CN-122026850-A - Cavity structure of wafer level packaging crystal oscillator and metal hot-press bonding method
Abstract
The invention discloses a cavity structure of a high-reliability wafer-level packaging crystal oscillator and a bonding method thereof, belonging to the technical field of micro-electronic mechanical systems and semiconductor packaging. The structure comprises a device wafer and a cap wafer, which are hermetically bonded through an annular composite metal bonding ring. The cap wafer is internally provided with a main cavity and an annular buffer microcavity surrounding the main cavity, the buffer microcavity is positioned at the inner side of the bonding ring and is adjacent to the bonding ring, the depth is smaller than that of the main cavity, and the side wall of the buffer microcavity is of an inclined plane or step-shaped structure, so that a mechanical stress buffer and particle capture well is formed. The bonding ring is of a gradient composite structure at least comprising a plastic deformation layer and a self-healing alloy layer. The invention synchronously realizes excellent stress isolation, ultrahigh air tightness, effective electromagnetic shielding and high production yield through the collaborative design of the cavity structure and the material process, and fundamentally solves the reliability bottleneck of the traditional wafer-level packaging crystal oscillator in high-end application.
Inventors
- WANG JIABIN
- HOU HAITAO
Assignees
- 北京晶宇兴科技有限公司
Dates
- Publication Date
- 20260512
- Application Date
- 20260126
Claims (10)
- 1. The utility model provides a cavity structure of wafer level encapsulation crystal oscillator, includes device wafer (2) and the cap wafer (1) that have the cavity that have resonator (5), device wafer (2) and cap wafer (1) pass through annular bonding ring (6) seal bonding, its characterized in that: The bottom of the cap wafer (1) is provided with a main cavity (3) and an annular buffer microcavity (4) surrounding the main cavity (3), and the annular buffer microcavity (4) is positioned at the inner side of the bonding ring (6) and is adjacent to the bonding ring; the depth of the annular buffer microcavity (4) is smaller than the depth of the main cavity (3); The bonding ring (6) is a composite metal structure symmetrically distributed at the bottom of the cap wafer (1) and the top of the device wafer (2), and at least comprises a plastic deformation metal layer (7) and a self-healing alloy layer (8) positioned on the surface of the plastic deformation metal layer (7).
- 2. The cavity structure of the wafer level package crystal oscillator of claim 1, wherein the side wall of the annular buffer microcavity (4) is an inclined plane or a stepped structure; The width of the annular buffer microcavity (4) is 5-50 mu m, the depth is 5-30 mu m, and the depth is 20% -60% of the depth of the main cavity (3).
- 3. The cavity structure of the wafer-level package crystal oscillator according to claim 1, wherein an electromagnetic shielding layer (9) is arranged on the inner wall of the annular buffer microcavity (4).
- 4. The cavity structure of a wafer level package crystal oscillator according to claim 1, characterized in that the composite metal structure further comprises a diffusion barrier layer (10) under the plastically deformable metal layer (7).
- 5. The cavity structure of the wafer level package crystal oscillator according to claim 4, wherein the plastic deformation metal layer (7) is ultra-fine grain copper or ultra-fine grain gold, the average grain size of the ultra-fine grain copper or ultra-fine grain gold is smaller than 200nm, the self-healing alloy layer (8) is an Au-Sn, au-Ge or Au-In eutectic alloy, and the diffusion barrier layer (10) comprises at least one of Ti, tiN, taN or Ru.
- 6. The cavity structure of a wafer level package crystal oscillator according to any one of claims 1-5, characterized in that the thickness of the plastically deformable metal layer (7) is 1-5 μm and the thickness of the self-healing alloy layer (8) is 0.5-2 μm.
- 7. The bonding method of the wafer level packaging crystal oscillator is characterized by comprising the following steps of: -providing a device wafer (2) and a cap wafer (1) according to any of claims 1-6; applying a first temperature T1 and a first pressure P1 in a vacuum or inert atmosphere environment to enable the plastic deformation metal layer (7) of the bonding ring (6) to be subjected to plastic deformation so as to realize pre-bonding; And (3) carrying out post-treatment on the bonding ring (6) after pre-bonding, and activating the self-healing alloy layer (8) at a second temperature T2 to realize self-healing reinforcement bonding.
- 8. The bonding method according to claim 7, wherein the first temperature T1 is 150-250 ℃, the first pressure P1 is 50-200MPa, and the second temperature T2 is 10-50 ℃ higher than the melting point of the self-healing alloy layer (8).
- 9. The bonding method according to claim 7 or 8, characterized in that the post-treatment is a laser selective annealing treatment, and the area of the bonding ring (6) is scanned at a scanning speed of 200-800mm/s by using a laser beam with a wavelength of 808nm or 1064 nm.
- 10. The bonding method according to claim 7 or 8, wherein the post-treatment is global thermal annealing, and the pre-bonded wafer pair is annealed at a temperature of 230-320 ℃ for 30 seconds to 15 minutes.
Description
Cavity structure of wafer level packaging crystal oscillator and metal hot-press bonding method Technical Field The invention relates to the technical field of micro-electro-mechanical systems and semiconductor packaging, in particular to a cavity structure of a wafer level packaging crystal oscillator and a metal hot-press bonding method. Background With the rapid development of the fields of 5G communication, internet of things, vehicle-mounted electronics, high-performance computing and the like, the clock source serving as a heart beat reference of an electronic system has increasingly stringent performance requirements. Quartz crystal resonators and microelectromechanical system (MEMS) resonators are widely used in various clock circuits due to their excellent frequency stability. Wafer level packaging technology has become the dominant packaging scheme for the resonators described above in pursuit of smaller size, lower cost and higher reliability. The conventional wafer-level packaging crystal oscillator generally adopts a structure that a single concave cavity is formed on a silicon cap wafer through dry etching or wet etching, and then the single concave cavity is sealed and jointed with a device wafer with a resonator by means of metal hot-press bonding, eutectic bonding or glass paste bonding and the like, so that a protection cavity is formed. However, this conventional architecture exposes several inherent technical bottlenecks when dealing with high-end application scenarios: The problem of mechanical stress sensitivity is that the bonding process itself generates large thermo-mechanical stress and the package is also subject to external stress in subsequent module assembly and use environments. The traditional single-cavity structure lacks an effective stress buffering mechanism, stress can be directly transmitted to a quartz wafer or an MEMS resonance beam which is extremely sensitive to the stress through a rigid bonding interface, so that the resonance frequency of the quartz wafer or the MEMS resonance beam has non-negligible drift, and the time sequence precision of the system is directly affected. Long term hermeticity and reliability challenges the quality of metal thermocompression bonding is highly dependent on interface flatness and cleanliness. Microscopic scale surface relief, oxide layers, or contaminants can cause micro-or even nano-scale gaps to form at the bonding interface. These initial defects may spread under the stress effects of temperature cycling, mechanical impact, etc., resulting in gradual degradation of the air tightness of the enclosure cavity, intrusion of moisture or harmful gases, accelerated degradation of resonator performance, and even failure. Although the use of ductile metals such as Au, cu, etc. can improve the fit, it is difficult for a single material system to achieve an optimal balance between plastic deformability, interfacial diffusion barrier, and high temperature stability. The electromagnetic interference shielding capability is insufficient, and the shielding effect of the silicon material on electromagnetic waves is limited. In a complex electromagnetic environment, high-frequency noise easily penetrates through the silicon cap cavity, and interferes with the electromechanical energy conversion process of the resonator, so that phase noise is deteriorated or spurious frequencies are generated. This problem is particularly acute for clock chips integrated near high speed SerDes, radio frequency front ends, and the like. The existing solutions add metal shields outside the package, but this adds size, weight and assembly complexity, which is not compatible with the original purpose of highly integrating wafer level packages. The difficult problem of controlling the pollution of process particles is that tiny particle pollutants are inevitably generated in the wafer-level bonding process. In conventional planar bonding interfaces, these particles may be present directly in the seal ring region, resulting in local bonding failure, formation of a blow-by channel, and severely reduced yield. To partially solve the above problems, some improvements have been tried in the industry. For example, there are proposals to design a raised sealing dam or a complex multi-layered metal structure around the bonding ring to improve the sealing property, and to coat a conductive layer on the inner wall of the cavity to provide a certain electromagnetic shielding. However, these improvements tend to focus on a single problem or introduce new complexity and cost that fail to systematically co-address multiple challenges of stress, air tightness, shielding, and pollution control. For example, complex multi-layer metal rings may exacerbate thermally mismatched stresses, while inner wall coatings are not beneficial for low frequency stress isolation. Therefore, an innovative wafer-level packaging crystal oscillator structure and bonding method are needed, which can be