A method for selectively removing a film from a 12-inch wafer based on plasma activation modification

Abstract

The application relates to the field of semiconductor manufacturing, and specifically discloses a 12-inch wafer selective film removal regeneration method based on plasma activation modification, aiming to solve the problems of poor selectivity, high damage risk and environmental protection in the existing wafer film removal process. The method comprises the following steps: performing plasma activation treatment on the wafer surface film layer, and precisely controlling the surface characteristics of the film layer by controlling the plasma process parameters, so that high-selectivity film removal can be realized in subsequent processing. By adopting the technical scheme, the 12-inch wafer surface film layer can be selectively removed with high precision, the damage risk to the substrate material is effectively reduced, the use of chemical reagents is reduced, and the environmental protection and economic efficiency of the process are improved.

Description

Technical Field

[0001] This application belongs to the field of semiconductor manufacturing, specifically relating to a selective film removal and regeneration method for 12-inch wafers based on plasma activation modification. Background Technology

[0002] As the fundamental carrier of integrated circuit manufacturing, the precision and complexity of semiconductor wafer manufacturing processes directly determine chip performance and yield. With the continuous evolution of semiconductor manufacturing process nodes towards advanced processes, 12-inch wafers have become the mainstream substrate material for high-end semiconductor products such as logic chips, memory devices, and power devices. In the wafer manufacturing and process testing stages, the number of process test wafers, process development wafers, and some process-decommissioned wafers continues to increase. Direct disposal would not only result in a huge waste of silicon substrate materials but also significantly increase the raw material costs of semiconductor manufacturing. Wafer regeneration technology removes functional films and process residues from the wafer surface using physical or chemical methods, restoring the wafer to specifications suitable for repeated process reprocessing. This is a crucial link in the green semiconductor manufacturing system.

[0003] The wafer regeneration film removal process mainly falls into three major technical routes: wet etching, plasma dry etching, and mechanical polishing. Wet etching removes surface materials by reacting chemical solutions with the film layer, offering advantages such as low equipment cost and simple operation. Plasma dry etching removes the film layer through the dual effects of high-energy particle bombardment and chemical reaction in plasma, exhibiting good anisotropy and minimal damage to the substrate. Mechanical polishing removes the surface film layer through the mechanical grinding action of abrasives. However, none of these individual technical routes can simultaneously meet the comprehensive requirements of high efficiency, high selectivity, low damage, and low cost.

[0004] Existing technologies attempt to overcome the limitations of single technologies by combining plasma treatment with wet etching. The technology disclosed in CN121865905A employs plasma bombardment of the outer edge region of the wafer to create microporous channels in the dielectric layer, followed by etching in an etching solution containing hydrofluoric acid and an oxidant. However, this technology has the following shortcomings: plasma bombardment is only performed on the outer edge region of the wafer, failing to achieve overall plasma activation and modification of the wafer surface, thus failing to improve the uniformity of subsequent wet etching solution wetting on the wafer surface; the plasma treatment area is limited to the edge, while the central region still relies on pure wet etching, failing to solve the lateral drilling problem caused by the isotropic nature of wet etching; and there is a lack of effective control between the dry and wet processes, making the surface prone to secondary contamination and oxide layer thickening during the transfer of the wafer from the plasma treatment chamber to the wet etching tank, negating the pretreatment effect. The technical solution disclosed in CN122602798A employs a three-step sequential wet process: Step 1 uses a mixture of hydrofluoric acid and hydrogen peroxide to remove the dense HK film; Step 2 uses aqua regia to remove other metal films; and Step 3 uses SC2 cleaning solution for controlled metal cleaning. However, this technology also has the following shortcomings: the pure wet process cannot avoid the lateral drilling effect caused by isotropic etching, which can easily damage the silicon substrate of patterned wafers and cannot meet the requirements for high-precision wafer regeneration; multiple films are etched simultaneously in a single tank or continuous wet process, which can easily lead to selective interference and make it difficult to achieve high-precision selective removal of multi-layer composite films; residual organic matter in the wet process will continuously contaminate the etching solution, significantly shortening the service life of the solution and increasing the cost of waste liquid treatment. The above problems indicate that the existing technical solution still has significant shortcomings in terms of effective utilization of plasma activation modification, control of dry and wet process transitions, and selective removal of multi-layer composite films. There is an urgent need for a wafer regeneration film removal method that can simultaneously meet the requirements of high efficiency, high selectivity, low damage, and low cost mass production. Summary of the Invention

[0005] The purpose of this invention is to provide a selective film removal and regeneration method for 12-inch wafers based on plasma activation modification, which can effectively solve the fundamental problem that the existing wafer regeneration and film removal processes cannot simultaneously meet the requirements of high efficiency, high selectivity, low damage, and low cost mass production.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A selective decoupling and regeneration method for 12-inch wafers based on plasma activation modification includes the following specific steps: Step 1: Material preparation: Provide a 12-inch wafer to be regenerated, the wafer surface of which is coated with an organic film layer and / or an inorganic film layer. The organic film layer includes photoresist and organic polymer residues, and the inorganic film layer includes at least one of silicon dioxide, silicon nitride, and a metal film layer; Step 2: Dry plasma pretreatment: Place the wafer to be regenerated in a plasma processing chamber, introduce process gas and excite plasma, simultaneously achieving a dual effect: firstly, removing the surface organic film layer through oxidation and ashing, and secondly, removing the inorganic film layer from the surface. The process involves several steps: Step 3: Segmented controlled transfer. The pre-treated wafer is transferred to the wet stripping station under controlled conditions to maintain surface activity and inhibit secondary contamination and oxidation thickening. Step 4: Selective wet stripping. A multi-tank etching system is used, sequentially employing corresponding selective etching solutions to remove the inorganic film layer on the wafer surface in stages, maintaining a high selectivity for the silicon substrate during etching. Step 5: Post-processing and inspection. The stripped wafer is rinsed with deionized water and dried. After passing film thickness and defect inspection, a regenerated wafer is obtained.

[0008] Preferably, in step 2, the plasma pretreatment adopts an oxygen-based plasma system, the process gas is selected from at least one of oxygen, nitrous oxide, a mixture of carbon tetrafluoride and oxygen, and a mixture of sulfur hexafluoride and oxygen, the radio frequency power is 300 to 1500W, the cavity pressure is 0.1 to 1 Torr, the substrate temperature is 150 to 300°C, the processing time is 30 to 300s, and the contact angle of the wafer surface with water after pretreatment is not greater than 10°.

[0009] Preferably, in step 3, the segmented controlled transfer includes three optional implementation methods: vacuum transfer method, which realizes the direct transfer of wafers in a vacuum environment through Load Lock vacuum cavity connection; inert gas protection method, which forms a positive pressure channel by continuously introducing nitrogen or argon gas for transfer; and timing control method, which is to perform real-time scheduling through the manufacturing execution system to ensure that the transfer interval is no more than 5 minutes.

[0010] Preferably, in step 4, the multi-tank series selective film removal system includes at least a dilute hydrofluoric acid etching tank, a hot phosphoric acid etching tank, and an SC1 / SC2 cleaning tank arranged in sequence, with a deionized water rinsing section between adjacent tanks to avoid cross-contamination of the chemical solutions. The dilute hydrofluoric acid etching tank is used to remove the silicon dioxide film, the hot phosphoric acid etching tank is used to remove the silicon nitride film, and the SC1 / SC2 cleaning tank is used to remove metal ion contamination.

[0011] Preferably, in step 4, the determination of the corrosion endpoint is achieved through online monitoring of the chemical solution conductivity, redox potential, or infrared reflectance spectrum to achieve precise endpoint control. When the monitoring signal reaches a preset threshold, the corresponding film layer is determined to be removed.

[0012] Preferably, in step 4, the dilute hydrofluoric acid corrosion tank uses a dilute hydrofluoric acid solution with a mass fraction of 1% to 5%, and corrosion is carried out under constant temperature conditions of 20 to 30°C. During the corrosion process, the corrosion endpoint is determined when the silicon dioxide film thickness drops to 0 by real-time monitoring of the infrared reflection spectrum.

[0013] Preferably, in step 4, the hot phosphoric acid etching tank uses a hot phosphoric acid solution at 165 to 175°C, and the etching endpoint of the silicon nitride film is determined by online monitoring of the oxidation-reduction potential. During the etching process, the selectivity of the silicon substrate is not less than 50 to 1.

[0014] Preferably, in step 5, the post-processing further includes sequential isopropanol dehydration and nitrogen drying processes. After drying, film thickness measurement and surface defect detection are performed, with the number of surface particles not exceeding 10 and the amount of silicon substrate removed not exceeding 5 nm.

[0015] Preferably, in step 2, when the surface of the wafer to be regenerated contains fluoropolymer residue, the process gas is a mixture of carbon tetrafluoride and oxygen in a volume ratio of 1:3, the radio frequency power is 800 to 1200W, the substrate temperature is 200 to 280°C, and the processing time is 150 to 250s.

[0016] Preferably, the present invention is also applicable to mass production line transformation scenarios. The wafer process is scheduled in real time through the manufacturing execution system. Wafers that have completed dry processing are set to enter the wet process first, ensuring that the interval between the end of dry processing and the start of wet processing is no more than 4 minutes, the activation effect decay is no more than 10%, and the film removal uniformity deviation is no more than 2%.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. The dry-wet segmented relay process design avoids the inherent defects of a single process. By dividing the labor into dry plasma for organic matter removal and surface activation, and wet plasma for selective removal of inorganic membranes, it solves the problems of incomplete organic matter removal and easy contamination of the solution in pure wet processes, while avoiding the defects of high equipment cost and low inorganic membrane removal efficiency in pure dry processes. Compared with pure dry processes, the equipment investment cost is reduced by more than 40%.

[0019] 2. The wettability control mechanism fundamentally solves the industry pain point that wet-process pharmaceutical solutions are difficult to uniformly wet hydrophobic surfaces. Plasma pretreatment introduces hydroxyl active groups on the surface of inorganic films, reducing the water contact angle on the wafer surface from 60 to 80° to less than 10°, significantly improving the uniformity of film removal and the consistency of film removal.

[0020] 3. Controlled connection design effectively maintains wafer surface activity. Through three segmented connection schemes—vacuum transfer, inert gas protection, or timing control—it effectively suppresses secondary surface contamination and oxide layer thickening during wafer transfer, ensuring that the gain effect of plasma pretreatment does not diminish.

[0021] 4. The multi-tank series film removal system achieves highly selective multi-layer removal. The independent etching scheme of separate tanks avoids mutual interference between the selectivity of different etching solutions in the mixed acid system, and can achieve step-by-step high-precision removal of silica, silicon nitride and metal contaminants, with controllable damage to the silicon substrate.

[0022] 5. The method is highly adaptable and easy to implement on production lines. The three segmented transfer schemes can be adapted to different levels of production line configurations. They can be used for the integrated design of brand-new production lines as well as for the low-cost upgrade and transformation of existing production lines, and have strong scalability.

[0023] Instruction manual illustrations

[0024] Figure 1 This is a schematic diagram of the overall technical architecture of the plasma-activated modification-based selective film removal and regeneration method for 12-inch wafers proposed in this invention, which shows the connection relationship between each process stage and the core functional modules.

[0025] Figure 2 This is a schematic diagram of the core principle framework of plasma activation modification to achieve selective film removal and surface regeneration in this invention, which shows the reaction mechanism and hydroxylation pathway between plasma and different films.

[0026] Figure 3 This is a schematic diagram of the key process stages of selective film removal and regeneration of 12-inch wafers in this invention, showing each decision node and process parameter feedback control loop. Detailed Implementation

[0027] Example 1

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0029] This embodiment details the core process flow and technical implementation of a selective film removal and regeneration method for 12-inch wafers based on plasma activation modification. This method achieves efficient, highly selective, and low-damage removal of both organic and inorganic films through a segmented relay mechanism of dry plasma pretreatment and wet selective film removal. The following provides a detailed explanation of each step.

[0030] S1 Material Preparation Stage

[0031] In the S1 material preparation stage, a 12-inch wafer to be recycled is provided as the starting point for the process. The substrate of this wafer to be recycled is a 12-inch silicon substrate wafer, with organic and / or inorganic film layers attached to its surface. Typical compositions of the organic film layer include at least one or more combinations of positive photoresist, negative photoresist, photoresist development residue, fluoropolymer byproducts, and residual organic polymer layers. Typical compositions of the inorganic film layer include at least one of silicon dioxide, silicon nitride, and metal film layers. The metal film layer may include metals or compounds commonly used in the process, such as copper, aluminum, titanium, tungsten, and cobalt.

[0032] The incoming material preparation stage includes the following sub-operations: First, the incoming wafers are batch registered and uniquely identified, and the entire wafer process is tracked through barcodes or RFID tags; then, the wafers to be recycled undergo an incoming appearance inspection, using automated optical inspection equipment to scan the wafer surface to confirm the absence of structural damage such as cracks, chipping, and fragments, and that the wafer thickness meets the 12-inch standard specification of 300mm ± 0.2mm; next, the wafer surface film layers are preliminarily characterized, and the thickness distribution and uniformity of each film layer are measured using an elliptic polarization film thickness gauge, while the surface defect density and defect type distribution are recorded; finally, based on the film layer composition and thickness data, the wafers to be recycled are classified into the corresponding process queues.

[0033] Specifically, when the surface of the wafer to be regenerated is simultaneously covered with a photoresist layer and a silicon dioxide dielectric layer, it is classified into the SiO2+ photoresist mixed contamination treatment queue; when the surface of the wafer to be regenerated is covered with fluoropolymer residue and silicon nitride film, it is classified into the Si3N4+ fluoropolymer contamination treatment queue; when the wafer to be regenerated involves a multilayer composite film structure, the process priority scheduling is based on the type and thickness of the uppermost film layer.

[0034] S2 Dry Plasma Pretreatment Stage

[0035] In the S2 dry plasma pretreatment stage, the wafer to be regenerated is placed in the plasma processing chamber, process gas is introduced, and plasma is excited. This stage simultaneously achieves two core functions: first, to remove the surface organic film layer through an oxidative ashing reaction; and second, to introduce hydroxyl active groups on the surface of the inorganic film layer, reducing the surface contact angle and improving the wettability of subsequent wet processing solutions.

[0036] The plasma pretreatment uses an oxygen-based plasma system as the process gas source. The optional combinations of process gases include the following four options: Option 1 is a pure oxygen system, using high-purity oxygen as the sole process gas; Option 2 is a nitrous oxide system, using nitrous oxide as the process gas, which decomposes in a plasma environment to generate oxygen free radicals; Option 3 is a mixture of carbon tetrafluoride and oxygen, where the selective removal of fluoropolymer-containing organic residues is achieved by adjusting the mixing ratio; Option 4 is a mixture of sulfur hexafluoride and oxygen, suitable for treating sulfur-containing polymers or special mixed fluorine and sulfur residues. Preferably, when the surface of the wafer to be regenerated contains fluoropolymer residues, the process gas is a mixture of carbon tetrafluoride and oxygen at a volume ratio of 1:3, the RF power is 800 to 1200 W, the substrate temperature is 200 to 280°C, and the processing time is 150 to 250 s.

[0037] The core process parameters for plasma pretreatment are set as follows: the RF power is set from 300W to 1500W, with a typical value of 800W to 1000W. This power range ensures that the plasma density is sufficient to achieve rapid ashing of organic materials while avoiding excessive bombardment damage to the wafer surface; the chamber pressure is set from 0.1 Torr to 1 Torr, with a typical value of 0.3 Torr to 0.5 Torr. Pressure control affects the mean free path and collision probability of plasma ions, thus affecting the processing uniformity; the substrate temperature is set from 150℃ to 300℃, with a typical value of 220℃ to 250℃. Appropriate temperature increases help accelerate the thermal decomposition reaction of organic materials; the processing time is set from 30s to 300s, with a typical value of 120s to 200s. The specific time is optimized according to the thickness and type of the organic film layer.

[0038] (I) Chemical mechanism and quantitative control of hydroxylation of different substrate materials

[0039] The chemical pathways for introducing hydroxyl groups through plasma activation differ fundamentally for different inorganic substrate materials. This invention establishes the following substrate-specific mechanism model:

[0040] (1) SiO2 substrate: High-energy oxygen free radicals (O·) in oxygen plasma combine with dangling bonds on the SiO2 surface to first form peroxy free radicals (Si-OO·), which then react with water molecules or hydrogen free radicals in the gas phase to transform into silanol groups (Si-OH). The standard Gibbs free energy change of this reaction is ΔG° = -78 kJ / mol, and the reaction rate constant under plasma conditions is... E a =0.18 eV. Surface hydroxyl density Γ OH With plasma dose D p=P·t (where P is the radio frequency power and t is the processing time) exhibits a Langmuir-type adsorption relationship: , where Γ max =4.8×10 14 cm -2 (Theoretical monolayer hydroxyl saturation density on SiO2 surface), k a =0.025 W -1 ·s -1 (Empirical adsorption constant).

[0041] (2) Si3N4 substrate: The activation pathway on the silicon nitride surface is more complex. Oxygen plasma first oxidizes the surface Si-N bonds to Si-O bonds, releasing N2 molecules. Subsequently, the newly formed Si-O sites undergo hydroxylation along the same pathway as SiO2. This reaction requires overcoming a higher activation energy barrier (E). a =0.32 eV), therefore, under the same plasma dose, the hydroxyl density on the Si3N4 surface is approximately 60-70% of that on the SiO2 surface. The quantitative relationship is: , where τ N =180s (characteristic time constant of oxidation saturation on silicon nitride surface).

[0042] (3) Metal substrate (Cu, Al, Ti, W, Co): The hydroxylation mechanism on the metal surface is that oxygen plasma first oxidizes the metal surface layer to form a thin metal oxide layer (MO). x Subsequently, the oxide layer reacts with hydrated hydrogen ions to form a metal hydroxyl complex (M-OH). This process is significantly affected by the standard electrode potential of the metal. Taking Cu as an example (E... 0 Cu2+ (Cu = +0.337 V), its oxidation rate is limited by the growth kinetics of the surface oxide layer, following a parabolic oxidation law: , where k p The temperature-dependent parabolic rate constant. k for Cu p (250°C) = 3.2 × 10 14 cm -2 ·s -1 / 2 .

[0043] (II) Quantitative mapping relationship between hydroxyl density and contact angle

[0044] This invention establishes a quantitative mapping relationship between hydroxyl density Γ_OH and surface water contact angle θ through experimental calibration. Based on the Young-Dupre equation and the Wenzel roughness correction model, the relationship between the contact angle θ and the hydroxyl density follows the following functional form:

[0045]

[0046] Where r is the surface roughness factor (calibrated by atomic force microscopy, typical value 1.05-1.15), θ Y For an ideal smooth surface, the Young contact angle, V m The molar volume of water (18.0 cm³ / mol), γ LV Let be the liquid-gas interfacial tension (72.1 mN / m), R be the ideal gas constant (8.314 J / mol·K), and T be the temperature (taken as 298 K). The simplified engineering fitting form is:

[0047]

[0048] Where θ0 = 78° (initial contact angle of untreated SiO2 surface), A = 8.2° (experimental calibration constant), Γ ref =1.0×10 13 cm -2 (Refer to hydroxyl density). When Γ OH Reaching 3.5×10 14 cm -2 At that time, the calculated contact angle θ≈9.2°, which matches the measured value ≤10°. This mapping relationship allows technicians to deduce the required hydroxyl density from the target contact angle and calculate the optimal plasma dose accordingly.

[0049] (III) Control strategies for hydroxyl density based on core process parameters

[0050] The present invention establishes a multi-parameter coupled hydroxyl density control equation as follows:

[0051]

[0052] The parameters and their typical value ranges are as follows: α = 2.1 × 10 -3 (Cooperation coefficient), β=0.42 (Power index), γ=0.55 (Time index), δ=0.38 (Oxygen concentration index, C O2 (where Ea is the oxygen partial pressure percentage) is the activation energy of the corresponding substrate material, and k is the activation energy. B This is the Boltzmann constant. This governing equation allows technicians to determine the contact angle of a given target (corresponding to the target Γ). OH Under the premise of adjusting the radio frequency power P, processing time t, and oxygen partial pressure C, O2 By considering four degrees of freedom—including substrate temperature T—multiple sets of equivalent process parameter combinations are obtained to adapt to the constraints of different equipment. An example control strategy table is shown below:

[0053] ≤10° <![CDATA[≥3.5×10 14 ]]> 800 180 80% 220 ≤8° <![CDATA[≥4.2×10 14 ]]> 1000 200 80% 250 ≤6° <![CDATA[≥4.6×10 14 ]]> 1200 240 90% 270

[0054] After plasma pretreatment, the contact angle of the wafer surface with water is no greater than 10°. This contact angle is a core parameter for evaluating the surface activation effect, and its physical meaning characterizes the degree of hydrophilicity of the wafer surface. The contact angle is measured using a static contact angle meter. A deionized water droplet is placed on the wafer surface, a side view of the droplet is photographed, and the contact angle value is calculated. When the contact angle drops below 10°, it indicates that the plasma treatment has introduced a sufficient concentration of hydroxyl active groups onto the inorganic film surface. These hydroxyl groups significantly improve the chemical activity and wetting properties of the wafer surface.

[0055] (iv) Technical details of plasma pretreatment stage

[0056] The plasma generator employs either capacitively coupled or inductively coupled plasma sources, with radio frequency frequencies of 13.56MHz or 2.45GHz. A matching network achieves impedance matching to maximize power transmission efficiency. The flow rate of process gases is precisely set via a mass flow controller, with typical oxygen flow rates ranging from 50 sccm to 200 sccm. The proportion of inert gases, such as argon, is adjusted according to specific process requirements. Adding argon can increase plasma density and improve processing uniformity. The wafers are placed in the chamber using horizontal trays. The substrate temperature is controlled in conjunction with a circulating cooling system and a heater, with a temperature control accuracy of ±2℃.

[0057] S3 segmented controlled transfer phase

[0058] In the S3 segmented controlled transfer stage, the plasma-pretreated wafer is transferred to the wet stripping station under a controlled environment. The core objective of this stage is to maintain the active state of the wafer surface, suppress secondary contamination and oxidation thickening, and ensure that the gain effect of plasma pretreatment does not significantly decrease before the wet stripping process begins.

[0059] Segmented controlled transfer includes three optional implementation methods, each suitable for different production line configurations and cost budget conditions.

[0060] The first implementation method is vacuum transfer. The present invention discloses in detail the engineering implementation scheme of this method as follows, to overcome the physical barriers to the transition between vacuum and atmospheric pressure:

[0061] (1) Equipment layout and cavity connection: A Load Lock transition cavity (LL cavity) is set between the plasma processing chamber (P1 cavity) and the wet film removal station (W1 tank). The LL cavity is an independent sealed cavity, and its two ends are isolated from the upstream and downstream by vacuum valves (V1 valve, connected to P1 cavity; V2 valve, connected to wet side buffer cavity). The LL cavity is equipped with an independent vacuum pump group (molecular pump + mechanical pump) and an inert gas return pipeline (N2 or Ar).

[0062] (2) Pressure Equalization and Air-Breaking Pressure Equalization Sequence: After the wafer is transferred from the P1 cavity (typical pressure 0.3-0.5 Torr) to the LL cavity, the pressure in the LL cavity needs to be increased to near atmospheric pressure (760 Torr) before the V2 valve can be opened to enter the wet process side. To prevent airflow impact damage to the wafer surface activation layer caused by sudden pressure changes, this invention specifies a graded pressure equalization sequence:

[0063] Phase 1: After valve V1 is closed, the LL chamber gradually increases from 0.3-0.5 Torr at a rate of 0.5 Torr / s to 1 Torr (lasting about 1-2 seconds).

[0064] Phase 2: From 1 Torr to 100 Torr, the boost rate is controlled within 5 Torr / s (lasting approximately 20 seconds).

[0065] Phase 3: From 100 Torr to 760 Torr, recharge with high-purity N2 (purity ≥ 99.999%), with the pressurization rate controlled within 20 Torr / s (lasting approximately 33s).

[0066] The total pressure equalization time is about 55-60 seconds. This slow-rise sequence ensures that the active layer on the wafer surface is not damaged by airflow impact, while avoiding secondary contamination caused by particle peeling due to rapid pressure change.

[0067] (3) Anti-acid mist backflow isolation structure: An acid mist atmosphere (containing volatile components such as HF and H3PO4) exists above the liquid surface of the wet corrosion tank (W1 tank). To prevent the acid mist from diffusing back into the LL chamber or even the P1 chamber when the V2 valve is opened, causing cross-contamination of the chambers, the present invention provides the following isolation structure:

[0068] Wet-side buffer chamber: An intermediate buffer chamber (chamber B) is added between the W1 tank and the V2 valve. The B chamber is isolated from the W1 tank by a liquid-sealed gate. Clean dry air or N2 is continuously introduced into the B chamber in a downward laminar flow at a velocity ≥0.5 m / s, forming an air curtain barrier to prevent acid mist from rising.

[0069] The V2 valve adopts a dual-gate series structure, forming a slightly positive pressure (5-10 Torr) N2 purging zone between the two gates to further isolate acid mist.

[0070] In the transfer path from the LL cavity to the B cavity, the wafer is carried horizontally by a vacuum robot. The end of the robot is equipped with an anti-drip baffle to prevent any condensed acid liquid that may be present at the bottom of the B cavity from dripping onto the wafer surface.

[0071] (4) Full-process timing control: The entire process from the completion of P1 cavity processing to the wafer entering the W1 etch solution is ≤90s, of which the pressure equalization of LL cavity is about 60s and the transfer from B cavity to W1 etch solution is about 15s. Under this method, the surface contact angle increment during the transfer process is ≤2°. This transfer scheme is the best in terms of surface activity retention among the three transfer methods and is suitable for high-precision regeneration scenarios with the most stringent requirements for activation effect.

[0072] The second implementation method is inert gas protection, which uses a positive pressure channel created by continuously introducing nitrogen or argon gas for transfer. The specific technical solution is as follows: the wafer transfer channel is designed as a sealed structure, with high-purity nitrogen or argon gas continuously introduced to create a slightly positive pressure environment to prevent external air infiltration; the purity of the inert gas is required to be no less than 99.999% to avoid the negative impact of impurities on the activated surface; an airflow guiding structure is installed inside the transfer channel to ensure uniform positive pressure distribution and eliminate dead zones. The transfer time under inert gas protection is ≤180s, and the surface contact angle change during the transfer process is ≤8°. This solution maintains surface activity while having a low implementation cost, making it suitable for medium-configuration wafer re-production lines.

[0073] The third implementation method is a timing control method, which uses a manufacturing execution system (MES) for real-time scheduling to ensure that the transfer interval is no more than 5 minutes. The specific technical solution is as follows: the existing plasma ashing machine and wet processing station on the production line are kept independently, requiring no new hardware. The MES schedules the wafer process in real time. Based on the estimated completion time of the plasma processing step, the system pre-arranges the occupancy time of the wet stripping station, ensuring that wafers enter the wet processing step first after dry processing. When the MES detects a signal indicating the completion of dry processing, it automatically triggers the preheating and chemical circulation preparation process of the wet processing station, minimizing waiting time. Verification results of the timing control method show that, under a transfer interval ≤300s, the surface contact angle attenuation is ≤10% (relative to the contact angle value before transfer). Taking the contact angle before transfer as 8.5°, it rises to approximately 9.3° after transfer, with an attenuation of approximately 0.8°, which translates to an attenuation ratio of approximately 9.4%, meeting the requirements of mass production processes. This solution requires no new hardware investment; process upgrades can be achieved simply through production line management optimization, adapting to the low-cost transformation needs of existing production lines.

[0074] It should be noted that the contact angle change evaluation indicators (increment, change value, and attenuation percentage) for the above three transfer methods are all for measuring the same physical process—that is, the degree of hydrophilic degradation of the activated surface during the transfer process. The three measurement methods can be converted to each other within the contact angle measurement accuracy range known to those skilled in the art. The core process requirement is that the contact angle of the wafer surface after controlled transfer and before entering the wet process should still be maintained within a process window of no more than 10°.

[0075] S4 Wet Process Selective Film Removal Stage

[0076] In the selective stripping stage of the S4 wet etching process, a multi-tank tandem etching system is used, sequentially applying etchants of corresponding selectivity to remove the inorganic film layer on the wafer surface step by step. A high selectivity to the silicon substrate is maintained during etching to avoid excessive damage to the substrate.

[0077] The multi-tank series selective film removal system includes at least a dilute hydrofluoric acid etching tank, a hot phosphoric acid etching tank, and an SC1 / SC2 cleaning tank arranged in sequence. A deionized water rinsing section is provided between adjacent tanks to avoid cross-contamination of the chemicals. The design concept of this system is to configure dedicated etching tanks for different types of inorganic films. Independent etching in separate tanks avoids mutual interference between the selectivity of different etching solutions in a mixed acid system, achieving high-precision selective removal of multi-layer composite films.

[0078] The function of a dilute hydrofluoric acid etching tank is to remove silica films, including those formed by thermal oxidation, chemical vapor deposition, and borosilicate glass. The tank uses a 1% to 5% (by mass) dilute hydrofluoric acid solution, typically 2%. The etching temperature is controlled at a constant 20 to 30°C, preferably 25°C. During etching, the infrared reflectance spectrum is monitored in real time, and the etching endpoint is determined when the silica film thickness drops to zero. The infrared reflectance monitoring principle is based on the difference in reflectivity of different film layers to infrared light. When the silica layer is completely removed, the reflectance spectrum signal changes significantly, triggering the etching endpoint determination logic. The selectivity ratio of the dilute hydrofluoric acid etching process to the silicon substrate is typically higher than 100:1, meaning that while removing a 1000nm silica film, the loss to the silicon substrate does not exceed 10nm.

[0079] The function of the hot phosphoric acid etching tank is to remove silicon nitride films, including low-pressure chemical vapor deposition (LPCVD) and plasma-enhanced chemical vapor deposition (PECVD). The hot phosphoric acid etching tank uses a hot phosphoric acid solution at 165-175°C, with a typical value of 170°C. The etching endpoint of the silicon nitride film is determined by online monitoring of the redox potential (ORP). During the etching process, the selectivity to the silicon substrate is no less than 50:1. The monitoring principle of ORP is as follows: when a silicon nitride film is present, the concentration of the reaction products between phosphoric acid and silicon nitride in the etching solution remains at a certain level, and the ORP signal is stable. When the silicon nitride is completely removed and the etching solution comes into contact with the silicon substrate, the type of reaction products changes, and the ORP signal shifts significantly. The system determines the etching endpoint based on this shift.

[0080] The SC1 cleaning tank is designed to remove particulate and metal ion contaminants. Its solution ratio is ammonia:hydrogen peroxide:water = 1:2:10, with a typical cleaning temperature of 65℃ and a cleaning time of 600 seconds. The SC2 cleaning tank is designed to remove heavy metal ion contaminants. Its solution ratio is hydrochloric acid:hydrogen peroxide:water = 1:2:8, with a typical cleaning temperature of 65℃ and a cleaning time of 600 seconds.

[0081] The process flow design of the multi-tank tandem selective film removal system is as follows: The wafer sequentially enters a dilute hydrofluoric acid etching tank to remove the silicon dioxide layer, followed by a deionized water rinsing tank for 120 seconds to remove residual etching solution. It then enters a hot phosphoric acid etching tank to remove the silicon nitride layer, followed by another deionized water rinsing tank for 120 seconds. Finally, it enters the SC1 / SC2 cleaning tank to remove particulate and metal contaminants. This process design ensures that cross-contamination does not occur between adjacent tanks, and each rinsing step is designed with sufficient rinsing time and deionized water flow rate to ensure cleaning effectiveness.

[0082] Detailed disclosure of the multi-parameter fusion algorithm for determining corrosion endpoint

[0083] The corrosion endpoint is determined by the conductivity σ of the chemical solution, the redox potential E, and the film thickness T by infrared reflectance spectroscopy. film Three-channel online monitoring enables precise endpoint control. The specific algorithm, weight allocation, threshold setting, and conflict handling logic of the fusion judgment function f disclosed in this invention are as follows:

[0084] (1) Extraction of feature signals for each channel:

[0085] Conductivity Channel: Real-time conductivity σ(t) of the corrosive solution is collected, and its first derivative dσ / dt is calculated. The conductivity characteristics of the dissolution products differ depending on the film layer. Taking the corrosion of SiO2 in dilute HF as an example, the dissolution product H2SiF6 increases the conductivity, and dσ / dt approaches zero from a positive value before the endpoint. Conductivity Normalized Characteristic Signal ,in The mean value of steady-state fluctuations is approximately 0.02 μS / cm·s. The noise standard deviation is approximately 0.005 μS / cm·s.

[0086] Redox potential channel: The ORP value E(t) of the etching solution is collected, and its first derivative dE / dt is calculated. Taking the etching of Si3N4 in hot phosphoric acid as an example, silicon nitride dissolves and releases NH4. + The presence of SiO2 colloid causes the ORP to drift negatively from approximately +200mV, with dE / dt jumping abruptly from negative to near zero at the endpoint. (Normalized characteristic signal) ,in ≈-0.5mV / s, ≈0.1 mV / s.

[0087] Infrared reflectance spectral channel: Real-time inversion of remaining film thickness T film(t) Calculate its first derivative dT film / dt (i.e., corrosion rate). Normalized characteristic signal. , where μ T S represents the average rate during the constant-rate corrosion phase. T This represents the standard deviation of the rate fluctuation.

[0088] (2) Algorithm form of fusion function f: This invention adopts weighted linear weighted fusion (weighted average method) as the main decision logic, supplemented by decision-level fusion of confidence assessment of each channel, the specific form of which is as follows:

[0089] Overall criteria for the finish line:

[0090] The weight allocation is adaptively set based on the signal-to-noise ratio (SNR) of each channel: W E W T Similarly, the typical weight value is... =0.25, W E =0.30, w T =0.45 (The infrared reflectance spectrum has the highest signal-to-noise ratio, so it is given the highest weight).

[0091] (3) Threshold setting and judgment logic:

[0092] The main criterion in normal mode is: when F(t) is lower than the threshold F for three consecutive sampling periods. th When the value is -0.75, an endpoint warning is triggered.

[0093] Auxiliary judgment condition: T must be satisfied simultaneously. film(t) ≤2 nm (the residual film thickness is close to zero), and the absolute values ​​of dσ / dt and dE / dt are less than 1.5 times their respective steady-state fluctuation thresholds.

[0094] Final endpoint trigger: When both the primary and secondary conditions are met, the system outputs an endpoint determination signal, triggering the process to move to the next step.

[0095] (4) Conflict handling logic: When the judgment conclusions of the three channels are inconsistent (for example, the infrared signal shows that the film thickness is 0, but the ORP signal has not yet reached the endpoint threshold), the following hierarchical arbitration strategy is adopted:

[0096] First priority: Infrared reflectance spectral signal S T (Direct measurement of film thickness has the clearest physical meaning and the lowest misjudgment rate). If S TIf the film thickness is ≤2nm and lasts for more than 5s, the system can independently trigger the endpoint (single-channel confirmation mode) even if other channels have not yet been confirmed.

[0097] Second priority: When the infrared signal is abnormal (e.g., the reflectivity of the wafer surface is distorted due to excessive roughness), activate the ORP signal S. E As the primary reference, the decision threshold is adjusted to F. th =-0.60, and cross-validation was performed using the conductivity signal as a reference.

[0098] Third priority: If the conductivity and ORP signal contradict each other (e.g., dσ / dt indicates that the etching has stopped while dE / dt still shows that the reaction is in progress), the system automatically enters "extended time mode": the etching continues at a slow speed (1 / 10 of the normal rate) and the infrared signal is continuously monitored until the infrared signal confirms the end point or the preset maximum extended time (≥30s) is reached, then the system is forcibly terminated and the wafer is marked for individual testing.

[0099] (5) Endpoint monitoring sampling frequency and window: The three-channel synchronous sampling rate is set to 10Hz, and the moving average window width is 5 sampling points (0.5s) to filter out high-frequency noise. The endpoint judgment window length is set to 3 sampling periods (0.3s) to ensure the stability and anti-jitter of the judgment results.

[0100] The determination of the corrosion endpoint is achieved through precise endpoint control via online monitoring of the chemical solution conductivity, redox potential, or infrared reflectance spectroscopy. Taking infrared reflectance spectroscopy as an example, its basic principle is as follows: When an infrared beam is incident on the wafer surface, different film layers have different refractive indices and absorption coefficients for infrared light, resulting in interference fringes in the reflectance spectrum related to film thickness. By acquiring the reflectance spectrum in real time and comparing it with a preset film thickness-spectrum database, the film thickness at the current moment can be accurately calculated. When the film thickness value corresponding to the monitored signal drops below a preset threshold, the removal of the corresponding film layer is considered complete, and the system automatically triggers the process transfer to the next step. This method achieves a monitoring accuracy of up to 1 nm, fully meeting the endpoint control requirements of wafer regeneration processes.

[0101] S5 Post-processing and Inspection Stage

[0102] In the S5 post-processing and inspection stage, the wafers after film removal undergo post-processing procedures and quality inspections to ensure that the regenerated wafers meet the Spec-In specification requirements.

[0103] The post-processing steps include the following: First, a final rinse with deionized water is performed. This step uses high-resistivity deionized water to thoroughly rinse the wafer surface to remove residual chemicals. The rinsing time is no less than 120 seconds, and the resistivity of the deionized water is no less than 18 MΩ·cm. Next, isopropanol dehydration is performed. The wafer is immersed in high-purity isopropanol to remove surface water stains. This step can effectively prevent water stains from forming water spot defects after drying. The dehydration time is 60 seconds. Finally, nitrogen drying is performed. Clean and dry nitrogen is used to blow and dry the wafer surface. The nitrogen dew point is below -60°C to ensure the drying effect.

[0104] After drying, film thickness measurement and surface defect detection are performed. Film thickness measurement is conducted using an elliptic polarization film thickness gauge to measure multiple points on key locations on the wafer surface, assessing the integrity of film removal and ensuring no residual film layer remains on the surface. Surface defect detection is performed using a dark-field particle detector or a bright-field defect detector to scan the entire wafer surface and count the number of surface particles and defects.

[0105] The acceptance criteria for reclaimed wafers are set as follows: the number of surface particles is not higher than 10 (the particle size is not less than 0.12μm), the amount of silicon substrate removed is not greater than 5nm, and there are no obvious structural defects such as scratches, chipping, or cracks on the surface. Wafers that meet the above conditions are judged as qualified reclaimed wafers and can be reused in subsequent processes.

[0106] The method of the present invention will be further explained below in conjunction with specific application scenarios.

[0107] Specific application example 1: Regeneration treatment of wafers contaminated with a mixture of SiO2 and photoresist.

[0108] The wafer to be regenerated is a 12-inch silicon substrate wafer with a 3μm thick positive photoresist coating on its surface and a 1000nm thick silicon dioxide dielectric layer grown underneath. After defect detection confirms that there is no structural damage, it enters the S2 dry plasma pretreatment process.

[0109] The wafer was placed in a plasma ashing chamber, evacuated to a chamber pressure of 0.5 Torr, and a mixture of oxygen and argon (volume ratio 4:1) was introduced. The RF power was set to 800W, the substrate temperature to 220℃, and the processing time to 120s. Under these process parameters, the photoresist removal rate reached over 99.5%, hydroxyl active groups were successfully introduced onto the silicon dioxide surface, the water contact angle on the wafer surface was reduced to 7.2°, and there were no obvious organic residues on the surface.

[0110] After plasma treatment, the wafer is directly transferred to the wet etching station in a vacuum environment through the Load Lock transfer cavity. There is no air exposure during the transfer process, the surface contact angle increment is ≤2°, and the activation effect is well maintained.

[0111] The wafer enters the first etching tank and is etched at a constant temperature of 25°C using a 2% (w / w) dilute hydrofluoric acid solution to remove the silicon dioxide layer. The film thickness is monitored in real time using infrared reflectance spectroscopy. The etching endpoint is determined when the silicon dioxide film thickness drops to 0, with a total etching time of approximately 180 seconds. After etching, the wafer is rinsed with deionized water for 120 seconds and then placed in an SC1 cleaning tank (ammonia:hydrogen peroxide:water = 1:2:10, 65°C) for 600 seconds to remove particulate and metal ion contaminants.

[0112] After being dehydrated with isopropanol and dried with nitrogen, the wafers underwent film thickness and surface defect testing. The results showed that the silicon substrate removal was less than 5 nm, and the number of surface particles (particle size not less than 0.12 μm) was less than 10, meeting the Spec-In regenerated wafer specifications.

[0113] Specific Application Example 2: Regeneration Treatment of Si3N4+ Fluoropolymer Contaminated Wafers

[0114] The wafers to be recycled are discarded wafers from the dry etching process, with residual fluoropolymer byproducts on the surface, and the substrate is a silicon wafer covered with silicon nitride. They then proceed to the S2 dry plasma pretreatment process.

[0115] A mixture of carbon tetrafluoride and oxygen (volume ratio 1:3) was introduced, with a chamber pressure of 0.3 Torr, RF power of 1000W, substrate temperature of 250℃, and processing time of 200s. Under these process parameters, the ashing removal and surface activation of the fluoropolymer were simultaneously achieved, and hydroxyl active groups were successfully introduced onto the silica surface.

[0116] The wafer is transferred through a closed transfer channel, in which high-purity nitrogen is continuously introduced to form a slight positive pressure. The transfer time is ≤180s, and the surface contact angle changes ≤8° during the transfer process.

[0117] The multi-tank series selective film removal process is as follows: The first tank is a pre-cleaning process with dilute hydrofluoric acid to remove the natural oxide layer on the surface for 60 seconds, followed by rinsing with deionized water for 90 seconds; the second tank is a hot phosphoric acid solution at 165℃ to etch and remove the silicon nitride film layer, with the endpoint determined by online monitoring of the oxidation-reduction potential, and the etching time is about 15 minutes; the third tank is a cleaning solution with SC2 (hydrochloric acid:hydrogen peroxide:water = 1:2:8, 65℃) to remove metal ion contaminants for 600 seconds.

[0118] After slow drying, the results showed that the silicon nitride was completely removed, the amount of corrosion on the silicon substrate was less than 2nm, and there were no obvious defects on the surface.

[0119] Example 2

[0120] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0121] This embodiment focuses on the regeneration of 12-inch wafers with multilayer composite film structures, illustrating the process implementation and technical advantages of the method of the present invention in complex film systems.

[0122] In regeneration scenarios involving multilayer composite film structures, the surface of the wafer to be regenerated typically contains multiple types of stacked films. For example, the bottom layer may be a silicon substrate, sequentially covered with a thermally oxidized silicon dioxide layer, a silicon nitride passivation layer, a metal interconnect layer, and an uppermost plasma-enhanced chemical vapor deposition silicon dioxide protective layer and a photoresist patterning layer. For such complex film structures, this invention employs a step-by-step, layer-by-layer removal process strategy to ensure the precision and selectivity of each film removal step.

[0123] The process flow design for the multilayer composite film structure is as follows: First, the uppermost visible light resist and organic polymer residue layer are removed through plasma pretreatment in the S2 dry section, while the exposed lower inorganic film layer surface is activated. Then, the first stage of the selective film removal process in the S4 wet section is initiated, using a dilute hydrofluoric acid etching bath to preferentially remove the top plasma-enhanced chemical vapor deposition silicon dioxide layer. This layer has a relatively loose structure and reacts quickly with dilute hydrofluoric acid. After the first stage, deionized water rinsing is performed, followed by the second stage, where a dedicated fluoropolymer etching solution is used to remove the residual fluoropolymer intermediate layer. After rinsing, the third stage is initiated, using a hot phosphoric acid etching bath to remove the silicon nitride passivation layer. This process requires a long etching time to ensure complete removal of the silicon nitride covering the sidewalls. Finally, the fourth stage is initiated, using a dilute hydrofluoric acid etching bath to remove the bottom thermally oxidized silicon dioxide layer. This layer is typically thin but requires high protection of the silicon substrate, thus necessitating precise control of the etching endpoint.

[0124] The endpoint determination of multilayer composite film structures employs a multi-parameter fusion monitoring strategy. The core monitoring parameters include signals from three channels: chemical conductivity, redox potential, and infrared reflectance spectroscopy. Chemical conductivity monitoring leverages the different conductivity characteristics of the dissolution products of different film materials in the corrosive solution; by acquiring the conductivity change curve of the corrosive solution in real time, the progress of the current corrosion reaction can be determined. Redox potential monitoring provides information on the chemical state of the corrosion reaction; different film materials correspond to different potential baseline values. Infrared reflectance spectroscopy monitoring provides direct film thickness information. These three monitoring signals are comprehensively determined through a data fusion algorithm, effectively avoiding potential misjudgments that may occur with a single monitoring channel and improving the reliability and accuracy of endpoint determination.

[0125] The algorithm for multi-parameter fusion monitoring can be implemented using the following logical expression for online judgment:

[0126] Where σ represents the conductivity of the drug solution, E represents the redox potential, and T represents the... filmThe value represents the film thickness monitored by infrared reflectance spectroscopy, and f represents the multi-parameter fusion judgment function. When the rate of change of the above three parameters simultaneously meets the preset endpoint judgment condition, the system outputs an endpoint trigger signal, stops the current etching process, and proceeds to the next process.

[0127]

[0128]

[0129]

[0130]

[0131] ,

[0132] ,

[0133] .

[0134] Step 3: Fusion Judgment Rules

[0135] The two-vote pass system is adopted for fusion judgment: when any two of the three endpoint markers are met simultaneously, the central controller outputs an endpoint trigger signal, immediately stops the etching solution circulation pump and the wafer lifting mechanism, and locks and archives the process parameters of this batch.

[0136]

[0137]

[0138] The advantage of this fusion judgment algorithm lies in utilizing the information redundancy of multiple independent monitoring channels. Even if a monitoring channel malfunctions due to noise interference or sensor failure, it can still be compensated and judged by information from other channels, significantly improving the robustness and reliability of the system.

[0139] Controlling the uniformity of film removal in multilayer composite film structures is one of the key process challenges. Due to potential differences in film thickness across different regions, the processing rates at the wafer center and edges may differ, resulting in film residue in some areas while other areas have undergone excessive etching after removal. This invention addresses this problem through the following technical means: First, precise control of pretreatment uniformity is achieved by adjusting gas and power distribution during the plasma pretreatment stage; second, the uniformity of the etching reaction is ensured by optimizing the flow field distribution of the etching solution during the wet film removal stage; third, statistical process control is performed using post-removal detection data, automatically triggering process parameter adjustments when a uniformity deviation exceeds a threshold.

[0140] The acceptance criteria for the regeneration treatment of multilayer composite film structures also follow the general standard of no more than 10 surface particles and no more than 5 nm of silicon substrate removal. However, since the total thickness of multilayer composite films is usually large, it is necessary to inspect the film thickness residue in each step of the film removal process to ensure that the final surface condition meets the requirements.

[0141] Example 3

[0142] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0143] This embodiment focuses on the scenario of mass production line transformation, illustrating the technical implementation scheme and verification results of the present invention to achieve low-cost upgrades on the basis of existing production lines.

[0144] In mass production line retrofit scenarios, the production line is typically equipped with independent plasma ashing machines and wet wafer laminators, with wafer transfer between the two sets of equipment conducted manually or semi-automatically. The core challenge in this scenario lies in how to effectively integrate dry and wet stripping processes using existing equipment while minimizing hardware modification costs.

[0145] The solution proposed in this invention is a time-optimized dry-wet segmented film removal process. Its core technical points are as follows: The manufacturing execution system (MES) schedules wafer processes in real time, prioritizing wafers that have completed dry processing before proceeding to the wet process, ensuring that the interval between the end of the dry process and the start of the wet process is ≤300 seconds. This solution requires no additional hardware such as a load lock transfer cavity or inert gas protection channel; it can be achieved solely through software upgrades and process scheduling optimization within the production line management system.

[0146] The detailed implementation of the timing optimization scheme is as follows: The manufacturing execution system maintains a dynamic process queue. Each wafer in the queue contains information such as the current process status, estimated completion time, and the next target process. When the plasma ashing process completes the processing of the current wafer, the system immediately queries the occupancy status of the wet processing station. If the wet processing station is idle and the chemical solution conditions meet the requirements, a wafer transfer command is immediately triggered. If the wet processing station is occupied, the system calculates the optimal waiting sequence scheme based on the queue information to ensure that the waiting time of the target wafer is minimized.

[0147] To verify the effectiveness of the timing optimization scheme, the following experiments were conducted: 100 12-inch test wafers with a surface contamination of photoresist and silicon dioxide were selected. Dry and wet stripping processes were performed according to the timing optimization scheme, with a target transfer interval of ≤300s (5 minutes). Experimental results showed that the average transfer interval for the 100 wafers was 3.7 minutes (approximately 222s), meeting the design requirement of no more than 5 minutes. After transfer, the average contact angle on the wafer surface was 9.3°, an increase of 0.8° compared to the initial contact angle of 8.5°, representing a contact angle attenuation of approximately 9.4% (≤10% acceptance standard). Regarding stripping uniformity, the standard deviation of silicon substrate removal was 1.2nm, and the stripping uniformity deviation was 1.8%, meeting the acceptance standard of no more than 2%.

[0148] The three transfer methods of this invention (vacuum transfer, inert gas protection, and timing control) differ in their time thresholds and allowable contact angle variations. The vacuum transfer method offers optimal surface activity retention (≤90s, contact angle increment ≤2°), suitable for regeneration scenarios with the highest precision requirements. The inert gas protection method provides good protection at a moderate cost (≤180s, contact angle variation ≤8°). The timing control method achieves process upgrades with zero hardware investment (≤300s, attenuation ≤10%). All three methods ensure that the wafer's surface contact angle remains within a process window of no more than 10° before entering the wet processing stage. Those skilled in the art can flexibly choose the most suitable transfer scheme based on actual production line configuration and cost budget.

[0149] Another technical advantage of the timing optimization solution is its strong scalability. When the production line needs to expand its capacity or upgrade its processes, a vacuum transfer solution or an inert gas protection solution can be added to the existing system to form a multi-level interconnected surface activity maintenance system. This solution can be used for the integrated design of brand-new production lines as well as for the low-cost upgrade and transformation of existing production lines, adapting to the configuration requirements of different levels of production lines.

[0150] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A selective film removal and regeneration method for 12-inch wafers based on plasma activation modification, characterized in that, Includes the following steps: S1. Perform incoming material inspection and film characterization on the 12-inch wafer to be regenerated. The wafer substrate is a silicon substrate with organic and / or inorganic film layers attached to its surface. S2. Place the wafer in a plasma chamber and introduce oxygen-based process gas to excite the plasma, simultaneously achieving the oxidative ashing removal of the organic film layer and the activation of hydroxyl groups on the surface of the inorganic film layer, so that the water contact angle on the surface of the treated wafer is ≤10°. S3. Transfer the plasma-treated wafer to the wet film removal station in a controlled environment to maintain the surface activation state. S4. A multi-tank series system is used to remove the inorganic film layer stepwise using selective etching solutions, with deionized water rinsing after each etching step. The etching endpoint is determined by conductivity σ, redox potential E, and film thickness T measured by infrared reflectance spectroscopy. film The three-channel online monitoring signals are fused and determined by the function. The determination is made in the fusion determination function F(t), where the characteristic signals Sσ(t), SE(t), and ST(t) of each channel are normalized signals. When F(t) is lower than the threshold F for three consecutive sampling periods... th The endpoint signal is triggered when the residual film thickness is ≤2nm; S5. Post-processing and inspection stage: The wafer after film removal is rinsed, dried and inspected.

2. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, The organic film layer includes at least one of positive photoresist, negative photoresist, photoresist development residue, fluoropolymer byproducts, and organic polymer residue layer; the inorganic film layer includes at least one of silicon dioxide layer, silicon nitride layer, and metal film layer, wherein the metal film layer includes at least one of copper, aluminum, titanium, tungsten, and cobalt.

3. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In the S2 stage, the oxygen-containing process gas is selected from a pure oxygen system, a nitrous oxide system, a carbon tetrafluoride and oxygen mixture system, or a sulfur hexafluoride and oxygen mixture system. When the surface of the wafer to be regenerated contains residual fluoropolymers, a carbon tetrafluoride and oxygen mixture is used with a volume ratio of 1:

3. The process parameters for plasma pretreatment are: RF power 300W to 1500W, cavity pressure 0.1 Torr to 1 Torr, substrate temperature 150℃ to 300℃, and processing time 30s to 300s.

4. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In stage S3, the controlled environment is achieved using a vacuum transfer method: a Load Lock transition chamber and a wet-side buffer chamber are set between the plasma processing chamber and the wet stripping station. The Load Lock transition chamber is connected to the plasma processing chamber via a first vacuum valve and to the wet-side buffer chamber via a second vacuum valve. A liquid-sealed gate is set between the wet-side buffer chamber and the etching tank. During the transfer, a graded pressure equalization sequence is executed: from 0.3-0.5 Torr, the pressure is increased at 0.5 Torr / s to 1 Torr, then at 5 Torr / s to 100 Torr, and then high-purity nitrogen is introduced at 20 Torr / s to increase the pressure to 760 Torr. A downward laminar clean gas is continuously introduced into the wet-side buffer chamber to form an air curtain barrier, with an airflow rate ≥0.5 m / s. Under the vacuum transfer method, the total transfer time from the completion of plasma processing to the wafer entering the etching solution is ≤90s, and the surface contact angle increment during the transfer is ≤2°.

5. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In the S3 stage, the controlled environment is achieved using an inert gas protection method: the wafer transfer channel is a closed structure, and nitrogen or argon gas with a purity of ≥99.999% is continuously introduced into the channel to form a micro-positive pressure environment. Under the inert gas protection method, the transfer time is ≤180s, and the surface contact angle change value during the transfer process is ≤8°.

6. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In the S3 stage, the controlled environment is implemented using a timing control method: the manufacturing execution system schedules the process in real time to ensure that the wafer after the completion of the S2 stage enters the S4 stage within 5 minutes. Under this method, the surface contact angle attenuation during the transfer process is ≤10%.

7. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In stage S4, the multi-tank series selective film removal system includes at least a dilute hydrofluoric acid etching tank, a hot phosphoric acid etching tank, an SC1 cleaning tank, and an SC2 cleaning tank arranged in sequence. The dilute hydrofluoric acid etching tank uses a dilute hydrofluoric acid solution with a mass fraction of 1% to 5% to remove the silicon dioxide film. The hot phosphoric acid etching tank uses a hot phosphoric acid solution at 165°C to 175°C to remove the silicon nitride film, with a silicon substrate selectivity ratio of not less than 50:

1. The chemical solution ratio of the SC1 cleaning tank is ammonia:hydrogen peroxide:water = 1:2:10, and the chemical solution ratio of the SC2 cleaning tank is hydrochloric acid:hydrogen peroxide:water = 1:2:

8.

8. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In the S4 stage, the characteristic signals of each channel in the fusion decision function F(t) are defined as follows: , , ,in μ E μ T This represents the average steady-state fluctuation of each channel. , , The standard deviation of noise; weights , , Typical weight value =0.25、 =0.30、 =0.45; in normal mode, the threshold F th =-0.75; the three-channel synchronous sampling rate is 10Hz, and the moving average window is 5 sampling points.

9. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In stage S4, the endpoint determination employs the following conflict resolution logic: when the determination conclusion of the fusion determination function F(t) is inconsistent with the direct film thickness determination conclusion of the infrared reflectance spectral channel S_T, S_T takes precedence. T The determination of S is of the highest priority; when S T When the signal is abnormal, the redox potential channel S E The decision is given as the second priority, and the threshold is adjusted to F. th =-0.60; when the conductivity channel With S E When the judgment conclusions are contradictory, the system enters the extended time mode, continues to corrode at 1 / 10 of the normal corrosion rate and continuously monitors the infrared signal until the infrared signal confirms the end point or reaches the preset maximum extended time of 30 seconds, after which it is forcibly terminated.

10. The selective film removal and regeneration method for 12-inch wafers based on plasma activation modification according to claim 1, characterized in that, In the S5 stage, the final rinsing time of the deionized water is ≥120s, and the resistivity of the deionized water is ≥18MΩ·cm; the nitrogen dew point of the nitrogen drying is ≤-60℃; the number of particles on the surface of the regenerated wafer that has passed the test is ≤10, and the amount of silicon substrate removed is ≤5nm.

Images