Polishing pad surface roughness monitoring device with AI system and method thereof
The AI-enabled polishing pad monitoring system addresses the inefficiencies of traditional offline inspections by enabling real-time monitoring and adaptive control, improving CMP process stability and reducing costs.
Patent Information
- Application Number
- CN202510706604.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-29
- Publication Date
- 2025-07-15
AI Technical Summary
In the prior art, the state detection of the polishing pad requires shutdown operation, and the detection equipment is separated from the dressing equipment, resulting in low detection efficiency and unstable polishing quality, and real-time monitoring cannot be achieved, making it difficult to accurately judge the wear state of the polishing pad.
The surface roughness monitoring device of polishing pad with an AI system is adopted, including an optical perceptron module, a data transmission module, a control and alarm module, a diamond disc fixed buckle and a waterproof cable trough. The surface roughness of the polishing pad is monitored in real time by using a micro-laser triangulation displacement sensor, and combined with a machine learning algorithm to analyze the wear trend of the polishing pad, and adaptively adjust the trimming parameters of the diamond disc.
Real-time monitoring and adaptive adjustment of the polishing pad status are realized, which improves the accuracy and stability of the CMP process, improves production efficiency, reduces costs, and extends the service life of the polishing pad and diamond disc.
Smart Images

Figure CN120307184A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of chemical mechanical polishing equipment, and particularly to a surface roughness monitoring device and method for a polishing pad with an AI system. Background Art
[0002] In the manufacturing process and technology of integrated circuit chips, due to the requirement of fine circuit linewidth design, the technology is changing with each passing day towards higher integration density. After thin films are deposited and etched layer by layer on the chip surface of the wafer, unevenness appears in micro copper circuits, tungsten circuits, oxide film dielectric layers, etc. Therefore, after the wafer chip is flattened by chemical mechanical polishing process, it is beneficial for the next process to proceed, and solves the problem that the exposure focusing is difficult due to poor flatness in the circuit lithography process. Therefore, the flatness technology of chemical mechanical polishing process is relatively important. Its main operating components include chemical polishing liquid, polishing pad, and polishing pad conditioner. The polishing pad conditioner repeatedly performs surface conditioning on the polishing pad to maintain good roughness of the polishing pad. The polishing pad is paired with the polishing liquid to achieve the goal of planarization through chemical mechanical polishing.
[0003] Chemical mechanical polishing (CMP) is a surface planarization technology commonly used in semiconductor manufacturing and precision component processing. A diamond disk is used to condition the polishing pad during the CMP process to maintain its flatness and roughness. However, in the prior art, the monitoring of the polishing pad state mainly relies on periodic downtime inspections or external equipment measurements, which leads to a decrease in production efficiency and potential quality fluctuations during the production process. Existing optical monitoring technologies, such as laser interferometers, confocal microscopes, and optical profilometers, although they can provide high-precision measurements, most of the equipment is large in size and difficult to be directly integrated into the dynamic components of CMP equipment; in addition, these technologies are highly sensitive to vibration and environmental conditions, increasing the complexity and cost of the overall system. Therefore, a solution that can monitor the polishing pad state in a lightweight, real-time, and efficient manner is needed to improve the stability of the CMP process and the product yield.
[0004] It is known that in the prior art, the detection of the polishing pad state usually requires downtime operation, and the detection equipment is separated from the conditioning equipment, resulting in low detection efficiency and unstable polishing quality. In addition, traditional detection methods cannot achieve real-time monitoring and it is difficult to accurately judge the wear state of the polishing pad. Summary of the Invention
[0005] Aiming at the deficiencies of the prior art, the present invention provides a method for monitoring the surface roughness of a polishing pad with an AI system to solve the problems mentioned in the above background art that in the prior art, the detection of the polishing pad state usually requires downtime operation, and the detection equipment is separated from the conditioning equipment, resulting in low detection efficiency and unstable polishing quality. In addition, traditional detection methods cannot achieve real-time monitoring and it is difficult to accurately judge the wear state of the polishing pad.
[0006] To achieve the above object, the present invention provides the following technical solution: A surface roughness monitoring device for a polishing pad with an AI system, comprising an optical sensor module, a data transmission module, a control and alarm module, a diamond disc fixing buckle, a waterproof cable trough, and a diamond disc swing arm, characterized in that: the optical sensor module is arranged on the side of the diamond disc fixing seat, the data transmission module is arranged on the side of the diamond disc fixing seat, and the control and alarm module is arranged on the side of the diamond disc fixing seat; The data transmission module is electrically connected to the optical sensor, and transmits the monitored data in real time to the data processing unit and the AI module on the operator's workbench for AI algorithm analysis, and inputs the workbench data to the control and alarm module; The dust-proof and waterproof sealing structure covers the optical sensor module, the data transmission module, and the control and alarm module to ensure its stable operation in the chemical mechanical polishing process environment; The diamond disc fixing buckle is arranged on the lower side of the diamond disc swing arm.
[0007] Preferably, the optical sensor module, the data transmission module, and the control and alarm module are respectively installed at equal intervals of 120 degrees on the outer edge of the diamond disc fixing seat.
[0008] Preferably, an optical sensor is embedded inside the optical sensor module. The optical sensor uses a micro laser triangulation displacement sensor with a measurement range of 1 micron to 1 millimeter and a measurement accuracy of ±1 micron. It is installed in the installation groove in a thin, light, short and small shape. The optical sensor emits laser and receives light beams to monitor the surface roughness and flatness of the polishing pad in real time.
[0009] Preferably, the data transmission module further includes a high-speed wireless transmission module. The high-speed wireless transmission module selects Wi-Fi 6 or Bluetooth 5.0 to ensure the stability and efficiency of real-time data transmission.
[0010] Preferably, the data processing unit receives the roughness and flatness data of the polishing pad from the data transmission module, and transmits the data to the AI module for algorithm analysis. The AI module is based on a machine learning algorithm to analyze the monitored data, learn and train, and predict the wear trend of the polishing pad; then the data is transmitted to the control and alarm module, and the adaptive control module automatically adjusts the dressing parameters of the diamond disc, including but not limited to the pressure, time, rotation speed, and path during the dressing process, so as to optimize the process and improve the accuracy, efficiency, and stability of the CMP process.
[0011] Preferably, the AI analysis module can self-optimize according to the device historical data and real-time data, and continuously improve the accuracy of the polishing pad state judgment and dressing parameter suggestions.
[0012] Preferably, the data processing unit further includes an early warning module, which can issue an alarm in real time when the roughness or flatness of the polishing pad surface exceeds the preset range, prompting the operator or automatically stopping the operation of the CMP equipment.
[0013] Preferably, the diamond disk swing arm is made of SUS430 stainless steel containing iron-chromium alloy, which has both the chemical properties of acid and alkali corrosion resistance and the physical property of high magnetic attraction. It is easy to obtain, reasonable in price and has good processability.
[0014] Preferably, a method for monitoring the surface roughness of a polishing pad with an AI system includes the following steps: Step A: Start the optical sensor to monitor the surface of the polishing pad in real time and collect roughness and flatness data. Step B: Transmit the collected data to the data processing unit of the workbench through the data transmission module. Step C: Analyze the data by the AI analysis module to judge the state of the polishing pad and predict its wear trend. Step D: The adaptive control module receives and adjusts the dressing parameters of the diamond disk according to the AI analysis result to optimize the CMP process. Step E: When an abnormal condition is detected on the surface of the polishing pad, the alarm module issues an alarm signal or automatically stops the process to avoid product defects.
[0015] Preferably, the AI analysis module includes using a regression model, extracting surface topography features through a convolutional neural network (CNN), and performing data training and predicting the wear trend of the polishing pad through a long short-term memory network (LSTM) or a deep learning model, so as to achieve precise process control. Compared with the prior art, the beneficial effects of the present invention are as follows: The device and method for monitoring the surface roughness of a polishing pad with an AI system realize the improvement of process precision and stability, the improvement of production efficiency and the reduction of costs, the simplification of diamond disk replacement and the extension of service life. This method can be applied to fields such as semiconductor and integrated circuit manufacturing, precision optical element processing, and high-precision metal surface treatment to improve product yield and process stability.
[0016] 1. Improve process precision and stability: Ensure that the polishing pad is always in the best state through real-time monitoring and AI data analysis, and improve product yield.
[0017] 2. Improve production efficiency: There is no need to frequently stop the machine to detect the polishing pad, reducing equipment idle time and improving the efficiency of the production line.
[0018] 3. Reduce production costs: Real-time monitoring and adaptive adjustment extend the service life of the polishing pad and diamond disk, reducing maintenance and replacement costs.
[0019] 4. Diamond disc simplifies dressing change and extends service life: Select a lightweight optical displacement sensor and install it on the diamond disc seat. Besides not affecting the normal operation of the swing arm of the original CMP equipment, it is easy to install and extends the service life of the diamond disc. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 It is a schematic diagram of the diamond disc seat device of the present invention; Figure 2 It is a schematic diagram of the connection structure between the CMP structure and the diamond disc seat of the present invention; Figure 3 It is a schematic diagram of the connection structure between the diamond disc swing arm and the diamond disc seat of the present invention; Figure 4 It is a schematic diagram of the working steps of the present invention; Figure 5 It is a schematic diagram of the program topology of the present invention; Figure 6 It is a schematic diagram of the working method program of the present invention.
[0021] In the figure: 1. Diamond disc seat device; 11. Optical sensor module; 111. Optical sensor transmitter; 112. Optical sensor receiver; 12. Data transmission module; 13. Control and alarm module; 14. Diamond disc fixing buckle; 15. Waterproof cable wire groove; 16. Workbench; 2. Diamond disc swing arm; 21. Interface between the fixed seat and the swing arm; 3. Diamond disc; 4. Polishing liquid device; 41. Polishing liquid seat; 42. Polishing liquid dropper; 43. Polishing liquid; 5. Wafer handling device; 51. Wafer handling drive shaft; 52. Wafer gripper; 53. Wafer; 6. Polishing pad device; 61. Polishing pad fixing seat; 62. Polishing pad; 63. Rough surface of the polishing pad. DETAILED DESCRIPTION OF THE INVENTION
[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a 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 those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0023] Please refer to FIGS. 1-5 for reference. The present invention discloses a surface roughness monitoring device for a polishing pad with an AI system, including an optical sensor module, a data transmission module, a control and alarm module, a diamond disc fixing buckle, a waterproof cable trough, and a diamond disc swing arm. The optical sensor module is arranged on the side of the diamond disc fixing seat. The data transmission module is arranged on the side of the diamond disc fixing seat. The control and alarm module is arranged on the side of the diamond disc fixing seat; The data transmission module is electrically connected to the optical sensor; The dust-proof and waterproof sealing structure covers the optical sensor module, the data transmission module, and the control and alarm module; The diamond disc fixing buckle is arranged on the lower side of the diamond disc swing arm.
[0024] Furthermore, the optical sensor module 11 is designed on the side of the diamond disc fixing seat for embedding the optical sensor. A light displacement sensor such as a micro laser triangulation sensor or a micro white light interferometer is used, which is thin, light, and short and installed in the installation groove. The optical sensor emits a laser and receives a light beam for real-time monitoring of the surface roughness and flatness of the polishing pad. The specifications of the optical sensor are as follows: measurement range: 1 micron to 1 mm, accuracy: ±1 micron, volume: ≤60 cubic millimeters, weight: ≤150 grams. The laser beam emitted by the sensor is vertically irradiated on the surface of the polishing pad, and the reflected light 112 is received by the photoelectric sensor inside the sensor and converted into surface topography data.
[0025] Furthermore, the data transmission module 12 is electrically connected to the optical sensor 11, inputs the surface topography data generated by real-time monitoring of the roughness and flatness of the polishing pad, and at the same time transmits the monitoring data in real-time to the data processing unit and the AI module of the external station workbench 16 for AI algorithm analysis, and sends the data of the workbench 16 to the control and alarm module 13.
[0026] Furthermore, the data processing unit and the AI module include: a data processing unit that receives the roughness and flatness data of the polishing pad from the data transmission module. The data is transmitted to the AI module for algorithm analysis. The AI module, based on machine learning algorithms, analyzes the monitoring data, learns and trains, and predicts the wear trend of the polishing pad, and makes a judgment by comparing with historical data.
[0027] Furthermore, the control and alarm module 13 can automatically generate the best dressing parameters to adjust and control the actuation of the diamond disc swing arm 2. The diamond disc 3 can achieve process optimization with the best effect, improve the accuracy, efficiency, and stability of the CMP process, and send an alarm to notify the operator or stop the equipment operation when an abnormal surface morphology of the polishing pad is detected.
[0028] Furthermore, the material of the fixing seat and the swing arm interface 21 is stainless steel containing iron-chromium alloy, which has the chemical property of acid and alkali corrosion resistance and the physical property of high magnetic attraction. It is SUS430 stainless steel material which is easy to obtain, reasonably priced and has good processability.
[0029] Furthermore, the polishing pad surface roughness monitoring method with AI system is characterized in that: the CMP process combines chemical and physical methods to polish the wafer surface. Under a certain pressure and in the presence of polishing liquid, the wafer and the polishing pad are made to move relative to each other, and the material is peeled off layer by layer from the wafer surface for polishing by means of the organic combination of the mechanical grinding and polishing effect of nano-abrasives and the chemical corrosion effect of oxidants, catalysts, etc., so as to achieve a highly flattened effect. The process is as follows: 1. Fix the wafer 53 under the polishing head wafer holder 52, and place the polishing pad 62 on the polishing disk device 6; 2. The wafer operation device 5 uses the wafer operation transmission shaft 51 to press the rotating wafer 53 onto the rotating polishing pad 62 with a certain pressure, between the surface of the wafer 53 and the rough surface 63 of the polishing pad. The polishing liquid device 4 adds the polishing liquid 43 into the gap through the polishing liquid seat 41 and the polishing liquid dropper 42. The polishing liquid 43 is diffused and transported on the polishing pad 62 and evenly coated under the action of centrifugal force, forming a chemical liquid film between the wafer 53 and the rough surface 63 of the polishing pad; 3. During the polishing process, the rough surface 63 of the polishing pad will inevitably be passivated by chemical corrosion and mechanical pressure, resulting in reduced friction and inability to polish. At the same time, the accumulated abrasive debris will scratch the wafer. However, after detection by the polishing pad surface roughness monitoring diamond disc seat device with AI system, the diamond disc swing arm 2 is automatically started to drive the diamond disc 3 on the interface 21 between the diamond disc seat device and the swing arm to trim the rough surface 63 of the polishing pad to restore its surface roughness. 4. Repeat the alternating process of chemical action of the above chemical film and mechanical friction until flatness is achieved.
[0030] Furthermore, the polishing pad surface roughness monitoring method with an AI system is characterized in that the method can be applied to fields such as semiconductor and integrated circuit manufacturing, precision optical component processing, and high-precision metal surface treatment to improve product yield and process stability.
[0031] Furthermore, the working method program of the polishing pad surface roughness monitoring device with an AI system includes Start => ①: The optical sensor end emits laser, and the optical sensor receiving end receives the laser => ②: The data transmission module preprocesses the data => ③: Reconstruct the surface profile of the polishing pad => ④: Calculate the surface roughness and flatness => ⑤: Through AI learning and analysis => ⑥: Make optimizations and decisions => ⑦: Autonomously control alarm and maintenance => End.
[0032] Furthermore, the method for monitoring the surface roughness of a polishing pad with an AI system is as follows: Its steps are divided into: Step A: Start the optical sensor, perform real-time monitoring on the surface of the polishing pad, and collect roughness and flatness data; Step B: Transmit the collected data to the workbench data processing unit through the data transmission module; Step C: Analyze the data by the AI analysis module, judge the state of the polishing pad, and predict its wear trend; Step D: The adaptive control module receives and adjusts the dressing parameters of the diamond disk according to the AI analysis result to optimize the CMP process; Step E: When an abnormal condition is detected on the surface of the polishing pad, the alarm module issues an alarm signal or automatically stops the process to avoid product defects.
[0033] Compared with the related technologies, a polishing pad surface roughness monitoring device and method with an AI system provided by the present invention have the following beneficial effects: improving the process precision and stability of the process, enhancing production efficiency and reducing costs, simplifying the replacement of the diamond disk and extending its service life.
[0034] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A surface roughness monitoring device for a polishing pad with an AI system, comprising an optical sensor module, a data transmission module, a control and alarm module, a diamond disc fixing buckle, a waterproof cable wire groove, and a diamond disc swing arm, characterized in that: The optical sensor module is arranged on the side of the diamond disc fixing seat, the data transmission module is arranged on the side of the diamond disc fixing seat, and the control and alarm module is arranged on the side of the diamond disc fixing seat; The data transmission module is electrically connected to the optical sensor; The dust-proof and waterproof sealing structure covers the optical sensor module, the data transmission module and the control and alarm module; The diamond disc fixing buckle is arranged on the lower side of the diamond disc swing arm.
2. The surface roughness monitoring device for a polishing pad with an AI system according to claim 1, characterized in that: The optical sensor module, the data transmission module and the control and alarm module are installed on the outer edge of the diamond disc fixing seat at equal intervals of 120 degrees.
3. The surface roughness monitoring device for a polishing pad with an AI system according to claim 2, characterized in that: An optical sensor is embedded inside the optical sensor module. The optical sensor adopts a micro laser triangulation displacement sensor with a measurement range of 1 micron to 1 millimeter and a measurement accuracy of up to ±1 micron.
4. The surface roughness monitoring device for a polishing pad with an AI system according to claim 2, characterized in that: The data transmission module further includes a high-speed wireless transmission module, and the high-speed wireless transmission module selects Wi-Fi 6 or Bluetooth 5.
0.
5. The surface roughness monitoring device of a polishing pad with an AI system according to claim 1 further includes a data processing unit, characterized in that: The data processing unit receives the roughness and flatness data of the polishing pad from the data transmission module, and transmits the data to the AI module for algorithm analysis. The AI module is based on a machine learning algorithm, analyzes and monitors the data for learning and training, and predicts the wear trend of the polishing pad; then transmits the data to the control and alarm module, and the adaptive control module automatically adjusts the dressing parameters of the diamond disc, including but not limited to the pressure, time, rotation speed and path during the dressing process, so as to optimize the process and improve the accuracy, efficiency and stability of the CMP process.
6. The surface roughness monitoring device of a polishing pad with an AI system according to claim 4, characterized in that: The AI analysis module can self-optimize according to the device historical data and real-time data, and continuously improve the accuracy of the polishing pad state judgment and dressing parameter suggestions.
7. A method for monitoring the surface roughness of a polishing pad with an AI system according to claim 6, characterized in that: The AI analysis module includes using a regression model, extracting surface topography features through a convolutional neural network (CNN), and performing data training and predicting the wear trend of the polishing pad through a long short-term memory network (LSTM) or a deep learning model, so as to achieve precise process control.
8. The surface roughness monitoring device of a polishing pad with an AI system according to claim 4, characterized in that: The data processing unit further includes a warning module. When the roughness or flatness of the polishing pad surface exceeds the preset range, this module can issue an alarm in real time to prompt the operator or automatically stop the operation of the CMP device.
9. The surface roughness monitoring device of a polishing pad with an AI system according to claim 1, characterized in that: The diamond disc swing arm is made of SUS430 stainless steel containing iron-chromium alloy, which has both the chemical properties of acid and alkali resistance and the physical property of high magnetic attraction. It is easy to obtain, has a reasonable price and good processability.
10. A method for monitoring the surface roughness of a polishing pad with an AI system according to claim 1, characterized in that: Step A: Start the optical sensor, monitor the surface of the polishing pad in real time, and collect roughness and flatness data; Step B: Transmit the collected data to the workbench data processing unit through the data transmission module; Step C: Analyze the data by the AI analysis module, judge the state of the polishing pad, and predict its wear trend; Step D: The adaptive control module receives and adjusts the dressing parameters of the diamond disc according to the AI analysis result to optimize the CMP process; Step E: When an abnormal condition is detected on the surface of the polishing pad, the alarm module issues an alarm signal or automatically stops the process to avoid product defects.
Citation Information
Cited By
Retainer for online detection of polishing solution state and detection method
CN120791641A