Silicon carbide substrate grinding method and device, terminal and storage medium

Abstract

The invention provides a silicon carbide substrate grinding method and device, a terminal and a storage medium, and relates to the technical field of semiconductor material preparation. The method comprises the steps that in the grinding process of a target silicon carbide substrate, technological parameters of grinding equipment in a first grinding stage and substrate data of the target silicon carbide substrate after the first grinding stage is finished are obtained, and the first grinding stage is any stage in the grinding process of the target silicon carbide substrate; inputting the process parameters and the substrate data into the grinding path prediction model, and outputting the operation data of the grinding equipment; the operation data comprises a grinding path, grinding pressure, grinding rotating speed, grinding time and grinding rate of the grinding equipment in a second grinding stage, and the second grinding stage is the next stage of the first grinding stage; and controlling the grinding equipment to grind the target silicon carbide substrate after the first grinding stage according to the operation data in the second grinding stage. The surface flatness of the silicon carbide substrate can be remarkably improved, and the grinding efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of semiconductor material preparation technology, and in particular to a silicon carbide substrate polishing method, apparatus, terminal and storage medium. Background Technology

[0002] Silicon carbide (SiC) substrates have become a core material for power electronic devices and radio frequency devices due to their high thermal conductivity, wide bandgap, and high breakdown electric field strength.

[0003] In silicon carbide substrate fabrication, grinding / polishing is a critical process; however, existing grinding / polishing processes typically employ fixed paths. Due to the high hardness and thermal conductivity of the substrate material, the removal rate varies significantly across different regions, leading to stress concentration and uneven surface thickness, particularly at the center and edges. Traditional fixed paths struggle to balance pressure and uniform abrasive grain distribution, easily resulting in uneven polishing. Existing grinding / polishing processes often use fixed patterns (such as circular, linear, or reciprocating), resulting in low efficiency; high repetition coverage can lead to over-polishing or stress concentration in localized areas; and uneven abrasive grain distribution can cause inconsistent surface roughness. Summary of the Invention

[0004] This application provides a silicon carbide substrate polishing method, apparatus, terminal, and storage medium to solve the problems of low polishing efficiency and inconsistent surface roughness of silicon carbide substrates caused by uneven abrasive particle density distribution in existing polishing methods.

[0005] In a first aspect, this application provides a method for polishing a silicon carbide substrate, comprising: During the grinding process of the target silicon carbide substrate, the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage are obtained; the process parameters include grinding pressure, grinding speed and disk surface temperature, and the substrate data includes abrasive density, substrate thickness and substrate surface flatness, and the first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate. The process parameters and substrate data are input into a pre-built grinding path prediction model, and the operating data of the grinding equipment is output. The operating data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage, and the second grinding stage is the next grinding stage after the first grinding stage. The grinding equipment is controlled to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

[0006] Secondly, this application provides a silicon carbide substrate polishing apparatus, comprising: The data acquisition module is used to acquire the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage is completed during the grinding process of the target silicon carbide substrate. The process parameters include grinding pressure, grinding speed and disk temperature, and the substrate data includes abrasive density, substrate thickness and substrate surface flatness. The first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate. The data prediction module is used to input the process parameters and the substrate data into a pre-built grinding path prediction model and output the operating data of the grinding equipment. The operating data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage, where the second grinding stage is the next grinding stage after the first grinding stage. The grinding module is used to control the grinding equipment to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

[0007] Thirdly, this application provides a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect or any possible implementation of the first aspect above.

[0008] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in the first aspect or any possible implementation of the first aspect.

[0009] This application provides a silicon carbide substrate polishing method, apparatus, terminal, and storage medium. During the polishing process of the target silicon carbide substrate, the process parameters of the polishing equipment in the first polishing stage and the substrate data of the target silicon carbide substrate after the first polishing stage are acquired. The process parameters include polishing pressure, polishing speed, and disk surface temperature. The substrate data includes abrasive density, substrate thickness, and substrate surface flatness. The first polishing stage is any polishing stage in the polishing process of the target silicon carbide substrate. The process parameters and substrate data are input into a pre-constructed polishing path prediction model, which outputs the operating data of the polishing equipment. The operating data includes the polishing path, polishing pressure, polishing speed, polishing time, and polishing rate of the polishing equipment in the second polishing stage, which is the next polishing stage after the first polishing stage. The polishing equipment is controlled to perform the second polishing stage on the target silicon carbide substrate after the first polishing stage according to the operating data. This application obtains the process parameters and substrate data of the first grinding stage and inputs them into the grinding path prediction model. Based on this real-time and comprehensive information, it can accurately predict the operating data required for the second grinding stage. This allows the grinding equipment to operate according to the optimal grinding path, pressure, speed, and other parameters, effectively reducing the microscopic undulations and defects on the substrate surface, thereby significantly improving the surface flatness of the silicon carbide substrate and meeting the substrate quality requirements of high-precision semiconductor device manufacturing. In addition, this application utilizes the grinding path prediction model to quickly and accurately predict the optimal operating data for the second grinding stage based on the actual data of the first grinding stage, avoiding repeated trials, greatly shortening the process development cycle, and improving production efficiency. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic flowchart of the silicon carbide substrate polishing method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the silicon carbide substrate polishing apparatus provided in the embodiments of this application; Figure 3 This is a schematic diagram of the terminal provided in the embodiments of this application. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0014] To address the problems of low grinding efficiency and inconsistent surface roughness of silicon carbide substrates due to uneven abrasive particle density distribution in existing grinding methods, this application proposes a silicon carbide substrate grinding method. The aim is to dynamically plan the optimal trajectory based on real-time data during the grinding process, thereby improving grinding uniformity and efficiency while reducing energy consumption and material waste. The concept involves introducing deep learning and reinforcement learning algorithms to achieve adaptive optimization of the grinding path, thereby improving grinding efficiency and surface smoothness, reducing surface defects, and increasing the substrate yield.

[0015] Figure 1 The implementation flowchart of the silicon carbide substrate polishing method provided in the embodiments of this application is described in detail below: In step 101, during the grinding process of the target silicon carbide substrate, the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage are acquired. The process parameters include grinding pressure, grinding speed and disk temperature, and the substrate data includes abrasive density, substrate thickness and substrate surface flatness. The first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate.

[0016] In this embodiment, when the target silicon carbide substrate needs to be polished, the process parameters of the polishing equipment during the first polishing stage and the substrate data of the target silicon carbide substrate after the first polishing stage are acquired in real time during the polishing process. The process parameters include the polishing pressure, polishing speed, and disk temperature (i.e., the temperature of the polishing disk surface of the polishing equipment), and the substrate data includes the abrasive density, substrate thickness, and substrate surface flatness of the target silicon carbide substrate.

[0017] In the embodiments of this application, the process parameters and substrate data of each grinding stage are acquired in real time, which can improve the grinding efficiency of the substrate.

[0018] In one possible implementation, obtaining substrate data of the target silicon carbide substrate after the completion of the first polishing stage may include: After the first grinding stage is completed, the surface image of the target silicon carbide substrate is acquired by an industrial camera installed on the grinding equipment, and the surface image is segmented and calculated using an image segmentation algorithm to obtain the number of abrasive grains per unit area on the surface image. Based on the number of abrasive grains per unit area, the abrasive grain density per unit area is calculated. The substrate thickness of the target silicon carbide substrate is acquired using a laser thickness gauge; the substrate thickness includes the thickness of the region corresponding to each unit area on the target silicon carbide substrate. The total thickness deviation of the target silicon carbide substrate is calculated using the maximum and minimum values ​​of the substrate thickness, and the surface flatness of the target silicon carbide substrate is determined using the total thickness deviation.

[0019] In this embodiment, the industrial camera can be a CCD camera.

[0020] Optionally, in each grinding stage, a CCD camera installed on the grinding equipment is used to collect surface images of the target silicon carbide substrate in real time after the grinding stage is completed. Then, the collected surface images are segmented using an image segmentation algorithm to statistically analyze the abrasive distribution in different parts of the target silicon carbide substrate. That is, the number of abrasive particles on the surface of the target silicon carbide substrate within a unit area after segmentation is calculated. The abrasive density within the unit area is obtained by calculating the ratio of the number of abrasive particles within the unit area to the unit area.

[0021] Then, a laser thickness gauge is used to monitor the substrate thickness of the target silicon carbide substrate in real time after each polishing stage. The substrate thickness after each polishing stage includes the substrate thickness of the region within each unit area on the target substrate.

[0022] After collecting the substrate thickness of the target silicon carbide substrate at the end of each grinding stage, the difference between the maximum and minimum substrate thickness at the end of each grinding stage is calculated as the Total Thickness Variation (TTV) for that grinding stage. This TTV value is used to evaluate the surface flatness of the target silicon carbide substrate at that grinding stage; a larger TTV value indicates poorer surface flatness. Accordingly, in this embodiment, the TTV value is also used to represent the removal rate of the target silicon carbide substrate.

[0023] In this embodiment, each grinding stage is divided according to a "time interval + thickness change dual triggering mechanism," specifically as follows: 1) Basic time interval: Determine a grinding stage based on the grinding time in the running data.

[0024] 2) Thickness change trigger: The grinding thickness of this grinding stage is determined based on the grinding time and grinding rate in the running data. When the grinding reaches the grinding thickness, the current grinding stage ends.

[0025] 3) Flatness Trigger: If the TTV value is high after a certain stage ends. If a preset deviation threshold is set, the process will proceed directly to the final grinding stage.

[0026] 4) Maximum duration limit: If the execution time of a certain grinding stage exceeds the preset maximum duration (e.g., 60 minutes) but does not reach the above-mentioned preset deviation threshold, the current grinding stage will be forcibly terminated, and the running data of the next grinding stage will be re-optimized through the grinding path prediction model.

[0027] In step 102, process parameters and substrate data are input into a pre-built grinding path prediction model, and the operating data of the grinding equipment is output. The operating data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage. The second grinding stage is the next grinding stage after the first grinding stage.

[0028] In this embodiment of the application, the process parameters and substrate data corresponding to the first grinding stage obtained in step 101 are input into a pre-built and trained grinding path prediction model to obtain the operating data of the grinding equipment in the second grinding stage. The operating data may include the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the second grinding stage, and the second grinding stage is the next grinding stage after the first grinding stage.

[0029] This application embodiment acquires the process parameters (grinding pressure, grinding speed, disk surface temperature) and substrate data (abrasive density, substrate thickness, substrate surface flatness) of the first grinding stage and inputs them into the grinding path prediction model. Based on this real-time and comprehensive information, it can accurately predict the operating data required for the second grinding stage. This allows the grinding equipment to operate according to the optimal grinding path, pressure, speed, and other parameters, effectively reducing microscopic undulations and defects on the substrate surface, thereby significantly improving the surface flatness of the silicon carbide substrate and meeting the substrate quality requirements of high-precision semiconductor device manufacturing.

[0030] In one possible implementation, the process of constructing the grinding path prediction model can be as follows: The process parameters of the grinding equipment for each grinding stage of the silicon carbide substrate sample during the grinding process, the substrate data of the silicon carbide substrate sample after each grinding stage, and the operation data of the grinding equipment for the next grinding stage corresponding to each grinding stage are obtained, and a training set is constructed. A grinding path prediction model is constructed by using convolutional neural networks and long short-term memory networks, combined with a reward function; A grinding path prediction model is trained based on the training set.

[0031] Optionally, a grinding path prediction model can be constructed by using a convolutional neural network (CNN) and a long short-term memory network (LSTM) and combining them with a reward function.

[0032] The convolutional layers in a convolutional neural network consist of three layers, with kernel sizes of [sizes to be filled in]. , , The corresponding numbers are 64, 128, and 256; the pooling layer uses max pooling, and the pooling kernel size is... The step size is 2; the activation function is the ReLU function.

[0033] The Long Short-Term Memory network has two hidden layers, each with 128 hidden units. The dropout rate is set to 0.2, and the activation function is the tanh function. The output layer is mapped to grinding operation data (grinding path, pressure, rotation speed, etc.) through a fully connected layer. The mapping relationship adopts a "multi-output regression model", and the output dimension corresponds one-to-one with the number of parameters in the operation data.

[0034] In addition, the reward function, as a feedback signal for reinforcement learning, is independent of the network loss function (MSE loss function). In this embodiment, the DQN algorithm is linked with the network parameter update. After each training round, the reward function value is calculated and fed back to the DQN experience replay pool. When updating the network parameters, the priority weights of the loss function gradient and the reward function feedback are combined to achieve the dual-objective optimization of "minimizing the fitting prediction error" and "optimizing the polishing effect".

[0035] Then, a training set is constructed, which involves acquiring the process parameters of the grinding equipment for each grinding stage of the silicon carbide substrate sample during the grinding process, as well as the substrate data of the silicon carbide substrate sample after each grinding stage. The operating data of the grinding equipment for the next grinding stage corresponding to each grinding stage is also acquired. The process parameters, substrate data, and operating data of the grinding equipment for the next grinding stage corresponding to each grinding stage are used as a training subset, and the training subsets of all grinding stages are combined into a training set.

[0036] For example, when the polishing process of silicon carbide substrate sample A has five stages, namely polishing stage one, polishing stage two, polishing stage three, polishing stage four, and polishing stage five, the process parameters 1, 2, 3, 4, and 5 of the polishing equipment in polishing stages one, two, three, four, and five are obtained respectively, as well as the substrate data 1, 2, 3, 4, and 5 of silicon carbide substrate sample A after the completion of polishing stages one, two, three, four, and five; and accordingly, the process parameters 1, 2, 3, 4, and 5 of the polishing equipment are obtained. The following data is used: 1) is the operating data of the grinding equipment before the start of the next grinding stage (i.e., before grinding stage 2); 2) is the operating data of the grinding equipment before the start of the next grinding stage (i.e., before grinding stage 3); 3) is the operating data of the grinding equipment before the start of the next grinding stage (i.e., before grinding stage 4); and 4) is the operating data of the grinding equipment before the start of the next grinding stage (i.e., before grinding stage 5). Since there is no next grinding stage after grinding stage 5, this stage can be used as the final reference stage. Accordingly, the operating data of the grinding equipment after grinding stage 5 is used as operating data 5. Then, process parameter 1, substrate data 1, and operating data 1 form training subset 1; process parameter 2, substrate data 2, and operating data 2 form training subset 2; process parameter 3, substrate data 3, and operating data 3 form training subset 3; process parameter 4, substrate data 4, and operating data 4 form training subset 4; and process parameter 5, substrate data 5, and operating data 5 form training subset 5. Finally, training subsets 1, 2, 3, 4, and 5 are combined to form the training set.

[0037] Finally, based on the training set mentioned above, a grinding path prediction model is trained. This involves using a convolutional neural network to extract features from the input process parameters and substrate data, with data processing strictly following the order of "process parameter standardization → substrate data normalization → feature matrix merging". Then, the extracted two-dimensional feature vectors are converted into one-dimensional temporal feature vectors in the convolutional neural network through "dimensional flattening → fully connected layer mapping", and then concatenated with the long short-term memory network in the manner of "channel dimension superposition" to ensure deep integration of spatial features and time series features.

[0038] The embodiments of this application utilize a grinding path prediction model to predict the operating data of the grinding equipment in the next grinding stage, thereby predicting the removal rate of silicon carbide substrates under different process paths.

[0039] In one possible implementation, after training the grinding path prediction model based on the training set, the method may further include: Obtain a first reward function value and a second reward function value. The first reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the target grinding stage, the substrate data of the silicon carbide substrate sample after the target grinding stage, and the operating data of the grinding equipment in the next grinding stage of the target grinding stage. The second reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the previous grinding stage of the target grinding stage, the substrate data of the silicon carbide substrate sample after the previous grinding stage of the target grinding stage, and the operating data of the grinding equipment in the target grinding stage. The target grinding stage is any grinding stage among the grinding stages of the silicon carbide substrate samples in the training set during the grinding process. Calculate the difference between the first reward function value and the second reward function value, and take the absolute value of the difference to obtain the target value; The substrate surface flatness in the substrate data of the silicon carbide substrate sample after the target grinding stage is taken as the first substrate surface flatness. Based on the target value and the flatness of the first substrate surface, the grinding path prediction model is deemed to have completed training.

[0040] Optionally, in the silicon carbide substrate polishing process, the embodiments of this application are most concerned with the uniformity of the substrate surface (i.e., minimal surface thickness deviation), maximization of removal rate, minimization of energy consumption (i.e., low motor power and polishing slurry consumption), and minimization of surface damage (i.e., avoiding local over-polishing and microcracks).

[0041] Based on the above, in the process of training the grinding path prediction model, the embodiments of this application also require model optimization judgments during the grinding stage. The process is as follows: The reward function values ​​of the grinding path prediction model for two adjacent grinding stages are obtained, namely the first reward function value of the target grinding stage and the second reward function value of the next grinding stage after the target grinding stage. Then, the difference between the first and second reward function values ​​is calculated, and the absolute value of the difference is taken to obtain the target value. The target grinding stage is any grinding stage among all grinding stages in the grinding process of the silicon carbide substrate sample in the training set.

[0042] The formula for calculating the reward function value is as follows:

[0043] in, For the target grinding stage The reward function value, For the target grinding stage Substrate surface flatness after completion. For energy consumption, , These are all weight coefficients, which can be 0.6 and 0.4 respectively, and their adjustment rule is: in the early stage of training (e.g., the first 100 iterations). Fixed at 0.6, The value is fixed at 0.4. After 100 rounds, it is dynamically adjusted based on the substrate surface flatness compliance rate. For every 10% increase in the compliance rate... Increase by 0.05 Reduce by 0.05 until... Reaching the maximum value of 0.8 and It reached the minimum value of 0.2.

[0044] For example, silicon carbide substrate sample A has five polishing stages, and the training set includes five training subsets. When the target polishing stage is polishing stage two, after training the polishing path prediction model using training subset 2, the first reward function value of the polishing path prediction model trained for polishing stage two is determined using the reward function value calculation formula. Correspondingly, after training the polishing path prediction model using training subset 1, the second reward function value of the polishing path prediction model trained for polishing stage one is determined using the reward function value calculation formula. Then, the absolute value of the difference between the first reward function value and the second reward function value is used as the target value for the target polishing stage.

[0045] After obtaining the target value, the substrate surface flatness in the substrate data of the silicon carbide substrate sample after the target grinding stage is obtained as the first substrate surface flatness.

[0046] For example, when the target grinding stage is grinding stage two, the first substrate surface flatness to be obtained is the substrate surface flatness in the substrate data of silicon carbide substrate sample A after the completion of grinding stage two.

[0047] Finally, the target value and the flatness of the first substrate surface are used to determine whether the grinding path prediction model has been trained successfully during the target grinding stage.

[0048] In one possible implementation, determining the completion of the grinding path prediction model training based on the target value and the flatness of the first substrate surface can include: Determine whether the target value is not greater than a preset threshold, and determine whether the surface flatness of the first substrate is not greater than a preset flatness. If the target value is not greater than the preset threshold and the flatness of the first substrate surface is not greater than the preset flatness, then the grinding path prediction model training is considered complete. If the target value is greater than the preset threshold, and / or the flatness of the first substrate surface is greater than the preset flatness, then it is determined that the training of the grinding path prediction model is not completed, and the step of training the grinding path prediction model based on the training set is returned to continue execution.

[0049] Optionally, determine the target value Is it not greater than the preset threshold? And the surface flatness of the first substrate Is it not greater than the preset flatness? ,Right now: like ,and, Then it is determined that the grinding path prediction model is in the target grinding stage. Training optimization was completed within the timeframe.

[0050] like , and / or Then it is determined that the grinding path prediction model is in the target grinding stage. The training optimization is not yet complete and needs to be continued using the training subset in the next grinding phase.

[0051] Among them, the preset threshold It can be set to 0.01, which indicates that the value of the reward function changes very little during continuous training, and the grinding path prediction model has become stable.

[0052] Preset flatness The range of values ​​can be set to The preferred value is The range of values ​​is determined based on the application requirements of silicon carbide substrates in power electronic devices and radio frequency devices. When the total thickness deviation of the substrate is within this range, it can meet the high precision requirements of the substrate surface flatness for subsequent device manufacturing, while taking into account grinding efficiency and production cost.

[0053] The embodiments of this application can also use a reinforcement learning algorithm (Deep Q-Network, DQN) to train the grinding path prediction model, specifically as follows: The thickness of silicon carbide substrate sample A is set to... Divided into The grid points and the initial thickness distribution matrix are as follows: This indicates the position of silicon carbide substrate sample A during the grinding process. The thickness distribution function at that location.

[0054] Define grinding path set ,in, It is a spiral path. For horizontal paths, It is a random path. Let be the position point of the grinding head of the grinding equipment on the grinding path in the planar coordinate system. and Let be a two-dimensional coordinate system on the substrate surface.

[0055] The grinding removal function is defined as:

[0056] in, For removal function, For grinding path type, This is the material removal function corresponding to the grinding path at this point, and its value is related to the grinding pressure, abrasive density, and grinding speed. , The range of values ​​for the correction factor can be set to... ; Grinding pressure, unit: , The pressure influence coefficient can be set to 0.8. Abrasive particle density, unit: , The coefficient representing the influence of abrasive particle density can be set to 0.3. Grinding speed, unit: , The value of the speed influence coefficient can be 0.5. , The position coordinate parameters are corrected by adjusting the position feature vector extracted by the convolutional neural network. The corrected formula is: ,in, The normalized value of the location feature is within the range of The substrate surface flatness TTV value was used as an evaluation index for uniformity.

[0057] By utilizing reinforcement learning algorithms and continuously optimizing the grinding path, the grinding effect is improved. Specifically, after 500 rounds of training, the standard deviation of the substrate thickness of silicon carbide substrate sample A after path optimization is improved from... Down to Uniformity improved by 51%.

[0058] The corresponding optimization process is as follows: initialization Value function ,in, For state, For the action (i.e., grinding path type). Set the initial thickness distribution. .

[0059] In each training round (i.e., the process of training using a subset of the training from each grinding phase), the current state is calculated. ,in, This represents the average thickness of the silicon carbide substrate sample. The path selection action is based on a greedy strategy. Update the thickness distribution (i.e., the grinding removal function) based on the path: Calculate the uniformity TTV value of the milled silicon carbide substrate sample, i.e. Calculate the reward function value, i.e. ,renew Value function The new state Input the data into the grinding path prediction model. Repeat the training until convergence, and output the optimal strategy. .

[0060] In addition, the reward function and The association between value functions is as follows:

[0061] in, The attenuation coefficient can be set to 0.9. To predict the reward function value, This represents the actual reward function value.

[0062] For example, a 4-inch silicon carbide substrate with a thickness of [missing information] is selected. A high-precision double-sided grinding machine is used, with the grinding pressure set to... The grinding speed is set to The polishing fluid used is diamond polishing fluid.

[0063] The control group used a fixed spiral path, while the experimental group used an optimized path obtained from the grinding path prediction model.

[0064] During the grinding process, every The new thickness distribution and reward function value of the silicon carbide substrate are output. Simultaneously, the surface of the silicon carbide substrate is scanned to obtain the abrasive grain distribution and thickness error map, and the path trajectory (spiral / straight / partitioned / custom) for the next grinding stage is selected based on the results. After 500 training rounds, the optimal path can be selected. Grinding stops when the thickness of the silicon carbide substrate meets the requirements. Experimental results are as follows: Table 1. Comparison of experimental results for different indicators between the control group and the experimental group.

[0065] The table above shows that the grinding path predicted by the grinding path prediction model given in the embodiments of this application can effectively improve the uniformity of grinding, while reducing energy consumption and increasing the removal rate.

[0066] This application embodiment updates and corrects the grinding path in real time based on the optimized grinding path prediction model, mainly including an adaptive spiral trajectory, a region-priority path based on abrasive particle density, and a dynamic avoidance path to prevent repeated coverage. Utilizing the grinding path prediction model enables intelligent grinding, breaking through the traditional fixed trajectory pattern, achieving adaptive path planning, reducing repeated grinding, lowering energy consumption, shortening processing time, improving surface uniformity and roughness, and reducing the probability of microcracks and stress concentration in silicon carbide substrates.

[0067] In step 103, the grinding equipment is controlled to perform a second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

[0068] In this embodiment, the grinding equipment is controlled to perform the second grinding stage using the predicted grinding equipment operation data output in step 102. The precise grinding parameter control in this embodiment allows for better regulation of the interaction between the abrasive grains and the substrate surface, resulting in more uniform and orderly cutting and friction of the abrasive grains on the substrate surface. This helps reduce the formation of surface roughness peaks, lowers surface roughness, provides a smoother substrate surface for subsequent semiconductor process steps, and improves device performance and reliability.

[0069] In one possible implementation, after the polishing equipment performs a second polishing stage on the target silicon carbide substrate following the completion of the first polishing stage, according to the operating data, the method may further include: After the second polishing stage, the substrate thickness of the target silicon carbide substrate in each unit area corresponding to the region is obtained, and the substrate thickness in each unit area corresponding to the region is taken as the first substrate thickness. Determine whether the thickness of the first substrate is not greater than the target thickness; If the thickness of the first substrate is not greater than the target thickness, then the grinding of the target silicon carbide substrate is considered complete. If there is a substrate thickness in the first substrate thickness that is greater than the target thickness, then the process parameters of the grinding equipment in the first grinding stage are updated using the process parameters of the grinding equipment in the second grinding stage, and the substrate data of the target silicon carbide substrate after the end of the second grinding stage is updated using the substrate data of the target silicon carbide substrate after the end of the first grinding stage. Then, the process parameters and substrate data are input into the pre-built grinding path prediction model, and the step of outputting the operating data of the grinding equipment continues to be executed.

[0070] Optionally, after acquiring the operating data of the grinding equipment in the second grinding stage, the grinding equipment is driven to perform grinding using the operating data. The substrate thickness of the target silicon carbide substrate in each unit area after the second grinding stage is acquired and used as the first substrate thickness. Then, it is determined whether the first substrate thickness is not greater than the target thickness. If it is not greater than the target thickness, it indicates that the grinding of the target silicon carbide substrate is complete. If there is a first substrate thickness greater than the target thickness, it indicates that the grinding of the target silicon carbide substrate is not complete. The process parameters of the grinding equipment in the first grinding stage are updated using the process parameters of the grinding equipment in the second grinding stage, and the substrate data of the target silicon carbide substrate after the second grinding stage is updated using the substrate data of the target silicon carbide substrate after the first grinding stage. Then, the process returns to step 102 to continue execution.

[0071] This application provides a silicon carbide substrate polishing method. During the polishing process of the target silicon carbide substrate, the method acquires the process parameters of the polishing equipment in a first polishing stage and the substrate data of the target silicon carbide substrate after the first polishing stage. The process parameters include polishing pressure, polishing speed, and disk surface temperature. The substrate data includes abrasive density, substrate thickness, and substrate surface flatness. The first polishing stage is any polishing stage in the polishing process of the target silicon carbide substrate. The process parameters and substrate data are input into a pre-built polishing path prediction model, which outputs the operating data of the polishing equipment. The operating data includes the polishing path, polishing pressure, polishing speed, polishing time, and polishing rate of the polishing equipment in a second polishing stage, which is the next polishing stage after the first polishing stage. The method controls the polishing equipment to perform the second polishing stage on the target silicon carbide substrate after the first polishing stage according to the operating data. This application obtains the process parameters and substrate data of the first grinding stage and inputs them into the grinding path prediction model. Based on this real-time and comprehensive information, it can accurately predict the operating data required for the second grinding stage. This allows the grinding equipment to operate according to the optimal grinding path, pressure, speed, and other parameters, effectively reducing the microscopic undulations and defects on the substrate surface, thereby significantly improving the surface flatness of the silicon carbide substrate and meeting the substrate quality requirements of high-precision semiconductor device manufacturing. In addition, this application utilizes the grinding path prediction model to quickly and accurately predict the optimal operating data for the second grinding stage based on the actual data of the first grinding stage, avoiding repeated trials, greatly shortening the process development cycle, and improving production efficiency.

[0072] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0073] The following are device embodiments of this application. For details not described in detail, please refer to the corresponding method embodiments described above.

[0074] Figure 2 A schematic diagram of the silicon carbide substrate polishing apparatus provided in an embodiment of this application is shown. For ease of explanation, only the parts relevant to the embodiment of this application are shown, and are described in detail below: like Figure 2 As shown, the silicon carbide substrate polishing apparatus 2 includes: The data acquisition module 21 is used to acquire the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage during the grinding process of the target silicon carbide substrate. The process parameters include grinding pressure, grinding speed and disk temperature, and the substrate data includes abrasive density, substrate thickness and substrate surface flatness. The first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate. The data prediction module 22 is used to input process parameters and substrate data into a pre-built grinding path prediction model and output the operation data of the grinding equipment. The operation data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage. The second grinding stage is the next grinding stage after the first grinding stage. The grinding module 23 is used to control the grinding equipment to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

[0075] This application provides a silicon carbide substrate polishing apparatus. During the polishing process of the target silicon carbide substrate, the process parameters of the polishing equipment in the first polishing stage and the substrate data of the target silicon carbide substrate after the first polishing stage are acquired. The process parameters include polishing pressure, polishing speed, and disk surface temperature. The substrate data includes abrasive density, substrate thickness, and substrate surface flatness. The first polishing stage is any polishing stage in the polishing process of the target silicon carbide substrate. The process parameters and substrate data are input into a pre-built polishing path prediction model, and the operating data of the polishing equipment is output. The operating data includes the polishing path, polishing pressure, polishing speed, polishing time, and polishing rate of the polishing equipment in the second polishing stage, which is the next polishing stage after the first polishing stage. The polishing equipment is controlled to perform the second polishing stage on the target silicon carbide substrate after the first polishing stage according to the operating data. This application obtains the process parameters and substrate data of the first grinding stage and inputs them into the grinding path prediction model. Based on this real-time and comprehensive information, it can accurately predict the operating data required for the second grinding stage. This allows the grinding equipment to operate according to the optimal grinding path, pressure, speed, and other parameters, effectively reducing the microscopic undulations and defects on the substrate surface, thereby significantly improving the surface flatness of the silicon carbide substrate and meeting the substrate quality requirements of high-precision semiconductor device manufacturing. In addition, this application utilizes the grinding path prediction model to quickly and accurately predict the optimal operating data for the second grinding stage based on the actual data of the first grinding stage, avoiding repeated trials, greatly shortening the process development cycle, and improving production efficiency.

[0076] In one possible implementation, the data acquisition module can be used for: After the first grinding stage is completed, the surface image of the target silicon carbide substrate is acquired by an industrial camera installed on the grinding equipment, and the surface image is segmented and calculated using an image segmentation algorithm to obtain the number of abrasive grains per unit area on the surface image. Based on the number of abrasive grains per unit area, the abrasive grain density per unit area is calculated. The substrate thickness of the target silicon carbide substrate is acquired using a laser thickness gauge; the substrate thickness includes the thickness of the region corresponding to each unit area on the target silicon carbide substrate. The total thickness deviation of the target silicon carbide substrate is calculated using the maximum and minimum values ​​of the substrate thickness, and the surface flatness of the target silicon carbide substrate is determined using the total thickness deviation.

[0077] In one possible implementation, the data prediction module can be used for: The process parameters of the grinding equipment for each grinding stage of the silicon carbide substrate sample during the grinding process, the substrate data of the silicon carbide substrate sample after each grinding stage, and the operation data of the grinding equipment for the next grinding stage corresponding to each grinding stage are obtained, and a training set is constructed. A grinding path prediction model is constructed by using convolutional neural networks and long short-term memory networks, combined with a reward function; A grinding path prediction model is trained based on the training set.

[0078] In one possible implementation, the device may further include a model optimization module, which can be used for: Obtain a first reward function value and a second reward function value. The first reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the target grinding stage, the substrate data of the silicon carbide substrate sample after the target grinding stage, and the operating data of the grinding equipment in the next grinding stage of the target grinding stage. The second reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the previous grinding stage of the target grinding stage, the substrate data of the silicon carbide substrate sample after the previous grinding stage of the target grinding stage, and the operating data of the grinding equipment in the target grinding stage. The target grinding stage is any grinding stage among the grinding stages of the silicon carbide substrate samples in the training set during the grinding process. Calculate the difference between the first reward function value and the second reward function value, and take the absolute value of the difference to obtain the target value; The substrate surface flatness in the substrate data of the silicon carbide substrate sample after the target grinding stage is taken as the first substrate surface flatness. Based on the target value and the flatness of the first substrate surface, the grinding path prediction model is deemed to have completed training.

[0079] In one possible implementation, the model optimization module can also be used for: Determine whether the target value is not greater than a preset threshold, and determine whether the surface flatness of the first substrate is not greater than a preset flatness. If the target value is not greater than the preset threshold and the flatness of the first substrate surface is not greater than the preset flatness, then the grinding path prediction model training is considered complete. If the target value is greater than the preset threshold, and / or the flatness of the first substrate surface is greater than the preset flatness, then it is determined that the training of the grinding path prediction model is not completed, and the step of training the grinding path prediction model based on the training set is returned to continue execution.

[0080] In one possible implementation, the device may further include a grinding determination module, which can be used to: After the second polishing stage, the substrate thickness of the target silicon carbide substrate in each unit area corresponding to the region is obtained, and the substrate thickness in each unit area corresponding to the region is taken as the first substrate thickness. Determine whether the thickness of the first substrate is not greater than the target thickness; If the thickness of the first substrate is not greater than the target thickness, then the grinding of the target silicon carbide substrate is considered complete. If there is a substrate thickness in the first substrate thickness that is greater than the target thickness, then the process parameters of the grinding equipment in the first grinding stage are updated using the process parameters of the grinding equipment in the second grinding stage, and the substrate data of the target silicon carbide substrate after the end of the second grinding stage is updated using the substrate data of the target silicon carbide substrate after the end of the first grinding stage. Then, the process parameters and substrate data are input into the pre-built grinding path prediction model, and the step of outputting the operating data of the grinding equipment continues to be executed.

[0081] Figure 3 This is a schematic diagram of the terminal provided in an embodiment of this application. For example... Figure 3 As shown, the terminal 3 in this embodiment includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program 32, it implements the steps in the various silicon carbide substrate polishing method embodiments described above, for example... Figure 1 Steps 101 to 103 are shown. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of each module are shown.

[0082] For example, the computer program 32 can be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 30 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 32 in the terminal 3. For example, the computer program 32 can be divided into... Figure 2 The modules shown.

[0083] The terminal 3 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of terminal 3 and does not constitute a limitation on terminal 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0084] The processor 30 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0085] The memory 31 can be an internal storage unit of the terminal 3, such as a hard disk or memory of the terminal 3. The memory 31 can also be an external storage device of the terminal 3, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal 3. Furthermore, the memory 31 can include both internal storage units and external storage devices of the terminal 3. The memory 31 is used to store the computer program and other programs and data required by the terminal. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0087] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0088] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0089] In the embodiments provided in this application, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0090] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0091] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0092] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various silicon carbide substrate polishing method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0093] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for polishing a silicon carbide substrate, characterized in that, include: During the grinding process of the target silicon carbide substrate, the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage are obtained. The process parameters include grinding pressure, grinding speed and disk surface temperature, and the substrate data include abrasive density, substrate thickness and substrate surface flatness. The first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate. The process parameters and substrate data are input into a pre-built grinding path prediction model, and the operating data of the grinding equipment is output. The operating data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage, and the second grinding stage is the next grinding stage after the first grinding stage. The grinding equipment is controlled to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

2. The silicon carbide substrate polishing method according to claim 1, characterized in that, Acquire substrate data of the target silicon carbide substrate after the completion of the first polishing stage, including: After the first grinding stage is completed, the surface image of the target silicon carbide substrate is acquired by an industrial camera installed on the grinding equipment, and the surface image is segmented and calculated using an image segmentation algorithm to obtain the number of abrasive grains per unit area on the surface image. Based on the number of abrasive grains per unit area, the abrasive grain density per unit area is calculated. The substrate thickness of the target silicon carbide substrate is acquired using a laser thickness gauge; the substrate thickness includes the thickness of the region corresponding to each unit area on the target silicon carbide substrate. The total thickness deviation of the target silicon carbide substrate is calculated using the maximum and minimum values ​​of the substrate thickness, and the surface flatness of the target silicon carbide substrate is determined using the total thickness deviation.

3. The silicon carbide substrate polishing method according to claim 1, characterized in that, The process of constructing the grinding path prediction model is as follows: The process parameters of the grinding equipment for each grinding stage of the silicon carbide substrate sample during the grinding process, the substrate data of the silicon carbide substrate sample after each grinding stage, and the operation data of the grinding equipment for the next grinding stage corresponding to each grinding stage are obtained, and a training set is constructed. A grinding path prediction model is constructed by using convolutional neural networks and long short-term memory networks, combined with a reward function; The grinding path prediction model is trained based on the training set.

4. The silicon carbide substrate polishing method according to claim 3, characterized in that, After training the grinding path prediction model based on the training set, the method further includes: Obtain a first reward function value and a second reward function value. The first reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the target grinding stage, the substrate data of the silicon carbide substrate sample after the target grinding stage, and the operating data of the grinding equipment in the next grinding stage of the target grinding stage. The second reward function value is the reward function value of a grinding path prediction model trained using the process parameters of the grinding equipment in the previous grinding stage of the target grinding stage, the substrate data of the silicon carbide substrate sample after the previous grinding stage of the target grinding stage, and the operating data of the grinding equipment in the target grinding stage. The target grinding stage is any grinding stage among the grinding stages of the silicon carbide substrate sample in the training set during the grinding process. Calculate the difference between the first reward function value and the second reward function value, and take the absolute value of the difference to obtain the target value; The substrate surface flatness in the substrate data of the silicon carbide substrate sample after the target grinding stage is taken as the first substrate surface flatness. Based on the target value and the surface flatness of the first substrate, it is determined that the grinding path prediction model has been trained successfully.

5. The silicon carbide substrate polishing method according to claim 4, characterized in that, The step of determining that the grinding path prediction model has been trained successfully based on the target value and the surface flatness of the first substrate includes: Determine whether the target value is not greater than a preset threshold, and determine whether the surface flatness of the first substrate is not greater than a preset flatness. If the target value is not greater than the preset threshold and the flatness of the first substrate surface is not greater than the preset flatness, then the grinding path prediction model is determined to have been trained. If the target value is greater than the preset threshold, and / or the flatness of the first substrate surface is greater than the preset flatness, then it is determined that the training of the grinding path prediction model is not completed, and the step of training the grinding path prediction model based on the training set is returned to continue execution.

6. The silicon carbide substrate polishing method according to claim 1, characterized in that, After the grinding equipment is controlled to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage has ended, according to the operating data, the method further includes: The substrate thickness of the target silicon carbide substrate in each unit area corresponding to the end of the second polishing stage is obtained, and the substrate thickness in each unit area corresponding to the end of the second polishing stage is taken as the first substrate thickness. Determine whether the thickness of the first substrate is not greater than the target thickness; If the thickness of the first substrate is not greater than the target thickness, then the grinding of the target silicon carbide substrate is determined to be complete. If there is a substrate thickness in the first substrate thickness that is greater than the target thickness, then the process parameters of the grinding equipment in the first grinding stage are updated using the process parameters of the grinding equipment in the second grinding stage, and the substrate data of the target silicon carbide substrate after the end of the second grinding stage is updated using the substrate data of the target silicon carbide substrate after the end of the first grinding stage. Then, the process parameters and the substrate data are input into the pre-built grinding path prediction model and the running data of the grinding equipment are returned to continue execution.

7. A silicon carbide substrate polishing apparatus, characterized in that, include: The data acquisition module is used to acquire the process parameters of the grinding equipment in the first grinding stage and the substrate data of the target silicon carbide substrate after the first grinding stage is completed during the grinding process of the target silicon carbide substrate. The process parameters include grinding pressure, grinding speed and disk surface temperature, and the substrate data include abrasive density, substrate thickness and substrate surface flatness. The first grinding stage is any grinding stage in the grinding process of the target silicon carbide substrate. The data prediction module is used to input the process parameters and the substrate data into a pre-built grinding path prediction model and output the operating data of the grinding equipment. The operating data includes the grinding path, grinding pressure, grinding speed, grinding time and grinding rate of the grinding equipment in the second grinding stage, where the second grinding stage is the next grinding stage after the first grinding stage. The grinding module is used to control the grinding equipment to perform the second grinding stage on the target silicon carbide substrate after the first grinding stage is completed, according to the operating data.

8. The silicon carbide substrate polishing apparatus according to claim 7, characterized in that, The data acquisition module is used for: After the first grinding stage is completed, the surface image of the target silicon carbide substrate is acquired by an industrial camera installed on the grinding equipment, and the surface image is segmented and calculated using an image segmentation algorithm to obtain the number of abrasive grains per unit area on the surface image. Based on the number of abrasive grains per unit area, the abrasive grain density per unit area is calculated. The substrate thickness of the target silicon carbide substrate was acquired using a laser thickness gauge. The substrate thickness includes the thickness of the region corresponding to each unit area on the target silicon carbide substrate; The total thickness deviation of the target silicon carbide substrate is calculated using the maximum and minimum values ​​of the substrate thickness, and the surface flatness of the target silicon carbide substrate is determined using the total thickness deviation.

9. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the silicon carbide substrate polishing method as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the silicon carbide substrate polishing method as described in any one of claims 1 to 6.

Images