Partitioned gradient composite seed layer for glass substrate and magnetron sputtering preparation method thereof

Abstract

The application provides a partition gradient composite seed layer for a glass substrate and a magnetron sputtering preparation method thereof. The seed layer comprises two first composite layers, which are arranged on the upper and lower surfaces of the glass substrate and block the upper and lower openings of the TGV through hole, and sequentially comprise a Ru-TiW gradient composite barrier layer and an ultrathin Cu seed layer from the glass surface; two second composite layers, which are located on the axial upper and lower sections of the inner surfaces of the first composite layers, and sequentially comprise an ultrathin TiW transition layer and a gradient Cu seed layer; and a third composite layer, which is located between the two second composite layers and is adjacent to the two second composite layers, and comprises a thick Cu seed layer and a Ti nucleation layer located between the thick Cu seed layer and the gradient Cu seed layer. The second and third composite layers are not in direct contact with the glass. During preparation, the upper and lower first composite layers are synchronously deposited, the upper and lower second composite layers are synchronously deposited, and finally, the third composite layer is deposited. The application greatly reduces the high-frequency signal insertion loss, the plating filling yield is greater than or equal to 99.5%, the adhesion is increased to more than 70 MPa, and is suitable for 5G / 6G radio frequency and Chiplet packaging.

Description

Technical Field

[0001] This invention relates to the field of semiconductor packaging, specifically to a partitioned gradient composite seed layer for glass substrates and its preparation method by magnetron sputtering. Background Technology

[0002] Through-Glass Vias (TGVs) are core vertical interconnect structures in advanced packaging, 5G / 6G RF, MEMS, and optoelectronic integration. The performance of their seed layer directly determines the yield of copper plating, interconnect reliability, and high-frequency signal transmission quality. In the core TGV process, the seed layer (typically a Ti / Cu or Ta / Cu stack) deposited by magnetron sputtering is the foundation for copper plating and must simultaneously meet three core requirements: low-loss transmission in high-frequency scenarios, preventing copper atoms from diffusing into the glass substrate, and ensuring uniform nucleation and void-free filling of the plated copper. Current TGV seed layer fabrication technology faces significant industry-wide challenges. Existing patents and mass production processes all employ a "uniform stacked structure across the entire surface," which cannot simultaneously address the aforementioned three requirements. Specific pain points are as follows: The contradiction between high-frequency loss and blocking performance: In 5G / 6G millimeter-wave RF scenarios, traditional Ti adhesive layers are ferromagnetic, which will generate severe eddy current loss and increase signal insertion loss. If non-magnetic high blocking materials such as Ta and TiW are used, although eddy current loss can be reduced and copper diffusion blocking performance can be improved, the high resistivity of the material itself will increase the series resistance, further aggravating high-frequency loss. Moreover, thickening the blocking layer will further exacerbate this contradiction.

[0003] The contradiction between barrier properties and electroplating nucleation: To prevent the diffusion of copper atoms, existing processes often use Ta or TiW layers with high barrier properties at the bottom of the hole. However, these materials have low surface energy, which makes it difficult for the initial nucleation of electroplated copper, and easily leads to defects such as voids at the bottom of the hole and peeling of the plating. This problem is more prominent in high aspect ratio TGV with a depth-to-width ratio of 1:30 or higher, which seriously reduces the electroplating yield.

[0004] The contradiction between over-plating at the orifice and thickness uniformity: The "line-of-sight deposition" characteristic of magnetron sputtering causes the seed layer thickness at the orifice of TGV through-holes to be generally higher than that at the bottom of the hole and in the planar area, resulting in over-plating at the orifice. This not only increases the difficulty of subsequent CMP planarization, but may also lead to electroplating sealing and the formation of internal voids. The existing seed layer with a uniform structure cannot alleviate this problem through material optimization, but can only be improved by fine-tuning the process parameters, with limited effect.

[0005] Existing TGV seed layer patents either focus on replacing a single material (such as replacing Ti with Ru) or optimizing sputtering parameters to improve in-hole coverage. None of them address the differentiated functional requirements of different regions of TGV vias (orifice / plane, wall, bottom), failing to fundamentally resolve the aforementioned performance contradictions. Furthermore, the insulating properties of glass substrates and the difficulty of depositing high aspect ratio vias also prevent the direct transfer of TSV (Through Silicon Via) partitioned seed layer technology to TGV processes.

[0006] Furthermore, with the rapid development of Chiplet technology, TGV glass interlayers are moving towards higher density, higher frequency, and thinner dimensions, which places higher demands on the seed layer's thickness uniformity, high frequency low loss, and high reliability. Existing seed layers with a uniform structure can no longer meet the mass production requirements of high-end TGV.

[0007] Therefore, it is necessary to provide a partitioned gradient composite seed layer for glass substrates and a method for preparing it by magnetron sputtering. Summary of the Invention

[0008] The present invention provides a partitioned gradient composite seed layer for glass substrates and its preparation method by magnetron sputtering, which effectively solves the problems of poor performance and high loss of existing seed layers.

[0009] The technical solution adopted in this invention is as follows: a partitioned gradient composite seed layer for a glass substrate, comprising: two first composite layers, which are respectively disposed on the upper and lower surfaces of the glass substrate, and respectively cover and block the upper and lower openings of the TGV via; each first composite layer consists of a Ru-TiW gradient composite barrier layer and an ultrathin Cu seed layer from the covered glass surface inwards; two second composite layers, which are respectively located on the inner surfaces of the two first composite layers and respectively correspond to the upper and lower axial sections of the TGV via; each second composite layer consists of an ultrathin TiW transition layer and a gradient Cu seed layer from the inner surface of the first composite layer; and a third composite layer, which is located between the two second composite layers and adjacent to the two second composite layers, the third composite layer comprising a thick Cu seed layer and a Ti nucleation layer located between the thick Cu seed layer and the gradient Cu seed layer. Neither the second composite layer nor the third composite layer is in direct contact with the glass substrate.

[0010] Furthermore, the Ru-TiW gradient composite barrier layer has a thickness of 15-30 nm, and from the side near the glass interface to the surface, the Ru content gradually decreases from 60% to 20%, while the TiW content gradually increases from 40% to 80%; the thickness of the ultrathin Cu seed layer is 20-50 nm; the total thickness of the first composite layer is 30-80 nm; the ultrathin Cu seed layer does not contact the glass substrate; and the Ru-TiW gradient composite barrier layer is disposed on the glass substrate.

[0011] Furthermore, the thickness of the ultrathin TiW transition layer is 5~10nm; the thickness of the gradient Cu seed layer is 20~120nm.

[0012] Furthermore, the thickness of the Ti nucleation layer is 5~10nm, and the thickness of the thick Cu seed layer is 80~120nm.

[0013] Furthermore, the aperture of the TGV via is 20~100μm, and the aspect ratio is 5:1~50:1; the glass substrate is borosilicate glass, alkali-free glass, or quartz glass, and the wafer size is 8 inches or 12 inches.

[0014] A magnetron sputtering method for preparing a partitioned gradient composite seed layer for a glass substrate includes the following steps: S1, substrate pretreatment: After cleaning and drying the glass wafer with through-hole TGV vias formed, it is placed in a multi-target co-sputtering cavity and evacuated to ≤5×10⁻⁶. -4S1. Introduce working gas and adjust the working gas pressure to 0.3~0.8Pa; S2. Simultaneously deposit the first composite layer: On the upper surface of the glass wafer and the upper opening of the TGV via, and on the lower surface of the glass wafer and the lower opening of the TGV via, simultaneously deposit a Ru-TiW gradient composite barrier layer using Ru target and TiW target co-sputtering, and then simultaneously deposit an ultrathin Cu seed layer using Cu target DC magnetron sputtering, forming two first composite layers, and simultaneously sealing the upper opening and the lower opening; During the deposition process, the wafer tilt angle is ≤15°, the rotation speed is 5~8rpm, and the substrate bias voltage is -20~-50V; S3. Simultaneously deposit the second composite layer: On the inner surfaces of the two first composite layers, corresponding to the axial upper and axial lower sections of the TGV via, simultaneously deposit an ultrathin TiW transition layer using TiW target RF magnetron sputtering, and then simultaneously deposit a gradient Cu seed layer using Cu target DC magnetron sputtering. S4. Depositing the third composite layer: Between the two second composite layers, a Ti nucleation layer is deposited by DC magnetron sputtering with a Ti target, and then a thick Cu seed layer is deposited by DC magnetron sputtering with a Cu target to form the third composite layer, so that the third composite layer is adjacent to the two second composite layers respectively; During the deposition process, the wafer tilt angle is 60°, the rotation speed is 5~8 rpm, the substrate bias is -80~-100V, and the deposition is terminated after the thickness reaches the target value by the online thickness measurement module; S5. Post-processing: Keep the cavity vacuum and Ar atmosphere cooled to room temperature, take out the wafer, rinse with deionized water overflow, blow dry with nitrogen and bake at 80°C for 1~2 min to complete the seed layer preparation.

[0015] Specifically, the parameters for co-sputtering Ru and TiW targets in S2 are as follows: in the initial stage, the Ru target power is 8~10kW and the TiW target power is 4~5kW; in the gradient transition stage, the Ru target power gradually decreases to 2~3kW and the TiW target power gradually increases to 8~10kW; in the final stage, the Ru target power is 2~3kW and the TiW target power is 8~10kW, with a total deposition time of 10~15min; when depositing an ultrathin Cu seed layer, the Cu target power is 6~8kW and the deposition time is 2~3min.

[0016] Specifically: The specific parameters for depositing the ultrathin TiW transition layer in S3 are: TiW target power 4~6kW, deposition time 1~2min; The specific parameters for depositing the gradient Cu seed layer are: Cu target power 6~7kW in the initial stage, then gradually increased to 10~12kW, total deposition time 12~15min.

[0017] Specifically: the parameters for depositing the Ti nucleation layer in S4 are: Ti target power 3~5kW, deposition time 1~2min; the parameters for depositing the thick Cu seed layer are: Cu target power 10~12kW, working gas pressure 0.4~0.6Pa, deposition time 8~10min.

[0018] Specifically, the multi-target co-sputtering cavity is equipped with four independent target sites respectively loaded with Ru target, TiW target, Ti target and Cu target, and the purity of each target material is 99.999%.

[0019] Beneficial effects of the invention: 1. By replacing the traditional Ti layer with a Ru-TiW non-magnetic gradient blocking layer, the signal insertion loss of TGV interconnect is reduced by more than 42% and crosstalk is reduced by 35% in the 28GHz millimeter wave band. It is fully compatible with the needs of 5G / 6G RF TR components and high-frequency millimeter wave devices, solving the industry pain point of high high-frequency loss of traditional seed layers.

[0020] 2. Ultra-thin Ti nucleation layer solves the problem of difficult nucleation of high barrier materials. The yield of void-free electroplating of TGV with a depth-to-width ratio of 1:50 is ≥99.5%, and the electroplating yield of TGV with a depth-to-width ratio of 1:30 is ≥99.8%. Compared with the existing traditional process (yield of 82%~88%), the yield is improved by more than 11%, and the mass production cost is significantly reduced.

[0021] 3. Through partitioned gradient design and dynamic parameter control, the seed layer thickness deviation between the orifice and the plane is ≤10%, and the thickness gradient between the orifice wall and the orifice bottom is uniform. This structurally avoids the problems of over-plating and electroplating sealing at the orifice. After CMP, the wafer surface flatness is improved by more than 80%, the depth of the dish pit is ≤5nm, and it is compatible with micro-bump flip-chip bonding process with a pitch of ≤10μm.

[0022] 4. The copper diffusion barrier performance of the Ru-TiW gradient barrier layer is superior to that of traditional Ti and Ta layers. The copper diffusion coefficient is reduced by an order of magnitude, the seed layer adhesion is increased from the conventional 30MPa to more than 70MPa, and there are no defects such as film peeling and pinholes. The long-term operating reliability of TGV devices is improved by more than 60%.

[0023] 5. Strong process compatibility and controllable cost: No new special equipment is required. Only target modification and parameter adjustment of existing multi-target magnetron sputtering equipment are required to adapt to the existing TGV mass production line. The materials used are all commonly used materials in semiconductor mass production. The cost is the same as that of traditional seed layers. No additional mask or etching process is required. The process complexity is reduced by 30% and the mass production efficiency is increased by 25%.

[0024] 6. Outstanding novelty and inventiveness: It is the first to propose a partitioned gradient composite design of "three zones and three structures" for TGV vias, which is different from the existing design concept of a unified stacked seed layer. It is specifically designed for the functional requirements of different regions of TGV vias, solving the performance contradictions that existing technologies cannot address simultaneously. It does not overlap with existing patents and has obvious novelty and inventiveness. At the same time, it is adapted to the insulating characteristics of glass substrates, which is different from the partitioned seed layer technology of TSV and cannot be directly transferred, further highlighting its innovation. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a partitioned gradient composite seed layer for a glass substrate provided in an embodiment of this application.

[0026] Figure 2 A flowchart illustrating the preparation method provided in the embodiments of this application.

[0027] The following are labeled in the figure: 1. Ru-TiW gradient composite barrier layer; 2. Ultrathin Cu seed layer; 3. Ultrathin TiW transition layer; 4. Gradient Cu seed layer; 5. Ti nucleation layer; 6. Thick Cu seed layer; 7. Glass substrate; 700. TGV via. Detailed Implementation

[0028] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0029] like Figure 1 As shown, the first embodiment provided in this application is a partitioned gradient composite seed layer for a glass substrate, comprising: two first composite layers, which are respectively disposed on the upper and lower surfaces of the glass substrate 7, and respectively cover and block the upper and lower openings of the TGV via 700; each first composite layer, from the covered glass surface inward, consists of a Ru-TiW gradient composite barrier layer 1 and an ultrathin Cu seed layer 2; two second composite layers, which are respectively located on the inner surfaces of the two first composite layers and respectively correspond to the upper and lower axial sections of the TGV via 700; each second composite layer, from the inner surface of the first composite layer, consists of an ultrathin TiW transition layer 3 and a gradient Cu seed layer 4; and a third composite layer, which is located between the two second composite layers and adjacent to the two second composite layers, the third composite layer including a thick Cu seed and a Ti nucleation layer 5 located between the thick Cu seed layer 6 and the gradient Cu seed layer 4. Neither the second composite layer nor the third composite layer is in direct contact with the glass substrate 7.

[0030] The above design employs an axially symmetrical structure with "end-to-end sealing and inner wall partitioning." Two first composite layers are deposited on the upper and lower surfaces of the glass substrate 7 and at the upper and lower openings of the TGV via 700, respectively. A dense Ru-TiW gradient barrier layer and an ultrathin Cu layer are formed by magnetron sputtering, completely sealing both ends of the via and forming a closed cavity. The Ru-TiW gradient design in the first composite layer results in a high Ru content near the glass, enhancing adhesion through Ru-O-Si chemical bonds; the high TiW content near the cavity provides excellent copper diffusion barrier performance. Two second composite layers are deposited on the upper and lower axial sections of the inner surface of the first composite layer, respectively. An ultrathin TiW transition layer 3 assists in adhesion, and the gradient Cu seed layer 4 gradually increases in thickness from both ends to the middle, compensating for the attenuation of atomic flux during deep-hole sputtering and ensuring uniform conductivity of the hole wall. A third composite layer is located in the middle axial section, connecting the upper and lower second composite layers. Its Ti nucleation layer 5 (high surface energy) promotes initial copper nucleation, and the thick Cu seed layer 6 provides a low-resistance conductive path. In the entire structure, the second and third composite layers do not come into direct contact with the glass, thus avoiding the interface defects between the traditional barrier layer and the glass and avoiding the risk of interface peeling.

[0031] Specifically: the Ru-TiW gradient composite barrier layer 1 has a thickness of 15~30nm, and from the side near the glass interface to the surface layer (the side of TGV via 700), the Ru content gradually decreases from 60% to 20%, and the TiW content gradually increases from 40% to 80%; the thickness of the ultrathin Cu seed layer 2 is 20~50nm; the total thickness of the first composite layer is 30~80nm, the ultrathin Cu seed layer 2 does not contact the glass substrate 7, and the Ru-TiW gradient composite barrier layer 1 is disposed on the glass substrate 7.

[0032] In the above design, the Ru-TiW gradient barrier layer of the first composite layer exhibits a continuous change in composition along its thickness. From the interface with the glass to the surface layer in contact with the ultrathin Cu seed layer 2, the Ru content gradually decreases from 60% to 20%, while the TiW content increases from 40% to 80%. This gradient design improves the adhesion of Ru to the glass (forming Ru-O-Si bonds) and the high copper diffusion barrier capability of TiW. The ultrathin Cu seed layer 2 does not directly contact the glass but grows on the surface layer with a high TiW content, ensuring both low electrical resistance and avoiding the adhesion degradation and diffusion problems caused by direct Cu contact with the glass.

[0033] Specifically: the thickness of the ultrathin TiW transition layer 3 is 5~10nm; the thickness of the gradient Cu seed layer 4 is 20~120nm.

[0034] In the above design, the ultrathin TiW transition layer 3 is extremely thin, mainly serving as an auxiliary adhesion and barrier; due to its small thickness, its contribution to resistance is negligible. The gradient Cu seed layer 4 has a thickness range of 20~120nm, which can adapt to the conductivity requirements of TGV vias 700 with different aspect ratios: for shallow holes or near the hole opening, a thinner layer (around 20nm) is sufficient for conductivity; for deep holes or near the hole bottom, a thicker layer (120nm) is required to compensate for insufficient sputtering atomic flux and ensure the overall conductivity continuity of the hole wall. This thickness range covers the requirements of mainstream TGV applications, while avoiding excessive thickness leading to over-plating at the hole opening or excessive thinness leading to film breakage.

[0035] Specifically, the thickness of the Ti nucleation layer 5 is 5~10nm, and the thickness of the thick Cu seed layer 6 is 80~120nm.

[0036] In the above design, the Ti nucleation layer 5 is located between the thick Cu and the gradient Cu layers. Utilizing the high surface energy of Ti, a large number of uniform nucleation sites are provided at the interface between the two Cu layers, allowing the electroplated copper to grow from the middle towards both ends, avoiding voids. This leverages the high surface energy of Ti (approximately 1.7 J / m²). 2 The Ti layer forms a highly active nucleation interface between the thick Cu layer and the gradient Cu layer, promoting uniform initial nucleation of electroplated copper in the axial mid-section region. Simultaneously, the thick Cu layer provides a low-resistance conductive path, ensuring uniform distribution of the electroplating current. The Ti layer, located between the two Cu layers, effectively blocks Cu atom diffusion and prevents interface oxidation. Furthermore, the Ti nucleation layer 5 acts as an intermediate barrier layer, effectively preventing inter-diffusion of atoms between the thick Cu layer and the gradient Cu layer, maintaining interface purity and improving long-term reliability.

[0037] Specifically: the diameter of the TGV via 700 is 20~100μm, and the aspect ratio is 5:1~50:1; the glass substrate 7 is borosilicate glass, alkali-free glass or quartz glass, and the wafer size is 8 inches or 12 inches.

[0038] In the above design, the glass substrate 7 is selected from borosilicate glass, alkali-free glass or quartz glass. These materials have low dielectric constant (ε≈4~6), low loss tangent (tanδ≈0.001~0.005), and their thermal expansion coefficient matches that of silicon chips. The aperture of the TGV via 700 covers the current mainstream mass production specifications of TGV and has broad industrial application value.

[0039] like Figure 2 As shown, the second embodiment provided in this application is a magnetron sputtering preparation method for preparing the partitioned gradient composite seed layer for the glass substrate, including the following steps: S1, substrate pretreatment: After cleaning and drying the glass wafer with the through-hole TGV 700 formed, it is placed in a multi-target co-sputtering cavity and evacuated to ≤5×10 -4S1. Introduce working gas and adjust the working gas pressure to 0.3~0.8Pa; S2. Simultaneously deposit the first composite layer: On the upper surface of the glass wafer and the upper opening of the TGV via 700, and on the lower surface of the glass wafer and the lower opening of the TGV via 700, simultaneously deposit a Ru-TiW gradient composite barrier layer 1 using Ru target and TiW target co-sputtering, and then simultaneously deposit an ultrathin Cu seed layer 2 using Cu target DC magnetron sputtering, forming two first composite layers, and simultaneously sealing the upper opening and the lower opening; During the deposition process, the wafer tilt angle is ≤15°, the rotation speed is 5~8rpm, and the substrate bias voltage is -20~-50V; S3. Simultaneously deposit the second composite layer: On the inner surfaces of the two first composite layers, corresponding to the axial upper section and axial lower section of the TGV via 700, simultaneously deposit an ultrathin TiW transition layer 3 using TiW target RF magnetron sputtering, and then simultaneously deposit a gradient layer 3 using Cu target DC magnetron sputtering. S4. Deposit a Cu seed layer 4 to form two second composite layers. During the deposition process, the wafer tilt angle is 30°~45°, the rotation speed is 10~15rpm, the substrate bias voltage is -50~-80V, and the Cu layer thickness gradient is controlled in real time by an online thickness measurement module. S5. Deposit a third composite layer: Between the two second composite layers, a Ti nucleation layer 5 is deposited by DC magnetron sputtering with a Ti target, and then a thick Cu seed layer 6 is deposited by DC magnetron sputtering with a Cu target to form the third composite layer, so that the third composite layer is adjacent to the two second composite layers respectively. During the deposition process, the wafer tilt angle is 60°, the rotation speed is 5~8rpm, the substrate bias voltage is -80~-100V, and the deposition is terminated after confirming that the thickness has reached the target value by an online thickness measurement module. S6. Post-processing: Keep the cavity vacuum and Ar atmosphere cooled to room temperature, take out the wafer, rinse with deionized water overflow, blow dry with nitrogen, and bake at 80°C for 1~2min to complete the seed layer preparation.

[0040] In the above design, two first composite layers are deposited simultaneously in S2, and two second composite layers are deposited simultaneously in S3. This saves approximately 40% of the process time compared to stepwise deposition, while ensuring top-to-bottom symmetry. The entire process is achieved using magnetron sputtering, eliminating the need for wafer transfer and avoiding contamination and oxidation. In S3 and S4, an online thickness measurement module monitors the Cu layer thickness gradient in real time, ensuring precise control over the thickness of both gradient Cu and thick Cu layers, thus improving repeatability and yield.

[0041] Specifically: The specific parameters for co-sputtering Ru target and TiW target in S2 are as follows: In the initial stage, the Ru target power is 8~10kW and the TiW target power is 4~5kW; in the gradient transition stage, the Ru target power gradually decreases to 2~3kW and the TiW target power gradually increases to 8~10kW; in the final stage, the Ru target power is 2~3kW and the TiW target power is 8~10kW, with a total deposition time of 10~15min; when depositing the ultrathin Cu seed layer 2, the Cu target power is 6~8kW and the deposition time is 2~3min.

[0042] In the above design, a continuous gradient is formed within a thickness of 15-30 nm by dynamically adjusting the power of the Ru and TiW targets, achieving a defect-free interface after deposition. A tilt angle ≤15° ensures perpendicular incidence of sputtered atoms, preferentially depositing them on the upper and lower surfaces and opening edges to form a dense, pore-free sealing layer. The power output range is 8-10 kW → 2-3 kW, and the deposition time is 10-15 min, facilitating parameter transfer for mass production.

[0043] Specifically: the parameters for depositing the ultrathin TiW transition layer 3 in S3 are: TiW target power 4~6kW, deposition time 1~2min; the parameters for depositing the gradient Cu seed layer 4 are: Cu target power 6~7kW in the initial stage, then gradually increased to 10~12kW, total deposition time 12~15min.

[0044] In the above design, the Cu target power is gradually increased from 6~7kW to 10~12kW to form a gradient layer from 20~30nm to 80~120nm, which compensates for the axial atomic flux attenuation. The TiW transition layer sputtering uses 4~6kW power and 1~2min time to obtain a thickness of 5~10nm, which provides adhesion without increasing resistance.

[0045] Specifically: the parameters for depositing the Ti nucleation layer 5 in S4 are: Ti target power 3~5kW, deposition time 1~2min; the parameters for depositing the thick Cu seed layer 6 are: Cu target power 10~12kW, working gas pressure 0.4~0.6Pa, deposition time 8~10min.

[0046] In the above design, a 5-10 nm Ti nucleation layer 5 is obtained in a short time of 1-2 minutes with low power of 3-5 kW, avoiding the increase in resistance due to excessive thickness or the incomplete coverage due to excessive thinness. A Cu seed layer 6 with a thickness of 80-120 nm is obtained in a short time of 8-10 minutes with high power of 10-12 kW, ensuring conductivity while shortening the process cycle.

[0047] Specifically, the multi-target co-sputtering cavity is equipped with four independent target sites respectively loaded with Ru target, TiW target, Ti target and Cu target, and the purity of each target material is 99.999%.

[0048] In the above design, the four independent target sites (Ru, TiW, Ti, Cu) and the dual-axis tilting rotary stage are standard configurations of existing magnetron sputtering equipment, requiring no customization, resulting in low cost. The 99.999% purity avoids the introduction of impurities, ensuring the electrical performance and long-term reliability of the seed layer.

[0049] To verify the beneficial effects of the present invention, two comparative examples were set up.

[0050] Comparative Example 1 uses a conventional Ti / Cu unified stacked seed layer: On the same TGV glass wafer (12-inch alkali-free glass, 50μm aperture, aspect ratio 10:1), a 20nm Ti layer and a 250nm Cu layer are deposited sequentially at a fixed tilt angle of 20°, a rotation speed of 10rpm, and a bias voltage of -30V. Test results show no improvement in insertion loss at 28GHz, a void-free plating yield of only 82%, a thickness deviation between the aperture and the plane as high as 220%, a seed layer adhesion of 30MPa, and a dish pit depth of 82nm after CMP.

[0051] Comparative Example 2 used a Ru / Cu unified stacked seed layer: a 25nm Ru layer and a 250nm Cu layer were deposited at a tilt angle of 15° and a bias voltage of -30V. Test results showed that the insertion loss in the 28GHz band was reduced by only 20%, the electroplating yield was 88%, the thickness deviation was 180%, and the adhesion was 65MPa, but the copper diffusion barrier performance was insufficient. In contrast, Example 1 of this invention, under the same conditions, achieved a 45% reduction in insertion loss, a 99.6% electroplating yield, a thickness deviation ≤8%, an adhesion of 72MPa, and a pit depth of only 3nm, with all performance characteristics significantly superior to the two comparative examples.

[0052] In further detail, it should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A partitioned gradient composite seed layer for a glass substrate, characterized in that: The system includes two first composite layers, which are respectively disposed on the upper and lower surfaces of the glass substrate (7), and the two first composite layers respectively cover and block the upper and lower openings of the TGV via (700). Each first composite layer consists of a Ru-TiW gradient composite barrier layer (1) and an ultrathin Cu seed layer (2) from the glass surface it covers inwards; two second composite layers, which are respectively located on the inner surfaces of the two first composite layers and correspond to the upper and lower axial sections of the TGV via (700) respectively. Each second composite layer consists of an ultrathin TiW transition layer (3) and a gradient Cu seed layer (4) from the inner surface of the first composite layer; and a third composite layer, which is located between the two second composite layers and adjacent to the two second composite layers respectively. The third composite layer includes a thick Cu seed layer (6) and a Ti nucleation layer (5) located between the thick Cu seed layer (6) and the gradient Cu seed layer (4). Neither the second composite layer nor the third composite layer is in direct contact with the glass substrate (7).

2. The partitioned gradient composite seed layer for glass substrates according to claim 1, characterized in that: The Ru-TiW gradient composite barrier layer (1) has a thickness of 15~30nm, and from the side near the glass interface to the surface, the Ru content gradually decreases from 60% to 20%, and the TiW content gradually increases from 40% to 80%; the thickness of the ultrathin Cu seed layer (2) is 20~50nm; the total thickness of the first composite layer is 30~80nm, the ultrathin Cu seed layer (2) does not contact the glass substrate (7), and the Ru-TiW gradient composite barrier layer (1) is disposed on the glass substrate (7).

3. The partitioned gradient composite seed layer for glass substrates according to claim 1, characterized in that: The thickness of the ultrathin TiW transition layer (3) is 5~10nm; the thickness of the gradient Cu seed layer (4) is 20~120nm.

4. The partitioned gradient composite seed layer for glass substrates according to claim 1, characterized in that: The thickness of the Ti nucleation layer (5) is 5~10nm, and the thickness of the thick Cu seed layer (6) is 80~120nm.

5. The partitioned gradient composite seed layer for glass substrates according to claim 1, characterized in that: The TGV via (700) has a diameter of 20~100μm and a depth-to-width ratio of 5:1~50:1; the glass substrate (7) is borosilicate glass, alkali-free glass or quartz glass, and the wafer size is 8 inches or 12 inches.

6. A magnetron sputtering method for preparing a partitioned gradient composite seed layer for a glass substrate as described in any one of claims 1 to 5, characterized in that: The process includes the following steps: S1, substrate pretreatment: After cleaning and drying the glass wafer with the through-hole TGV (700) formed, it is placed in a multi-target co-sputtering chamber and evacuated to ≤5×10. -4 Pa, introduce working gas and adjust the working gas pressure to 0.3~0.8Pa; S2, simultaneously deposit the first composite layer: on the upper surface of the glass wafer and the upper opening of the TGV via (700), on the lower surface of the glass wafer and the lower opening of the TGV via (700), simultaneously deposit a Ru-TiW gradient composite barrier layer (1) using Ru target and TiW target co-sputtering, and then simultaneously deposit an ultrathin Cu seed layer (2) using Cu target DC magnetron sputtering to form two first composite layers, and simultaneously seal the upper opening and the lower opening; during the deposition process, the wafer tilt angle is ≤15°, the rotation speed is 5~8rpm, and the substrate bias voltage is -20~-50V; S3, simultaneously deposit the second composite layer: on the inner surface of the two first composite layers, corresponding to the axial upper section and axial lower section of the TGV via (700), simultaneously deposit an ultrathin TiW transition layer (3) using TiW target RF magnetron sputtering, and then simultaneously deposit an ultrathin TiW transition layer (3) using Cu target DC magnetron sputtering. Deposit a gradient Cu seed layer (4) to form two second composite layers; during the deposition process, the wafer tilt angle is 30°~45°, the rotation speed is 10~15rpm, the substrate bias voltage is -50~-80V, and the Cu layer thickness gradient is controlled in real time by the online thickness measurement module; S4, deposit the third composite layer: between the two second composite layers, a Ti nucleation layer is deposited by DC magnetron sputtering with a Ti target (5), and then a thick Cu seed layer is deposited by DC magnetron sputtering with a Cu target (6) to form the third composite layer, so that the third composite layer is adjacent to the two second composite layers respectively; during the deposition process, the wafer tilt angle is 60°, the rotation speed is 5~8rpm, the substrate bias voltage is -80~-100V, and the deposition is terminated after confirming that the thickness reaches the target value by the online thickness measurement module; S5, post-processing: keep the cavity vacuum and Ar atmosphere cooled to room temperature, take out the wafer, rinse with deionized water overflow, blow dry with nitrogen and bake at 80°C for 1~2min to complete the seed layer preparation.

7. The magnetron sputtering preparation method according to claim 6, characterized in that: The specific parameters for co-sputtering Ru and TiW targets in S2 are as follows: in the initial stage, the Ru target power is 8~10kW and the TiW target power is 4~5kW; in the gradient transition stage, the Ru target power gradually decreases to 2~3kW and the TiW target power gradually increases to 8~10kW; in the final stage, the Ru target power is 2~3kW and the TiW target power is 8~10kW, and the total deposition time is 10~15min; when depositing the ultrathin Cu seed layer (2), the Cu target power is 6~8kW and the deposition time is 2~3min.

8. The magnetron sputtering preparation method according to claim 6, characterized in that: The specific parameters for depositing the ultrathin TiW transition layer (3) in S3 are: TiW target power 4~6kW, deposition time 1~2min; the specific parameters for depositing the gradient Cu seed layer (4) are: Cu target power 6~7kW in the initial stage, then gradually increased to 10~12kW, and total deposition time 12~15min.

9. The magnetron sputtering preparation method according to claim 6, characterized in that: The specific parameters for depositing the Ti nucleation layer (5) in S4 are: Ti target power 3~5kW, deposition time 1~2min; the specific parameters for depositing the thick Cu seed layer (6) are: Cu target power 10~12kW, working gas pressure 0.4~0.6Pa, deposition time 8~10min.

10. The magnetron sputtering preparation method according to claim 6, characterized in that: The multi-target co-sputtering cavity is equipped with four independent target sites, each containing a Ru target, a TiW target, a Ti target, and a Cu target, with each target material having a purity of 99.999%.

Images