Fragmentation risk active prevention and control system for semiconductor chemical mechanical polishing (CMP)

Abstract

The invention discloses an active prevention and control system and method for fragment risks in a chemical mechanical polishing (CMP) process, and aims to solve the problems of capacity loss and yield reduction caused by a fragment passive response mechanism in the existing chemical mechanical polishing (CMP) process. The core of the invention lies in that early prediction, accurate positioning and automatic disposal of fragment risks are realized through a torque current and friction noise composite monitoring system, so that fragment prevention and control is upgraded from post emergency to pre-blocking.

Description

Technical Field

[0001] The present invention relates to a debris risk active prevention and control system for semiconductor chemical mechanical polishing (CMP). Background Art

[0002] In the semiconductor manufacturing process, chemical mechanical polishing (CMP) is a key step in wafer flattening, especially for 12-inch regenerated wafers, which have large thickness differences and may have film residues of various material coatings on the wafer surface, resulting in differences in friction and pressure on the wafers processed simultaneously on the polishing disk, further exacerbating the risk of debris during the grinding process. Grinding debris not only directly leads to wafer scrapping and material loss, but may also cause equipment contamination, downtime for maintenance, and subsequent process interruptions, resulting in significant economic losses.

[0003] The mainstream response plan in the industry currently focuses on rapid response after debris occurs, such as detecting debris signals through optical sensors or pressure feedback systems and triggering emergency shutdown of equipment to limit the spread of debris and reduce secondary damage. However, such methods are essentially passive remedial measures that cannot fundamentally prevent the occurrence of debris, and the shutdown and troubleshooting process will still lead to production capacity losses.

[0004] In order to reduce the fragmentation rate, existing technologies attempt to start from two aspects: process parameter optimization and equipment improvement: (1) At the process level, mechanical stress can be reduced by adjusting the pressure distribution of the grinding head, optimizing the grinding fluid formula, or reducing the table speed; (2) At the equipment level, introduce vibration suppression devices or real-time thickness monitoring modules to predict potential debris risks by analyzing multi-sensor data during the grinding process. However, the above methods still have the limitation of passive processing. Therefore, how to achieve active prevention and control of debris risks under complex and changeable grinding conditions is still a technical bottleneck that the industry urgently needs to break through. Summary of the invention

[0005] The object of the present invention is to provide a debris risk active prevention and control system for semiconductor chemical mechanical polishing (CMP) to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a debris risk active prevention and control system for semiconductor chemical mechanical polishing (CMP), comprising: The dynamic stress field reconstruction module includes a quantum dot fiber sensor array embedded in the polishing disk to obtain multi-dimensional mechanical data of the wafer surface in real time; an intelligent pressure control module, comprising a grinding head having an independently controllable magnetorheological fluid pressure unit, and dynamically adjusting the local pressure in response to an output signal of the dynamic stress field reconstruction module; Adaptive lubrication module, including a grinding fluid delivery system loaded with nano-lubricant, triggering the directional release of nano-lubricant according to the real-time friction state; The collaborative control unit synchronously coordinates the operation timing and parameters of the dynamic stress field reconstruction module, the intelligent pressure regulation module and the adaptive lubrication module by integrating the stress distribution prediction model and the reinforcement learning algorithm.

[0007] Preferably, the quantum dot fiber optic sensor array is arranged at a pitch of 5-20 μm, and each sensor node integrates a quantum tunneling effect enhanced probe, which can detect dynamic stress changes in the range of 0.05-1N and simultaneously collect temperature and vibration signals.

[0008] Preferably, the number of magnetorheological fluid pressure units is 512-2048, and each unit comprises: A magnetorheological fluid cavity, the interior of which is filled with a magnetorheological fluid having a magnetorheological particle concentration of 30%-60%; a miniature electromagnetic coil, arranged around the cavity, generating a local magnetic field of 0.1-1T in response to a control current; The pressure conduction diaphragm converts the viscosity change of the magnetorheological fluid caused by the magnetic field into a local pressure output of 0-50kPa.

[0009] Preferably, the nanolubricant is a core-shell structure nanocapsule, the core of which comprises heterojunction nanosheets of molybdenum disulfide and graphene, and the shell is a pH-responsive copolymer, and the critical response temperature is 40-50°C.

[0010] Preferably, the adaptive lubrication module further comprises a surface acoustic wave generator, the operating frequency of which is 10-100 MHz, for inducing directional migration and rupture release of the nanocapsules at the grinding interface.

[0011] Preferably, the collaborative control unit performs the following operations: a) Based on the time reversal algorithm, a dynamic stress cloud map of the wafer-polishing pad contact interface is generated every 20-100 μs; b) Predict the debris risk area within the next 200-500ms using a thermal-mechanical coupled finite element model; c) Dynamically allocate the action modes of the magnetorheological fluid pressure unit according to the risk level, including gradient pressure reduction, peripheral compensation and pulse buffering modes.

[0012] Preferably, a systematic CMP process debris prevention and control method is characterized by comprising the steps of: S1. During the polishing process, the wafer surface stress, temperature and vibration parameters are monitored in real time through a quantum dot fiber sensor array; S2. Use time reversal algorithm to reconstruct the dynamic stress field and identify stress concentration areas and potential debris risk locations; S3. According to the risk position coordinates, control the corresponding magnetorheological fluid pressure unit to adjust the local pressure within 2-10ms; S4. When the local friction temperature is detected to exceed a threshold, the surface acoustic wave generator is activated and the nano-lubricant is released; S5. Iteratively optimize the pressure distribution strategy and lubricant release logic based on historical data.

[0013] Preferably, the pressure adjustment in step S3 includes: Implement linear pressure reduction in the stress concentration area, with a pressure reduction rate of 5-20kPa / ms; The compensatory pressure increase is started within a radius of 1-3mm around the stress concentration area, and the pressure increase amplitude is 10%-30% of the base pressure.

[0014] Preferably, it comprises by weight percentage: MoS2 / graphene heterojunction nanosheets 15%-25%; pH responsive styrene-acrylic acid copolymer 30%-45%; Surfactant 5%-10%; The balance is deionized water and organic solvent; The nanosheet has a thickness of 1-5 nm, a lateral size of 50-200 nm, and is surface-modified with carboxyl functional groups.

[0015] Preferably, the glass transition temperature (Tg) of the pH responsive copolymer is 45-55°C, and swelling and rupture occur when pH <4 or pH >9.

[0016] Compared with the prior art, the beneficial effects of the present invention are: the present invention improves debris detection from "after-the-fact response" to "pre-prediction", and the warning time window is expanded by 3 orders of magnitude; through the spatiotemporal coordination of nano-lubrication and stress field, the stress unevenness on the wafer surface is reduced to 18% of the traditional process; the dynamic control system can uniformly control the fragmentation rate of wafers of different thicknesses to below 0.02‰, which improves the reliability by 20 times compared with the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a process flow chart of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of 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.

[0019] See also Figure 1 The present invention provides a technical solution: a debris risk active prevention and control system for semiconductor chemical mechanical polishing (CMP), comprising: The dynamic stress field reconstruction module includes a quantum dot fiber sensor array embedded in the polishing disk to obtain multi-dimensional mechanical data of the wafer surface in real time; an intelligent pressure control module, comprising a grinding head having an independently controllable magnetorheological fluid pressure unit, and dynamically adjusting the local pressure in response to an output signal of the dynamic stress field reconstruction module; Adaptive lubrication module, including a grinding fluid delivery system loaded with nano-lubricant, triggering the directional release of nano-lubricant according to the real-time friction state; The collaborative control unit synchronously coordinates the operation timing and parameters of the dynamic stress field reconstruction module, the intelligent pressure regulation module and the adaptive lubrication module by integrating the stress distribution prediction model and the reinforcement learning algorithm.

[0020] Preferably, the quantum dot fiber optic sensor array is arranged at a pitch of 5-20 μm, and each sensor node integrates a quantum tunneling effect enhanced probe, which can detect dynamic stress changes in the range of 0.05-1N and simultaneously collect temperature and vibration signals.

[0021] Preferably, the number of magnetorheological fluid pressure units is 512-2048, and each unit comprises: A magnetorheological fluid cavity, the interior of which is filled with a magnetorheological fluid having a magnetorheological particle concentration of 30%-60%; a miniature electromagnetic coil, arranged around the cavity, generating a local magnetic field of 0.1-1T in response to a control current; The pressure conduction diaphragm converts the viscosity change of the magnetorheological fluid caused by the magnetic field into a local pressure output of 0-50kPa.

[0022] Preferably, the nanolubricant is a core-shell structure nanocapsule, the core of which comprises heterojunction nanosheets of molybdenum disulfide and graphene, and the shell is a pH-responsive copolymer, and the critical response temperature is 40-50°C.

[0023] Preferably, the adaptive lubrication module further comprises a surface acoustic wave generator, the operating frequency of which is 10-100 MHz, for inducing directional migration and rupture release of the nanocapsules at the grinding interface.

[0024] Preferably, the collaborative control unit performs the following operations: a) Based on the time reversal algorithm, a dynamic stress cloud map of the wafer-polishing pad contact interface is generated every 20-100 μs; b) Predict the debris risk area within the next 200-500ms using a thermal-mechanical coupled finite element model; c) Dynamically allocate the action modes of the magnetorheological fluid pressure unit according to the risk level, including gradient pressure reduction, peripheral compensation and pulse buffering modes.

[0025] Preferably, a systematic CMP process debris prevention and control method is characterized by comprising the steps of: S1. During the polishing process, the wafer surface stress, temperature and vibration parameters are monitored in real time through a quantum dot fiber sensor array; S2. Use time reversal algorithm to reconstruct the dynamic stress field and identify stress concentration areas and potential debris risk locations; S3. According to the risk position coordinates, control the corresponding magnetorheological fluid pressure unit to adjust the local pressure within 2-10ms; S4. When the local friction temperature is detected to exceed a threshold, the surface acoustic wave generator is activated and the nano-lubricant is released; S5. Iteratively optimize the pressure distribution strategy and lubricant release logic based on historical data.

[0026] Preferably, the pressure adjustment in step S3 includes: Implement linear pressure reduction in the stress concentration area, with a pressure reduction rate of 5-20kPa / ms; The compensatory pressure increase is started within a radius of 1-3mm around the stress concentration area, and the pressure increase amplitude is 10%-30% of the base pressure.

[0027] Preferably, it comprises by weight percentage: MoS2 / graphene heterojunction nanosheets 15%-25%; pH responsive styrene-acrylic acid copolymer 30%-45%; Surfactant 5%-10%; The balance is deionized water and organic solvent; The nanosheet has a thickness of 1-5 nm, a lateral size of 50-200 nm, and is surface-modified with carboxyl functional groups.

[0028] Preferably, the glass transition temperature (Tg) of the pH responsive copolymer is 45-55°C, and swelling and rupture occur when pH <4 or pH >9.

[0029] Example 1: 12-inch regenerated wafer CMP process System Configuration The polishing plate has a diameter of 810 mm and is embedded with 5,120 quantum dot fiber optic sensors; The grinding head is equipped with 1024 magnetorheological fluid units, with a unit diameter of 2 mm; The amount of nano-lubricant added is 3% of the total mass of the grinding fluid.

[0030] Operation process Step 1: Start the grinding machine, and the quantum dot sensor collects stress data at a frequency of 20kHz; Step 2: The edge computing node generates a dynamic stress cloud map through a time inversion algorithm to identify areas where the stress exceeds the threshold (>8kPa); Step 3: The DRL model calculates the pressure adjustment strategy, applying a gradient pressure reduction (rate 10 kPa / ms) to the stress concentration area, while increasing the pressure of the surrounding units by 15%; Step 4: When the temperature sensor detects that the local temperature reaches 47°C, the surface acoustic wave is activated to trigger the nanocapsule to rupture and release the lubricant; Step 5: After each wafer is polished, the system automatically updates the process parameter mapping table and optimizes the control logic for the next wafer.

[0031] Effect: After testing 500 regenerated wafers, the fragmentation rate was reduced from 0.5‰ in the traditional process to 0.02‰; The wafer surface stress non-uniformity (standard deviation) was reduced from 12.3 kPa to 2.2 kPa.

[0032] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A debris risk active prevention and control system for semiconductor chemical mechanical polishing (CMP), characterized in that: include: The dynamic stress field reconstruction module includes a quantum dot fiber sensor array embedded in the polishing disk to obtain multi-dimensional mechanical data of the wafer surface in real time; an intelligent pressure control module, comprising a grinding head having an independently controllable magnetorheological fluid pressure unit, and dynamically adjusting the local pressure in response to an output signal of the dynamic stress field reconstruction module; Adaptive lubrication module, including a grinding fluid delivery system loaded with nano-lubricant, triggering the directional release of nano-lubricant according to the real-time friction state; The collaborative control unit synchronously coordinates the operation timing and parameters of the dynamic stress field reconstruction module, the intelligent pressure regulation module and the adaptive lubrication module by integrating the stress distribution prediction model and the reinforcement learning algorithm.

2. The system according to claim 1, characterized in that The quantum dot optical fiber sensor array is arranged at a pitch of 5-20 μm, and each sensor node integrates a quantum tunneling effect enhanced probe, which can detect dynamic stress changes in the range of 0.05-1N and synchronously collect temperature and vibration signals.

3. The system according to claim 1, characterized in that The number of the magnetorheological fluid pressure units is 512-2048, and each unit comprises: A magnetorheological fluid cavity, the interior of which is filled with a magnetorheological fluid having a magnetorheological particle concentration of 30%-60%; a miniature electromagnetic coil, arranged around the cavity, generating a local magnetic field of 0.1-1T in response to a control current; The pressure conduction diaphragm converts the viscosity change of the magnetorheological fluid caused by the magnetic field into a local pressure output of 0-50kPa.

4. The system according to claim 1, characterized in that The nano-lubricant is a core-shell structure nano-capsule, the core of which contains heterojunction nano-sheets of molybdenum disulfide and graphene, and the shell is a pH-responsive copolymer, and the critical response temperature is 40-50°C.

5. The system according to claim 4, characterized in that The self-adaptive lubrication module also includes a surface acoustic wave generator, whose operating frequency is 10-100 MHz, which is used to induce the directional migration and rupture release of the nanocapsules at the grinding interface.

6. The system according to claim 1, characterized in that The collaborative control unit performs the following operations: a) Based on the time reversal algorithm, a dynamic stress cloud map of the wafer-polishing pad contact interface is generated every 20-100 μs; b) Predict the debris risk area within the next 200-500ms using a thermal-mechanical coupled finite element model; c) Dynamically allocate the action modes of the magnetorheological fluid pressure unit according to the risk level, including gradient pressure reduction, peripheral compensation and pulse buffering modes.

7. A method for preventing and controlling CMP process debris based on the system of any one of claims 1 to 6, characterized in that: Includes steps: S1. During the polishing process, the wafer surface stress, temperature and vibration parameters are monitored in real time through a quantum dot fiber sensor array; S2. Use time reversal algorithm to reconstruct the dynamic stress field and identify stress concentration areas and potential debris risk locations; S3. According to the risk position coordinates, control the corresponding magnetorheological fluid pressure unit to adjust the local pressure within 2-10ms; S4. When the local friction temperature is detected to exceed a threshold, the surface acoustic wave generator is activated and the nano-lubricant is released; S5. Iteratively optimize the pressure distribution strategy and lubricant release logic based on historical data.

8. The method according to claim 7, characterized in that The pressure adjustment in step S3 includes: Implement linear pressure reduction in the stress concentration area, with a pressure reduction rate of 5-20kPa / ms; The compensatory pressure increase is started within a radius of 1-3mm around the stress concentration area, and the pressure increase amplitude is 10%-30% of the base pressure.

9. A self-repairing nano-lubricant for CMP process, characterized in that: Contains by weight: MoS2 / graphene heterojunction nanosheets 15%-25%; pH responsive styrene-acrylic acid copolymer 30%-45%; Surfactant 5%-10%; The balance is deionized water and organic solvent; The nanosheet has a thickness of 1-5 nm, a lateral size of 50-200 nm, and is surface-modified with carboxyl functional groups.

10. The nano-lubricant according to claim 9, characterized in that: The pH responsive copolymer has a glass transition temperature (Tg) of 45-55° C. and swells and ruptures when pH is less than 4 or when pH is greater than 9.

Images