A semiconductor device surface cleaning method and system

Abstract

The present invention discloses a semiconductor device surface cleaning method and system, comprising: placing a semiconductor device that has been preliminarily treated to remove large particle pollutants on a card seat of a cleaning platform; a sensor module collects and obtains first cleaning data of the semiconductor device, and a shooting module obtains second cleaning data of the semiconductor device; performing a first analysis and processing on the second cleaning data to identify pollutant data; forming a data set with the first cleaning data and the pollutant data; performing a second analysis and processing on the data set to calculate an actual cleaning path; based on the actual cleaning path, using a pure water cleaning solution for cleaning; after the pure water cleaning is completed, using nitrogen gas for blow-drying and cleaning; the present invention can improve the cleaning efficiency and cleaning effect of the semiconductor device surface, effectively reduce the cleaning cost, and further improve the accuracy and reliability of the cleaning process.

Description

Technical Field

[0001] The invention belongs to the technical field of computer data processing, and in particular relates to a semiconductor device surface cleaning method and system. Background Art

[0002] In the field of semiconductor manufacturing equipment, various processes pay special attention to the content of impurities and dust on the surface of products and particulate matter in the air. Currently, most of the impurities on the surface of semiconductor devices are still removed by overflow cleaning with a large amount of pure water. However, a large amount of water is wasted during overflow, and the cleaning is incomplete and has poor reliability. Different small particles of impurities cannot be cleaned thoroughly, and some places with small angles cannot be cleaned, resulting in low cleaning efficiency, waste of water resources, and increased costs. Summary of the invention

[0003] In view of the defects existing in the above-mentioned prior art, the present invention provides a method for cleaning the surface of a semiconductor device, comprising the following steps: Step S101, placing the semiconductor device that has been preliminarily treated to remove large particle contaminants on a holder of a cleaning platform.

[0004] Step S102, the sensor module collects and obtains first cleaning data of the semiconductor device, the first cleaning data including the height, width, coordinate position and temperature of the semiconductor device, and the shooting module obtains second cleaning data of the semiconductor device, the second cleaning data including image data of the semiconductor device.

[0005] Step S103: performing a first analysis process on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration.

[0006] Step S104, forming a data set S with the first cleaning data and the contaminant data, wherein the mathematical calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the contaminant, and L_i represents the type of the contaminant; performing a second analysis and processing on the data set S, wherein the second analysis and processing includes calculating an actual cleaning path, wherein the mathematical calculation formula for calculating the actual path P_actual from the starting node n_start (x_start, y_start) to the target node n_target (x_target, y_target) is: P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor*g_temp(n_i))).

[0007] Wherein: g(n_i) represents the actual cost from the starting node n_start (x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target (x_target, y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path.

[0008] The T_factor*g_temp(n_i) represents the additional cost based on temperature after being affected by the temperature influencing factor; the resources are cleaning time and physical distance.

[0009] Step S105: Based on the actual cleaning path, use pure water cleaning solution for cleaning.

[0010] Step S106, after the pure water cleaning is completed, nitrogen is used for drying and cleaning to remove various small particle impurities.

[0011] Wherein, the holder is used to fix the semiconductor device.

[0012] The cleaning platform is provided with an external transparent PVC cover for observing the internal cleaning situation and providing cleaning protection.

[0013] Among them, the mathematical calculation formula of g(n_i) and h(n_i) is: g(n_i)=d(n_start, n_i)+C_cleaning(n_i), where d(n_start, n_i) represents the physical distance from the starting node n_start (x_start, y_start) to the node n_i, and C_cleaning(n_i) represents the cleaning time from the starting node n_start (x_start, y_start) to the node n_i.

[0014] h(n_i)=k*d(n_i,n_target), where k represents the heuristic factor, reflecting the relationship between cleaning efficiency and target distance, and d(n_i,n_target) represents the estimated distance from node n_i to target node n_target (x_target,y_target).

[0015] Among them, the mathematical calculation formula of T_factor and g_temp(n_i) is: T_factor=(T_current-T_min) / (T_max-T_min), wherein T_current represents the current temperature of the pure water cleaning liquid, T_min represents the lower limit of the temperature of the pure water cleaning liquid, and T_max represents the upper limit of the temperature of the pure water cleaning liquid.

[0016] g_temp(n_i)=C_temp*|T_current-T_opt|, where C_temp represents the temperature cost coefficient, reflecting the impact of temperature deviation on cleaning effect, and T_opt represents the set optimal cleaning temperature.

[0017] Among them, different small particle impurities are cleaned at different frequencies and angles; the frequency range is: f∈{40 kHz, 80 kHz, 120 kHz}, different frequencies correspond to different water pressures, and the water pressure range is: Pw∈[0.02, 0.25]MPa; the different angles are: 30°, 60°, 90°; f represents frequency, and Pw represents water pressure.

[0018] The data set S is used to analyze and process data and automatically identify different types of pollutants, wherein the mathematical calculation formula is: R=sum{i=1}^{N}∑(C_i*phi(h_i,w_i)*F(x_i,y_i)*e^{-{T_i} / t}).

[0019] Among them, R represents the comprehensive score of recognition; N represents the number of semiconductor devices; phi(h_i, w_i) represents the effective area function, defined as: phi(h_i, w_i)=(h_i*w_i) / (h_i+w_i), reflecting the effective area of ​​the semiconductor device; F(x_i, y_i) represents the position influence function, defined as: F(x_i, y_i)=1 / (1+|(x_i, y_i)|²); represents the influence of the position of the semiconductor device on the recognition efficiency; t represents the temperature attenuation constant.

[0020] In the step S105, during the cleaning process using pure water cleaning solution, the temperature of the pure water cleaning solution is monitored to ensure that the temperature is within the safety range of [10°C, 35°C]. If the temperature exceeds the safety range, an early warning is issued and the temperature is adjusted.

[0021] Wherein, the temperature is adjusted as follows: if the temperature is lower than 10°C, the heater is automatically activated and the temperature rises to the safety range; if the temperature is higher than 35°C, the cooling device is automatically activated and the temperature drops to the safety range.

[0022] The present invention also proposes a semiconductor device surface cleaning system, comprising: a holder for fixing the semiconductor device that has been preliminarily treated to remove large particle contaminants on a cleaning platform; a laser ranging sensor for obtaining the height and width of the semiconductor device; an optical positioning sensor for obtaining the coordinate position of the semiconductor device; a temperature sensor for measuring the temperature of a pure water cleaning liquid; and a shooting module for obtaining second cleaning data of the semiconductor device, wherein the second cleaning data includes image data of the semiconductor device.

[0023] The first analysis and processing module is used to perform a first analysis and processing on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration.

[0024] The second analysis and processing module forms a data set S with the first cleaning data and the pollutant data, wherein the calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the pollutant, and L_i represents the type of the pollutant; the data set S is subjected to a second analysis and processing, and the second analysis and processing includes calculating an actual cleaning path, wherein the calculation formula for calculating the actual cleaning path P_actual from the starting node n_start (x_start, y_start) to the target node n_target (x_target, y_target) is: P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor*g_temp(n_i))).

[0025] Wherein: g(n_i) represents the actual cost from the starting node n_start (x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target (x_target, y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path.

[0026] The T_factor*g_temp(n_i) represents the additional cost based on temperature after being affected by the temperature influencing factor; the resources are cleaning time and physical distance.

[0027] The pure water cleaning module is used to perform cleaning using pure water cleaning liquid based on the actual cleaning path.

[0028] The blow-dry cleaning module is used to blow-dry and clean with nitrogen after pure water cleaning to remove various small particle impurities.

[0029] Compared with the prior art, the present invention has the following beneficial technical effects.

[0030] By optimizing the cleaning path and dynamically adjusting the temperature factor, it can be ensured that the pure water cleaning fluid covers the surface to be cleaned more effectively, thereby improving the cleaning efficiency and cleaning effect.

[0031] It can effectively reduce resource consumption and time cost in the cleaning process and improve economic efficiency.

[0032] By real-time monitoring of temperature and adjustment of cleaning paths, the solution has strong adaptability and can cope with cleaning needs under different environmental conditions and equipment states, ensuring consistency of cleaning effects.

[0033] By using the normalization coefficient and temperature influence factor, pure water cleaning fluid and other resources can be reasonably allocated and used, reducing waste and improving resource utilization.

[0034] By controlling the temperature of the pure water cleaning fluid and other parameters during the cleaning process, equipment damage and other potential risks caused by excessive temperature can be effectively prevented, thus extending the service life of the equipment.

[0035] Combining sensor data and deep learning algorithms, the solution provides data-based decision support that can automatically identify different types of contaminants and generate optimized cleaning paths, improving the accuracy and reliability of the cleaning process. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The flowchart of a semiconductor device surface cleaning method according to an embodiment of the present invention is shown.

[0037] Figure 2 FIG. 4 is a block diagram showing a semiconductor device surface cleaning system according to an embodiment of the present invention.

[0038] Figure 3 The figure is a comparative experimental test diagram of particles on the surface of a semiconductor device after cleaning using an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms of "a", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, and unless the context clearly indicates other meanings, "multiple" generally includes at least two. The optional embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0041] Embodiment 1, as Figure 1 As shown, the present invention discloses a method for cleaning the surface of a semiconductor device, comprising the following steps: Step S101, placing a semiconductor device that has been preliminarily treated to remove large particle contaminants on a holder of a cleaning platform, wherein the holder is used to fix the semiconductor device.

[0042] Step S102, the sensor module collects and obtains first cleaning data of the semiconductor device, the first cleaning data including the height, width, coordinate position and temperature of the semiconductor device, and the shooting module obtains second cleaning data of the semiconductor device, the second cleaning data including image data of the semiconductor device.

[0043] Step S103: performing a first analysis process on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration.

[0044] Step S104, the first cleaning data and the contaminant data are formed into a data set S, and the mathematical calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the contaminant, and L_i represents the type of the contaminant; the data set S is subjected to a second analysis and processing, and the second analysis and processing includes calculating an actual cleaning path, wherein the mathematical calculation formula for calculating the actual cleaning path P_actual from the starting node n_start (x_start, y_start) to the target node n_target (x_target, y_target) is as follows.

[0045] P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor*g_temp(n_i))).

[0046] Wherein: g(n_i) represents the actual cost from the starting node n_start (x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target (x_target, y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path; the resources are cleaning time and physical distance.

[0047] Step S105: Based on the actual cleaning path, use pure water cleaning solution for cleaning.

[0048] Step S106, after the pure water cleaning is completed, nitrogen is used for drying and cleaning to further remove various small particle impurities.

[0049] In one embodiment, the nitrogen blowing and cleaning process can be consistent with the pure water cleaning process, that is, nitrogen blowing and cleaning are repeated according to steps S104-S105 until the cleaning is completely completed.

[0050] Embodiment 2, the present invention provides a method for cleaning the surface of a semiconductor device, comprising the following steps: Step S101, placing the semiconductor device that has been preliminarily treated to remove large particle contaminants on a holder of a cleaning platform.

[0051] Step S102, the sensor module collects and obtains first cleaning data of the semiconductor device, the first cleaning data including the height, width, coordinate position and temperature of the semiconductor device, and the shooting module obtains second cleaning data of the semiconductor device, the second cleaning data including image data of the semiconductor device.

[0052] Step S103: performing a first analysis process on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration.

[0053] Step S104, forming a data set S with the first cleaning data and the contaminant data, wherein the mathematical calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the contaminant, and L_i represents the type of the contaminant; performing a second analysis process on the data set S, wherein the second analysis process includes calculating an actual cleaning path, wherein the following formula (1) is used to calculate an actual cleaning path P_actual from a starting node n_start (x_start, y_start) to a target node n_target (x_target, y_target). Formula (1): P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor* g_temp(n_i))).

[0054] Wherein: g(n_i) represents the actual cost from the starting node n_start (x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target (x_target, y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path; the resources are cleaning time and physical distance.

[0055] Among them, the temperature influence factor T_factor is determined through experimental data analysis and modeling, reflecting the influence of temperature on the cleaning effect. By establishing a relationship model between cleaning effect and temperature, this factor can be obtained to provide a basis for the optimization of the cleaning process. Conduct a series of experiments or field tests to collect data on the cleaning effect under different temperature conditions. The parameters that need to be recorded include: temperature, water temperature or ambient temperature under different cleaning conditions; cleaning effect indicators, which can be the amount of residual pollutants after cleaning, cleaning time, energy consumption or other related indicators. Define the measurement of cleaning effect based on experimental data. A common measurement method can be E(T)=cleaning effect / (cleaning time or resource consumption); where E(T) is the cleaning effect of pure water cleaning fluid at temperature T, and T represents temperature.

[0056] Through statistical analysis methods, such as linear regression, nonlinear regression or machine learning methods, a relationship model between temperature and cleaning effect is established. The following form of analysis model can be used: E(T)=a+bT+cT²; where a, b, c, etc. are coefficients obtained by fitting experimental data. Through the analysis model, the temperature influence factor T_factor can be defined and obtained by calculating the derivative of the model: T_factor=dE(T) / dT, which indicates the sensitivity of the cleaning effect to temperature changes at a specific temperature.

[0057] Step S105: Based on the actual cleaning path, use pure water cleaning solution for cleaning.

[0058] Step S106, after the pure water cleaning is completed, nitrogen is used for drying and cleaning to further remove various small particle impurities.

[0059] Wherein, the holder is used to fix the semiconductor device.

[0060] The cleaning platform is provided with an external transparent PVC cover for observing the internal cleaning situation and providing cleaning protection.

[0061] Among them, the mathematical calculation formula of g(n_i) and h(n_i) is: g(n_i)=d(n_start, n_i)+C_cleaning(n_i), where d(n_start, n_i) represents the physical distance from the starting node n_start (x_start, y_start) to the node n_i, and C_cleaning(n_i) represents the cleaning time from the starting node n_start (x_start, y_start) to the node n_i.

[0062] h(n_i)=k*d(n_i,n_target), where k represents the heuristic factor, reflecting the relationship between cleaning efficiency and target distance, and d(n_i,n_target) represents the estimated distance from node n_i to the target node n_target (x_target,y_target).

[0063] Among them, the mathematical calculation formulas of T_factor and g_temp(n_i) are as follows.

[0064] T_factor=(T_current-T_min) / (T_max-T_min), where T_current represents the current temperature of the pure water cleaning liquid, T_min represents the lower temperature limit of the pure water cleaning liquid, and T_max represents the upper temperature limit of the pure water cleaning liquid.

[0065] g_temp(n_i)=C_temp*|T_current-T_opt|, where C_temp represents the temperature cost coefficient, reflecting the impact of temperature deviation on cleaning effect, and T_opt represents the set optimal cleaning temperature.

[0066] Among them, C_cleaning(n_i)=(C_time(n_i)+C_resource(n_i)+C_efficiency(n_i))*q.

[0067] Among them, C_time(n_i): represents the time required for cleaning at node n_i, which is expressed by the following formula: C_time(n_i)=T_cleaning*d(n_start,n_i) / v, where T_cleaning is the cleaning efficiency per unit time, d(n_start, n_i) is the distance from the starting node n_start (x_start,y_start) to n_i, and v is the moving speed of the cleaning equipment.

[0068] C_resource(n_i) represents the materials consumed for cleaning at node n_i, including the amount of pure water cleaning fluid and other consumables. The calculation formula is C_resource(n_i)=R_fluid*V_used(n_i)+C_supplies(n_i).

[0069] Among them, R_fluid is the cost of unit pure water cleaning fluid, V_used(n_i) is the volume of pure water cleaning fluid used by node n_i, and C_supplies (n_i) is the cost of other consumables related to cleaning.

[0070] C_efficiency(n_i) represents the impact of the equipment on the cleaning efficiency at node n_i, which can be calculated by the following formula: C_efficiency(n_i)=E_base / E(n_i), where E_base is the base efficiency of the equipment and E(n_i) is the actual cleaning efficiency at node n_i, which varies according to factors such as equipment status and temperature, and can be obtained based on actual measurements.

[0071] q is the adjustment coefficient, which reflects the characteristics of the overall cleaning process and ensures the rationality of the final cleaning cost. It is set as a constant based on experience.

[0072] The calculation formula of T_factor and g_temp(n_i) is: T_factor=(T_current-T_min) / (T_max-T_min), wherein T_current represents the current temperature of the pure water cleaning liquid, T_min represents the lower limit of the temperature of the pure water cleaning liquid, and T_max represents the upper limit of the temperature of the pure water cleaning liquid.

[0073] g_temp(n_i)=C_temp*|T_current-T_opt|, where C_temp represents the temperature cost coefficient, reflecting the impact of temperature deviation on cleaning effect, and T_opt represents the set optimal cleaning temperature.

[0074] The temperature cost coefficient C_temp is calculated as follows: C_temp=α*|T_current-T_opt|+β*(T_current-T_min) / (T_max-T_min); T_current represents the current temperature of the pure water cleaning liquid, the value actually measured, T_opt represents the set optimal cleaning temperature, the ideal temperature determined based on experiments or literature, T_min represents the lower limit of temperature, the lowest temperature used for pure water cleaning liquid, such as 10°C; T_max represents the upper limit of temperature, the highest temperature used for cleaning liquid, such as 35°C; α represents the sensitivity factor of the influence of temperature deviation on the cleaning effect, which is usually determined by experiments, and β represents the degree of influence of temperature range on the cleaning effect, which is usually determined by experiments or statistical analysis. Using regression analysis or other statistical methods to determine the specific values ​​of α and β can be achieved by fitting the experimental data.

[0075] Among them, different frequencies can be selected for different small particle impurities.

[0076] The frequency is f∈{40kHz, 80kHz, 120kHz}, different frequencies correspond to different water pressures, and the water pressure range is Pw∈[0.02, 0.25] MPa. f represents frequency, and Pw represents water pressure.

[0077] Wherein, the step S104 includes: analyzing and processing the data set S to automatically identify different types of pollutants, and the calculation formula is: R=sum{i=1}^{N}∑(C_i*phi(h_i,w_i)*F(x_i,y_i)*e^{-{T_i} / t}).

[0078] Where R represents the comprehensive score of identification; N represents the number of devices.

[0079] phi(h_i,w_i) represents the effective area function, which is defined as: phi(h_i,w_i)=(h_i*w_i) / (h_i+w_i), reflecting the effective area of ​​the semiconductor device; F(x_i,y_i) represents the position influence function, which is defined as: F(x_i,y_i)=1 / (1+|(x_i,y_i)|²); it represents the influence of the position of the semiconductor device on the recognition efficiency; t represents the temperature attenuation constant.

[0080] Commonly used deep learning algorithms include convolutional neural networks (CNN), recurrent neural networks (RNN), long short-term memory networks (LSTM), autoencoders, generative adversarial networks (GAN), and Transformer models (natural language processing models).

[0081] Select the above deep learning model, such as CNN or RNN, build the network structure, and the input features are: C_i, h_i, w_i, x_i, y_i, T_i.

[0082] Use the labeled dataset S to train the model and optimize the loss function to improve the recognition accuracy. The cross entropy loss function is usually used to evaluate the gap between the model output and the true label.

[0083] Input the newly collected data into the trained model and use formula (2) to calculate and obtain the recognition probability of each pollutant.

[0084] According to the value of R, the pollutant type with the highest probability is selected as the identification result, where formula (2) is expressed as follows.

[0085] R=sum{i=1}^{N}∑(C_i*phi(h_i,w_i)*F(x_i,y_i)*e^{-{T_i} / t}); Formula (2).

[0086] Among them, R represents the comprehensive score of recognition, and the larger the value, the better the recognition effect.

[0087] N represents the number of semiconductor devices.

[0088] C_i represents the contaminant concentration of the i-th semiconductor device. h_i represents the height of the i-th semiconductor device; w_i represents the width of the i-th semiconductor device; x_i represents the x value of the two-dimensional coordinate of the i-th semiconductor device; y_i represents the y value of the two-dimensional coordinate of the i-th semiconductor device.

[0089] phi(h_i,w_i) represents the effective area function.

[0090] Defined as phi(h_i,w_i)=(h_i*w_i) / (h_i+w_i), it reflects the effective area of ​​the semiconductor device.

[0091] F(x_i,y_i) represents the position influence function, which is defined as: F(x_i,y_i)=1 / (1+|(x_i,y_i)|²); it represents the influence of the semiconductor device position on the recognition efficiency.

[0092] T_i represents the temperature of the pure water cleaning solution corresponding to the i-th semiconductor device; t represents the temperature attenuation constant, which affects the effect of temperature on identification.

[0093] Contaminant concentrations are identified using the following steps: Use a high-definition camera to capture images or videos of the cleaning process. These images should clearly show contaminants on the surface of the semiconductor device.

[0094] Before analysis, the image needs to be preprocessed, including: De-noising: using filters, such as Gaussian filtering, to remove noise from the image. Adjusting contrast and brightness: enhancing image quality to make contaminants more visible. Image scaling: adjusting the image to a size suitable for model input.

[0095] Use computer vision techniques to extract features related to pollutants: Color analysis: Analyze the color distribution in the image. Some pollutants may have specific color characteristics. Texture analysis: Use texture features, such as local binary patterns (LBP); to identify different types of pollutants.

[0096] Edge Detection: Use methods such as Canny edge detection to locate the boundaries of pollutants. Canny represents a multi-step algorithm for edge detection of images.

[0097] Machine learning or deep learning models are used to estimate pollutant concentrations. Deep learning models such as convolutional neural networks (CNNs) can be selected, which are specifically used for image classification and regression tasks.

[0098] Use a labeled dataset for training. The dataset needs to contain images of pollutants of different concentrations and types and their corresponding concentration labels.

[0099] Optimize model performance by tuning hyperparameters and using methods such as cross-validation.

[0100] Use the trained model to make predictions on new images. Input image: Input the newly acquired image into the model.

[0101] Predicted pollutant concentration: The model outputs a predicted value representing the concentration of pollutants in the image.

[0102] To elaborate on how to use the provided formula to calculate the actual cleaning path P_actual from the starting node n_start (x_start, y_start) to the target node n_target (x_target, y_target).

[0103] First, let's explain the basic concepts of this application. In graph theory and path planning, actual cost and heuristic cost are important concepts, which are usually used to calculate the cost from one node to another. The following is a detailed explanation of these two concepts.

[0104] The actual cost g(n_i) is the actual cost from the starting node n_start(x_start,y_start) to the node n_i. This cost is usually based on the weight of the edge passed through the graph, representing the cost resources required for movement or transmission: such as cleaning time, physical distance, energy, etc.

[0105] In this application, the actual cost can be understood as the sum of the costs of each node when the starting node first passes through the coordinate x-axis and then passes through the coordinate y-axis. Example: Assume that the cost per unit physical distance of two adjacent coordinate nodes is 1, that is, the actual cost of the cost (0,0) to (0,1) is 1. For example, the actual cost of node (0,0) to node (3,4) is from (0,0) to (0,4) via the x-axis, and then from (0,4) to (3,4), so its actual cost is 3+4=7.

[0106] The heuristic cost h_i is an estimate of the cost from node n_i to the target node. It is not necessarily the actual cost, but an estimate based on some heuristic information to help the algorithm find the path more efficiently. The heuristic cost usually uses Euclidean distance or Manhattan distance to calculate the minimum resource consumed from the current node to the target node.

[0107] The additional cost refers to the additional cost caused by certain specific conditions such as temperature and environment changes when selecting the cleaning path. It is reflected by multiplying T_factor by the additional cost related to temperature g_temp(n_i). T_factor can be regarded as a weight factor to adjust the impact of temperature on the total cost.

[0108] Temperature extra cost is the additional cost associated with the operating temperature of the equipment or system. As the temperature changes, the efficiency, lifespan and maintenance frequency of the equipment may be affected, resulting in additional path costs. In path planning, all these costs are combined together, and formula (1) represents the optimal cleaning path found by considering the actual cost, heuristic cost and temperature-related extra cost. This combination enables the search algorithm to consider both the cost already traveled and the potential cost ahead when finding the path, so as to find the optimal path more efficiently. By minimizing these costs, the system can optimize resource usage while ensuring efficiency and safety.

[0109] Assume the starting node is (0,0) and the target node is (10,20).

[0110] g_temp(n_i) in formula (1) is the additional cost based on temperature, which is actually related to the type of pollutant.

[0111] Before calculating the cleaning path, we first define possible nodes. Assume that a grid is divided during the cleaning process, and the nodes include: (0,0), (1,0), (2,0), (3,0), and so on (10,20).

[0112] Actual cost g(n_i); for example, calculate the actual cost from each node to the starting node. Example: Assume that the cost per unit physical distance of two adjacent coordinate nodes is 1.

[0113] g((1,0))=1g((1,0))=1g((1,0))=1.

[0114] g((2,0))=2g((2,0))=2g((2,0))=2.

[0115] By analogy, g((10,20))=10+20=30g((10, 20))=10+20=30g((10,20))=10+20=30.

[0116] Use Euclidean distance as the heuristic cost.

[0117] h(n_i)=(x_target-x_i)²+(y_target-y_i)², calculate the value of h(n_i) and then take the square root.

[0118] h((1,0))=(10-1)²+(20-0)²=481, the square root of 481≈21.93.

[0119] h((2,0))=(10-2)²+(20-0)²=464, the square root of 464≈21.54.

[0120] Similarly, h((10, 20))=0.

[0121] Assume that the temperature influence factor T_factor = 0.5. Assume that the temperature at (3,5) causes an additional cost g_temp((3,5)) = 2; the value can be obtained from the sensor data.

[0122] T_factor*g_temp(n_i) is calculated for each node. T_factor represents the temperature influence factor, which reflects the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature. The T_factor*g_temp(n_i) represents the additional cost based on temperature after being affected by the temperature influence factor.

[0123] The total cost of the cleaning path = actual cost + heuristic cost + additional cost based on temperature after the influence of temperature influencing factor.

[0124] Example: For the node (3,5), the calculation is as follows.

[0125] T_factor*g_temp((3,5))=0.5*2=1.

[0126] The total cost of the cleaning path is: cost(n_i)=g(n_i)+h(n_i)+T_factor*g_temp(n_i).

[0127] For example, the example calculates the total cost of node (3,5). cost((3,5))=g((3,5))+h((3,5))+T_factor*g_temp((3,5)). g((3,5))=8. h((3,5))≈16.52. T_factor*g_temp((3,5))=1. cost((3,5))≈8+16.52+1=25.52.

[0128] The additional temperature cost g_temp(n_i) is calculated through the following steps. This cost is usually related to the operating temperature, environmental factors and operating efficiency of the equipment.

[0129] Before calculating g_temp(n_i), some key parameters need to be determined: operating temperature range, determine the normal operating temperature range of the equipment; real-time monitoring of the current temperature of the equipment; temperature influence coefficient; according to the equipment type and material, set the temperature influence coefficient C_temp on the cost. The coefficient can be determined through historical data analysis; operating time, the time the equipment runs at a specific temperature.

[0130] Calculate the additional temperature cost g_temp(n_i) using the temperature deviation and temperature influence coefficient, g_temp(n_i)=C_temp*(T_max-T_min)*t i In some cases, other factors may also be considered to influence additional costs, such as: Maintenance costs: When equipment is operated in a high temperature environment, more frequent maintenance may be required. Energy consumption: High temperatures may lead to increased energy consumption, which in turn increases operating costs. i Indicates the operating time of a device at a specific temperature.

[0131] After calculating the cost of all nodes, the cleaning path with the minimum cost is selected. Assume that by traversing all possible cleaning paths, the minimum cost cleaning path is finally calculated as: P_actual={(0,0),(1,0),(2,0),(3,5),(10,20)}.

[0132] By calculating the actual cost, heuristic estimated cost, temperature impact and additional cost of each node, the optimal cleaning path P_actual from the starting node (0,0) to the target node (10, 20) can be found. This process takes into account the type of contaminants, device location and temperature impact, making the cleaning process more efficient, accurate and reliable.

[0133] Wherein, the step S105 includes: during the cleaning process, monitoring the temperature of the pure water cleaning liquid to ensure that it is within the safety range of [10°C, 35°C]. If it exceeds this safety range, an early warning is issued and the temperature is adjusted.

[0134] In one embodiment, a high-precision temperature sensor, such as PT100 or thermocouple, is used to monitor the temperature of the pure water cleaning fluid in real time. The sensor should be installed in the pipe where the pure water cleaning fluid circulates to ensure accurate measurement. The upper and lower limits of the temperature are set in the control system; the lower limit: 10°C and the upper limit: 35°C.

[0135] When the temperature exceeds the set range, the system should immediately issue an early warning; Visual alarm: a red alarm is displayed on the control panel. Sound alarm: an alarm sounds to alert the operator. Message notification: notify relevant personnel via SMS or email. If the temperature is below 10°C, the heater is automatically activated. The heater must have a fast heating capability to ensure that the temperature rises quickly to a safe range. If the temperature is above 35°C, a cooling device, such as a water cooling system or refrigeration equipment, is automatically activated to quickly reduce the temperature to a safe range.

[0136] Use PID controllers to precisely regulate the output of heating and cooling equipment. This feedback control system can adjust the working state of the equipment based on the real-time temperature to keep the temperature within the set range. PID: stands for proportional, integral, derivative controller.

[0137] The temperature module should include: temperature sensor, heater, cooling device, and control system.

[0138] Temperature sensors are installed in the pipes where the cleaning fluid flows to ensure that temperature data can be collected in real time. A heater with appropriate power is selected to quickly heat up and connect to the temperature sensor for automatic control. A water cooler or refrigeration equipment can be selected to quickly reduce the temperature of the liquid and link with the control system. The control system uses a microcontroller or PLC as the control center to receive temperature sensor data in real time and control the heating and cooling equipment; PLC stands for: Programmable Logic Controller.

[0139] A user interface may also be provided to display temperature data, set temperature and alarm status in real time.

[0140] Read the temperature sensor data in real time to determine whether the temperature is within the set range.

[0141] When the temperature exceeds or falls below the set threshold, the corresponding heating or cooling operation is triggered and an early warning is issued. The temperature adjustment includes: if the temperature is below 10°C, the heater is automatically activated to ensure that the temperature rises quickly to the safe range.

[0142] If the temperature is higher than 35°C, the cooling device will be automatically activated to quickly reduce the temperature to the safety range.

[0143] Embodiment 3: The present invention further proposes a semiconductor device surface cleaning system, which includes: a holder for fixing the semiconductor device that has been preliminarily treated to remove large particle contaminants on a cleaning platform.

[0144] A laser ranging sensor is used to obtain the height and width of the semiconductor device.

[0145] An optical positioning sensor is used to obtain the coordinate position of the semiconductor device.

[0146] A temperature sensor is used to measure the temperature of the pure water cleaning fluid.

[0147] A shooting module is used to obtain second cleaning data of the semiconductor device, wherein the second cleaning data includes image data of the semiconductor device.

[0148] The first analysis and processing module is used to perform a first analysis and processing on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration.

[0149] A second analysis and processing module forms a data set S with the first cleaning data and the contaminant data. The mathematical calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the contaminant, and L_i represents the type of the contaminant; a second analysis and processing is performed on the data set S, and the second analysis and processing includes calculating an actual cleaning path, wherein the following formula is used to calculate an actual cleaning path P_actual from a starting node n_start (x_start, y_start) to a target node n_target (x_target, y_target).

[0150] P_actual=argmin(Σ( g(n_i)+h(n_i)+T_factor*g_temp(n_i))).

[0151] Wherein: g(n_i) represents the actual cost from the starting node n_start (x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target (x_target, y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path; the resources are cleaning time and physical distance.

[0152] A pure water cleaning module is used to perform cleaning using pure water cleaning fluid based on the actual cleaning path.

[0153] The blow-dry cleaning module is used to blow dry with nitrogen after pure water cleaning to remove various small particle impurities.

[0154] Laser distance sensors determine the distance of an object by emitting a laser beam and measuring the time it takes to reflect back. Place the laser distance sensor above or on the side of the device, aiming directly at the device surface. Based on the output signal of the sensor, which is usually a voltage or digital signal, obtain the height and width data of the device.

[0155] Optical positioning sensors use cameras to capture device images and identify the device's position through image processing algorithms. They can be installed on a mobile platform to monitor the platform's movement to obtain the device's coordinate position. Use a camera to capture the device and use image processing software to identify its position in the image: x-coordinate, y-coordinate. By reading the encoder's value, the position of the semiconductor device on the x-axis and y-axis is calculated.

[0156] The temperature sensor uses a thermocouple or PT100 temperature sensor, where the thermocouple generates voltage at the contact point of two different metals to measure temperature. PT100 uses a platinum resistor to measure temperature with high accuracy. The temperature sensor is installed in the pipe where the pure water cleaning fluid flows, or directly in contact with the surface of the semiconductor device. The temperature data is obtained in real time through the output signal of the sensor: voltage or resistance change.

[0157] Among them, a semiconductor device surface cleaning system also includes the following structures and devices: a nozzle, which is used to discharge water or blow dry gas at different angles to remove impurities at different heights and angles on the substrate surface, and the hole size is 5mm; a rotating shaft; a variable frequency ultrasonic device, the frequency is: {40kHz, 80kHz, 120kHz}, different frequencies have different effects on the removal of impurities on the product surface, and different frequencies can be selected for impurities of different particle sizes to better remove impurities; a temperature control module: the temperature is: [10°C, 35°C], which is used to control the temperature of ultrasonic water to prevent the workpiece from being oxidized and spotting due to excessive temperature; a movable cleaning axis: it moves up and down, left and right, and forward and backward to achieve automatic cleaning of the product; a control panel: it can store a variety of cleaning program solutions; a measuring probe: using the three-point positioning principle, this principle uses three mutually perpendicular coordinate axes: X-axis, Y-axis, Z-axis; to control a probe to move on the surface of the object, and record the coordinate values ​​of the probe at each position. It is the same as the positioning principle in the three-coordinate measurement system. The real-time position of the nozzle is monitored by the measuring probe sensor and is precisely controlled using the three-point positioning principle. The present invention can realize the height and position of the system, and at different times, use the different positions of the rotating shaft to change the different injection angles of the cleaning water and gas to perform pure water cleaning and nitrogen cleaning. The water pressure is changed by using a variable frequency ultrasonic device, and the movable cleaning shaft is used to adjust the position of the cleaning equipment to free up manpower and achieve a higher precision cleaning effect while reducing operating costs. The program control system is rotated to control the different angles of water or gas outlet of the equipment at different times, and the optional angles are 30°, 60°, and 90°. The program control device can move the cleaning equipment to perform pure water cleaning and nitrogen cleaning on semiconductor devices, which can achieve better cleaning and drying effects. The software and the device are linked by a data cable.

[0158] Embodiment 4: In a certain embodiment, the present invention adopts the following process for cleaning.

[0159] After the initial treatment to remove large particles, the semiconductor devices are placed on the cleaning platform holder. The holder is used to fix the semiconductor devices to prevent displacement during the cleaning process. A transparent PVC cover is placed outside the platform. The function of the transparent cover is to check the internal cleaning status and also to play a cleaning protection role.

[0160] The camera collects data through various sensors and system algorithms: product shape, height, width, coordinate position, temperature and other data.

[0161] After collecting the data, we analyze and process it. Different semiconductor devices have different specifications and sizes, so the system analyzes the data and gives different cleaning paths.

[0162] The cleaning nozzle is cleaned according to the path analyzed by the system settings. The cleaning water is pure water. The variable frequency ultrasonic device is located at the nozzle. The variable frequency ultrasonic device changes different frequencies. The frequencies are {40kHz, 80kHz, 120kHz}. The water pressure corresponding to different frequencies is also different. The controllable water pressure range is: [0.02, 0.25] MPa. The three different frequencies have different effects on the removal of impurities on the product surface. Different frequencies can be selected for different small particles of impurities to better remove impurities.

[0163] The temperature is constant during the cleaning process. If the pure water temperature rises or the working environment temperature rises, the temperature sensor collects temperature data, and the system performs early warning processing and constant temperature control to ensure the constant cleaning water temperature. The constant temperature range is: [10°C, 35°C].

[0164] After the semiconductor device is cleaned with pure water, the semiconductor device is cleaned with air blowing, the purpose is to dry the semiconductor device, and further clean the semiconductor device of different small particles of impurities, and the gas cleaning is completed; the gas pressure is: 0.5 Mpa; the gas refers to nitrogen (N2). Pure water refers to purified water (H2O), and the resistivity is >18.25MΩ.CM.

[0165] The cleaning angles of pure water cleaning and air blowing cleaning can be different variable angles.

[0166] The multi-angle cleaning system includes three different angles: 30°, 60°, and 90°. Different angles can remove small particles of impurities at different locations.

[0167] like Figure 3The figure shows the comparative experimental data of the semiconductor device after being cleaned by the system of the present invention, and then soaking it in pure water cleaning solution again, and detecting the particles soaked in pure water cleaning solution; the samples 1 and 2 of the same batch were sampled for comparative testing respectively, and the particle counts of the semiconductor device after comparative cleaning were measured under the same testing instrument, the same cleaning time, and the same temperature safety environment. The particle count data of the semiconductor device soaked in pure water cleaning solution after cleaning is shown in FIG. Figure 3 .

[0168] According to comparative experiments, Figure 3 It can be concluded that different frequencies have different effects on removing different particles, and multi-angle cleaning can clean different positions of the product; the number of particles contained in the semiconductor device after multi-angle and multi-frequency cleaning is the least, and the best cleaning effect and the highest cleaning efficiency can be achieved through multi-angle and multi-frequency cleaning.

[0169] The disclosed embodiment of the present invention provides a non-volatile computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the method steps described in the above embodiment.

[0170] It should be noted that the computer-readable medium disclosed in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0171] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0172] Computer program code for performing the operations disclosed in the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (AN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0173] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to the various embodiments disclosed in the present invention. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0174] The units involved in the embodiments disclosed in the present invention may be implemented by software or hardware, wherein the name of a unit does not limit the unit itself in some cases.

[0175] The above introduces the preferred embodiments of the present invention, which is intended to make the spirit of the present invention clearer and easier to understand, but is not intended to limit the present invention. All modifications, substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection outlined by the claims attached to the present invention.

Claims

1. A method for cleaning the surface of a semiconductor device, comprising the following steps: Step S101, placing the semiconductor device that has been preliminarily treated to remove large particle contaminants on a holder of a cleaning platform; Step S102, the sensor module collects and obtains the first cleaning data of the semiconductor device, the first cleaning data includes the height, width, coordinate position and temperature of the semiconductor device, and the shooting module obtains the second cleaning data of the semiconductor device, the second cleaning data includes the image data of the semiconductor device; Step S103, the second cleaning data is subjected to a first analysis and processing to identify the pollutant data, the pollutant data includes the pollutant type and the pollutant concentration; Step S104, the first cleaning data and the pollutant data are formed into a data set S, the mathematical calculation formula of the data set S is: S={(x_i, y_i, h _i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the pollutant, and L_i represents the type of the pollutant; performing a second analysis process on the data set S, the second analysis process includes calculating the actual cleaning path, wherein the starting node n_start (x_start, y_start) is calculated. The mathematical calculation formula of the actual cleaning path P_actual to the target node n_target (x_target, y_target) is: P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor*g_temp(n_i))); where: g(n_i) represents the actual cost from the starting node n_start(x_start, y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target(x_target, y_target); the heuristic cost table Indicates: the resources consumed by the node to the target node movement; T_factor represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: the resources consumed due to temperature changes when selecting the cleaning path; the T_factor*g_temp(n_i) represents the additional cost based on temperature after being affected by the temperature influence factor; the resources are the cleaning time and the physical distance; step S105, based on the actual cleaning path, use pure water cleaning liquid for cleaning; step S106, after the pure water cleaning is completed, use nitrogen to blow dry and clean to remove different small particles of impurities;The mathematical calculation formula of g(n_i) and h(n_i) is: g(n_i)=d(n_start,n_i)+C_cleaning(n_i), where d(n_start,n_i) represents the physical distance from the starting node n_start(x_start,y_start) to the node n_i, and C_cleaning(n_i) represents the cleaning time from the starting node n_start(x_start,y_start) to the node n_i; h(n_i)=k*d(n_i,n_target), where k represents the heuristic factor, reflecting the relationship between the cleaning efficiency and the target distance, and d(n_i,n_target) represents the distance from the node n_i to the target node n_target (x_target, y_target); The mathematical calculation formula of T_factor and g_temp(n_i) is: T_factor=(T_current-T_min) / (T_max-T_min), where T_current represents the current temperature of the pure water cleaning liquid, T_min represents the lower limit of the temperature of the pure water cleaning liquid, and T_max represents the upper limit of the temperature of the pure water cleaning liquid; g_temp(n_i)=C_temp*|T_current-T_opt|, where C_temp represents the temperature cost coefficient, reflecting the impact of temperature deviation on the cleaning effect, and T_opt represents the set optimal cleaning temperature. ; 2. A semiconductor device surface cleaning method as claimed in claim 1, characterized in that: The holder is used to fix the semiconductor device.

3. A semiconductor device surface cleaning method as claimed in claim 1, characterized in that: The cleaning platform is provided with an external transparent PVC cover for observing the internal cleaning situation and providing cleaning protection.

4. A semiconductor device surface cleaning method as claimed in claim 1, characterized in that: Different small particle impurities are cleaned at different frequencies and angles. The frequency range is: f∈{40kHz, 80kHz, 120kHz}. Different frequencies correspond to different water pressures. The water pressure range is: Pw∈[0.02, 0.25] MPa; the different angles are: 30°, 60°, 90°; f represents frequency, and Pw represents water pressure.

5. A semiconductor device surface cleaning method as claimed in claim 1, characterized in that: The data set S is analyzed and processed, and different types of pollutants are automatically identified, wherein the mathematical calculation formula is: R=sum{i=1}^{N}∑(C_i*phi(h_i,w_i)*F(x_i,y_i)*e^{-{T_i} / t}); wherein R represents the comprehensive score of identification; N represents the number of semiconductor devices; phi(h_i,w_i) represents the effective area function, defined as: phi(h_i,w_i)=(h_i*w_i) / (h_i+w_i), reflecting the effective area of ​​the semiconductor device; F(x_i,y_i) represents the position influence function, defined as: F(x_i,y_i)=1 / (1+|(x_i,y_i)|²); represents the influence of the position of the semiconductor device on the identification efficiency; t represents the temperature attenuation constant.

6. A semiconductor device surface cleaning method as claimed in claim 1, characterized in that: In the step S105, during the cleaning process using pure water cleaning solution, the temperature of the pure water cleaning solution is monitored to ensure that the temperature is within the safety range of [10°C, 35°C]. If the temperature exceeds the safety range, an early warning is issued and the temperature is adjusted.

7. A semiconductor device surface cleaning method as claimed in claim 6, characterized in that: The temperature is adjusted as follows: if the temperature is lower than 10°C, the heater is automatically activated and the temperature rises to the safety range; if the temperature is higher than 35°C, the cooling device is automatically activated and the temperature drops to the safety range.

8. A semiconductor device surface cleaning system, comprising a semiconductor device surface cleaning method as claimed in claim 1, further comprising: A holder for fixing the semiconductor device, which has been preliminarily treated to remove large particle contaminants, on the cleaning platform; A laser distance measuring sensor, used to obtain the height and width of the semiconductor device; An optical positioning sensor is used to obtain the coordinate position of the semiconductor device; a temperature sensor is used to measure the temperature of the pure water cleaning liquid; a shooting module is used to obtain second cleaning data of the semiconductor device, wherein the second cleaning data includes image data of the semiconductor device; a first analysis and processing module is used to perform a first analysis and processing on the second cleaning data to identify pollutant data, wherein the pollutant data includes pollutant type and pollutant concentration; a second analysis and processing module forms a data set S with the first cleaning data and the pollutant data, wherein the calculation formula of the data set S is: S={(x_i, y_i, h_i, w_i, T_i, C_i, L_i)}, wherein i represents the i-th semiconductor device, x_i represents the x value of the two-dimensional coordinate of the semiconductor device, y_i represents the y value of the two-dimensional coordinate of the semiconductor device, h_i is the height of the semiconductor device, w_i is the width of the semiconductor device, T_i is the temperature of the semiconductor device, C_i represents the concentration of the pollutant, and L_i represents the type of the pollutant; the data set S is subjected to a second analysis process, the second analysis process includes calculating an actual cleaning path, wherein the actual cleaning path P_ from the starting node n_start (x_start, y_start) to the target node n_target (x_target, y_target) is calculated. The calculation formula of actual is: P_actual=argmin(Σ(g(n_i)+h(n_i)+T_factor*g_temp(n_i))); where: g(n_i) represents the actual cost from the starting node n_start(x_start,y_start) to the node n_i; the actual cost represents: the resources consumed by moving from the starting node to the node; h(n_i) represents the heuristic cost from the node n_i to the target node n_target(x_target,y_target); the heuristic cost represents: the resources consumed by moving from the node to the target node; T_fact or represents the temperature influence factor, and the temperature influence factor represents the weight factor, which is used to reflect the influence of temperature on the cleaning effect; g_temp(n_i) represents the additional cost based on temperature; the additional cost of temperature represents: when selecting the cleaning path, the resources consumed due to temperature changes; the T_factor*g_temp(n_i) represents the additional cost based on temperature after being affected by the temperature influence factor; the resources are cleaning time and physical distance; a pure water cleaning module is used to clean with pure water cleaning liquid based on the actual cleaning path; a blow-dry cleaning module is used to use nitrogen to blow-dry and clean after the pure water cleaning is completed to remove different small particle impurities.

Images