A chip stacked interconnect system and its manufacturing method

By employing a three-stage process of wafer-level bonding and self-grown interconnects, the thermal paths of logic chips and memory chips are decoupled, solving the problems of high process difficulty, high parasitic parameters, and thermal coupling in existing chip stacking interconnect schemes. This achieves efficient chip stacking interconnects and heat dissipation, improving the thermal reliability and electrical signal transmission performance of the system.

CN123124822APending Publication Date: 2026-11-17上海曜感科技有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611365285.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-03
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing chip stacking interconnect solutions suffer from problems such as narrow process windows within the grooves, high alignment difficulty, high parasitic parameters in the interconnect structure, and heating of the memory chips due to heat dissipation path coupling, which affect the thermal reliability and electrical signal transmission performance of the system.

Method used

The process employs a three-stage approach: wafer-level bonding and self-grown interconnects, dicing to form stacked units, and overall embedding into the substrate groove. By using logic chips as the first chip to set up heat sinks and embedding memory chips on the substrate, combined with decoupling capacitors and thermoelectric coolers, thermal path decoupling and power supply noise suppression are achieved.

Benefits of technology

It significantly reduces process difficulty, improves manufacturing efficiency and yield, enhances system thermal reliability and electrical signal integrity, and meets the requirements of high-frequency and high-speed signal transmission.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123124822A_ABST
    Figure CN123124822A_ABST
Patent Text Reader

Abstract

This invention relates to the field of semiconductor packaging technology, specifically to a chip stacking interconnect system and its manufacturing method. The method includes: S1: providing a substrate, etching grooves on the upper surface of the substrate, and forming through-silicon vias (TSVs) in the substrate; and providing a first bonding pad on the upper surface of a single-crystal silicon substrate to obtain an intermediate structure; S2: providing a wafer and at least one second functional chip, pre-forming a plurality of first functional chips on the wafer; and forming a second bonding pad and a third bonding pad on opposite sides of the first and second functional chips, respectively. This invention avoids the difficulties of operation within the grooves and simplifies the process through a three-stage process of wafer-level bonding and self-grown interconnection, and overall embedding into the substrate grooves after dicing. Simultaneously, by having the logic chip's back side directly contact the heat sink and the memory chip embedded in the grooves, the thermal paths of the two are decoupled, effectively solving the heat dissipation problem of heat-sensitive memory and improving system reliability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of semiconductor packaging, and more specifically to a chip stacking interconnect system and its manufacturing method. Background Technology

[0002] Heterogeneous chip-to-chip integration technology achieves high-performance, high-bandwidth system-level integration by three-dimensionally stacking and interconnecting multiple chips with different functions and process nodes. It has become an important technological path for continuing Moore's Law in the post-Moore era. In chip-to-chip stacked interconnect systems, memory chips and logic chips are typically stacked vertically, and high-density interconnection is achieved through micro-bumps or hybrid bonding between chips.

[0003] In existing chip stacking interconnect solutions, a substrate with grooves is typically fabricated first, and then chips are placed one by one within the grooves, followed by underfilling and interconnection processes. This method requires multiple high-precision operations within the confined space of the grooves, resulting in a narrow process window, high alignment difficulty, and low production efficiency. Furthermore, the interconnection between chips in existing solutions typically employs a solder microbump reflow soldering process. As interconnection density increases and pad spacing decreases, the risk of short circuits due to microbump bridging significantly increases, and the resistance and parasitic capacitance of solder interconnects are relatively large, making it difficult to meet the requirements of high-frequency, high-speed signal transmission. Simultaneously, the electrical connection between chips and the substrate in existing solutions is usually achieved through wire bonding or edge wiring, resulting in long signal transmission paths and large parasitic parameters, limiting further improvements in system bandwidth and signal integrity.

[0004] In terms of heat dissipation, in existing vertical stacking structures, logic chips and memory chips are tightly stacked. The logic chips, as the primary heat source, easily generate heat that can inflate the temperature-sensitive memory chips through heat conduction paths, leading to increased operating temperatures, deterioration of data retention characteristics, and even failure of the memory chips. Since memory chips are extremely sensitive to temperature changes (typically requiring operating temperatures below 85°C), while logic chips can operate at temperatures exceeding 100°C, the thermal coupling between the two severely restricts the system's thermal reliability and performance. How to effectively decouple the thermal paths between logic chips and memory chips while achieving high-density vertical interconnects has become a pressing technical challenge in this field.

[0005] In addition, the power distribution network inside the existing substrate usually only contains multiple wiring layers and lacks local decoupling capability. The power noise generated when the logic chip switches at high speed can easily be coupled to the memory chip through the power distribution network, affecting the read and write reliability and signal integrity of the memory chip. Summary of the Invention

[0006] The present invention aims to overcome at least some of the problems in existing chip stacking interconnect schemes, such as narrow process windows in the grooves, high alignment difficulty, high parasitic parameters of interconnect structure, and heating of memory chips due to heat dissipation path coupling, and provides a chip stacking interconnect system and its manufacturing method.

[0007] To address the above problems, this invention provides a method for manufacturing a chip stacked interconnect system, comprising the following steps: S1: A substrate is provided, a groove is etched on the upper surface of the substrate, and a through-silicon via is formed in the substrate; a first bonding pad is disposed on the upper surface of the substrate to obtain an intermediate structure; S2: A wafer and at least one second functional chip are provided. A plurality of first functional chips are pre-formed on the wafer. Second and third bonding pads are formed on opposite sides of the first and second functional chips, respectively. The second functional chips are interconnected and bonded to the first functional chips, with the second and third bonding pads perpendicularly aligned and bonded to achieve electrical interconnection between them. The interconnected wafer is diced to obtain independent stacking units. In the stacking units, the coverage area of ​​the first functional chips is larger than the coverage area of ​​the second functional chips. S3: The stacked unit is placed in the groove of the intermediate structure; wherein the second functional core is located in the groove, and the peripheral portion of the first functional core is supported on the top periphery of the groove; a fourth solder pad is provided on the surface of the first functional core, which is perpendicularly aligned with the first solder pad and electrically interconnected; a filling layer is formed in the groove, such that the filling layer fills the gap inside the groove and its surface is coplanar with the surface of the first core; a heat dissipation structure is provided on the back side of the first functional core; S4: Thin the side of the intermediate structure away from the stacking unit to expose the bottom end of the through silicon via; form conductive bumps at the exposed bottom end of the through silicon via; align and press the intermediate structure with the external multilayer substrate to electrically connect each conductive bump to the corresponding conductive pad on the multilayer substrate. S5: Divide the composite wafer array after lamination to obtain independent chip stacked interconnect packaging units.

[0008] Optionally, the first functional chip is a logic chip, and the second functional chip is a storage chip.

[0009] Optionally, the fourth pad of the first functional core is disposed in the peripheral area of ​​its bonding surface, and the first pad on the substrate surface is disposed around the groove and perpendicularly corresponds to the fourth pad.

[0010] Optionally, it may also include bonding and interconnecting N functional cores on the back side of the first functional core, where N≥1.

[0011] Optionally, the bonding layer is a permanent dry film, which is laminated to the bonding surface of the second functional chip and / or the wafer by hot rolling, and the second functional chip is mechanically fixed to the wafer by hot pressing.

[0012] Optionally, the depth of the groove matches the thickness of the second functional core, such that after the stacking unit is embedded, the back side of the second functional core contacts or is close to the bottom of the groove, and the top surface of the second functional core is flush or substantially flush with the surface of the stacking unit.

[0013] Optionally, after the stacked unit is embedded in the groove, the method further includes: The gap between the groove sidewall and the second functional core and the peripheral area of ​​the first functional core are filled with a medium, such that the filling medium is flush with the back substrate of the first functional core.

[0014] Optionally, it also includes decoupling capacitors, resistors, inductors, and thermoelectric coolers that are electrically connected to the through-silicon vias integrated within the substrate; The decoupling capacitor is arranged close to the first functional core and / or the second functional core. The resistor and / or inductor, together with the decoupling capacitor, form a filter network for the power distribution network, used to suppress noise in the power distribution network. The thermoelectric cooler is electrically connected to an external power source through an independent through-silicon via and is located between the bottom of the groove and the second functional chip to provide active cooling for the second functional chip.

[0015] Optionally, the bonding interconnection method is any one of hybrid bonding, thermo-press bonding, solder bonding, or self-grown interconnection.

[0016] Optionally, the self-grown interconnect is achieved through a selective deposition process, which may be electroless plating, electroplating, chemical vapor deposition, or atomic layer deposition.

[0017] Optionally, the heat dissipation structure is a thermally conductive microchannel disposed on the back side of the first functional core, and the thermally conductive microchannel is in communication with the outside; or, A heat sink is located on the back of the first functional core.

[0018] A chip stack interconnect system, comprising: An intermediate structure includes a substrate, a groove is formed on the upper surface of the substrate, a through-silicon via is formed in the substrate, and a first bonding pad is disposed on the upper surface of the substrate; A stacking unit is disposed on the intermediate structure. The stacking unit includes a first functional core and a second functional core disposed below the first functional core. The coverage area of ​​the first functional core is larger than that of the second functional core. The second functional core is located within the groove, and the peripheral portion of the first functional core is supported by the top periphery of the groove. A second solder pad and a third solder pad are respectively disposed on the opposite surfaces of the first and second functional cores. The second functional core is bonded to the first functional core, such that the second and third solder pads are vertically aligned and bonded together to achieve electrical interconnection between them. A fourth solder pad is disposed on the surface of the first functional core opposite to the intermediate structure. The fourth solder pad is aligned with and electrically interconnected with the first solder pad. A filling layer fills the gap inside the groove, and its surface is coplanar with the surface of the first functional core particle; A heat dissipation structure is disposed on the back side of the first functional core; and Conductive bumps are disposed on the side of the intermediate structure opposite to the stacked unit and electrically connected to the bottom end of the through-silicon via for interconnection with an external circuit board.

[0019] The present invention has at least the following beneficial effects: 1. This solution employs a three-stage process: wafer-level bonding and self-growing interconnection, dicing to form stacked units, and overall embedding into the substrate groove. First, high-precision bonding and interconnection of the first and second cores are completed at the wafer level. Then, the cores are diced to form independent stacked units. Finally, the stacked units are embedded into the substrate groove and electrical connection with the substrate is completed simultaneously. The interconnection between the stacked units and the substrate can be achieved with a single alignment, avoiding the difficulty of performing multiple high-precision process operations directly in the groove. This significantly reduces the process difficulty and improves manufacturing efficiency and yield.

[0020] 2. This solution places the logic chip as the first chip above the second chip, and directly places a heat sink on the back of the logic chip. This allows the large amount of heat generated by the logic chip during operation to be directly conducted to the heat sink through the thermal interface material on its back and dissipated to the external environment. The heat dissipation path does not pass through the memory chip. At the same time, the memory chip is embedded in the substrate groove, and passive heat dissipation is achieved through the substrate. The thermal paths of the two are decoupled from each other, avoiding the heat generated by the logic chip from heating the memory chip and improving the thermal reliability of the system.

[0021] 3. This solution provides a fourth pad for connecting the intermediate structure in the peripheral area of ​​the first functional core bonding surface, and a first pad is provided at the corresponding position outside the groove of the intermediate structure. This allows the fourth pad and the first pad to achieve shortest path alignment and interconnection in the vertical direction after the stacked unit is embedded in the groove, eliminating the need for additional bridging wiring. This helps reduce parasitic parameters and signal transmission delay, and improves high-speed signal integrity. At the same time, a decoupling capacitor is provided inside the intermediate structure near the groove to provide local decoupling for the first and second functional cores, effectively suppressing noise in the power distribution network and improving the power integrity of the system. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the structure in step S1 of embodiment one of the present invention, which provides a substrate and integrates decoupling capacitors and through-silicon vias; Figure 2 This is a schematic diagram of the wafer structure provided in step S2 of embodiment one of the present invention; Figure 3 This is a schematic diagram of the structure of multiple second functional cores provided in step S2 of embodiment one of the present invention; Figure 4 This is a schematic diagram of the structure in step S2 of embodiment one of the present invention, in which a bonding layer is formed on the first functional chip; Figure 5 This is a schematic diagram of the structure in step S2 of embodiment one of the present invention, in which the second functional core is bonded to the substrate; Figure 6 This is a schematic diagram of the selective deposition and self-growth interconnection of the first functional core and the second functional core in step S2 of embodiment one of the present invention; Figure 7 This is a schematic diagram of the structure in step S2 of embodiment one of the present invention, in which the interconnected wafer is cut into stacked units; Figure 8 This is a schematic diagram of the structure in step S3 of embodiment one of the present invention, in which the stacked units are embedded in the base groove and interconnected. Figure 9 This is a schematic diagram of the filling medium material in step S3 of embodiment one of the present invention; Figure 10 This is a schematic diagram of the structure of the heat sink in step S3 of embodiment one of the present invention; Figure 11This is a schematic diagram of the structure of the stacked unit and the substrate bonded together and then connected to the multilayer carrier board in step S4 of embodiment one of the present invention. Figure 12 This is a schematic diagram of the stacked third core particle structure in Embodiment 2 of the present invention.

[0024] Icon labels: 10-Substrate; 11-Groove; 12-Decoupling capacitor; 13-Through silicon via; 14-First bonding pad; 20-Wafer; 21-First functional chip; 22-Second bonding pad; 23-Bonding layer; 30-Second functional chip; 31-Third bonding pad; 40-Metal pillar; 50-Stacking unit; 51-Fourth bonding pad; 60-Fill layer; 70-Heat sink; 80-Multilayer carrier; 81-Conductive bump; 90-Third functional chip. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0026] In the description of this invention, it should be understood that the terms "center", "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0027] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0028] Example 1

[0029] This embodiment provides a method for manufacturing a chip stacked interconnect system, characterized by the following steps: S1: A substrate 10 is provided, a groove 11 is etched on the upper surface of the substrate 10, and a through-silicon via 13 is formed in the substrate 10; a first bonding pad 14 is provided on the upper surface of the substrate 10 to obtain an intermediate structure; refer to Figure 1 As shown, a substrate 10 is provided, which is a monocrystalline silicon substrate, for supporting the stacked unit 50 and providing electrical connection and heat dissipation path to external systems. The substrate 10 is a monocrystalline silicon substrate 10, preferably with a thickness of 800-1000μm, and contains multiple wiring layers for signal routing and power distribution.

[0030] At least one groove 11 is formed on the upper surface of the substrate 10 for accommodating the second functional core 30: the groove 11 is formed by ultraviolet laser ablation, CO2 laser ablation, high-precision CNC machine milling or deep reactive ion etching (DRIE) process on the silicon interlayer.

[0031] The depth of the groove 11 matches the thickness of the second functional core 30, with an additional 5-20 μm added to the thickness of the second functional core 30 to reserve space for the adhesive layer. This ensures that after the stacking unit 50 is embedded, the top surface of the second functional core 30 is flush with or substantially flush with the surface of the intermediate structure, with a height difference of <5 μm. The planar dimensions of the groove 11 are 50-200 μm larger than the second functional core 30, with a 25-100 μm gap reserved on each side to facilitate the placement and alignment of the stacking unit 50. The angle between the sidewall and the bottom surface of the groove 11 is 80°-120°, preferably 90°.

[0032] A first pad 14 is provided on the surface of the substrate 10 at the corresponding position around the groove 11. The first pad 14 is electrically connected to the through silicon via 13 and the external connection point through the wiring layer inside the substrate 10. The top surface of the first pad 14 is planarized, with a roughness Ra < 0.5 nm, and the surface height is flush with or slightly higher than the surface of the substrate 10 by 0-2 μm.

[0033] A through-silicon via (TSV) 13 is also provided inside the substrate 10, connecting the first bonding pad 14 to the back side of the substrate 10 or the internal wiring layer. For the silicon interposer substrate 10, the via diameter is 10-100 μm, preferably 30-50 μm, with a depth-to-width ratio of 5:1 to 20:1 (preferably 10:1). The insulating layer is SiO2 or SiN with a thickness of 0.2-1 μm, the barrier layer is Ti / TiN or Ta / TaN with a thickness of 50-200 nm, and the filler metal is electroplated copper. For the organic substrate 10, the via is formed by laser drilling with a diameter of 50-150 μm and filled with electroplated copper. The TSV 13 is a blind hole extending downward from the upper surface of the substrate 10, and its depth is greater than or equal to the final thickness of the substrate 10 after subsequent thinning. In the back side thinning process of step S4, the bottom end of the TSV 13 is exposed on the back side of the substrate 10, forming a through-silicon via structure.

[0034] In a preferred embodiment, a decoupling capacitor 12 electrically connected to the through-silicon via 13 is also integrated within the substrate 10. This decoupling capacitor 12 is positioned close to the subsequently placed first functional chip 21 and / or second functional chip 30 for noise suppression in the power distribution network. The decoupling capacitor 12 is a metal-insulator-metal capacitor, formed between wiring layers within the substrate 10, with a capacitance density of 1-10 fF / μm². In other variations, the decoupling capacitor 12 may also be a deep trench capacitor (formed in a silicon interposer through deep trench etching and dielectric / metal filling, with a capacitance density of 10-100 fF / μm²) or a thin-film capacitor using a high-k dielectric material (capacitance density 50-200 fF / μm²). The decoupling capacitor 12 is positioned close to the first functional chip 21 and / or second functional chip 30, with a power supply loop length < 500 μm to minimize parasitic inductance.

[0035] Furthermore, a resistor, inductor, and thermoelectric cooler (not shown) can also be integrated within the substrate 10. The resistor and / or inductor, together with the decoupling capacitor 12, constitute a filter network for the power distribution network, used to suppress noise in the power distribution network. This network can be formed through a wiring layer or thin-film process within the substrate 10. The thermoelectric cooler is electrically connected to an external power supply through an independent through-silicon via (TSV), and is disposed between the bottom of the groove 11 and the second functional chip 30. It provides active cooling for the second functional chip 30. The thermoelectric cooler can be made of bismuth telluride-based thermoelectric thin film material, deposited on the bottom of the groove 11 by magnetron sputtering or electroplating.

[0036] S2: A wafer 20 and at least one second functional chip 30 are provided. A plurality of first functional chips 21 are pre-formed on the wafer 20. The first functional chips 21 and the second functional chips 30 are placed opposite each other and mechanically fixed by a bonding layer 23. Second solder pads 22 and third solder pads 31 are formed on the opposite sides of the first functional chips 21 and the second functional chips 30, respectively. The solder pads are vertically aligned and electrically interconnected by selective deposition process, forming interconnect metal pillars 40 in the gap between the second solder pads 22 and the third solder pads 31. The interconnected wafer 20 is divided to obtain independent stacking units 50. In the stacking unit 50, the coverage area of ​​the first functional chips 21 is larger than the coverage area of ​​the second functional chips 30. In this step, the bonding interconnection between the first functional core 21 and the second functional core 30 can be any one of hybrid bonding, thermo-press bonding, solder bonding, or self-growing interconnection.

[0037] refer to Figure 2As shown, a wafer 20 with completed front-end processes is provided. The wafer 20 includes a plurality of first functional granules 21 arranged in a predetermined pattern. The first functional granules 21 are logic granules, which can be CPU, GPU, or AI processor cores. At the same time, second bonding pads 22 are formed on the bonding surface of the wafer 20.

[0038] The second solder pad 22 is disposed in the peripheral region of the bonding surface of the first functional core 21 for electrical connection with the intermediate structure. The diameter of the second solder pad 22 is preferably 30-50 μm, and the material is copper or copper / nickel / gold laminate.

[0039] The second pad 22 is electrically connected to the logic circuit inside the first functional chip 21 through the redistribution layer (RDL) inside the wafer 20.

[0040] refer to Figure 3 As shown, at least one second functional chip 30 is provided, which has completed front-end processing and testing screening and is a known good chip (KGD). Each second functional chip 30 includes a memory circuit, which may be a DRAM, SRAM, or HBM memory cell. The thickness of the second functional chip 30 is preferably 50-100 μm, and its planar dimensions are from 5 mm × 5 mm to 20 mm × 20 mm.

[0041] A third bonding pad 31 is provided on the bonding surface of the second functional chip 30, that is, on the side opposite to the first functional chip 21. The third bonding pad 31 corresponds one-to-one with the second bonding pad 22 on the wafer 20. The diameter, spacing and material of the third bonding pad 31 are matched with those of the second bonding pad 22 (diameter 5-30μm, spacing 10-50μm, material is copper) to ensure the reliability of subsequent alignment and interconnection.

[0042] The back (bottom) of the second functional chip 30 has no solder pads, and is a flat silicon surface or covered with a protective layer.

[0043] refer to Figure 4 As shown, before bonding, a permanent dry film is laminated onto the bonding surface of wafer 20 using hot roller lamination as bonding layer 23. The lamination temperature is 80-120℃, the pressure is 0.3-0.8MPa, the speed is 0.5-2m / min, and the dry film thickness is 10-50μm. Then, the dry film is patterned by photolithography to form openings above the second pad 22 and the third pad 31 to expose the metal surfaces to be interconnected. The photolithography process includes exposure (dose 150-300mJ / cm²), post-exposure baking (65℃ / 2min+95℃ / 5min), development and hard baking (150℃ / 10min). After patterning, the dry film is retained in the non-pad areas as a permanent bonding substrate.

[0044] This solution uses a permanent dry film as the bonding layer 23, which is laminated by hot roller and then cured by hot pressing. It does not need to be removed in subsequent processes, eliminating the need for temporary bonding / debonding steps and simplifying the process flow. At the same time, the surface of the permanent dry film does not have catalytic activity and can also serve as a mask layer during selective deposition. The metal grows only on the surface of the pads and does not deposit on the surface of the dry film, eliminating the need for additional masking. This achieves a synergistic effect of mechanical fixation and selective deposition, and the dry film thickness is uniform and controllable, ensuring the consistency of the gap between the pads and improving the reliability of the interconnect.

[0045] refer to Figure 5 As shown, next, a high-precision flip-chip bonding machine is used to pick up each of the second functional chips 30 and place them one by one onto the corresponding positions of the first functional chips 21 on the wafer 20, with an alignment accuracy of ±1-3μm, so that the third pad 31 of the second functional chip 30 is precisely aligned with the second pad 22 of the first functional chip 21; then, hot-press bonding is performed, and the bonding temperature is maintained at 150-240℃, pressure at 5-30N / cm², and in a nitrogen or vacuum environment for 5-30 minutes to cure the dry film and mechanically fix the two together. After bonding, the second pad 22 and the third pad 31 are positioned opposite each other with a gap of 1-5μm. After bonding, palladium activation solution (palladium chloride 0.1-0.5g / L, hydrochloric acid 1-5ml / L) is delivered to the pad area through the dry film opening, and treated at room temperature for 1-3 minutes, so that the activation solution selectively forms palladium catalytic centers on the exposed copper pad surface, while the dry film surface does not have catalytic activity.

[0046] refer to Figure 6 As shown, finally, the bonded structure is immersed in a chemical copper plating solution for selective deposition. The chemical copper plating solution contains 5-15 g / L copper sulfate pentahydrate, 5-15 ml / L formaldehyde or 10-30 g / L sodium hypophosphite, 20-40 g / L EDTA or 15-30 g / L sodium potassium tartrate, 5-20 mg / L 2,2'-bipyridine, pH 11.5-12.5, temperature 50-70℃, and deposition time 20-60 minutes. During this process, copper ions are reduced and deposited on the surface of the pads with palladium catalytic centers. Since both the second pad 22 and the third pad 31 have catalytic activity, copper grows simultaneously from the surfaces of the upper and lower pads, growing towards each other in a 1-5 μm gap and eventually bridging to form a complete interconnected metal pillar 40. This allows the first functional core 21 and the second functional core 30 to form an electrically connected and mechanically fixed stacked unit 50, and the metal pillar 40 grows only on the pad surface and does not deposit on the dry film surface.

[0047] The selective deposition process described above can be achieved through any one of electroless plating, electroplating, chemical vapor deposition, or atomic layer deposition. This embodiment uses electroless plating as an example. In other embodiments, the bonding interconnect can also be achieved through any one of hybrid bonding, thermo-press bonding, or solder bonding, all of which can achieve the technical effects of the present invention.

[0048] refer to Figure 7 As shown, the interconnected wafer 20 is diced to form multiple independent stacking units 50. Each stacking unit 50 includes a first functional core 21 and a second functional core 30. During dicing, mechanical dicing is performed using a diamond blade, or laser dicing is performed using an ultraviolet / femtosecond laser. The dicing width is 30-80 μm (preferably 50 μm), and the dicing depth is from the front side of the wafer 20 to 80-90% of the thickness of the wafer 20, leaving 10-20% of the thickness on the back side to maintain the integrity of the stacking unit 50. The mechanical dicing speed is 10-30 mm / s, and the laser dicing speed is 50-200 mm / s. Deionized water at 18-25°C is used for cooling during the dicing process.

[0049] After cutting, the stacked cell 50 is cleaned to remove debris and contaminants generated during cutting. After cutting, each stacked cell 50 includes a first functional chip 21 and a second functional chip 30, wherein the first functional chip 21 is a logic chip and the second functional chip 30 is a storage chip; the bonding surface area of ​​the first functional chip 21 remains exposed except for the second pad 22 that is interconnected with the second functional chip 30, for subsequent electrical connection with the interposer structure.

[0050] S3: The stacking unit 50 is placed in the groove 11 of the intermediate structure; wherein the second functional core 30 is located in the groove 11, and the peripheral portion of the first functional core 21 is supported on the top periphery of the groove 11; a fourth solder pad 51 is provided on the surface of the first functional core 21, which is perpendicularly aligned with the first solder pad 14 and electrically interconnected; a filling layer 60 is formed in the groove 11, such that the filling layer 60 fills the internal gap of the groove 11 and its surface is coplanar with the surface of the first core; a heat dissipation structure is provided on the back side of the first functional core 21; refer to Figure 8As shown, an adhesive material with a thickness of 5-20 μm is coated on the bottom and sidewalls of the intermediate structure groove 11 to fix the second functional core 30 of the stacking unit 50 in the groove 11. A high-precision pick-and-place machine is used to pick up and align the stacking unit 50 to the position of the groove 11 of the intermediate structure, so that the fourth solder pad 51 on the bonding surface of the first functional core 21 of the stacking unit 50 is precisely aligned with the first solder pad 14 on the surface of the intermediate structure, with an alignment accuracy of ±2-10 μm. During placement, the second functional core 30 of the stacking unit 50 is partially embedded in the groove 11, and its back side contacts the bottom of the groove 11 through the adhesive layer. The top surface of the second functional core 30 is flush or substantially flush with the surface of the intermediate structure.

[0051] Subsequently, hot-press bonding is performed, and the bonding layer is maintained for 5-30 minutes at a bonding temperature of 150-250℃, a pressure of 5-30N / cm², and a nitrogen atmosphere to solidify the adhesive layer and fix the first functional core 21 in the groove 11. At the same time, the fourth solder pad 51 and the first solder pad 14 form a metal bond under pressure and temperature, realizing the mechanical fixation and electrical connection between the stacked unit 50 and the intermediate structure.

[0052] refer to Figure 9 As shown, after the stacked unit 50 is embedded in the groove 11, and after the stacked unit is embedded in the groove 11, the gap between the sidewall of the groove 11 and the second functional core 30, as well as the peripheral area of ​​the first functional core 21, is filled with a dielectric material. Specifically, a capillary underfill process is used, in which liquid underfill adhesive is injected along the edge of the gap at room temperature or 80-100°C, filling by capillary action, and then cured at 150-165°C for 60-120 minutes. After curing, the surface of the filling medium is ground to make the surface of the filling medium flush with the back surface of the first functional core 21, forming a flat and coplanar composite surface. The dielectric filler plays a role in protecting the interconnect structure, dispersing thermal stress, and preventing contaminants from entering.

[0053] In other embodiments, the interconnection between the fourth solder pad 51 and the first solder pad 14 can also be achieved by hybrid bonding, solder bonding or self-growing interconnection; the filler layer 60 can also be achieved by compression molding process, in which solid EMC particles are heated to 175-200°C to melt and then injected into the mold cavity under pressure for filling.

[0054] refer to Figure 9 As shown, in this embodiment, the heat dissipation structure adopts a heat sink 70, such as a heat sink. A thermal interface material (TIM) is coated on the back of the first functional core 21. The thermal interface material is any one of thermally conductive silicone grease, thermally conductive gel, phase change material, or sintered silver / sintered copper, with a thickness of 20-100 μm, preferably 30-50 μm.

[0055] The heat sink 70 is then attached to the thermal interface material, forming a thermal coupling between the heat sink 70 and the back side of the first functional chip 21 through the thermal interface material. The heat sink 70 can be a heat sink fin or a heat radiator. Since the first functional chip 21 is a logic chip, the heat generated during operation is directly conducted to the heat sink 70 through its back side via the thermal interface material, and dissipated to the external environment along the shortest path, without passing through the second functional chip 30. This avoids the heat from the logic chip heating the storage chip, effectively solving the heat dissipation problem of heat-sensitive memory.

[0056] In another optional embodiment, the heat dissipation structure is a thermally conductive microchannel disposed on the back side of the first functional chip 21, and the thermally conductive microchannel is connected to the outside. Specifically, microchannel channels are formed on the back side of the first functional chip 21 by etching or machining. The channel width is 50-200μm and the depth is 100-300μm. Then, a sealing cover plate is bonded to the back side of the first functional chip 21 to form a closed flow channel. The heat generated by the first functional chip 21 is carried away by the flow of a cooling medium (such as water or refrigerant) within the microchannel, achieving efficient heat dissipation. This method is particularly suitable for logic chips with high heat generation power.

[0057] Both of the above heat dissipation structures can achieve independent heat dissipation for the first functional core 21, preventing heat from being transferred downwards to the second functional core 30.

[0058] S4: Thin the side of the intermediate structure away from the stacking unit 50 to expose the bottom end of the through silicon via 13; form conductive bumps 81 at the exposed bottom end of the through silicon via 13; align and press the intermediate structure with the external multilayer substrate 80 so that each conductive bump 81 is electrically connected to the corresponding conductive pad on the multilayer substrate 80. refer to Figure 11 As shown, the side of the intermediate structure facing away from the stacking unit 50 is thinned by mechanical grinding and / or chemical mechanical polishing (CMP) until the bottom end of the through silicon via 13 is exposed on the back side of the intermediate structure, so as to achieve electrical communication between the through silicon via 13 and the outside.

[0059] Conductive bumps 81 are formed at the bottom end of the exposed through-silicon via 13. The conductive bumps 81 are electrically connected to the first pad 14 through the through-silicon via 13 inside the intermediate structure. The conductive bumps 81 are made of solder (such as SnAgCu alloy), copper, or gold, and can be formed by ball-planting, electroplating, or printing.

[0060] An external multilayer carrier board 80 is provided, the external multilayer carrier board 80 has multiple wiring layers inside, and the side of the external multilayer carrier board 80 facing the intermediate structure has conductive pads, the conductive pads corresponding one-to-one with the conductive bumps 81.

[0061] The intermediate structure with completed conductive protrusions 81 is placed in a pressing device, so that each conductive protrusion 81 on the back of the intermediate structure is perpendicularly aligned with the corresponding conductive pad on the external multilayer carrier board 80, with an alignment accuracy of ±5-15μm; then, thermosetting bonding (or reflow soldering) is performed to achieve pressing, with a pressing temperature of 150-250℃ (200-260℃ for reflow soldering), a pressing pressure of 5-30N / cm², and a holding time of 5-30 minutes. Under the combined action of pressure and temperature, the conductive protrusions 81 and the conductive pads form a metallurgical bond, realizing the mechanical fixation and electrical connection between the intermediate structure and the external multilayer carrier board 80.

[0062] Through the aforementioned lamination process, the chip stack interconnect system can be directly mounted on a printed circuit board or package substrate using surface mount technology, enabling signal transmission and power distribution with external systems. The conductive bumps 81 provide a standardized external interface, allowing the entire chip stack structure to be easily integrated into end-user electronic systems.

[0063] S5: The composite wafer 20 array that has been pressed is diced to obtain independent chip stacked interconnect packaging units.

[0064] After the interposer structure and the external multilayer substrate 80 are bonded together, a composite wafer array 20 is formed. The composite wafer array 20 is then diced using the same dicing process as in step S2. During dicing, the external multilayer substrate 80 and the interposer structure are sequentially cut through, separating each stacked unit 50 one by one to form multiple independent die-stacked interconnect packaging units. After dicing, the diced packaging units are cleaned to remove debris and contaminants generated during dicing.

[0065] After the above-described cutting and separation, each independent chip stack interconnect package unit includes: an interposer structure, stacked units 50 (including a first functional chip 21 and a second functional chip 30) stacked within the recess 11 of the interposer structure, a filler layer 60 filling the gaps inside the recess 11, a heat sink 70 disposed on the back side of the first functional chip 21, conductive bumps 81 disposed on the side of the interposer structure opposite to the stacked units 50, and an external multilayer carrier board 80 portion pressed onto the back side of the interposer structure. Each package unit is independent of each other and can be directly used for subsequent surface mounting or system integration.

[0066] refer to Figure 11 As shown, the present invention also provides a chip stack interconnect system, comprising: An intermediate structure includes a substrate 10, a groove 11 is formed on the upper surface of the substrate 10, a through-silicon via 13 is formed in the substrate 10, and a first bonding pad 14 is disposed on the upper surface of the substrate 10. A stacking unit 50 is disposed on the intermediate structure. The stacking unit 50 includes a first functional core 21 and a second functional core 30 disposed below the first functional core 21. The coverage area of ​​the first functional core 21 is larger than the coverage area of ​​the second functional core 30. The second functional core 30 is located within the groove 11, and the peripheral portion of the first functional core 21 is supported on the top periphery of the groove 11. A second solder pad 22 and a third solder pad 31 are respectively disposed on the opposite surface of the first functional core 21 and the second functional core 30. The second functional core 30 is bonded to the first functional core 21, such that the second solder pad 22 and the third solder pad 31 are vertically aligned and bonded together to achieve electrical interconnection between them. A fourth solder pad 51 is disposed on the surface of the first functional core 21 opposite to the intermediate structure. The fourth solder pad 51 is aligned with and electrically interconnected with the first solder pad 14. A filling layer 60 fills the gap inside the groove 11, and its surface is coplanar with the surface of the first functional core 21. A heat dissipation structure is disposed on the back side of the first functional core 21; and Conductive bumps 81 are disposed on the side of the intermediate structure opposite to the stacking unit 50 and are electrically connected to the bottom end of the through silicon via 13 for interconnection with an external circuit board.

[0067] Example 2

[0068] refer to Figure 12 As shown, the difference between this embodiment and Embodiment 1 is that N functional cores are continued to be bonded and interconnected on the back side of the first functional core 21, where N≥1.

[0069] Specifically, after the stacking unit 50 of the first functional core 21 and the second functional core 30 is embedded in the intermediate structure groove 11, a third functional core is stacked on the back side of the first functional core 21 using dry film bonding and self-growing interconnection processes. The third functional core is perpendicularly interconnected with the first functional core 21, forming a three-layer stacked structure consisting of the second functional core 30, the first functional core 21, and the third functional core, starting from the intermediate structure. The heat dissipation structure can be disposed on the back side of the first functional core 21, allowing the heat generated by the first functional core 21 to be dissipated through the heat dissipation structure. In another arrangement, the heat dissipation structure can also be set on the back of the third functional core 90, so that the heat generated by the uppermost third functional core 90 is directly conducted to the heat sink 70 through its back, and the heat of the lower core is conducted through the intermediate structure to achieve graded heat dissipation.

[0070] By stacking a third functional chip on top of the first functional chip 21, forming a three- or more-layer stacked structure, more functional chips can be vertically integrated within the same area occupied by the intermediate structure, significantly improving system integration density and functional density. Specifically, the second functional chip 30 is a storage chip embedded in the intermediate structure groove 11; the first functional chip 21 is a logic chip serving as the main heat source; and the third functional chip is an I / O chip or an additional storage chip.

[0071] Regarding the arrangement of the heat dissipation structure: A heat dissipation structure is provided on the back side of the first functional core 21, that is, between the first functional core 21 and the third functional core 90, so that the heat generated by the first functional core 21 is dissipated through this heat dissipation structure and is not transferred downward to the second functional core 30. In another arrangement, the heat dissipation structure can also be provided on the back side of the third functional core 90, so that the heat generated by the uppermost third functional core 90 is directly conducted to the heat sink 70 for dissipation through its back side, while the heat of the first functional core 21 is conducted downward through the intermediate structure, achieving tiered heat dissipation.

[0072] When stacking continues, the bonding interconnection method between each layer of core particles is the same as the bonding interconnection method between the first functional core particle 21 and the second functional core particle 30 in Example 1. It can be achieved by permanent dry film bonding combined with selective deposition self-growth interconnection, or by hybrid bonding, hot-press bonding or solder bonding.

[0073] N functional cores, where N≥1, can be bonded and interconnected on the back side of the first functional core 21 to form a multilayer stacked structure. The bonding and interconnection method between the layers is the same as the bonding and interconnection method between the first functional core 21 and the second functional core 30 in Embodiment 1.

[0074] Example 3

[0075] The difference between this embodiment and Embodiment 1 is that the interconnection between the fourth pad 51 and the first pad 14 adopts a self-growing interconnection method instead of thermo-press bonding.

[0076] In step S3, after the stacked unit 50 is embedded in the intermediate structure groove 11, a gap of 1-10 μm is maintained between the fourth pad 51 and the first pad 14, instead of direct contact; then, an interconnect metal pillar 40 is formed between the fourth pad 51 and the first pad 14 by chemical plating.

[0077] Specifically, a palladium activation solution (palladium chloride 0.1-0.5 g / L, hydrochloric acid 1-5 ml / L) is supplied to the area of ​​the fourth pad 51 and the first pad 14 through the gap between the sidewall of the groove 11 and the first functional core 21. The mixture is treated at room temperature for 1-3 minutes to form palladium catalytic centers on the surfaces of the fourth pad 51 and the first pad 14. Subsequently, a chemical copper plating solution (copper sulfate pentahydrate 10 g / L, EDTA 30 g / L, formaldehyde 8 ml / L, pH 12.5, temperature 60°C) is supplied, causing copper ions to be reduced and deposited on the surface of the pads with catalytic centers. These ions grow towards each other from the surfaces of the fourth pad 51 and the first pad 14, bridging to form complete interconnect metal pillars 40 after 30-60 minutes. Finally, the surface is rinsed with deionized water and annealed in a nitrogen atmosphere at 150-250°C for 30-60 minutes. This variation does not require precise control of the contact height between the fourth pad 51 and the first pad 14, and has lower requirements for bonding pressure, making it suitable for pressure-sensitive devices.

[0078] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, such as the combination of technical features between embodiments, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method of manufacturing a core particle stack interconnect system, characterized by, Includes the following steps: S1: A substrate is provided, a groove is etched on the upper surface of the substrate, and a through-silicon via is formed in the substrate; a first bonding pad is disposed on the upper surface of the substrate to obtain an intermediate structure; S2: A wafer and at least one second functional chip are provided. A plurality of first functional chips are pre-formed on the wafer. Second and third bonding pads are formed on opposite sides of the first and second functional chips, respectively. The second functional chips are interconnected and bonded to the first functional chips, with the second and third bonding pads perpendicularly aligned and bonded to achieve electrical interconnection between them. The interconnected wafer is diced to obtain independent stacking units. In the stacking units, the coverage area of ​​the first functional chips is larger than the coverage area of ​​the second functional chips. S3: The stacked unit is placed in the groove of the intermediate structure; wherein the second functional core is located in the groove, and the peripheral portion of the first functional core is supported on the top periphery of the groove; a fourth solder pad is provided on the surface of the first functional core, which is perpendicularly aligned with the first solder pad and electrically interconnected; a filling layer is formed in the groove, such that the filling layer fills the gap inside the groove and its surface is coplanar with the surface of the first core; a heat dissipation structure is provided on the back side of the first functional core; S4: Thin the side of the intermediate structure away from the stacking unit to expose the bottom end of the through silicon via; form conductive bumps at the exposed bottom end of the through silicon via; align and press the intermediate structure with the external multilayer substrate to electrically connect each conductive bump to the corresponding conductive pad on the multilayer substrate. S5: Divide the composite wafer array after lamination to obtain independent chip stacked interconnect packaging units.

2. The method of claim 1, wherein, The first functional chip is a logic chip, and the second functional chip is a storage chip.

3. The method of claim 1, wherein, The fourth pad of the first functional core is disposed in the peripheral area of ​​its bonding surface, and the first pad on the substrate surface is disposed around the groove and perpendicularly corresponding to the fourth pad.

4. The method of claim 1, wherein, It also includes bonding and interconnecting N functional cores on the back side of the first functional core, where N≥1.

5. The method of claim 1, wherein, The bonding layer is a permanent dry film, which is laminated to the bonding surface of the second functional chip and / or the wafer by hot rolling, and the second functional chip is mechanically fixed to the wafer by hot pressing.

6. The method of claim 1, wherein, The depth of the groove matches the thickness of the second functional core, so that after the stacking unit is embedded, the back side of the second functional core contacts or is close to the bottom of the groove, and the top surface of the second functional core is flush or substantially flush with the surface of the stacking unit.

7. The method of claim 1, wherein, After the stacked unit is embedded in the groove, it further includes: The gap between the groove sidewall and the second functional core and the peripheral area of ​​the first functional core are filled with a medium, such that the filling medium is flush with the back substrate of the first functional core.

8. The method of claim 1, wherein, It also includes decoupling capacitors, resistors, inductors, and thermoelectric coolers that are electrically connected to the through-silicon vias within the substrate; The decoupling capacitor is arranged close to the first functional core and / or the second functional core. The resistor and / or inductor, together with the decoupling capacitor, form a filter network for the power distribution network, used to suppress noise in the power distribution network. The thermoelectric cooler is electrically connected to an external power source through an independent through-silicon via and is located between the bottom of the groove and the second functional chip to provide active cooling for the second functional chip.

9. The method of claim 1 or 4, wherein, The bonding interconnection method is any one of hybrid bonding, thermo-press bonding, solder bonding, or self-growing interconnection.

10. The method of claim 9, wherein, The self-grown interconnects are achieved through a selective deposition process, which may be electroless plating, electroplating, chemical vapor deposition, or atomic layer deposition.

11. The method of claim 1, wherein, The heat dissipation structure is a thermally conductive microchannel disposed on the back side of the first functional core, and the thermally conductive microchannel is connected to the outside; or, A heat sink is located on the back of the first functional core.

12. A core particle stack interconnect system, characterized by, include: An intermediate structure includes a substrate, a groove is formed on the upper surface of the substrate, a through-silicon via is formed in the substrate, and a first bonding pad is disposed on the upper surface of the substrate; A stacking unit is disposed on the intermediate structure. The stacking unit includes a first functional core and a second functional core disposed below the first functional core. The coverage area of ​​the first functional core is larger than that of the second functional core. The second functional core is located within the groove, and the peripheral portion of the first functional core is supported by the top periphery of the groove. A second solder pad and a third solder pad are respectively disposed on the opposite surfaces of the first and second functional cores. The second functional core is bonded to the first functional core, such that the second and third solder pads are vertically aligned and bonded together to achieve electrical interconnection between them. A fourth solder pad is disposed on the surface of the first functional core opposite to the intermediate structure. The fourth solder pad is aligned with and electrically interconnected with the first solder pad. A filling layer fills the gap inside the groove, and its surface is coplanar with the surface of the first functional core particle; A heat dissipation structure is disposed on the back side of the first functional core. as well as Conductive bumps are disposed on the side of the intermediate structure opposite to the stacked unit and electrically connected to the bottom end of the through-silicon via for interconnection with an external circuit board.