An embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips and its fabrication method
By embedding a unidirectional self-driven microfluidic heat dissipation chip within the satellite computing chip, and utilizing capillary force and wettability gradient to drive the circulation of the cooling working fluid, the problems of insufficient heat dissipation reliability and heat dissipation capacity of the satellite computing chip are solved. This achieves efficient phase change heat dissipation and flow stability, making it suitable for lightweight design of satellite computing chips.
Patent Information
- Application Number
- CN202611291022.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-25
- Publication Date
- 2026-11-17
AI Technical Summary
Existing heat dissipation solutions for satellite computing chips have reliability issues and insufficient heat dissipation capacity. In particular, traditional air cooling and liquid cooling solutions are unable to meet the high heat flux density requirements in microgravity environments.
An embedded unidirectional self-driven microfluidic heat dissipation chip is designed. By integrating a microfluidic heat dissipation structure layer in a semiconductor chip substrate, the self-driving force of the cooling working fluid generated by capillary force and wettability gradient is utilized to achieve spontaneous circulation. Combined with the phase change heat dissipation zone and vapor conduction channel, a closed self-driven phase change heat dissipation circuit is formed.
It achieves chip-level phase change heat dissipation without external pumping power, improves heat dissipation capacity, reduces package thermal resistance, and enhances flow stability and heat dissipation efficiency, making it suitable for lightweight design of satellite computing chips.
Smart Images

Figure CN123124813A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of thermal management technology for aerospace electronic devices, specifically to an embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips and its fabrication method. Background Technology
[0002] With the rapid development of high-performance computing chips for spaceborne applications, chip power consumption density has increased dramatically. However, satellites operate in a vacuum and microgravity environment, rendering traditional air cooling methods ineffective. Existing heat dissipation solutions face two fundamental contradictions: Firstly, although the pump-driven liquid cooling solution has strong heat dissipation capabilities, it relies on external micro-pumps and other moving parts, which poses a risk to long-term reliability in space, and the pump power consumption is in conflict with the satellite's power consumption budget. Secondly, while passive heat pipe solutions have no moving parts and high reliability, their heat transfer power is limited and they are unable to meet the demands of continuously increasing local heat flux density. Furthermore, traditional pulsating heat pipes are at risk of burning out in the evaporation section under microgravity conditions, resulting in severe performance degradation.
[0003] The shortcomings of the above two types of solutions constitute the core technical bottleneck of current satellite computing chip thermal management, and there is an urgent need for a new chip-level heat dissipation solution that combines high heat dissipation capacity with the characteristics of no moving parts. Summary of the Invention
[0004] To address the shortcomings of existing technologies, such as pump-driven liquid cooling solutions relying on external moving parts leading to reliability issues, and passive heat pipe solutions lacking sufficient heat dissipation capacity to meet the demands of high heat flux density chips, this application provides an embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing chips and its fabrication method. This chip integrates a microfluidic heat dissipation structure layer with a unidirectional wettability gradient within a semiconductor chip substrate. It utilizes the directional driving force generated by capillary force and the wettability gradient to collaboratively drive the unidirectional spontaneous circulation of the cooling medium, achieving chip-level phase change heat dissipation without external pumping power, fundamentally eliminating the reliability risks associated with moving parts.
[0005] Specifically, this application provides the following technical solutions: This application provides an embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips, comprising: A semiconductor chip substrate includes a computing circuit layer and a microfluidic heat dissipation structure layer; wherein, the microfluidic heat dissipation structure layer includes several channels, a capillary driving region disposed at the inlet end of the channels, a phase change heat transfer region disposed in the downstream region of the channels, and a vapor conduction channel communicating with the channels. A sealing cover plate is hermetically bonded to the semiconductor chip substrate to form a closed flow channel; An external condensation radiation unit is connected to the steam channel and the capillary drive zone inlet via a pipeline, forming a self-driven phase change heat dissipation circuit.
[0006] Optionally, the flow channels are arranged in an array, and the contact angle of the surface of the flow channels changes continuously and gradually along the flow direction of the working fluid.
[0007] Optionally, the first end contact angle of the flow channel corresponds to a hydrophobic state, and the second end contact angle corresponds to a superhydrophilic state.
[0008] Optionally, the capillary drive region is provided with a micro-nano porous structure, which is at least one of a porous film, a porous silicon layer, or a porous capillary core.
[0009] Optionally, the micropillar array can be any one of a cylindrical, square, or conical structure.
[0010] This application also provides a method for fabricating an embedded unidirectional self-driven microfluidic heat dissipation chip, comprising: Several flow channels are processed on the back or inside of a semiconductor chip substrate; Micro-nano porous structures are fabricated at the inlet end of the flow channel to form a capillary driving region; and a micro-pillar array is fabricated in the downstream region of the flow channel to form a phase change heat transfer region. The inner wall of the flow channel is surface treated along its length to continuously increase the wettability of the inner wall of the flow channel along the flow direction of the working fluid. The sealing cover is hermetically bonded to the semiconductor chip substrate to form a flow channel; An external condensation radiation unit is connected, and the steam channel is connected to the inlet of the capillary drive area through a pipeline to form a self-driven phase change heat dissipation circuit.
[0011] Optionally, the flow channel is fabricated using a deep reactive ion etching process, and the micropillar array is fabricated in the downstream region of the flow channel using a secondary deep reactive ion etching process.
[0012] Optionally, an electrochemical etching process is used to form a porous silicon layer at the inlet end of the flow channel as the capillary driving region, wherein the pore density of the porous silicon layer is greater on the side closer to the phase change heat transfer zone than on the side farther away from the phase change heat transfer zone.
[0013] Optionally, a cooling medium is injected into the closed flow channel, and the injection port is sealed after the injection is completed.
[0014] Optionally, the cooling medium is at least one of water, ammonia, ethanol, or a mixture thereof.
[0015] Compared with the prior art, this application has the following beneficial effects: 1. The unidirectional wettability gradient and the directional driving force generated by the capillary porous structure drive the spontaneous circulation of the working fluid, eliminating dependence on external pump power and the risk of failure of moving parts.
[0016] 2. The micro-pillar array in the phase change heat transfer zone increases the gas-liquid interface area, enhances thin-film evaporation heat transfer, and efficiently extracts heat by utilizing the latent heat of phase change of the working fluid, thereby significantly improving the effective heat dissipation capacity.
[0017] 3. The directional driving force generated by the unidirectional wettability gradient can effectively suppress the backflow of the working fluid, avoid the drastic temperature fluctuations caused by disordered gas-liquid flow in traditional two-phase cooling, and improve flow stability.
[0018] 4. The heat dissipation structure is directly integrated into the back or inside of the semiconductor chip substrate, reducing the thermal resistance of the package and facilitating the lightweight design of satellite payloads. Attached Figure Description
[0019] Figure 1 This is a schematic cross-sectional view of the chip heat dissipation structure provided in an embodiment of this application.
[0020] Figure 2 This is a schematic diagram of the closed self-driven phase change heat dissipation cycle principle provided in the embodiments of this application.
[0021] Figure 3 This is a schematic diagram of the unidirectional wettability gradient of the inner wall of the flow channel provided in an embodiment of this application.
[0022] In the figure, 1 is the semiconductor chip substrate; 22 is the capillary drive area; 23 is the phase change heat transfer area; 24 is the vapor conduction channel; 25 is the airtight sealing area; and 3 is the sealing cover. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0024] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0025] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments.
[0026] Example 1 See Figure 1The embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing chips provided in this application includes a semiconductor chip substrate 1, a microfluidic heat dissipation structure layer, a sealing cover plate 3, and an external condensation radiation unit. A computing circuit layer is disposed on the front side of the semiconductor chip substrate 1, and the microfluidic heat dissipation structure layer is integrated on the back side or inside of the semiconductor chip substrate 1, forming an integrated structure with the computing circuit layer. This allows the heat dissipation path to be directly coupled to the heat source, effectively reducing the packaging thermal resistance.
[0027] The thermal path model of the heat dissipation chip is as follows: From the computing circuit layer (i.e., the heat source) to the external condensation radiation unit (i.e., the final heat dissipation surface), the heat passes through the following thermal resistance links in sequence: (1) Thermal resistance of silicon substrate It depends on the thickness of the silicon material between the front computing circuit layer and the back microfluidic structure layer of the semiconductor chip substrate 1. With thermal conductivity (Approximately 150 watts per meter Kelvin); (2) Phase change thermal resistance The thermal resistance is determined by the thin film evaporation or nucleation boiling heat transfer on the surface of the micropillar array in the phase change heat zone 23, and is the core thermal resistance link in the entire thermal circuit. (3) Steam transport thermal resistance The pressure drop loss, calculated by the steam transport in the steam channel and pipeline, is usually very small and can be approximated as negligible. (4) Condensation thermal resistance The thermal resistance is determined by the heat released during condensation of steam in the external condensation and radiation unit. Of the four thermal resistances, and This is a major contribution; by directly integrating the microfluidic heat dissipation structure layer onto the back side of the semiconductor chip substrate 1 (instead of the traditional solution of using an external heat sink), the interface thermal resistance between the chip and the heat sink (typically...) can be eliminated. to This reduces the total package thermal resistance by 30% to 50% compared to traditional external heat dissipation solutions, effectively improving the chip junction temperature margin.
[0028] See Figure 1 The microfluidic heat dissipation structure layer includes several flow channels, a capillary drive region 22, a phase change heat transfer region 23, and a vapor channel 24. Each functional region is arranged sequentially within the flow channels along the working fluid flow direction. The flow channels are arranged in an array, and the inner walls of the flow channels have a unidirectional wettability gradient with continuously increasing wettability along the working fluid flow direction. Under the action of this unidirectional wettability gradient, the resultant capillary force on the liquid cooling working fluid always points towards the outlet end, forming a continuous directional driving force.
[0029] The flow channels are arranged in a parallel array on the back side of the semiconductor chip substrate 1, and the length direction of the flow channels is aligned with the main heat-generating area of the computing circuit layer, so that each flow channel can cover the heat-generating area and achieve effective cooling of local hot spots.
[0030] For example, for a spaceborne computing chip with an area of 10 mm by 10 mm, the channel array includes 20 to 50 parallel channels, each channel having a width of 100 micrometers and a depth of 200 micrometers, with an adjacent channel spacing of 100 micrometers. The total coverage area of the channel array is not less than 80% of the area of the heat-generating region of the computing circuit layer, to ensure that each region of the chip is uniformly cooled.
[0031] In computing chip applications with multiple dispersed hotspots (local heat flux density more than twice the average heat flux density), localized densification design can be implemented for the flow channels corresponding to the hotspot areas: The channel spacing at the hot spot location is reduced to 50 micrometers, and the spacing of the micro-pillar array in the same region is reduced to 50 to 80 micrometers to increase the gas-liquid interface area in the hot spot region, improve local heat dissipation capacity, and achieve thermal uniformity control under non-uniform heat flux density distribution conditions.
[0032] The inlet end of the flow channel is provided with a flow equalization structure (an inlet flow distribution cavity with a width of 500 micrometers to 1000 micrometers). The flow equalization structure ensures that the liquid cooling working fluid is evenly distributed to each parallel flow channel after flowing back from the external condensation radiation unit, avoiding the problem of insufficient liquid supply to some flow channels due to uneven distribution of the working fluid.
[0033] The capillary drive region 22 is located at the inlet end of the flow channel. The capillary drive region 22 contains a micro / nano porous structure with a high specific surface area, which can generate a large capillary pressure difference, providing driving force for the intake and unidirectional transport of the cooling working fluid. The phase change heat transfer region 23 is located in the downstream region of the flow channel. The phase change heat transfer region 23 contains a micropillar array. The micropillar array expands the gas-liquid interface area, promoting the formation of a thin liquid film in the cooling working fluid in the high heat flux density region and facilitating efficient evaporation or boiling phase change. The latent heat of vaporization of the working fluid is used to remove the large amount of heat generated by the chip.
[0034] The heat transfer mechanism of the phase change heat transfer zone 23 is as follows: Under low heat flux density (30% to 50% below the critical heat flux density), the liquid cooling medium forms a continuous thin liquid film on the surface of the micropillars of the micropillar array. The main heat transfer mode is thin film evaporation, that is, the liquid film directly vaporizes at the evaporation phase interface. The heat transfer coefficient of the thin film evaporation is much higher than that of conventional convection heat transfer. Under high heat flux density (50% higher than the critical heat flux density), the liquid cooling medium forms vapor nuclei at the gas-liquid interface at the root of the micro-column, entering the nucleus boiling heat exchange mode. The vapor bubbles generated by the nucleus boiling carry away a large amount of latent heat of vaporization when they leave the surface of the micro-column, further enhancing the heat exchange efficiency.
[0035] The structural design of the micropillar array enables the liquid cooling medium to continuously replenish the surface of the micropillars. Because the capillary gaps between the micropillars are continuously supplied with liquid medium by the capillary drive region 22, the liquid film on the surface of the micropillars is prevented from drying out. Vapor bubbles escape from the gaps between the micropillars along the direction of the vapor channel 24, without blocking the liquid supply channel to the surface of the micropillars. This allows the phase change heat transfer zone 23 to maintain efficient phase change heat transfer over a wide range of heat flux densities, avoiding film boiling (i.e., "burn-out") caused by vapor film coverage in traditional boiling heat dissipation.
[0036] The steam channel 24 is connected to the flow channel 21 and is used to unidirectionally export the steam generated by the phase change heat zone 23 to the outside of the chip.
[0037] Meanwhile, the cross-sectional area of the steam channel 24 is greater than the sum of the cross-sectional areas of each flow channel in the phase change heat zone 23, so as to ensure that steam can be smoothly collected and discharged from the phase change heat zone 23, and to avoid steam accumulating in the flow channel to form a steam plug and block the supply of liquid working fluid.
[0038] For example, in a microfluidic heat dissipation structure layer with 20 parallel flow channels, each with a cross-sectional area of 100 micrometers by 200 micrometers, the cross-sectional area of the steam conduction channel 24 is designed to be 0.5 square millimeters to 2 square millimeters to meet steam conduction requirements. The inner wall of the steam conduction channel 24 is hydrophobically treated (contact angle greater than...). This is to prevent the liquid cooling medium from condensing on the inner wall of the steam channel and forming a liquid film that blocks the steam flow channel. The connection between the steam guiding channel 24 and the phase change heat transfer zone 23 is located in the top region of the phase change heat transfer zone 23. Utilizing the characteristic that the density of steam is lower than that of the liquid working medium, surface tension in a microgravity environment causes steam to preferentially migrate towards the steam guiding channel 24, achieving natural gas-liquid separation.
[0039] The sealing cover plate 3 is hermetically bonded to the semiconductor chip substrate 1 to form a closed flow channel. The external condensation radiation unit is connected to the vapor channel 24 and the inlet of the capillary drive region 22 through a pipeline, and together with the microfluidic heat dissipation structure layer, it forms a closed self-driven phase change heat dissipation circuit.
[0040] During chip operation, the liquid cooling medium flows unidirectionally from the inlet into the phase change heat transfer zone 23 under the combined action of capillary force generated by the micro-nano porous structure of the capillary drive zone 22 and directional driving force generated by the wettability gradient of the inner wall of the flow channel 21. A thin liquid film forms on the surface of the micropillar array, and after absorbing heat from the chip, the cooling medium undergoes a phase change, transforming into vapor. The vapor flows out of the chip through the vapor channel 24 and enters the external condensation radiation unit, where it condenses into liquid at the radiant heat sink. The condensed liquid medium then flows back to the inlet of the capillary drive zone 22 along the pipeline due to capillary force and wettability gradient force, completing a unidirectional, pump-free, self-driven circulation. (See working principle for details.) Figure 2 .
[0041] The external condensation radiation unit includes a radiant heat sink and a capillary return core. The radiant heat sink is made of alumina ceramic or carbon fiber reinforced aluminum matrix composite material with high emissivity (emissivity...). (Not less than 0.85), dissipating heat to the outer space environment through thermal radiation. The capillary return core is embedded inside the radiative heat dissipation plate and is used to transport the condensed liquid cooling medium from the condensation zone to the return pipeline outlet through capillary force, ensuring that the liquid working medium can continuously and stably flow back to the inlet of the capillary drive zone 22 and maintain the working medium circulation of the closed loop.
[0042] The external condensation radiation unit is connected to the microfluidic heat dissipation structure layer via a flexible metal braided hose. This flexible metal braided hose has excellent vibration resistance, capable of withstanding vibration and impact loads during rocket launch, ensuring the reliability of the pipeline connection. The area of the radiating heat sink is designed based on the satellite platform's heat dissipation surface resources and the chip's maximum heat dissipation (for example, for a spaceborne computing chip with a heat dissipation of 10 watts to 50 watts, the area of the radiating heat sink is 0.05 square meters to 0.5 square meters) to meet the heat dissipation requirements under the most severe thermal conditions in orbit.
[0043] See Figure 3 The contact angle of the unidirectional wettability gradient continuously decreases from the inlet end (i.e., the first end) to the outlet end (i.e., the second end) along the flow direction of the working fluid. The contact angle at the first end corresponds to the hydrophobic state, and the contact angle at the second end corresponds to the superhydrophilic state. This continuous contact angle gradient forms a continuous capillary pressure gradient on the inner wall of the flow channel, causing the liquid working fluid to be subjected to a net driving force pointing towards the hydrophilic end (outlet end), spontaneously flowing towards the phase change heat transfer zone without any external pump input.
[0044] The physical mechanism by which the unidirectional wettability gradient generates a directional driving force is as follows: According to the Young-Laplace equation, the pressure difference across a curved liquid surface (i.e., capillary pressure difference) With the radius of curvature of the liquid surface and liquid surface tension The relationship is .
[0045] Contact angle of the inner wall of the flow channel With half width of the flow channel The geometric relationship between them, the capillary radius can be approximated as: Therefore, the capillary pressure difference is expressed as .
[0046] Due to the unidirectional wettability gradient, the contact angle Along the length of the flow channel from the inlet end (i.e., the first end), to The number of particles continuously decreases until it reaches the outlet end (i.e., the second end). The capillary pressure difference increases monotonically along the flow channel direction, thereby generating a net capillary driving force pointing towards the outlet end. The net capillary driving force drives the liquid cooling working fluid to flow unidirectionally along the gradient direction.
[0047] The design direction of the unidirectional wettability gradient (hydrophobic at the inlet and superhydrophilic at the outlet) ensures that the working fluid is always transported unidirectionally towards the phase change heat transfer zone under the combined effect of capillary forces, and will not flow back under the action of gradient forces. If the gradient direction is reversed (hydrophilic at the inlet and hydrophobic at the outlet), the gradient force will drive the working fluid away from the phase change heat transfer zone, which is contrary to the heat dissipation objective.
[0048] The effective range of the net capillary driving force is constrained by the channel length, for example, for a channel length of 5 mm to 20 mm and a half-width... For flow channels ranging from 25 micrometers to 100 micrometers, when water is used as the working medium ( The net capillary driving force It can reach tens to hundreds of Pa, which is much greater than the viscous flow resistance of the flow channel, and can meet the driving requirements of steady-state working fluid circulation.
[0049] The micropillar array of the phase change heat transfer zone 23 can be a cylindrical, square, or conical structure with a height of 50 micrometers to 500 micrometers and a spacing of 50 micrometers to 200 micrometers. This parameter range achieves a balance between increasing the gas-liquid interface area and reducing flow resistance, which helps to expand the gas-liquid interface and improve the liquid film evaporation efficiency.
[0050] The selection of the geometric parameters of the micropillar array follows the following principles: the height of the micropillar determines the expansion factor of the gas-liquid interface per unit base area. The greater the height, the larger the effective heat exchange area. However, excessively tall micropillars will increase the flow resistance of the working fluid in the gap between the micropillars and cause the capillary radius to increase. At the same time, the spacing between the micropillars determines the capillary pore size. The smaller the spacing, the greater the capillary suction force. However, excessively small spacing will make it difficult for steam to escape, resulting in increased vapor resistance in the gap between the micropillars.
[0051] For example, in the case of water as the working fluid and heat flux density of 20 W / cm² to 100 W / cm², the height of the microcolumn is preferably 100 μm to 300 μm, and the spacing between the microcolumns is preferably 80 μm to 150 μm. This combination of parameters can achieve a better balance between the expansion of the gas-liquid interface area and the resistance to vapor conduction.
[0052] The cross-sectional shape of the micro-pillars has little effect on capillary force. The capillary driving performance of cylindrical, square, and conical columns is similar under the same height and spacing conditions. The cone angle design of the conical column helps the vapor bubbles to detach from the column root, which can further reduce the initial superheat of nucleate boiling.
[0053] This embodiment provides a method for fabricating a chip-embedded unidirectional self-driven microfluidic heat dissipation chip, specifically including the following steps: S100, which processes several flow channels on the back side or inside of a semiconductor chip substrate. In this embodiment, a 4-inch single-crystal silicon wafer (500 micrometers thick) is selected as the semiconductor chip substrate, and the computing circuit layer has been fabricated on the front side of the single-crystal silicon wafer.
[0054] Photoresist is coated onto the back of the single-crystal silicon wafer and then photolithographically developed to define the flow channel mask pattern. The photolithography process includes the following sub-steps: First, a negative thick adhesive (e.g., SU-8 3050, with a thickness of 50 to 100 micrometers, to serve as a hard mask layer for deep reactive ion etching) is spin-coated onto the back of the single-crystal silicon wafer at a spin speed of 1500 to 2000 rpm for 30 seconds. Then, pre-baking (holding at 65 degrees Celsius for 3 minutes, followed by holding at 95 degrees Celsius for 20 to 30 minutes) is performed to allow the solvent in the photoresist to fully evaporate; The flow channel mask pattern is exposed onto the photoresist layer using a contact lithography machine (or a projection lithography machine) with an exposure energy of 200 to 300 millijoules per square centimeter. Exposure followed by baking (PEB, 65 degrees Celsius for 1 minute, then 95 degrees Celsius for 5 minutes) promotes photoresist crosslinking; Develop with SU-8 special developer for 4 to 6 minutes, rinse with isopropanol (IPA) and dry with nitrogen to obtain a photoresist mask layer with flow channel pattern; Finally, a hard film baking process (150 degrees Celsius, 30 minutes) is performed to enhance the etching resistance of the photoresist mask layer in the subsequent deep reactive ion etching process.
[0055] Here, deep reactive ion etching (DRIE, also known as Bosch process) is used to fabricate a channel array on the back side of the single-crystal silicon wafer. The width of the channels ranges from 50 micrometers to 200 micrometers, the depth from 100 micrometers to 300 micrometers, and the spacing between adjacent channels ranges from 100 micrometers to 400 micrometers. The total length of the channels covers the chip's heat-generating area corresponding to the computing circuit layer. The DRIE process forms high aspect ratio vertical sidewall channels on the silicon substrate by alternating etching and passivation steps. These vertical sidewalls facilitate the uniform construction of subsequent wettability gradients.
[0056] The Bosch cycle parameters for the deep reactive ion etching process are as follows: the etching step adopts... Gas flow rate (150 to 200 standard cubic centimeters per minute), RF power of 800 to 1200 watts, etching time of 8 to 12 seconds, and etching depth of approximately 0.8 to 1.5 micrometers per etching cycle; passivation step using... Gas flow rate of 80 to 120 standard cubic centimeters per minute, RF power of 600 to 800 watts, passivation time of 5 to 8 seconds, and a fluorocarbon polymer protective layer deposited on the flow channel sidewall to suppress lateral etching.
[0057] The etching and passivation steps are performed alternately and cyclically, with the number of cycles determined according to the target etching depth. For a flow channel with a target depth of 200 micrometers, the number of cycles is approximately 180 to 250, and the total etching time is approximately 60 to 90 minutes.
[0058] After etching, the single-crystal silicon wafer is placed in an oxygen plasma cleaner (300 watts, 10 minutes) to remove the residual fluorocarbon polymer passivation layer on the flow channel sidewall, ensuring that chemical reagents can uniformly contact the inner wall of the flow channel in subsequent process steps.
[0059] When verifying the geometry of the flow channel, a scanning electron microscope (SEM) was used to characterize the cross-section of the flow channel, measuring the actual width, depth, and sidewall perpendicularity (the target sidewall angle does not exceed...). (Deviating from the vertical direction). If the sidewall verticality does not meet the standard, the process can be corrected by adjusting the etching to passivation time ratio or reducing the RF bias power.
[0060] In S200, a micro-nano porous structure is fabricated at the inlet end of the flow channel to form a capillary-driven region; a micro-pillar array is fabricated in the downstream region of the flow channel to form a phase transformation heat transfer zone. Before processing begins, the inner wall of the channel after etching must be thoroughly cleaned to remove polymer byproducts (mainly fluorocarbon compounds) and metal contaminants remaining on the inner wall of the channel during deep reactive ion etching, ensuring the uniformity of subsequent electrochemical etching and wettability gradient construction processes.
[0061] Local mask protection is applied to the inlet region of the flow channel, and a micropillar array is fabricated in the downstream region of the flow channel using a secondary deep reactive ion etching process to form a phase change heat transfer zone.
[0062] The mask alignment steps for the secondary deep reactive ion etching process are as follows: Using the flow channel pattern in S100 as a reference, a double-sided alignment lithography machine (or an infrared alignment lithography machine) is used to precisely align the micropillar array mask pattern with the downstream region of the flow channel, with an alignment accuracy of not less than [missing information]. The micrometers are used to ensure that the micropillar array falls precisely within the phase change heat transfer zone of the flow channel without extending into the capillary drive zone.
[0063] After alignment is completed, a protective adhesive (AZ 5214E photoresist with a thickness of 3 micrometers) is spin-coated and cured in the inlet region of the flow channel (i.e., the position corresponding to the capillary drive region) to protect the inlet region and ensure that subsequent micropillar array etching is performed only in the downstream region of the flow channel.
[0064] The micropillars are cylindrical with a diameter of 30 to 100 micrometers, a height of 50 to 200 micrometers, and a spacing of 50 to 150 micrometers. The micropillar array significantly increases the gas-liquid contact area within the flow channel, promoting the formation of a uniform thin liquid film of the liquid cooling medium on the micropillar surface, which is beneficial for efficient evaporation and phase change heat transfer.
[0065] In other embodiments of this application, the micropillar array may also be selected as square pillars or conical micropillars to meet different heat dissipation enhancement requirements.
[0066] When square pillars are selected, their cross-sectional side length is 30 micrometers to 100 micrometers, and their height and spacing are the same as those of cylindrical pillars. They can be achieved using the same deep reactive ion etching process as cylindrical pillars. The mask pattern can be changed from a circle to a square, and the process parameters do not need to be adjusted.
[0067] When a conical micropillar is selected, the cross-section of the micropillar gradually narrows from the base to the top, with a base diameter of 30 to 100 micrometers, a top diameter of 10 to 50 micrometers, a height of 50 to 300 micrometers, and a cone half-angle of 2 to 10 degrees.
[0068] The conical micropillars are fabricated by adjusting the process parameters of deep reactive ion etching (DRIE): during etching, the C4F8 gas flow rate in the passivation step is gradually reduced, linearly decreasing from the initial flow rate (120 standard cubic centimeters per minute) to the final flow rate (40 standard cubic centimeters per minute). This causes the fluorocarbon passivation layer on the sidewalls to gradually thin with increasing etching depth, while the lateral etching amount increases with depth, thus forming a conical cross-section that is narrower at the top and wider at the bottom. After fabrication, the cross-section of the conical micropillars is characterized using a scanning electron microscope, measuring the top diameter, base diameter, and height to verify that the cone angle meets the design requirements (the allowable deviation of the cone half-angle is ±1 degree). The conical sidewalls of the micropillars gradually widen the gap between the micropillars from the top to the root, facilitating the detachment of vapor bubbles from the root of the micropillars and their escape along the sidewalls. This reduces the initial superheat of nucleate boiling by approximately 10% to 20%, delays the appearance of the critical heat flux density, and expands the safe operating heat flux density range of the heat dissipation system.
[0069] Secondly, at the inlet end of the flow channel, a porous silicon layer is formed on the surface of the single-crystal silicon wafer using an electrochemical etching method (see...). Figure 1 The capillary drive region 22 is formed.
[0070] The electrochemical etching is carried out in a hydrofluoric acid-based electrolyte, and by controlling the etching current density and time, a porous silicon layer with a porosity of 40% to 60% and a pore size of 10 nanometers to 500 nanometers is formed.
[0071] In this embodiment, the porous silicon layer employs a gradient pore size design, meaning that the region closer to the phase change heat transfer zone has a higher pore density and smaller pore size to increase capillary force; while the region farther from the phase change heat transfer zone has a lower pore density and larger pore size to reduce flow resistance, thereby optimizing the working fluid transport efficiency throughout the entire capillary drive region. The high specific surface area structure of the porous silicon layer allows the cooling working fluid to rapidly penetrate at the inlet end, providing a continuous capillary driving force for subsequent unidirectional flow.
[0072] The specific process parameters for the electrochemical etching are as follows: An electrolyte solution (hydrofluoric acid mass fraction of 49%) was prepared by mixing hydrofluoric acid and ethanol at a volume ratio of 1:1. The etching current density was 10 mA / cm² to 50 mA / cm², and the etching time was 5 min to 20 min. The single-crystal silicon wafer was used as the anode, and a platinum mesh was used as the cathode. The etching was performed at room temperature.
[0073] In the fabrication of the gradient-pore porous silicon layer in this embodiment, a partitioned mask method is used to achieve a pore density gradient: a denser photoresist aperture pattern (aperture ratio of 60% to 70%) is formed on the side near the phase change heat transfer zone (liquid outlet, i.e., high capillary force region), and a sparser aperture pattern (aperture ratio of 30% to 40%) is formed on the side away from the phase change heat transfer zone (liquid inlet, i.e., low flow resistance region). A transition region is set between the two regions to achieve a gradual transition in pore density. After etching, the single-crystal silicon wafer is ultrasonically cleaned in deionized water for 10 minutes and annealed at 200 degrees Celsius for 10 minutes under a nitrogen atmosphere to stabilize the microstructure of the porous silicon layer and improve its hydrophilicity.
[0074] S300, the inner wall of the flow channel is surface-treated along the length of the flow channel to create a unidirectional wettability gradient. In this embodiment, a unidirectional wettability gradient (i.e., roughness gradient method) is constructed on the inner wall of the flow channel using wet chemical etching. Before wet chemical etching, a copper thin film needs to be pre-deposited on the inner wall of the flow channel as a reaction substrate. First, the inner wall of the flow channel of the single crystal silicon wafer is cleaned by argon ion sputtering to remove the natural oxide layer; Subsequently, a copper film with a thickness of 200 nm to 500 nm was deposited on the inner wall of the flow channel using a magnetron sputtering process (the target material was a 99.99% pure copper target, the sputtering power was 100 W to 200 W, and the argon pressure was 0.5 Pa). The copper film served as the reaction substrate for the subsequent wet chemical etching growth of copper hydroxide nanowire arrays.
[0075] After copper thin film deposition is completed, the single crystal silicon wafer is ultrasonically cleaned in deionized water for 5 minutes and then dried with nitrogen gas for later use.
[0076] The single-crystal silicon wafer with the copper thin film deposited is then placed vertically along the length of the channel (channel inlet facing upwards), and slowly immersed at a uniform and controllable rate (5 mm / min to 10 mm / min) in a solution containing 1.5 mol / L sodium hydroxide (NaOH) and 0.06 mol / L ammonium persulfate (i.e., In the etching solution, the total immersion time is controlled to be 15 minutes, allowing the etching solution to gradually immerse the inner wall of the channel along its length. The vertical immersion direction ensures that the copper film at the outlet end (second end) contacts the etching solution first, with the longest immersion time and the densest growth of the copper hydroxide nanowires; the inlet end (first end) is immersed last, with the shortest immersion time and the sparsest growth of the nanowires, thereby forming a copper hydroxide nanowire array with a gradually varying density along the length of the channel on the inner wall of the channel.
[0077] This gradual immersion process causes the surface of the channel inner wall to... The density of the nanograss structure (i.e., copper hydroxide nanowire array) exhibits a gradient distribution along the length of the flow channel: The inlet end (corresponding to the first end) has the shortest immersion time, a sparse nano-grass structure, and a surface contact angle of approximately 80 to 83 degrees, corresponding to a hydrophobic state. The outlet end (corresponding to the second end) has the longest immersion time, a dense nano-grass structure, and a surface contact angle close to 0 degrees, corresponding to a superhydrophilic state. The contact angle continuously decreases along the length of the flow channel, forming a unidirectional wettability gradient surface.
[0078] The unidirectional wettability gradient causes the liquid cooling medium to be subjected to a directional capillary driving force from the inlet end to the outlet end. This force works in conjunction with the capillary force generated by the capillary driving zone 22 to drive the cooling medium to flow spontaneously in one direction without the need for any external pumping.
[0079] The uniformity of the unidirectional wettability gradient was verified using a contact angle measuring instrument (accuracy: [insert accuracy here]). The static contact angle is measured at 1 to 2 millimeters every 1 millimeter along the length of the flow channel, and the distribution curve of the contact angle along the flow channel position is recorded. Under the exemplary process parameters of this embodiment, the linearity error of the contact angle distribution curve does not exceed [a certain value]. Confirm that the uniformity of the gradient distribution meets the requirements. If the linearity error exceeds the standard, the process can be corrected by adjusting the immersion rate (faster immersion results in a steeper gradient change, while slower immersion results in a gentler gradient) or by adjusting the NaOH concentration (increased concentration increases the nanowire growth rate and density).
[0080] The S400 hermetically bonds a sealing cover to the semiconductor chip substrate and connects it to an external condensation radiation unit, forming a closed, self-driven phase-change heat dissipation circuit. Before performing hermetic bonding, the bonding surfaces of the microfluidic structure region on the back of the monocrystalline silicon wafer and the glass cover plate need to be surface treated to ensure bonding quality.
[0081] A Pyrex 7740 borosilicate glass cover was then selected as the sealing cover, and a steam outlet through-hole corresponding to the steam venting channel was pre-machined on the glass cover. In the hermetic sealing region 25, an anodic bonding process (bonding temperature 400 degrees Celsius, bonding voltage 800 volts) was used to hermeticly seal the glass cover to the back of the single-crystal silicon wafer, forming a closed microfluidic channel. The anodic bonding creates strong chemical bonds at the silicon wafer-glass interface, ensuring that the sealing reliability meets aerospace-grade hermeticity requirements.
[0082] The quality control measures for the anodic bonding process are as follows: During the bonding process, the bonding progress is judged by monitoring the trend of the bonding current. The bonding current is usually large at the beginning of the bonding and decreases monotonically as the bonding interface is gradually formed. When the current drops to less than 10% of the initial value, the bonding is considered complete. If the current fluctuates abnormally or fails to decrease normally, the bonding is stopped and the cleanliness and flatness of the bonding surface are checked.
[0083] After bonding is completed, the bonding interface is non-destructively inspected using a scanning ultrasonic microscope (C-SAM) to observe the bonding uniformity and identify any potential bonding voids. The area of any bonding void must not exceed 5% of the total bonding area and must not be located within the sealed flow channel area. For airtightness testing, a helium mass spectrometer is used to test the leakage rate of the sealed flow channel; the leakage rate must not exceed [a certain value]. (Aerospace-grade airtightness indicators) verify that the sealing reliability meets the requirements for long-term on-orbit use of satellites.
[0084] The bonded chip also needs to pass a thermal shock test. to (100 cycles) and vibration test (random vibration spectrum, on the order of magnitude of) The test was conducted over a period of 120 seconds to verify the reliability of the bonding interface under extreme mechanical and thermal conditions.
[0085] A stainless steel capillary tube (1.6 mm outer diameter) is connected to the steam outlet through hole. The steam channel is connected to the external radiant heat sink (i.e., external condensation radiant unit) through the pipeline. At the same time, the condensate outlet of the radiant heat sink is connected to the inlet of the capillary drive area 22 through the pipeline, forming a closed self-driven phase change heat dissipation circuit.
[0086] Then, cooling medium is injected into the closed flow channel. Before injection, the closed flow channel is placed in a vacuum chamber and evacuated until the pressure is lower than [previous pressure]. Pressurize for 30 minutes to remove residual gas from the flow channel and prevent the accumulation of non-condensable gas from affecting the phase change heat transfer efficiency.
[0087] After vacuuming is completed, the degassed cooling medium (in this embodiment, deionized water with a resistivity of not less than 18 megohms·cm) is injected into the closed flow channel through the filling pipeline under vacuum-assisted conditions. The filling rate is controlled between 40% and 60% (i.e., the volume of liquid working medium accounts for 40% to 60% of the total volume of the flow channel). If the filling rate is too high, the steam space will be insufficient, and if the filling rate is too low, the liquid working medium supply will be insufficient, both of which will affect the phase change heat dissipation efficiency.
[0088] After filling, epoxy resin sealant is dripped onto the filling port and cured (curing temperature is 80 degrees Celsius, curing time is 30 minutes), or the metal filling tube is welded by laser spot welding to seal the filling port and complete the entire preparation process.
[0089] In other embodiments of this application, the cooling medium may also be ammonia, ethanol, or a mixture of the above-mentioned mediums to adapt to different operating temperature ranges and material compatibility requirements. The specific selection principles are as follows: (1) When ammonia is used as the cooling medium, it has a high latent heat of vaporization (approximately 1370 kJ / kg at 0°C), making it suitable for low-temperature conditions ranging from -40°C to 80°C. This significantly improves the heat exchange capacity of the cooling system in the low-temperature environment of the satellite in orbit. Since ammonia is highly corrosive to copper, when using ammonia as the working medium, all components in contact with the working medium, including the semiconductor chip substrate, sealing cover, pipelines, and external condensation radiation unit, must be made of stainless steel (316L) or titanium alloy (TC4). Simultaneously, the porous structure of the capillary drive area should use porous titanium capillary cores or anodized aluminum nanoporous films (AAO films) to prevent the copper-based porous structure from being corroded and failing by ammonia. The degassing treatment before filling is the same as in this embodiment, and the filling rate is also controlled between 40% and 60%.
[0090] (2) When ethanol is used as the cooling medium, its wettability is better than that of water (surface tension is about 22 millinewtons per meter, much lower than water's 72 millinewtons per meter), resulting in better spreadability on the wettability gradient surface of the inner wall of the flow channel. This helps to form a more uniform thin liquid film in the phase change heat transfer zone, making it suitable for scenarios with high requirements for rapid wettability of the working medium. Ethanol has no significant corrosiveness to copper, stainless steel, and titanium, allowing for a wide range of material selection. Due to the lower surface tension of ethanol, the capillary pressure difference generated in the capillary drive zone is reduced by about 70% compared to the water working medium. Therefore, when using ethanol as the working medium, the pore size of the porous structure should be adjusted to 5 to 15 micrometers (smaller than the pore size range of the water working medium solution) to compensate for the capillary force loss caused by the reduced surface tension and ensure that the working medium circulation driving force meets the steady-state heat dissipation requirements. The filling rate is also controlled between 40% and 60%.
[0091] (3) When using water-ethanol as a working medium, the physical properties of the working medium can be continuously adjusted between pure water (high surface tension, high latent heat of vaporization) and pure ethanol (low surface tension, good wettability) by adjusting the mixing ratio, so as to meet the comprehensive optimization requirements of specific working temperature range and heat flux density range.
[0092] For example, using deionized water as the cooling medium, with a channel width of 100 micrometers, a channel depth of 200 micrometers, a channel length of 10 millimeters, a porous silicon layer in the capillary drive region with a pore size of 100 nanometers, and a micropillar array spacing in the phase change heat transfer region of 100 micrometers, the steady-state operating performance of the closed self-driven phase change heat transfer circuit can be estimated.
[0093] The maximum capillary pressure difference generated by the porous silicon layer in the capillary drive region ,in The surface tension of water, The contact angle is in a superhydrophilic state. (Half the aperture), substituting, we get .
[0094] The viscous flow resistance of the flow channel ,in The dynamic viscosity of the working fluid. The length of the flow channel. Volumetric flow rate, The diameter is the hydraulic diameter. During steady-state operation, the capillary driving force balances the flow resistance, maintaining continuous circulation of the working fluid. The above estimation results indicate that the capillary driving force (in the kilopascal range) is far higher than the gravitational potential energy difference (approximately 98 Pa for a water column with a 10 mm height difference), demonstrating that the working fluid circulation driving force of the closed self-driven phase change heat dissipation circuit does not depend on gravity and can operate normally in microgravity environments.
[0095] It should be noted that at the initial moment of chip power-on, the closed flow channel is filled with liquid cooling medium. Under the unidirectional wettability gradient directional driving force, the micro-nano porous structure of the capillary drive region 22 causes the liquid medium to rapidly wet towards the phase change heat transfer zone, establishing a stable liquid film supply. As the heat generation of the computing circuit layer 11 increases, the liquid film on the surface of the micropillar array of the phase change heat transfer zone 23 begins to evaporate. The resulting vapor pressure pushes the vapor through the vapor guiding channel 24 to the external condensation radiation unit.
[0096] The startup time for the heat dissipation system from initial power-on to establishing a stable phase transition cycle is typically tens of seconds to several minutes, depending on the chip power consumption rise rate and the working fluid filling rate. When the power consumption of the computing circuit layer undergoes a step change (such as switching from standby to full-load operation), the directional driving force generated by the unidirectional wettability gradient can respond quickly, allowing the working fluid flow rate to automatically adjust with the increase in power consumption without the need for an additional active flow control mechanism, demonstrating the adaptive adjustment capability of the heat dissipation system.
[0097] Example 2 Based on Example 1, this embodiment enhances and optimizes the capillary driving region and the phase transformation heat transfer region, providing a chip-embedded unidirectional self-driven microfluidic heat dissipation chip fabrication scheme that further improves the heat dissipation limit.
[0098] S100 (flow channel processing) in this embodiment is the same as in Embodiment 1, and will not be described again here.
[0099] S201: The capillary drive region is enhanced using anodized aluminum nanoporous films, and the phase transformation heat transfer region is enhanced using femtosecond laser micro / nano processing. In the processing of the capillary-driven region, an anodized aluminum oxide nanoporous film (i.e., an AAO nanoporous film, prepared from an anodized aluminum oxide template) is embedded at the inlet end of the flow channel. The pore size of the AAO nanoporous film is 20 nanometers to 100 nanometers, and the thickness is 50 micrometers to 200 micrometers. The small pore size of the AAO nanoporous film can generate a large capillary pressure difference at the inlet end, allowing the cooling medium to rapidly penetrate into the flow channel under capillary force and continuously supply liquid to the phase change heat transfer zone.
[0100] Compared with the porous silicon layer in Example 1, the AAO nanoporous film has more uniform and regular pores, which is conducive to the formation of a more uniform thin liquid film of the cooling working fluid and further improves the capillary evaporation efficiency.
[0101] The preparation process of the AAO nanoporous film is as follows: First, a thin aluminum film with a thickness of 500 nanometers to 1000 nanometers (purity of 99.999%) is deposited on the bottom wall of the inlet end of the flow channel using magnetron sputtering. The aluminum-containing thin film substrate was then placed in an oxalic acid (3% to 5% by mass) electrolyte and subjected to a first anodizing at a DC voltage of 20 to 40 volts (temperature 0 to 5 degrees Celsius, time 2 hours). After removing the irregular oxide layer formed by the first anodizing, a second anodizing was performed under the same conditions (time 4 to 8 hours) to form a highly ordered AAO nanopore array. Finally, the AAO nanoporous film is subjected to pore-expansion treatment in a phosphoric acid (5% by mass) solution (at a temperature of 30 degrees Celsius for 20 to 40 minutes) to expand the pore size to the target range (20 nanometers to 100 nanometers).
[0102] After the AAO nanoporous film is prepared, aluminum oxide with a thickness of 2 to 5 nanometers is deposited on the inner wall of the pores using atomic layer deposition (ALD) technology. A thin film was prepared to enhance the hydrophilicity of the AAO nanoporous film (contact angle after deposition is approximately...). to It also improves the mechanical strength and chemical stability of the inner wall of the pores, preventing radiation damage in the aerospace environment.
[0103] The surface of the AAO nanoporous film is hydrophilically modified (using a hydrophilic silane coupling agent) to ensure that the cooling medium is continuously and stably supplied to the phase change heat transfer zone under capillary force, thus avoiding the film drying out and affecting heat dissipation capacity.
[0104] In the fabrication of the phase change heat transfer zone, the surface of the micropillar array is processed using a femtosecond laser (pulse width on the order of 100 femtoseconds to 1 picosecond) to create a micro-nano composite structure. The femtosecond laser scans the micropillar surface at high peak power, forming a multi-level capillary structure on the micropillar surface that combines micron-scale grooves with nano-scale particles. This multi-level capillary structure significantly increases the effective specific surface area of the micropillars and simultaneously makes the micropillar surface superhydrophilic. This facilitates the spreading of the cooling medium on the micropillar surface to form an extremely thin liquid film, further enhancing the efficiency of thin liquid film formation and evaporation, and improving the heat dissipation limit of the phase change heat transfer zone.
[0105] After fabrication, the morphology of the multi-level micro / nano composite structure was characterized using SEM and atomic force microscopy (AFM) to confirm the superposition characteristics of micron-scale grooves and nano-scale particles. The static contact angle of the multi-level micro / nano composite structure was measured using a contact angle meter (target value not exceeding...). The superhydrophilicity of the micropillars was verified to meet the design requirements. In an exemplary embodiment, the effective specific surface area of the micropillars after the multi-level micro-nano composite structure treatment is increased by 2 to 5 times compared with that before treatment, which helps to further increase the gas-liquid contact interface area and reduce the liquid film spreading resistance.
[0106] The other steps are the same as in Example 1, and will not be repeated here.
[0107] In other embodiments of this application, the capillary drive region 22 may also employ a sintered porous capillary core instead of a porous silicon layer or anodized aluminum nanoporous film to further expand the range of material choices and working fluid compatibility. The porous capillary core is a porous metal structure prepared by powder sintering, as detailed below: Spherical copper powder with an average particle size of 5 to 50 micrometers (purity not less than 99.9%) is uniformly spread at the inlet end of the flow channel, with a powder thickness of 50 to 200 micrometers. Subsequently, it is sintered at 800 to 900 degrees Celsius for 30 to 60 minutes in a hydrogen protective atmosphere (volume fraction greater than 99.9%) to form a metallurgical bond between the copper powder particles, resulting in a porous copper capillary core with a porosity of 35% to 55% and an average pore size of 5 to 30 micrometers. This porous copper capillary core forms a tight fit with the inner wall of the flow channel inlet end, generating a capillary pressure difference on the order of several thousand to tens of thousands of Pascals, providing a continuous capillary driving force for the cooling working fluid.
[0108] In applications using ammonia or ethanol as the cooling medium, since ammonia is corrosive to copper, the porous capillary core can be replaced by sintering spherical titanium powder (average particle size of 10 to 50 micrometers) to prepare a porous titanium capillary core. The sintering temperature is 1000 to 1200 degrees Celsius, and the atmosphere is high vacuum (pressure below 1×10⁻³ Pa), resulting in a porous titanium structure with a porosity of 40% to 60%. The porous titanium capillary core has excellent corrosion resistance to both ammonia and ethanol and can maintain stable capillary transport performance under long-term on-orbit conditions.
[0109] Through the synergistic effect of capillary evaporation of the AAO nanoporous film and boiling heat transfer of the multi-level micro-nano composite structure of the micropillar array, the heat dissipation limit of the heat dissipation chip prepared in this embodiment is further improved, which can effectively meet the thermal management requirements of chips with higher heat flux density.
[0110] Compared with Embodiment 1, this embodiment has improvements in the following aspects: First, the pore size (20 nm to 100 nm) of the AAO nanoporous film is smaller than that of the porous silicon layer in Example 1 (wide distribution of 10 nm to 500 nm), and the pore size distribution is more uniform, which can generate a more stable capillary pressure difference, which is beneficial to the stable transport of the working fluid under different working conditions. Second, the multi-level micro-nano composite structure processed by femtosecond laser increases the effective gas-liquid interface area on the surface of the micropillar by 2 to 5 times compared with the smooth micropillar in Example 1, which helps to further improve the phase change heat transfer coefficient under the same micropillar array density. Third, the AAO nanoporous film is deposited via atomic layer deposition. The modified material exhibits excellent chemical stability and better compatibility with ammonia (which is corrosive to copper), thus broadening the range of suitable working media. The preparation method in this embodiment is more complex than that in Example 1, making it suitable for high-end spaceborne computing chip applications with higher heat dissipation requirements.
[0111] In other embodiments of this application, the unidirectional wettability gradient can also be constructed using a chemical gradient method: by plasma treatment or chemical vapor deposition, hydrophobic or hydrophilic functional groups of different densities are deposited along the length of the flow channel to achieve a continuous change in the contact angle; an exemplary implementation process of the chemical gradient method is as follows: the semiconductor chip substrate with the completed flow channel is placed in a plasma cleaner, and the inner wall of the flow channel is hydrophilically activated by oxygen plasma (power of 100 watts, pressure of 50 Pa, time of 5 minutes), so that the contact angle of the inner wall is reduced to The semiconductor chip substrate is then placed vertically along the length of the flow channel, and a 1% (volume ratio) solution of octadecyltrichlorosilane (OTS, hydrophobic modifier) in hexane is injected into the flow channel at a rate of 0.2 ml / min. The hydrophobic modifier solution has the longest residence time at the inlet end of the flow channel and the shortest residence time at the outlet end, forming a flow channel from the inlet end (hydrophobic, contact angle approximately...) to From the outlet end (hydrophilic, contact angle approximately) to The chemical contact angle gradient was calculated; finally, the inner wall of the flow channel was flushed with pure hexane to remove excess modifier, and the flow was dried with nitrogen to complete the chemical gradient method construction.
[0112] Alternatively, nanoimprinting can be used to prepare polymer nanopillar arrays with continuously varying diameters using an anodic porous alumina mold, thereby achieving a wettability gradient surface.
[0113] For example, the implementation process of the nanoimprinting method is as follows: the AAO mold (the aperture continuously varies from 50 nanometers to 200 nanometers along the length of the mold) is heated to above the softening temperature of the polymer (for polymethyl methacrylate, i.e., PMMA, it is heated to 130 degrees Celsius to 150 degrees Celsius), and the AAO mold is pressed into the polymer film coated on the inner wall of the flow channel with a pressure of 0.5 MPa to 2 MPa. After holding the pressure for 5 to 10 minutes, it is cooled and demolded, forming an array of polymer nanopillars with gradually changing diameters on the inner wall of the flow channel; the smaller the diameter of the nanopillars, the greater the surface roughness and the smaller the contact angle (more hydrophilic), and the larger the diameter, the larger the contact angle (more hydrophobic), thereby forming a unidirectional wettability gradient.
[0114] Alternatively, laser micro-nano fabrication (laser texturing) can be used, where a femtosecond laser is used to directly scan the surface of the flow channel metal or silicon substrate. By continuously changing the laser power, scanning speed, or pulse frequency, a micro-nano composite structure with gradually varying density and height can be constructed, achieving a continuous gradient transition of wettability from hydrophobic to superhydrophilic.
[0115] The above-mentioned methods for constructing wettability gradients can all be combined with the flow channel structure of this application to achieve unidirectional self-driven heat dissipation. This application is not limited to a specific construction process.
[0116] In other embodiments of this application, the sealing cover can also be hermetically connected to the semiconductor chip substrate by eutectic bonding or fusion bonding to adapt to different material systems and packaging process requirements.
[0117] When using eutectic bonding, a gold-tin (Au-Sn, mass ratio 80:20) eutectic alloy can be selected as the bonding intermediate layer. The bonding temperature is 320°C to 340°C (Au-Sn eutectic point is 280°C), the bonding pressure is 0.5 MPa to 1 MPa, and the bonding is completed under an inert gas protective atmosphere (nitrogen or argon). The eutectic bonding has the advantages of low bonding temperature and high mechanical strength, and is suitable for chip packaging scenarios with strict thermal budget constraints.
[0118] When using fusion bonding (direct bonding), the bonding surfaces of the silicon wafer and the silicon cover plate can be chemically and mechanically polished (roughness reduced). The material (not exceeding 0.5 nanometers) undergoes hydrophilic activation treatment followed by pre-bonding at room temperature, and then covalent bonding is formed through high-temperature annealing at 800°C to 1100°C (inert atmosphere, for 2 hours). The strength of the molten bonding interface is close to that of the bulk silicon material, making it suitable for aerospace-grade packaging scenarios with extremely high hermeticity requirements. The external condensation radiation unit can be adjusted according to the heat dissipation radiation surface resources of the satellite platform. By optimizing the area of the radiating heat sink and the pipeline layout, the performance and space adaptability of the overall heat dissipation system can be further improved.
[0119] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips, characterized in that, include: A semiconductor chip substrate includes a computing circuit layer and a microfluidic heat dissipation structure layer; wherein, the microfluidic heat dissipation structure layer includes several channels, a capillary driving region disposed at the inlet end of the channels, a phase change heat transfer region disposed in the downstream region of the channels, and a vapor conduction channel communicating with the channels. A sealing cover plate is hermetically bonded to the semiconductor chip substrate to form a flow channel; An external condensation radiation unit is connected to the steam channel and the capillary drive zone inlet via a pipeline, forming a self-driven phase change heat dissipation circuit.
2. The embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips according to claim 1, characterized in that, The flow channels are arranged in an array, and the contact angle of the surface of the flow channels changes continuously and gradually along the flow direction of the working fluid.
3. The embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips according to claim 2, characterized in that, The first end contact angle of the flow channel corresponds to the hydrophobic state, and the second end contact angle corresponds to the superhydrophilic state.
4. The embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips according to claim 1, characterized in that, The capillary drive region is provided with a micro-nano porous structure, which is at least one of a porous thin film, a porous silicon layer, or a porous capillary core.
5. The embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips according to claim 4, characterized in that, The micropillar array can be any one of a cylindrical, square, or conical structure.
6. A method for fabricating an embedded unidirectional self-driven microfluidic heat dissipation chip for satellite computing power chips, characterized in that, include: Several flow channels are processed on the back or inside of a semiconductor chip substrate; Micro-nano porous structures are fabricated at the inlet end of the flow channel to form a capillary drive region; And a micropillar array is fabricated in the downstream region of the flow channel to form a phase change heat transfer zone; The inner wall of the flow channel is surface treated along its length to continuously increase the wettability of the inner wall of the flow channel along the flow direction of the working fluid. The sealing cover is hermetically bonded to the semiconductor chip substrate to form a flow channel; An external condensation radiation unit is connected, and the steam channel is connected to the inlet of the capillary drive area through a pipeline to form a self-driven phase change heat dissipation circuit.
7. The preparation method according to claim 6, characterized in that, The flow channel is fabricated using a deep reactive ion etching (DRIE) process, and the micropillar array is fabricated in the downstream region of the flow channel using a secondary DRIIE process.
8. The preparation method according to claim 7, characterized in that, A porous silicon layer is formed at the inlet of the flow channel by electrochemical etching as the capillary driving region. The pore density of the porous silicon layer is greater on the side closer to the phase change heat transfer zone than on the side farther away from the phase change heat transfer zone.
9. The preparation method according to claim 6, characterized in that, Cooling medium is injected into the closed flow channel, and the injection port is sealed after the injection is completed.
10. The preparation method according to claim 9, characterized in that, The cooling medium is at least one of water, ammonia, ethanol, or a mixture thereof.
