Method of treating a polishing pad, apparatus and polishing equipment

Abstract

This application relates to the field of semiconductor manufacturing technology, and provides a method, apparatus, and polishing equipment for processing polishing pads. The method includes: when the polishing equipment is in a non-processing state, spatially sampling the surface of the polishing pad based on the environmental field distribution characteristics inside the equipment to obtain current state distribution data; predicting the state trend of the polishing pad surface within a future set time window based on the current state distribution data and real-time environmental parameters, and determining risk areas; if the state trend meets preset conditions, solving for the optimal intervention strategy in a multi-objective optimization space and controlling the actuator to perform state maintenance operations. This application achieves precise maintenance of the polishing pad state, reducing resource consumption and the risk of scratches.

Description

Technical Field

[0001] This application relates to the field of semiconductor manufacturing technology, specifically to a method, apparatus, and polishing equipment for processing polishing pads. Background Technology

[0002] In the chemical mechanical polishing (CMP) process for semiconductor wafers, the polishing pad must be kept in a constant state of moisture to ensure process stability and surface quality. When the polishing equipment enters a non-processing state due to fault alarms, program interruptions, or waiting for materials, the deionized water on the surface of the polishing pad evaporates rapidly, and the residual chemicals in the polishing slurry easily crystallize and solidify, causing the polishing pad to dry out and harden locally or entirely. When the polishing equipment resumes operation, the dry polishing pad is one of the main causes of scratches on the wafers during subsequent polishing.

[0003] In existing technologies, humidity control for polishing pads typically employs passive response mechanisms such as single-point sensor feedback or timed cleaning. However, due to uneven airflow distribution and heat source effects within the polishing equipment, the drying process on the polishing pad surface exhibits significant spatial non-uniformity. Single-point monitoring cannot accurately perceive the overall drying gradient of the polishing pad. Furthermore, simple threshold triggering or timed control lacks predictability regarding drying trends, often intervening only after a drying risk has actually occurred, or resorting to excessive cleaning to mitigate the risk. This delayed maintenance approach not only fails to completely eliminate the risk of scratches but also results in significant waste of maintenance fluids, energy, and consumables, and reduces the overall uptime of the equipment. Summary of the Invention

[0004] This application provides a method, apparatus, and polishing equipment for processing polishing pads, enabling predictive and precise maintenance of the polishing pad's condition.

[0005] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, this application provides a method for processing a polishing pad, comprising: when the polishing equipment is in a non-processing state, spatially sampling the surface of the polishing pad of the polishing equipment based on the environmental field distribution characteristics inside the polishing equipment to obtain the current state distribution data of the polishing pad surface; predicting the state trend of the polishing pad surface within a future set time window based on the current state distribution data and real-time environmental parameters, and determining risk areas; if the state trend meets preset conditions, solving for the optimal intervention strategy in a multi-objective optimization space, and controlling the execution mechanism to perform state maintenance operations on the polishing pad according to the optimal intervention strategy; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption, and maintenance action time.

[0006] As an optional implementation, the environmental field distribution characteristics include the distribution of drying acceleration factors; the spatial sampling of the polishing pad surface based on the environmental field distribution characteristics inside the polishing equipment includes: allocating a higher sampling density in the region where the drying acceleration factors are concentrated to other regions using a non-uniformly deployed sensor array, in order to construct the current state distribution data reflecting the drying gradient.

[0007] As an optional implementation, the drying acceleration factors include the dominant airflow direction or the location of the heat source; the sensor array is deployed in a specific fan-shaped area upstream of the dominant airflow direction or the location of the heat source, with a first density and a second density in the other areas, wherein the first density is greater than the second density; the current state distribution data includes the overall average humidity, regional drying unevenness, and the instantaneous drying rate at each sampling point.

[0008] As an optional implementation, predicting the state trend of the polishing pad surface within a future set time window based on the current state distribution data and real-time environmental parameters includes: calling a drying kinetic model, using the current state distribution data and the real-time environmental parameters as input, predicting the expected state value of each spatial point on the polishing pad surface within the future set time window, and determining the risk area that reaches the danger threshold based on the expected state value.

[0009] As an optional implementation, the input of the drying kinetics model further includes historical state data of the polishing pad, which includes processing time and material parameters; the output of the drying kinetics model further includes the expected time for each spatial point to reach the danger threshold; predicting the state trend of the polishing pad surface within a future set time window further includes: calculating the expected time and identifying the corresponding area where the expected time is less than a preset time threshold as the risk area.

[0010] As an optional implementation, the multi-objective optimization space also includes physical wear indicators of the polishing pad; the solution of the optimal intervention strategy in the multi-objective optimization space includes: taking the minimization of drying risk, the minimization of maintenance fluid consumption, the minimization of maintenance action time, and the minimization of physical wear indicators as optimization objectives, and solving the globally optimal intervention strategy through a multi-objective optimization algorithm.

[0011] As an optional implementation, the optimal intervention strategy includes a target intervention area, an intervention mode, a duration, a fluid flow rate, and rotation parameters of the polishing pad; the control actuator performs a state maintenance operation on the polishing pad according to the optimal intervention strategy, including: driving the multi-degree-of-freedom cleaning arm to position itself to the target intervention area, opening the nozzle corresponding to the intervention mode, spraying maintenance fluid according to the fluid flow rate and the duration, and controlling the polishing pad to rotate according to the rotation parameters.

[0012] As an optional implementation, after the control actuator performs a state maintenance operation on the polishing pad according to the optimal intervention strategy, the method further includes: continuously monitoring the state changes of the polishing pad surface and comparing the measured state change data with the expected trend data; if the measured state change data does not reach the predetermined target range, adjusting the execution parameters of the optimal intervention strategy and re-executing the state maintenance operation; if the measured state change data stabilizes within the predetermined target range, stopping the state maintenance operation and archiving the full-link data of this event to update the parameters and optimization decision rules of the drying kinetic model used to predict the state trend.

[0013] Secondly, this application also provides a polishing pad processing device, comprising: a spatial sampling module, used to perform spatial sampling on the surface of the polishing pad based on the environmental field distribution characteristics inside the polishing equipment when the polishing equipment is in a non-processing state, to obtain the current state distribution data of the polishing pad surface; a trend prediction module, used to predict the state trend of the polishing pad surface within a future set time window based on the current state distribution data and real-time environmental parameters, and to determine the risk area; an optimization decision module, used to solve for the optimal intervention strategy in a multi-objective optimization space if the state trend meets preset conditions; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption, and maintenance action time; and an execution control module, used to control the execution mechanism to perform state maintenance operations on the polishing pad according to the optimal intervention strategy.

[0014] Thirdly, this application also provides a polishing apparatus, including: a polishing pad; a maintenance fluid supply assembly; and a polishing pad processing device as described above, wherein the polishing pad processing device is communicatively connected to the maintenance fluid supply assembly to control the maintenance fluid supply assembly to provide maintenance fluid to the polishing pad.

[0015] The technical solution provided in this application overcomes the shortcomings of traditional single-point or uniform sampling, which cannot capture the local drying gradient caused by drying acceleration factors such as airflow and heat sources, by using non-uniform spatial sampling based on environmental field distribution characteristics. This provides a data foundation for precise maintenance. On this basis, the system uses a drying dynamics model to predict the state trend within a future set time window, enabling the system to intervene in advance before the actual occurrence of drying risks, fundamentally avoiding wafer scratches caused by hardening of the polishing pad. Furthermore, by solving for the optimal intervention strategy in an optimization space that includes state maintenance effect, resource consumption, action time, and physical wear, and combining closed-loop feedback and self-learning mechanisms, the system ensures that each maintenance action is the globally optimal solution for the current working condition. This significantly reduces maintenance fluid consumption and equipment energy consumption, and improves the overall uptime of the equipment, while ensuring that the polishing pad is always in a recoverable production state. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of the polishing pad processing device according to an embodiment of this application; Figure 2 This is a schematic diagram of sensor deployment according to an embodiment of this application; Figure 3 This is a flowchart of a polishing pad processing method according to an embodiment of this application.

[0017] Figure Labels 100-Polishing equipment main control system, 200-Adaptive moisturizing control unit, 300-Maintenance fluid supply component, 400-Upper plate, 500-Lower plate, 600-Polishing pad, 800-Wafer, 900-Multi-mode nozzle, 10-First sensor, 20-Second sensor. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms "a" or "one," and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms "connected" or "linked," and similar terms, are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right," etc., are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship also changes accordingly.

[0020] Example 1 First refer to Figure 1 and Figure 2 , Figure 1 This illustration shows the overall architecture and sensor layout of the polishing pad processing device provided in an embodiment of this application. The device includes a polishing equipment main control system 100, an adaptive moisturizing control unit 200, a maintenance fluid supply assembly 300, and a sensor array distributed within the equipment. Figure 2 As shown, the sensor array includes a first sensor 10 and a second sensor 20, which are arranged in a non-uniform strategy in the circumferential area of ​​the polishing pad 600. The second sensor 20 is concentrated in a specific fan-shaped area corresponding to drying acceleration factors (such as upstream of the dominant airflow direction or the location of a heat source), while the first sensor 10 covers the remaining area. A maintenance fluid supply assembly 300 is connected to a multi-mode nozzle 900 via piping. The multi-mode nozzle 900 is located between the upper fixed plate 400 and the lower fixed plate 500 and is used to precisely spray maintenance fluid onto the surface of the polishing pad 600 under the control of the adaptive moisturizing control unit 200. The maintenance fluid includes polishing slurry and / or deionized water. The above hardware architecture provides the physical execution environment for the following processing method. Figure 3 As shown, this embodiment provides a method for processing a polishing pad, which includes the following steps: Step S100: When the polishing equipment is in a non-processing state, based on the environmental field distribution characteristics inside the polishing equipment, spatial sampling is performed on the surface of the polishing pad of the polishing equipment to obtain the current state distribution data of the polishing pad surface.

[0021] In this embodiment, the non-processing state refers to the idle period when the polishing equipment is idle due to reasons such as fault alarm, program interruption, waiting for materials, or planned shutdown, resulting in the upper platen 400 being detached from the lower platen 500 and no wafer 800 polishing operation being performed. In this state, the surface of the polishing pad 600 loses the continuous replenishment of process fluid and is highly susceptible to uneven drying due to environmental factors such as airflow and temperature inside the equipment. In this embodiment, the surface of the polishing pad of the polishing equipment can be spatially sampled by a non-uniformly deployed sensor array. Unlike the uniformly distributed or single-point monitoring methods in the prior art, the spatial sampling strategy in this step has significant non-uniformity, that is, the deployment density and position of the sensor array (including the first sensor 10 and the second sensor 20) are pre-planned or dynamically adjusted according to the environmental field distribution characteristics. For example, in areas with faster airflow or concentrated heat sources, the sensors are deployed more densely, using a first density deployment to capture the more drastic state change gradients in that area; while in relatively stable environmental areas, a second density deployment is used, wherein the first density is greater than the second density. This sampling method can more accurately reproduce the true state distribution of the polishing pad 600 surface with less data, avoiding the problems of insufficient sampling data in key areas or wasted resources in stable areas by uniform sampling.

[0022] Step S200: Based on the current state distribution data and real-time environmental parameters, predict the state trend of the polishing pad surface within a future set time window, and determine the risk area.

[0023] Specifically, the adaptive humidity control unit 200 receives current state distribution data from the sensor array and real-time temperature and humidity parameters from the environmental monitoring component, and analyzes the data using a built-in predictive model. The future set time window can be set based on process experience or model accuracy, such as 5 minutes, 10 minutes, or 15 minutes. Unlike traditional real-time threshold judgment, this embodiment predicts which areas will first exceed the safety boundary and enter a dangerous state at a specific future time. The identified risk areas are those local or overall ranges where the expected state value will be lower than the preset safety threshold. Through the technical solution of this embodiment, the maintenance timing can be shifted from after drying to before drying occurs, fundamentally eliminating the risk of scratches on the wafer 800 caused by the hardening of the polishing pad 600 during drying.

[0024] Step S300: If the state trend meets the preset conditions, solve the optimal intervention strategy in the multi-objective optimization space, and control the actuator to perform state maintenance operation on the polishing pad according to the optimal intervention strategy; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption and maintenance action time.

[0025] When the predicted state trend in step S200 indicates the existence of a risk area, or the risk level exceeds the allowable tolerance, the intervention decision process is triggered. Specifically, the multi-objective optimization space defines several mutually constraining or synergistic objective functions: state maintenance aims to ensure that the humidity or cleanliness of the polishing pad 600 returns to a safe process range after intervention; maintenance resource consumption aims to minimize the use of deionized water, electricity, and chemical reagents; and maintenance action time aims to shorten non-productive maintenance time to improve the overall equipment uptime. The adaptive moisturizing control unit 200 uses a multi-objective optimization algorithm to find the Pareto optimal solution or weighted optimal solution among these objectives, generating an optimal intervention strategy containing specific execution parameters. Subsequently, the polishing equipment main control system 100 drives the maintenance fluid supply component 300 and the multi-mode nozzle 900, among other actuators, to strictly implement maintenance on the polishing pad 600 according to this strategy. This decision-making mechanism based on multi-objective optimization ensures that each maintenance action is the best choice that balances quality, cost, and efficiency under the current operating conditions, effectively overcoming the resource waste or insufficient maintenance problems caused by traditional fixed-mode cleaning. Among them, multi-objective optimization algorithms refer to finding a set of solutions in the objective space that "cannot be fully improved" in optimization problems with multiple conflicting objectives, rather than a single optimal solution.

[0026] It should be understood that although this embodiment describes the above processing flow in a specific order of steps, in other embodiments, the data interaction between the steps can be parallel or asynchronous, as long as the logical causal dependency is maintained.

[0027] Example 2 Building upon Example 1, this example further refines the specific implementation of spatial sampling. Specifically, the environmental field distribution characteristics include the distribution of drying acceleration factors. Spatial sampling of the polishing pad surface is performed based on the environmental field distribution characteristics within the polishing equipment. This includes using a non-uniformly deployed sensor array to allocate a higher sampling density in areas where drying acceleration factors are concentrated compared to other areas. This is because the drying process of the polishing pad 600 in its non-processing state is not uniform but is influenced by the physical environmental field within the equipment. Drying acceleration factors refer to environmental variables that can accelerate the evaporation of moisture on the polishing pad 600 surface or promote the crystallization and solidification of the polishing slurry; their spatial distribution directly determines the spatial distribution of drying risk. By correlating the sampling density of the sensor array with the intensity distribution of drying acceleration factors, high-resolution state data can be obtained in critical areas where the drying gradient changes most drastically, while only basic monitoring is maintained in relatively stable areas. This allows for accurate reconstruction of the true drying situation with optimal hardware cost and data volume, avoiding the shortcomings of traditional uniform sampling, such as insufficient resolution in critical areas or resource redundancy in stable areas.

[0028] In a preferred embodiment, the drying acceleration factors include the dominant airflow direction or the location of the heat source. The sensor array is arranged in a first density in a specific fan-shaped area upstream of the dominant airflow direction or the location of the heat source, and in a second density in other areas, wherein the first density is greater than the second density.

[0029] like Figure 2 As shown, during the operation of a semiconductor double-sided polishing machine, the cooling fan inside the equipment or the air vents of the external cleanroom often form a fixed dominant airflow direction. The airflow velocity is fastest and the convective heat transfer coefficient is highest in this direction, resulting in a significantly higher evaporation rate of moisture on the polishing pad 600 surface located upstream of the airflow compared to the leeward or side areas. Therefore, this embodiment divides the sensor array into two categories: second sensors 20 and first sensors 10. Multiple second sensors 20 are densely deployed in a specific fan-shaped area upstream of the dominant airflow direction (e.g., within the range of 0° to 60° relative to the airflow centerline), forming a first-density sampling area; while in other angular ranges (e.g., 90°, 180°, 270°, etc.), first sensors 10 are sparsely deployed, forming a second-density sampling area. It should be understood that although the figure shows specific angular divisions and sensor numbers, this is only illustrative and not limiting. In other embodiments, if the main drying driving factor becomes the thermal radiation generated by the local heating element, the layout area of ​​the first density should be adjusted accordingly to a fan-shaped area close to the heat source; or, a movable sensor module can be configured to dynamically adjust the orientation of the high-density sampling according to the real-time detected environmental hot spots, so as to adapt to the environmental field changes of different equipment models or different process stages.

[0030] Furthermore, the current state distribution data obtained through the above non-uniform sampling includes the overall average humidity, regional dryness unevenness, and instantaneous drying rate at each sampling point.

[0031] The overall average humidity is calculated by weighting or arithmetically averaging the humidity values ​​measured at all sampling points (including the second sensor 20 and the first sensor 10). It characterizes the overall moisture level of the polishing pad 600 and serves as an indicator for determining whether maintenance is required. Regional drying unevenness is typically quantified using the standard deviation or coefficient of variation of the humidity values ​​at each sampling point. Its magnitude directly reflects the severity of the drying gradient. When this indicator exceeds a preset limit, even if the overall average humidity is within a safe range, it indicates severe localized dryness and wetness differentiation, posing a very high risk of scratches. The instantaneous drying rate at each sampling point is obtained by differentiating or calculating the humidity data over a continuous time series. It is a key parameter for predicting future trends. For example, in a high-density sampling area upstream of the airflow, if a sharp increase in the instantaneous drying rate is detected along with a synchronous increase in regional drying unevenness, it can be determined that the area is undergoing rapid drying, thus identifying risk areas in advance.

[0032] Example 3 In this embodiment, the state trend of the polishing pad surface within a future set time window is predicted based on the current state distribution data and real-time environmental parameters. This includes calling a drying kinetics model, using the current state distribution data and real-time environmental parameters as input, to predict the expected state values ​​of each spatial point on the surface of the polishing pad 600 within the future set time window, and determining the risk areas that reach the danger threshold based on the expected state values. The drying kinetics model is a computational model that characterizes the nonlinear mapping relationship between the moisture evaporation rate on the surface of the polishing pad 600 and multi-dimensional variables.

[0033] To further improve prediction accuracy and adapt to different working conditions, the drying kinetics model also includes historical state data of the polishing pad 600 as input. This historical state data includes processing time and material parameters. Specifically, the processing time reflects the aging degree of the polishing pad 600 and the wear condition of its surface microstructure. As processing time accumulates, the pores on the surface of the polishing pad 600 may become partially blocked due to residual polishing fluid or smoothed due to mechanical friction, leading to significant changes in its water retention capacity and moisture release characteristics. For example, the surface moisture evaporation rate of a polishing pad 600 nearing the end of its service life is often significantly higher than that of a new polishing pad. Meanwhile, material parameters determine the basic physical properties of the polishing pad 600. For example, the polyurethane matrix and the nonwoven fabric matrix have fundamental differences in water absorption, air permeability, and thermal conductivity, directly affecting the parameters of the drying kinetics model. By using historical state data as prior input to the model, the drying kinetics model can make predictions for each specific polishing pad 600, effectively eliminating prediction biases caused by individual differences and aging effects, and achieving accurate predictions.

[0034] The output of the drying kinetics model also includes the expected time for each spatial point to reach the danger threshold. Predicting the state trend of the polishing pad 600 surface within a set time window also involves calculating the expected time and identifying areas where the expected time is less than the preset time threshold as risk areas. Traditional methods typically only determine an anomaly when humidity is detected to be below a certain fixed threshold. At this point, the polishing pad 600 is often already at the drying critical point or even completely dry, easily leading to untimely maintenance. In this embodiment, the preset time threshold is assumed to be 5 minutes, meaning that if an area is expected to dry within 5 minutes, immediate intervention is required. Even if the humidity reading for that area is currently within a safe range, if the drying kinetics model calculates that the point will exceed the danger threshold in 3 minutes, it is still identified as a risk area and subsequent processes are triggered. Conversely, if the current humidity at a point is slightly low but the model predicts it will not reach the danger threshold within the next 30 minutes, no intervention is required, thus avoiding unnecessary resource waste. It should be understood that the preset time threshold is not a fixed constant; it can be dynamically configured based on the standard time required for equipment restart preparation, the response speed of the maintenance fluid supply component 300, or the process's tolerance for scratches.

[0035] Example 4 In this embodiment, the multi-objective optimization space also includes physical wear indicators for the polishing pad 600. In semiconductor polishing processes, the micro-texture and pore structure of the polishing pad 600 surface are crucial for maintaining slurry retention and wafer polishing quality. Excessive mechanical scouring or improper fluid impact can lead to texture smoothing or surface scratches, thereby shortening the lifespan of the polishing pad 600 and affecting subsequent processing yield. Therefore, this embodiment incorporates physical wear indicators into the optimization space to ensure that the generated maintenance strategy restores the polishing pad 600 to a wet state while controlling the physical stress it experiences within a safe threshold.

[0036] The optimal intervention strategy is sought within a multi-objective optimization space, including minimizing drying risk, minimizing maintenance fluid consumption, minimizing maintenance action time, and minimizing physical wear indicators as optimization objectives. A multi-objective optimization algorithm is used to find the globally optimal intervention strategy. For example, when equipment is in a short-term waiting state and is expected to resume operation soon, the weight of "minimizing maintenance action time" can be automatically increased, generating a strategy of rapid water replenishment but slightly higher water consumption to prioritize equipment uptime. Conversely, when equipment enters a long-term shutdown maintenance state due to a malfunction, the weights of "minimizing maintenance fluid consumption" and "minimizing physical wear indicators" are increased, employing a low-frequency, low-flow intermittent moisturizing strategy to maximize resource conservation and protect the mat surface while preventing cracking. It should be understood that the multi-objective optimization algorithm in this application can be implemented using various forms such as genetic algorithms, particle swarm optimization, reinforcement learning, or rule-based heuristic search, as long as it can output decision results that take into account all objectives within the aforementioned multi-dimensional space, and is not limited to a specific mathematical model.

[0037] Furthermore, the optimal intervention strategy includes the target intervention area, intervention mode, duration, fluid flow rate, and rotation parameters of the polishing pad 600. The target intervention area is precisely defined based on the risk area identified in Example 3, and can be a specific fan-shaped area, annular band, or irregular patch, avoiding ineffective spraying of wet areas. The intervention mode is selected according to the degree of dryness and risk level, including, for example, a fine atomization mode and a fan-shaped rinsing mode. The fine atomization mode produces droplets with small diameter and low kinetic energy, suitable for mild dryness or preventative moisturizing, minimizing physical impact on the polishing pad 600 surface while replenishing moisture; the fan-shaped rinsing mode has a larger flow rate and coverage area, suitable for severe dryness or scenarios with residual polishing fluid crystals. Fluid flow rate and duration are key variables for quantifying and controlling maintenance intensity; they work in conjunction with the rotation parameters to determine the fluid accumulation per unit area and the duration of action.

[0038] The control actuator performs state maintenance operations on the polishing pad 600 according to the optimal intervention strategy, including driving the multi-degree-of-freedom cleaning arm to position it to the target intervention area, opening the nozzle corresponding to the intervention mode, spraying maintenance fluid according to the fluid flow rate and duration, and controlling the polishing pad 600 to rotate according to the rotation parameters. Figure 1As shown, the multi-mode nozzle 900, acting as the end effector, is carried and precisely moved above the preset target intervention area by a multi-degree-of-freedom cleaning arm (not separately marked in the figure, but a necessary mechanical component for driving the multi-mode nozzle 900). During spraying, the polishing equipment's main control system 100 synchronously controls the upper stationary plate 400 to drive the polishing pad 600 to rotate according to specified rotation parameters (such as rotation speed, direction of rotation, and acceleration / deceleration curves). On the one hand, the relative motion ensures that the maintenance fluid evenly covers every surface within the target area, avoiding localized over-wetting or physical pitting caused by a fixed nozzle directly spraying the same point for a long time; on the other hand, by precisely matching the flow rate and rotation speed, the liquid film thickness and residence time on the surface of the polishing pad 600 can be dynamically adjusted to ensure consistent maintenance results. For example, when dealing with a heavily dry fan-shaped area upstream of the airflow, the multi-degree-of-freedom cleaning arm may be instructed to hover radially above this area, activating the fan-shaped rinsing mode and spraying at a high flow rate, while simultaneously controlling the polishing pad 600 to rotate at a lower speed to increase the wetting time at a single point. Conversely, when dealing with overall slight dryness, a fine atomization mode may be switched to, coupled with a higher rotation speed, to achieve rapid and uniform humidification throughout the entire circumference. The technical solution of this application can flexibly adapt to various complex drying conditions, truly achieving a globally optimal balance between maintenance precision, resource efficiency, and consumable lifespan.

[0039] Example 5 In this embodiment, after the control actuator performs a state maintenance operation on the polishing pad 600 according to the optimal intervention strategy, the method further includes continuously monitoring the state changes of the polishing pad 600 surface and comparing the measured state change data with the expected trend data. During and after the maintenance operation, the sensor array does not enter a dormant or blind zone state, but continues to maintain high-frequency sampling of the polishing pad 600 surface to obtain measured state change data reflecting the actual intervention effect. At the same time, the adaptive humidification control unit 200 internally runs in parallel an expected trend curve generated based on a drying kinetic model. This curve represents the theoretical trajectory of the polishing pad 600 humidity change over time after executing the current optimal intervention strategy under ideal working conditions. By comparing the measured data with the expected trend point by point or in a sliding window manner, deviations in the actual effect caused by factors such as sudden environmental changes, slight nozzle blockage, or local aging differences in the polishing pad 600 can be keenly detected, providing a basis for subsequent dynamic adjustments or model corrections.

[0040] If the measured state change data does not reach the predetermined target range, the execution parameters of the optimal intervention strategy are adjusted and the state maintenance operation is re-executed. When the measured data shows that the humidity recovery rate is slower than expected, or the final stable value is still below the target lower limit, it is determined that the current strategy is insufficient to offset the current drying load, and an online re-optimization or parameter fine-tuning mechanism is triggered. This dynamic adjustment capability can still ensure the reliable completion of maintenance tasks through real-time feedback compensation when there are deviations in model predictions or unexpected disturbances in external conditions, effectively preventing the risk of dryness of the polishing pad 600 due to a single execution failure.

[0041] If the measured state change data stabilizes within the predetermined target range, the state maintenance operation is stopped, and the entire data for this event is archived to update the parameters and optimization decision rules of the drying kinetic model used to predict state trends. The entire data encompasses all information from the initial environmental field distribution, raw sensor readings, model-predicted risk areas and expected times, strategy parameters generated by optimization decisions, measured feedback curves during execution, to the final achieved target state. Using this data, key physical parameters in the drying kinetic model can be calibrated through online learning or periodic batch training, gradually approximating the actual physical characteristics of the equipment. Simultaneously, the optimization decision rules can be iterated based on historical success rates and actual statistical results of resource consumption, for example, automatically adjusting the weight allocation of each objective in the multi-objective optimization algorithm.

[0042] Example 6 like Figure 1 and Figure 2 As shown, this embodiment provides a polishing pad processing device. Specifically, the polishing pad processing device includes a spatial sampling module, a trend prediction module, an optimization decision module, and an execution control module. It should be understood that each module in this embodiment is a virtual unit based on functional logic partitioning. Its physical implementation can be a software program integrated into the adaptive moisturizing control unit 200, an application-specific integrated circuit, a field-programmable gate array, or a combination thereof, as long as it can realize the functions and data interaction logic defined by each module below. This should not be construed as a specific limitation on the hardware circuit structure.

[0043] The spatial sampling module is used to perform spatial sampling on the surface of the polishing pad based on the environmental field distribution characteristics inside the polishing equipment when the polishing equipment is not in a processing state, thereby obtaining the current state distribution data of the polishing pad surface. Specifically, the spatial sampling module establishes a communication connection with the first sensor 10 and the second sensor 20 to receive the raw humidity signals of each sampling point in real time.

[0044] The trend prediction module is used to predict the state trend of the polishing pad surface within a set time window in the future, based on the current state distribution data and real-time environmental parameters, and to identify risk areas. Specifically, the trend prediction module receives the current state distribution data from the spatial sampling module and the temperature and humidity parameters provided by external environmental sensors, and calls the built-in drying kinetics model for calculation. This module not only outputs the expected state value of each spatial point at a future time, but also further calculates the expected time for each point to reach the danger threshold, and marks areas where the expected time is less than the preset time threshold as risk areas.

[0045] The optimization decision module is used to solve for the optimal intervention strategy in a multi-objective optimization space if the state trend meets the preset conditions; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption, and maintenance action time.

[0046] The execution control module controls the actuator to perform state maintenance operations on the polishing pad according to the optimal intervention strategy. Specifically, this module establishes a communication connection with the maintenance fluid supply component 300, the multi-degree-of-freedom cleaning arm, and the drive motor of the upper stationary plate 400. For example, when receiving a fine atomization moisturizing strategy for the upstream fan-shaped area of ​​the airflow, the execution control module will issue the following instructions sequentially or in parallel: drive the cleaning arm to move to a specified angle, open the atomizing nozzle and adjust the proportional valve to the target flow rate, control the upper stationary plate to rotate at a preset speed, and automatically shut off the fluid and reset the robotic arm after the duration is reached.

[0047] Example 7 This embodiment provides a polishing device, specifically, the polishing device includes a polishing pad 600, a maintenance fluid supply assembly 300, and a polishing pad processing device as described in Embodiment 6.

[0048] The polishing pad processing device is communicatively connected to the maintenance fluid supply assembly 300 to control the maintenance fluid supply assembly 300 to provide maintenance fluid to the polishing pad 600. Specifically, the adaptive moisturizing control unit 200 establishes a bidirectional communication link with the intelligent proportional regulating valve, multi-mode nozzle 900, and pumping unit in the maintenance fluid supply assembly 300 via a high-speed industrial Ethernet or dedicated fieldbus. In the downlink direction, the adaptive moisturizing control unit 200 parses the optimal intervention strategy generated in Example 4 into specific control commands, such as target flow rate values, nozzle mode switching signals, and valve opening percentages, and sends them to the execution end of the maintenance fluid supply assembly 300 in real time. In the uplink direction, the maintenance fluid supply assembly 300 transmits the actual fluid pressure, flow rate feedback, and valve status back to the adaptive moisturizing control unit 200 for closed-loop verification and anomaly monitoring in Example 5. Simultaneously, the fluid output end of the maintenance fluid supply assembly 300 is connected to the multi-mode nozzle 900 via precision piping. This nozzle is configured to controllably spray various types of maintenance fluids, such as deionized water, cleaning fluid, or pH adjuster, onto the surface of the polishing pad 600.

[0049] When the polishing equipment enters a non-processing state due to malfunction, waiting for materials, or planned shutdown, and the upper fixed plate 400 swings off the lower fixed plate 500 to expose the surface of the polishing pad 600, the main control system 100 of the polishing equipment sends a status trigger signal to the adaptive moisturizing control unit 200. The adaptive moisturizing control unit 200 then activates the spatial sampling module, driving the first sensor 10 and the second sensor 20 to acquire the current state distribution data of the polishing pad 600 surface; subsequently, it calls the trend prediction module and the optimization decision module to generate the optimal intervention strategy; finally, through the execution control module, it coordinates the rotation drive mechanism of the maintenance fluid supply component 300 and the upper fixed plate 400 to perform precise maintenance operations on the polishing pad 600.

[0050] Example 8 This embodiment uses a specific maintenance scenario where a semiconductor double-sided polishing machine enters a non-processing state due to a fault alarm as an example for detailed explanation. It should be understood that although this embodiment focuses on a semiconductor double-sided polishing machine, the processing method provided in this application is also applicable to single-sided polishing machines, edge polishing machines, or other precision processing equipment equipped with polishing pads and which pose a risk of uneven drying during downtime.

[0051] In this application scenario, when the semiconductor double-sided polishing machine shuts down urgently due to a sudden malfunction or program interruption, and the upper platen 400 automatically detaches from the lower platen 500 to the maintenance position, the equipment enters a non-processing state. At this time, the polishing pad 600 surface loses continuous replenishment of process fluid, and due to the continuous operation of the internal cooling fan or cleanroom air vents, a significant airflow direction dominates on the polishing pad 600 surface. In conventional technologies, operators often need to perform multiple long-duration high-pressure cleanings based on experience after the equipment resumes operation. If the polishing pad 600 has become severely dry and hardened, time-consuming pad repair work is also required. This not only reduces the overall equipment uptime but also wastes a large amount of deionized water and electrical energy. In contrast, the polishing pad processing device provided in this application can perform polishing pad maintenance in this non-processing state.

[0052] Specifically, when the main control system 100 of the polishing equipment detects that the equipment has entered a non-processing state and the upper plate 400 has been deployed, the adaptive moisturizing control unit 200 is immediately activated and executes the processing flow described in the aforementioned embodiment. First, based on the environmental field distribution characteristics inside the equipment, spatial sampling is performed on the surface of the polishing pad 600 using a non-uniformly deployed sensor array. In this scenario, the second sensor 20 is densely distributed in a fan-shaped area upstream of the dominant airflow direction, which can keenly capture the rapid drying gradient caused by the high-velocity airflow in this area; while the first sensor 10 covers other relatively stable areas. This non-uniform sampling strategy can accurately reproduce the true wetting state of the polishing pad 600 surface with a very small amount of data, avoiding the problem of insufficient resolution in key drying areas in traditional uniform sampling.

[0053] Subsequently, the adaptive humidification control unit 200 invokes the drying dynamics model, combining current state distribution data, real-time ambient temperature and humidity, and historical processing data of the polishing pad 600 to predict the state evolution trend within a set time window. For example, the model may calculate that a certain area upstream of the airflow will exceed the dangerous humidity threshold in 3 minutes, while other areas will remain within a safe range for the next 15 minutes. Based on this prediction, the optimal intervention strategy is solved within a multi-objective optimization space that includes state maintenance effect, maintenance resource consumption, maintenance action time, and physical wear indicators. During execution, the multi-degree-of-freedom cleaning arm precisely positions itself to the target area, the multi-mode nozzle 900 activates the atomization mode and sprays maintenance fluid at an optimized flow rate, while the upper plate 400 rotates at a low speed to ensure uniform coverage. After execution, the actual state changes are continuously monitored and compared with the expected trend. If the target is not met, the parameters are dynamically adjusted and the execution is repeated until the state stabilizes within the predetermined target range.

[0054] Compared to traditional timed, full-circumferential high-pressure cleaning methods, the treatment method provided in this application can save 50% to 75% of deionized water consumption. This embodiment intervenes only when the risk of drying is predicted, completely eliminating ineffective spraying of wet areas and excessive maintenance of low-risk areas. Simultaneously, since the polishing pad 600 is always maintained in an optimal wet state during non-processing periods, allowing for immediate resumption of production, lengthy cleaning or pad repair procedures are unnecessary after equipment restart, reducing equipment preparation time by more than 60% and directly improving production capacity and overall equipment uptime. Furthermore, the consideration of physical wear indicators in the multi-objective optimization decision-making effectively avoids erosion damage to the micro-texture of the polishing pad 600 surface by high-pressure water flow, extending the service life of consumables.

[0055] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for processing a polishing pad, characterized in that, include: When the polishing equipment is in a non-processing state, based on the environmental field distribution characteristics inside the polishing equipment, spatial sampling is performed on the surface of the polishing pad of the polishing equipment to obtain the current state distribution data of the surface of the polishing pad; Based on the current state distribution data and real-time environmental parameters, the state trend of the polishing pad surface within a set time window is predicted, and risk areas are identified. If the state trend meets the preset conditions, the optimal intervention strategy is solved in the multi-objective optimization space, and the actuator is controlled to perform state maintenance operation on the polishing pad according to the optimal intervention strategy; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption and maintenance action time.

2. The method according to claim 1, characterized in that, The environmental field distribution characteristics include the distribution of drying acceleration factors; The step of spatially sampling the surface of the polishing pad based on the environmental field distribution characteristics inside the polishing equipment includes: By using a non-uniformly deployed sensor array, a higher sampling density is allocated to the region where the drying acceleration factors are concentrated, compared to other regions, to construct the current state distribution data that reflects the drying gradient.

3. The method according to claim 2, characterized in that, The factors that accelerate drying include the dominant airflow direction or the location of the heat source; The sensor array is deployed in a first density in a specific fan-shaped area upstream of the dominant airflow direction or at the location of the heat source, and in other areas in a second density, wherein the first density is greater than the second density. The current state distribution data includes the overall average humidity, regional dryness unevenness, and instantaneous drying rate at each sampling point.

4. The method according to claim 1, characterized in that, The step of predicting the state trend of the polishing pad surface within a future set time window based on the current state distribution data and real-time environmental parameters includes: The drying kinetics model is invoked, and the current state distribution data and the real-time environmental parameters are used as inputs to predict the expected state values ​​of each spatial point on the surface of the polishing pad within the future set time window, and the risk area that reaches the danger threshold is determined based on the expected state values.

5. The method according to claim 4, characterized in that, The input to the drying kinetics model also includes the historical state data of the polishing pad, which includes the processing time and material parameters; The output of the drying kinetics model also includes the expected time for each spatial point to reach the danger threshold; The prediction of the state trend of the polishing pad surface within a future set time window also includes: Calculate the expected time, and determine the corresponding area where the expected time is less than a preset time threshold as the risk area.

6. The method according to claim 1, characterized in that, The multi-objective optimization space also includes physical wear indicators for the polishing pad; Solving for the optimal intervention strategy within a multi-objective optimization space includes: With the optimization objectives of minimizing drying risk, minimizing maintenance fluid consumption, minimizing maintenance action time, and minimizing physical wear indicators, the optimal intervention strategy is solved globally using a multi-objective optimization algorithm.

7. The method according to claim 6, characterized in that, The optimal intervention strategy includes the target intervention area, intervention mode, duration, fluid flow rate, and rotation parameters of the polishing pad; The control actuator performs state maintenance operations on the polishing pad according to the optimal intervention strategy, including: The multi-degree-of-freedom cleaning arm is driven to position itself in the target intervention area, the nozzle corresponding to the intervention mode is turned on, maintenance fluid is sprayed according to the fluid flow rate and the duration, and the polishing pad is controlled to rotate according to the rotation parameters.

8. The method according to claim 4, characterized in that, After the control actuator performs a state maintenance operation on the polishing pad according to the optimal intervention strategy, the method further includes: Continuously monitor the state changes of the polishing pad surface and compare the measured state change data with the expected trend data; If the measured state change data does not reach the predetermined target range, adjust the execution parameters of the optimal intervention strategy and re-execute the state maintenance operation; If the measured state change data stabilizes within the predetermined target range, the state maintenance operation is stopped, and the entire chain data of this event is archived to update the parameters and optimization decision rules of the drying kinetic model used to predict the state trend.

9. A polishing pad processing apparatus, characterized in that, include: The spatial sampling module is used to perform spatial sampling on the surface of the polishing pad based on the environmental field distribution characteristics inside the polishing equipment when the polishing equipment is in a non-processing state, and to obtain the current state distribution data of the surface of the polishing pad. The trend prediction module is used to predict the state trend of the polishing pad surface within a set time window in the future, based on the current state distribution data and real-time environmental parameters, and to identify risk areas. An optimization decision module is used to solve for the optimal intervention strategy in a multi-objective optimization space if the state trend meets preset conditions; the multi-objective optimization space includes at least the state maintenance effect, maintenance resource consumption, and maintenance action time. The execution control module is used to control the actuator to perform state maintenance operations on the polishing pad according to the optimal intervention strategy.

10. A polishing device, characterized in that, include: Polishing pad; Maintain fluid supply components; And the polishing pad processing apparatus as claimed in claim 9, wherein the polishing pad processing apparatus is communicatively connected to the maintenance fluid supply assembly to control the maintenance fluid supply assembly to supply maintenance fluid to the polishing pad.

Images