Preparation method of graphite for large semiconductor

Abstract

The invention discloses a preparation method of graphite for a large semiconductor, and belongs to the technical field of graphite for semiconductors. The method comprises the steps of raw material pretreatment, forming process, graphitization, purification and post-treatment, wherein needle coke and a carbon-based functional material are mixed, and a carbon-based dispersing agent is added under the action of a gradient magnetic field for dispersion to obtain a pretreated raw material; filling the mixture into a carbon-based mold, carrying out cold isostatic pressing molding through cooperation of gradient pressure and a gradient magnetic field, and absorbing stress through a carbon-based buffer material to obtain a molded green body; placing the green body in a large graphitization furnace, regulating and controlling a magnetic field in sections, introducing mixed gas, strengthening the reaction through a carbon-based catalytic material, and prolonging the reaction time of a core part by utilizing gas backflow to complete graphitization and purification; and finally, carrying out plasma surface activation, boron nitride coating deposition and annealing treatment to obtain a graphite product for a large semiconductor. The problems that the purity of a large green body is out of control, the impurity removal gradient difference is large and the stress is large can be effectively solved, and the product performance is excellent.

Description

Technical Field

[0001] This invention belongs to the field of graphite technology for semiconductors, specifically a method for preparing large-scale semiconductor graphite. Background Technology

[0002] Graphite materials for semiconductors are crucial foundational materials for core processes such as chip manufacturing and single crystal growth. They must meet stringent requirements including ultra-high purity (ash content ≤10ppm), excellent mechanical properties (bending strength ≥80MPa), and extremely low performance fluctuations (radial resistivity difference ≤5%). Their quality directly impacts the yield and reliability of semiconductor devices. As the semiconductor industry moves towards larger wafer sizes (e.g., 12 inches and above), the supporting graphite thermal components (e.g., single crystal furnace crucibles and heaters with diameters ≥1m) are also evolving towards larger sizes. However, existing fabrication technologies face significant bottlenecks in the production of large-scale preforms, specifically as follows:

[0003] Firstly, the presence of metal-based functional materials leads to uncontrolled purity: Existing technologies often employ metal-containing magnetic response materials (such as Fe and Co-catalyzed carbon nanotubes) and chemical additives (such as sulfur-containing dispersants and chlorine-containing binders) to enhance multi-field coupling effects. These materials are difficult to completely remove from large green bodies due to size effects: residual metal impurities result in product ash content generally ≥50ppm, far exceeding industry standards; sulfur and chlorine residues generated from the high-temperature decomposition of chemical additives further contaminate the materials and easily induce localized agglomeration during dispersion, exacerbating the compositional unevenness of large green bodies.

[0004] Secondly, there are gradient differences in the removal of impurities from large billets: conventional gas purification processes rely on surface diffusion, while the mass transfer path difference between the core and the surface of large billets can be more than 500 mm, and the core temperature is 200-300℃ lower than the surface temperature, resulting in a significant decrease in impurity removal efficiency.

[0005] Third, the size effect causes serious stress problems: During the forming and heat treatment process, large blanks are prone to significant gradient stress due to their large radial dimensions (≥1m). The pressure transmission delay in the forming stage leads to a stress difference of more than 200MPa between the core and the surface. The temperature gradient in the graphitization stage further aggravates the internal stress, which significantly increases the cracking rate of the blank.

[0006] Therefore, how to provide a method for preparing large-scale semiconductor graphite that can effectively overcome the problems of uncontrolled purity, large differences in impurity removal gradients, and high internal and external stresses in the product has become a key issue that urgently needs to be addressed. Summary of the Invention

[0007] The purpose of this invention is to provide a method for preparing graphite for large semiconductors, in order to solve the problems of uncontrolled purity, large differences in impurity removal gradients, and severe stress caused by size effects in existing graphite preforms for large semiconductors, thereby obtaining large semiconductor graphite products that meet the requirements of ultra-high purity, excellent mechanical properties, and low performance fluctuations.

[0008] The objective of this invention is achieved through the following technical solution:

[0009] A method for preparing graphite for large-scale semiconductors includes the following steps:

[0010] S1. Raw material pretreatment: Needle coke is mixed with carbon-based functional materials, and a carbon-based dispersant is added under the action of a gradient magnetic field. Dispersion is achieved through stirring and shearing to obtain pretreated raw materials. In this step, the gradient magnetic field induces the magnetic response carbon nanotubes in the carbon-based functional materials to oriented. The fullerene coating on its surface and the carbon-based dispersant form a "core-shell" synergistic dispersion system. The fullerene coating acts as the "core" to provide magnetic field response sites, and the fullerene-derived groups of the dispersant act as the "shell" to enhance steric hindrance. This dual action solves the agglomeration problem caused by the large amount and high mixing difficulty of raw materials in large blanks (diameter ≥1m). At the same time, it lays the foundation for the dispersion consistency in the subsequent molding stage and achieves macroscopic uniform dispersion of needle coke and carbon-based functional materials.

[0011] S2. Molding Process: Pretreated raw materials are filled into a carbon-based mold, and cold isostatic pressing is performed by applying gradient pressure and coordinating with a gradient magnetic field. Stress is absorbed by a carbon-based buffer material, resulting in a molded green body. In this step, the gradient pressure and gradient magnetic field work synergistically. The carbon nanotube-graphene composite framework of the carbon-based mold provides radial support, and the carbon-based buffer material absorbs stress through interlayer sliding, forming a synergistic stress control system of "support-buffer-constraint" with the outer constraint structure of the mold. The porous structure of the all-carbon-based framework guides the flow of raw materials, preventing local voids in large green bodies due to the "arching effect." Real-time monitoring by the carbon-based piezoelectric sensing layer provides precise feedback for stress control.

[0012] S3. Graphitization and Purification: The formed green body is placed in a large graphitization furnace, the magnetic field strength is controlled in stages, a mixed gas is introduced, and the reaction is enhanced by carbon-based catalytic materials. The gas reflux is used to extend the core reaction time, thus completing graphitization and purification to obtain a graphitized green body. In this step, the staged magnetic field control promotes the orderly arrangement of carbon atoms, the carbon-based catalytic materials guide the diffusion of the mixed gas, and the gas reflux mechanism extends the core reaction time. The nitrogen atoms of the nitrogen-doped graphene in the carbon-based catalytic materials not only serve as catalytic active sites to accelerate the reaction between impurities and gas, but also provide a covalent bond anchoring basis for the subsequent coating, forming a synergistic effect of "catalytic purification - anchoring preparation", which solves the problems of uneven graphitization and incomplete impurity removal in the core and surface of the large green body.

[0013] S4. Post-processing: First, the graphitized preform is subjected to plasma surface activation, followed by deposition of a boron nitride coating. Annealing is then performed using a carbon-based elastic component to apply pressure and control the cooling rate, resulting in a large-scale graphite product for semiconductors. In this step, plasma activation increases the surface energy of graphite, facilitating uniform deposition of the boron nitride coating. Nitrogen atoms in the nitrogen-doped graphene remaining in the carbon-based catalytic material form covalent bonds with boron atoms in the boron nitride coating, enhancing the adhesion between the coating and the substrate. During annealing, the carbon-based elastic component (made of the same material as the S2 buffer material) applies pressure, which, combined with the gradient matching of the thermal conductivity of the carbon-based materials throughout the process, forms a "stress relief-thermal matching" synergistic system, preventing cracking of the large preform due to differences in thermal expansion.

[0014] This application, through the above-described process, can effectively overcome the technical problems of uncontrolled purity, large differences in impurity removal gradients, and high internal and external stresses in the existing technology.

[0015] As some possible embodiments of this application, the needle coke has a purity of ≥99.99% and a particle size distribution of 5-30 μm.

[0016] As some possible implementations of this application, in step S1, the carbon-based functional material includes magnetically responsive carbon nanotubes and carbon-based quantum dots, with a mass ratio of 1:3 to 1:7; the magnetically responsive carbon nanotubes have a diameter of 10-30 nm, a length of 100-300 nm, and are coated with a fullerene layer, with a saturation magnetization ≥5 emu / g.

[0017] Carbon-based functional materials possess both magnetic field responsiveness and dispersibility: Magnetic-responsive carbon nanotubes coated with a fullerene layer enhance magnetic field response while preventing metal residue; carbon-based quantum dots fill the gaps between needle-like particles, resulting in stronger interfacial bonding of the raw materials. Furthermore, the fullerene layer can work in conjunction with carbon-based dispersants to enhance dispersion. Magnetic-responsive carbon nanotubes can align themselves under a magnetic field, improving the material's thermal conductivity; the filling of gaps by carbon-based quantum dots further strengthens interfacial bonding; and the all-carbon-based material system ensures functional requirements while avoiding the introduction of metal impurities, thus balancing functionality and purity.

[0018] As some possible implementations of this application, the gradient magnetic field is 0.05-0.5T in the raw material pretreatment stage and 0.1-1.0T in the molding stage.

[0019] Gradient magnetic fields are crucial for the directional alignment of carbon-based functional materials. The magnetic field strength during pretreatment and molding stages must be matched to the material state. If the magnetic field is too strong during pretreatment, carbon nanotubes are prone to agglomeration; if the magnetic field is insufficient during molding, the aligned structure cannot be fixed. Large preforms are large in size, and the magnetic field distribution is prone to unevenness. If the strength is not appropriate, the functional materials will be disordered, affecting the uniformity of product performance. The magnetic field in the pretreatment stage can initially disperse the functional materials, while the magnetic field in the molding stage can fix the aligned structure. The combination of the two can ensure a uniform distribution of functional materials in large preforms and reduce performance fluctuations caused by large size.

[0020] As one possible implementation of this application, in step S1, the carbon-based dispersant is polyfullerene glycol ether, and its addition amount is 0.005-0.02wt%.

[0021] In the raw materials for large preforms, carbon-based functional materials are prone to agglomeration due to their large specific surface area and strong intermolecular forces, requiring carbon-based dispersants to reduce surface energy. Traditional chemical additives contain elements such as sulfur and chlorine, which leave impurities after high-temperature decomposition, contaminating the product. Carbon-based dispersants, however, interact with the fullerene layer on the surface of magnetically responsive carbon nanotubes, enhancing dispersion without introducing impurities. The carbon-based dispersant forms a coating layer with the fullerene layer, preventing agglomeration.

[0022] As some possible implementations of this application, in step S2, the pressure range of gradient pressurization is 50-600MPa, and the pressure is applied in a stepped manner: 50-200MPa for the surface layer, 200-400MPa for the middle layer, and 400-600MPa for the core.

[0023] Large billets, with radial dimensions exceeding 1 meter, experience significant pressure differences between the surface and core. Applying high pressure all at once results in an overly dense surface layer, while the core, due to slower pressure transmission, lacks sufficient density, leading to radial stress differences and cracking. Gradient pressurization, by gradually increasing pressure, allows the density of the surface and core to become more uniform, adapting to the size characteristics of large billets. This results in more uniform density between the surface and core, reduces the radial stress difference, lowers the likelihood of cracking during molding, and creates a more intact billet structure.

[0024] As one possible implementation method of this application, the specific steps of step S2 are as follows:

[0025] S21. Mold preparation: A carbon-based mold with an embedded carbon nanotube-graphene composite framework is used. The porosity of the framework is distributed radially in a gradient of 20% to 40%, and carbon quantum dots are loaded on the surface of the framework. A carbon-based piezoelectric sensing layer is embedded in the inner wall of the mold, and a Ni-Ti / carbon composite shape memory alloy ring is wrapped on the outer layer.

[0026] S22. Raw material filling: Fill the pretreated raw material into the mold prepared in step S21;

[0027] S23. Coupling treatment: Apply pressure in stages at a set rate, successively reaching 50-200MPa, 200-400MPa, and 400-600MPa; while applying pressure, simultaneously apply axial rotating magnetic fields of 0.1-0.2T, 0.3-0.7T, and 0.8-1.0T, with a magnetic field rotation frequency of 1-5Hz; when the pressure rises to the 200-400MPa range, introduce 20-50kHz focused ultrasonic waves, focusing on the core region of the billet;

[0028] During the coupling process, the carbon-based piezoelectric sensing layer monitors the internal stress of the blank. When the local stress exceeds the set value, the carbon-based buffer material (expanded graphite / graphyne composite layer) undergoes reversible deformation, and at the same time, the outer carbon-based shape memory alloy ring of the mold undergoes radial shrinkage.

[0029] S24. Removal of blank: After processing is completed, the shaped blank is obtained.

[0030] When forming large billets, the mold structure and process parameters must be well-matched: the carbon-based mold skeleton must provide radial support to prevent material flow from causing shape deformation; the stress monitoring and buffering system must be able to respond promptly to local stress concentrations to avoid cracking; and ultrasonic assistance can solve the problem of core material agglomeration, making the dispersion more uniform. Traditional molds lack suitable gradient structures and dynamic control capabilities for large billets, easily leading to uneven material accumulation and stress concentration. The carbon-based skeleton can guide the flow of raw materials and reduce the "arching effect"; the stress monitoring and buffering system can alleviate local stress; and ultrasonic assistance improves the dispersion of core material, thus improving the overall forming quality and structural uniformity of large billets.

[0031] As one possible implementation of this application, in step S3, the mixed gas is a mixture of chlorine trifluoride and carbon tetrafluoride with a volume ratio of 1:2 to 1:5, and is introduced at a temperature of 2000-2800°C.

[0032] There is a temperature difference between the core and surface of large billets, with the core typically 200-300°C lower than the surface, resulting in different impurity removal efficiencies. Traditional gas purification processes do not adjust parameters according to the temperature gradient of large billets, leading to incomplete impurity removal from the core. It is necessary to adjust the ratio of the mixed gas and the inlet temperature to enhance the reactivity of the core and improve the impurity removal rate.

[0033] As one possible implementation of this application, in step S3, the carbon-based catalytic material is a porous graphite medium with 0.3-0.8 wt% nitrogen-doped graphene loaded on its surface.

[0034] Nitrogen atoms in nitrogen-doped graphene can serve as catalytic active sites and also provide a bonding basis for subsequent boron nitride coatings, thus enhancing the adhesion of the coatings.

[0035] As some possible implementations of this application, in step S4, the surface activation is performed using CH4 plasma treatment with a power of 200-500W for 3-10 minutes; during the deposition of the boron nitride coating, a carbon quantum dot sublayer with a thickness of 0.05-0.2nm is inserted every 30-70 cycles.

[0036] Large preforms have a large surface area, making coatings prone to uneven thickness and insufficient adhesion due to size limitations. Plasma activation increases the energy of the graphite surface, making the coating easier to wet and adhere; the carbon quantum dot sublayer fills interfacial defects, mitigating the difference in thermal expansion between the coating and the substrate, and preventing traditional coatings from detaching during thermal cycling. Plasma activation allows for more uniform deposition of boron nitride coatings; the carbon quantum dot sublayer enhances the interfacial bonding between the coating and the substrate, improving the stability of the coating on large preform surfaces and preventing coating detachment during thermal cycling.

[0037] Compared with the prior art, the beneficial effects of the present invention are:

[0038] The method for preparing large-scale semiconductor graphite provided in this application can effectively overcome the technical problems of uncontrolled purity, large differences in impurity removal gradients, and high internal and external stresses in the existing technology, as detailed below:

[0039] ① Performance uniformity of graphite products for large semiconductors: Through the synergistic effect of each step, small radial resistivity differences and uniform density distribution are achieved, meeting the semiconductor industry's requirement for "low thermal conductivity fluctuations" in thermal field components.

[0040] ② Improve product mechanical reliability: The synergistic effect of stress regulation and coating anchoring enables the product to withstand repeated thermal shocks during single crystal growth, reducing the risk of cracking.

[0041] ③ Meets ultra-high purity requirements: The all-carbon-based material system avoids the introduction of metal impurities. Combined with a synergistic purification mechanism, it ensures that the impurity content of the product is extremely low, avoiding device failure caused by impurity diffusion during semiconductor material preparation. Detailed Implementation

[0042] Example 1

[0043] S1: Raw material pretreatment.

[0044] S11. Weigh 100 kg of needle coke with a purity of 99.99% and a particle size distribution of 5-30 μm, place it in a drying oven and dry it at 120℃ for 2 h to obtain dried needle coke.

[0045] Another method involves taking 1 kg of magnetically responsive carbon nanotubes (diameter 10-30 nm, length 100-300 nm, surface coated with a fullerene layer, saturation magnetization 5 emu / g) and 4 kg of carbon-based quantum dots, mixing and grinding them in an agate mortar for 10 min until homogeneous to obtain carbon-based functional materials. The magnetically responsive carbon nanotubes are prepared as follows: They are added to a 0.5 wt% fullerene-toluene solution (carbon nanotube to solution mass ratio 1:20), ultrasonically dispersed for 60 min (power 300 W), allowed to stand for 2 h, and then centrifuged (speed 8000 r / min, time 10 min). The product is collected and vacuum dried at 80℃ for 4 h to obtain magnetically responsive carbon nanotubes with a fullerene layer on the surface.

[0046] S12. The dried needle coke and carbon-based functional materials were added together to the mixing jar of a double planetary mixer and stirred at 500 r / min for 5 min. Then, 0.01 wt% of polyfullerene glycol ether (based on the mass of the needle coke) was added to the mixing jar. The magnetic field strength was then set to 0.3 T, with the magnetic field direction increasing linearly from the container wall to the center along the stirring axis. At the same time, the stirring speed was adjusted to 1500 r / min, and stirring and shearing treatment was performed for 30 min to obtain the pretreated raw material.

[0047] S2: Molding process.

[0048] S21. Mold Preparation: A carbon-based mold with an inner diameter of 1000mm and a height of 1000mm is used. Its structure from the inside out is as follows:

[0049] Innermost layer: Carbon-based piezoelectric sensing layer. The carbon-based piezoelectric sensing layer is a flexible film (5mm thick) composed of piezoelectric graphene and carbon nanotubes. It is continuously laid in a ring along the inner wall of the mold. Stress acquisition points are evenly distributed in the film at intervals of 50-100mm. It can monitor the stress changes at different radial positions during the blank forming process in real time (detection range 0-1000MPa, sampling frequency 1kHz). When the local stress exceeds the set threshold, it directly triggers the reversible deformation of the carbon-based buffer material and the radial shrinkage of the shape memory alloy ring.

[0050] Intermediate layer: Expanded graphite / graphyne composite buffer material (20mm thick), completely laid on the outside of the carbon-based piezoelectric sensing layer, fully covering the mold radially;

[0051] Framework layer: Carbon nanotube-graphene composite framework (50mm thick), located on the outside of the buffer material, with its porosity distributed radially in a gradient of 20%→30%→40%, and the framework surface loaded with 0.5wt% carbon quantum dots.

[0052] Outermost layer: A Ni-Ti / carbon composite shape memory alloy ring with a diameter of 1100mm is fitted, and its phase transformation temperature is 60℃.

[0053] S22. Raw material filling: The pretreated raw material obtained in step S1 is filled into the mold prepared in S21 in 5 portions. After each filling, the mold is vibrated for 3 minutes using a 20kHz ultrasonic vibration device. After the final filling is completed, the filling density of the raw material is measured to be 1.2 ± 0.02 g / cm³. 3 .

[0054] S23. Coupling Process

[0055] Start the pressure system of the cold isostatic pressing equipment. The initial pressure is 0 MPa. Increase the pressure to 50 MPa at a rate of 5 MPa / min. After reaching the pressure, maintain the pressure for 10 minutes.

[0056] After the pressure holding period, the pressure is increased to 250 MPa at a rate of 10 MPa / min and held for 15 minutes. During this process, the axial rotating magnetic field generator is simultaneously activated, with the magnetic field strength set to 0.5 T and the magnetic field rotation frequency set to 3 Hz.

[0057] When the pressure rises to 250MPa and is held for 5 minutes, a 35kHz focused ultrasonic wave (500W power) is introduced, with the focal point located on the central axis of the billet, 400mm from the bottom, and continues to act until the pressure rises to the next stage.

[0058] Next, the pressure was increased to 600 MPa at a rate of 8 MPa / min and held for 20 minutes. At the same time, the axial rotating magnetic field strength was adjusted to 0.9 T and the frequency was maintained at 3 Hz.

[0059] Throughout the pressurization process, the carbon-based piezoelectric sensing layer monitors the internal stress of the billet in real time. When the local stress is ≥300MPa, the expanded graphite / graphyne composite buffer material undergoes 3% reversible deformation, while the Ni-Ti / carbon composite shape memory alloy ring on the outer layer of the mold undergoes 0.5±0.1mm radial shrinkage (monitored in real time by a laser diameter gauge).

[0060] After holding the pressure at 600MPa for 20 minutes, the pressure is slowly released to 0MPa at a rate of 5MPa / min. During this period, the internal stress of the billet is continuously monitored until the stress stabilizes below 50MPa, at which point the coupling process is complete.

[0061] S24. Removal of the blank: After the pressure is released and the stress is stable, first remove the outermost Ni-Ti / carbon composite shape memory alloy ring of the mold, then remove the carbon nanotube-graphene composite skeleton, expanded graphite / graphyne composite buffer material and carbon-based piezoelectric sensing layer in sequence, and remove the formed blank.

[0062] S3. Graphitization and purification:

[0063] The shaped blank obtained in step S2 is placed vertically into the center of the furnace cavity of a large graphitization furnace (1500mm inner diameter), with a uniform gap maintained between the blank and the furnace wall.

[0064] Inside the graphitization furnace, magnetic field coils with the same magnetic field direction as in step S2 are arranged, and the magnetic field gradient is set to continue the distribution of the forming stage in S2, that is, the magnetic field strength along the radial direction of the billet from the surface to the core is 0.9T→0.5T→0.15T.

[0065] Close the furnace door and begin gradient heating:

[0066] The temperature was raised from room temperature to 1000℃ at a rate of 5℃ / min, and then held at 1000℃ for 2 hours.

[0067] The temperature was increased from 1000℃ to 2000℃ at a rate of 3℃ / min, and then held at 2000℃ for 4 hours.

[0068] The temperature was increased from 2000℃ to 2500℃ at a rate of 2℃ / min.

[0069] When the furnace temperature reaches 2500℃, the gas introduction system is turned on, and a mixture of chlorine trifluoride and carbon tetrafluoride is introduced into the furnace at a volume ratio of 1:3 and a total flow rate of 10L / min, and the concentration of the mixed gas in the furnace is controlled at 10%.

[0070] The mixed gas is guided through a carbon-based catalyst material (a hollow cylindrical structure with an outer diameter of 1200 mm, an inner diameter of 1050 mm, and a height matching that of the billet) pre-placed around the billet. This hollow cylindrical catalyst material is tightly positioned around the billet, placing the billet at its center. This ensures that the mixed gas can fully contact the nitrogen-doped graphene loaded on the surface as it flows through the catalyst material, effectively guiding the gas to diffuse to all parts of the billet, especially compensating for the mass transfer difference between the core and surface of large billets, enhancing the reaction between impurities and the gas, and improving graphitization and purification effects. Simultaneously, the gas reflux device at the bottom of the furnace is activated, ensuring that the mixed gas has a longer residence time in the core of the billet than on the surface, with continuous gas flow for 8 hours.

[0071] After the ventilation is completed, the furnace temperature is increased from 2500℃ to 2900℃ at a rate of 1℃ / min, and then held at 2900℃ for 15 hours.

[0072] After the heat preservation is completed, turn off the heating system and allow the furnace temperature to cool down naturally to 2000℃. Then turn off the magnetic field coil and continue to cool down to room temperature before taking out the graphitized billet.

[0073] S4. Post-processing:

[0074] S41. Plasma Surface Activation: The graphitized preform is placed in the plasma treatment chamber, the chamber is closed and evacuated to 0.1 Pa, and then CH4 gas is introduced, with the gas flow rate controlled at 50 sccm. The radio frequency power supply is turned on and set to 350 W, and the surface of the preform is subjected to plasma activation treatment for 5 min. During the treatment, the surface temperature of the preform is monitored by thermocouples and maintained at 100±5℃.

[0075] S42. Deposition of boron nitride coating:

[0076] The activated graphitized preform was transferred to the deposition chamber of a hot-wire chemical vapor deposition (CVD) system. Using BN as the target material, a mixed reactive gas of NH3 and BCl3 (volume ratio 3:1) was introduced, with a total flow rate controlled at 20 sccm. The deposition temperature was 800℃, and the chamber pressure was 1 Pa. The deposition program was initiated, with each deposition cycle consisting of "30 seconds of deposition (during which the coating thickness increases by approximately 0.1 μm) → 10 seconds of gas interruption."

[0077] After completing 50 of the above deposition cycles, the deposition is paused, and a carbon quantum dot ethanol solution is uniformly deposited on the coating surface using a spray device to form a carbon quantum dot sublayer with a thickness of 0.1 nm. Then, vacuum drying is started for 30 seconds.

[0078] The above deposition cycle and carbon quantum dot sublayer insertion operation were repeated for a total of 100 deposition cycles, finally obtaining a boron nitride coating with a total thickness of 5 μm.

[0079] S43. Annealing treatment: Place the coated billet into an annealing furnace, place a carbon-based elastic component (expanded graphite / graphyne composite layer, 10mm thick) of the same material as the buffer material in step S2 on the outside of the billet, and apply a gradient pressure of 50MPa→30MPa→10MPa from the surface of the billet to the core through a pressurizing device.

[0080] Set the cooling program: from 2000℃ to 1000℃ at a cooling rate of 5℃ / min; from 1000℃ to room temperature at a cooling rate of 3℃ / min.

[0081] Throughout the annealing process, nitrogen gas was introduced into the furnace for protection at a flow rate of 20 L / min.

[0082] After annealing is completed, the annealing furnace is turned off. Once the furnace temperature has dropped to room temperature, the product is removed, which is the graphite product for large semiconductors.

[0083] The preparation method of the relevant materials in Example 1 is as follows:

[0084] The preparation method of expanded graphite / graphyne composite buffer material is as follows:

[0085] Expanded graphite (300 mesh) and graphylene are mixed at a mass ratio of 7:3. A small amount of ethanol is then added to form a paste. The paste is then cold-pressed into a 20 mm thick sheet under a pressure of 20 MPa and calcined in an inert atmosphere at 800 °C for 2 hours to obtain the final product.

[0086] The preparation method of carbon nanotube-graphene composite framework is as follows:

[0087] Multi-walled carbon nanotubes (10-30 nm in diameter) and graphene (monolayer) were mixed at a mass ratio of 1:2, and 5 wt% phenolic resin was added as a binder (based on the total mass of carbon nanotubes and graphene). The mixture was pre-formed in a mold under a pressure of 50 MPa, and then carbonized in an inert atmosphere at 1000 °C for 3 hours to achieve a radial porosity gradient distribution of 20%-40% (achieved through mold design).

[0088] The preparation method of Ni-Ti / carbon composite shape memory alloy ring is as follows:

[0089] Take a commercially available Ni-Ti shape memory alloy ring (Ni:Ti = 55:45, mass ratio), immerse it completely in 65wt% concentrated nitric acid for 30 min for etching, remove it and rinse it with deionized water until pH = 7, then vacuum dry it at 80℃ for 2 h.

[0090] The etched alloy ring was placed in a container, and 1 wt% graphene dispersion was added, with the liquid level 10 mm above the top of the ring. The ring was then allowed to stand and soak for 2 hours.

[0091] Remove the alloy ring, drain the surface liquid, place it in a tube furnace, and heat treat it at 600℃ for 1 hour under nitrogen protection (flow rate 20L / min). After natural cooling, a Ni-Ti / carbon composite shape memory alloy ring is obtained.

[0092] The preparation method of carbon-based catalytic materials is as follows:

[0093] Nitrogen-doped graphene preparation: A graphite rod (10 mm in diameter) was used as the anode and a copper rod as the cathode, placed in an arc discharge reactor. Urea and graphite powder were mixed at a mass ratio of 1:5 and pressed into rod-shaped doping sources with a diameter of 10 mm, which were then fixed next to the anode. After evacuating to 0.1 Pa, argon gas was introduced to 0.05 MPa, and the arc power supply (current 80 A, voltage 30 V) was turned on, discharging for 30 min. The product on the inner wall of the reactor was then collected, washed three times with 5 wt% dilute hydrochloric acid, rinsed with deionized water until neutral, and vacuum dried at 60 °C for 6 h to obtain nitrogen-doped graphene.

[0094] Porous graphite pretreatment: Take commercially available porous graphite (500mm in diameter), and adjust the density by pressing with a mold to achieve an overall porosity of 30%.

[0095] Weigh 100g of the above porous graphite, add 36g of 0.5wt% nitrogen-doped graphene dispersion (ethanol as solvent), soak for 12h, remove and drain the liquid, and calcine at 800℃ for 2h under argon protection (flow rate 15L / min) to obtain carbon-based catalytic material.

[0096] It is worth noting that all materials whose preparation methods are not specified above are commercially available materials.

[0097] Example 2

[0098] Compared to Example 1, the following differences exist:

[0099] S1. Raw material pretreatment.

[0100] The carbon-based functional material consists of 1.5 kg of magnetically responsive carbon nanotubes and 4.5 kg of carbon-based quantum dots (mass ratio 1:3), with a saturation magnetization of 6 emu / g for the magnetically responsive carbon nanotubes.

[0101] The amount of polyfullerene glycol ether added is 0.02 wt%.

[0102] The gradient magnetic field strength was 0.5T, the stirring and shearing speed was 2000r / min, and the time was 25min.

[0103] S2: Molding process.

[0104] S21. Mold preparation: The mold has an inner diameter of 1200mm and a height of 1200mm. The expanded graphite / graphyne composite buffer material is 25mm thick, and the outer sleeve is a Ni-Ti / carbon composite shape memory alloy ring with a diameter of 1300mm.

[0105] S23. Coupling process: The pressure gradient is 0MPa→50MPa→150MPa→400MPa (loading interval 100MPa, 250MPa), corresponding to a magnetic field strength of 0.2T→0.7T→1.0T (frequency 5Hz); the ultrasonic wave is introduced when the pressure rises to 300MPa (50kHz), with the focal point 500mm from the bottom.

[0106] S24. Removal of the blank: The diameter of the formed blank is 1200mm and the height is 950mm.

[0107] S3. Graphitization and purification.

[0108] The magnetic field gradient continues from 1.0T to 0.7T to 0.2T (radially from the surface to the core);

[0109] The mixed gas is chlorine trifluoride and carbon tetrafluoride (volume ratio 1:5), introduced at 2800℃, with a total flow rate of 12L / min.

[0110] S4. Post-processing.

[0111] Plasma activation power 500W, time 10min;

[0112] The boron nitride coating inserts a 0.2 nm carbon quantum dot sublayer every 70 deposition cycles, for a total of 140 cycles (total thickness 7 μm);

[0113] Annealing gradient pressure: 60MPa→40MPa→20MPa, cooling rate: 3℃ / min (2000→1000℃), 2℃ / min (1000→room temperature).

[0114] The remaining steps are the same as in Example 1.

[0115] Comparative Example 1

[0116] Compared to Example 1, in stage S1, the carbon-based functional material is replaced with Fe-catalyzed carbon nanotubes (containing 0.5 wt% Fe, without fullerene coating), and the remaining steps are the same as in Example 1.

[0117] Comparative Example 2

[0118] Compared to Example 1, there is no expanded graphite / graphyne composite buffer material in stage S21, and no Ni-Ti / carbon composite shape memory alloy ring in the outer layer; there is no stress regulation in stage S23, and the rest is the same as in Example 1.

[0119] Comparative Example 3

[0120] Compared to Example 1, no carbon-based catalyst material is used in the S3 stage, and the mixed gas diffuses naturally. The remaining steps are the same as in Example 1.

[0121] Comparative Example 4

[0122] Compared to Example 1, in stage S4, no carbon quantum dot sublayer is inserted when depositing the boron nitride coating, and 100 deposition cycles are completed directly. The remaining steps are the same as in Example 1.

[0123] Experimental Example

[0124] The products of Example 1, Example 2, and Comparative Examples 1-4 were selected for the following performance tests, and the test results are shown in Table 1.

[0125] Ash content (ppm): determined by a high-frequency infrared carbon-sulfur analyzer;

[0126] Radial resistivity difference (%): Four-probe method (10 points are taken uniformly along the radial direction);

[0127] Coating adhesion (N / cm): Cross-cut test (1mm spacing, tape peel test);

[0128] Bending strength (MPa): Three-point bending method (span 50mm, loading rate 2mm / min);

[0129] Cracking rate after thermal cycling (%): Visual observation was conducted after 10 thermal cycles (20-2000℃), and the proportion of cracked areas to the total area of ​​the sample was statistically analyzed.

[0130] Core-surface impurity difference (ppm): High-frequency inductively coupled plasma mass spectrometry.

[0131] Table 1:

[0132]

[0133]

[0134] Results analysis:

[0135] 1. The ash content of Examples 1-2 is only 3-4 ppm, far below the industry standard (≤10 ppm), and significantly lower than that of Comparative Example 1 (55 ppm) which uses Fe-catalyzed carbon nanotubes. This indicates that the "all-carbon-based functional material system" of the present invention (magnetically responsive carbon nanotubes coated with fullerene layers, carbon-based dispersants, etc.) effectively avoids the introduction of metal impurities and completely solves the problem of purity control failure.

[0136] 2. The core-to-surface impurity difference in Examples 1-2 was only 4-5 ppm, while in Comparative Example 3, which lacked carbon-based catalyst material and gas reflux, the difference reached 35 ppm. This demonstrates that the synergistic mechanism of "carbon-based catalyst-guided diffusion + gas reflux extending core reaction time" effectively eliminates the gradient difference in impurity removal and achieves uniform purification of the core and surface.

[0137] 3. The cracking rate after thermal cycling in Examples 1-2 was only 2-3%, and the flexural strength reached 90-92 MPa. In contrast, the cracking rate in Comparative Example 2, which lacked the buffer material and shape memory alloy ring, reached 20%, and the flexural strength was only 68 MPa. This indicates that the stress regulation system of "carbon-based buffer material + shape memory alloy ring" and the gradient pressure design during the annealing stage successfully alleviated the internal stress caused by the size effect and reduced the risk of cracking.

[0138] In summary, the preparation method of this invention comprehensively solves the technical problems of uncontrolled purity, large differences in impurity removal gradients, and high internal and external stresses in existing large-scale graphite products, meeting the stringent requirements for purity, uniformity, and mechanical reliability of large-scale semiconductor graphite. It is worth noting that the process of this invention is designed for large blanks with a diameter ≥ 1m. If small equipment (processing blanks with a diameter < 500mm) is used, the gradient pressure will cause the blank to become excessively dense or porous, carbon-based catalysis and gas reflux will cause excessive fluorine gas reaction, multi-field coupling will cause resistivity fluctuations, and post-processing will lead to a decrease in coating adhesion, thus having an adverse effect. Therefore, this method is not suitable.

Claims

1. A method for preparing graphite for large-scale semiconductors, characterized in that, Includes the following steps: S1. Raw material pretreatment: Needle coke is mixed with carbon-based functional materials, and a carbon-based dispersant is added under the action of a gradient magnetic field. Dispersion is achieved through stirring and shearing to obtain pretreated raw materials. S2. Molding process: The pretreated raw material is filled into a carbon-based mold, and gradient pressure is applied in sequence and combined with a gradient magnetic field for cold isostatic pressing. The stress is absorbed by the carbon-based buffer material to obtain the molded blank. S3. Graphitization and purification: The formed green body is placed in a large graphitization furnace, the magnetic field strength is controlled in stages, a mixed gas is introduced and the reaction is enhanced by carbon-based catalytic materials, and the reaction time in the core is extended by gas reflux to complete graphitization and purification and obtain a graphitized green body. S4. Post-processing: First, the graphitized preform is subjected to plasma surface activation, then a boron nitride coating is deposited, and annealing is performed by applying pressure with a carbon-based elastic component and controlling the cooling rate to obtain a large-scale graphite product for semiconductors.

2. The method for preparing graphite for large-scale semiconductors according to claim 1, characterized in that, The needle coke has a purity of ≥99.99% and a particle size distribution of 5-30 μm.

3. The method for preparing graphite for large-scale semiconductors according to claim 2, characterized in that, In step S1, the carbon-based functional material includes magnetically responsive carbon nanotubes and carbon-based quantum dots, with a mass ratio of 1:3 to 1:7; the magnetically responsive carbon nanotubes have a diameter of 10-30 nm, a length of 100-300 nm, and are coated with a fullerene layer, with a saturation magnetization ≥5 emu / g.

4. The method for preparing graphite for large-scale semiconductors according to claim 3, characterized in that, The gradient magnetic field is 0.05-0.5T in the raw material pretreatment stage and 0.1-1.0T in the molding stage.

5. The method for preparing graphite for large-scale semiconductors according to claim 4, characterized in that, In step S1, the carbon-based dispersant is polyfullerene glycol ether, and its addition amount is 0.005-0.02 wt%.

6. The method for preparing graphite for large-scale semiconductors according to claim 1, characterized in that, In step S2, the pressure range of gradient pressurization is 50-600MPa, with the surface layer at 50-200MPa, the middle layer at 200-400MPa, and the core at 400-600MPa in a stepped manner.

7. The method for preparing graphite for large-scale semiconductors according to claim 6, characterized in that, The specific steps of step S2 are as follows: S21. Mold preparation: A carbon-based mold with an embedded carbon nanotube-graphene composite framework is used. The porosity of the framework is distributed radially in a gradient of 20% to 40%, and carbon quantum dots are loaded on the surface of the framework. A carbon-based piezoelectric sensing layer is embedded in the inner wall of the mold, and a Ni-Ti / carbon composite shape memory alloy ring is wrapped on the outer layer. S22. Raw material filling: Fill the pretreated raw material into the mold prepared in step S21; S23. Coupling treatment: Apply pressure in stages at a set rate, successively reaching 50-200MPa, 200-400MPa, and 400-600MPa; while applying pressure, simultaneously apply axial rotating magnetic fields of 0.1-0.2T, 0.3-0.7T, and 0.8-1.0T, with a magnetic field rotation frequency of 1-5Hz; when the pressure rises to the 200-400MPa range, introduce 20-50kHz focused ultrasonic waves, focusing on the core region of the billet; During the coupling process, the carbon-based piezoelectric sensing layer monitors the internal stress of the blank. When the local stress exceeds the set value, the carbon-based buffer material (expanded graphite / graphyne composite layer) undergoes reversible deformation, and at the same time, the outer carbon-based shape memory alloy ring of the mold undergoes radial shrinkage. S24. Removal of blank: After processing is completed, the shaped blank is obtained.

8. The method for preparing graphite for large-scale semiconductors according to claim 1, characterized in that, In step S3, the mixed gas is a mixture of chlorine trifluoride and carbon tetrafluoride with a volume ratio of 1:2 to 1:5, and is introduced at a temperature of 2000-2800°C.

9. The method for preparing graphite for large-scale semiconductors according to claim 8, characterized in that, In step S3, the carbon-based catalytic material is a porous graphite medium with 0.3-0.8 wt% nitrogen-doped graphene loaded on its surface.

10. The method for preparing graphite for large-scale semiconductors according to claim 1, characterized in that, In step S4, the surface activation is performed using CH4 plasma treatment with a power of 200-500W for 3-10 minutes.

Images