A stress regulation processing method and system for single-side polished heat sink

Abstract

This application proposes a stress-controlled processing method and system for single-sided polished heat sinks, comprising: determining a first surface and a second surface; polishing the first surface and performing gradient ion implantation to form a doped layer; bombarding the second surface with gradient plasma to form a gradient pre-compensated stress field and a gradient temperature response layer; performing metastable stress pre-release treatment on the heat sink to be processed; and performing stress field locking treatment on the heat sink that has completed metastable stress pre-release treatment to obtain a single-sided polished heat sink. This application achieves precise compensation of static thermal pressure stress and adaptation to dynamic deformation across the entire temperature range through gradient plasma bombardment and gradient ion implantation, effectively bridging the thermal expansion difference between the chip and the heat sink, continuously ensuring tight adhesion between the chip and the heat sink, significantly improving interface thermal conductivity, and significantly reducing the risk of chip cracking and interface debonding.

Description

Technical Field

[0001] This application relates to the field of heat sink preparation, and more specifically, to a stress-controlled processing method and system for a single-sided polished heat sink. Background Technology

[0002] High-power laser chips generate a lot of heat during packaging and use. They are usually dissipated by metal heat sinks that are polished on one side. These heat sinks have only one side with a high-precision polished surface for attaching and fixing the light-emitting chip, while the other side serves as a heat dissipation surface that directly participates in heat exchange. With the structural advantage of single-sided processing, they are suitable for compact packaging and assembly scenarios.

[0003] Existing conventional single-sided polishing heat sink processing technology focuses solely on the precision polishing of the chip bonding surface to reduce flatness and roughness, unilaterally pursuing the smoothness of the bonding interface, and lacking overall control over stress state during processing. After the single-sided precision polishing process, uneven residual stress is easily retained inside the heat sink substrate. When the laser chip is subjected to alternating high and low temperatures and dynamic power changes for a long time, the thermal stress generated by the chip operation will continuously act on the single-sided bonded heat sink structure. The original processing stress retained by the heat sink itself and the external thermal stress are superimposed and coupled, which can easily cause the heat sink to slowly deform and warp. On the one hand, it will cause gaps at the chip-heat sink bonding interface, significantly reducing the interface thermal conductivity and causing local heat accumulation and overheating problems in the device; on the other hand, continuous stress pulling can easily cause chip cracking and debonding of the packaging interface, seriously shortening the overall service life of optoelectronic devices. Summary of the Invention

[0004] To address the problems existing in the current technology, this application provides a stress-controlled processing method and system for single-sided polishing heat sinks. The specific solution is as follows:

[0005] A stress-controlled processing method for single-sided polishing heat sinks includes: Obtain the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink, and determine the polishing scheme based on the material characteristics of the first surface and the second surface. First, the first surface is polished according to the polishing scheme. Then, gradient ion implantation is performed on the polished first surface to form a doped layer with a continuously varying coefficient of thermal expansion, thus obtaining a functional surface. The functional surface is then sealed and protected. The polishing scheme employs gradient plasma bombardment of the second surface to form a gradient pre-compensation stress field and a gradient temperature response layer on the second surface, thereby synergistically compensating for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite in direction to the thermal stress of the chip. Remove the sealing protection of the functional surface and perform metastable stress pre-release treatment on the heat sink to be processed to accelerate the natural relaxation process of the residual stress introduced after plasma bombardment. The heat sink that has completed metastable stress pre-release is subjected to stress field locking treatment to solidify the stress distribution state of the gradient pre-compensated stress field and gradient temperature response layer, resulting in a single-sided polished heat sink.

[0006] A stress-controlled processing system for single-sided polishing heat sinks, comprising: The heat sink preparation unit is used to acquire the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink to be processed, and determine the polishing scheme according to the material characteristics of the first surface and the second surface. The polishing unit is used to first polish the first surface based on the polishing scheme, then perform gradient ion implantation on the polished first surface to form a doped layer with a continuously varying coefficient of thermal expansion, thereby obtaining a functional surface, and then seal and protect the functional surface. The stress compensation unit is used to bombard the second surface with gradient plasma based on the polishing scheme to form a gradient pre-compensation stress field and a gradient temperature response layer on the surface of the second surface, so as to synergistically compensate for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite to the direction of the chip thermal stress. The first stabilizing unit is used to remove the sealing protection of the functional surface and to perform metastable stress pre-release treatment on the heat sink to be processed, thereby accelerating the natural relaxation process of the residual stress introduced after plasma bombardment. The second stabilizing unit is used to perform stress field locking treatment on the heat sink that has completed metastable stress pre-release, solidify the stress distribution state of the gradient pre-compensated stress field and the gradient temperature response layer, and obtain a single-sided polished heat sink.

[0007] Beneficial effects: This application proposes a stress-controlled processing method and system for single-sided polished heat sinks. By using gradient plasma bombardment and gradient ion implantation, it achieves precise cancellation of static hot-pressure stress and dynamic deformation adaptation across the entire temperature range, effectively bridging the thermal expansion difference between the chip and the heat sink, continuously ensuring tight adhesion between the chip and the heat sink, significantly improving interface thermal conductivity, significantly reducing the risk of chip cracking and interface debonding, effectively extending the overall service life of high-power optoelectronic devices, and meeting the current high-precision and high-reliability requirements for heat sinks used in high-power lasers.

[0008] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the stress-controlled processing method of this application; Figure 2 This is a schematic diagram illustrating the principle of the stress-controlled processing method of this application; Figure 3 This is a schematic diagram of the double-sided processing flow of this application; Figure 4 This is a schematic diagram of the generation process of the multifunctional layer in this application; Figure 5 This is a schematic diagram of the layered preparation system module of this application.

[0011] Reference numerals in the attached figures: 1-Heat sink preparation unit; 2-Polishing processing unit; 3-Stress compensation unit; 4-First stabilizing unit; 5-Second stabilizing unit. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0013] This application proposes a stress-controlled processing method for single-sided polishing heat sinks. Through a systematic stress management approach covering the entire process, it achieves active control and long-term stability of the internal stress of the heat sink. A schematic diagram of the stress-controlled processing method is attached. Figure 1 As shown in the attached diagram, the principle is as follows. Figure 2 As shown, the specific solution is as follows: A stress-controlled processing method for single-sided polishing heat sinks includes: 101. Obtain the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink, and determine the polishing scheme based on the material characteristics of the first surface and the second surface. 102. First, polish the first surface based on the polishing scheme, then perform gradient ion implantation on the polished first surface to form a doped layer with a continuously varying coefficient of thermal expansion, thus obtaining the functional surface, and then seal and protect the functional surface. 103. Based on the polishing scheme, gradient plasma is used to bombard the second surface to form a gradient pre-compensation stress field and a gradient temperature response layer on the surface of the second surface, so as to synergistically compensate for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite to the direction of the chip thermal stress. 104. Remove the sealing protection of the functional surface and perform metastable stress pre-release treatment on the heat sink to be processed to accelerate the natural relaxation process of the residual stress introduced after plasma bombardment. 105. Perform stress field locking treatment on the heat sink that has completed metastable stress pre-release, solidify the stress distribution state of the gradient pre-compensated stress field and gradient temperature response layer, and obtain a single-sided polished heat sink.

[0014] This application first determines a targeted polishing scheme by combining the material characteristics of both sides of the heat sink. While ensuring the flatness and roughness accuracy of the chip bonding surface, it controls the initial residual stress introduced during polishing within a reasonable range, avoiding the generation of uneven primary processing stress from the source of processing and eliminating the fundamental cause of stress superposition and coupling in traditional processes. On this basis, a doped layer with a continuously gradient change in the coefficient of thermal expansion is formed on the functional surface by gradient ion implantation, effectively bridging the difference in the coefficient of thermal expansion between the chip and the heat sink, eliminating the root cause of thermal mismatch stress at the interface level, and avoiding stress concentration problems at the interface. The core innovation of this method lies in using gradient plasma to bombard the second surface, while simultaneously forming a gradient pre-compensation stress field and a gradient temperature response layer on the surface. The stress introduced by the gradient pre-compensation stress field is completely opposite in direction to the thermocompression stress generated by the chip operation and is precisely matched in distribution, which can directly offset most of the static thermocompression stress. The gradient temperature response layer can automatically adjust its own thermal expansion state with temperature changes, dynamically supplementing the stress compensation capability under different operating conditions. The synergistic effect of the two achieves stress balance across the entire temperature and power range, avoiding warping deformation of the heat sink caused by stress superposition. Furthermore, through metastable stress pre-release treatment, the natural relaxation process of metastable residual stress introduced after plasma bombardment is accelerated, ensuring that the residual stress field reaches a thermodynamically stable state before leaving the factory. This completely solves the problem of warpage changes over time due to spontaneous stress relaxation during storage and transportation in traditional processes, guaranteeing the consistency of batch products. Finally, through stress field locking treatment, the stress distribution state of the gradient pre-compensation stress field and gradient temperature response layer is solidified, preventing stress migration and decay under long-term temperature and power cycling conditions, significantly improving the long-term service stability of the heat sink.

[0015] Step 101 is the foundational preparation stage of the entire process, its core function being to establish a unified processing benchmark and process adaptation system. The heat sink to be processed is typically a metal-based or composite-based heat dissipation material such as oxygen-free copper-molybdenum alloy or diamond-copper composite. Its two surfaces serve completely different functions. The first surface, as the chip bonding surface, requires extremely high flatness and surface quality to ensure tight bonding and efficient heat conduction between the chip and the heat sink. The second surface, as the heat dissipation surface, is mainly responsible for heat conduction and dissipation, with relatively lower surface precision requirements but needing good heat dissipation performance. Therefore, the functional positioning of the two surfaces must be clearly defined before a targeted processing plan can be developed. Material characteristics are the core basis for determining the polishing scheme; heat sinks of different materials exhibit significant differences in hardness, thermal conductivity, elastic modulus, and processing damage characteristics. For example, oxygen-free copper has low hardness and good ductility, making it prone to plastic deformation and surface scratches during polishing. Therefore, a softer polishing pad and fine-grained abrasive are needed, along with controlled polishing pressure and speed. In contrast, diamond-copper composites have high hardness and are more brittle, requiring even harder abrasives and higher polishing pressure to effectively remove material and achieve a good surface finish. Materials with high thermal conductivity dissipate heat more easily during polishing, reducing the likelihood of localized thermal deformation. Materials with high elastic modulus retain residual stress more easily after processing, making it difficult to relax. The processing damage characteristics determine the thickness of material to be removed in each polishing step to completely eliminate the damage layer from previous processes. Only a polishing plan tailored to the material characteristics can ensure processing accuracy while controlling the initial residual stress introduced during polishing within a reasonable range, laying the foundation for subsequent stress control.

[0016] Step 102 is the functional surface fabrication and protection stage. The process sequence of polishing followed by ion implantation and then sealing for protection has clear technical rationale. Polishing the first surface gradually removes surface defects and damaged layers generated during rough machining, obtaining a high-precision surface that meets chip bonding requirements. Simultaneously, the mechanical action between the abrasive and the heat sink surface during polishing naturally introduces residual stress, typically compressive stress, which can improve the fatigue resistance and damage resistance of the functional surface to some extent. The subsequent gradient ion implantation process forms a doped layer with a continuously varying coefficient of thermal expansion on the functional surface. The core function of this structure is to eliminate the thermal mismatch stress caused by the difference in thermal expansion coefficients between the chip and the heat sink. The coefficient of thermal expansion of the chip material is usually much lower than that of the heat sink substrate material. For example, the coefficient of thermal expansion of gallium nitride chips is approximately 5.5 ppm per degree Celsius, while that of oxygen-free copper is approximately 17 ppm per degree Celsius. When the two are directly bonded, enormous thermal stress is generated at the interface during temperature changes. This stress is sufficient to cause chip cracking or interface debonding. Continuous gradient ion implantation avoids abrupt stress changes and stress concentrations at the interface, allowing thermal stress to be gradually released within the doped layer, significantly improving the reliability and lifespan of the bonding interface. Gradient ion implantation is achieved by continuously controlling the energy and dosage of ion implantation, causing the concentration of dopants on the functional surface to continuously vary along the depth direction, thus achieving continuous gradient control of the coefficient of thermal expansion. Sealing the functional surface after fabrication is crucial to prevent subsequent plasma processing from etching and contaminating it, while maintaining the established surface morphology and stress state. Sealing is typically achieved by covering the functional surface with an insulating material, then using vacuum adsorption or mechanical fixation to ensure tight adhesion. Without sealing, high-energy particles during plasma processing bombard the surface, increasing surface roughness and decreasing flatness. Furthermore, it introduces additional non-uniform stress, disrupting the established stress balance and severely impacting the precision and performance of the functional surface.

[0017] Step 103 is the core stress control step of the entire process. Gradient plasma bombardment of the second surface layer simultaneously forms a gradient pre-compensation stress field and a gradient temperature response layer. These two elements work together to precisely compensate for the thermal stress of the chip during operation. During chip operation, a large amount of heat is generated, with a distribution that is high in the center and low at the edges. This non-uniform heat distribution causes non-uniform thermal expansion of the heat sink; the central region of the heat sink has a high temperature and large expansion, while the edge region has a low temperature and small expansion, resulting in non-uniform thermal stress. This stress can cause the heat sink to warp upwards in a bowl shape, and in severe cases, it can lead to gaps at the chip-heat sink interface or even chip cracking. The gradient pre-compensation stress field introduces stress in the opposite direction to the chip's thermal stress, and its stress distribution perfectly matches the chip's thermal stress distribution. It can directly offset most of the thermal stress during heat sink operation, keeping the heat sink flat during operation. The gradient pre-compensation stress field is achieved by controlling the power density distribution of the plasma, so that different regions of the second surface receive different plasma bombardment intensities, thereby introducing residual stresses of varying magnitudes. For example, for chips with high central heat generation, the plasma power density can be set to be low at the center and high at the edges, thus creating a gradient tensile stress field on the second surface layer that is low at the center and high at the edges. This field precisely cancels out the thermocompressive stress field generated during chip operation, which is high at the center and low at the edges. The gradient temperature response layer can automatically adjust its thermal expansion state according to changes in ambient temperature, supplementing the compensation capability of the gradient pre-compensation stress field in different temperature ranges. The gradient pre-compensation stress field is designed for the chip's rated operating temperature. Under low-temperature start-up and overload high-temperature conditions, its compensation effect will decrease. The gradient temperature response layer can automatically adjust its thermal expansion according to temperature changes, contracting at low temperatures to generate additional tensile stress and expanding at high temperatures to generate additional compressive stress. The synergistic effect of these two processes can achieve stress balance across the entire temperature range, ensuring that the heat sink remains flat throughout the entire operating temperature range from low-temperature start-up to overload high temperatures. The gradient plasma bombardment process can precisely control the spatial distribution of stress and thermal expansion coefficient by adjusting the plasma power distribution and application time, adapting to the heating characteristics of chips of different sizes and power.

[0018] Step 104 is the metastable stress pre-release stage, whose core function is to eliminate the time lag effect of warpage. The plasma bombardment process introduces a large number of high-energy particles into the heat sink surface. These high-energy particles collide with the atoms on the heat sink surface, causing them to deviate from their equilibrium positions and forming a large number of lattice distortions. These lattice distortions are in an unstable metastable state and have a tendency to spontaneously transition to lower energy states. In traditional processes, these metastable stresses will spontaneously relax with changes in storage time and ambient temperature. This process usually takes tens or even hundreds of hours to complete at room temperature, causing the warpage of the heat sink to change slowly after leaving the factory, seriously affecting the consistency and reliability of the product. It is common for the warpage to be acceptable at the time of leaving the factory, but exceed the standard after a month of storage. This embodiment uses metastable stress pre-release treatment to apply uniform external energy to the heat sink, activating the diffusion and dislocation movement of surface atoms, accelerating the relaxation process of metastable lattice distortions, and enabling the residual stress field to reach a thermodynamically stable state before leaving the factory. Metastable stress pre-release treatment is typically achieved by placing the heat sink in a controlled thermal environment. Heating allows atoms to gain sufficient energy for diffusion and rearrangement, thereby accelerating the stress relaxation process. After metastable stress pre-release treatment, the warpage of the heat sink during subsequent storage and use is controlled within a very small range, solving the problem of spontaneous deformation during storage and use, and ensuring the consistency and long-term stability of batch products.

[0019] Step 105 is the stress field locking process, used to further solidify the stress distribution state and improve the long-term service reliability of the heat sink. Although the metastable stress pre-release treatment stabilizes the stress field, under long-term temperature and power cycling conditions, atoms inside the heat sink will still slowly diffuse and rearrange, leading to slow stress migration and changes. Over time, this can still cause heat sink warping and deformation. The stress field locking treatment, through the synergistic effect of multiple physical fields, induces ordered rearrangement of atoms on the heat sink surface, stabilizes the lattice distortion structure corresponding to the gradient pre-compensation stress field, and strengthens the interface bonding between the gradient temperature response layer and the heat sink substrate, permanently solidifying the stress distribution state of both. The stress field locking treatment is typically achieved by simultaneously applying a thermal field and an auxiliary physical field to the heat sink in a vacuum environment. Through the synergistic effect of thermal activation and the auxiliary physical field, the atomic arrangement becomes more stable, and the interface bonding becomes stronger. After thousands of temperature cycles, the warping change of the heat sink treated with stress field locking can still be controlled within a very small range, far superior to heat sinks processed by traditional methods.

[0020] In some embodiments, material characteristics include hardness, thermal conductivity, elastic modulus, and processing damage characteristics. Based on these material characteristics, at least the abrasive grain size, polishing pressure, polishing disc rotation speed, and polishing time for each stage on the first surface are determined. The magnitude and distribution of residual surface stress introduced after polishing the first surface are predicted. Combined with the plasma etching rate of the heat sink, the plasma power gradient distribution, gas flow ratio, and processing time for each stage on the second surface are determined. The inherent hardness, thermal conductivity, elastic modulus, and processing damage characteristics of the material are selected as the core judgment criteria. This enables the linkage setting of process parameters for the two core processing steps, ensuring that all processing parameters perfectly match the material properties of the heat sink substrate. This controls the generation magnitude and distribution law of processing stress from the source of the process, achieving mutual adaptation and synergistic matching of the processing processes on both sides of the single-sided polishing heat sink.

[0021] First, based on the fundamental properties of various materials, the polishing process parameters for the first chip bonding surface are precisely determined. Material hardness directly determines the appropriate mechanical force and abrasive selection during polishing. Softer pure copper heat sink substrates are prone to surface plastic deformation under external friction and pressure. These materials require smaller abrasive particles and lower polishing pressure to prevent surface wrinkles and excessive compressive stress buildup. Copper-molybdenum composite materials, with higher hardness, exhibit stronger wear and extrusion resistance. Larger abrasive particles and appropriately increased polishing pressure can be used to improve material removal efficiency. Material thermal conductivity directly affects the rate of heat dissipation during polishing. Substrate materials with excellent thermal conductivity can quickly dissipate frictional heat generated at the polishing contact surface, preventing temporary thermal stress caused by localized heat accumulation. For such materials, a higher polishing disc rotation speed can be set to shorten the overall processing cycle. Substrate materials with weaker thermal conductivity require a lower disc rotation speed and a slower processing pace to avoid minor deformations caused by localized high temperatures. The elastic modulus of a material represents its ability to resist deformation and recover from deformation. Materials with a high elastic modulus are less likely to rebound and recover after being subjected to external forces, resulting in slower dissipation of residual mechanical stress after polishing. This necessitates extending the intervals between polishing stages and the overall operation time. Materials with a lower elastic modulus have stronger deformation recovery capabilities, allowing for rapid release of instantaneous processing stress, thus enabling more efficient reduction of polishing time at each stage. Processing damage characteristics specifically refer to the inherent defect structures such as grain breakage layers, internal microcrack layers, and plastic deformation layers remaining on the surface of the heat sink after initial roughing. The thickness and layers of damage layers formed by different materials vary significantly. Based on this characteristic, the operation time for each stage of semi-finishing, finishing, and final polishing can be clearly defined. This allows for the complete removal of all processing damage left over from previous stages and avoids unnecessary material loss due to over-polishing. By combining these multiple material characteristics, the abrasive grit size, polishing pressure, actual polishing disc speed, and polishing operation time for each sub-stage within the entire process can be accurately determined.

[0022] Based on the determination of complete polishing process parameters and the finalization of the first-side processing parameters, it is possible to accurately predict the overall magnitude of residual stress generated on the surface of the first side after the entire polishing process, as well as the actual distribution of stress at different locations on the surface. This allows for advance prediction and quantitative control of processing stress. Furthermore, by combining the inherent plasma etching rate and stress introduction efficiency of the heat sink substrate material, the planning and setting of all process parameters for the gradient plasma processing of the second-side heat dissipation surface are completed. The plasma etching rate represents the rate at which different substrate materials are stripped of their surface material by high-energy particle bombardment under the same plasma environment. Materials with faster etching rates are more prone to surface material loss in a short time. To ensure the uniformity of the overall thickness and the integrity of the surface structure of the second side, the overall plasma treatment time needs to be appropriately reduced. Materials with slower etching rates have stronger resistance to particle bombardment, so the processing time can be appropriately extended to ensure the full formation of the target functional structure. Stress introduction efficiency refers to the forming speed and effect of a substrate material forming a gradient pre-compensated stress field and a gradient temperature response layer under the same plasma operation conditions. Materials with high stress introduction efficiency only need to set a small range of power differences to form a gradient stress distribution that meets the design standards. Materials with low stress introduction efficiency need to increase the difference in plasma output power between different regions to compensate for the material's insufficient forming efficiency. By utilizing the inherent processing characteristics of the two types of materials, operators can accurately plan the overall plasma power gradient arrangement of the second surface, clarify the power rise and fall pattern from the center region to the edge region, and rationally adjust the flow rate ratio between inert protective gas and functional reactive gas to ensure that the two effects of particle bombardment stress formation and gas phase deposition layer formation are carried out simultaneously and stably. Finally, the processing time corresponding to each stage of the plasma step-by-step operation is determined.

[0023] In some embodiments, polishing the first surface specifically includes: semi-finish polishing the first surface with an abrasive of a first grit size to completely remove the damaged layer generated by roughing; finishing polishing the first surface with an abrasive of a second grit size to achieve a first predetermined flatness value; and finally polishing the first surface with an abrasive of a third grit size to achieve a second predetermined surface roughness value, and introducing a residual compressive stress field on the surface layer of the first surface by controlling the polishing parameters; wherein the first grit size is larger than the second grit size, and the second grit size is larger than the third grit size. The processing of the first surface is as shown in the attached figure. Figure 3 As shown, a three-stage polishing mode with progressively decreasing particle size is used to complete the overall finishing of the first surface. This processing mode follows the processing logic of rough shaping, then fine shaping, and finally fine finishing. It can not only remove various surface defects left over from the previous processing layer by layer, steadily improve the geometric accuracy and surface smoothness of the bonding surface, but also actively shape the surface mechanical stress state at the end of the processing.

[0024] First, a semi-finish polishing operation is carried out using a larger-sized abrasive grain. The larger-sized abrasive grain has stronger overall cutting and surface peeling capabilities. In the actual processing, the continuous grinding action between the abrasive grains and the first surface of the heat sink can completely remove the surface damage structure formed after rough processing such as cutting and milling. This type of damage structure includes fine processing cracks in the surface plastic deformation area and grain misalignment damage layer. If it remains for a long time, it will directly affect the subsequent polishing accuracy and interface bonding stability. Semi-finish polishing can completely remove all the damage layers, creating a flat and clean base surface and eliminating the interference of structural defects caused by the previous processing.

[0025] After the semi-finish polishing is completed, a second-size abrasive with a smaller particle size is used for fine polishing. Compared with the previous abrasive, this size abrasive has a smoother and more uniform grinding force, and will not cause a large amount of material removal from the substrate. During the processing, it can effectively repair the coarse grinding marks left by the semi-finish polishing process, gradually correct the height differences and local deformation problems on the surface of the heat sink, and continuously optimize the overall flatness of the first surface until the flatness parameters of the entire bonding surface of the heat sink stably reach the preset standard range. The core purpose of this step is to calibrate the macroscopic surface morphology to ensure that there will be no local suspension or misalignment when the heat sink and the chip are connected.

[0026] Finally, the third-finest abrasive with the smallest particle size is selected for the final polishing process. This ultrafine abrasive can only slightly modify the surface microstructure. During processing, a low-pressure, stable polishing condition is used to continuously refine the surface microstructure, gradually reducing the surface roughness of the first surface. This ensures the surface roughness meets the predetermined values ​​required for assembly, guaranteeing a sufficiently tight bond between the chip and the heat sink, minimizing interfacial thermal resistance, and ensuring efficient heat transfer. In this final polishing step, operators can flexibly adjust controllable parameters such as polishing contact pressure, polishing disc operation mode, and overall polishing time. Relying on the long-term, uniform, and gentle squeezing friction of the ultrafine abrasive, a stable and uniformly distributed residual compressive stress field is generated on the first surface. This compressive stress effectively improves the density and fatigue resistance of the bonding surface structure. When the device experiences tensile stress due to temperature changes, the compressive stress inherent in the surface layer creates a balancing effect, reducing defects such as warping, cracking, and interface delamination.

[0027] Throughout the entire process, the particle size distribution rule is strictly followed: the first particle size is larger than the second particle size, and the second particle size is larger than the third particle size. This design method of gradually reducing the abrasive particle size conforms to the basic laws of precision machining. In the early stage, large-particle abrasive is used to quickly complete defect cleaning and base surface shaping. In the middle stage, medium-particle abrasive is used to calibrate the overall planar shape. In the later stage, small-particle abrasive is used to complete mirror finishing and stress control. Throughout the process, there will be no problem of coarse-particle abrasive scratching the finely machined surface. While ensuring the overall processing efficiency, the three core indicators of geometric dimensional accuracy, surface appearance quality, and surface stress state are simultaneously achieved.

[0028] In some embodiments, the doped layer includes a first doped layer and a second doped layer whose doping concentration varies continuously along the depth direction. First metal ions are individually implanted into the polished first surface, with the ion implantation energy continuously and linearly decreasing while the ion implantation dose is continuously and linearly increasing, forming a first doped layer on the surface of the first surface. Based on the first doped layer, first metal ions and second non-metal ions are implanted, with the implantation energy of both ions continuously and linearly decreasing over time, and the implantation dose ratio remaining constant, forming a second doped layer in the shallow part of the functional surface. The formation process of the doped layer is as follows: Figure 3 As shown, the doped layer constructed on the functional surface is divided into a first doped layer and a second doped layer that are closely connected. Both layers can achieve a continuous and stable gradient change in the concentration of doped elements along the depth direction of the heat sink substrate, without any abrupt boundary regions of concentration. The entire structure is formed by two stages of different ion implantation processes. By dynamically adjusting the energy value and implantation dose during the ion implantation process, the distribution of various doped elements in the surface layer can be precisely controlled. The dual-layer gradient doping structure achieves a smooth transition of the coefficient of thermal expansion, effectively bridging the thermal property differences between the heat sink substrate and the bonded chip.

[0029] First, the preparation of the inner first doped layer is completed. During the preparation phase, only a single type of first metal ions are introduced into the polished first surface to complete the implantation operation. In the actual processing, two dynamic parameter adjustments are performed simultaneously: the ion implantation energy is continuously and linearly decreased with the operation time, while the overall ion implantation dose is linearly increased with the operation time. The ion implantation energy directly determines the penetration depth of ions into the heat sink substrate. The higher the implantation energy, the deeper the ions can reach into the substrate. Continuously decreasing the implantation energy makes it difficult for subsequently injected metal ions to penetrate into the substrate, and they can only stay in the relatively outer area. On the other hand, simultaneously increasing the implantation dose means that the number of metal ions doped per unit area is continuously increasing. With the synergistic change of the two parameters, a distribution pattern is naturally formed in which the metal ion doping concentration is higher closer to the substrate and gradually decreases closer to the outer surface. This forms the first doped layer with a continuous gradient characteristic in the depth direction. This layer is set close to the heat sink substrate and is mainly used to support the material properties of the substrate itself, initially adjust the basic value of thermal expansion of the overall area, and build a transition foundation between the substrate and the outer doped structure.

[0030] After the first doped layer has solidified, the implantation process for the second doped layer continues. In this stage, both first metal ions and second non-metal ions are simultaneously introduced for co-implantation. Both types of ions follow a consistent rule where the implantation energy decreases linearly with the duration of the process, maintaining the same energy change rhythm as the previous step. This ensures a smooth transition in elemental distribution between the two doped structures, preventing structural delamination. Throughout the co-implantation process, the ratio of the two ion doses remains strictly constant, without arbitrary changes to the supply ratio. This constant ratio ensures a balanced composition of the two elements within the mixed doping region, preventing uncontrolled fluctuations in thermal expansion coefficients caused by localized imbalances. With the continuously decreasing implantation energy, the two types of mixed ions are concentrated at the outermost, shallowest layer of the functional surface, ultimately adhering to the outermost surface of the first doped layer to form the second doped layer. Within the second doped layer, a continuous, gradually changing doping concentration gradient also forms along the depth direction.

[0031] The inner first doped layer primarily interfaces with the heat sink substrate material, smoothly transitioning the inherent thermal expansion parameters of the substrate. The outer second doped layer directly approaches the chip structure to be assembled, relying on a composite doping combination of metal and non-metal ions to finely adjust the thermal expansion values ​​of the surface region, making the thermal expansion characteristics of the outermost surface of the functional surface infinitely close to the thermal expansion performance of the chip material itself. The entire implantation process precisely controls the ion injection depth through linear energy changes and regulates the element enrichment degree through dosage changes, achieving a smooth transition of the thermal expansion coefficient along the depth direction of the substrate throughout the process. This eliminates the potential for stress concentration at the interface. Furthermore, the uniform and consistent parameter adjustment method is simple and controllable, enabling stable replication of the double-layer gradient doped structure in actual mass production, ensuring consistent thermal expansion adaptation of the functional surfaces in batch products.

[0032] In some embodiments, the first metal ion is aluminum ion, used to regulate the low to medium range of the thermal expansion coefficient of the doped layer; the second non-metal ion is silicon ion, used to regulate the low range of the thermal expansion coefficient of the doped layer; the synergistic effect of aluminum ion and silicon ion achieves a wide range of continuously adjustable thermal expansion coefficients.

[0033] Aluminum ions themselves possess moderate thermal expansion characteristics. After being incorporated into the lattice structure of the heat sink surface layer, there will be no drastic fluctuations in thermal expansion parameters. It can stably and flexibly control the range of medium and low thermal expansion coefficients in the doped layer. In the first doped layer formed by only implanting aluminum ions, the doping concentration of aluminum ions gradually changes along the depth direction, which can gradually reduce the original high thermal expansion value of the heat sink along the substrate depth. It can smoothly complete the transition adjustment from the inherent thermal expansion performance of the heat sink substrate to the adaptation range, which can cover the medium and low thermal expansion parameter range required for the adaptation of most conventional optoelectronic devices and meet the basic interface adaptation conditions of general-purpose chips.

[0034] Silicon ions, as non-metallic dopants, possess an extremely low intrinsic coefficient of thermal expansion. When incorporated into the substrate surface, they can significantly reduce the overall expansion capacity of the corresponding region. They are specifically designed for precise control of the low coefficient of thermal expansion range, filling the gap in ultra-low expansion parameter adjustment that cannot be achieved by aluminum ions alone. In particular, they are well-suited for the needs of high-power chips with low coefficients of thermal expansion, such as gallium nitride. In the second doped layer formed by the co-doping of the two ions, silicon ions can finely adjust the thermal expansion value of the outer region of the surface layer by maintaining a constant ratio, allowing the thermal expansion performance of the outermost functional surface to closely approximate the properties of the chip's core material.

[0035] Aluminum ions are responsible for adjusting parameters over a wide range of medium-range conditions, while silicon ions are responsible for fine-tuning voltage drops in the low-range conditions. The two ions work together in a layered manner with their concentrations changing in sync to create a gradient effect. This allows for a wide-range, continuous, and uninterrupted adjustment of the thermal expansion coefficient of the overall doped structure. This approach can handle conventional assembly scenarios where the thermal expansion difference between the heat sink and the chip is small, as well as special high-power device assembly scenarios where the thermal expansion parameters of the two components differ significantly. Furthermore, both types of ions exhibit excellent compatibility with mainstream heat sink substrates such as copper-based alloys. After doping, there will be no issues with impurity precipitation or large-area lattice distortion. This ensures the long-term stable maintenance of the thermal expansion adjustment capability of the doped layer, continuously weakens the thermal mismatch stress generated at the interface during temperature alternation, and guarantees the integrity and stability of the bonding interface structure over the long term.

[0036] In some embodiments, the formation of the gradient pre-compensated stress field includes: introducing an inert working gas, controlling the plasma power density to form a continuous gradient distribution along a predetermined spatial direction of the second surface, and introducing controllable lattice distortion by bombarding the surface layer of the second surface with plasma, thereby forming a gradient pre-compensated stress field with a continuous gradient change in stress value along the predetermined spatial direction; the formation of the gradient temperature response layer includes: simultaneously introducing a reactive gas and a metal-organic precursor, and co-depositing a negative thermal expansion material gradient layer in situ on the surface of the gradient pre-compensated stress field, thereby forming a gradient temperature response layer with a continuous gradient change in thermal expansion coefficient along the predetermined spatial direction. The generation process of the multifunctional layer is as follows: Figure 4 As shown, a gradient pre-compensated stress field is first prepared by pure physical plasma bombardment, and then a gradient temperature response layer is prepared by vapor phase co-deposition under plasma atmosphere. The two processes are completed by the same set of processing equipment and maintain the same spatial gradient arrangement direction. By relying on two different mechanisms of stress formation by physical deformation and film formation by chemical reaction, static stress cancellation and dynamic temperature deformation adaptation are respectively achieved, thus comprehensively improving the stress control system on the heat sink heat dissipation side.

[0037] The formation process of the gradient pre-compensation stress field is a purely physical plasma bombardment modification process, with only high-purity inert working gas introduced throughout. High-purity argon gas with a purity of no less than 99.999% is preferentially selected as the core working medium. Argon gas possesses extremely stable chemical properties and, under plasma ionization, will not undergo any chemical reactions such as oxidation, combination, or doping with various heat sink substrates, including copper-based, copper-molybdenum-based, diamond-based, and copper-based substrates. It can ionize in a sealed cavity to form a stable high-energy plasma atmosphere, providing a continuous and uniform energy medium for particle bombardment. It also protects the secondary substrate material of the heat sink throughout the process, solving problems such as surface oxidation and thermal conductivity degradation during processing, ensuring that the inherent thermal conductivity and structural integrity of the heat dissipation surface substrate remain unaffected. Argon atoms have a relatively large mass, and the high-energy argon ions formed after ionization possess sufficient kinetic energy to impact the substrate surface, efficiently introducing controllable lattice distortion. Furthermore, argon gas is inexpensive and readily available, making it suitable for large-scale industrial production. Besides argon, other inert gases such as helium and neon can also be selected according to actual processing requirements. Helium, with its smaller atomic mass, is suitable for precise stress control, while neon, with its lower ionization energy, can maintain a stable plasma discharge state at lower power. In actual processing, based on the non-uniform heating characteristics of high-power chips, where the temperature is concentrated at the center and decreases at the edges, a spatial gradient direction extending from the geometric center to the surrounding edges of the second surface is pre-set. By independently controlling the output power of the plasma generating array in different zones, the plasma power density is made to present a smooth, continuous, and abrupt gradient distribution along this preset direction. The level of plasma power density directly determines the kinetic energy flux and bombardment intensity of the ionized inert high-energy particles. The higher the power density, the more high-energy particles that collide with the substrate surface per unit time, resulting in stronger impact energy. When high-energy particles continuously impact the regular lattice structure of the heat sink surface, they exert directional compression and shearing effects on the lattice atoms, forcing the originally regularly arranged periodic lattice atoms to undergo slight displacement, dislocation slip, and lattice distortion, forming a controllable and uniform micro-lattice distortion that is only distributed on the surface layer, without damaging the lattice structure and mechanical properties of the deeper layers of the heat sink substrate. Because the plasma bombardment intensity varies across different regions of the second surface, the lattice distortion density and distortion degree in the corresponding regions simultaneously form an ordered gradient. The central region, with its higher bombardment intensity, exhibits a higher lattice distortion density and a larger amplitude of the reverse residual stress accumulated on the surface. Conversely, the distortion density and stress amplitude decrease synchronously in the edge regions where the bombardment intensity decreases. Ultimately, this results in a gradient pre-compensated stress field on the second surface where the stress value gradually changes along a predetermined spatial direction. The core technical feature of this stress field is that the overall stress direction is completely opposite to the thermocompression stress generated during chip operation. The thermocompression stress generated by the chip's heat generation will compress the heat sink, causing it to warp upwards. The gradient reverse stress introduced in this process can preemptively counteract and offset this thermocompression stress, achieving static stress pre-compensation at the microstructural level.

[0038] Based on the complete formation of the gradient pre-compensated stress field without moving the heat sink workpiece or altering the plasma gradient distribution parameters, the reaction gas path is switched directly within the same cavity to initiate the in-situ preparation process of the gradient temperature response layer. The single inert gas supply is stopped, and activation reaction gas and metal-organic precursor raw materials are simultaneously introduced. High-purity oxygen with a purity of no less than 99.999% is preferentially selected as the reaction gas. Oxygen can form active oxygen ions in the high-energy plasma environment, reacting with the metal components produced by the decomposition of the metal-organic precursor to generate oxide-based negative thermal expansion functional materials. Oxygen has moderate reactivity and stable reaction products, making it the optimal choice for preparing oxide negative thermal expansion materials. For nitride negative thermal expansion materials with special requirements, high-purity nitrogen can be used as the reaction gas, while high-purity methane is used for carbide negative thermal expansion materials. The metal-organic precursor is selected based on the type of the target negative thermal expansion material. When preparing the most widely used zirconium tungstate negative thermal expansion material, tetraisopropoxyzirconium and hexacarbonyltungsten are chosen as bimetallic organic precursors. Tetraisopropoxyzirconium is the core raw material providing zirconium; its low thermal decomposition temperature allows for rapid decomposition in a plasma environment to generate active zirconium components. Hexacarbonyltungsten is the core raw material providing tungsten, possessing good volatility and thermal stability, and can decompose simultaneously with tetraisopropoxyzirconium and react according to a stoichiometric ratio. For preparing other types of negative thermal expansion materials, tetraethoxytitanium can be used to prepare lead titanate negative thermal expansion materials, and niobium pentachloride can be used to prepare lithium niobate negative thermal expansion materials. This process employs an in-situ co-deposition molding method, where all functional components are directly deposited on the pre-formed gradient stress-modified surface layer. Relying on the gradient distribution characteristics of the original plasma power density, differentiated deposition molding is achieved. Regions with higher plasma power density exhibit higher precursor decomposition efficiency, greater functional component nucleation density, and higher material enrichment. The spatial law of power density gradient directly replicates the gradient law of functional material component concentration. Negative thermal expansion materials possess the unique property of shrinking in volume with increasing temperature and expanding in volume with decreasing temperature. Their core mechanism lies in the rotation and expansion / contraction of the rigid polyhedral structure within the material. When the temperature rises, the bonds between the polyhedra twist, causing the overall structure to shrink; when the temperature decreases, the bonds return to their original state, causing the overall structure to expand. The concentration of the components in a negative thermal expansion material directly determines the local coefficient of thermal expansion; the higher the component concentration, the lower the regional coefficient of thermal expansion. Based on the continuous gradient change in component concentration, a gradient temperature response layer with a continuously and smoothly varying coefficient of thermal expansion along a predetermined direction is ultimately formed on the surface of the gradient pre-compensated stress field.

[0039] The two functional layers are seamlessly integrated and complementary, overcoming the limitations of single-stress compensation. Relying solely on the gradient pre-compensation stress field formed by lattice distortion results in a static stress structure with a fixed stress amplitude. This only achieves precise stress compensation at the chip's rated operating temperature. Under complex conditions such as low-temperature start-up, high-temperature overload, and frequent temperature cycling, the accuracy of static stress compensation drops significantly, failing to meet dynamic temperature deformation requirements. In contrast, the surface gradient temperature response layer enables dynamic adaptive adjustment across the entire temperature range. It autonomously adjusts its micro-deformation state in real-time, following changes in the chip's operating temperature. At high temperatures, it compensates for the thermal expansion deformation of the heat sink substrate through its own contraction; at low temperatures, it compensates for the substrate's contraction deformation through its own slight expansion, dynamically compensating for the difference in static stress field compensation. The synergistic effect of these two layers provides dual protection through static stress balance and dynamic deformation adaptation, ensuring the heat sink maintains stress balance and morphological flatness under all temperature and power conditions. This solves the failure problems of single-sided polished heat sinks caused by non-uniform thermal stress, such as warping, interface debonding, and chip stress cracking.

[0040] In some embodiments, the second surface is divided into multiple adjacent and independently controllable processing units, and the plasma output parameters of each processing unit can be adjusted individually. During processing, the plasma electron density, electron temperature, and emission spectrum characteristics of each processing unit are collected in real time. Based on the pre-established mapping relationship between process parameters, stress field, and thermal expansion coefficient, the plasma power density and reactant gas injection rate of each unit are dynamically adjusted in a closed loop. A spatially continuous, stepless gradient plasma action field is formed on the surface of the second surface, thereby causing the stress amplitude of the gradient pre-compensated stress field and the thermal expansion coefficient amplitude of the gradient temperature response layer to change synchronously and continuously along the direction from the center to the edge of the second surface. The plasma processing operation on the second surface is carried out using a partitioned and independently controlled processing mode, breaking the limitations of traditional whole-surface uniform parameter processing. Relying on refined partitioned control combined with online monitoring and closed-loop adjustment, the gradient action effect that highly matches the chip heating law is accurately replicated, ensuring the uniformity and accuracy of the gradient distribution of the double-layer functional structure from the processing control level.

[0041] During actual processing, based on the overall dimensions and stress control design requirements of the second surface of the heat sink, the complete heat dissipation surface is divided into multiple independent processing units that are adjacent to each other and have no gaps between them. The entire plasma generation control system is equipped with a dedicated independent control path for each processing unit, so that the core operating parameters such as plasma output power and gas supply rate of each unit can be changed and adjusted independently without the overall setting. This partitioned design can adapt to heat sink workpieces with conventional regular shapes, as well as meet the differentiated processing requirements of irregular heat dissipation surfaces, and flexibly adapt to the diverse stress distribution design schemes corresponding to chips with different power levels.

[0042] Throughout the entire plasma bombardment and deposition process, the system's online monitoring module continuously collects status data from each independent processing unit. It focuses on acquiring three core characteristic parameters in real time: plasma electron density, electron temperature, and emission spectrum. Electron density directly reflects the total amount of high-energy particles in the region, directly determining the degree of surface lattice distortion. Electron temperature represents the energy level of the plasma, affecting both the formation efficiency of the stress field and the decomposition reaction rate of the precursor material. The emission spectrum characteristic parameter can determine the fullness of the reaction in the cavity and the composition of the deposition components in real time. This multi-dimensional data acquisition allows for a complete understanding of the actual processing conditions in each region, resolving the discrepancy between preset parameters and actual effects.

[0043] The equipment has established a comprehensive mapping relationship based on a large amount of experimental data in advance, clearly defining the stress field range and thermal expansion coefficient variation range that different plasma operating parameters can ultimately form. During the processing, various monitoring data collected in real time are synchronously imported into the mapping system, automatically comparing the difference between the current forming state and the preset target state. Based on this, a closed-loop dynamic adjustment operation is completed, specifically increasing or decreasing the plasma power density of the corresponding processing unit, while simultaneously fine-tuning the amount of reactant gas introduced, so that the operating state of each unit is close to the preset standard. The parameter correction is completed automatically throughout the process without the need for manual on-site intervention.

[0044] By employing a processing method that combines independent regional control with real-time closed-loop correction, the parameter abruptness issues that were originally prone to occur between different units are eliminated. Ultimately, a gradient plasma field with a smooth and continuous overall flow without significant numerical differences or hierarchical steps is constructed on the entire second surface. Under the continuous action of this uniformly gradient field, the stress amplitude inside the gradient pre-compensated stress field obtained by subsequent molding, as well as the thermal expansion coefficient amplitude corresponding to the gradient temperature response layer, will uniformly and continuously change in an orderly manner from the center of the heat sink to the outer edge. The gradient change rhythm of the two functional structures is completely consistent, precisely matching the actual working condition of concentrated thermal stress in the central area and gradually weakening thermal stress in the edge area during chip operation. This allows the stress compensation capability and temperature deformation adaptation capability of the heat sink heat dissipation side to achieve uniform matching across the entire domain, further enhancing the morphological stability and stress balance effect of the heat sink under long-term temperature cycling conditions.

[0045] In some embodiments, the metastable stress pre-release treatment includes: placing the heat sink in an inert protective atmosphere or vacuum environment, applying a uniform and stable thermal field to activate atomic diffusion and dislocation motion on the surface of the heat sink, accelerating the relaxation process of the metastable lattice distortion formed after plasma bombardment, and enabling the residual stress field to reach a thermodynamically stable state before leaving the factory. The metastable stress pre-release treatment is a pre-stabilization process designed for unstable lattice structures remaining on the surface after plasma processing. It can artificially accelerate the natural dissipation of unstable stress and complete the stress state stabilization in advance.

[0046] During this process, the heat sink with the completed plasma structure is first placed in a sealed processing chamber. An inert protective atmosphere or a high vacuum environment is selected as the working environment. The use of inert gas to isolate oxygen and water vapor in the outside air can prevent oxidation and deterioration of the gradient temperature response layer and the distorted lattice region on the second surface of the heat sink. It can also prevent changes in the composition of the doped layer on the functional surface. The use of a vacuum environment can eliminate the slight differences in heat conduction caused by the gas medium, making the overall environment of the heat sink purer and more uniform. Both environments can protect the various functional surface structures formed in the early stage throughout the process, and will not cause structural damage or performance degradation during the stress release stage.

[0047] Once the environmental conditions are set up, a uniformly distributed and stable thermal field is continuously output to the cavity, eliminating the need for localized, fixed-point heating. This ensures that all areas of the heat sink are heated to a completely consistent degree, preventing the generation of new temporary thermal stresses due to uneven heating and avoiding the superposition of old and new stresses that could disrupt the established gradient stress distribution. The stable and continuous thermal energy provides sufficient kinetic energy for the microscopic particles within the heat sink surface, successfully activating the free diffusion of atoms within the surface area, while simultaneously driving the previously stagnant dislocation structures within the crystal lattice to undergo ordered slip and regularization.

[0048] After being bombarded by high-energy plasma particles, a large number of lattice atoms on the surface of the heat sink deviate from their original standard arrangement, forming a large-scale metastable lattice distortion. This type of distorted structure has a spontaneous tendency to transform into a regular lattice shape, but the transformation rate is very slow under normal temperature conditions, resulting in a slow change in residual stress over a long period of time. With the energy provided by a constant thermal field, the relaxation transition rate of this metastable lattice distortion can be significantly accelerated, allowing the displaced atoms to gradually return to their standard lattice positions, and the randomly distributed dislocation structure to gradually achieve a regular arrangement, rapidly completing the self-adjustment of the unstable microstructure.

[0049] After a set period of thermal insulation, all unstable residual stresses introduced by plasma processing are released and properly arranged in advance. The overall residual stress field inside the heat sink escapes its easily fluctuating metastable state and directly reaches a stable state of thermodynamic equilibrium. After this process, the heat sink will not exhibit changes in flatness or slight warping caused by spontaneous stress relaxation during subsequent storage, transportation, and initial assembly and use. This effectively ensures the consistency of the dimensions and stress state of batch-processed products. Furthermore, the gentle heat treatment throughout the process does not alter the overall distribution of the gradient pre-compensated stress field, nor does it damage the gradual thermal expansion characteristics of the gradient temperature response layer, thus fully preserving the performance characteristics of all previous stress control and thermal adaptation structures.

[0050] In some embodiments, the stress field locking process includes: simultaneously applying a uniform thermal field and an auxiliary physical field to the heat sink in a vacuum environment; through the synergistic effect of thermal activation and the auxiliary physical field, inducing ordered rearrangement of atoms on the surface of the heat sink, stabilizing the lattice distortion structure corresponding to the gradient pre-compensation stress field, and simultaneously strengthening the interfacial bonding between the gradient temperature response layer and the substrate, thereby solidifying the stress distribution state of the gradient pre-compensation stress field and the gradient temperature response layer. The synergistic effect of multiple physical fields completes the microstructure shaping and interlayer bonding strengthening, locking the predetermined stress arrangement pattern of the two types of gradient functional structures constructed in the previous stage.

[0051] During the lock-up process, the heat sink is placed entirely within a sealed vacuum chamber. The vacuum environment isolates the heat sink from oxygen, moisture, and various impurities, preventing oxidation and performance degradation of the gradient temperature response layer. It also avoids interference from external media on the physical field's effectiveness and eliminates temperature unevenness caused by convective heat transfer under normal pressure, ensuring highly uniform and stable overall processing conditions. Simultaneously, a uniformly distributed thermal field and a directional auxiliary physical field are applied to the entire heat sink within the vacuum environment. Both fields act simultaneously and completely cover the entire heat sink area, preventing any localized unevenness in intensity.

[0052] A uniform thermal field primarily serves as the foundation for thermal activation. By relying on stable and continuous thermal energy, it reduces the resistance to the movement of various atoms on the heat sink surface, allowing the previously relatively stable surface atoms to regain sufficient kinetic energy for orderly recombination. This provides the necessary energy conditions for the orderly rearrangement of atoms. The uniform heating throughout the process does not generate new additional thermal stress, thus perfectly preserving the previously adjusted overall stress balance. The auxiliary physical field can provide directional guidance and constraint on the atomic motion trend, working efficiently with the thermal field to change the disordered motion of atoms under natural conditions. It precisely induces various atoms within the heat sink surface area to complete an orderly rearrangement according to a predetermined pattern.

[0053] Under the synergistic drive of the dual action fields, the lattice distortion structure within the corresponding gradient pre-compensation stress field region is further stabilized and fixed. The lattice offset morphology, which has been straightened and regulated by the previous pre-release process, is fixed in this process. The misalignment angle and arrangement position of the lattice atoms no longer have room for autonomous change, firmly maintaining the orderly and gradual stress amplitude distribution from the center to the edge, thus solving the problem of stress reduction caused by slow lattice reset under long-term operating conditions. At the same time, the activation effect generated by the action field can also act on the interface between the gradient temperature response layer and the heat sink substrate, promoting the mutual penetration and fusion of the substrate surface atoms and the deposited functional layer atoms, filling the tiny gaps and weak bonding areas at the interface, effectively strengthening the interfacial bonding strength between the two layers, and effectively avoiding undesirable phenomena such as interlayer separation and peeling during repeated temperature changes.

[0054] After the entire locking process is completed, the stress distribution of the gradient pre-compensation stress field and the matching stress adaptation state formed by the gradient temperature response layer based on its own thermal expansion characteristics are all solidified and finalized. The stress synergistic compensation effect originally possessed by the two types of structures will not gradually weaken with the increase of usage time. This allows the processed single-sided polished heat sink to maintain a stable stress balance and flat shape in continuous power switching working scenarios in high and low temperature alternating environments. This greatly improves the service stability and overall service life of the heat sink product and also ensures that the stress control effect of products processed in the same batch remains highly consistent.

[0055] A stress-controlled processing system for single-sided polishing heat sinks, the system modules are shown in the attached figure. Figure 5 As shown, the system includes: The heat sink preparation unit 1 is used to acquire the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink to be processed, and determine the polishing scheme according to the material characteristics of the first surface and the second surface. Polishing unit 2 is used to first polish the first surface based on the polishing scheme, then perform gradient ion implantation on the polished first surface to form a doped layer with a continuously gradient change in thermal expansion coefficient, thus obtaining a functional surface, and then seal and protect the functional surface. The stress compensation unit 3 is used to bombard the second surface with gradient plasma based on the polishing scheme to form a gradient pre-compensation stress field and a gradient temperature response layer on the surface of the second surface, so as to jointly compensate for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite to the direction of the chip thermal stress. The first stabilizing unit 4 is used to remove the sealing protection of the functional surface and to perform metastable stress pre-release treatment on the heat sink to be processed, thereby accelerating the natural relaxation process of the residual stress introduced after plasma bombardment. The second stabilizing unit 5 is used to perform stress field locking treatment on the heat sink that has completed metastable stress pre-release, solidify the stress distribution state of the gradient pre-compensation stress field and gradient temperature response layer, and obtain a single-sided polished heat sink.

[0056] Those skilled in the art will understand that the components of this application described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage system for execution by the computing system. Alternatively, they can be fabricated as separate integrated circuit components, or multiple components or steps can be fabricated as a single integrated circuit component. Thus, this application is not limited to any particular combination of hardware and software.

[0057] Note that the above description is merely a preferred embodiment and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this application, and the scope of this application is determined by the scope of the appended claims.

[0058] The above disclosures are only a few specific implementation scenarios of this application. However, this application is not limited to these. Any variations that can be conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A stress-controlled processing method for single-sided polishing heat sinks, characterized in that, include: Obtain the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink, and determine the polishing scheme based on the material characteristics of the first surface and the second surface. First, the first surface is polished according to the polishing scheme. Then, gradient ion implantation is performed on the polished first surface to form a doped layer with a continuously varying coefficient of thermal expansion, thus obtaining a functional surface. The functional surface is then sealed and protected. The polishing scheme employs gradient plasma bombardment of the second surface to form a gradient pre-compensation stress field and a gradient temperature response layer on the second surface, thereby synergistically compensating for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite in direction to the thermal stress of the chip. Remove the sealing protection of the functional surface and perform metastable stress pre-release treatment on the heat sink to be processed to accelerate the natural relaxation process of the residual stress introduced after plasma bombardment. The heat sink that has completed metastable stress pre-release is subjected to stress field locking treatment to solidify the stress distribution state of the gradient pre-compensated stress field and gradient temperature response layer, resulting in a single-sided polished heat sink.

2. The stress-regulating processing method according to claim 1, characterized in that, The material characteristics include hardness, thermal conductivity, elastic modulus, and processing damage characteristics. Based on the material characteristics, at least the abrasive particle size, polishing pressure, polishing disc rotation speed, and polishing time of each stage of the first surface are determined. The magnitude and distribution of residual stress introduced after polishing the first surface are predicted. Combined with the plasma etching rate of the heat sink, the plasma power gradient distribution, gas flow ratio, and processing time of each stage of the second surface are determined.

3. The stress-regulating processing method according to claim 1, characterized in that, Polishing the first surface specifically includes: The first surface is semi-polished using an abrasive of the first grit size to completely remove the damaged layer generated by roughing; the first surface is fine-polished using an abrasive of the second grit size to achieve a first predetermined flatness value; the first surface is finally polished using an abrasive of the third grit size to achieve a second predetermined surface roughness value, and a residual compressive stress field is introduced into the surface layer of the first surface by controlling the polishing parameters; wherein, the first grit size is larger than the second grit size, and the second grit size is larger than the third grit size.

4. The stress-regulating processing method according to claim 1, characterized in that, The doped layer includes a first doped layer and a second doped layer whose doping concentration changes continuously along the depth direction; First metal ions are injected separately into the polished first surface. During the injection process, the ion implantation energy is continuously and linearly decreased while the ion implantation dose is continuously and linearly increased to form a first doped layer on the surface of the first surface. Based on the first doped layer, the first metal ions and the second non-metal ions are implanted. The implantation energy of both ions decreases linearly over time, and the implantation dose ratio of the two remains constant, forming a second doped layer in the shallow part of the functional surface.

5. The stress-regulating processing method according to claim 4, characterized in that, The first metal ion is an aluminum ion, used to regulate the low to medium range of the thermal expansion coefficient of the doped layer; the second non-metal ion is a silicon ion, used to regulate the low range of the thermal expansion coefficient of the doped layer; the aluminum ion and the silicon ion work together to achieve a wide range of continuously adjustable thermal expansion coefficients.

6. The stress-regulating processing method according to claim 1, characterized in that, The formation of the gradient pre-compensation stress field includes: introducing an inert working gas, adjusting the plasma power density to be continuously gradient distributed along a predetermined spatial direction of the second surface, and introducing controllable lattice distortion by bombarding the surface layer of the second surface with plasma to form a gradient pre-compensation stress field in which the stress value continuously changes along the predetermined spatial direction. The formation of the gradient temperature response layer includes: simultaneously introducing a reactive gas and a metal-organic precursor, and co-depositing a negative thermal expansion material gradient layer in situ on the surface of the gradient pre-compensated stress field to form a gradient temperature response layer with a continuously varying thermal expansion coefficient along the predetermined spatial direction.

7. The stress-regulating processing method according to claim 6, characterized in that, The second surface is divided into multiple adjacent and independently controllable processing units, and the plasma output parameters of each processing unit can be adjusted individually. During the process, the plasma electron density, electron temperature and emission spectrum characteristics of each processing unit are collected in real time. Based on the pre-established mapping relationship between process parameters and stress field and thermal expansion coefficient, the plasma power density and reactant gas injection rate of each unit are dynamically adjusted in a closed loop. A spatially continuous, stepless gradient plasma field is formed on the surface of the second surface, which causes the stress amplitude of the gradient pre-compensated stress field and the thermal expansion coefficient amplitude of the gradient temperature response layer to change synchronously and continuously along the direction from the center to the edge of the second surface.

8. The stress-regulating processing method according to claim 1, characterized in that, Metastable stress pre-release treatment includes: placing the heat sink in an inert protective atmosphere or vacuum environment, applying a uniform and stable thermal field to activate atomic diffusion and dislocation movement on the surface of the heat sink, accelerating the relaxation process of the metastable lattice distortion formed after plasma bombardment, so that the residual stress field reaches a thermodynamically stable state before leaving the factory.

9. The stress-regulating processing method according to claim 1, characterized in that, The stress field locking process includes: simultaneously applying a uniform thermal field and an auxiliary physical field to the heat sink in a vacuum environment; through the synergistic effect of thermal activation and the auxiliary physical field, inducing orderly rearrangement of atoms on the surface of the heat sink, stabilizing the lattice distortion structure corresponding to the gradient pre-compensation stress field, and strengthening the interfacial bonding between the gradient temperature response layer and the substrate, thereby solidifying the stress distribution state of the gradient pre-compensation stress field and the gradient temperature response layer.

10. A stress-regulating processing system for a single-sided polishing heat sink, characterized in that, include: The heat sink preparation unit is used to acquire the heat sink to be processed, determine the first surface to be polished and the second surface for heat dissipation on the heat sink to be processed, and determine the polishing scheme according to the material characteristics of the first surface and the second surface. The polishing processing unit is used to first polish the first surface based on the polishing scheme, then perform gradient ion implantation on the polished first surface to form a doped layer with a continuously varying coefficient of thermal expansion, thereby obtaining a functional surface, and then seal and protect the functional surface. The stress compensation unit is used to bombard the second surface with gradient plasma based on the polishing scheme to form a gradient pre-compensation stress field and a gradient temperature response layer on the surface of the second surface, so as to synergistically compensate for the thermal stress generated during chip operation; the introduced stress of the gradient pre-compensation stress field is opposite to the direction of the chip thermal stress. The first stabilizing unit is used to remove the sealing protection of the functional surface and to perform metastable stress pre-release treatment on the heat sink to be processed, thereby accelerating the natural relaxation process of the residual stress introduced after plasma bombardment. The second stabilizing unit is used to perform stress field locking treatment on the heat sink that has completed metastable stress pre-release, solidify the stress distribution state of the gradient pre-compensated stress field and the gradient temperature response layer, and obtain a single-sided polished heat sink.

Images