A substrate with bumps, a flip-chip quantum chip, and a quantum computer
By employing a composite structure of support platform and metal layer in the flip-chip quantum chip, the problem of poor bump structure quality is solved, achieving uniform bump height and strong mechanical properties, thus ensuring the stability of the flip-chip and the reliability of signal transmission.
Patent Information
- Application Number
- CN202522543503.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-01
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2035-12-01
AI Technical Summary
In existing technologies, the quality of bump structures is poor, and they are prone to differences in height from expectations, uneven height, and weak mechanical properties, which affect the performance and stability of flip chips.
The system employs a composite structure of a support platform and a metal layer. The support platform is made of rigid material and provides mechanical support, while the metal layer is used for signal transmission, forming a heterogeneous interconnect system. The support platform and the metal layer are integrally covered on the substrate surface to ensure the uniformity and stability of the bump height.
This achieves uniformity of bump height and improved mechanical properties, ensuring the stability of flip chips and the reliability of signal transmission, avoiding deformation of bumps during manufacturing and packaging, and improving the packaging quality of the chip.
Smart Images

Figure CN224987941U_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of quantum chip fabrication technology, and in particular relates to a substrate with bumps, a flip-chip quantum chip, and a quantum computer. Background Technology
[0002] For current superconducting quantum chips, wire bonding is no longer suitable as the scale of qubits increases. Flip bonding is currently the better solution for expanding superconducting qubits. By precisely flip-bonding multiple sub-chips onto a common silicon interposer, flip bonding enables low-loss, low-crosstalk quantum bus connections between chips.
[0003] In flip chip structures, a highly precise and uniform spacing must be maintained between the upper and lower layers. This spacing not only affects the electromagnetic field distribution within the flip chip but also ensures the mechanical stability of the flip chip and achieves uniform bonding. Currently, using bumps (such as indium pillars and indium tin pillars) for inter-chip bonding is the most common method. These bump structures can provide both inter-chip support and signal conduction.
[0004] Correspondingly, the spacing of flip-chip bonding pads after bump interconnection has a significant impact on the performance of the flip chip itself. The spacing affects the coupling between quantum devices carried by the two substrates. Therefore, providing high-quality bump structures on the substrate is an urgent problem to be solved. Utility Model Content
[0005] The purpose of this application is to provide a substrate with bumps, a flip-chip quantum chip, and a quantum computer to solve the problems of poor bump structure quality in the prior art, such as differences in height from the expected height, uneven height, and weak mechanical properties. It can provide a bump with decoupled support function and information transmission function, so as to have stronger mechanical properties and the height obtained by fabrication is more in line with expectations.
[0006] To solve the above-mentioned technical problems, the present invention provides a substrate with bumps in a first aspect, comprising: Substrate; A support stage, located on the surface of the substrate; and A metal layer is integrally covered on the surfaces of the substrate and the support stage, the metal layer comprising a first portion covering the top surface of the support stage and a second portion covering the surface of the substrate; In a direction perpendicular to the surface of the substrate, the first portion is higher than the second portion.
[0007] Optionally, the support platform is integrally formed with the substrate; Alternatively, the support platform is fixedly connected to the substrate.
[0008] Optionally, there may be multiple support platforms, and each support platform shall have the same height in a direction perpendicular to the surface of the substrate.
[0009] Optionally, the support platform is frustum-shaped.
[0010] Optionally, the material of the metal layer is one of niobium, tantalum, titanium nitride, and aluminum.
[0011] Optionally, the substrate is a silicon substrate.
[0012] Optionally, it further includes a buffer layer, which is stacked on the surface of the first portion facing away from the substrate.
[0013] Optionally, the buffer layer is an indium layer.
[0014] In a second aspect, this invention provides a flip-chip quantum chip, comprising: the aforementioned substrate with bumps.
[0015] In a third aspect, this invention provides a quantum computer comprising: the aforementioned flip-chip quantum chip.
[0016] Compared with existing technologies, this application decouples the support function and signal transmission function of bumps on the substrate. The support stage is disposed on the substrate surface, and the metal layer is integrally covered on the support stage and the substrate surface to realize signal conduction, forming a heterogeneous interconnect system in which structural support and signal transmission each perform their respective functions. In this way, the height of the bumps provided is easy to fabricate and has strong retention, and the height and spacing of the bumps on the substrate can be kept uniform while having signal transmission function. Attached Figure Description
[0017] Figure 1 This is a schematic diagram of the structure of a substrate with bumps provided in an embodiment of this application; Figure 2 A schematic diagram illustrating the process of fabricating the support stage and substrate provided in an embodiment of this application; Figure 3 A schematic diagram of a structure in which a buffer layer is further provided on the convex surface provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of the flip-chip quantum chip provided in an embodiment of this application.
[0018] Figure labeling: 1-substrate, 2-support stage, 3-metal layer, 4-mask, 5-buffer layer. Detailed Implementation
[0019] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.
[0020] The specific embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. The advantages and features of the present invention will become clearer from the following description and claims. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.
[0021] In the description of this invention, it should be understood that the terms "center", "upper", "lower", "left", "right", 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. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0022] In the evolution of superconducting quantum computing towards large-scale fault-tolerant architectures, the manufacturing limits of a single wafer have gradually become a core bottleneck. Constrained by the coherence requirements of qubits, the uniformity control of superconducting materials, and process yield limitations, the number of qubits that can be integrated on a single wafer and the complexity of functional modules are insufficient to meet the demands of practical quantum processors. Therefore, heterogeneous integration architecture based on silicon interposers has become a key technology to overcome this limitation. Its core logic lies in the orderly integration of multiple functional sub-chips (such as qubit array chips, control readout circuit chips, and signal routing chips) onto the same silicon interposer using high-precision flip-chip bonding technology. This constructs a low-loss, low-crosstalk quantum bus connection across chips, ultimately achieving a large-scale quantum processor that surpasses the physical limits and functional boundaries of a single wafer.
[0023] The core advantages of this integrated architecture lie not only in its breakthrough in scale and performance assurance, but also in its exceptional scalability and compatibility. Through the modular design of the sub-chips, the number of qubit sub-chips can be flexibly increased or decreased according to the needs of the quantum processor, or different control and readout sub-chips can be replaced without reconstructing the overall architecture. This architecture provides a feasible path for the realization of large-scale fault-tolerant quantum processors: by continuously optimizing the alignment accuracy of flip-chip bonding, the superconducting transmission performance of the TSV structure, and the crosstalk suppression capability of the interposer layer, the number of qubits can be expected to scale to tens of thousands.
[0024] Currently, common flip-chip quantum chips are complex systems that achieve multi-chip collaboration through heterogeneous integration technology. Their structural design needs to take into account the coherence of quantum bits, low-loss transmission of cross-chip signals, and system scalability. The core consists of three main parts: functional sub-chip array, silicon interposer, and flip-chip bonding structure.
[0025] In flip-chip quantum chips, the spacing between the upper and lower layers is one of the core process parameters that determines system performance. This spacing is not a simple physical gap, but a space that is precisely controlled based on the height of the bonding bumps, the flatness of the substrate, and the bonding process. Its precision is directly related to the electromagnetic properties, mechanical reliability, and electrical connection quality of the quantum chip.
[0026] From the perspective of electromagnetic field manipulation, this spacing is the "boundary condition" of the electromagnetic environment of the qubit, directly shaping the coupling capacitance network between the qubit and its surrounding structure. If the spacing is too small, the parasitic capacitance increases, which may cause the qubit frequency to deviate by more than 100MHz, or even fall into the frequency overlap region with the control signal, causing crosstalk. Uneven spacing will cause the capacitance consistency of qubits in the same array to deteriorate, leading to an increase in frequency dispersion between qubits and forcing the calibration complexity to increase exponentially. More importantly, spacing fluctuations will destroy the effectiveness of the electromagnetic shielding around the qubit, allowing environmental noise (such as substrate charge fluctuations and external magnetic flux changes) to invade the quantum state through capacitive coupling, directly leading to a shortening of the coherence time (T1, T2).
[0027] In terms of mechanical stability, uniform spacing is the core guarantee for balancing the thermal stress and mechanical load of flip-chip structures. If there is a local deviation in spacing, the bonding bumps in that area will bear a greater compressive load, while areas with excessively large spacing may form stress concentration zones due to insufficient load. This uneven stress can cause substrate bending and even microcracks at the edge of the silicon substrate. Such mechanical damage can directly undermine the structural integrity of the Josephson junction, leading to qubit failure. In addition, uniform spacing can ensure the parallelism of the upper and lower layers, avoiding shear stress caused by tilting on the bonding bumps. Soft metal bumps such as indium are prone to plastic deformation under shear stress, which may cause bump breakage or interface delamination, resulting in loss of electrical connectivity.
[0028] From the perspective of bonding technology and electrical connection, consistent microbump height is a prerequisite for achieving uniform inter-chip spacing, which directly determines the reliability of cross-chip interconnects. During flip bonding, the pressure from bonding is transmitted to the entire chip interface through the bumps, causing the bumps to undergo controllable plastic deformation to eliminate contact gaps caused by surface roughness. If the spacing is uneven, the bumps in some areas may not compress enough (too large a spacing), resulting in insufficient pressure distribution and an inability to form effective ohmic contacts, causing a sharp increase in quantum signal transmission loss. On the other hand, areas with too small a spacing may experience excessive pressure overload, leading to excessive compression of the bumps. Metal overflow may then bridge adjacent bumps (forming a short circuit) or damage the superconducting wiring beneath the bumps (such as aluminum film breakage). For quantum signal paths, this connection inconsistency is fatal. Fluctuations in contact resistance can cause uneven attenuation of microwave pulse amplitude, amplifying the differences in manipulation errors between different qubits. Short circuits may directly connect the qubits to ground, causing them to completely lose their quantum state storage capability.
[0029] Currently, flip bonding of quantum chips typically uses bumps for soldering. Bump structures, such as indium pillars and indium tin pillars, can be used. These bump structures provide both support and enable superconducting signal conduction. In current fabrication processes, bump structures can be prepared using thermal evaporation or electroplating. However, bump structures obtained through these methods often suffer from uneven bump heights. This can lead to uneven spacing between the upper and lower chip layers and chip tilting during flip bonding.
[0030] For example, the height of the bump structure needs to be 5 to 10 micrometers. This is because if the height is too low, the component structures of the upper and lower chip layers will come into contact with each other, such as the air bridge structure (1 to 3 micrometers in height). At the same time, if the height is too low, the two chip layers will not be able to bond effectively and firmly. If the height is too high, the pressure will increase during thermoforming, and the chip will bear unnecessary pressure.
[0031] The bump structure needs to be 5 to 10 micrometers high. When using thermal evaporation to prepare bumps of this height, a stripping method is used, which usually requires a thicker photoresist. To prepare a 10-micrometer high bump, the photoresist thickness needs to be 12 to 15 micrometers. When spin-coating such a thick layer of photoresist, the "edge bead" effect and uneven thickness across the entire wafer are likely to occur. For the preparation of the bump structure, even a small percentage change in the photoresist thickness will result in a significant difference in the absolute height of the bump structure on the 10-micrometer scale. This will undoubtedly affect the inter-die spacing after flip-chip bonding. In addition, during the photoresist stripping process, uneven solvent penetration and photoresist expansion rates may cause some bumps to be physically pushed or pulled before the photoresist is completely dissolved, resulting in uneven bump tops. This will also affect the uniformity of the chip spacing after flip-chip bonding.
[0032] When the above bump structure is fabricated using electroplating, due to the pattern density effect, when the bump array is unevenly distributed across the entire wafer, such as dense distribution in some areas and sparse distribution in others; in dense areas, each bump carries a smaller current and grows slowly, while in sparse areas, each bump carries a larger current and grows faster. This results in height differences between the bumps fabricated in different areas, which in turn affects the chip spacing for subsequent flip-chip bonding.
[0033] As can be seen, while various existing processes can fabricate chip-to-chip interconnect bump structures, the currently fabricated bump structures also suffer from problems such as discrepancies between the expected and actual heights, uneven height distribution, and susceptibility to deformation. Undoubtedly, bump structures with these issues will have an uncontrollable impact on the chip packaging quality and may even lead to chip failure. Therefore, high-quality bumps on the substrate are a challenge that must be overcome.
[0034] To address the aforementioned problems, this application provides a substrate 1 with bumps, wherein the support of the bumps is provided by a rigid support platform 2 to prevent deformation during manufacturing and packaging, and its signal transmission function is provided by a metal layer 3.
[0035] refer to Figure 1 In one embodiment of this application, the substrate 1 with bumps includes: a substrate 1; a support stage 2 located on the surface of the substrate 1; and a metal layer 3 integrally covering the surfaces of the substrate 1 and the support stage 2. The metal layer 3 includes a first portion covering the top surface of the support stage 2 and a second portion covering the surface of the substrate 1. In a direction perpendicular to the surface of the substrate 1, the first portion is higher than the second portion.
[0036] In flip chip technology, the bumps on substrate 1 are the cornerstone and key feature of the entire technology system. Bumps are microstructures directly fabricated on the active surface of the chip, used for electrical and mechanical interconnection between the chip and the packaging substrate or another chip. In flip chip technology, the bumps on substrate 1 appear as short, thick vertical paths, greatly shortening the signal transmission path and significantly reducing signal delay. Furthermore, the bumps also serve as the chip's primary support structure, bearing the chip's own weight and the stress generated during subsequent packaging processes and operation. In addition, bumps also serve functions such as heat dissipation, grounding, and providing signal transmission density.
[0037] To address the mechanical stability bottleneck inherent in traditional indium pillar interconnects under large-scale integration scenarios, this embodiment proposes a composite structure of a support platform 2 and a metal layer 3 to replace the indium pillars. The core design decouples the support and signal transmission functions of traditional interconnects. The support platform 2 on the surface of substrate 1 provides the support, while the metal layer 3 integrally covers the surfaces of substrate 1 and support platform 2 (the top and side surfaces of support platform 2). Therefore, the metal layer 3 is used for signal transmission, thus forming a heterogeneous interconnect system where structural support and signal transmission each perform their respective functions.
[0038] Specifically, the support platform 2 is made of a rigid material with extremely high mechanical properties, ensuring that it will not deform under pressure during flip-chip bonding of the substrate 1. Therefore, the support platform 2, located on the surface of the substrate 1, provides a more stable support effect and even ensures the spacing accuracy and uniformity provided by the support platform 2 after long-term chip use. The material of the support platform 2 does not have to be metal; its mechanical strength is significantly higher than indium, effectively resisting the pressure impact during subsequent flip-chip bonding and the thermal stress during low-temperature cycling, reducing the risk of substrate 1 bending and bump collapse, and ensuring uniform support for large-scale bump arrays. Most importantly, the support platform 2 can be fabricated according to preset parameters, ensuring that all fabricated support platforms 2 have the same height in the direction perpendicular to the substrate 1, thus guaranteeing the height uniformity of the bumps.
[0039] The fabrication of the support platform 2 is the core supporting foundation of this structure, and its design and process must be fully matched to the chip's operating temperature environment and mechanical stability requirements. For example, the substrate 1 used to fabricate the quantum chip (typically a high-resistivity silicon substrate 1 to reduce interference from the substrate 1 to the qubits) can be patterned using high-precision photolithography to define the array pattern of the support platform 2. The pattern layout needs to correspond one-to-one with the subsequent signal transmission metal layer 3 wiring to ensure spatial matching between the support platform 2 and the signal path. To form the support platform 2 on the surface of the substrate 1, a deep silicon etching (DRIE) process can be used to selectively etch the substrate 1, precisely controlling the etching depth and sidewall perpendicularity to form a support platform 2 with uniform height and regular shape. Alternatively, selective silicon epitaxy (SESi), nanoimprint lithography (NIL)-dry etching composite processes, and sacrificial layer-assisted patterning processes can be used to form the substrate 1 with the support platform 2 on its surface.
[0040] For metal layer 3, it is integrally covered on the surfaces of substrate 1 and support stage 2 (including the side and top surfaces of support stage 2), with the metal layer 3 on the surface of substrate 1 serving as the large structural layer of the chip. For example, metal layer 3, as the large structural layer of a superconducting quantum chip, is uniformly disposed on the surface of substrate 1 and patterned using techniques such as photolithography and etching to form structures such as resonant cavities, control lines, and buses, or structures such as Josephson junctions and bit capacitors. The portion of metal layer 3 on the surface of support stage 2 mainly serves as the signal interconnection path between chips.
[0041] It should be noted that the metal layer 3 includes a first part covering the top surface of the support stage 2 and a second part covering the surface of the substrate 1. The first part is higher than the second part in the direction perpendicular to the substrate 1. On the one hand, this is because both the support stage 2 and the metal layer 3 are components of bumps, the purpose of which is to achieve the support effect between chips so that the chips have a fixed and uniform spacing. On the other hand, other devices, such as air bridges, will also be formed on the surface of the substrate 1. The air bridges have a considerable height in the direction perpendicular to the substrate 1. Therefore, the first part being higher than the second part is also to ensure that there must be space between the chips to accommodate other devices to meet the integrity of the flip chip.
[0042] After the support stage 2 is fabricated, a continuous metal layer 3 is formed in the exposed areas on the surfaces of the support stage 2 and the substrate 1. This metal layer 3 replaces the core signal transmission carrier of the indium pillar. The fabrication process of the metal layer 3 needs to balance coverage integrity and superconducting compatibility. It typically employs techniques such as magnetron sputtering, electron beam evaporation, and atomic layer deposition to uniformly deposit the target metal material on the surfaces of the substrate 1 and the support stage 2 in a vacuum environment, thereby achieving integrated coverage of the top surface, side surface, and surface of the support stage 2 and the substrate 1.
[0043] For example, the substrate material can be silicon (high lattice matching degree and low coefficient of thermal expansion), and the whole is encapsulated in a low-temperature compatible frame (such as an oxygen-free copper frame). The frame is connected to the cold stage of the dilution refrigerator to achieve efficient heat conduction and avoid local temperature rise that leads to degradation of superconducting performance.
[0044] Furthermore, a highly uniform support stage 2 is formed on the surface of substrate 1 (especially the high-resistivity silicon substrate 11 commonly used in quantum chips). The core requirements for the support stage 2 are size consistency, morphological regularity, and array consistency. In addition, the process must be compatible with the low-pollution and high-cleanliness requirements of superconducting quantum chips to avoid introducing impurities that may affect the coherence of the quantum bits.
[0045] In one embodiment of this application, the support stage 2 is integrally formed with the substrate 1. During the specific fabrication of the support stage 2, a single substrate can be provided, and then the substrate is processed to form the substrate 1 with the support stage 2. Directly etching a single substrate into an integral substrate 1 and support stage 2 using methods such as deep silicon etching is relatively simple, and since the support stage 2 and substrate 1 are integrally formed, it provides stronger mechanical properties, reducing the risk of the support stage 2 detaching in subsequent processes.
[0046] Realization of forming support stage 2 on silicon substrate: A mixture of TMAH-tetramethylammonium hydroxide (CH3)4NOH and silicic acid or ammonium persulfate can be used to selectively etch the silicon substrate to prepare 'frustum-shaped' microbumps.
[0047] TMAH-Tetramethylammonium hydroxide (CH3)4NOH, and silicic acid mixture: Add 5 wt% silicic acid to a TMAH base solution with a tetramethylammonium hydroxide concentration of 20-25 wt%.
[0048] TMAH-Tetramethylammonium hydroxide (CH3)4NOH, and ammonium persulfate mixture: Add 7-10 g / L of ammonium persulfate to a TMAH base solution with a tetramethylammonium hydroxide concentration of 20-25 wt%.
[0049] TMAH etching of single-crystal silicon is a typical anisotropic etching process, with the etching rate relationship being: (100)>(110)>>(111). Among them, the etching rate of the (111) plane is tens or even hundreds of times slower than that of the (100) plane, and can be regarded as the etching cutoff plane. On the (100) wafer: using a square mask window, since the lateral etching almost stops on the (111) plane, a frustum-shaped support stage 2 will be naturally etched, and the sidewall angle is a fixed 54.74°, which is determined by the angle between the (100) plane and the (111) plane.
[0050] For the etching of the substrate, the etching depth is directly proportional to the width of the mask 4 window. The depth D and the mask window opening width W satisfy the relationship: D≈(W / 2)*tan(54.74°). In this process, it is not necessary to control the depth by precisely controlling the etching time. Only by designing the pattern size of the mask 4 can a highly consistent bump structure be obtained, that is, a highly consistent support stage 2 array can be obtained.
[0051] Alternatively, the support platform 2 can be fixedly connected to the substrate 1; a compatible material can be deposited on the surface of the substrate 1 to form the support platform 2. This method involves less engineering work and has relatively lower processing difficulty, and the support platform 2 is fixed to the substrate 1 after deposition. For example, by using a mask-defined, vapor-phase directional growth mechanism, silicon material can be precisely deposited in a predetermined area of the substrate 1 to form a support platform 2 that is homogeneous with the substrate 1: First, a SiO2 or SiN4 mask is prepared on the surface of the substrate 1 and patterned with windows to expose the target bump area; then, in an ultra-high vacuum epitaxial furnace, using SiH4 or Si2H6 as the silicon source, selective growth of silicon atoms is achieved only in the windowed area at 600-850℃. By strictly controlling the temperature uniformity, gas flow distribution, and growth time, the bump height is ensured to remain consistent from the center to the edge, while avoiding parasitic growth in the mask area.
[0052] like Figure 2 As shown, the integrally formed support stage 2 and substrate 1 can be fabricated using the following method: Step 1: A patterned mask 4 is set on the surface of the substrate; The second step is to etch the substrate using the mask 4. Step 3: Remove the mask 4.
[0053] A mask 4 is fabricated on the surface of substrate 1 (such as the (100) high-resistivity silicon substrate 1 commonly used in quantum chips). Its core function is to serve as a "patterned template" to achieve precise region selection and process isolation. On the one hand, it can transfer the preset pattern through photolithography and act as an etching hard mask when preparing silicon bumps using deep silicon etching process, protecting the integrity of the substrate in the non-etched area and ensuring the dimensional uniformity and spacing accuracy of the support stage 2 array. In addition, the mask 4 needs to be compatible with PECVD (plasma-enhanced chemical vapor deposition) silicon nitride preparation, photolithography, etching and other processes. Through precise pattern definition, the functions of "structural support (support stage 2) - signal transmission (metal layer 3)" are decoupled, providing a foundation for high-precision alignment of flip bonding and stable control of interlayer spacing, ultimately supporting the scalability and process yield improvement of large-scale quantum chips. Wherein, the material of the substrate is silicon; and / or, the material of the mask 4 is silicon nitride.
[0054] A patterned silicon nitride mask 4 can be fabricated on the surface of a substrate using plasma-enhanced chemical vapor deposition (PECVD). In semiconductor and quantum chip manufacturing, PCVD is one of the core technologies for fabricating silicon nitride masks 4. It generates plasma by exciting the reactive gas with radio frequency (RF) or microwave energy, which significantly reduces the deposition temperature while providing high deposition rates, beneficial step coverage, and good process compatibility.
[0055] The core of plasma-enhanced chemical vapor deposition is to utilize the non-equilibrium characteristics of plasma to decompose reactive gases (such as silicon and nitrogen sources) into active species (ions, free radicals, etc.) and form a thin film by chemical reaction on the surface of substrate 1.
[0056] To obtain a high-quality silicon nitride mask, the temperature is typically controlled between 200 and 400°C. This not only avoids thermal damage to the substrate but also reduces impurity diffusion. In subsequent dry etching (such as deep silicon etching, metal etching) or wet etching processes, silicon nitride must exhibit high selectivity for the etched material (such as silicon or metal) to ensure that the mask layer is not easily eroded during the etching process, thereby enabling precise pattern transfer.
[0057] After the mask 4 is formed on the surface of the substrate, it needs to be patterned. Patterning can be performed by using a photolithography machine to define the square mask window and position of the support stage 2 on the silicon nitride mask 4, and by using plasma etching (such as RIE) to etch the square mask of the support stage 2.
[0058] After the patterned mask 4 is prepared, the wafer with the patterned silicon oxide mask 4 is immersed in the TMAH solution described in the previous embodiment at a temperature of 80-90°C to etch the substrate and finally obtain a device in which the support stage 2 and the substrate 1 are integrally formed.
[0059] After etching the substrate to the required depth, the substrate is removed from the TMAH solution, and the silicon oxide mask 4 on the surface of the substrate is completely removed using buffered oxide etchant (BOE), thus exposing the substrate 1 and the support stage 2.
[0060] It should be noted that the etching method of the substrate is wet etching, but dry etching can also be selected. When etching the high-resistivity silicon substrate 1, wet etching removes materials by isotropic reaction between chemical solutions (such as KOH, TMAH) and silicon. The etching rate depends on the silicon crystal orientation (e.g., the etching rate of (100) crystal orientation is higher than that of (111)). It has extremely high selectivity for mask layers such as SiO2 and Si3N4, and there is no lattice damage caused by plasma bombardment. Dry etching (such as SF6 plasma etching) uses a combination of ion physical bombardment and chemical reaction to achieve anisotropic etching. It can precisely control high aspect ratio structures, but it will introduce plasma-induced damage (such as lattice defects and surface residues) and has relatively low selectivity for mask layers.
[0061] In summary, wet etching is the preferred method for etching high-resistivity silicon substrates because of its low-damage characteristics. The carrier mobility and quantum coherence of high-resistivity silicon are highly sensitive to lattice defects. Wet etching avoids the ion bombardment damage of dry etching and can preserve the electrical properties of the substrate to the greatest extent. At the same time, its high selectivity for the mask layer reduces mask consumption, resulting in lower process costs and simpler operation. It is especially suitable for high-resistivity silicon applications such as quantum chips that are sensitive to damage to the substrate.
[0062] Specifically, the etchant used to etch the substrate is a mixed solution comprising tetramethylammonium hydroxide and silicic acid; more specifically, approximately 50 grams of silicic acid powder is added to every 1000 mL of TMAH solution. Alternatively, the etchant used to etch the substrate is a mixed solution comprising tetramethylammonium hydroxide and ammonium sulfate; more specifically, 7 to 10 grams of ammonium persulfate crystals are added to every 1000 mL of TMAH solution.
[0063] It should be noted that the substrate is prepared by wet etching to form an integral support stage 2 and substrate 1, and then the silicon nitride hard mask is removed from the top surface of the support stage 2. Although an integral support stage 2 and substrate 1 are obtained after this series of processes, the surface roughness of the support stage 2 and substrate 1 is relatively large. This will result in poor performance or poor coverage of the electrically connected metal layer 3 formed on the surface of the substrate 1 and the support stage 2.
[0064] The rough surface can be oxidized into silicon dioxide by thermal oxidation, and then the silicon dioxide prepared by thermal oxidation can be completely removed by buffer oxide etchant (BOE) to obtain the support stage 2 and substrate 1 surface with small roughness.
[0065] In one embodiment of this application, there are multiple support platforms 2, and each support platform 2 has the same height in a direction perpendicular to the surface of the substrate 1.
[0066] As part of the bump array, the support platform 2 serves a supporting function. Its design, characterized by a large number of bumps and high uniformity, is a core requirement for balancing mechanical stability, electrical reliability, and performance retention. For example, in flip-chip quantum chips, the uniform distribution of multiple bumps can disperse stress across the entire chip interface, preventing substrate 1 bending or cracking due to localized stress concentration, or breakage of the support platform 2 itself. Insufficient support platforms 2 may lead to stress imbalance and structural failure. Furthermore, quantum chips require the transmission of a large number of parallel signals. Each support platform 2 and its surface metal layer 3 correspond to an independent signal path. The multi-bump array enables one-to-one signal routing, avoiding crosstalk caused by non-standard signals sharing the same channel. In addition, some bumps also need to perform grounding and shielding functions, which undoubtedly increases the required number of support platforms 2.
[0067] Furthermore, the electromagnetic environment of a qubit is highly sensitive to the interlayer spacing. Fluctuations in spacing directly alter the coupling capacitance between the qubit and the surrounding metal structure, causing frequency shifts (up to 100MHz or more), disrupting inter-qubit consistency, and increasing calibration complexity. Highly consistent bumps (height deviation < ±0.1μm) can strictly lock the interlayer spacing, ensuring a uniform electromagnetic environment for all qubits and preventing a decrease in coherence time due to uneven spacing.
[0068] During flip-chip thermocompression bonding of a substrate 1 with a bump structure, the bonding pressure (10-50 MPa) needs to be transmitted to the entire interface through the bumps, causing controlled plastic deformation of the bumps (the height of the support platform 2 remains constant, while the metal layer 3 or layers above it deforms) to form an ohmic contact. If the bump heights are inconsistent, shorter bumps will not deform sufficiently due to insufficient pressure, leading to increased contact resistance and quantum signal transmission loss; taller bumps may be crushed due to pressure overload, causing short circuits between adjacent bumps or bump breakage. Bumps with consistent heights ensure pressure balance at each contact point, guaranteeing consistent and stable contact resistance across all signal paths.
[0069] Furthermore, differences in bump height can cause the interface between the sub-chip and the interposer to tilt, generating shear stress that may cause the edge bumps to bear several times the stress of the center bumps. Under long-term low-temperature cycling, this can easily lead to bump detachment or substrate cracks. Bumps with consistent height can maintain interface parallelism, ensuring uniform stress distribution and guaranteeing the long-term mechanical reliability of the chip in low-temperature environments.
[0070] In one embodiment of this application, the support platform 2 is frustum-shaped. It is known that the frustum-shaped support platform 2 has higher mechanical strength and a more robust structure compared to cylindrical microbumps prepared by the traditional peeling method. The height of the microbump mainly comes from the support platform 2. To prepare a 10μm high microbump, the support platform 2 can be designed to be 9~9.5μm, and the metal layer 3 to be 0.5~1μm. The metal layer 3 only needs to be thin, greatly avoiding the height difference caused by a thicker coating; moreover, the surface smoothness of the thin metal layer 3 is higher than that of a cylindrical shape, which is more conducive to flip-chip bonding.
[0071] In one embodiment of this application, the material of the metal layer 3 can be selected as niobium, tantalum, titanium nitride, aluminum, etc. For example, when the metal layer 3 is formed on the substrate 1, a niobium layer of 100 to 300 nm is deposited on the entire surface of the substrate 1 and the surface of the support stage 2. Then, the niobium layer is patterned by photolithography and dry etching to form structures such as bits; or, the niobium layer is patterned by photolithography and dry etching to form wiring, such as resonant cavities, control lines, and buses.
[0072] Furthermore, the material chosen for substrate 1 is silicon substrate 1. The insulating properties of high-resistivity silicon can significantly reduce leakage current and charge relaxation of substrate 1, and reduce quantum state decoherence; most importantly, when fabricating substrate 1 with support stages 2 on its surface, deep silicon etching (DRIE) can produce high-quality (highly consistent, uniform, and low roughness) silicon support stages 2, ensuring the consistency of support for flip bonding; selective silicon epitaxy can precisely grow homogeneous silicon structures, avoiding defects and losses caused by heterogeneous material interfaces.
[0073] In this application, Figure 3 In one embodiment shown, the substrate 1 with bumps further includes a buffer layer 5, which is stacked on the surface of the first portion facing away from the substrate 1.
[0074] This is because the devices with bumps on the substrate 1 in this embodiment are mainly used for flip chips, meaning that the substrate 1 will also be bonded to another circuit board via thermocompression. In flip chips, the support platform 2 provides support between the chips, while the metal layer 3 serves as a signal transmission layer. However, since the metal layer 3 also needs to be fabricated as a circuit structure (such as a Josephson junction, capacitor, etc.) on the surface of the substrate 1, the thickness of the metal layer 3 on the surface of the support platform 2 is set to 100 to 300 nm, which is undoubtedly thin. The force of thermocompression may cause this thin metal layer 3 on the surface of the support platform 2 to crack, be damaged, or have changes in physical properties. Therefore, to avoid this situation, a buffer layer 5 can be formed on the surface of the metal layer 3 on the top surface of the support platform 2. This buffer layer 5 is mainly used to absorb the relative pressure brought by thermocompression to ensure that the performance of the metal layer 3 remains unchanged.
[0075] Furthermore, the buffer layer 5, serving as a relative pressure absorption layer, needs to be made of a material with a surface softer than that of the metal layer 3, making it more prone to deformation. The buffer layer 5 also needs to facilitate signal transmission between the metal layers 3 on the top surfaces of the upper and lower support platforms 2 after hot pressing; therefore, the buffer layer 5 must also possess electrical conductivity. Thus, indium can be chosen as the material for the buffer layer 5.
[0076] An indium layer of 0.5 to 1 μm is prepared on the surface of the support stage 2. Its core function is to serve as a stress buffer layer 5 during the flip-chip bonding process, absorbing the mechanical load during the hot-pressing process through its own plastic deformation. Indium, as a low-melting-point (156.6℃) and highly ductile metal, is prone to significant plastic flow under bonding temperature (typically 120-180℃, below the melting point of indium but within its active plastic deformation range) and pressure (10-50MPa). This deformation characteristic allows the indium layer to effectively disperse and absorb bonding pressure, preventing the pressure from being directly transmitted to the 100-300nm thick niobium layer below. Niobium, as a superconducting material, possesses excellent electrical properties, but it has high mechanical brittleness. If directly subjected to concentrated stress, it is prone to structural damage such as lattice distortion and microcracks.
[0077] The plastic deformation mechanism of the indium layer provides crucial protection for the structural integrity of the niobium layer, thereby maintaining its superconducting properties. Niobium layers are often used as key interconnect layers or electrode materials in superconducting quantum chips, and their core parameters, such as the superconducting critical temperature (Tc≈9.2K) and critical current density (Jc), are highly sensitive to structural defects. Deformation of the niobium layer can lead to changes in the local cross-sectional area of the superconducting pathway, an increase in lattice defect density, resulting in increased resistance, degradation of superconducting properties, or even failure. The indium layer absorbs most of the thermal bonding stress, keeping the stress on the niobium layer within a safe threshold, ensuring that its lattice structure and superconducting properties are not damaged, thus guaranteeing the overall electrical reliability of the flip-chip bonded device.
[0078] In this application, Figure 4 In one embodiment shown, a flip-chip quantum chip is also proposed, comprising a substrate 1 with bumps as described in the foregoing embodiments.
[0079] The core of flip-chip quantum computing is to interconnect two or more chips with different functions (such as a quantum bit chip and a control / readout chip) through a high-precision bonding process. The bump structure on substrate 1 is the key structure for achieving mechanical fixation and electrical connection between the two chips.
[0080] Specifically, the support platform 2 provides physical support between the two chips, separating and fixing them at a uniform and reasonable height, thereby precisely controlling the interlayer spacing; the metal layer 3 can be regarded as a miniature wire between the chips to establish an ohmic connection. Through the metal layer 3, the control chip can send precise microwave signals to the quantum bit chip to perform quantum gate operations, and at the same time, it can also receive state readout signals from the quantum bit.
[0081] In summary, the bumps on substrate 1 are the core hub for realizing flip-chip interconnection of quantum chips. They not only provide physical bridges and electrical wires for the two flip-chip interconnects, but also control the electromagnetic environment between chips, ensure mechanical stability, and are key to achieving high-density interconnection.
[0082] In one embodiment of this application, a quantum computer is also proposed, comprising the flip-chip quantum chip described in the foregoing embodiments.
[0083] It should be noted that the bumped substrates used in flip-chip quantum chips and quantum computers are similar to the bumped substrates discussed above and have the same beneficial effects as the flip-chip quantum chip embodiments described above, therefore, they will not be described again. For technical details not specifically disclosed in other sections, those skilled in the art should refer to the description of the bump structure (support stage and metal layer) above for understanding, and for the sake of brevity, they will not be repeated here.
[0084] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," or "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.
[0085] The above description, based on the embodiments shown in the drawings, details the structure, features, and effects of this application. The above description is only a preferred embodiment of this application, but this application does not limit the scope of implementation to what is shown in the drawings. Any changes made in accordance with the concept of this application, or modifications to equivalent embodiments, that do not exceed the spirit covered by the specification and drawings, should be within the protection scope of this application.
Claims
1. A substrate (1) with bumps, characterized in that, include: Substrate (1); A support platform (2) is located on the surface of the substrate (1); as well as A metal layer (3) is integrally covered on the surfaces of the substrate (1) and the support stage (2). The metal layer (3) includes a first portion covering the top surface of the support stage (2) and a second portion covering the surface of the substrate (1). In the direction perpendicular to the surface of the substrate (1), the first part is higher than the second part.
2. The substrate (1) with bumps according to claim 1, characterized in that, The support platform (2) and the substrate (1) are integrally formed; Alternatively, the support platform (2) is fixedly connected to the substrate (1).
3. The substrate (1) with bumps according to claim 1, characterized in that, The number of the support platforms (2) is multiple, and the height of each support platform (2) is consistent in the direction perpendicular to the surface of the substrate (1).
4. The substrate (1) with bumps according to any one of claims 1 to 3, characterized in that, The support platform (2) is frustum-shaped.
5. The substrate (1) with bumps according to claim 1, characterized in that, The metal layer (3) is made of one of niobium, tantalum, titanium nitride and aluminum.
6. The substrate (1) with bumps according to claim 1, characterized in that, The substrate (1) is a silicon substrate (1).
7. The substrate (1) with bumps according to claim 1, characterized in that, It also includes a buffer layer (5), which is stacked on the surface of the first portion away from the substrate (1).
8. The substrate (1) with bumps according to claim 7, characterized in that, The buffer layer (5) is an indium layer.
9. A flip-chip quantum chip, characterized in that, include: The substrate with bumps as described in any one of claims 1 to 8 (1).
10. A quantum computer, characterized in that, include: The flip-chip quantum chip according to claim 9.
