure.

Images