Three-step fine polishing method for 19nm wafer regeneration

Through a three-step fine polishing method, combined with gradient-decreasing chemical mechanical force and dynamic feedback control, the technical contradiction between macroscopic flatness and microscopic integrity in 19nm wafer regeneration is resolved, achieving ultra-low defect density and chemically stable wafer surface, supporting the high efficiency and reliability of semiconductor manufacturing.

CN120696913APending Publication Date: 2025-09-26ANHUI FULLERDE CHANGJIANG SEMICON MATERIALS CO LTD

Patent Information

Application Number
CN202511061570.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-29
Publication Date
2025-09-26

AI Technical Summary

Technical Problem

Existing wafer polishing technology has difficulty balancing macroscopic flatness and microscopic surface integrity when processing 19nm high-process recycled wafers, resulting in microscopic defects and subsurface damage, affecting the electrical performance and reliability of semiconductor devices.

Method used

A three-step fine polishing method is adopted to remove the accumulated defects of the regenerated wafer and repair the micro damage in stages through gradient-decreasing chemical mechanical force, dynamic feedback control and surface energy state regulation. It includes the first fine polishing step, the second fine polishing step and the third fine polishing step, each using specific hardware and consumables configuration, and supplemented by online process monitoring and adaptive feedback control.

Benefits of technology

The surface quality of the regenerated wafers has reached the ultra-low defect density and chemical stability required by the 19nm process node, ensuring the high efficiency and reliability of semiconductor manufacturing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120696913A_ABST
    Figure CN120696913A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of three-step fine polishing methods for 19nm wafer regeneration, discloses a fine polishing method named'three-step fine polishing method for 19nm wafer regeneration ', and aims to solve the technical contradiction between macroscopic planarization and microscopic surface integrity of regenerated wafers in advanced process nodes. The method is characterized by comprising a first fine polishing step, a second fine polishing step and a third fine polishing step, and in each step, gradient decreasing polishing pad hardness, abrasive particle size and concentration, polishing pressure and pH value are adopted; and an on-line process monitoring and self-adaptive feedback control subsystem is used for assistance. By adopting the technical scheme, defects can be efficiently removed, damage can be repaired step by step, the wafer surface with ultralow defect density and excellent chemical stability is obtained, the requirements of the manufacturing process of 19nm and more advanced semiconductors are met, and the yield is increased.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of semiconductor technology, and in particular relates to a three-step fine polishing method for 19nm wafer regeneration. Background Art

[0002] As the semiconductor industry evolves in line with Moore's Law, the feature sizes of integrated circuits have entered the nanometer scale. Advanced process nodes such as 19nm place unprecedented demands on the control precision of various materials and process parameters during the manufacturing process. Against this backdrop, wafer reconditioning technology, a key link in the semiconductor manufacturing industry chain, has become increasingly important. This technology primarily treats and regenerates control wafers and test wafers used for process monitoring or equipment verification during production, enabling them to be reused. This significantly reduces wafer fabrication plant (FAB) operating costs and improves the efficiency of critical resource utilization without compromising product quality. Accordingly, the technical requirements for wafer reconditioning have also risen accordingly, evolving from its early service capabilities for mature processes such as 60nm and 90nm to gradually extending to 19nm and even more advanced process nodes. This sets extremely stringent standards for the surface quality of reconditioned wafers, particularly surface roughness and micro-defect density.

[0003] In the existing field of wafer polishing technology, various solutions have been disclosed for improving wafer surface flatness. For example, patent publication number CN101670546B discloses a semiconductor wafer polishing method that processes both the front and back sides of the wafer using chemical mechanical polishing (CMP). The core of this technical concept is to differentially control the amount of material removed from the front and back sides based on the profile variation of the wafer diameter, in order to achieve better overall thickness uniformity. This profile-controlled material removal strategy had positive practical significance at the time for addressing macroscopic deformation of the wafer caused by processing stress and improving overall flatness. Another technical solution, publication number CN111823120B, achieved more precise control of wafer geometric parameters by designing a double-sided polishing device with deformation adjustment capabilities, and made significant progress in improving polishing efficiency. This solution focuses on actively compensating and adjusting the pressure distribution during the polishing process through a sophisticated mechanical structure and control system, effectively improving key geometric indicators such as the total thickness variation (TTV) and bow / warp of the wafer. When these technologies were first proposed, they all provided effective solutions to specific problems in wafer manufacturing at the time.

[0004] However, with the continued penetration of related technologies into ultra-high-precision nodes such as 19nm and the specialized application scenarios (i.e., wafer regeneration), some inherent design characteristics of the above-mentioned technical solutions have gradually revealed deep-seated limitations in meeting new challenges. The reason is that these technical solutions are essentially more focused on correcting the wafer's macroscopic geometry and optimizing the material removal rate. This optimization strategy exists in an inherent technical contradiction with the extreme requirements of the 19nm node for the microscopic state of the wafer surface. Specifically, to achieve efficient material removal and macroscopic planarization, existing technologies generally use a relatively simple and high-intensity chemical mechanical polishing process parameter combination. Although this process can effectively improve the wafer's geometric parameters, its high polishing pressure and rotation speed, as well as the polishing pad or polishing slurry with strong cutting ability, can easily introduce new microscopic defects on the wafer surface and subsurface layers, such as micro-scratch, residual stress, and more subtle sub-surface damage (SSD). At larger process nodes like 60nm and 90nm, the size of these microscopic defects is still tiny relative to device feature sizes, and their impact on device electrical performance is within acceptable limits. However, as the technology node advances to 19nm, transistor gate dimensions and metal wiring pitches have been reduced to their limits. Any tiny surface undulations or lattice damage on the wafer can become a fatal weakness in electrical performance, directly impacting the quality of subsequent thin film deposition and the focusing accuracy of the photolithography process, ultimately leading to serious problems such as increased device leakage and reduced reliability. This contradiction is further amplified in the field of wafer reprocessing. Reprocessed wafers are not pristine silicon wafers. Their surface and subsurface layers have accumulated a more complex and deep-seated defect history due to multiple complex process cycles such as thin film deposition, etching, and ion implantation. A single polishing strategy, primarily focused on macroscopic planarization, is unable to completely eradicate these deep-seated defects. In fact, while removing existing defects, this "one-size-fits-all" processing approach may even introduce new, more difficult-to-treat microscopic damage, creating a cycle of treating the symptoms but not the root cause.

[0005] In summary, the core challenge faced by existing wafer polishing technology when processing 19nm high-process recycled wafers is no longer a simple flattening efficiency or macro-geometry control problem, but how to build a step-by-step, refined process system that can take into account both macro-defect removal and micro-surface integrity repair. This system needs to be able to efficiently remove the cumulative surface and sub-surface damage unique to recycled wafers, while at the same time systematically repairing the micro-texture and damage that may be introduced by the previous process in subsequent steps, and ultimately achieve ultra-low surface roughness and particle residue levels of less than 100ea that meet the requirements of advanced nodes. Therefore, how to design a new type of fine polishing method that works together through multi-step process logic and differentiated parameter configuration to systematically solve the above-mentioned deep-seated technical contradictions has become a key technical problem that technicians in this field urgently need to overcome. Summary of the Invention

[0006] The purpose of the present invention is to overcome the above-mentioned defects of the prior art and provide a three-step fine polishing method for 19nm wafer regeneration, which aims to systematically solve the deep-seated technical contradiction between macroscopic flatness and microscopic surface integrity of regenerated wafers at advanced process nodes. In order to achieve the above-mentioned purpose of the invention, the present invention provides a three-step fine polishing method for 19nm wafer regeneration, which constructs an integrated process system including gradient-decreasing chemical mechanical force, dynamic feedback control and surface energy state regulation, while efficiently removing the cumulative defects of the regenerated wafer, and gradually repairing the microscopic damage that may be introduced by the previous process, and finally generating a wafer surface with ultra-low defect density and excellent chemical stability, so as to meet the stringent requirements of 19nm and more advanced semiconductor manufacturing processes for regenerated wafers.

[0007] The core of the technical solution disclosed in this invention lies in deconstructing the fine polishing process into three independent steps with progressive functionality and precise parameter coupling, supported by a consistent online process control subsystem. These three steps—the first, second, and third fine polishing steps—are performed in separate chemical mechanical polishing units, each with differentiated hardware and consumables configurations tailored to specific surface restoration objectives.

[0008] Specifically, the core purpose of the first fine polishing step of the present invention is to efficiently and deeply remove material from the regenerated wafer after front-end stripping and rough polishing, thereby completely eradicating the complex film components, embedded particles, and lattice damage regions accumulated in the subsurface layer from multiple heat treatment and ion implantation cycles, known as subsurface damage (SSD). To achieve this goal, the first fine polishing step is performed on a first fine polishing unit. The first fine polishing unit is equipped with a first fine polishing pad characterized by being a rigid polyurethane polishing pad with a specific macro-groove structure. The macro-groove structure is designed as a concentric array with a rectangular cross-section. The depth is precisely controlled between 0.45 mm and 0.55 mm, the width is controlled between 0.4 mm and 0.6 mm, and the center-to-center spacing (i.e., pitch) between adjacent grooves is set to 2.0 mm to 3.0 mm. The presence of this groove structure significantly enhances the transport and exchange efficiency of the polishing liquid at the interface between the wafer and the polishing pad, ensuring that fresh chemical reactants continuously reach the reaction interface and efficiently removes reaction products and ground particles from the polishing area, thereby maintaining a stable and efficient material removal rate. Furthermore, the intermittent contact surface formed by the grooves increases the local pressure on the wafer surface on a macroscale, enhancing the mechanical grinding action and playing a decisive role in removing deep subsurface damage. The main material of the first fine polishing pad is rigid polyurethane with a Shore D hardness set in the range of 55 to 65. This high hardness ensures that the polishing pad is less susceptible to elastic deformation when subjected to high polishing pressure, thereby maintaining the overall flatness of the wafer and effectively controlling the total thickness variation (TTV).

[0009] In conjunction with the first fine polishing pad, the first fine polishing step uses a first fine polishing liquid. The first fine polishing liquid is a highly alkaline chemical mechanical polishing liquid using colloidal silica as an abrasive. The average primary particle size of the colloidal silica abrasive is precisely controlled within the range of 50 to 70 nanometers, and its particle size distribution presents a unimodal normal distribution with a standard deviation of less than 5 nanometers. This relatively large abrasive particle size provides strong mechanical cutting capabilities, enabling effective physical grinding of silicon substrates. The mass concentration of the abrasive in the polishing liquid is set to 20% to 30%. In order to maximize the chemical removal rate, the pH value of the first fine polishing liquid is precisely controlled within the strong alkaline range of 10.5 to 11.5 by adding a strong alkaline regulator such as potassium hydroxide (KOH) or tetramethylammonium hydroxide (TMAH). At this pH value, the chemical reaction rate of silicon and hydroxide ions reaches its peak, forming an efficient synergy with the mechanical action of the abrasive, thereby achieving efficient removal of wafer material.

[0010] In the process parameter setting of the first fine polishing step, in order to achieve the above-mentioned defect removal goal, a higher mechanical energy input is adopted. Specifically, the polishing pressure applied to the back of the wafer is set to 3.0 pounds per square inch (PSI) to 4.0 pounds per square inch; the polishing disk speed of the first fine polishing unit is set to 80 revolutions per minute (RPM) to 100RPM; the speed of the polishing head carrying the wafer is set to 80RPM to 95RPM per minute; the supply rate of the first fine polishing liquid is set to 180 ml to 220 ml per minute. Through the specific combination of the above-mentioned hardware, consumables and process parameters, the first fine polishing step can effectively remove the most stubborn defects in the surface and sub-surface layer of the regenerated wafer within a specified time, laying a clean and uniform foundation for subsequent fine repair.

[0011] Furthermore, the core purpose of the second fine polishing step of the present invention is to systematically repair the new microscopic surface topography introduced by the first fine polishing step. These topography primarily includes the microtexture left on the wafer surface by the hard grooved pad and the fine scratches caused by the larger abrasive. To achieve this goal, the wafers after the first fine polishing step are transferred to a second fine polishing unit. This second fine polishing unit is equipped with a second fine polishing pad, which differs from the first fine polishing pad in that it is a medium-hard microporous polyurethane polishing pad without a macroscopic groove structure. This groove-free design avoids the reproduction of any macroscopic patterns on the wafer surface, resulting in a more uniform polishing force at the microscale. The second fine polishing pad is made of polyurethane, with a Shore D hardness set between 45 and 55, lower than that of the first fine polishing pad. This lower hardness provides the polishing pad with better surface conformability, allowing it to better conform to the microscopic undulations of the wafer surface, thereby effectively flattening the microtexture. At the same time, its interior has a rich open microporous structure with an average pore size of 25 microns to 35 microns. These micropores serve as microscopic channels for storing and transporting polishing liquid, ensuring the uniform distribution of liquid on the polishing interface.

[0012] In conjunction with the second fine polishing pad, the second fine polishing step utilizes a second fine polishing slurry. This second fine polishing slurry differs significantly from the first fine polishing slurry in its compositional design. Its core feature is the use of colloidal silica abrasive with a bimodal particle size distribution. Specifically, the abrasive system comprises a blend of two colloidal silica particles of varying particle sizes: the primary abrasive, which accounts for 80% to 90% of the total abrasive mass, has an average primary particle size within a narrow range of 20 to 30 nanometers and is used for fine smoothing the wafer surface. The secondary abrasive, which accounts for 10% to 20% of the total abrasive mass, has an average primary particle size within a range of 40 to 50 nanometers and is used to efficiently remove deep microscratches and surface textures left over from the first step. This synergistic effect of the bimodal particle size distribution achieves both reduced surface roughness and efficient removal of specific microscopic defects. The total abrasive mass concentration in the second fine polishing slurry is reduced to 12% to 18%. At the same time, the pH value is adjusted by using highly volatile ammonia water (NH4OH) and controlled within the range of 9.5 to 10.5, which is lower than that in the first step, in order to reduce the chemical corrosion rate and allow the mechanical smoothing effect to dominate.

[0013] In the process parameter settings for the second fine polishing step, the overall mechanical energy input is significantly reduced compared to the first step. Specifically, the polishing pressure applied to the back of the wafer is reduced to 2.0PSI to 2.8PSI; the polishing disk speed of the second fine polishing unit is reduced to 60RPM to 75RPM per minute; the polishing head speed is reduced to 55RPM to 70RPM per minute; and the supply rate of the second fine polishing liquid is increased to 230ml to 270ml per minute to ensure sufficient lubrication and chemical action at lower pressures. Through the refined treatment of the second fine polishing step, the roughness of the wafer surface is greatly reduced, the microscopic damage introduced in the first step is effectively repaired, and the surface state tends to be smooth and uniform.

[0014] Furthermore, the third fine polishing step of the present invention functions as the final surface modification and passivation treatment. Its core purpose is no longer material removal or roughness reduction, but rather restoring the wafer surface to a physically and chemically "perfect" state. This state specifically includes achieving atomic-level surface flatness, controlling the number of residual surface particles below a target threshold (for the 19nm node, this means fewer than 100 particles larger than 19nm), and constructing a surface passivation layer with low surface energy and high hydrophilicity to inhibit re-adsorption of particles from the environment during subsequent cleaning, drying, and transport. To achieve this ultimate goal, wafers that have undergone the second fine polishing step are transferred to a third fine polishing unit. This third fine polishing unit is equipped with a third fine polishing pad, a soft, dense, microporous polyurethane polishing pad without a macrogrooved structure. Its Shore D hardness has been further reduced to the range of 35 to 45, making it the softest polishing pad of the three steps. Its extreme softness enables maximum atomic-level contact with the wafer surface, and the polishing force is transmitted almost entirely through the liquid, achieving a pure hydrodynamic polishing or chemical glow polishing mode. Its internal micropore structure is more dense, with an average pore size of less than 20 microns, to support the formation of a stable and uniform ultra-thin liquid film.

[0015] In conjunction with the third fine polishing pad, the third fine polishing step uses a third fine polishing liquid, which has an original formulation. It is characterized in that it is a chemically modified liquid with ultra-low abrasive concentration or completely no abrasive. In a preferred embodiment, the polishing liquid contains colloidal silica abrasive with an average primary particle size of less than 10 nanometers, and its mass concentration is strictly controlled at an extremely low level of 0.5% to 1.5%. Its function is only a slight glow wipe, not mechanical grinding. More importantly, a specific surface passivator is added to the third fine polishing liquid. The surface passivator is a cationic surfactant, specifically dodecyltrimethylammonium bromide (DTAB) or hexadecyltrimethylammonium chloride (CTAC), and its added concentration in the polishing liquid is 0.01% to 0.05% (mass percentage). During the polishing process, the hydrophobic carbon chain end of the cationic surfactant will generate electrostatic repulsion with the similarly negatively charged silica abrasive (if present) and the silicon surface, while its hydrophilic quaternary ammonium salt head will be physically adsorbed on the silicon surface after polishing, forming a single molecule or sub-single molecule temporary hydrophilic passivation layer. This passivation layer significantly improves the hydrophilicity of the wafer surface (manifested as a water contact angle of less than 5 degrees) and changes the Zeta potential of the surface, so that an electrostatic repulsion force is generated between the surface and the negatively charged particles in the solution, thereby greatly inhibiting the redeposition of the particles at the moment of polishing. At the same time, the pH value of the third fine polishing liquid is precisely controlled in a near-neutral range of 7.5 to 8.5 by using a non-ionic buffer (such as Tris buffer) to minimize any chemical corrosion and ensure the final surface integrity.

[0016] The process parameters for the third fine polishing step are set to minimize mechanical energy input. Specifically, the polishing pressure applied to the wafer backside is set to 1.0 to 1.8 PSI; the polishing disk speed is set to 30 to 45 RPM; and the polishing head speed is set to 25 to 40 RPM. The supply rate of the third fine polishing solution is further increased to 280 to 320 ml per minute to ensure stable hydrodynamic support.

[0017] As a core technical feature of the present invention, in order to ensure the accuracy and robustness of the above-mentioned three-step gradient-decreasing repair process, the present invention also includes an online process monitoring and adaptive feedback control subsystem. The subsystem is configured on the wafer conveying path between adjacent fine polishing units, or integrated in the non-working area of ​​the polishing unit. The subsystem includes an optical measurement module and a central process control module. The optical measurement module includes a laser light source, which emits a beam of collimated laser with a central wavelength of 633 nanometers, which is irradiated at a preset incident angle on the surface of the wafer that has just completed the previous polishing step. An array detector composed of multiple photodiodes is configured to receive scattered light from the wafer surface. Among them, the central detector located in the direction of the mirror reflection angle is used to measure the intensity of the reflected light to evaluate the surface roughness; and the detectors distributed at different angles around are used to measure the angular distribution of the scattered light intensity to calculate the haze value of the surface.

[0018] The central process control module is configured to receive and process the detection data from the optical measurement module in real time. Before the wafer completes the first fine polishing step and is transferred to the second fine polishing unit, the system performs the first optical measurement on the wafer to obtain a first intermediate surface parameter group (including roughness value R1 and haze value H1). The central process control module compares R1 and H1 with the preset first target range.

[0019] [R target1_min ,R target1_max ] and [H target1_min ,H target1_max ] is compared. If the measured value falls within the target range, the second fine polishing step is performed according to the preset standard process parameters. If the measured value deviates from the target range, the central process control module will dynamically adjust one or more key process parameters of the second fine polishing step according to a preset control algorithm. For example, if the measured haze value H1 is higher than the target upper limit, indicating that the removal of sub-surface damage in the first step is not complete, the control module will instructively increase the execution time or polishing pressure of the second fine polishing step, and the adjustment amount is proportional to the degree of deviation. The control algorithm can be specifically a proportional-integral-differential (PID) controller, the input of which is the error between the measured value and the target value, and the output is the adjustment amount for the downstream process parameters. Similarly, after the second fine polishing step is completed and before entering the third fine polishing step, a second optical measurement is performed to obtain a second intermediate surface parameter group (R2, H2), and based on the comparison result with the second target range, the process parameters of the third fine polishing step are fine-tuned.

[0020] By introducing this online process monitoring and adaptive feedback control subsystem, the method of the present invention is no longer a fixed, open-loop process, but a dynamic, closed-loop intelligent repair system. It can sense and respond in real time to the state differences of incoming regenerated wafers and any minor fluctuations that may exist between process steps, ensuring that the surface condition of each wafer, upon leaving the current step, converges to a narrow process window optimized for the next step. This significantly improves the ultimate consistency, yield, and process robustness of the entire three-step fine polishing process.

[0021] Beneficial effects of the present invention: The present invention adopts an innovative three-step gradient-decreasing fine polishing method, which combines differentiated hardware configurations, gradient consumables formulations, step-by-step decreasing process parameters and closed-loop feedback control throughout the entire process. The method first effectively removes deep damage through the first step of high energy input, then repairs micro-textures and scratches through the second step of medium energy input, and finally achieves atomic-level flatness and surface passivation through the third step of minimum energy input. This systematic, divide-and-conquer design concept accurately solves the inherent contradiction between efficient removal and fine repair in the existing technology, so that the surface quality of the regenerated wafer can stably reach the ultra-high quality standard required by the 19nm process node, with the number of particles larger than 19nm being less than 100, providing solid technical support for the sustainable development of the semiconductor manufacturing industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic flow chart of a three-step fine polishing method for 19nm wafer regeneration according to the present invention;

[0023] Figure 2 It is a system structure block diagram of the present invention for implementing the method;

[0024] Figure 3 It is a structural schematic diagram of the first fine polishing unit in the present invention;

[0025] Figure 4 It is a schematic structural diagram of the second fine polishing unit in the present invention;

[0026] Figure 5 Schematic diagram of the structure of the third fine polishing unit in the present invention;

[0027] Figure 6 It is a schematic diagram of the principle of the online process monitoring and adaptive feedback control subsystem in the present invention.

[0028] The figure numbers are as follows: 10, first fine polishing unit; 11, first fine polishing pad; 20, second fine polishing unit; 21, second fine polishing pad; 30, third fine polishing unit; 31, third fine polishing pad; 40, online process monitoring and adaptive feedback control subsystem; 41, optical measurement module; 411, laser light source; 412, array detector; 42, central process control module. DETAILED DESCRIPTION

[0029] To make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0030] See also Figures 1 to 6 This paper discloses a three-step fine polishing method for 19nm wafer regeneration. Its core is to collaboratively address complex surface and subsurface defects on regenerated wafers through a systematic, phased process flow. This method deconstructs the entire fine polishing process into three functionally interconnected steps with progressively decreasing process parameters. This method, supplemented by a consistent closed-loop process control mechanism, achieves a precise balance between macroscopic flatness and microscopic integrity.

[0031] Reference Figure 1 and Figure 2 The overall process flow and system architecture shown in the figure begins with the reclaimed wafer to be processed, typically a 300mm silicon wafer that has undergone front-end chemical stripping, mechanical grinding, and rough polishing. It first enters the first fine polishing unit 10. Within this unit, the wafer undergoes the first fine polishing step, whose primary goal is to efficiently remove material from the wafer to eradicate deep sub-surface damage (SSD) and residual contaminants. After completing the first fine polishing step, the wafer is transferred to the online process monitoring and adaptive feedback control subsystem 40 for a first intermediate surface quality inspection. Based on the inspection results, this subsystem 40 dynamically adjusts the process parameters for the upcoming second fine polishing step. The wafer then enters the second fine polishing unit 20 for a second fine polishing step, which aims to repair microscratches and surface texture that may have been introduced in the first step and significantly reduce surface roughness. After completing the second fine polishing step, the wafer undergoes a second intermediate surface quality inspection again by the online process monitoring and adaptive feedback control subsystem 40, which fine-tunes the parameters for the third fine polishing step accordingly. Finally, the wafer enters the third fine polishing unit 30 for final surface modification and passivation, known as the third fine polishing step. The goal of this step is to achieve atomic-level surface flatness, minimize surface particle contamination, and create a hydrophilic surface with excellent chemical stability.

[0032] Specifically, the first fine polishing step is performed in a first fine polishing unit 10, the detailed structure of which can be referred to in Figure 3 . The first fine polishing unit 10 is a standard chemical mechanical polishing equipment, and its core components include a rotatable polishing disc, a polishing head for bearing and applying pressure, and a polishing liquid supply system. In the specific implementation of the present invention, the first fine polishing unit 10 is equipped with a first fine polishing pad 11. The physical properties and structural morphology of the first fine polishing pad 11 are the key to achieving the technical goals of the first step. Its main material is made of hard polyurethane, and its Shore D hardness is precisely controlled at 60. Such a high hardness ensures that the overall deformation of the polishing pad is extremely small when it is subjected to a polishing pressure of up to 3.0 to 4.0 PSI, so that the pressure can be effectively transferred to the wafer surface, maintaining the macroscopic flatness of the wafer, which is crucial to the control of total thickness variation (TTV).

[0033] Furthermore, the surface of the first fine polishing pad 11 is machined with a specific macro-groove structure. This structure is designed as a concentric circle array, and the cross-section of the groove is rectangular. In a specific embodiment, the groove depth is set to 0.50 mm, the width is set to 0.5 mm, and the center-to-center distance between adjacent grooves, i.e., the pitch, is set to 2.5 mm. This deep and wide groove design greatly improves the fluid dynamics of the polishing liquid at the pad-wafer interface. It constructs an efficient channel network, allowing fresh polishing liquid to be quickly replenished to the contact area, while quickly discharging the silicate products, detached silicon particles, and abrasive particles generated after the polishing reaction, avoiding a decrease in the polishing rate due to product accumulation and effectively reducing the risk of scratches on the wafer surface. In addition, the presence of the grooves forms an intermittent contact area, causing the boss portion of the polishing pad to generate a higher local pressure on the wafer surface. This enhanced mechanical force plays an irreplaceable role in penetrating the surface layer and effectively removing lattice damage areas hidden tens to hundreds of nanometers deep below the subsurface.

[0034] In conjunction with the first fine polishing pad 11, the first fine polishing step uses a specially formulated first fine polishing liquid. The polishing liquid is a highly alkaline water-based dispersion with colloidal silica as an abrasive. The average primary particle size of the colloidal silica abrasive is controlled at 60 nanometers. Through transmission electron microscopy (TEM) analysis, the abrasive particle size distribution presents a highly concentrated single-peak normal distribution, and the standard deviation of its distribution is less than 4 nanometers, ensuring the uniformity of the grinding action. Such relatively large abrasive particles provide sufficient mechanical cutting force. In the formula, the mass concentration of the abrasive is set to 25%. In order to drive efficient chemical reactions, the chemical environment of the first fine polishing liquid is set to be strongly alkaline. By precisely adding potassium hydroxide (KOH) as a pH regulator, the pH value of the polishing liquid is stably maintained at 11.0. In such a high pH environment, silicon surface atoms react violently with hydroxide ions to form soluble silicates. This greatly accelerates the chemical removal rate of the material. This high degree of synergy between strong mechanical action and strong chemical action is the basis for achieving the goal of efficiently removing deep damage in this step.

[0035] In terms of process parameter setting, the first fine polishing step adopts a set of high-energy input parameter combinations. Specifically, the downforce applied to the back of the wafer by the polishing head is set to 3.5 pounds per square inch (PSI). The polishing disk speed of the first fine polishing unit 10 is set to 90 revolutions per minute (RPM), while the polishing head speed carrying the wafer is set to 85 revolutions per minute (RPM). The polishing head and the polishing disk rotate in the same direction but with a speed difference, which helps to form a uniform polishing track on the wafer surface. The supply rate of the first fine polishing liquid is set to 200 milliliters per minute to ensure sufficient lubrication of the polishing interface and the supply of chemical reactants. Through the precise combination of the above-mentioned hardware, consumables and process parameters, the first fine polishing step can stably remove silicon material of about 1.5 microns thick within a process time of about 90 seconds, thereby ensuring that the subsurface damage layer accumulated during the multiple cycles of use of the regenerated wafer is completely removed, providing a clean and uniform substrate for the subsequent fine repair step.

[0036] After the first fine polishing step, the wafer is transferred to the second fine polishing unit 20 for the second fine polishing step. The core task of this step is to systematically repair the new, relatively minor surface damage that is inevitably introduced by the high energy input in the first step, mainly including the micro-texture copied by the hard groove pad on the wafer surface and the fine scratches caused by the 60 nm abrasive. Figure 4 The second fine polishing unit 20 is different from the first unit in hardware configuration, and the key lies in the configuration of the second fine polishing pad 21.

[0037] The second fine polishing pad 21 is a microporous polyurethane polishing pad with a medium hardness and no macro-grooves. Its surface is flat and free of any macro-geometric patterns, fundamentally avoiding any periodic pressure distribution on the wafer surface, thereby helping to eliminate the microtexture left by the first step. Its base material is also polyurethane, but its formulation has been adjusted to reduce its Shore D hardness to 50. This medium hardness imparts enhanced surface compliance, allowing it to better conform to the micro-contours of the wafer surface and effectively smooth out any remaining undulations. Furthermore, the polishing pad is uniformly distributed with a large number of open micropores, with an average pore size of approximately 30 microns as observed by scanning electron microscopy (SEM). These interconnected micropores function as a capillary network, effectively absorbing and storing the polishing liquid and evenly releasing it to the polishing interface during the polishing process. This ensures that a stable, uniform liquid film forms between the wafer and the polishing pad even at low polishing pressures.

[0038] In conjunction with the second fine polishing pad 21, the second fine polishing step utilizes a specially designed second fine polishing slurry. This polishing slurry features an innovative abrasive system, employing a colloidal silica abrasive with a bimodal particle size distribution. This abrasive system consists of two colloidal silica particles of different particle sizes mixed in a specific ratio. The primary component, comprising 85% of the total abrasive mass, is ultrafine silica particles with an average primary particle size of 25 nanometers. This component primarily smoothes and thins the wafer surface, effectively reducing overall surface roughness. The secondary component, comprising 15% of the total abrasive mass, is silica particles with an average primary particle size of 45 nanometers. This component is intended to more efficiently and targetedly remove microscratches of a certain depth (typically ranging from a few to tens of nanometers) left over from the first step. Larger particles are more effective in "plowing" and removing these scratches. Through the synergistic effect of large and small particles, this step achieves rapid surface roughness reduction while also effectively removing discrete microdefects. In the formulation, the total mass concentration of the abrasive was reduced to 15%. Regarding the chemical environment, the pH of the polishing solution was controlled at 10.0 by using highly volatile ammonia (NH4OH) as a pH adjuster. Compared to the first step, the pH reduction is intended to moderately slow the chemical corrosion rate, allowing mechanical smoothing to play a more important role in this step, thereby achieving a more optimal surface morphology.

[0039] In terms of process parameter setting, the overall mechanical energy input of the second fine polishing step is significantly lowered. Specifically, the polishing pressure applied to the back of the wafer is reduced to 2.5PSI. The polishing disc speed of the second fine polishing unit 20 is reduced to 70RPM per minute, and the polishing head speed is reduced to 60RPM per minute. In order to compensate for the lack of lubrication that may result from the pressure reduction, the supply rate of the second fine polishing liquid is correspondingly increased to 250 ml per minute. Within a process time of about 120 seconds, this step can remove about 0.5 microns of silicon material. After the systematic repair of the second fine polishing step, the root mean square (Rq) roughness of the wafer surface can be reduced from about 0.3 nanometers after the first step to below 0.1 nanometers, the microscopic damage introduced in the first step is effectively eliminated, and the surface presents a highly smooth and uniform state.

[0040] Furthermore, after completing the second fine polishing step, the wafer is sent to the third fine polishing unit 30 to perform the last step of the method of the present invention, namely the third fine polishing step. The function of this step is to final surface cleaning, modification and chemical passivation. Its goal is no longer to significantly remove material or reduce roughness, but to achieve atomic-level surface flatness, control the number of surface particle residues (Light Point Defects, LPDs) larger than 19 nanometers to less than 100 per wafer, and construct a hydrophilic surface passivation layer that can inhibit the re-adsorption of particles. Figure 5 The third fine polishing unit 30 is equipped with a third fine polishing pad 31 .

[0041] The third fine polishing pad 31 is a soft, dense, microporous polyurethane polishing pad without a macro-groove structure. Its Shore D hardness has been further reduced to 40, making it the softest of the polishing pads used in the three steps. This extremely high softness enables the polishing pad to achieve maximum conformal contact with the wafer surface, achieving a near-perfect fit at the microscopic level. Under these conditions, the polishing pressure is mainly transmitted through the fluid dynamic pressure film formed by the polishing liquid, and mechanical contact is reduced to a minimum, achieving the so-called "glow polishing" or "chemical-based" polishing mode, thereby avoiding any new mechanical scratches. The microporous structure inside the polishing pad is also denser, with an average pore size of less than 15 microns. This structure is conducive to forming and maintaining an ultra-thin, stable, and uniform liquid film even under extremely low polishing pressure.

[0042] In conjunction with the third fine polishing pad 31, the third fine polishing step adopts a third fine polishing liquid with an innovative formula. This is a chemically modified liquid with ultra-low abrasive concentration or even no abrasive. In a preferred embodiment, the polishing liquid contains ultrafine colloidal silica abrasive with an average primary particle size of 8 nanometers, and its mass concentration in the polishing liquid is strictly controlled at an extremely low level of 1.0%. The role of these trace abrasives is only to gently "wipe" the surface to remove any possible organic residues or surface electrostatic adsorbents, rather than mechanical grinding. The key to the polishing liquid formula is the addition of a specific surface passivator. In a specific embodiment, the surface passivator is dodecyltrimethylammonium bromide (DTAB), a cationic surfactant, and its added concentration in the polishing liquid is 0.03% (mass percentage). Its mechanism of action is as follows: During the polishing process, the hydrophobic dodecyl long chain ends of the DTAB molecules generate electrostatic repulsion with the negatively charged silica abrasive (if present) and the similarly negatively charged silicon surface (at high pH), thereby helping to prevent abrasive particles from depositing on the wafer surface at the polishing endpoint. More importantly, after polishing is completed, the positively charged hydrophilic quaternary ammonium salt head preferentially adsorbs to the newly formed, negatively charged exposed silanol (Si-OH) surface through electrostatic attraction, forming a temporary monomolecular or sub-monomolecular hydrophilic passivation layer. This passivation layer significantly enhances the hydrophilicity of the wafer surface, resulting in a stable water contact angle below 5 degrees. Simultaneously, it shifts the zeta potential of the wafer surface from negative to positive or near zero, generating strong electrostatic repulsion with the negatively charged particulate contaminants commonly found in the cleaning solution, significantly suppressing redeposition of particles during post-polishing cleaning, drying, and transport. In addition, in order to maximize the protection of the final surface integrity, the pH value of the third polishing solution is precisely controlled by using non-ionic Tris buffer and stabilized in the near-neutral range of 8.0.

[0043] In terms of process parameter settings, the mechanical energy input in the third fine polishing step is reduced to a minimum. Specifically, the polishing pressure applied to the back of the wafer is set to 1.5 PSI. The polishing disk speed is set to 35 RPM, and the polishing head speed is set to 30 RPM. To ensure stable hydrodynamic support, the supply rate of the third fine polishing solution is further increased to 300 ml / min. Within a process time of approximately 60 seconds, this step only removes approximately 0.05 microns of material, and its primary function is surface modification rather than material removal.

[0044] As the core difference between the present invention and traditional fixed recipe process, the present method integrates an online process monitoring and adaptive feedback control subsystem 40. Figure 6This subsystem 40 is deployed on the wafer transport path between adjacent fine polishing units (e.g., 10 and 20, 20 and 30). This subsystem includes an optical measurement module 41 and a central process control module 42. The optical measurement module 41 contains a laser light source 411, for example, a helium-neon laser with a central wavelength of 633 nanometers and a power of 5 milliwatts. This laser emits a collimated laser beam at a fixed angle of incidence (e.g., 45 degrees) onto the surface of a wafer that has just completed the previous polishing step. An array detector 412 is configured to receive scattered light from the wafer surface. Specifically, the detector consists of a central photodiode located in the direction of the specular reflection angle and a plurality of ring-shaped photodiodes surrounding it at different scattering angles. The central detector measures the intensity of specularly reflected light, a value that is highly correlated with the surface microroughness (Rq). The surrounding ring-shaped detectors measure the scattered light intensity at different angles. Using a specific algorithm, integration of this intensity can be used to calculate the surface haze value, which is particularly sensitive to subsurface defects.

[0045] The central process control module 42, typically an industrial computer, runs a preset control algorithm. It receives and processes detection data from the optical measurement module 41 in real time. In a specific process flow, after a wafer completes the first fine polishing step and before being sent to the second fine polishing unit 20, the optical measurement module 41 scans and measures it, obtaining a first set of intermediate surface parameters, primarily including the haze value H1. The central process control module 42 compares the measured H1 value with a preset first target range, such as [0.15ppm, 0.25ppm]. If the H1 value is within this range, indicating that the first step has achieved the desired subsurface damage removal effect, the second fine polishing step will be performed according to the aforementioned standard process parameters. However, if the measured H1 value, such as 0.30ppm, is above the upper limit of the target range, this indicates that the initial subsurface damage on this particular wafer is relatively severe and was not completely removed in the first step. In this case, the central process control module 42 activates feedback control logic and, based on a proportional-integral (PI) control algorithm, instructs the polishing pressure to increase during the second fine polishing step. For example, the pressure adjustment amount ΔP2 = K p ·(H1-H 1,target,upper ), where K pis a preset proportional gain coefficient. This tiny pressure increment will enhance the material removal capability of the second step to compensate for the shortcomings of the first step. Similarly, after the second fine polishing step is completed and before entering the third fine polishing step, the system performs a second optical measurement to obtain a second intermediate surface parameter group, mainly including the surface roughness Rq2. If the measured Rq2 value is higher than the target value (for example, 0.09nm), the control module will fine-tune the process time of the third fine polishing step to ensure that the final surface reaches atomic level flatness. Through this closed-loop feedback control, the method of the present invention can actively adapt to the individual differences of the incoming wafers and the slight fluctuations in the process, ensuring that the final quality of each wafer is highly consistent, and significantly improving the stability and yield of the process.

[0046] The adaptive feedback control logic executed by the central process control module 42 is centered around establishing a quantitative mapping relationship between measurement parameter deviations and process parameter adjustments. In a preferred embodiment of the present invention, an incremental proportional-integral (PI) control algorithm is employed.

[0047] Adjust the parameters of the second fine polishing step based on the first haze value H1:

[0048] 1. Deviation calculation: Calculate the difference between the measured first haze value H1 and the preset first target haze range upper limit value H1 target,upper Deviation from the lower limit (e.g. 0.25ppm) or the lower limit (depending on the control target, here we use the upper limit as an example to adjust the pressure to remove more damage):

[0049]

[0050] If H 1,measured >H 1,target,upper , then e(k)>0, indicating that the damage removal is insufficient and the material removal capacity of the second step needs to be increased (e.g., increasing the pressure);

[0051] If H 1,measured <H 1,target,upper , then e(k)<=0, indicating that the damage removal meets the standard and the standard parameters are maintained

[0052] 2. Calculation of pressure adjustment (ΔP2): The pressure adjustment is determined by the proportional term and the integral term: ΔP2(k) = k p *e(k)+K i *∑[e(j)]; (from j=0 to j=k),

[0053] ΔP2(k): The calculated adjustment amount for the polishing pressure in the second finishing polishing step (unit: PSI);

[0054] K p: Proportional gain coefficient (unit: PSI / ppm).

[0055] K i: Integral gain coefficient (unit: PSI / (ppm*cycle)).

[0056] e(k): Deviation of this measurement (unit: ppm).

[0057] ∑[e(j)]: Cumulative sum of historical deviations (unit: ppm).

[0058] 3. Apply the adjustment: Apply the calculated adjustment ΔP2(k) to the standard polishing pressure setting value P2_standard in the second fine polishing step: P 2applied =P 2standard +ΔP2(k),

[0059] At the same time, it is necessary to ensure that P 2applied It falls within the safe process window allowed by the second fine polishing step (2.0 to 2.8 PSI as defined in claim 5). If it exceeds, the boundary value is taken.

[0060] 4. Parameter determination: the proportional gain coefficient K p and integral gain coefficient K i The specific value of needs to be determined through experimental calibration, the process includes:

[0061] a) Prepare a set of wafer samples with known subsurface damage of varying degrees.

[0062] b) Fix the first fine polishing step as standard process.

[0063] c) For each sample, measure its haze value H1 after completing the first step.

[0064] d) In the second fine polishing step, a series of controllable pressure deviations (ΔP2_test) are intentionally introduced and the second step is performed.

[0065] e) Measure the haze value (H_final) and / or other key quality indicators (such as LPD, roughness) of the final polished wafer.

[0066] f) Analyze the data and establish a response model between the H1 deviation (e) and the optimal ΔP2 required to achieve the target H_final.

[0067] g) Based on the response model, combined with the stability and response speed requirements of the PI controller (such as avoiding overshoot and fast convergence), determine a set of applicable K p and Ki values, typical K p The value range is between 0.1PSI / ppm and 0.5PSI / ppm, K iThe value range is 0.01PSI / (ppm*cycle) to 0.05

[0068] The integral term is accumulated and usually sets a limit or reset logic to prevent integral saturation;

[0069] For the control logic of adjusting the parameters of the third step (such as time Δt3) based on the roughness Rq2 after the second step, a similar PI algorithm structure can be used:

[0070]

[0071] e Rq (k) = R q 2 measured -R q 2 target

[0072] K p_time : proportional gain (unit: s / nm);

[0073] K i_time : integral gain (unit: s / (nm*cycle));

[0074] Its gain coefficient K p_time ,K i_time Also obtained through experimental calibration for the third step target (surface roughness). Typical Kp_time ranges may be between 5s / nm and 20s / nm.

[0075] Example

[0076] In order to verify the technical effect of the present invention, a batch of 300 mm P-type <100> Test-grade silicon wafers with a 28nm orientation were reprocessed. These wafers had previously been used as monitor wafers in the 28nm logic device manufacturing process and had a complex surface texture of residual films, embedded particles, and accumulated subsurface damage from multiple thermal annealing and ion implantation cycles. After standard film stripping, grinding, and rough polishing, the wafers' initial state showed a total thickness variation (TTV) of approximately 2.0 microns, a surface haze greater than 1.0 ppm, and numerous scratches and defects.

[0077] The pre-treated wafer is subjected to the three-step fine polishing method of the present invention for processing.

[0078] First fine polishing step: Using the first fine polishing unit 10 described in detail above, the first fine polishing pad 11 (60D hardness, concentric grooves), and the first fine polishing slurry (60nm silica abrasive, 25% concentration, pH 11.0), the process parameters were set as follows: polishing pressure 3.5 PSI, polishing disk speed 90 RPM, polishing head speed 85 RPM, polishing slurry flow rate 200 ml / min, and process time 90 seconds.

[0079] After the first step, the average haze H1 on the wafer surface, as measured by the online process monitoring and adaptive feedback control subsystem 40, was 0.22 ppm, falling within the target range of [0.15 ppm, 0.25 ppm]. Therefore, the central process control module 42 decided to use standard process parameters for the second fine polishing step.

[0080] Second fine polishing step: Using the second fine polishing unit 20, the second fine polishing pad 21 (50D hardness, non-grooved microporous pad), and the second fine polishing slurry (25nm / 45nm bimodal particle size abrasive, 15% total concentration, pH 10.0), the process parameters were set as follows: polishing pressure 2.5 PSI, polishing disk speed 70 RPM, polishing head speed 60 RPM, polishing slurry flow rate 250 ml / min, and process time 120 seconds.

[0081] After the second step, the root mean square roughness Rq2 of the wafer surface was measured again by the online process monitoring system 40 to be 0.085 nm, which is lower than the target upper limit of 0.09 nm. Therefore, the third fine polishing step also uses standard process parameters.

[0082] The third fine polishing step uses the third fine polishing unit 30, the third fine polishing pad 31 (40D hardness, soft, dense microporous pad), and the third fine polishing solution (8 nm silica abrasive, 1.0% concentration, containing 0.03% DTAB passivating agent, pH 8.0). The process parameters are set as follows: polishing pressure 1.5 PSI, polishing disk speed 35 RPM, polishing head speed 30 RPM, polishing solution flow rate 300 ml / min, and process time 60 seconds.

[0083] Comparative Example

[0084] For comparison, a conventional two-step method was used to regenerate and fine-polish the pre-processed wafers from the same batch.

[0085] The first step (rough polishing) is the same as the first fine polishing step of the embodiment, and is intended to remove deep damage.

[0086] Step 2 (Final Polishing): This step attempts to simultaneously smooth the surface and achieve final cleaning. A non-grooved, microporous polishing pad with a hardness of 50D was used, along with a fine polishing slurry with a unimodal particle size distribution (30nm silica abrasive, 10% concentration, pH 10.5, and no surfactant). Process parameters were set as follows: polishing pressure 2.0 PSI, polishing pad speed 50 RPM, polishing head speed 45 RPM, polishing slurry flow rate 280 ml / min, and process time 150 seconds. This process was conducted without online monitoring or feedback control.

[0087] Results comparison

[0088] Final quality inspection was performed on the wafers after processing in the examples and comparative examples. The inspection instruments included an atomic force microscope (AFM) for measuring surface roughness, a laser surface scanner (such as the KLA-Tencor SP5) for measuring haze and LPD, and a contact angle meter for evaluating surface hydrophilicity. The results are summarized in the following table:

[0089]

[0090] From the data comparison in the above table, it can be clearly seen that the regenerated wafers finally obtained by adopting the three-step fine polishing method described in the present invention show significant superiority in various key quality indicators. Especially in terms of surface particle defects (LPD) and subsurface damage residues (Haze), which are of greatest concern at the 19nm process node, the results of the method of the present invention are orders of magnitude higher than those of the traditional two-step method. The extremely low number of LPDs (<80) and excellent surface hydrophilicity (water contact angle <5 degrees) in the embodiment directly prove the effectiveness of the surface passivator in the third step. The extremely low Haze value (<0.08ppm) shows that the subsurface damage is thoroughly and stably removed through the synergistic effect of the first and second steps and the guarantee of online feedback control. In summary, the method provided by the present invention can stably and reliably produce ultra-high-quality regenerated wafers that meet the requirements of 19nm and more advanced process nodes, and has extremely high industrial application value.

Claims

1. A three-step fine polishing method for 19nm wafer regeneration, characterized in that: The following steps are involved: A first fine polishing step is performed on a wafer equipped with a first fine polishing unit (10), wherein the first fine polishing step is performed on the wafer by applying a first set of process parameters and using a combination of a first fine polishing pad (11) and a first fine polishing liquid to perform a chemical mechanical polishing process with a high material removal rate to remove subsurface damage of the wafer; After the first fine polishing step is completed, the wafer is transferred to a second fine polishing unit (20) and a second fine polishing step is performed, wherein the second fine polishing step is performed by applying a second set of process parameters and using a combination of a second fine polishing pad (21) and a second fine polishing liquid to perform a chemical mechanical polishing process on the wafer to repair the microscopic morphological defects introduced by the first fine polishing step; After the second fine polishing step is completed, the wafer is transferred to a third fine polishing unit (30) and a third fine polishing step is performed, wherein the third fine polishing step is performed on the surface of the wafer by applying a third set of process parameters and using a combination of a third fine polishing pad (31) and a third fine polishing liquid; And, it includes an online process monitoring and adaptive feedback control step, which is executed by an online process monitoring and adaptive feedback control subsystem (40) deployed between the first fine polishing unit (10), the second fine polishing unit (20) and the third fine polishing unit (30), and the subsystem (40) is configured to measure the intermediate surface parameters of the wafer after at least one of the fine polishing steps is completed, and dynamically adjust the process parameters of at least one subsequent fine polishing step based on the comparison of the measurement results of the intermediate surface parameters with the preset target range.

2. The method according to claim 1, characterized in that In the first fine polishing step: The first fine polishing pad (11) is a hard polyurethane polishing pad with a Shore D hardness of 55 to 65, and a macro-groove structure is provided on its working surface; the macro-groove structure is a concentric circle array, the cross-sectional shape of the grooves constituting the macro-groove structure is rectangular, the depth is 0.45 mm to 0.55 mm, the width is 0.4 mm to 0.6 mm, and the center distance between adjacent grooves is 2.0 mm to 3.0 mm; In addition, the first fine polishing liquid is a highly alkaline chemical mechanical polishing liquid, which uses colloidal silica as an abrasive, the average primary particle size of the colloidal silica abrasive is 50 nanometers to 70 nanometers, and its mass concentration in the first fine polishing liquid is 20% to 30%; the pH value of the first fine polishing liquid is regulated in the range of 10.5 to 11.

5.

3. The method according to claim 2, characterized in that The first set of process parameters in the first fine polishing step includes: The polishing pressure applied to the back side of the wafer is set to 3.0 psi to 4.0 psi; The polishing disc rotation speed of the first fine polishing unit (10) is set to 80 to 100 revolutions per minute; The polishing head carrying the wafer is set to rotate at a speed of 80 to 95 revolutions per minute; The supply rate of the first fine polishing liquid is set to 180 ml to 220 ml per minute; And, the process time of the first finish polishing step is set to 60 seconds to 120 seconds.

4. The method according to claim 1, wherein In the second fine polishing step: The second fine polishing pad (21) is a medium-hardness microporous polyurethane polishing pad without a macro-groove structure, and has a Shore D hardness of 45 to 55, which is lower than the Shore D hardness of the first fine polishing pad (11); the interior of the body of the second fine polishing pad (21) has an open microporous structure with an average pore size of 25 microns to 35 microns; In addition, the second fine polishing liquid is a chemical mechanical polishing liquid, which uses colloidal silica with a bimodal particle size distribution as an abrasive; the colloidal silica abrasive with a bimodal particle size distribution is a mixture of a main abrasive and an auxiliary abrasive, wherein the average primary particle size of the main abrasive is 20 nanometers to 30 nanometers, accounting for 80% to 90% of the total abrasive mass, and the average primary particle size of the auxiliary abrasive is 40 nanometers to 50 nanometers, accounting for 10% to 20% of the total abrasive mass; the total mass concentration of the abrasive in the second fine polishing liquid is 12% to 18%; the pH value of the second fine polishing liquid is regulated within the range of 9.5 to 10.

5.

5. The method according to claim 4, characterized in that The second set of process parameters in the second fine polishing step includes: The polishing pressure applied to the back side of the wafer is set to 2.0 psi to 2.8 psi; The polishing disc speed of the second fine polishing unit (20) is set to 60 to 75 revolutions per minute; The polishing head carrying the wafer is set to rotate at a speed of 55 to 70 revolutions per minute; The supply rate of the second fine polishing liquid is set to 230 ml to 270 ml per minute; And, the process time of the second finish polishing step is set to 90 seconds to 150 seconds.

6. The method according to claim 1, characterized in that In the third fine polishing step: The third fine polishing pad (31) is a soft, dense, microporous polyurethane polishing pad without a macro-groove structure, and has a Shore D hardness of 35 to 45, which is lower than the Shore D hardness of the second fine polishing pad (21); the interior of the body of the third fine polishing pad (31) has a microporous structure with an average pore size of less than 20 microns; In addition, the third fine polishing liquid is a nearly neutral chemical modification liquid, and its pH value is controlled within the range of 7.5 to 8.5; a cationic surfactant with a concentration of 0.01% to 0.05% by mass is added to the third fine polishing liquid as a surface passivator; and the mass concentration of the colloidal silica abrasive contained in the third fine polishing liquid is controlled within the range of 0% to 1.5%, and if abrasive is contained, the average primary particle size of the abrasive is less than 10 nanometers.

7. The method according to claim 6, characterized in that The surface passivating agent is selected from at least one of dodecyltrimethylammonium bromide and hexadecyltrimethylammonium chloride; Furthermore, the third set of process parameters in the third fine polishing step includes: The polishing pressure applied to the back side of the wafer is set to 1.0 psi to 1.8 psi; The polishing disc speed of the third fine polishing unit (30) is set to 30 to 45 revolutions per minute; The polishing head carrying the wafer is set to rotate at a speed of 25 to 40 revolutions per minute; The supply rate of the third fine polishing liquid is set to 280 ml to 320 ml per minute; And, the process time of the third finish polishing step is set to 45 seconds to 75 seconds.

8. The method according to claim 1, characterized in that The online process monitoring and adaptive feedback control subsystem (40) includes an optical measurement module (41) and a central process control module (42); The optical measurement module (41) comprises a laser light source (411) and an array detector (412) configured to receive scattered light from the wafer surface. The optical measurement module (411) is configured to emit a beam of collimated laser light of a preset wavelength to illuminate the wafer surface, and use the array detector (412) to measure the scattered light to obtain the intermediate state surface parameters, wherein the intermediate state surface parameters include a surface roughness value and / or a surface haze value. The central process control module (42) is electrically connected to the optical measurement module (41), and is configured to receive the intermediate surface parameters and perform the comparison and dynamic adjustment.

9. The method according to claim 8, characterized in that The online process monitoring and adaptive feedback control steps specifically include: After the first fine polishing step is completed and before the second fine polishing step begins, using the optical measurement module (41) to measure a first intermediate surface parameter group of the wafer, the parameter group including at least a first haze value (H1); The central process control module (42) compares the first haze value (H1) with a preset first target haze range; If the first haze value (H1) exceeds the first target haze range, the central process control module (42) dynamically adjusts the polishing pressure and / or execution time in the second set of process parameters according to a preset control algorithm.

10. The method according to claim 1, characterized in that The first, second and third fine polishing steps present a gradient decreasing relationship in process intensity, which is specifically reflected as follows: The Shore D hardness of the first fine polishing pad (11) is higher than the Shore D hardness of the second fine polishing pad (21), and the Shore D hardness of the second fine polishing pad (21) is higher than the Shore D hardness of the third fine polishing pad (31); The polishing pressure applied in the first fine polishing step is higher than the polishing pressure applied in the second fine polishing step, and the polishing pressure applied in the second fine polishing step is higher than the polishing pressure applied in the third fine polishing step; Furthermore, the pH value of the first fine polishing liquid is higher than the pH value of the second fine polishing liquid, and the pH value of the second fine polishing liquid is higher than the pH value of the third fine polishing liquid.

Citation Information

Patent Citations

  • Methods for polishing semiconductor wafers

    CN101670546B

  • Semiconductor wafer polishing equipment and polishing method

    CN111823120B

Cited By

  • Wafer process parameter optimization method and system based on laser scanning

    CN120914128A

  • Wafer thinning device and thinning method

    CN121018328A

  • Wafer heavy scratch three-dimensional shape repairing process based on DFG 8560 single-axis accurate grinding mode

    CN121374375A