A gradient modulus structured polyurethane composite polishing pad and method of making the same

Abstract

The present application relates to the technical field of semiconductor polishing, in particular to a polyurethane composite polishing pad with gradient modulus structure and a preparation method thereof.The present application overcomes the problem that the mechanical properties and polishing durability of the composite polishing pad are difficult to be considered in the prior art.The present application builds a gradient transition layer with chemical bonding between the hard working layer and the soft buffer layer through the thermal penetration of 1,4-butanediol;meanwhile, the microspheres pore-forming and water chemical foaming technologies are used to form micro-porous and macro-porous structures in the hard layer and the soft layer respectively; the hard layer of the polishing pad has high hardness and wear resistance, and the soft layer has high resilience and hydrolysis resistance, which can realize efficient planarization and low defect rate polishing of the semiconductor wafer, and maintain stable removal rate and physical properties in long-term use.

Description

Technical Field

[0001] This invention relates to the field of semiconductor polishing technology, specifically to a polyurethane composite polishing pad with a gradient modulus structure and its preparation method. Background Technology

[0002] As semiconductor integrated circuit manufacturing processes enter the 14nm and below nodes, chemical mechanical polishing (CMP) is currently a key technology for achieving global planarization of wafer surfaces in integrated circuit manufacturing. As a core consumable in the CMP process, the surface microporous structure, hardness, hydrophilicity, and elastic recovery ability of the polishing pad directly determine the stability of the polishing rate and the flatness of the wafer surface. Among many polishing pad materials, polyurethane has become the mainstream choice due to its excellent wear resistance and the flexibility of adjusting the modulus through formulation. Its performance directly determines the processing quality.

[0003] Existing commercial polishing pads are mainly divided into single-layer pads and double-layer composite pads. Although single-layer hard pads have strong planarization capabilities, they lack cushioning and are prone to causing scratches on the wafer surface. Traditional double-layer composite pads use pressure-sensitive adhesive to bond the hard upper pad and the soft lower pad, which improves the cushioning performance, but has obvious defects.

[0004] Firstly, the interface bonding is fragile. Under high-speed rotation and strong shearing conditions during polishing, the physical bonding interface is prone to delamination due to fatigue or immersion in polishing fluid, leading to catastrophic wafer scrap. Secondly, it is difficult to balance mechanical properties and durability. In pursuit of high removal rates, the upper pad is usually designed with a high-hardness microporous structure, but the micropore walls are prone to breakage under high loads, resulting in poor wear resistance. At the same time, in pursuit of low defects, the lower pad is usually made of soft foam, but it is prone to creep and hydrolytic collapse under long-term pressure and acid and alkali corrosion. This causes the overall resilience and compressibility of the polishing pad to decrease nonlinearly over time, seriously affecting the stability of the polishing rate. Therefore, a gradient structure polyurethane composite polishing pad that balances mechanical properties and polishing durability is needed to meet the application range of electronic semiconductors.

[0005] To address this, a polyurethane composite polishing pad with a gradient modulus structure and its preparation method are proposed. Summary of the Invention

[0006] The purpose of this invention is to provide a polyurethane composite polishing pad with a gradient modulus structure and its preparation method. This invention utilizes the thermal permeation of 1,4-butanediol to construct a chemically bonded gradient transition layer between the hard working layer and the soft buffer layer; simultaneously, it employs microsphere pore-forming and hydrochemical foaming techniques to form microporous and macroporous structures in the hard and soft layers, respectively. This polishing pad exhibits high hardness and wear resistance in the hard layer and high resilience and hydrolysis resistance in the soft layer, enabling efficient planarization and low defect rate polishing of semiconductor wafers, while maintaining stable removal rates and physical properties during long-term use.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] This invention provides a method for preparing a polyurethane composite polishing pad with a gradient modulus structure, comprising the following steps:

[0009] Preheat the mold to 80°C, spray with fluorine-based release agent, add the hard layer mixture into it and let it flow out evenly. At the same time, rotate the mold to quickly spread the mixture to the bottom of the mold, maintaining a thickness of 1.2mm. Start timing at this point and enter the wire drawing monitoring stage to pour the soft layer.

[0010] The casting step utilizes the time window of the semi-gel state of the hard layer. After casting the hard layer, the operator must closely monitor the material surface, lightly touch the surface with a needle and lift it. When the surface loses its luster, feels sticky, and a fine filament can be drawn out when the needle is lifted, but it does not break and retract into droplets, the next step should be performed immediately. At this point, the hard layer is in a semi-gel state, possessing load-bearing capacity but with molecular chains not yet fully cross-linked and locked. If it is liquid and no filament is drawn when the needle is lifted, further waiting is required. If a skin has formed on the surface and no fingerprint is left when pressed, it indicates that it is too late and casting cannot continue. The optimal filament drawing time of this invention is 60-90 seconds after casting, at which time the viscosity of the hard layer is 35,000-45,000 cP.

[0011] The soft layer mixture is poured into the hard layer of the mold during the drawing stage at multiple points. Since the temperature of the soft material is 70°C and the viscosity is low at this time, it will quickly flow and cover the entire mold. The 1,4-butanediol in the liquid penetrates downward under the heat drive, locking the interface. About 20-30 seconds after pouring, the surface of the liquid turns white and begins to generate tiny bubbles. Then, within 60-90 seconds, the volume of the liquid begins to expand. At this time, the mold cover is closed to obtain the reaction mold. The reaction mold is heated to 90-100°C and kept at this temperature for 20-30 minutes to cure. The mold is opened, and the polishing pad blank is taken out by blowing air with an air gun or mechanical ejection. It is placed flat on a stainless steel plate and pushed into a precision oven for a three-stage curing process. First, the temperature is raised to 80-90°C for low-temperature diffusion for 2-4 hours. Then, the temperature is raised to 100-105°C for medium-temperature enhancement for 4-6 hours. Finally, the temperature is raised to 115-120°C for high-temperature setting for 12-16 hours to obtain the polyurethane polishing pad. A linear heating rate of 0.8℃ / min is set between each stage to avoid thermal shock.

[0012] The hard layer mixture is obtained by stepwise mixing of hard layer prepolymer, thermally expandable polymer microspheres, hydrophobic nano-vaporized silica, triethylenediamine and 3,3'-dichloro-4,4'-diaminodiphenylmethane;

[0013] The soft layer mixture is obtained by stirring and dispersing soft layer prepolymer, 3,5-dimethylthiotoluene diamine, 1,4-butanediol, deionized water and carbodiimide.

[0014] Preferably, the stepwise mixing includes the following steps: adding 1.8-2.2 parts of thermally expandable polymer microspheres to 10 parts of hard layer prepolymer, stirring at 50 rpm for 20 min to obtain a microsphere slurry; adding 90 parts of hard layer prepolymer to a mixing tank, controlling the temperature at 50-55℃, adding the microsphere slurry, and stirring at 100 rpm for 10 min; then adding 0.02-0.05 parts of triethylenediamine and 0.5-1 parts of hydrophobic nano-fumed silica, stirring at 1500 rpm for 10-15 min under a vacuum of -0.095 MPa to obtain a mixture; adding 18.2-19.5 parts of molten 3,3'-dichloro-4,4'-diaminodiphenylmethane to the mixture, increasing the stirring speed to 3000 rpm, and stirring for 30-45 s to obtain the hard layer mixture.

[0015] Preferably, the molten 3,3'-dichloro-4,4'-diaminodiphenylmethane is prepared by adding 3,3'-dichloro-4,4'-diaminodiphenylmethane particles into a melting tank. After the bottom portion melts at 100-105°C, stirring is started and the mixture is stirred at 100 rpm until it melts. The molten 3,3'-dichloro-4,4'-diaminodiphenylmethane liquid should be a clear amber color. If it turns dark brown or black, it must be discarded and cannot be used. This product should be melted as needed. During the preparation process, the molten 3,3'-dichloro-4,4'-diaminodiphenylmethane has a high temperature, and its addition will instantly raise the temperature of the mixture, and the reaction will begin immediately. If the stirring time is too long, it will cause explosive polymerization due to heat generation, while if it is too short, MOCA filaments will remain. Therefore, the stirring time needs to be strictly controlled.

[0016] Preferably, the preparation of the rigid layer prepolymer includes the following steps: 100 parts of polytetrahydrofuran ether diol with a molecular weight of 1000 and 33.5-35.5 parts of 2,4-toluene diisocyanate are placed in a vacuum drying oven and dried at 80°C for 2 hours; the polytetrahydrofuran ether diol is added to a reaction flask, and under nitrogen protection, 0.03-0.05 parts of benzoyl chloride and 0.05 parts of defoamer are added and stirred evenly; 2,4-toluene diisocyanate is slowly added dropwise, and the temperature is raised to 70-90°C for polymerization reaction for 3-4 hours. The NCO content is determined by di-n-butylamine titration to be stable at 5.5-6.2%, thus obtaining the rigid layer prepolymer, which is then cooled to 50-55°C for later use.

[0017] Preferably, the stirring and dispersing includes the following steps:

[0018] Add 5.5-6.5 parts of 3,5-dimethylthiotoluene diamine to a dry container, and add 1.5-2.5 parts of 1,4-butanediol and 0.15-0.25 parts of deionized water under stirring at 30°C and 100 rpm. Then add 0.8-1.2 parts of polyether-modified silicone oil foam stabilizer and 0.5-1.0 parts of carbodiimide. Stir and mix at 500 rpm for 5-10 minutes to obtain a mixed component. Add the mixed component to the soft layer prepolymer at 300°C. The mixture is stirred and dispersed at 0-4000 rpm for 10-15 seconds to obtain a soft layer mixture. Due to the high temperature of the soft layer and the presence of highly active, small-molecule BDO, the BDO preferentially penetrates into the surface of the hard layer under the influence of gravity and thermal diffusion, reacting with the residual NCO groups in the hard layer to form an interpenetrating network. Electron microscopy observation of cross-sectional morphology reveals an intermediate transition layer with a depth of 0.2-0.5 mm. Because the interface is chemically bonded and contains an interpenetrating network, its peel strength is far higher than that of adhesives. There is absolutely no risk of delamination during continuous CMP polishing lasting tens of hours. Simultaneously, the water in the soft layer reacts with NCO to release carbon dioxide gas, forming a uniform macroporous structure under the action of a foam stabilizer. The macroporous structure is evenly distributed, and due to the semi-gel state of the hard layer, the pores do not penetrate to the surface of the hard layer.

[0019] Preferably, the preparation of the soft layer prepolymer includes the following steps:

[0020] 100 parts of polyether polyol with a molecular weight of 2000-3000 and 30-33 parts of diphenylmethane diisocyanate were placed in a vacuum drying oven and dried at 80°C for 2 hours. The polyether polyol was a mixture of PPG-3000 and PTMEG-2000 in a 7:3 ratio. The polyether polyol was added to a reaction flask, and under nitrogen protection, diphenylmethane diisocyanate was added. The temperature was raised to 70-75°C and the polymerization reaction was carried out for 2-3 hours until the NCO content stabilized at 4.8-5.2%. The soft layer prepolymer was obtained and cooled to 70°C for later use.

[0021] Preferably, the soft layer uses an MDI / PPG system with a higher loss factor, which can effectively absorb grinding vibration; the hard layer uses TDI / PTMEG to ensure rigidity; the intermediate penetration layer ensures that energy transfer is not abrupt and reduces the vibration lines on the wafer surface; the temperature of the soft layer prepolymer needs to be higher than that of the hard layer to reduce viscosity and promote subsequent penetration; the soft layer mixture is poured using multi-point pouring or by using a flow guide plate to buffer the flow rate, so that the high-temperature soft material gently covers the surface of the hard material.

[0022] Preferably, the three-stage ripening process includes the following steps:

[0023] The temperature is raised to 80-90℃ for low-temperature diffusion for 2-4 hours; then raised to 100-105℃ for medium-temperature enhancement for 4-6 hours; finally, the temperature is raised to 115-120℃ for high-temperature setting for 12-16 hours to obtain the polyurethane polishing pad. A linear heating rate of 0.8℃ / min is set between each stage to avoid thermal shock.

[0024] The present invention also provides a method for preparing a polyurethane composite polishing pad with a gradient modulus structure. The raw materials for preparing the composite polishing pad include a hard layer prepolymer, a soft layer prepolymer, thermally expandable polymer microspheres, hydrophobic nano-fumed silica, and 1,4-butanediol.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0026] 1. This invention introduces hydrophobic nano-fumed silica into the hard layer prepolymer system, and utilizes its physical cross-linking points and reinforcing effect in the micropore walls to significantly improve the tear resistance of the pore walls. This design effectively avoids the brittle fragmentation of the micropore walls under diamond dressing. Compared with the prior art, which often leads to increased material brittleness and decreased wear resistance by simply pursuing high hardness, this invention ensures the long-term stability of the micro-texture on the surface of the polishing pad.

[0027] 2. This invention employs a wet-bonding and wet-casting process, utilizing the high diffusion coefficient and thermal driving effect of 1,4-butanediol to rapidly penetrate to the surface depth of the hard layer during the semi-gel window period. In addition, the active hydroxyl groups carried by BDO react in situ with the NCO groups remaining in the hard layer to construct a gradient transition layer of an interpenetrating polymer network structure. Compared with the problem of easy delamination at the adhesive interface of traditional double-layer polishing pads, this invention has high interlayer peel strength, avoiding the hidden danger of interface delamination.

[0028] 3. This invention introduces carbodiimide as an anti-hydrolysis agent into the soft layer formulation and uses a polyether-modified silicone oil foam stabilizing system to construct a closed-cell / semi-open-cell macroporous structure with extremely strong anti-hydrolysis ability. Carbodiimide can effectively capture carboxyl groups generated by hydrolysis, cut off the self-catalytic degradation chain, and prevent the soft layer from becoming spongy. Compared with existing soft pads that often suffer from decreased resilience due to unstable open-cell structure or water absorption, the polishing pad of this invention has stable compression and resilience after long-term polishing, ensuring continuous follow-up on wafer surface morphology and uniform pressure transmission.

[0029] 4. This invention achieves full life-cycle stability of polishing performance through the synergistic design of micropores in the hard layer and macropores in the soft layer, as well as the elimination of internal stress by a three-stage curing process; the wear-resistant micropores in the hard layer ensure continuous delivery of polishing fluid, and the anti-creep macropores in the soft layer provide stable support; it has excellent rate stability, significantly extends the service life of polishing pads, and reduces the cost of consumables and the frequency of downtime for pad replacement in semiconductor manufacturing.

[0030] 5. This invention constructs an asymmetric pore structure with upper micropores and lower macropores. The upper micropores are prepared by microsphere pre-dispersion process, with uniform pore size, which can effectively store grinding fluid and form hydrodynamic lubrication, improving planarization efficiency. The lower macropores are prepared by water chemical foaming, with larger pore size, providing high compressibility similar to air springs, which can effectively buffer mechanical vibration. This staggered pore size distribution ensures both the rigid cutting ability of the contact surface and provides overall flexible buffering, effectively balancing the contradiction between high removal rate and low surface defects. Attached Figure Description

[0031] Figure 1 The graph shows the changes in polishing efficiency of the polyurethane polishing pads prepared in Examples 1-4 and Comparative Examples 10-13 of this invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] The PTMEG-1000 described in this invention is polytetrahydrofuran ether diol with a molecular weight of 1000 g / mol and a hydroxyl value of 112 KOH / g; TDI-100 is 2,4-toluene diisocyanate; MOCA is 3,3'-dichloro-4,4'-diaminodiphenylmethane with CAS number 101-14-4; the thermally expandable polymer microspheres are commercially available products with an initial particle size of 10-40 μm, a foamed particle size of 30-50 μm, an initial expansion temperature of 90-100℃, and a maximum expansion temperature of 130-140℃; the hydrophobic nano-fumed silica is hydrophobic fumed silica treated with dimethyldichlorosilane, with a particle size of 10-30 nm and a specific surface area of ​​100- 200 m² / g; PPG-3000 is polypropylene glycol with a molecular weight of 3000 g / mol; PTMEG-2000 is polytetrahydrofuran ether glycol with a molecular weight of 2000 g / mol and a hydroxyl value of 56 KOH / g; MDI is commercially available diphenylmethane diisocyanate with an NCO content of 29%-33%; DMTDA is 3,5-dimethylthiotoluene diamine with an amine value of 530-540 mg KOH / g; BDO ​​is 1,4-butanediol; the carbodiimide used is a polymeric carbodiimide with CAS number 13515-40-7; triethylenediamine is a dipropylene glycol solution system with a TEDA content of 33%; the defoamer is polyether-modified silicone oil.

[0034] Please see Figure 1This invention provides a polyurethane composite polishing pad with a gradient modulus structure and its preparation method. The technical solution is as follows:

[0035] Example 1

[0036] 100 parts of polytetrahydrofuran ether diol with a molecular weight of 1000 and 34.5 parts of 2,4-toluene diisocyanate were placed in a vacuum drying oven and dried at 80°C for 2 hours. The polytetrahydrofuran ether diol was added to a reaction flask, and under nitrogen protection, 0.05 parts of benzoyl chloride and 0.05 parts of defoamer were added and stirred until homogeneous. 2,4-toluene diisocyanate was slowly added dropwise, and the temperature was raised to 80°C for polymerization reaction for 4 hours. The NCO content was determined by di-n-butylamine titration and found to be stable at 6.0%, thus obtaining a rigid layer prepolymer, which was then cooled to 50-55°C for later use.

[0037] 100 parts of polyether polyol and 32 parts of diphenylmethane diisocyanate were placed in a vacuum drying oven and dried at 80°C for 2 hours. The polyether polyol was a mixture of PPG-3000 and PTMEG-2000 in a 7:3 ratio. The polyether polyol was added to a reaction flask, and under nitrogen protection, diphenylmethane diisocyanate was added. The mixture was heated to 75°C and polymerized for 3 hours until the NCO content stabilized at 5.0%. The soft layer prepolymer was obtained and kept at 65-70°C for later use.

[0038] Two parts of thermally expandable polymer microspheres were added to ten parts of rigid layer prepolymer, and stirred at 50 rpm for 20 min to obtain a microsphere slurry. Ninety parts of the rigid layer prepolymer were added to a mixing tank, and the temperature was controlled at 50℃. The microsphere slurry was then added, and the mixture was stirred at 100 rpm for 10 min. Then, 0.03 parts of triethylenediamine and 0.8 parts of hydrophobic nano-fumed silica were added, and the mixture was stirred at 1500 rpm for 10 min under a vacuum of -0.095 MPa to obtain a mixed solution. 18.8 parts of molten 3,3'-dichloro-4,4'-diaminodiphenylmethane were added to the mixture, and the stirring speed was increased to 3000 rpm for 40 seconds to obtain a hard layer mixture. The mold was preheated to 80°C, and a fluorinated release agent was sprayed on it. The hard layer mixture was then added to it and allowed to flow out evenly. At the same time, the mold was rotated to quickly spread the mixture to the bottom of the mold, maintaining a thickness of 1.2 mm. At this point, the timing was started, and the soft layer was poured during the wire drawing monitoring stage.

[0039] Six parts of 3,5-dimethylthiotoluene diamine were added to a dry container. At 30°C and 100 rpm, 2 parts of 1,4-butanediol and 0.2 parts of deionized water were added, followed by 1 part of polyether-modified silicone oil foam stabilizer and 0.8 parts of carbodiimide. The mixture was stirred at 500 rpm for 8 minutes to obtain a mixed component. This mixed component was then added to the soft layer prepolymer and dispersed at 3000 rpm for 15 seconds to obtain a soft layer mixture. The soft layer mixture was poured into the hard layer of the mold during the fiber drawing stage. Since the soft material temperature was 70°C and the viscosity was low, it quickly flowed and covered the entire mold. The 1,4-butanediol in the liquid penetrated downwards under thermal drive, locking the interface. 20 seconds after pouring, the surface of the liquid turned white and began to generate tiny bubbles. Subsequently, the volume of the liquid began to expand. If precise control of density and thickness was required, the mold cover was then placed on top to obtain a reaction mold.

[0040] The reaction mold was heated to 90°C and cured for 30 minutes. The mold was then opened, and the polishing pad blank was removed using an air gun or mechanical ejection. It was placed flat on a stainless steel plate and pushed into a precision oven for a three-stage curing process: first, the temperature was raised to 80°C for low-temperature diffusion for 3 hours; then, the temperature was raised to 105°C for medium-temperature enhancement for 5 hours; finally, the temperature was raised to 120°C for high-temperature setting for 15 hours to obtain the polyurethane polishing pad. A linear heating rate of 0.8°C / min was set between each stage to avoid thermal shock.

[0041] Examples 2-4 follow the same preparation method and parameter conditions as Example 1, with differences shown in Table 1.

[0042] Table 1. Parameter variations in Examples 1-4

[0043] Comparative Example 1 is the same as Example 1, except that hydrophobic nano-vaporized silica is not added to the hard layer mixture, while the amounts of the other components remain unchanged.

[0044] Comparative Example 2 is the same as Example 1, except that 3,3'-dichloro-4,4'-diaminodiphenylmethane is not added to the hard layer mixture, while the amounts of the other components remain unchanged.

[0045] Comparative Example 3 is the same as Example 1, except that 3,5-dimethylthiotoluene diamine is not added to the soft layer mixture, while the amounts of the other components remain unchanged.

[0046] Comparative Example 4 is the same as Example 1, except that only PPG-3000 is used in the preparation of the soft layer prepolymer, and PTMEG-2000 is not used, while the amounts of other components remain unchanged.

[0047] Comparative Example 5 is the same as Example 1, except that only PTMEG-2000 is used in the preparation of the soft layer prepolymer, and PPG-3000 is not used, while the amounts of other components remain unchanged.

[0048] Comparative Example 6 is the same as Example 1, except that the polishing pad curing process does not use a three-stage curing process, but only uses a high temperature of 120°C for 15 hours.

[0049] Comparative Example 7 is the same as Example 1, except that the polishing pad curing process does not use a three-stage curing process, but only uses a high temperature of 100°C for 20 hours.

[0050] Experimental Example 1: Mechanical Property Testing

[0051] The mechanical properties and wear resistance of the polyurethane polishing pad were tested; the Shore D hardness was tested according to the national standard GB / T 531.1-2008. A type D hardness tester was used to measure five points at different locations on the sample surface, and the average value was taken. Hardness reflects the material's ability to resist indentation deformation and is positively correlated with planarization efficiency. Hard and soft layers were sampled separately for hardness testing. A Shore type D hardness tester was used, and the hard layer was removed before measuring on the surface of the hard layer. A Shore type A hardness tester was used, and the hard layer was removed before measuring on the solid part of the soft layer. Taber abrasion resistance was tested according to the national standard ASTM D4060 or the industry-standard test method. A Taber abrasion tester was used with an H-18 grinding wheel, a load of 1000g, and 1000 rotations. The mass loss before and after the test was measured in mg. The lower the wear, the better the abrasion resistance and the less wear during polishing pad dressing. Tensile strength was tested according to GB / T528-2009 standard, with a tensile speed of 500mm / min. The maximum force value at sample fracture was recorded, and the tensile strength was calculated. The test results are shown in Table 2.

[0052] Table 2 Performance Tests of Examples and Comparative Examples

[0053] Example Overall tensile strength / MPa Hardness of hard layer / Shore D Soft layer hardness / Shore A Tiber wear rate / mg Example 1 24.3 70.1 80.2 35.2 Example 2 25.1 72.3 80.1 28.5 Example 3 36.4 69.8 79.5 36.1 Example 4 23.8 69.2 78.4 35.8 Comparative Example 1 22.1 66.5 80.3 85.4 Comparative Example 2 3.2 / 80.2 >500 Comparative Example 3 4.5 70.2 32.6 35.5 Comparative Example 4 18.2 70.1 75.2 35.3 Comparative Example 5 20.5 70.2 88.4 35.1 Comparative Example 6 15.6 70.5 81.1 38.2 Comparative Example 7 19.3 62.4 72.3 62.1

[0054] According to the results in Table 2, the polyurethane polishing pads prepared in the comparative examples show significant differences in mechanical properties and wear resistance compared to those in the examples. In Comparative Example 1, the hydrophobic nano-fumed silica acts as a physical crosslinking point and reinforcing filler in the hard layer. Without it, the micropore walls lack skeletal support and are easily broken and detached under the shearing force of the grinding wheel, resulting in increased wear and a slight decrease in hardness. In Comparative Example 2, MOCA is the main chain extender of the hard layer prepolymer. Without MOCA, the prepolymer can only react with trace amounts of moisture in the air or a small amount of soft layer components that subsequently come into contact with it, and cannot form rigid urethane hard segment regions. This results in the hard layer being in a sticky, semi-fluid, uncured state, with a significant reduction in tensile strength and wear resistance. In Comparative Example 3, DMTDA was the main chain extender for the soft layer. Without DMTDA, the soft layer relied solely on a small amount of BDO and water for chain extension, resulting in extremely low crosslinking density. This prevented the formation of an effective macroporous framework, leading to very low hardness and a soft, mushy state. The overall tensile strength decreased significantly due to the loss of cohesion in the soft layer. Comparative Examples 4-5 showed that using a blend of PPG and PTMEG resulted in PTMEG having a regular structure, high strength, and better compatibility with the hard layer. Using only PPG resulted in excessively flexible molecular chains, poor strength, and slightly inferior interfacial compatibility with the hard layer, leading to a decrease in overall tensile strength. Conversely, PPG's lower viscosity facilitated water foaming and the formation of macropores, while using only PTMEG increased prepolymer viscosity, increased foaming resistance, and lower closed-cell ratio. Excessive temperature causes the soft layer hardness to spike, losing the high resilience and low hardness characteristics that a buffer layer should possess, thus impairing compressibility. Comparative results in Examples 6-7 show that low-temperature diffusion, medium-temperature strengthening, and high-temperature setting effectively reduce residual stress. Directly heating to 120℃ causes the reaction rate to increase exponentially. The BDO at the interface is instantly locked on the surface before it can penetrate deep into the hard layer, resulting in a very thin gradient layer, weak interfacial bonding, and a significant decrease in tensile strength. At the same time, the thermal shock caused by excessively rapid heating will generate internal stress. In addition, 100℃ is lower than the optimal post-curing temperature of the MOCA system, resulting in incomplete curing of the hard layer, low NCO conversion rate, and incomplete formation of hard segment micro-regions, manifested as lower hardness, poorer wear resistance, and a significant reduction in overall strength.

[0055] Comparative Example 8 is the same as Example 1, except that 1,4-butanediol is not added during the preparation of the soft layer mixture, while the amounts of the other components remain unchanged.

[0056] Comparative Example 9 is the same as Example 1, except that 1,4-butanediol is not added during the preparation of the soft layer mixture, but 2.6 parts of 1,6-hexanediol are added, and the amounts of the other components remain unchanged.

[0057] Comparative Example 10 is the same as Example 1, except that deionized water is not added to the soft layer mixture for foaming.

[0058] Experimental Example 2: Peel Strength Test

[0059] Interlayer adhesion was assessed to examine the bonding strength between the hard and soft layers. According to GB / T 2791-1995 "Test Method for Peel Strength of Adhesives - Flexible Materials to Flexible Materials", the polishing pad was cut into strips 30mm wide and 200mm long. A pre-peel of approximately 50mm was manually performed. The base fabric layer and foam layer were clamped in the upper and lower fixtures of a tensile testing machine, and a 180-degree peel test was conducted at a speed of 100mm / min. The average load during the peel process was recorded, and the peel strength was calculated. The overall compression rate and resilience of the polishing pad were tested according to JIS L 1096. The initial load was 300g, and the thickness D1 was tested. A main load of 1800g / cm² was applied and held for 60s, and the thickness D2 was tested. After unloading and allowing it to stand for 60s, the thickness D3 was tested. The resilience and compression rate were also tested. The test results are summarized in Table 3.

[0060] Table 3 Test Results of Examples and Comparative Examples

[0061] Example interlayer peel strength / N / 30 mm Compression rate / % Rebound rate / % Example 1 95.2 2.5 96.5 Example 2 88.4 2.3 95.8 Example 3 94.5 2.6 96.2 Example 4 94.6 2.8 97.4 Comparative Example 3 <5.0 (Unable to test) <20 (collapse) <10 Comparative Example 4 72.3 3.5 91.2 Comparative Example 5 85.6 0.8 98.1 Comparative Example 8 12.4 2.4 96.1 Comparative Example 9 25.6 2.8 96.3 Comparative Example 10 90.1 0.4 98.5

[0062] As shown in Table 3, the polyurethane polishing pads prepared in the comparative examples exhibit significant differences in peel strength and resilience compared to those in the examples. In Comparative Example 3, DMTDA is the main chain extender for the soft layer, providing necessary physical crosslinking points and hard segment support. Its absence results in extremely low crosslinking density in the soft layer, failing to lock in the pore structure generated by water foaming, leading to a muddy, shapeless soft layer with extremely high compression ratio and very low resilience. Furthermore, due to the near-zero cohesive strength of the soft layer, peel strength cannot be tested. The results of Comparative Examples 4-5 show that PP... G molecular chains have good flexibility but low strength, and their compatibility with rigid layers is not as good as PTMEG. A full PPG formulation results in a softer soft layer, increased compressibility, and decreased interfacial compatibility, leading to reduced peel strength. Because PTMEG has a higher viscosity than PPG, the soft layer prepolymer has a higher viscosity, increasing water foaming resistance, resulting in excessively high closed-cell ratio and smaller pore size. This leads to increased hardness and extremely low compressibility in the soft layer, losing its energy-absorbing properties as a buffer layer. Although it has good resilience, it cannot adapt to the wafer's microstructure. Comparative examples 8-9 show that B... DO is the core penetrant for constructing gradient interfaces. Without BDO, the soft layer cannot microscopically penetrate to the surface of the hard semi-gel layer; the two layers rely solely on extremely weak physical adhesion, resulting in a sharp drop in peel strength and easy delamination. BDO preferentially penetrates to a depth of 0.2-0.5 mm into the surface of the hard layer under the influence of gravity and thermal diffusion, reacting with the residual NCO groups in the hard layer to form an interpenetrating network. Conversely, using 1,6-hexanediol, due to its large molecular size and high diffusion resistance, results in poor diffusion during the wet-to-wet contact window. Most of the 6-hexanediol remained on the soft layer side and could not penetrate into the hard layer as deeply as BDO. Therefore, the interpenetrating network at the interface was very shallow, resulting in a final interlayer bonding strength that was better than Comparative Example 8, but far worse than Example 1. The results of Comparative Example 10 showed that water was a chemical foaming agent responsible for generating macroporous structures. Without the addition of deionized water, the soft layer became a dense solid polyurethane. Although it had high strength, it lost the compressible space provided by the pores, resulting in an extremely low compressibility. It could not be used as a CMP buffer layer and lost its planarization adjustment ability.

[0063] Comparative Example 11 is the same as Example 1, except that thermally expandable polymer microspheres are not added to the hard layer mixture.

[0064] Comparative Example 12 is the same as Example 1, except that the thermally expandable polymer microspheres are directly added to the hard mixture without undergoing a microsphere slurry premixing process.

[0065] Comparative Example 13 is the same as Example 1, except that carbodiimide is not added to the soft layer mixture.

[0066] Experiment Example 3 Polishing Durability Test

[0067] The polyurethane polishing pads prepared in Examples 1-4 and Comparative Examples 10-13 were subjected to long-term durability tests. The tests were modified according to the industry-standard SEMI standard and GB / T 35474-2017 "Test Method for Polishing Removal Rate of Hard and Brittle Materials". Marathon continuous polishing tests were conducted. A 300mm (12-inch) chemical mechanical polishing machine was used. The wafer was a 12-inch TEOS (tetraethyl orthosilicate) silicon dioxide cover sheet. The polishing pads were used in conjunction with commercially available fumed silica alkaline polishing slurry and a 3M A165 diamond dressing disk. In-situ dressing mode was used.

[0068] After installing the new pad, polish with a dummy wafer for 20 minutes to stabilize the surface. Continuous polishing: set the pressure to 3.0 psi, the rotation speed to 90 / 93 rpm (head / pan), and the polishing slurry flow rate to 250 ml / min. Perform continuous polishing for 20 hours; Polishing rate decay rate (RR, %): test one monitoring wafer for 1 minute every 1 hour, and measure the change in film thickness before and after polishing using an optical film thickness gauge to calculate the removal rate; calculation formula: RR1 is the initial value, and RR0 is the final value. The lower the value, the better the polishing stability. A contact thickness gauge was used to measure the thickness change at a specified location on the polishing pad before and after the test, and the average thickness loss per hour was calculated. The average wear rate reflects the physical wear rate of the polishing pad. Simultaneously, longitudinal cross-sectional samples of the polishing pad were prepared using the cryogenic fracture method. After gold sputtering, they were observed and tested using a scanning electron microscope. Fifty micropores in the hard layer region were randomly selected, their diameters were measured, and the average value was calculated. Additionally, 30 macropores in the soft layer region were randomly selected, their diameters were measured, and the average value was calculated. The test results are shown in Table 5. The polishing efficiency changes of the polyurethane polishing pads prepared in Examples 1-4 and Comparative Examples 10-13 are shown in Table 5. Figure 1 As shown.

[0069] Table 5 Polishing durability tests of the examples and comparative examples

[0070] Example Average pore size of hard layer / μm Average pore size of soft layer / μm Rate decay rate / % Average wear rate / μm / h Example 1 35.2 185.1 3.7 32.5 Example 2 32.8 182.4 3.8 28.1 Example 3 35.5 186.7 4.1 33.2 Example 4 38.1 190.3 4.3 31.8 Comparative Example 10 35.3 / 6.6 24.5 Comparative Example 11 / 184.5 34.6 8.2 Comparative Example 12 48.6 185.3 19.2 55.4 Comparative Example 13 35.4 184.8 27.5 34.1

[0071] As shown in Table 5, the polyurethane polishing pads prepared in the comparative example have significantly different polishing durability compared to those in the examples. The pore size of the hard layer is stable in the range of 30-40 μm, while the pore size of the soft layer is in the range of 180-200 μm. The microsphere pre-dispersion process ensures the uniformity of the micropores in the hard layer, and the water foaming and foam stabilizer system constructs a stable macroporous structure in the soft layer. At the same time, the reinforcing effect of nano-silica keeps the wear rate within the ideal range, ensuring surface renewal without excessively consuming the polishing pad's lifespan. In Comparative Example 10, without the addition of deionized water, the soft layer could not form a macroporous structure, becoming a dense solid polyurethane. Therefore, it lost the compressible space provided by the pores, resulting in an extremely low overall compressibility of the polishing pad. This prevented the polishing pad from adhering to the micro-ripples on the wafer surface, significantly reducing the effective contact area, and thus resulting in a significantly lower removal rate than in the examples. In Comparative Example 11, without the addition of thermally expandable polymer microspheres, the hard layer surface lacked a microporous structure. Micropores play a crucial role in storing and transporting polishing slurry in CMP. The absence of micropores caused the polishing slurry to form a water film and slide between the pad and the wafer, failing to generate effective mechanical friction, leading to an extremely low removal rate and easy surface glazing, resulting in rapid performance degradation. The results of Comparative Example 12 show that omitting the premixing process of the microsphere slurry resulted in a denser microsphere layer. With extremely low viscosity, the microspheres are prone to agglomeration when directly added to high-viscosity prepolymers. After solidification, the agglomerated microspheres form huge voids, resulting in uneven pore size. During polishing, the diamond dresser easily tears the agglomerated areas, causing a surge in wear rate and extremely unstable surface conditions, leading to significant fluctuations and attenuation in polishing rate. In Comparative Example 13, without the addition of carbodiimide, the soft layer uses polyester / polyether polyol and is foamed with water. Under long-term immersion in acidic or alkaline polishing fluids and water, the macropore walls are prone to hydrolysis and chain breakage. Carbodiimide, as an anti-hydrolysis agent, can capture the carboxyl groups generated by hydrolysis. After its absence, the soft layer structure gradually softens and collapses, causing the overall resilience of the polishing pad to drop significantly in the later stages of the test, making it unable to maintain stable downward pressure transmission, resulting in a sharp drop in the endpoint rate.

[0072] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for preparing a polyurethane composite polishing pad with a gradient modulus structure, the composite polishing pad being used for chemical mechanical polishing of semiconductor wafers, characterized in that, Includes the following steps: Preheat the mold, spray with fluorine-based release agent, inject the hard layer mixture into the mold and level it, covering the bottom of the mold to obtain the mold hard layer; monitor during the wire drawing period; pour the soft layer mixture into the mold hard layer at multiple points to obtain the reaction mold; The reaction mold is solidified, a polishing pad blank is taken, and a polyurethane polishing pad is obtained through a three-stage curing process; the hard layer mixture is obtained by stepwise mixing of hard layer prepolymer, thermally expandable polymer microspheres, hydrophobic nano-fumed silica, triethylenediamine, and 3,3'-dichloro-4,4'-diaminodiphenylmethane; the soft layer mixture is obtained by stirring and dispersing soft layer prepolymer, 3,5-dimethylthiotoluenediamine, 1,4-butanediol, deionized water, and carbodiimide. The stepwise mixing includes the following steps: adding the thermally expandable polymer microspheres to the rigid layer prepolymer and stirring to obtain a microsphere slurry; adding the rigid layer prepolymer to a mixing tank, adding the microsphere slurry, and stirring to mix; then adding the triethylenediamine and the hydrophobic nano-fumed silica and stirring to obtain a mixture; adding molten 3,3'-dichloro-4,4'-diaminodiphenylmethane to the mixture and stirring to obtain the rigid layer mixture; The stirring and dispersion includes the following steps: adding the 3,5-dimethylthiotoluene diamine to a dry container, stirring and adding the 1,4-butanediol and the deionized water; then adding the defoamer and the carbodiimide, stirring and mixing to obtain a mixed component; adding the mixed component to the soft layer prepolymer, stirring and dispersing to obtain the soft layer mixture; The three-stage curing process includes the following steps: curing the polishing pad blank at 80-90℃ for 2-4 hours; then heating it to 100-105℃ for 4-6 hours; and finally heating it to 115-120℃ for high-temperature shaping for 12-16 hours to obtain the polyurethane polishing pad.

2. The method for preparing a polyurethane composite polishing pad with a gradient modulus structure according to claim 1, characterized in that, The preparation of the rigid layer prepolymer includes the following steps: vacuum drying of polytetrahydrofuran ether diol and 2,4-toluene diisocyanate; adding them to a reaction flask, under nitrogen protection, adding benzoyl chloride and defoamer and stirring evenly; adding 2,4-toluene diisocyanate dropwise, and polymerizing to obtain the rigid layer prepolymer.

3. The method for preparing a polyurethane composite polishing pad with a gradient modulus structure according to claim 1, characterized in that, The preparation of the soft layer prepolymer includes the following steps: drying the polyether polyol and diphenylmethane diisocyanate; then adding them to a reaction flask, under nitrogen protection, adding diphenylmethane diisocyanate, and polymerizing to obtain the soft layer prepolymer.

4. A polyurethane composite polishing pad with a gradient modulus structure, characterized in that, The composite polishing pad is obtained by the preparation method described in any one of claims 1-3; the raw materials for preparing the composite polishing pad include a hard layer prepolymer, a soft layer prepolymer, thermally expandable polymer microspheres, hydrophobic nano-fumed silica, and 1,4-butanediol; the initial chemical CMP polishing rate of the composite polishing pad is 6185-6315 Å / min, and the final chemical CMP polishing rate is 5920-6098 Å / min.

Images