Dual-recessed inorganic self-aligning optical coupling device with quasi-telecentric beam relay for co-packaged optics applications, and method applied to the same

Abstract

The present invention discloses a double-concave inorganic self-aligned optical coupling device and method for CPO applications. A double-concave inorganic self-aligned optical coupling device for CPO applications according to the present invention is an optical coupling device for coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a first optical block mounted in a trench of the silicon photonics chip; a second optical block detachably disposed opposite to the first optical block; a first lens and a first beam folding element provided in the first optical block for collimating light from the edge coupler into an expanded beam; a second lens and a second beam folding element provided in the second optical block for refocusing the expanded beam; and an optical coupling gap formed between the first lens and the second lens, wherein the first lens and the second lens are recessed within the surface of each corresponding optical block or are disposed oppositely to maintain a mutually non-contact state during coupling and separation, and the first beam folding element and the second beam folding element provide double 90° beam folding for the optical path between the edge coupler and the fiber array unit, and All optical components of the optical path between the edge coupler and the fiber array unit are characterized by being composed of inorganic materials.

Description

Technology Field

[0001] The present invention relates to a dual 90° beam folding optical coupling device for co-packaging optics (CPO) applications, and more specifically, to a dual concave inorganic self-aligned optical coupling device and method for CPO applications for coupling between an edge coupler of a silicon photonics (SiPh) chip and a fiber array unit (FAU).

[0002] Specifically, the present invention optimizes the expanded beam diameter and cavity air gap distance based on Gaussian beam coupling theory, thereby achieving self-alignment characteristics that achieve coupling loss of 0.5 dB or less in ±15 μm lateral misalignment, while maintaining an ultra-thin total system height of about 0.90 to 1.22 mm. Background Technology

[0003] Recently, AI accelerators and high-performance computing systems are facing significantly higher demands than in the past regarding data transfer bandwidth both within and between packages, driven by improvements in computational performance. In particular, with the rapid advancement of AI accelerators, data transfer bandwidth is trending toward evolving from 1.6 Tbps to 3.2 Tbps; consequently, the limitations of relying solely on electrical interconnects in terms of signal loss, power consumption, heat generation, and wiring density are becoming increasingly evident. For this reason, the need for Co-packaging Optics (CPO) technology, which aims to directly integrate optical interconnects at the package level, is gaining significant prominence. Specifically, in high-density HBM4-class packaging environments, optical coupling structures must be miniaturized, maintain low-loss characteristics at high transmission bandwidths, and ensure compatibility with mass production processes. Furthermore, as the market for immersion-cooled data centers grows rapidly, optical coupling structures are no longer sufficient with mere optical performance; they must also be able to ensure chemical stability and thermal reliability even in environments where they are exposed to cooling fluids for extended periods.

[0004] To address these technical demands, various optical coupling technologies have been proposed; however, while each technology may be effective under specific conditions, they have revealed limitations in simultaneously satisfying requirements such as high-bandwidth AI packaging, ultra-small packaging, assembly based on passive alignment, compatibility with immersion cooling, and high-temperature process compatibility. For example, metal PIC coupler-based technologies often rely on active alignment, leading to complex assembly processes and increased costs. Although active alignment methods offer the advantage of maximizing optical performance, they require separate measurements and corrections during the alignment process, prolong assembly times, and can place a heavy burden on equipment when applied to mass production processes. Furthermore, structures utilizing metal mirrors may raise questions regarding long-term chemical stability in immersion cooling environments, and dependence on the intellectual property of specific companies could potentially limit design freedom or commercialization autonomy. Moreover, such structures may be disadvantageous in meeting the requirements of next-generation high-performance packages, as bandwidths generally remain at the approximately 30 nm level or insertion losses can be as low as 4 to 5 dB.

[0005] Polymer-based PhotonicBump or Plug-type technologies are noteworthy for introducing the concept of self-alignment; however, due to their polymer structures based on multiple reflections, they face structural limitations in maintaining reliability under harsh conditions, such as 260°C reflow processes or immersion cooling environments. Generally, polymer materials may be at a disadvantage compared to inorganic materials in terms of thermal stability, chemical resistance, and long-term reliability, and the possibility of optical surface deformation or material degradation cannot be ruled out. Furthermore, surface coupling-based structures may be structurally constrained in terms of bandwidth, which may limit their ability to handle ultra-high-speed optical signal transmission. Therefore, while these technologies may be suitable for a certain range of applications, they are unlikely to be a sufficient solution in environments requiring high speed, high reliability, and high-temperature processing capabilities simultaneously.

[0006] Metasurface or grating-based COUPE-based technologies can be viewed as an advanced approach in terms of high-density integration and optical path control; however, their significant dependence on specific foundry processes can act as a limitation. This process dependence restricts manufacturing options and makes it difficult to scale to other processes or design platforms. Furthermore, since they are often designed with non-removable structures, they offer low flexibility in terms of post-assembly maintenance, component replacement, and rework. Bandwidth also tends to remain at the approximately 30 nm level, which can be a constraint when wider broadband characteristics are required. Ultimately, while such technologies offer strengths in high-density implementation, they have limitations in terms of repairability, system operational flexibility, and manufacturing independence.

[0007] Past GF / IBM family technologies adopted a structure centered on back cavities and gratings, which differs in structural premise from approaches that combine light at the edge. While back cavity or grating methods may be advantageous for specific optical path formation methods, they face difficulties in directly meeting requirements such as optical path placement within ultra-thin packages, passive insertion assembly, detachability, and high-precision edge coupling structures. In other words, while these technologies were meaningful in specific eras and under specific structural conditions, they can be seen as somewhat diverging in direction from the assembly, maintainability, high integration, and reliability requirements demanded in the recent CPO package environment.

[0008] Meanwhile, extended beam free-space coupling technology has long been utilized in multi-core optical connectors for military and aerospace applications to ensure resistance to dust and contamination. For instance, the extended beam method, utilizing beam diameters of 80 to 200 μm, offers the advantage of lower contamination sensitivity compared to contact-type structures that directly abut optical fiber cross-sections, and allows for a certain level of alignment margin through free-space transmission. However, these connectors are fundamentally designed based on the premise of independent fiber-to-fiber interconnects at the cable or backplane level. Consequently, they cannot be viewed as structures that account for on-chip integration with silicon photonics edge couplers, passive alignment using lithography-defined trenches formed on photonics dies, dual 90° beam folding in environments that strictly limit the package's Z-height, and the thermal and chemical conditions unique to CPO, such as 260°C reflow and immersion cooling. In other words, while extended beam technology itself is known, it does not directly meet the requirements of highly integrated CPO packages.

[0009] Intel's silicon photonics edge coupling technology also holds significant technical importance in that it changes the direction of light propagation through a 45° mirror. However, this method is generally designed based on the premise of permanent fiber attachment and differs from structures that consider repeated coupling and separation through detachable interfaces. Furthermore, separate cavity structures to protect lenses or optical surfaces during coupling and separation may not be sufficiently considered, and it is difficult to view systematic light beam design to expand alignment tolerances as being necessarily incorporated. In other words, simply changing the beam direction using a 45° mirror alone cannot resolve all requirements regarding repeatability, alignment tolerance, and high-reliability operating conditions.

[0010] A more fundamental problem with existing self-alignment designs is the tendency to treat optical self-alignment tolerances merely as a matter of mechanical insertion precision, or to regard the concept of using an extended beam as sufficient. However, actual optical coupling performance is determined by the closely interrelated physical relationships between beam diameter, cavity air gap, and misalignment tolerance. For example, if a certain level of lateral misalignment occurs when the beam diameter is not sufficiently large, insertion loss increases sharply. In particular, in designs with a beam diameter of approximately 50 μm, according to Gaussian beam theory, a loss of about 3.13 dB can occur at the level of ±15 μm misalignment, making it difficult to secure practical self-alignment performance. This demonstrates that explanations at the level of simply "using an extended beam" cannot satisfy the actually required assembly tolerances and loss performance. Ultimately, a quantitative design methodology is required regarding how large the beam should be extended, what range the air gap through which the extended beam passes should be set, and what level the target self-alignment tolerance should be.

[0011] In summary, although conventional technologies each possess their own advantages, they fall short in comprehensively satisfying requirements for high-bandwidth AI packaging, including ultra-small optical coupling structures, low insertion loss, wide bandwidth, manual assembly capability, repeatable coupling and disassembly, 260°C reflow compatibility, chemical and thermal reliability in immersion cooling environments, and quantitative self-alignment design based on Gaussian beam physics. In particular, while existing technologies are meaningful at the level of individual component technologies, they have limitations in that they fail to simultaneously meet the multidimensional conditions required in actual next-generation CPO application environments.

[0012] Therefore, it can be said that a more sophisticated technical approach is required, going beyond simple optical path changes or the application of simple extended beams, by considering optical tolerances, package constraints, manufacturing process conditions, and operating environment conditions together. Prior art literature

[0013] Republic of Korea Published Patent Application No. 10-2025-0025510 (Published Feb. 21, 2025) The problem to be solved

[0014] Accordingly, the present invention was created to solve the above problems, and the objective of the present invention is to provide a double concave inorganic self-aligned optical coupling device and method for CPO applications that can maximize compatibility with mass production processes by reducing dependence on active alignment equipment or complex optical correction procedures during the alignment process of the optical coupling structure and enabling stable assembly with only manual alignment.

[0015] In addition, the present invention provides a dual concave inorganic self-aligned optical coupling device and method for CPO applications that can secure quantitative self-alignment characteristics, which are suppressed to a loss of 0.5 dB or less even if lateral misalignment of ±15 μm occurs during the optical coupling process, by designing the expanded beam diameter to be 125 μm or more based on Gaussian beam theory.

[0016] In addition, the present invention provides a double concave inorganic self-aligned optical coupling device and method for CPO applications, which can stably maintain beam collimation characteristics by optimizing the cavity air gap to a range of 50 μm to 300 μm so that the beam is transmitted over a short distance of about 1% of the Rayleigh range, while ensuring sufficient protective clearance to prevent direct contact between lenses during coupling and separation processes.

[0017] In addition, the present invention provides a double concave inorganic self-aligned optical coupling device and method for CPO applications that can maintain chemical stability, thermal stability, and structural reliability even when exposed to a 260°C reflow process and a immersion cooling environment for a long period of time by implementing an optical coupling interface based on an all-inorganic material.

[0018] Furthermore, the present invention provides a dual-concave inorganic self-aligned optical coupling device and method for CPO applications that can be flexibly applied to various manufacturing environments and packaging conditions and improve design freedom by adopting a non-dependent structure that is not dependent on a specific foundry process or a specific manufacturing platform.

[0019] In addition, the present invention provides a dual concave inorganic self-aligned optical coupling device and method for CPO applications that can improve system operational convenience and serviceability by forming the optical coupling interface as a detachable structure, thereby enabling replacement, inspection, reassembly, and on-site maintenance of the optical fiber-side component or coupling interface.

[0020] In addition, the present invention provides a double concave inorganic self-aligned optical coupling device and method for CPO applications that can suppress the height of the entire optical coupling structure to a level of about 0.90 mm to 1.22 mm by efficiently arranging the optical path through a double 90° beam folding structure and a horizontal FAU side coupling structure.

[0021] In addition, the present invention provides a dual concave inorganic self-aligned optical coupling device and method for CPO applications suitable for high-bandwidth optical signal transmission environments by simultaneously achieving broadband characteristics of 300 nm or more and low loss characteristics of 1 dB or less through an edge coupling-based structure.

[0022] The purpose of the invention is not limited to the purposes mentioned above, and other unmentioned purposes will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0023] A double-concave inorganic self-aligned optical coupling device for CPO applications according to a first aspect of the present invention for achieving the above objective is an optical coupling device for coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a first optical block mounted in a trench of the silicon photonics chip; a second optical block detachably disposed opposite to the first optical block; a first lens and a first beam folding element provided in the first optical block for collimating light from the edge coupler into an expanded beam; a second lens and a second beam folding element provided in the second optical block for refocusing the expanded beam; and an optical coupling gap formed between the first lens and the second lens, wherein the first lens and the second lens are recessed within the surface of each corresponding optical block or are disposed oppositely to maintain a mutually non-contact state during coupling and separation, and the first beam folding element and the second beam folding element are double 90° with respect to the optical path between the edge coupler and the fiber array unit. It is characterized by providing beam folding, and all optical components of the optical path between the edge coupler and the fiber array unit are composed of inorganic materials.

[0024] In addition, the dual concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized in that the expanded beam diameter is 80 to 130 μm, more preferably 110 to 130 μm, and the beam radius (w) is 55 to 65 μm, and is determined within the above range by lens manufacturing tolerance and mode field variation of the silicon photonics chip.

[0025] In addition, the double concave inorganic self-aligned optical coupling device for CPO application of the present invention is characterized in that the height of the first optical block is 0.15 to 0.60 mm and the height of the second optical block is 0.40 to 0.70 mm.

[0026] In addition, the double concave inorganic self-aligned optical coupling device for CPO application of the present invention is characterized in that the depth of the trench is 50 to 100 μm, the width in the X direction is 100 to 200 μm, and the length in the Y direction is 1,500 to 3,000 μm.

[0027] In addition, the dual concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized by including an array comprising 16 or more channels and having a channel pitch of 127 μm.

[0028] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the inorganic material comprises fused silica (SiO₂), a metal reflective coating comprising aluminum or gold is formed on the 45° beam folding surface, and is characterized by having broadband optical transmittance over O-band, C-band, and L-band wavelength ranges.

[0029] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the fiber array unit comprises a V-groove-based SMF or PM-SMF and is horizontally permanently bonded to the side of the second optical block with epoxy, and the detachable interface is implemented only in the optical coupling gap.

[0030] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, the first optical block and the second optical block are aligned by a flat reference plane and a fiducial, and the first lens received in the first recess and the second lens received in the second recess maintain a non-contact state with each other during coupling and separation to form a detachable structure.

[0031] In addition, the double concave inorganic self-aligning optical coupling device for CPO application of the present invention is characterized by having a total system height of 1.50 mm or less.

[0032] In addition, the double concave inorganic self-aligning optical coupling device for CPO applications of the present invention is characterized by having a peripheral UV epoxy glue stop formed around the trench for 260°C reflow resistance.

[0033] In addition, in a dual concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the first optical block is an MPLA, and the MPLA is characterized in that a lens array and a 45° prism are integrally formed on a single inorganic substrate by a wafer-level lithography process, and the lens ROC is configured to generate an expanded beam of 50 μm or more by providing a magnification of 15 to 25 times compared to a PIC MFD (3 to 6 μm).

[0034] In addition, in the double concave inorganic self-aligned optical coupling device for CPO application of the present invention, the double 90° beam folding is characterized by forming a seven-step optical path of PIC edge coupler output, H→V folding by 45° TIR, MPLA collimation, expanded beam delivery in the detachable cavity air gap, refocusing in the second optical block, V→H folding by 45° mirror, and fiber array unit coupling.

[0035] In addition, the dual concave inorganic self-aligning optical coupling device for CPO applications of the present invention is characterized by having an expanded beam diameter in the range of about 80 to 100 μm, and achieving a loss of 0.55 dB or less to 0.51 dB or less in lateral misalignment of ±10 to 12 μm depending on the corresponding beam radius.

[0036] In addition, in a double-concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the silicon photonics chip further comprises one or more grooves surrounding the trench opening on the upper surface of the chip substrate, and the grooves block the capillary flow of the adhesive so that the adhesive is captured in a reservoir channel before reaching the waveguide cross-section inside the trench, thereby protecting the waveguide cross-section from adhesive contamination during assembly.

[0037] In addition, the double concave inorganic self-aligning optical coupling device for CPO application of the present invention is characterized in that the refractive index of the anti-reflective coating formed on the lens surface is located between the ideal anti-reflective refractive index for air and the ideal anti-reflective refractive index for immersion cooling fluid, so that the residual reflectance of each coating surface is maintained below a predetermined threshold value in both air and immersion cooling fluid environments.

[0038] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, a monolithic standoff rail surrounding a lens accommodated in the first recess and the second recess is integrally formed in each of the first optical block and the second optical block, and when coupled, the standoff rail of the first optical block and the standoff rail of the second optical block come into bidirectional contact to form a labyrinth seal geometry that inhibits the penetration of an immersion cooling fluid into the recessed lens cavity formed by the first recess and the second recess.

[0039] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the first optical block is an MPLA, and an inorganic anti-reflective coating is additionally formed on the silicon photonics chip trench-facing surface of the first optical block to reduce Fresnel reflection on the facing surface.

[0040] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the first optical block is an MPLA, and an inclined surface is formed on the lower part of the trench insertion portion of the first optical block, and a corresponding inclined surface is formed on the upper part of the trench of the silicon photonics chip, so that when vertically dropped, the inclined surfaces passively guide and seat the first optical block, thereby providing transverse self-centering and vertical height determination by wedge coupling.

[0041] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, one or more non-datum inclined surfaces include a clearance or relief area, and are configured such that manual insertion is guided by the inclined surfaces and final seating is defined by a part rather than all of the opposing inclined surfaces.

[0042] In addition, the double concave inorganic self-aligning optical coupling device for CPO applications of the present invention is characterized in that the inclined surfaces form surface contact on substantially all opposing surfaces upon seating.

[0043] In addition, the double concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized in that the vertical landing height is further defined by a shoulder landing structure formed on the surface of a silicon photonic chip adjacent to the trench.

[0044] In addition, a dual concave inorganic self-aligned optical coupling device for CPO applications according to a second aspect of the present invention for achieving the above objective is an optical coupling device for coupling through an expansion beam relay between a silicon photonics chip and a fiber array unit, comprising a first optical element that converts light from a silicon photonics chip into a collimated expansion beam having a beam diameter of 50 μm or more, and a second optical element that is positioned opposite to the first optical element with a free-space air gap between them to refocus the expansion beam, wherein the air gap distance is 5% or less of the Rayleigh range z_R = πw² / λ of the expansion beam, and the expansion beam diameter is set to satisfy the relationship Loss(dB) = 8.686·Δx² / w² to achieve a coupling loss of 1.0 dB or less in a lateral misalignment of ±15 μm, and the first optical element and the second optical element are mutually detachable in the air gap.

[0045] In addition, in the double concave inorganic self-aligned optical coupling device for CPO application of the present invention, the air gap distance is characterized as being 5% or less of the Rayleigh range z_R = πw² / λ of the expansion beam.

[0046] In addition, the dual concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized by satisfying Loss(dB) = 8.686·Δx² / w², and setting the beam radius w to be 62.5 μm or more to achieve a coupling loss of 0.5 dB or less in lateral misalignment of ±15 μm.

[0047] In addition, a double-concave inorganic self-aligned optical coupling device for CPO applications according to a third aspect of the present invention for achieving the above objective is an optical coupling device for co-packaging optics in an immersion cooling environment, comprising a first silicon optical block including a 45° prism that achieves total reflection without a metal reflective coating, a second silicon optical block including a 45° mirror that achieves total reflection without a metal reflective coating, an inorganic anti-reflective coating formed on the silicon-air interface of the first silicon optical block and the second silicon optical block, and a removable air gap formed between the first silicon optical block and the second silicon optical block, wherein the first silicon optical block is bonded to a silicon photonics chip and withstands 260°C Pb-free reflow, the second silicon optical block is bonded to a fiber array unit, and the total reflection conditions in the 45° prism and the 45° mirror are maintained even when an immersion cooling liquid with a refractive index of 1.80 or less penetrates the air gap region.

[0048] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the first silicon optical block and the second silicon optical block are formed of silicon, the 45° prism and the 45° mirror are implemented without a metal coating by total internal reflection within silicon, and the silicon-air interface is configured such that a Si₃N₄ anti-reflective coating with a thickness of about 164 nm is applied to the silicon-air interface so that residual reflection is less than 0.5%.

[0049] In addition, in the double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the ROC of the silicon microlens is optimized for the change in thermo-optical refractive index at an operating temperature of 105°C, and is configured to achieve optimal coupling efficiency under full-load operating conditions rather than room-temperature assembly conditions.

[0050] In addition, the double concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized by being configured such that a 45° total reflection condition is maintained even when an immersion cooling liquid with a refractive index of 1.80 or less penetrates into the air gap due to the high refractive index of silicon.

[0051] In addition, in a double concave inorganic self-aligned optical coupling device for CPO application of the present invention, the second silicon optical block is formed of silicon, and epoxy adhesion with the fiber array unit is provided by a native oxide film on the silicon surface, and epoxy fillet beads are formed on four sides of the fiber array unit bonding line to prevent capillary wicking of the immersion cooling liquid.

[0052] In addition, a dual-concave inorganic self-aligned optical coupling device for CPO applications according to a fourth aspect of the present invention for achieving the above objective is an optical coupling device for optical coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a dual-recess cavity structure in which an aspherical convex lens accommodated in a first recess on the upper surface of a first optical block and an aspherical convex lens accommodated in a second recess on the lower surface of a second optical block are arranged oppositely with an air gap of 50 to 300 μm; a monolithic MPLA in which a 45° TIR prism and an aspherical convex microlens are integrally formed on a single inorganic substrate to generate a collimated expanded beam of 50 μm or more; an integrated mirror-microlens structure in which a 45° mirror and a microlens are integrally formed; a passive alignment structure in which the MPLA is passively inserted into an etched trench of a silicon photonics chip; and such that the expanded beam diameter achieves a coupling loss of 0.5 dB or less at ±15 μm misalignment according to Gaussian beam theory. It is characterized by comprising a set quasi-telecentric self-aligned beam relay, an all-inorganic structure in which all optical components within the optical path are composed of inorganic materials; and a horizontal FAU side alignment structure in which a fiber array unit is horizontally joined to the side of a second optical block, wherein optical coupling is achieved through a double 90° beam folding of a first beam folding (H→V) by the TIR prism of the MPLA and a second beam folding (V→H) by the mirror of the second optical block.

[0053] And, a dual-concave inorganic self-aligned optical coupling method for CPO applications according to the fifth aspect of the present invention for achieving the above objective is an optical coupling method between an edge coupler of a silicon photonics chip and a fiber array unit, comprising the steps of: inserting an MPLA block including a monolithic MPLA in which a 45° TIR prism and an aspherical convex microlens are integrally formed on a single inorganic substrate into a trench of the silicon photonics chip; positioning a connector block including a 45° mirror opposite the MPLA block so that the lenses of both blocks are arranged oppositely with an optical coupling gap of 50 to 500 μm; forming a quasi-telecentric self-aligned relay by the opposing lenses with an expanded beam diameter of 50 μm or more; and coupling to a horizontal fiber array unit by a dual 90° beam folding of H→V folding by the 45° TIR of the MPLA block and V→H folding by the 45° mirror of the connector block, wherein all components of the optical path are made of inorganic material It is characterized by being composed.

[0054] In addition, the method for a double concave inorganic self-aligned optical coupling for CPO applications of the present invention is characterized by further including the step of fabricating an MPLA by a wafer-level lithography process before step (a), wherein the lens ROC is configured to provide a magnification of 15 to 25 times compared to a PIC MFD (3 to 6 μm).

[0055] In addition, the dual concave inorganic self-aligned optical coupling method for CPO application of the present invention is characterized by further including vision-based non-contact passive insertion and peripheral UV epoxy glue stop application.

[0056] In addition, in the double concave inorganic self-aligned optical coupling method for CPO applications of the present invention, step (b) is characterized by being performed after 260°C Pb-free reflow of the silicon photonics chip in an MPLA mounted state.

[0057] In addition, the double concave inorganic self-aligned optical coupling method for CPO applications of the present invention further includes the step of horizontally permanently bonding a fiber array unit to the side of a connector block with epoxy, and is characterized in that the detachable interface is implemented only in an air gap.

[0058] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, at least one of the first lens and the second lens is formed as a protruding convex lens protruding from the surface of the corresponding optical block, and the optical coupling gap is defined by the protrusion height of the protruding lens and a mechanical standoff structure.

[0059] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, an inorganic spacer made of glass, fused silica, or silicon is disposed between the first optical block and the second optical block to define an optical coupling gap between the protruding lenses, and the thickness of the spacer is determined by the height of the protruding lenses and the target optical coupling gap.

[0060] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, the first optical block comprises a protruding convex lens and the second optical block comprises a recessed convex lens, and the optical coupling gap is characterized by being determined by the protrusion height of the protruding lens, the recess depth of the recessed lens, and a mechanical standoff structure.

[0061] In addition, the double concave inorganic self-aligning optical coupling device for CPO applications of the present invention is characterized in that the entire optical path through which the optical beam passes is devoid of epoxy, polymer, or resin, and the adhesive is used only in the mechanical fixing part outside the optical beam path.

[0062] In addition, the double concave inorganic self-aligning optical coupling device for CPO application of the present invention further comprises a receptacle that accommodates the first optical block, and is characterized by including a two-stage alignment structure in which coarse alignment is achieved by a mechanical guide structure when the receptacle is inserted, and precise alignment is achieved by contact of the standoff rail when coupled.

[0063] In addition, in the double concave inorganic self-aligned optical coupling method for CPO application of the present invention, the wafer-level lithography process is characterized by forming a recess depth of 25 to 30 μm and an aspherical convex lens in a single process without a separate DRIE etching step, and the non-etched periphery forms a standoff rail.

[0064] In addition, in the dual concave inorganic self-aligned optical coupling method for CPO applications of the present invention, the second optical block is detachably coupled to the first optical block after the completion of Pb-free reflow of the silicon photonics chip, and the detachable coupling is achieved without separate active alignment by mechanical contact of the standoff rail.

[0065] In addition, the dual concave inorganic self-aligned optical coupling device for CPO applications of the present invention is characterized in that a plurality of first optical blocks are each mounted in a plurality of trenches of a single silicon photonics chip, or the number of channels of a single first optical block is expanded to 32, 42, 64, 96 or more, and the channel pitch is 127 μm or 80 μm or less.

[0066] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, the groove is formed by two concentric grooves with a depth of 10 to 20 μm and a width of 20 to 30 μm and a spacing of 30 to 50 μm from each other, forming a closing ring that surrounds the trench opening.

[0067] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, the standoff rail of the first optical block and the standoff rail of the second optical block come into contact with a mechanical interface gap of substantially 0 μm when coupled, and the optical coupling gap is formed separately within a recess cavity formed by the first recess and the second recess.

[0068] Additionally, in a double-concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the device further comprises a receptacle (200) permanently mounted on a silicon photonics chip substrate, wherein the receptacle (200) comprises a frame having three or more walls, a bottom opening providing a window to the trench, and one or more spring-loaded ball detents (220) each embedded in the respective wall, wherein a plug body (210) for receiving the second optical block and fiber array unit is detachably received in the receptacle (200), and wherein the plug body (210) has one or more dimples (214) on its outer surface corresponding to the one or more ball detents (220), and wherein, upon vertical insertion of the plug body (210), the spring-loaded ball (221) engages with the dimple (214) to provide a detachable mechanical retention.

[0069] In addition, in the double concave inorganic self-aligning optical coupling device for CPO application of the present invention, the force direction of each ball detent (220) is horizontal and is orthogonal to the vertical optical axis of the extension beam relay between the first optical block and the second optical block.

[0070] In addition, in a double concave inorganic self-aligned optical coupling device for CPO applications of the present invention, the plug body (210) comprises a first region (211) having upper and lower window openings for optical coupling and heat dissipation of the second optical block; and a second region (212) having a sandwich structure with closed upper and lower frame surfaces to hold the fiber array unit substrate, wherein the wall height of the first region (211) has a stepped wall (213) shorter than the wall height of the second region (212) to prevent interference with the upper surface protrusion of the first optical block.

[0071] In addition, in a double concave inorganic self-aligning optical coupling device for CPO applications of the present invention, the receptacle (200) and the plug body (210) are absent from a micro-alignment feature, optical alignment between the first optical block and the second optical block is achieved entirely by direct contact of the standoff rail, and the ball detent (220) retention is mechanically independent of the optical alignment interface. Effects of the invention

[0072] Accordingly, the present invention allows for quantitative prediction and control of coupling loss even when lateral misalignment occurs, by setting the expanded beam diameter, beam radius, and optical coupling gap based on Gaussian beam theory. Accordingly, it has the effect of securing excellent self-aligned optical coupling performance of 0.50 dB or less even with a misalignment of ±15 μm and 0.22 dB or less even with a misalignment of ±10 μm.

[0073] In addition, the present invention has the effect of preventing direct contact between lenses during the coupling and coupling process by forming an apex gap of about 94 μm between opposing lenses and securing a protective clearance of about 47 μm for each lens, thereby reducing the possibility of damage to the optical surface during repeated coupling and coupling and improving the long-term reliability of the coupling interface.

[0074] In addition, the present invention is configured so that an optical block is aligned and mounted based on a trench structure formed on a silicon photonics chip, thereby allowing it to be flexibly applied to various foundry processes that support DRIE trench formation and has the effect of reducing dependence on a specific manufacturer or a specific process platform.

[0075] In addition, since the present invention can implement an optical coupling structure by combining a passive alignment structure rather than an active alignment structure with a wafer-level optical (WLO) placement process, the assembly process can be simplified and compatibility with mass production processes can be improved, which has an advantageous effect in terms of mass producibility.

[0076] In addition, the present invention has the effect of improving long-term reliability by ensuring that the main optical components within the optical path are composed of all-inorganic materials such as fused silica or silicon, thereby maintaining thermal stability, chemical stability, and structural reliability even under harsh conditions such as a 260°C reflow process and a immersion cooling environment.

[0077] In addition, the present invention can be implemented with a height of approximately 0.90 mm to 1.22 mm and can satisfy a profile of up to 1.50 mm or less, and in particular, through a rail-contact standoff design and an apex gap structure of approximately 94 μm, it has the effect of stably positioning the optical coupling structure even within a limited Z-direction package budget.

[0078] In addition, the present invention can secure broadband characteristics of 300 nm or more by adopting an edge coupling-based optical coupling structure and can suppress insertion loss to a level of 1 dB or less, thereby providing excellent optical performance suitable for co-packaging optical environments where high-bandwidth optical signal transmission is required.

[0079] In addition, the present invention can efficiently redirect the optical path through a dual 90° beam folding structure using a 45° TIR prism and a 45° mirror and a horizontal FAU coupling structure, thereby enabling the coupling of fiber array units in the horizontal direction without further increasing the height in the Z direction, which has an advantageous effect for realizing an ultra-thin package.

[0080] The effects of the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description in the claims. Brief explanation of the drawing

[0081] FIG. 1 is a schematic diagram showing an optical coupling device according to one embodiment of the present invention. Figure 2 is a diagram showing Z stacking as a YZ cross-sectional view (channel array plane) of the optical coupling device of Figure 1. Figure 3 is a diagram showing a 7-step beam path as an XZ cross-sectional view (optical propagation plane) of the optical coupling device of Figure 1. FIG. 4 is an XY cross-sectional plan view (detachable interface surface) of the optical coupling device of FIG. 1, showing the double recess cavity geometry, 16-channel array arrangement, DRIE trench details, fiducial marks, and recess lens shape. Figure 5 is a diagram showing the assembly and detachment flow for the optical coupling device of Figure 1. Figure 6 is a graph showing misalignment versus loss by beam diameter for the optical coupling device of Figure 1. Figure 7 is a drawing for comparing and explaining the optical coupling device of Figure 1 with the prior art. Figure 8 is a drawing showing the labyrinth groove placement plane on the upper surface of the PIC substrate for the optical coupling device of Figure 1. Figure 9 is a drawing showing a three-layer immersion cooling defense architecture for the optical coupling device of Figure 1. FIG. 10 is a drawing showing the bidirectional inclined plane-induced magnetic centering, vertical height determination by inclined plane contact and / or shoulder landing, and selective relief of a non-reference inclined plane for the optical coupling device of FIG. 1. FIG. 11 is a flowchart illustrating an optical coupling method according to one embodiment of the present invention. FIG. 12 is an XY plan view of a receptacle of an optical coupling device according to one embodiment of the present invention, showing the arrangement of a bottom frame, a window, a bore in which a ball detent is placed, and a clearance hole. And, FIG. 13 is an XZ cross-sectional view showing the DFAU seating state of an optical coupling device according to one embodiment of the present invention, showing the state in which a ball detent in the receptacle is fastened to the dimple of the plug body, the formation of a Z-datum by contact between the standoff rail of the first optical block and the standoff rail of the second optical block, and the floating state of the guide pin. Specific details for implementing the invention

[0082] Hereinafter, embodiments will be described in detail with reference to the attached drawings. However, the scope of the patent application is not limited or restricted by these embodiments. Identical reference numerals in each drawing indicate identical components.

[0083] Various modifications may be made to the embodiments described below. The embodiments described below are not intended to limit the forms of practice and should be understood to include all modifications, equivalents, and substitutions thereof.

[0084] Terms such as "first" or "second" may be used to describe various components, but these terms should be understood solely for the purpose of distinguishing one component from another. For example, a first component may be named a second component, and similarly, a second component may be named a first component.

[0085] The terms used in the embodiments are used merely to describe specific embodiments and are not intended to limit the embodiments. A singular expression includes a plural expression unless the context clearly indicates otherwise. In this specification, phrases such as “A or B,” “at least one of A and B,” “at least one of A or B,” “A, B or C,” “at least one of A, B and C,” and “at least one of A, B, or C” may each include any one of the items listed together with the corresponding phrase, or any possible combination thereof. In this specification, terms such as “comprising” or “having” are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0086] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0087] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. When describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0088] The dual-concave inorganic self-aligned optical coupling device for Co-Packaged Optics (CPO) applications of the present invention is an optical coupling device for coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a first optical block mounted in a trench of the silicon photonics chip; a second optical block detachably disposed opposite to the first optical block; a first lens and a first beam folding element provided in the first optical block for collimating light from the edge coupler into an expanded beam; a second lens and a second beam folding element provided in the second optical block for refocusing the aforementioned expanded beam; and an optical coupling gap formed between the first lens and the second lens, wherein the aforementioned first lens and the second lens are recessed within the surface of each corresponding optical block or are disposed oppositely to maintain a mutually non-contact state during coupling and separation, and the aforementioned first beam folding element and the second beam folding element provide dual 90° beam folding for the optical path between the edge coupler and the fiber array unit, and the aforementioned All optical components of the optical path between the mentioned edge coupler and the fiber array unit are characterized by being composed of inorganic materials.

[0089] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0090] FIG. 1 is a schematic diagram showing an optical coupling device according to one embodiment of the present invention.

[0091] As illustrated in FIG. 1, the first optical block may correspond to the MPLA block (120), and the second optical block may correspond to the connector block (110). Additionally, the first lens may correspond to the MPLA recessed lens (122) formed in the MPLA block (120), and the second lens may correspond to the Block-1 recessed lens (112) formed in the connector block (110).

[0092] Additionally, the first beam folding element may correspond to a 45° TIR (Total Internal Reflection) prism (124), the second beam folding element may correspond to a Block-1 45° mirror (114), the optical coupling gap may correspond to a cavity air gap (130) formed between the connector block (110) and the MPLA block (120), and the trench of the silicon photonics chip may correspond to a DRIE trench (170). Furthermore, the expansion beam transmitted in the optical path may correspond to an expansion beam (180).

[0093] Here, MPLA stands for 'Microprism-Lens Array', and the MPLA block (120) is mounted in a DRIE (Deep Reactive Ion Etching) trench (170) formed on one side of the PIC (Photonic Integrated Circuit, 150) to receive light output from the edge coupler of the PIC (150), and the connector block (110) is positioned opposite the MPLA block (120) to redirect the light transmitted from the MPLA block (120) toward the FAU (Fiber Array Unit, 160). At this time, the MPLA block (120) can be implemented as a block that is permanently mounted to the PIC (150), and the connector block (110) can be implemented as a detachable block that can be separated and reassembled as needed. Accordingly, the optical coupling device (100) can be configured to allow replacement, inspection, or reassembly of the FAU (160) side while maintaining the optical interface on the PIC (150) side, thereby ensuring both maintenance convenience and field serviceability. This configuration may be an embodiment in which the MPLA block (120) and the connector block (110) are applied as heterogeneous blocks performing different functions, but in other embodiments, both blocks may be designed to perform different optical functions while being formed from the same inorganic substrate series.

[0094] Specifically, the MPLA block (120) is mounted in a DRIE trench (170) formed on one side of the PIC (150) and performs the role of receiving light output from the edge coupler of the PIC (150), and the connector block (110) is positioned opposite the MPLA block (120) and performs the role of redirecting the light transmitted from the MPLA block (120) toward the FAU (160). At this time, the MPLA block (120) can be implemented as a block that is permanently mounted to the PIC (150), and the connector block (110) can be implemented as a detachable block that can be separated and reassembled as needed. Accordingly, the optical coupling device (100) can be configured to allow replacement, inspection, or reassembly of the FAU (160) side while maintaining the optical interface on the PIC (150) side, and as a result, both maintenance convenience and field serviceability can be secured. This configuration may be an embodiment in which the MPLA block (120) and the connector block (110) are applied as different types of blocks that perform different functions, but in other embodiments, both blocks may be designed to be formed from the same inorganic substrate series while performing different optical functions.

[0095] Additionally, the MPLA recess lens (122) formed in the MPLA block (120) can be configured to receive light with a relatively small mode field diameter output from the edge coupler of the PIC (150) and then collimate it into an expanded beam (180) in an enlarged state. At this time, the MPLA recess lens (122) is formed in an aspherical convex shape to expand the beam to a target beam diameter while correcting the light divergence characteristics, and in a preferred embodiment, the diameter of the expanded beam (180) can be designed to be in the range of about 110 to 130 μm, more preferably about 125 μm nominally. Accordingly, self-alignment characteristics can be secured so that optical coupling loss does not increase rapidly even if mechanical assembly errors exist, and in particular, the effect of expanding the tolerance range for lateral misalignment can be expected. In other embodiments, the actual diameter of the extended beam (180) may be set to 80 to 100 μm depending on the lens manufacturing tolerance, the mode field variation of the PIC (150), or the target application wavelength range, and in this case, it may also be configured to enable effective optical coupling within the design purpose and allowable loss range.

[0096] The 45° TIR prism (124) can function as a first beam folding element that redirects light output in a substantially horizontal direction from the edge coupler of the PIC (150) into a vertical direction. In this case, the 45° TIR prism (124) may be formed integrally within the main body of the MPLA block (120), or in another embodiment, it may be formed in a shape protruding downward from the bottom surface of the MPLA block (120) and configured to be partially inserted into the DRIE trench (170). Accordingly, the direction of light propagation in the waveguide inside the PIC (150) and the direction of light propagation in the cavity air gap (130) section can be switched to be mutually orthogonal, and light paths that are difficult to directly combine on the same plane can be efficiently converted within a short distance. In particular, the 45° TIR prism (124) can be formed to fold light by total reflection without a metal reflective coating, and in an embodiment where the material is silicon, stable total reflection conditions can be secured by utilizing a larger difference in refractive index. On the other hand, in an embodiment applying fused silica, it may be implemented by further applying a metal reflective layer to the 45° folding surface.

[0097] Additionally, the Block-1 45° mirror (114) can function as a second beam-folding element that redirects the light refocused by the Block-1 recess lens (112) back to a horizontal direction. Accordingly, the light path, which has been converted once from horizontal to vertical by the 45° TIR prism (124), can be converted again from vertical to horizontal by the Block-1 45° mirror (114), and as a result, the optical coupling device (100) can implement a double 90° beam-folding path of H→V→H. With such a double 90° beam-folding structure, the optical coupling device (100) can minimize the increase in the height of the entire device while horizontally positioning the FAU (160) on the side of the connector block (110), and can be advantageously applied to ultra-thin mounting conditions required in a co-packaging optical environment. In another embodiment, the Block-1 45° mirror (114) may be formed as a mirror surface having a separate reflective layer, and in yet another embodiment, it may be formed as a structural reflective surface utilizing the difference in the refractive index of the material.

[0098] Additionally, the cavity air gap (130) is an optical free space gap formed between the MPLA block (120) and the connector block (110), and can be understood as a concept distinct from the mechanical interface gap. That is, the cavity air gap (130) is not a simple gap but refers to an optical coupling section where the expansion beam (180) is transmitted in a non-contact state, and in a preferred embodiment, it can be set to a range of 50 to 300 μm, more preferably at a nominal level of about 94 μm. When the cavity air gap (130) is formed to a level that is very small compared to the Rayleigh range, beam divergence in that section becomes substantially negligible, and the collimated state of the expansion beam (180) can be stably maintained. At the same time, the cavity air gap (130) can serve to prevent lens damage during repeated removal and attachment of the connector block (110) by providing a protective clearance that prevents direct contact between the MPLA recess lens (122) and the Block-1 recess lens (112). In a preferred embodiment, a protective clearance of approximately 47 μm per lens may be provided by the recess depth of both blocks, but in other embodiments, the value may be adjusted according to changes in the block structure, recess depth, or standoff structure.

[0099] The DRIE trench (170) can function as a receiving portion where the MPLA block (120) is aligned and mounted to the PIC (150). In a preferred embodiment, the depth of the DRIE trench (170) may be formed to be in the range of 50 to 100 μm, the width in the X direction to be 100 to 200 μm, and the length in the Y direction to be 1,500 to 3,000 μm, and these dimensions may be set considering the insertion stability of the MPLA block (120), optical axis alignment precision, and multi-channel array length. The vertical sidewalls of the DRIE trench (170) may function as mechanical reference planes for the MPLA block (120) to enable manual drop-in alignment, and if necessary, a glue stop or labyrinth structure may be further formed around the trench to suppress the internal inflow of adhesive. In another embodiment, a corresponding inclined surface may be formed on the upper part of the DRIE trench (170) or on the lower part of the insertion portion of the MPLA block (120) so that self-centering can be more easily achieved during vertical insertion. Accordingly, the optical coupling device (100) can be implemented to enable repeatable passive alignment without active alignment.

[0100] The FAU (160) can be implemented as a component that is horizontally positioned on the side of the connector block (110) to finally receive light that has passed through the Block-1 45° mirror (114). In a preferred embodiment, the FAU (160) may include a V-groove-based SMF or PM-SMF array, with a channel pitch of 127 μm and at least 16 channels. Accordingly, multi-channel optical signals can be stably transmitted within a narrow array width, and connection with an external optical ribbon or external optical communication module can be easily implemented through the horizontally aligned FAU (160). In other embodiments, the number of channels may be increased or decreased, or the channel pitch may be changed to suit the design environment, and the side coupling structure of the connector block (110) may also be adjusted accordingly. Additionally, the FAU (160) can be attached to the connector block (110) by epoxy bonding, alignment reference plane bonding, or other fixing methods, and the detachable function can be implemented in the cavity air gap (130) between the connector block (110) and the MPLA block (120) rather than in the FAU (160) itself.

[0101] Additionally, the flat reference plane (190) and the fiducial can function as reference structures for aligning the relative positions of the MPLA block (120) and the connector block (110). The flat reference plane (190) can act as a mechanical reference plane that regulates the placement direction and rotational state of both blocks, and the fiducial can function as a reference mark recognized by a vision system or alignment equipment to reduce alignment errors in the channel arrangement direction. Accordingly, the MPLA block (120) and the connector block (110) can be positioned opposite each other with high channel alignment precision even with the cavity air gap (130) between them, and as a result, uniform coupling characteristics can be secured throughout the multi-channel optical path. In other embodiments, the shape, position, or number of the flat reference plane (190) may be changed, and the fiducial may also be modified into a single mark, multiple marks, a linear pattern, or a dot pattern.

[0102] Additionally, the optical coupling device (100) illustrated in FIG. 1 represents a typical recessed lens embodiment, but in other embodiments, it may be modified into a protruding lens structure in which one or both of the MPLA recessed lens (122) and the Block-1 recessed lens (112) protrude from each block surface. In such a protruding lens embodiment, the optical coupling gap between opposing lens surfaces may be determined by the protrusion height and the mechanical standoff structure, and in some cases, may be configured to fill the gap with a refractive index matching medium such as optical epoxy or gel. Even in this case, the overall self-alignment and optical path redirection scheme by the MPLA block (120), connector block (110), 45° TIR prism (124), Block-1 45° mirror (114), DRIE trench (170), expansion beam (180), flat reference plane (190), and fiducial may be maintained. Accordingly, the city form of FIG. 1 corresponds only to one preferred example of the present invention, and the lens shape, recess depth, block height, material type and alignment method can be varied in various ways without departing from the technical spirit of the present invention.

[0103] Furthermore, the optical coupling device (100) of FIG. 1 can be described based on a single-channel optical path, but in an actual implementation, multiple channels may be repeatedly arranged in the channel pitch direction to form a multi-channel optical coupling array. In this case, for each channel, an MPLA recessed lens (122), a Block-1 recessed lens (112), a 45° TIR prism (124), a Block-1 45° mirror (114), and a corresponding optical fiber channel may be repeatedly arranged, and the entire array alignment by a flat reference plane (190) and a fiducial may be applied together. Accordingly, the present embodiment is not limited to a single-channel structure but can be extended to a multi-channel structure of 16 channels or more required in a high-bandwidth CPO package.

[0104] More specifically, the MPLA block (120) can be implemented as a block permanently mounted on the PIC (150) side, and the connector block (110) can be implemented as a detachable block together with the FAU (160). Accordingly, the optical coupling device (100) can be configured to allow the connector block (110) or the FAU (160) side to be separated, inspected, replaced, or reassembled while maintaining the optical interface on the PIC (150) side. In particular, the MPLA block (120) can be stably positioned relative to the PIC (150) through a structure that accommodates the DRIE trench (170), and the connector block (110) can be positioned opposite the MPLA block (120) in a predetermined direction to form a cavity-type detachable interface. By this structure, the present embodiment can be extended to a structure that secures field maintainability and repetitive mate / demate characteristics, rather than a simple permanent adhesive optical coupler.

[0105] Additionally, the MPLA recess lens (122) may be formed to receive light output from the edge coupler of the PIC (150) and collimate it into an expanded beam (180) in an enlarged state. In a preferred embodiment, the diameter of the expanded beam (180) may be formed in the range of 80 to 130 μm, in a more preferred embodiment in the range of 110 to 130 μm, and in an even more preferred embodiment in the nominal level of about 125 μm. At this time, the beam radius (w) of the expanded beam (180) may be set in the range of 55 to 65 μm, preferably about 62.5 μm, and the above values ​​may be determined according to the radius of curvature of the MPLA recess lens (122), the edge coupler mode field diameter of the PIC (150), the operating wavelength, and the system magnification. In particular, since the actual diameter of the expanded beam (180) may vary within the above range depending on the lens manufacturing tolerance and the mode field variation of the PIC (150), this embodiment can be understood as a design parameter that encompasses the entire range of 80 to 130 μm, rather than being limited to a specific single value. This configuration is advantageous for securing self-alignment performance based on Gaussian beam coupling theory.

[0106] Figure 2 is a diagram showing Z stacking as a YZ cross-sectional view (channel array plane) of the optical coupling device of Figure 1.

[0107] As shown in FIG. 2, an FAU (160) is positioned at the top, a detachable connector block (110) is positioned at the bottom, an MPLA block (120) that is permanently mounted on a PIC (150) is positioned at the bottom, and a PIC (150) and a mounting substrate structure can be sequentially stacked at the bottom. With such a stacked structure, the optical coupling device (100) can maintain a multi-channel optical coupling function while keeping the height in the Z direction within a limited package budget.

[0108] Additionally, FIG. 2 indicates that the detachable interface between the connector block (110) and the MPLA block (120) is not formed only by an open gap, but includes a rail contact structure in which the outer non-etched portions of both blocks come into contact with each other. That is, a standoff rail may be provided on the opposing surfaces of both blocks separately from the internal cavity where the recessed lens is formed, and a mechanical reference surface is formed by the contact of the standoff rail, while a cavity air gap (130) is maintained internally to ensure a non-contact state between the lenses. In a preferred embodiment, the cavity height may be formed to be approximately 100 μm and the apex gap between the lens vertices to be approximately 94 μm, thereby preventing lens damage during detachment while securing an optically stable free-space transmission section.

[0109] Additionally, FIG. 2 shows that the FAU (160) can be implemented as an array of 16 channels or more with a channel pitch of 127 μm, and accordingly, the connector block (110), MPLA block (120), and PIC (150) can also be extended to a corresponding length along the channel array direction. Therefore, the cross-section shown in FIG. 2 can be understood as representing a multi-channel array type optical interconnect structure in which the same cross-sectional structure is repeated in the channel array direction, rather than a single-channel dedicated structure.

[0110] Additionally, as shown in FIG. 2, the height of the connector block (110) can be formed in the range of about 0.50 to 0.60 mm, and the height of the MPLA block (120) can be formed in the range of about 0.45 to 0.55 mm, and the PIC (150), solder bumps and underfill, and the lower substrate or interposer structure can be stacked together so that the overall system height can be realized at a level of about 0.90 to 1.22 mm. Accordingly, while satisfying the design condition of limiting the total system height to 1.50 mm or less, sufficient optical and mechanical interface space can be secured between the upper FAU (160) and the lower PIC (150).

[0111] Furthermore, FIG. 2 illustrates a substantial CPO package cross-section through a structure in which an MPLA block (120) is mounted to a PIC (150) via a DRIE trench (170), and a connector block (110) is positioned oppositely on top thereof. Accordingly, the optical coupling device (100) of the present invention can be implemented not as a simple combination of optical components, but as a system-level interconnect combined with a package mounting structure. In particular, the embodiment illustrated in FIG. 2 represents a stacking example that simultaneously satisfies an all-inorganic structure excluding polymers within the optical path, a high-temperature reflow-compatible structure, and an immersion-cooling-compatible structure, and can be understood as a representative embodiment suitable for a co-packaging optical environment for high-density data centers.

[0112] In addition, regarding the Z stacking described above, you can refer to Table 1 below.

[0113] floor Components Z dimension (mm) note ① Solder bump + underfill 0.05 - 0.10 FCBGA / Si Interposer ② PIC die (thin SOI) 0.08 - 0.18 ~80 ~ 180 μm ③ MPLA protrusion (H_MPLA - Trench) 0.35 - 0.48 0.45-0.55 mm - trench (70-100 μm) ④ Cavity Air Gap (Self-alignment Optimization) 0.10 - 0.30 Nominal ~94 μm ⑤ Block-1 (Compact WLO Mirror+MLA) 0.50 - 0.60 fused silica or silicon ⑥ upper shell of the housing 0.10 - 0.15 mechanical enclosure total Total system height ~0.90 - 1.22 ≤ 1.50 mm ?

[0114] Table 1 described above quantitatively shows how the optical coupling device (100) of the present invention distributes the height of each stacked component to limit the overall system height. According to Table 1, the solder bump and underfill can be formed to a level of about 0.05 to 0.10 mm, the thin SOI-based PIC die to a level of about 0.08 to 0.18 mm, the MPLA protrusion height to a level of about 0.35 to 0.48 mm, the self-aligned optimized cavity air gap to a level of about 0.10 to 0.30 mm, Block-1 to a level of about 0.50 to 0.60 mm, and the housing top wall to a level of about 0.10 to 0.15 mm, and accordingly, the overall system height can be implemented to a level of about 0.90 to 1.22 mm and can be suppressed to a maximum of 1.50 mm or less. In addition, since the PIC die thickness may vary depending on the thin SOI specifications of the foundry, the MPLA protrusion height is determined by subtracting the trench depth from H_MPLA, and may vary slightly depending on the etching non-uniformity of the DRIE process, it shows that the stacked structure of the present invention is presented as a design range considering process conditions rather than a fixed value.

[0115] Figure 3 is a diagram showing a 7-step beam path as an XZ cross-sectional view (optical propagation plane) of the optical coupling device of Figure 1.

[0116] Specifically, FIG. 3 is a diagram that focuses on how a single-channel optical propagation path from the PIC (150) to the FAU (160) is formed in the XZ plane, rather than the stacked structure or channel array arrangement described in FIG. 1 and FIG. 2. That is, FIG. 3 is a diagram for explaining the optical operation principle of the optical coupling device (100), and can be understood as an embodiment visualizing a series of seven-step beam paths in which light output from the edge coupler of the PIC (150) is transmitted to the connector block (110) and the FAU (160) through the optical function of the MPLA block (120).

[0117] As illustrated in FIG. 3, in the first step, an input light with a relatively small mode field diameter may be output from the edge coupler of the PIC (150). The input light travels in the X direction and may reach a 45° TIR prism (124) formed inside or adjacent to the MPLA block (120) in the second step, thereby redirecting the input light from the horizontal direction to the vertical direction. At this time, the 45° TIR prism (124) may not only simply change the direction of the light but also perform the function of aligning the propagation axis of the light toward the MPLA recess lens (122) so that beam expansion by a subsequent lens system is possible. Subsequently, in the third step, the redirected light may be collimated as it passes through the MPLA recess lens (122) to form an expanded beam (180), and in a preferred embodiment, the diameter of the expanded beam (180) may be designed to be approximately 120 μm to 130 μm.

[0118] Additionally, in the fourth step, the expansion beam (180) may pass through the cavity air gap (130) formed between the MPLA block (120) and the connector block (110). Since the cavity air gap (130) section corresponds to an optically free-space transmission section, maintaining a sufficiently large beam diameter in this section may increase the tolerance for mechanical assembly errors. That is, FIG. 3 indicates that the cavity air gap (130) is the section where the expansion beam (180) is formed most significantly on the single-channel optical path, thereby demonstrating that stable optical transmission can be achieved even in a non-contact detachable interface. In other embodiments, the diameter of the expansion beam (180) may be formed to a level of approximately 80 μm to 100 μm depending on the required insertion loss and alignment tolerance, but FIG. 3 should be understood as representatively illustrating an expansion beam center design that is more advantageous for self-alignment performance.

[0119] Additionally, in the fifth step, the expanded beam (180) passing through the cavity air gap (130) can be refocused by the Block-1 recess lens (112) formed at the bottom of the connector block (110). That is, while the MPLA recess lens (122) performs the role of expanding and collimating the light, the Block-1 recess lens (112) can perform the role of converging the expanded light again and converting it into a beam shape suitable for the subsequent reflection structure and the optical fiber incidence conditions of the FAU (160). Subsequently, in the sixth step, the refocused light can be folded again in a horizontal direction by the Block-1 45° mirror (114) formed on the connector block (110), and accordingly, the light path can finally be aligned with the optical fiber array direction of the FAU (160). Finally, in the seventh step, the light redirected in the horizontal direction can be coupled to the FAU (160) and transmitted to an external optical fiber or an external optical communication path.

[0120] Additionally, FIG. 3 shows that the 7-step beam path is not merely a sequence of beam propagation steps, but forms an optical relay system in which the edge output structure on the PIC (150) side, the prism-lens integrated structure of the MPLA block (120), the expanded beam delivery in the cavity air gap (130), and the lens-mirror refocusing structure of the connector block (110) are sequentially combined. Accordingly, the embodiment illustrated in FIG. 3 can be understood as an embodiment that comprehensively describes, from an optical perspective, H→V→H dual 90° beam folding, self-aligned delivery using the expanded beam (180), and low-loss horizontal coupling to the FAU (160) in a single optical propagation plane. In other embodiments, the radius of curvature of the MPLA recess lens (122) and the Block-1 recess lens (112), the height of the cavity air gap (130), or the reflection shape of the Block-1 45° mirror (114) may be slightly different, but the basic optical propagation concept in which light output from the PIC (150) is magnified in the MPLA block (120), refocused in the connector block (110), and then transmitted to the FAU (160) may be maintained the same.

[0121] FIG. 4 is an XY cross-sectional plan view (detachable interface surface) of the optical coupling device of FIG. 1, showing the double recess cavity geometry, 16-channel array arrangement, DRIE trench details, fiducial marks, and recess lens shape.

[0122] Specifically, FIG. 4 is a drawing showing the detachable interface surface of the optical coupling device (100) in the XY plane, illustrating the planar geometry of the double recess cavity and the channel array arrangement formed with the lower surface of the connector block (110) and the upper surface of the MPLA block (120) facing each other. That is, FIG. 4 is a drawing suitable for explaining how each channel is repeatedly arranged in the XY plane with a certain spacing and arrangement rule, and how the outer mechanical reference structure of the detachable interface is formed in a planar shape, unlike the YZ stacking structure of FIG. 2 or the XZ optical propagation path of FIG. 3.

[0123] As illustrated in FIG. 4, a plurality of Block-1 recess lenses (112) may be repeatedly arranged along the channel arrangement direction on the lower surface of the connector block (110), and a plurality of corresponding MPLA recess lenses (122) may be arranged on the upper surface of the MPLA block (120) according to the same arrangement rule. In a preferred embodiment, the plurality of Block-1 recess lenses (112) and the plurality of MPLA recess lenses (122) may be formed with 16 or more channels, and the pitch between adjacent channels may be formed with 127 μm. Accordingly, a channel array type optical interface that directly corresponds to a multi-channel optical fiber array, rather than a single-channel-centered optical coupling structure, may be implemented. In other embodiments, the number of channels may be increased or decreased, and the channel pitch may also be changed according to the specifications of the applied FAU (160), and accordingly, the Y-direction lengths of the connector block (110) and the MPLA block (120) may also be adjusted.

[0124] Additionally, as illustrated in FIG. 4, a non-etched area surrounding the lens array may be formed on the outer edge of the lower surface of the connector block (110) and the outer edge of the upper surface of the MPLA block (120), and this non-etched area may function as a standoff rail when facing each other. That is, from a planar perspective, the lens array has a structure placed inside the outer frame, and the outer frame may act as a reference plane for mechanical contact and alignment at the detachable interface. Accordingly, the lens array in the central part is responsible for non-contact optical transmission, and the rail structure in the outer part is responsible for maintaining mechanical spacing and suppressing fluid penetration, thereby creating a division of roles. In other embodiments, the width, outer shape, or corner shape of the rail may be changed, and it may be modified into a closed-loop frame, a partially segmented frame, or a plurality of parallel rail structures.

[0125] Additionally, the right detailed view of FIG. 4 illustrates a double recessed cavity and rail contact structure corresponding to the planar interface. According to this, a lens forming portion can be arranged in a recessed state on the opposing surfaces of the connector block (110) and the MPLA block (120), and the outer rails of both blocks can come into contact with each other to form a substantial mechanical reference surface. By adopting a structure in which the outer rail and the internal lens arrangement are separated in the planar view, a cavity air gap (130) is maintained in the central part, and a stable contact reference can be provided in the outer part. In particular, this structure can be extended to a labyrinth seal geometry that prevents direct contact between lens surfaces even during repeated separation and reassembly, and suppresses the penetration of immersion cooling fluid into the internal cavity.

[0126] Additionally, FIG. 4 is also useful for explaining the planar arrangement of the DRIE trench (170) and the corresponding relationship with the channel array. That is, the MPLA block (120) can be positioned at a location corresponding to the DRIE trench (170) formed on the upper part of the PIC (150), and the DRIE trench (170) can be formed to align with the central axis of the channel array. Accordingly, relative position alignment between the array of MPLA recessed lenses (122) repeatedly arranged on the upper surface of the MPLA block (120) and the optical waveguide channel array on the PIC (150) side can be facilitated. In other embodiments, the width, length, and opening shape of the DRIE trench (170) may be changed, and a groove that suppresses the capillary inflow of adhesive may be further formed around the trench opening.

[0127] Additionally, as illustrated in FIG. 4, the flat reference plane (190) / fiducial can be used to determine the reference axis of the channel array. The flat reference plane (190) can function as a mechanical reference that limits the rotational and insertion directions of the connector block (110) and the MPLA block (120), and the fiducial can function as a flat reference mark recognized by a vision alignment device or an inspection device. Accordingly, even in a multi-channel structure expanded to 16 channels or more, the accumulation of lens center axis alignment errors of each channel can be suppressed, and uniform optical coupling precision can be secured for the entire channel array. In other embodiments, the fiducial can be changed to a point, linear, cross, or multi-pattern, and the flat reference plane (190) can also be formed as a multiple reference plane structure rather than a single reference plane.

[0128] In addition, the lower right side of FIG. 4 schematically illustrates an advanced form of the lens shape, suggesting that compared to the initial protruding lens structure, the recessed lens structure can provide a more stable protective effect in a planar interface, and furthermore, that a double recessed structure with both sides recessed is more suitable for a detachable interface. That is, by forming both the Block-1 recessed lens (112) and the MPLA recessed lens (122) by retracting them into each block, the lenses can face each other precisely in a planar alignment state while avoiding physical contact. Accordingly, FIG. 4 can be understood as a drawing indicating that the double recessed cavity structure is not a simple lens arrangement method, but a planar structural solution suitable for a multi-channel detachable optical interface.

[0129] Additionally, in terms of optical design parameters and self-alignment performance, the dual concave inorganic self-aligned optical coupling device (100) for CPO application of the present invention is an optical coupling device for coupling through an expansion beam relay between a silicon photonics chip and a fiber array unit, comprising a first optical element that converts light from a silicon photonics chip into a collimated expansion beam having a beam diameter of 50 μm or more, and a second optical element that is positioned opposite to the first optical element with a free-space air gap between them to refocus the expansion beam, wherein the aforementioned air gap distance is 5% or less of the Rayleigh range z_R = ðw² / ë of the expansion beam, and the aforementioned expansion beam diameter is set to satisfy the relationship Loss(dB) = 8.686 x² / w² to achieve a coupling loss of 1.0 dB or less in a lateral misalignment of ±15 μm, and the first optical element and the second optical element may be mutually detachable in the air gap.

[0130] Here, the first optical element may correspond to an optical section including an MPLA recess lens (122) and a 45° TIR prism (124) provided in the MPLA block (120), and the second optical element may correspond to an optical section including a Block-1 recess lens (112) and a Block-1 45° mirror (114) provided in the connector block (110). Additionally, the free space air gap may correspond to the cavity air gap (130), and the collimated expansion beam may correspond to the expansion beam (180). Accordingly, the dual concave inorganic self-aligned optical coupling device (100) for the application of the CPO of the present invention can be described with respect to the optical relationship of the expansion beam (180) transmitted through the cavity air gap (130) between the optical section on the MPLA block (120) side and the optical section on the connector block (110) side in the drawing.

[0131] Specifically, in terms of optical design parameters and self-alignment performance, the optical coupling device (100) can be understood as focusing on the optical relay relationship between the optical section on the MPLA block (120) side and the optical section on the connector block (110) side. That is, the optical section including the MPLA recess lens (122) and 45° TIR prism (124) provided in the MPLA block (120) can function as a first optical element that receives light output from a silicon photonics chip, redirects it, and converts it into a collimated expanded beam (180). Additionally, the optical section including the Block-1 recess lens (112) and Block-1 45° mirror (114) provided in the connector block (110) can function as a second optical element that receives the expanded beam (180) again, refocuses it, and redirects the light path toward the FAU (160). Therefore, the present embodiment can be understood as a structure in which the optical part on the MPLA block (120) side and the optical part on the connector block (110) side form a single self-aligned optical relay with the cavity air gap (130) in between, rather than the mechanical arrangement of the block itself.

[0132] Additionally, the optical section on the MPLA block (120) may be configured not to directly transmit light with a relatively small mode field diameter output from the edge coupler of the PIC (150), but rather to first convert the light from H to V direction through a 45° TIR prism (124), and then convert the beam into an expanded beam (180) with an enlarged diameter by collimating it through an MPLA recess lens (122). At this time, the diameter of the expanded beam (180) may be formed to be 50 μm or more, and in a preferred embodiment, in a range of 80 to 130 μm, in a more preferred embodiment, in a range of 110 to 130 μm, and in an even more preferred embodiment, at a nominal level of about 125 μm. In addition, the beam radius (w) of the expanded beam (180) may be set to a range of 55 to 65 μm, preferably about 62.5 μm or more. Such values ​​can be set by taking into account the radius of curvature of the MPLA recess lens (122), the mode field diameter of the PIC (150), the operating wavelength, lens manufacturing tolerances, and actual assembly error conditions, and thus can be understood as optical parameters adjustable according to design purposes rather than being limited to a specific single value.

[0133] Additionally, the cavity air gap (130) may not be a simple mechanical gap, but may act as an optical relay section through which the expansion beam (180) is transmitted in a free-space state. At this time, the length of the cavity air gap (130) may be set to be 5% or less of the Rayleigh range z_R = ðw² / ë of the expansion beam (180). In a preferred embodiment, the air gap length may be set to be about 1% of the Rayleigh range, and accordingly, beam divergence in the cavity air gap (130) section is substantially suppressed so that the expansion beam (180) can be transmitted to the connector block (110) side while maintaining a nearly constant beam diameter. For example, when the beam radius is approximately 62.5 μm and the operating wavelength is 1.31 μm to 1.55 μm, the Rayleigh range can be formed to be several mm or more, so the cavity air gap (130) of approximately 100 μm can function as a stable relay section corresponding to 5% or less of the Rayleigh range, more specifically, about 1%. Accordingly, the present embodiment can suppress optical axis shaking or rapid beam diffusion inside the cavity air gap (130) while simultaneously securing the detachable function between blocks.

[0134] Regarding the Rayleigh range analysis, please refer to Table 2 below.

[0135] Beam diameter D Beam radius w Rayleigh range z_R Maximum cavity air gap (≤ z_R) 50 μm 25 μm 1.50 mm ≤ 1.50 mm 80 μm 40 μm 3.84 mm ≤ 3.84 mm 125 μm ★ 62.5 μm 9.36 mm ≤ 9.36 mm 150 μm 75 μm 13.5 mm ≤ 13.5 mm

[0136] In the aforementioned Table 2, the Rayleigh range analysis quantitatively demonstrates that for the beam to maintain a collimated state within the cavity air gap, the cavity air gap distance must be set within the beam's Rayleigh range zR, preferably at a level much smaller than that. According to Table 2, as the beam diameter increases to 50 μm, 80 μm, 125 μm, and 150 μm, the beam radius and Rayleigh range also increase, respectively. In particular, for a diameter of 125 μm, which is the preferred design point of the present invention, the beam radius is calculated to be 62.5 μm and the Rayleigh range to be approximately 9.36 mm. Accordingly, since a cavity air gap of approximately 94 μm is only about 1% of the Rayleigh range, beam divergence in that section is practically negligible, and it can be seen that optical relay can be performed while the expanded beam is stably maintained.

[0137] Additionally, the optical section on the connector block (110) side can be configured to converge and refocus the expanded beam (180) again through the Block-1 recess lens (112). That is, while the MPLA recess lens (122) performs the role of expanding and collimating the light, the Block-1 recess lens (112) performs the role of reducing the expanded light again to convert it into a beam state that can be matched with the optical fiber core of the FAU (160). Subsequently, the refocused light is folded again in the V→H direction by the Block-1 45° mirror (114) to finally form an optical path that is substantially close to a straight line with the optical fiber array direction of the FAU (160). In this way, by the optical section on the MPLA block (120) side and the optical section on the connector block (110) side constituting a single optical relay, the present embodiment can provide a quantitatively designed self-aligned optical coupling structure that includes optical profile control of expansion, transmission, and reduction, rather than a simple opposing coupling between two parts.

[0138] Additionally, this embodiment may be designed to satisfy the Gaussian beam coupling relationship Loss(dB) = 8.686 x² / w². Here, Äx represents the amount of lateral misalignment, and w represents the beam radius of the extension beam (180). In a preferred embodiment, the beam radius w is set to 62.5 μm or more, so that a coupling loss of 0.5 dB or less can be achieved even if a lateral misalignment of ±15 μm occurs. This is not merely intended to form a large beam, but to quantitatively accommodate the misalignment tolerance that may occur during the actual assembly process and on-site reassembly process. Furthermore, as a broader independent design concept, the diameter of the extension beam (180) may be set to achieve a coupling loss of 1.0 dB or less at a lateral misalignment of ±15 μm. Accordingly, the present embodiment may include, on the one hand, an optical performance structure having a wider independent design range, and on the other hand, a more preferred detailed embodiment that secures a beam radius w of 62.5 μm or more to achieve a loss of 0.5 dB or less even with ±15 μm misalignment.

[0139] In addition, this embodiment may also incorporate alternative numerical designs. For example, when the diameter of the expansion beam (180) is formed to a level of approximately 80 μm, it may be designed to achieve a loss of approximately 0.55 dB or less for a lateral misalignment of ±10 μm, and when the diameter of the expansion beam (180) is formed to a level of approximately 100 μm, it may be designed to achieve a loss of approximately 0.51 dB or less for a lateral misalignment of ±12 μm. Accordingly, the present invention is not necessarily limited to only one nominal design point, but may selectively adopt a plurality of self-alignment optimization design points depending on the channel pitch, package budget, wavelength range, loss tolerance, and manufacturing tolerance level. That is, a design achieving a loss of 0.5 dB or less at ±15 μm is suitable for embodiments that prioritize high-precision self-alignment performance, and a design applying alternative beam diameters at the 80 μm or 100 μm level may be suitable for other embodiments that consider block size, Z stacking limit, or optical system simplification.

[0140] In addition, the optical parameter design is further significant when combined with the fact that the MPLA block (120) and the connector block (110) have a mutually detachable structure. That is, when the cavity air gap (130) is set to 5% or less of the Rayleigh range and the beam radius of the extension beam (180) is formed to be sufficiently large, the optical coupling loss may not increase rapidly even if there is a minute positional deviation during the re-coupling process of the blocks. Accordingly, this embodiment can be extended to a structure that simultaneously satisfies optical high-alignment performance and mechanical detachability, and can be applied advantageously in terms of field maintenance, partial replacement, or module reuse. In another embodiment, the optical parameter design is applied together with a mechanical alignment structure based on a flat reference plane / fiducial, so that substantial high-precision optical coupling can be achieved with only manual alignment.

[0141] Additionally, the optical parameter design described in this embodiment may also be linked to specific material combinations of the MPLA block (120) and the connector block (110). For example, in a silicon-based embodiment, a target expansion beam (180) can be formed with a smaller lens radius of curvature due to a high refractive index, and in a fused silica-based embodiment, similar optical performance can be achieved by combining a gentler curvature with a separate reflective coating. Furthermore, the refractive index and thickness of the anti-reflective coating on the lens surface may be adjusted to maintain effective residual reflectance and coupling loss in an immersion cooling fluid environment as well as in an air environment. Thus, the relationship between the first optical element, the second optical element, the cavity air gap (130), and the expansion beam (180) in this embodiment can be understood as a quantitatively designed optical relay structure that goes beyond a simple structural correspondence and takes into account optical loss, self-alignment tolerance, material refractive index, medium environment, and detachment repeatability.

[0142] Figure 5 is a diagram showing the assembly and detachment flow for the optical coupling device of Figure 1.

[0143] In step (a) of FIG. 5, the manufacturing of an MPLA block (120) may be performed. In a preferred embodiment, the MPLA block (120) may be formed by applying a wafer-level optical process on a single inorganic wafer of silicon or fused silica, and in this process, a 45° TIR prism (124) and an MPLA recess lens (122) may be formed integrally. Subsequently, by dicing into individual chips, a plurality of MPLA blocks (120) that can be mounted on a PIC (150) may be provided. In another embodiment, the recess pocket may be formed through a first DRIE process and then the lens shape may be formed through a second WLO process, and the total reflection conditions or anti-reflection coating structure may vary depending on the selection of materials and processes.

[0144] In step (b) of FIG. 5, an MPLA-to-PIC Assembly step may be performed in which an MPLA block (120) is assembled to a PIC (150). At this time, a DRIE trench (170) is formed in advance in the PIC (150), and the MPLA block (120) can be inserted into the DRIE trench (170) by a manual drop-in method and aligned. Additionally, peripheral UV epoxy may be applied around the DRIE trench (170) to mechanically secure the MPLA block (120), and a groove or labyrinth groove formed around the trench opening may prevent the adhesive from flowing into the waveguide cross-section, thereby preventing optical contamination. Subsequently, even with the MPLA block (120) mounted, the PIC (150) can be configured to withstand a Pb-free reflow process at a level of approximately 260°C, so the present invention can be implemented as a structure compatible with a semiconductor package assembly process.

[0145] In step (c) of FIG. 5, assembly of the connector block (110) and the FAU (160) may be performed. In a preferred embodiment, the connector block (110) may be formed to include a Block-1 recessed lens (112) and a Block-1 45° mirror (114) through a separate wafer-level optical process, then diced into individual blocks, and then the FAU (160) may be horizontally bonded to its side. The FAU (160) may be formed as a V-groove-based SMF or PM-SMF array, and may be implemented with a channel pitch of 127 μm and a number of channels of 16 or more. Accordingly, the connector block (110) itself may be prepared as a detachable upper module that simultaneously performs optical coupling and external optical fiber interface functions.

[0146] In step (d) of FIG. 5, a step in which the connector block (110) and the MPLA block (120) are joined together may be performed. In this step, both blocks may be aligned with each other by a flat reference plane / fiducial, and when joined, the standoff rails on both outer sides may come into contact with each other to form a mechanical reference plane. Accordingly, mechanical contact of substantially 0 μm may be achieved at the rail contact portions of both blocks, while a cavity air gap (130) having an apex gap of approximately 94 μm may be formed internally by the central double recess structure. Due to this structure, the expansion beam (180) formed on the MPLA block (120) side is transmitted in a non-contact state through the cavity air gap (130) and subsequently refocused on the connector block (110) side. That is, step (d) of FIG. 5 can be understood not as a simple assembly, but as a step in which mechanical joining and optical relay formation are completed simultaneously.

[0147] In step (e) of FIG. 5, a demate / remate flow for field maintenance is shown. According to an embodiment of the present invention, the connector block (110) can be non-destructively separated from the MPLA block (120), and during this process, the recessed lenses on both sides can be protected from direct contact. In a preferred embodiment, a protective clearance of about 47 μm per lens is secured, so that the possibility of lens damage can be suppressed even during repeated separation and reassembly processes. Therefore, if a malfunction occurs on the FAU (160) or the connector block (110), only the upper module can be replaced while maintaining the MPLA block (120) permanently mounted on the PIC (150), thereby greatly improving field serviceability and repairability.

[0148] As such, FIG. 5 is a diagram illustrating the manufacturing of the MPLA block (120), mounting on the PIC (150), modularization of the connector block (110) and FAU (160), alignment and coupling of both blocks, and on-site detachment and maintenance as a single continuous flow. Accordingly, the embodiment illustrated in FIG. 5 can be understood as a representative process embodiment explaining that the optical coupling device (100) of the present invention is a structure that satisfies wafer-level manufacturability, package process suitability, and maintainability based on the detachment interface.

[0149] Figure 6 is a graph showing misalignment versus loss by beam diameter for the optical coupling device of Figure 1.

[0150] As shown in FIG. 6, the horizontal axis represents the amount of lateral misalignment Äx (μm), and the vertical axis represents the corresponding coupling loss (dB), and each curve can represent cases where the diameter D of the expansion beam (180) is 50 μm, 80 μm, 100 μm, 125 μm, 150 μm, and 200 μm, respectively.

[0151] Specifically, FIG. 6 clearly shows that even under the same misalignment conditions, the rate of increase in coupling loss becomes gentler as the diameter of the expansion beam (180) increases. That is, a beam with a smaller diameter is sensitive to lateral positional deviation, so the loss may increase rapidly, whereas a beam with a larger diameter can maintain a relatively smaller increase in loss even for the same misalignment. Accordingly, the present invention can be designed not merely to magnify light, but to quantitatively secure the self-alignment performance required in a detachable structure by appropriately setting the diameter of the expansion beam (180). In particular, FIG. 6 shows that the selection of the diameter of the expansion beam (180) is not merely a selection of optical size, but a key design variable that defines the allowable positional error during the actual assembly and reassembly processes.

[0152] Additionally, as illustrated in FIG. 6, a lateral misalignment point of ±15 μm can be presented as a practical design reference point. In a preferred embodiment, when the diameter of the extension beam (180) is about 125 μm and the beam radius w is set to about 62.5 μm or more, a loss of about 0.5 dB can be achieved at a misalignment of ±15 μm. This can be set as a preferred design point in the present invention and can be understood as a self-alignment design reference suitable for securing substantial optical coupling performance while maintaining a structural detachable interface. In other words, FIG. 6 supports the fact that forming the diameter of the extension beam (180) to about 125 μm is not merely an example, but corresponds to a practical optimization result for maintaining low loss while allowing a lateral misalignment of ±15 μm.

[0153] FIG. 6 may display a 0.5 dB baseline and a 1.0 dB baseline together, allowing different allowable standards to be selectively applied according to design goals. For example, in embodiments requiring stricter optical performance, a loss standard of 0.5 dB or less may be applied, while in embodiments prioritizing wider manufacturing tolerances or low-cost implementation, a loss standard of 1.0 dB or less may be applied. Accordingly, the present invention does not imply only a single fixed design, but may have a structure expandable to multiple design points according to the loss allowable standards. That is, an extended beam (180) design of 125 μm is suitable for high-performance embodiments that prioritize loss of 0.5 dB or less, and a beam design with a smaller diameter may be applied to other embodiments that allow loss of 1.0 dB or less.

[0154] Additionally, FIG. 6 may indicate that when the diameter is at the level of 50 μm or 80 μm, the loss at the same ±15 μm misalignment may increase significantly. This means that if an excessively small extension beam (180) is selected, it may be difficult to sufficiently absorb assembly errors or repeatable coupling errors of the detachable interface. On the other hand, when the diameter becomes larger at the level of 150 μm or 200 μm, the sensitivity to misalignment may be lower, but in this case, other constraints such as enlargement of the lens aperture size, increase in block width, potential for channel pitch interference, and increase in the total package area may be involved. Therefore, FIG. 6 should be understood not simply as meaning that a larger diameter is always desirable, but as a diagram showing that an appropriate beam diameter should be selected by considering the balance between self-alignment performance and package integration.

[0155] In addition, in a preferred embodiment of the present invention, the design can be based on the Gaussian beam coupling relationship Loss(dB) = 8.686 alignment x² / w². Accordingly, the relationship in which the loss for the same misalignment amount Äx decreases as the beam radius w increases can be quantitatively explained, and FIG. 6 can serve to illustrate this theoretical relationship in the form of an actual design curve. That is, FIG. 6 visually indicates that the self-alignment performance of the optical coupling device (100) is not determined by simple empirical judgment, but is designed according to a mathematical relationship based on Gaussian beam theory. Through such a quantitative design method, the present invention can secure a design basis that simultaneously satisfies structural detachability and optical low-loss characteristics.

[0156] Additionally, FIG. 6 is more meaningful when considered in conjunction with the length of the cavity air gap (130). That is, when the diameter of the extension beam (180) is sufficiently large and, at the same time, the cavity air gap (130) is set to a very short section relative to the Rayleigh range of the beam, the beam diameter can be maintained substantially constant even at the non-contact interface between blocks. Accordingly, the misalignment-loss characteristics of FIG. 6 can be understood not as a simple free-space propagation model, but as an actual application design combined with the dual recessed cavity structure of the present invention. In other words, the present invention can ensure optical stability in the detachable structure by performing both the optimization of the diameter of the extension beam (180) and the optimization of the length of the cavity air gap (130).

[0157] In other embodiments, an extension beam (180) with a diameter of approximately 80 μm can be applied to achieve a loss of approximately 0.55 dB or less in lateral misalignment of ±10 μm, and an extension beam (180) with a diameter of approximately 100 μm can be applied to achieve a loss of approximately 0.51 dB or less in lateral misalignment of ±12 μm. Accordingly, FIG. 6 can be understood as a diagram indicating that the optical coupling device (100) of the present invention is not limited to a single beam diameter but can be extended to various self-alignment design points depending on the target loss, allowable assembly error, channel density, and package constraints. That is, FIG. 6 is significant in that it shows that the preferred nominal design point of the present invention is at the level of 125 μm, while simultaneously providing a design range that can also encompass alternative embodiments having various beam diameters such as 80 μm, 100 μm, 150 μm, and 200 μm.

[0158] Refer to Table 3 below for the previously described self-spirit performance matrix.

[0159] Beam diameter (D=2w) 5 μm ㆍ10 μm 15 μm 20 μm 25 μm 50 μm 0.35 dB 1.39 dB 3.13 dB 5.56 dB 8.69 dB 80 μm 0.14 dB 0.54 dB 1.22 dB 2.17 dB 3.39 dB 100 μm 0.09 dB 0.35 dB 0.78 dB 1.39 dB 2.17 dB 125 μm ★ 0.06 dB 0.22 dB 0.50 dB 0.89 dB 1.39 dB 150 μm 0.04 dB 0.15 dB 0.35 dB 0.62 dB 0.97 dB 200 μm 0.02 dB 0.09 dB 0.20 dB 0.35 dB 0.54 dB

[0160] The aforementioned self-alignment performance matrix compares the coupling loss (dB) calculated for various combinations of beam diameter and lateral misalignment distance, quantitatively demonstrating that the loss decreases at the same misalignment as the beam diameter increases. According to Table 3, when the beam diameter is 50 μm, the loss reaches 3.13 dB at ±15 μm misalignment, but decreases to 1.22 dB at 80 μm, 0.78 dB at 100 μm, 0.50 dB at 125 μm, 0.35 dB at 150 μm, and 0.20 dB at 200 μm. In particular, a diameter of 125 μm is indicated as the design point, and since the loss is maintained at 0.5 dB or less even at ±15 μm misalignment, it is presented as a reference point that satisfies the self-alignment performance objective of the present invention. Therefore, Table 3 supports the fact that setting the expanded beam diameter to approximately 125 μm in the present invention is suitable for simultaneously securing practical misalignment tolerance and low-loss optical coupling in a detachable optical interface.

[0161] Figure 7 is a drawing for comparing and explaining the optical coupling device of Figure 1 with the prior art.

[0162] As illustrated in FIG. 7, the present invention may first employ a structure comprising a double recess cavity of a size of 50 to 300 μm. This double recess cavity is distinguished from a simple gap existing between blocks and may represent a composite interface in which opposing lenses are positioned in a recessed state within each block, a cavity for light transmission is formed in the center, and a mechanical reference or contact structure is formed on the outer edge. Accordingly, the present invention can form a structure suitable for repeated attachment and detachment mechanically, while securing a non-contact transmission section optically. On the other hand, the comparative subjects in FIG. 7 are depicted as not employing such a double recess cavity structure or failing to simultaneously provide the same level of lens protection and attachment / detachment interface structure.

[0163] In addition, the present invention may secondly include a monolithic MPLA and a double 90° beam folding structure. That is, the present invention may be configured such that a prism and lens structure formed in the MPLA block (120) and a lens and reflection structure on the connector block (110) side are coupled to perform two optical path switchings of H→V and V→H. Accordingly, the present invention can efficiently switch the optical path in the Z direction while ultimately realizing horizontal coupling with the FAU (160). While some prior art may include optical path switching using prisms or mirrors, FIG. 7 should be understood as indicating that an integrated configuration combining a monolithic MPLA structure and a double 90° folding, as in the present invention, is not substantially presented in prior art structures.

[0164] In addition, the present invention may include, thirdly, a manual drop-in alignment structure. This may mean a structure that allows the MPLA block (120) to be repeatedly inserted and seated into the DRIE trench (170) of the PIC (150) without active alignment equipment. In particular, the present invention can be extended to enable the manual drop-in alignment to be achieved in conjunction with a trench, reference plane, fiducial, or inclined plane induction structure, thus ensuring excellent compatibility with mass production processes and package assembly processes. FIG. 7 suggests that while prior art may have a simple alignment structure, a permanent bond structure, or a separate alignment-dependent structure, they do not incorporate the concept of manual drop-in alignment as a core feature as the present invention.

[0165] In addition, the present invention may include, fourthly, a Gaussian-optimized telecentric relay structure. That is, the present invention can be designed to ensure self-alignment performance even in a detachable free-space transmission section by quantitatively linking the diameter of the expansion beam (180), the beam radius, the length of the cavity air gap (130), and the allowable misalignment range. This is distinguished from simply widening the beam or performing free-space transmission, and corresponds to a feature that quantitatively reflects the relationship between the misalignment tolerance and the loss at the design stage. FIG. 7 can be understood as indicating that such a quantitatively optimized self-alignment relay is one of the key differentiating features of the present invention.

[0166] In addition, fifth, the present invention may employ an all-inorganic structure. That is, core optical components forming the optical path may be formed from inorganic materials such as silicon or fused silica, and an inorganic anti-reflective coating may be applied as needed. Accordingly, the present invention can secure relatively excellent thermal and chemical stability even in high-temperature processes such as 260°C reflow or immersion cooling environments. Although FIG. 7 indicates that some comparative technologies may also partially possess characteristics similar to an all-inorganic structure, the present invention is distinguished in that it is a structure that integrally satisfies the all-inorganic structure together with other key elements.

[0167] Furthermore, the present invention may include, sixth, a horizontal FAU side coupling structure. That is, the FAU (160) may be coupled horizontally to the side of the connector block (110), thereby enabling connection with an external optical fiber array while minimizing the increase in height of the entire optical interface. This can be seen as a very important configuration in a highly integrated CPO package and can be an advantage that distinguishes it from a simple vertical coupling structure or an on-chip fixed structure. FIG. 7 can be understood as showing that prior art does not simultaneously possess such a horizontal FAU side coupling structure along with other key elements such as the present invention.

[0168] Accordingly, FIG. 7 can be understood as a comparative illustration showing that while each of the prior art technologies may partially include some individual elements or aim for similar purposes, only the present invention simultaneously satisfies multiple key configurations such as a double recess cavity, a monolithic MPLA and double 90° folding, passive drop-in alignment, a Gaussian-optimized telecentric relay, an all-inorganic structure, and horizontal FAU lateral coupling. In other words, FIG. 7 may serve as a representative comparative example illustrating that the present invention is an integrated optical interface in which multiple structural, optical, and packaging features are organically combined, rather than being based on a simple substitution of a single element or a design change.

[0169] Figure 8 is a drawing showing the labyrinth groove placement plane on the upper surface of the PIC substrate for the optical coupling device of Figure 1.

[0170] That is, FIG. 8 can be understood as a planar drawing showing the adhesive flow control and assembly protection structure on top of the PIC (150) rather than the optical propagation path or the block stacking structure itself. The present invention may additionally include one or more grooves surrounding the DRIE trench (170) opening on the upper surface of the chip substrate of the PIC (150), and said grooves may be formed to block the capillary flow of the adhesive so that the adhesive is captured in a reservoir channel before reaching the waveguide cross-section inside the trench.

[0171] As illustrated in FIG. 8, the groove may be formed in the form of a double closed loop consisting of at least an inner groove and an outer groove. In a preferred embodiment, the inner groove may be positioned closer to the outer edge of the DRIE trench (170), and the outer groove may be positioned spaced apart from the outer side of the inner groove. Accordingly, even if the adhesive moves from the outside to the inside, the flow velocity is first reduced in the outer groove and then captured or blocked again in the inner groove, thereby further reducing the possibility of direct inflow into the DRIE trench (170). That is, FIG. 8 can be understood as an embodiment showing a planar arrangement that can be extended not only to a blocking structure by a single groove but also to a multi-defense structure using multiple grooves.

[0172] Additionally, in a preferred embodiment, the depth of each groove may be formed to be about 10 to 20 μm and the width to be about 20 to 30 μm, and the gap between the inner groove and the outer groove may be formed to be about 30 to 50 μm. Additionally, the offset between the outer edge of the DRIE trench (170) and the inner groove may be formed to be about 50 to 80 μm. These values ​​may be adjusted considering the viscosity of the adhesive, the amount applied, the surface tension, the package process conditions, and the usable area on the PIC (150), and are not necessarily limited to a single fixed value. In other embodiments, the depth and width of the groove may be formed smaller or larger, and the gap between grooves or the distance from the DRIE trench (170) may also be changed according to the application process.

[0173] Additionally, FIG. 8 indicates that the groove can be formed in a closed-loop ring shape that completely surrounds the DRIE trench (170). Due to this closed-loop structure, even if adhesive is introduced from one direction, it can be prevented from bypassing through a specific open end and entering the interior of the DRIE trench (170). That is, the groove is not a simple linear groove, but can function as a perimeter-shaped storage channel that protects the entire waveguide array corresponding to the DRIE trench (170) and its lower portion. Accordingly, there is an advantage that the entire channel array can be protected simultaneously, rather than selectively contaminating only the cross-section of a single channel, particularly in multi-channel arrays of 16 channels or more.

[0174] Additionally, as illustrated in FIG. 8, a 16-channel edge coupler array may be correspondingly arranged inside or on one side of the DRIE trench (170), and lower waveguide cross-sections may be repeatedly formed along the inner wall of the DRIE trench (170). Therefore, since there is a possibility that multiple waveguide cross-sections, rather than a single optical waveguide, may be contaminated simultaneously if adhesive penetrates into the DRIE trench (170), the groove structure may be particularly important in a multi-channel optical coupling device. FIG. 8 can be understood as showing a planar layout in which grooves are continuously formed along the trench to protect the front of such a multi-channel array.

[0175] Additionally, the groove structure can be implemented so as not to require a separate additional complex process. In a preferred embodiment, the groove can be formed together with the process of forming the DRIE trench (170) in the same mask layer and the same etching step. Accordingly, the groove structure can be implemented without increasing the number of manufacturing processes or requiring a separate precision alignment process, while providing the additional function of blocking adhesive inflow. In other embodiments, the groove may be additionally formed in a separate microfabrication process, but the method of simultaneous formation in the same process may be advantageous for reducing manufacturing costs and process complexity.

[0176] Additionally, in the embodiment illustrated in FIG. 8, a configuration may be presented in which, even if the adhesive causes capillary flow from the outer edge to the inner edge, the flow is captured in the inner groove and its progression toward the DRIE trench (170) is blocked. This can be understood not merely as a structure that stops the adhesive, but as a structure that dissipates flow energy by providing a storage space where the adhesive can stay.

[0177] Figure 9 is a drawing showing a three-layer immersion cooling defense architecture for the optical coupling device of Figure 1.

[0178] The dual concave inorganic self-aligned optical coupling device (100) for CPO application of the present invention is an optical coupling device for co-packaging optics in an immersion cooling environment, comprising a first silicon optical block including a 45° prism that achieves total reflection without a metal reflective coating, a second silicon optical block including a 45° mirror that achieves total reflection without a metal reflective coating, an inorganic anti-reflective coating formed on the silicon-air interface of the first silicon optical block and the second silicon optical block, and a removable air gap formed between the first silicon optical block and the second silicon optical block, wherein the first silicon optical block is bonded to a silicon photonics chip and withstands 260°C Pb-free reflow, and the second silicon optical block is bonded to a fiber array unit, and the total reflection conditions in the 45° prism and the 45° mirror are maintained even when an immersion cooling liquid with a refractive index of 1.80 or less penetrates the air gap region.

[0179] The first silicon optical block described above may correspond to the MPLA block (120), and the second silicon optical block may correspond to the connector block (110). Additionally, the 45° prism may correspond to the 45° TIR prism (124) formed in the MPLA block (120), and the 45° mirror may correspond to the Block-1 45° mirror (114) formed in the connector block (110). The removable air gap may correspond to the cavity air gap (130), the silicon photonics chip may correspond to the PIC (150), and the fiber array unit may correspond to the FAU (160). Accordingly, the dual concave inorganic self-aligned optical coupling device (100) for CPO applications of the present invention can be described in terms of a silicon-based embodiment with a common structure including an MPLA block (120), a connector block (110), a cavity air gap (130), a PIC (150), and a FAU (160).

[0180] Specifically, FIG. 9 can be understood as a drawing illustrating a three-layer immersion cooling defense architecture that ensures the detachable interface between the connector block (110) and the MPLA block (120) in a silicon-based embodiment remains stable even in an immersion cooling environment. That is, FIG. 9 is not merely a drawing illustrating the presence of a cavity air gap (130), but a drawing describing a structure that ensures optical and mechanical protection functions are secured in stages even when immersion cooling fluid approaches from the outside. In particular, in the embodiment illustrated in FIG. 9, both the connector block (110) and the MPLA block (120) are formed from silicon material, and a double recess structure and a four-sided continuous standoff rail are formed together on the opposing surfaces of both blocks, thereby enabling structural defense that blocks or delays the access of external cooling fluid, and optical defense that maintains total internal reflection conditions even if some cooling fluid approaches.

[0181] Additionally, the first protective layer can be understood as a continuous standoff rail contact structure formed on the outer periphery of the connector block (110) and the MPLA block (120). As illustrated in FIG. 9, a closed-loop standoff rail may be formed along four sides on the outer periphery of both blocks, and when combined, the standoff rails may come into contact with each other to form a continuous periphery contact area with substantially no openings. Accordingly, even if the immersion coolant attempts to flow into the interior from the side or the outer periphery, it may first be mechanically blocked at the standoff rail contact area or the flow path may be extended, making it difficult to reach the central recess cavity directly. That is, the standoff rail in this embodiment may function not as a simple spacing structure, but as a structural seal that forms a 4-side perimeter seal to primarily block access to the coolant. In other embodiments, the width, height, cross-sectional shape, or contact surface roughness of the standoff rail may be adjusted, and accordingly, sealing performance and assembly tolerance allowance may be adjusted together.

[0182] Additionally, the second protective layer can be understood as the cavity air gap (130) itself formed inside the standoff rail. That is, as illustrated in FIG. 9, the inner lens forming portions of both blocks are recessed, and as a result, a closed or semi-closed cavity filled with air can be formed in the central portion. In a preferred embodiment, the recess depth of each block can be formed to a level of approximately 50 μm, and the apex gap between lens vertices can be formed to a level of approximately 94 μm. Accordingly, optically, the expansion beam (180) can pass through the cavity air gap (130) in a non-contact state, and mechanically, direct contact between lenses can be prevented. Furthermore, since the cavity is surrounded by an outer rail contact structure, the external coolant cannot immediately replace the inside of the cavity and can maintain an air state for a certain period of time. That is, the second protective layer can function not merely as a light transmission space, but as a lens protection space and simultaneously as an intermediate buffer layer that delays the inflow of immersion coolant. In other embodiments, the cavity height, apex gap, and recess depth may be changed, but the basic concept of maintaining an internal low-refractive-index medium space in combination with an outer sealing structure may be applied in the same way.

[0183] Additionally, the third defense layer can be understood as a total reflection maintaining structure based on the silicon material itself. In this embodiment, the MPLA block (120) can function as a first silicon optical block bonded to the PIC (150), and the connector block (110) can function as a second silicon optical block bonded to the FAU (160). Furthermore, the 45° TIR prism (124) formed in the MPLA block (120) can perform optical path switching in the H→V direction by total reflection without a metal reflective coating, and the Block-1 45° mirror (114) formed in the connector block (110) can also perform optical path switching in the V→H direction by structural total reflection without a metal reflective coating. At this time, since the refractive index of silicon is very high at approximately 3.48, even if some immersion cooling liquid with a refractive index of 1.80 or less is present in the cavity air gap (130) region or adjacent area, the total reflection condition in the 45° TIR prism (124) and Block-1 45° mirror (114) can be maintained. Therefore, this embodiment can implement multi-layer defense against an immersion cooling environment by first suppressing the inflow of cooling liquid by the standoff rail sealing, secondarily stabilizing the internal environment by the air cavity, and thirdly maintaining the silicon-based total reflection condition itself even if some cooling liquid approaches.

[0184] In addition, in this embodiment, an inorganic anti-reflective coating may be formed on the silicon-air interface of the MPLA block (120) and the connector block (110) together with the silicon-based total reflection structure. In a preferred embodiment, the anti-reflective coating may be formed from a Si₃N₄ series, and its thickness and refractive index may be set to reduce residual reflection in both air environments and submerged coolant environments. Accordingly, this embodiment not only prevents the inflow of coolant but also maintains optical performance by reducing interfacial reflection loss. In other embodiments, the material, thickness, or layer structure of the anti-reflective coating may be adjusted, and an additional coating may be applied not only to the lens surface but also to the incident surface of the MPLA block (120) facing the PIC (150).

[0185] Additionally, FIG. 9 shows an embodiment in which the immersion cooling protection structure simultaneously satisfies optical performance maintenance and mechanical maintainability. That is, the connector block (110) is a detachable block that can be non-destructively separated from the MPLA block (120), and each recess lens can maintain a non-contact state during the separation process. In this case, the outer standoff rail can function as a mechanical reference surface during repeated mate / demate, and the central cavity can maintain lens protection and extended beam transmission functions. Therefore, this embodiment can be understood not as a structure that sacrifices detachability for the sake of immersion cooling, but rather as a structure that is extended to allow for on-site maintenance even in an immersion cooling environment.

[0186] In addition, this embodiment can also ensure compatibility with a 260°C Pb-free reflow process. That is, the MPLA block (120) can be formed to withstand high-temperature reflow while bonded to the PIC (150), and the connector block (110) can subsequently be attached to the MPLA block (120) in a detachable state. Accordingly, this embodiment can be implemented as a silicon-based optical interface structure that takes into account compatibility with the semiconductor package assembly process, long-term reliability against immersion cooling environments, and field detachability. In other embodiments, instead of adopting a silicon-based All-Silicon Variant, some inorganic material combinations may be changed or coating conditions adjusted, but the basic intent of this drawing, which interprets a common structure including the connector block (110), MPLA block (120), cavity air gap (130), PIC (150), and FAU (160) in terms of a silicon-based embodiment, can be maintained.

[0187] FIG. 10 is a drawing showing the bidirectional inclined plane-induced magnetic centering, vertical height determination by inclined plane contact and / or shoulder landing, and selective relief of a non-reference inclined plane for the optical coupling device of FIG. 1.

[0188] As illustrated in FIG. 10, an inclined surface may be formed on the lower part of the trench insertion portion of the MPLA block (120), and a corresponding inclined surface may be formed on the upper part of the DRIE trench (170) on the PIC (150) side. Accordingly, when the MPLA block (120) is vertically dropped in from above to below, the two inclined surfaces act as guide surfaces to each other to naturally guide the MPLA block (120) toward the center, and upon final seating, the alignment state can be stably maintained by wedge-lock. In particular, since the drawings and specifications describe this structure as a taper-guided seating interface (NOT face-lock), this embodiment can be understood as a self-centering seating structure with a guiding function rather than a simple face-fixing structure.

[0189] Additionally, the inclined surface structure can provide a self-centering function in the Y-axis direction, that is, in the channel array direction. Since Y self-centering (bilateral taper) is specified in the drawing and the specification states that Y-axis self-centering and Z-axis height determination are simultaneously achieved by the surface contact of both inclined surfaces, this embodiment can be understood as a structure in which both inclined surfaces act symmetrically to automatically correct the left and right eccentricity of the MPLA block (120). On the other hand, since the position in the X-axis direction can be determined by the contact between the end wall of the pocket-type DRIE trench (170) and the end surface of the MPLA block (120), this embodiment can be extended into an axis-specific role division structure in which the Y-axis is inclined surface-induced self-centering, the X-axis is datum wall alignment, and the Z-axis is height determination by the inclined surface and shoulder.

[0190] Additionally, in this embodiment, one or more non-datum inclined surfaces may be formed to include a clearance or relief area. This allows for a structure where, rather than all inclined surfaces making full contact simultaneously to be sensitive to manufacturing tolerances or particulate contamination, some non-datum inclined surfaces are configured with a margin, while the remaining inclined surfaces performing a reference function are responsible for self-centering. In other words, this embodiment can be understood as a semikinematic passive insertion structure with a higher capacity to absorb manufacturing tolerances than a simple full-constrained structure. In other embodiments, it may instead be implemented as a full-contact type structure where all opposing inclined surfaces form substantially surface contact.

[0191] Additionally, the final landing height in the Z-axis direction may be determined solely by the inclined surface contact, or it may be additionally defined by the shoulder landing structure on the PIC (150) as shown in the drawing. That is, in this embodiment, after the inclined surface guides the descent, the final vertical height may be determined by either the inclined surface contact point or the shoulder contact, or a combination of both. Since this shoulder landing structure can provide a more stable final Z height reference regardless of the inclined surface machining error or wear, it may be advantageous in multi-channel optical coupling structures where optical axis height alignment is important. In another embodiment, the Z height may be determined solely by the inclined surface contact without shoulder landing, and in yet another embodiment, the shoulder may be designed to act as the primary reference and the inclined surface as the secondary reference.

[0192] Additionally, FIG. 10 shows that the present embodiment is a structure suitable for high-speed assembly without active alignment. The drawings and specifications describe that a wide inclined surface inlet accommodates a placement error of ±5 μm level of pick-and-place equipment, and that precise alignment is achieved by induction of the inclined surface upon final placement. Accordingly, assembly time and productivity can be significantly improved by a passive drop-in method compared to active alignment. In other words, the present embodiment is significant not only in terms of optical alignment performance but also in terms of mass production processes, and can be extended to a structure that increases process efficiency, particularly in a CPO package environment where a multi-channel optical coupling device (100) must be repeatedly assembled.

[0193] Accordingly, FIG. 10 can be understood as a drawing illustrating an embodiment that simultaneously implements vertical drop-in assembly, Y-axis self-centering, X-axis reference wall alignment, Z-axis inclined surface / shoulder reference height determination, and tolerance absorption by non-reference inclined surface relief by utilizing the inclined surface structure between the MPLA block (120) and the DRIE trench (170). Furthermore, this embodiment is not limited to a single fixed shape and can be modified in various ways within the scope of maintaining the technical concept of inclined surface-induced self-centering and manual drop-in assembly, even if the inclination angle, relief depth, shoulder position, number of reference surfaces, and contact method are changed.

[0194] And, the dual concave inorganic self-aligned optical coupling device (100) for CPO application of the present invention is an optical coupling device for optical coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a dual recess cavity structure in which an aspherical convex lens received in a first recess on the upper surface of a first optical block and an aspherical convex lens received in a second recess on the lower surface of a second optical block are arranged oppositely with an air gap of 50 to 300 μm; a monolithic MPLA in which a 45° TIR prism (124) and an aspherical convex microlens are integrally formed on a single inorganic substrate to generate a collimated expanded beam of 50 μm or more; an integrated mirror-microlens structure in which a 45° mirror and a microlens are integrally formed; a passive alignment structure in which the MPLA is passively inserted into an etched trench of a silicon photonics chip; and a coupling loss of 0.5 dB or less at ±15 μm misalignment of the expanded beam diameter according to Gaussian beam theory. It may be characterized by including a quasi-telecentric self-aligned beam relay configured to achieve an all-inorganic structure in which all optical components within the optical path are composed of inorganic materials; and a horizontal FAU side alignment structure in which a fiber array unit is horizontally joined to the side of a second optical block, wherein optical coupling is achieved through a double 90° beam folding of a first beam folding (H→V) by the TIR prism of the MPLA and a second beam folding (V→H) by the mirror of the second optical block.

[0195] Here, the first optical block may correspond to the MPLA block (120), and the second optical block may correspond to the connector block (110). Additionally, the aspherical convex lens received in the first recess may correspond to the MPLA recess lens (122) formed in the MPLA block (120), and the aspherical convex lens received in the second recess may correspond to the Block-1 recess lens (112) formed in the connector block (110). The double recess cavity structure may correspond to a structure in which the MPLA recess lens (122) and the Block-1 recess lens (112) are arranged opposite each other with a cavity air gap (130) in between, the monolithic MPLA may correspond to the MPLA block (120), and the 45° TIR prism (124) may correspond to the 45° TIR prism (124). The integrated mirror-microlens structure may correspond to a structure including a Block-1 recessed lens (112) and a Block-1 45° mirror (114) provided in the connector block (110). Furthermore, the passive alignment structure may correspond to a structure in which the MPLA block (120) is mounted in the DRIE trench (170) and an alignment structure by a flat reference plane / fiducial. Additionally, the quasi-telecentric self-aligning beam relay may correspond to an optical relay formed by the MPLA recessed lens (122), cavity air gap (130), Block-1 recessed lens (112), and extension beam (180), and the horizontal FAU side alignment structure may correspond to a structure in which the FAU (160) is horizontally coupled to the connector block (110).

[0196] Specifically, in this embodiment, the optical coupling device (100) can be understood as a structure in which a double recess cavity structure, a monolithic MPLA, an integrated mirror-microlens structure, a passive alignment structure, a quasi-telecentric self-aligning beam relay, an all-inorganic structure, and a horizontal FAU lateral alignment structure are organically combined on a single common platform, rather than a structure in which independent components are simply arranged in parallel. That is, the MPLA block (120), connector block (110), cavity air gap (130), DRIE trench (170), extension beam (180), and FAU (160) are not designed to perform only individual functions, but are interconnected to simultaneously achieve detachability, passive alignment, optical low-loss characteristics, and high-reliability package compatibility. In particular, this embodiment is significant in that it provides an integrated CPO optical interface that is distinguished from conventional simple edge coupling structures through a combination of seven key elements.

[0197] Additionally, the double recess cavity structure may be formed by arranging the MPLA recess lens (122) on the upper surface of the MPLA block (120) and the Block-1 recess lens (112) on the lower surface of the connector block (110) opposite each other with a cavity air gap (130) in between. In a preferred embodiment, the non-etched periphery of each block forms a standoff rail, and a mechanical reference between the blocks is established by the contact of both standoff rails, while maintaining an optical free-space cavity inside. Accordingly, the lens surfaces can maintain a stable optical alignment state without direct contact during assembly and separation. In other embodiments, the recess may be formed by combining DRIE and WLO, or may be formed by a single WLO process without a separate DRIE, and furthermore, an embodiment in which one or both lenses are formed as protruding instead of recessed is also possible. In the protruding embodiment, the optical coupling gap can be determined by the protrusion height and mechanical standoff, and can also be applied by filling a refractive index matching medium.

[0198] Additionally, the monolithic MPLA has a structure in which a 45° TIR prism (124) and an aspherical convex microlens are integrally formed on a single inorganic substrate, thereby constituting the internal optical functional part of the MPLA block (120). Correspondingly, the connector block (110) may have an integrated mirror-microlens structure including a Block-1 recessed lens (112) and a Block-1 45° mirror (114). As the optical functional parts of both blocks are formed correspondingly as a prism-lens and a lens-mirror structure, respectively, on one side, an H→V optical path switching and beam expansion can be performed, and on the other side, beam refocusing and V→H optical path switching can be performed. Therefore, the dual 90° beam folding of this embodiment can be understood not as a simple direction change, but as a structure that simultaneously achieves beam profile control and horizontal FAU (160) coupling. In addition, regarding materials, both fused silica-based and silicon-based embodiments are possible, and in a preferred all-silicon embodiment, it can be implemented to operate solely through structural total reflection without a metal reflective coating.

[0199] Additionally, the manual alignment structure may be formed such that the MPLA block (120) is inserted into the DRIE trench (170) of the PIC (150) in a manual drop-in manner, and the sidewalls, endwalls, and flat reference planes / fiducials of the trench may function as mechanical alignment references. In a preferred embodiment, in addition to the basic manual alignment structure, a wedge-coupled self-aligning structure may be applied to form corresponding inclined surfaces on the lower part of the insertion portion of the MPLA block (120) and the upper part of the DRIE trench (170). In this case, the inclined surfaces may simultaneously provide magnetic centering in the Y-axis direction and height determination in the Z-axis direction, and relief or clearance may be formed on some non-reference inclined surfaces to relieve over-constraint. Thus, the present embodiment is not limited to a simple insertion type structure but can be extended to a reactive manual assembly structure having higher repeatability and tolerance absorption capabilities.

[0200] Additionally, the quasi-telecentric self-aligned beam relay can be formed by the interrelationship of the MPLA recess lens (122), cavity air gap (130), Block-1 recess lens (112), and extension beam (180). In this embodiment, the relationship between the beam diameter, beam radius, misalignment tolerance, and loss can be quantitatively set according to Gaussian beam theory, and at a preferred design point, the beam radius w can be configured to be 62.5 μm or greater so that a coupling loss of 0.5 dB or less is achieved at ±15 μm misalignment. Additionally, the cavity air gap (130) is set to be about 1% of the Rayleigh range so that beam divergence within it can be substantially ignored. In another embodiment, the beam diameter may be set to 80 to 100 μm, taking into account actual manufacturing tolerances and MFD variations of the PIC (150), and in this case, it may be designed to maintain a practical loss range for misalignment of ±10 to ±12 μm. Thus, this embodiment does not mean only a specific single value, but may encompass multiple self-aligned design points depending on the target loss and assembly tolerance conditions.

[0201] In addition, the all-inorganic structure and the horizontal FAU side alignment structure can function as elements that enhance the package suitability of the present embodiment. That is, the MPLA block (120) and the connector block (110) are formed from inorganic materials such as fused silica or silicon, and a Si₃N₄ anti-reflective coating or a metallic reflective coating is optionally applied as needed, thereby maintaining optical and chemical stability even in a 260°C Pb-free reflow and immersion cooling environment. At the same time, the FAU (160) can be permanently bonded horizontally to the side of the connector block (110), so it can be directly connected to an external optical fiber array without further increasing the Z stacking height. Accordingly, the present embodiment can be implemented as an optical coupling device (100) for CPO that simultaneously satisfies ultra-thin system height, multi-channel scalability, detachable interface, and high-reliability environmental responsiveness.

[0202] The combination of seven elements of the aforementioned embodiment can be understood not as a structure in which only one element is selectively adopted, but as a structure in which each element acts complementarily to produce a comprehensive technical effect. For example, sufficient self-alignment performance may not be secured with only a double recess cavity, detachment reliability and maintainability may not be secured with only a Gaussian optimization relay, and it may be difficult to fully explain the ultra-thin package effect resulting from horizontal FAU (160) coupling with only an all-inorganic structure. On the other hand, since the above elements are implemented simultaneously within a common structure, this embodiment enables mass production application with only manual alignment, maintains lens protection and low-loss optical coupling even during repeated mate / demate processes, and can satisfy the Z budget required in high-speed package environments and immersion cooling compatibility. Therefore, it is desirable to understand this embodiment not as a simple juxtaposition of individual elements, but as an integrated embodiment that comprehensively considers the actual usage conditions of optical interconnects for CPO.

[0203] And, FIG. 11 is a flowchart illustrating an optical coupling method according to one embodiment of the present invention.

[0204] As illustrated in FIG. 11, the double recess inorganic self-aligned optical coupling method for CPO applications of the present invention is an optical coupling method between an edge coupler of a silicon photonics chip and a fiber array unit performed by an optical coupling device (100), which can be understood as a process flow configured such that the manufacturing of an MPLA block (120), alignment insertion into a PIC (150), formation of a detachable interface with a connector block (110), and final horizontal coupling to a FAU (160) are interconnected, rather than a procedure of simply sequentially arranging optical components. In particular, the method of this embodiment is significant in that it procedurally demonstrates how the double recess cavity, passive alignment, quasi-telecentric self-aligned relay, and double 90° beam folding, which are structural features of the optical coupling device (100), are implemented in an actual manufacturing and assembly process.

[0205] Additionally, in some embodiments, prior to step (S100), a preparation step may be added in which a wafer-level lithography process is performed on a single inorganic wafer to first form an optical functional part for the MPLA block (120). At this time, the wafer-level lithography process may be configured to integrally form a 45° TIR prism (124) and an aspherical convex microlens on the same substrate, and the ROC of the lens may be set to provide an optical magnification of about 15 to 25 times, taking into account that the edge coupler mode field diameter of the PIC (150) is 3 to 6 μm. Accordingly, the MPLA block (120) may be fabricated in a ready state capable of converting light of a small mode field diameter output from the PIC (150) into a collimated expanded beam (180) of 50 μm or more. In another embodiment, a DRIE process may be combined with the wafer-level lithography process to form the recess depth and lens shape separately, and in yet another embodiment, the recess and lens may be formed together using only a single WLO process.

[0206] Additionally, the process of inserting the MPLA block (120) into the trench of the PIC (150) in step (S100) can be performed not by simple mechanical loading, but in combination with a vision-based non-contact manual insertion structure. That is, the equipment can recognize a flat reference plane / fiducial to determine the approximate position of the MPLA block (120), and then install the MPLA block (120) in a manual drop-in manner using a side wall, reference wall, or inclined plane guidance structure of the DRIE trench (170). At this time, a peripheral UV epoxy glue stop with 260°C reflow resistance can be formed around the trench, thereby reducing the subsequent excessive diffusion of the adhesive and contamination of the waveguide cross-section. In some embodiments, the vision-based alignment and the inclined plane guidance insertion structure may be applied together to ensure repeatable insertion precision without active optical alignment. Such a process can be understood as a methodological feature that implements structural manual alignment in the actual manufacturing stage.

[0207] Additionally, step (S102) can be performed after a 260°C Pb-free reflow on the PIC (150) has been performed prior to the MPLA block (120) being mounted on the PIC (150). That is, in this embodiment, since the MPLA block (120) can function as a permanent mounting block on the PIC (150) side, it is first fixed in a state compatible with the semiconductor package mounting process, and then a connector block (110), which is a detachable upper module, can be positioned opposite it. Accordingly, the thermal stress of the reflow process is absorbed in the permanent bonded state of the MPLA block (120) and the PIC (150), and a structure is provided in which only the connector block (110) side can be selectively separated or re-bonded if field service or replacement is required thereafter. Therefore, step (S102) can be understood not as a simple assembly step, but as a step that simultaneously ensures package reliability and maintainability by distinguishing between the permanent bonded part and the detachable bonded part.

[0208] Additionally, in some embodiments, an additional step of horizontally permanently bonding the FAU (160) to the side of the connector block (110) with epoxy may be performed before, during, or as part thereof, step (S102). In this case, the connector block (110) may be prepared in advance with the Block-1 recessed lens (112) and the Block-1 45° mirror (114), and the FAU (160) may be fixed to the side of the connector block (110) as a V-groove-based SMF or PM-SMF array. Accordingly, the detachable interface is implemented only in the cavity air gap (130) between the connector block (110) and the MPLA block (120), and a non-detachable permanent bond may be maintained between the FAU (160) and the connector block (110). This process configuration allows the entire upper module to be treated as a single replacement unit, which is advantageous for a field-serviceable structure.

[0209] Additionally, in step (S104), a quasi-telecentric self-alignment relay can be formed by the interrelationship between the MPLA recess lens (122), the cavity air gap (130), and the Block-1 recess lens (112). In a preferred embodiment, the relay can be formed with an expansion beam (180) diameter of 50 μm or more, in a more preferred embodiment, it can be configured to form an expansion beam (180) of approximately 80 to 130 μm, and in an even more preferred embodiment, a nominal expansion beam (180) of approximately 125 μm. At this time, since the two lenses are positioned opposite each other with the cavity air gap (130) in between, an optical relay for magnification, transmission, and refocusing can be established while maintaining a non-contact state between the lenses. In other embodiments, the beam diameter or cavity length may be adjusted according to the required loss budget, misalignment tolerance, and channel pitch, but the basic method concept of a self-alignment relay by opposing lenses can be maintained.

[0210] Additionally, in step (S106), H→V folding by the 45° TIR prism (124) of the MPLA block (120) and V→H folding by the Block-1 45° mirror (114) of the connector block (110) are performed sequentially, thereby finally achieving horizontal optical coupling for the FAU (160). That is, the method of the present embodiment is not limited to a procedure of simply transmitting a beam between opposing lenses, but can be configured so that the direction change of the input light, beam expansion, free-space transmission, refocusing, re-direction change, and FAU (160) coupling are implemented as a single continuous 7-step optical coupling procedure. Accordingly, the present embodiment can implement an ultra-thin CPO interface suitable for horizontal FAU (160) placement, and since all major components of the optical path are formed of inorganic materials, high reliability can be maintained even in high-temperature processes and immersion cooling environments. Therefore, the method illustrated in FIG. 11 is not merely a simple sequence of assembly steps, but is preferably understood as an integrated optical coupling method that simultaneously reflects structural features, optical performance, package reliability, and maintainability.

[0211] FIGS. 12 and FIGS. 13 are drawings showing a detachable fiber array unit (DFAU) structure that can be additionally provided in the optical coupling device (100) of the present invention.

[0212] That is, FIG. 12 is an XY plan view of a receptacle (200) of an optical coupling device (100) according to one embodiment of the present invention, showing the arrangement of a three-wall frame structure, a bottom frame, a central window, a bore (223) in which a ball detent (220) is placed, and a clearance hole (231) into which a guide pin (230) is inserted.

[0213] As illustrated in FIG. 12, the optical coupling device (100) may include a receptacle (200) permanently mounted on a silicon photonics chip substrate, and the receptacle (200) may be formed as a frame structure with an open top so that a plug body (210) can be inserted from above. Specifically, the receptacle (200) may be formed as a three-wall frame structure including at least one X-direction wall, two Y-direction walls spaced apart from each other, and a bottom frame connecting them, and a window may be formed in the center so that a lower optical coupling structure can be exposed. Accordingly, the receptacle (200) may function as a receiving frame that mechanically accommodates the plug body (210) while providing an opening for the lower optical coupling area.

[0214] Additionally, the X-wall of the receptacle (200) can be formed with a thickness of approximately 1.5 mm to stably accommodate the ball detent (220) and bore (223) on the X-direction side, and the two Y-walls can also be formed with a thickness of approximately 1.5 mm each. In a preferred embodiment, the bore (223) formed in each Y-wall accommodates the ball detent (220), and can be configured so that a residual thickness of approximately 250 μm is secured on each side of the wall even after the bore is formed. Additionally, the total length of the receptacle (200) in the Y-direction can be formed to a level of approximately 6.2 mm, and accordingly, sufficient mechanical strength and assembly stability can be secured while the two Y-walls, the central window, and the bottom frame structure are arranged together.

[0215] Here, Block-2 accessed through the central window corresponds to the first optical block or MPLA block (120) in this specification, and Block-1 received by the plug body (210) inserted into the receptacle (200) corresponds to the second optical block or connector block (110) in this specification.

[0216] Additionally, one or more clearance holes (231) may be formed in the bottom frame of the receptacle (200). These clearance holes (231) may be spaced apart along a diagonal reference line on the bottom frame, and a corresponding guide pin (230) may also be formed to be diagonally aligned on the lower surface of the plug body (210), thereby improving the stability of the alignment in the X and Y directions when the plug body (210) is inserted.

[0217] The clearance hole (231) may be formed to correspond to the position where a guide pin (230) protruding from the lower surface of the plug body (210) is inserted, and may be formed to have an inner diameter larger than the guide pin (230) and sufficient depth so that the tip of the guide pin (230) does not come into contact with the bottom of the clearance hole (231) even when the plug body (210) is fully seated. Accordingly, it is preferable to understand that the guide pin (230) and the clearance hole (231) are structured to provide only coarse alignment in the X and Y directions and do not participate in position determination in the Z direction. That is, a precise fine alignment feature may not be separately formed between the receptacle (200) and the plug body (210), and the combination of the guide pin (230) and the clearance hole (231) may function as an auxiliary alignment structure for inducing insertion and approximate position alignment.

[0218] Additionally, one or more bores (223) may be formed in the X-direction wall and the Y-direction wall of the receptacle (200), and a ball detent (220) may be accommodated within each bore (223). At this time, the second and third bores (223) formed in each of the two Y-direction walls may be positioned to be aligned on corresponding X coordinates, and accordingly, the holding force of the ball detent (220) acting on both sides of the Y direction may be formed in a balanced manner.

[0219] More specifically, the ball detent (220) may be formed with a structure including a ball (221) and a spring (222) disposed within a bore (223), and the ball (221) may be maintained to protrude toward the inside of the receptacle (200) by the elastic force of the spring (222). With such a configuration, when the plug body (210) is vertically inserted into the receptacle (200), the ball (221) may be secured to a dimple (214) formed on the outer surface of the plug body (210) to provide a detachable mechanical retention. In a preferred embodiment, the bore (223) may be formed with at least one on the X-direction wall and at least one on each of the Y-direction walls to provide retention forces in multiple directions, thereby improving the resistance to tilting or vibration of the plug body (210).

[0220] Additionally, according to the planar arrangement of the receptacle (200) as illustrated in FIG. 12, the force direction of each ball detent (220) can act in a horizontal direction toward the inside from the wall surface of the receptacle (200). Accordingly, the holding force by the ball detent (220) can be provided in a direction perpendicular to the vertical direction of the optical axis where the expansion beam relay between the first optical block and the second optical block is formed, and thus the ball detent (220) can function as an element responsible for the mechanical fastening and holding function of the plug body (210) rather than an element that directly forms optical alignment. That is, the ball detent (220) structure provided in the receptacle (200) can form a holding structure that is mechanically independent of the optical alignment interface, and the fine alignment function can be achieved by the optical coupling structure described later rather than the ball detent (220).

[0221] Meanwhile, since FIG. 12 is a drawing centered on the planar arrangement of the receptacle (200), the specific shape and arrangement of the first region (211), second region (212), stepped wall (213), and dimple (214) of the plug body (210) are understood as the opposing structure inserted into the receptacle (200) rather than as the main subject of illustration directly shown in this drawing. Accordingly, matters such as the window-frame cradle structure of the plug body (210), the distinction between the first region (211) and the second region (212), the interference prevention function of the stepped wall (213), and optical alignment and the formation of an optical coupling gap within the recess cavity through direct contact of the standoff rail can be explained more specifically in the detailed description of FIG. 13 described later.

[0222] As illustrated in FIG. 13, the receptacle (200) may be formed into a frame structure that is permanently mounted on a silicon photonics chip substrate. The receptacle (200) may have a three-wall frame structure including an X-direction wall, two Y-direction walls, and a bottom frame, the bottom may be open to allow access to the MPLA block (120), which is a first optical block mounted in the DRIE trench (170), and the right side may be open for the fiber ribbon exit of the FAU (160).

[0223] The correspondence between Block-1 and Block-2 in FIG. 13 is the same as described in FIG. 12, where Block-2 corresponds to the first optical block or MPLA block (120) and Block-1 corresponds to the second optical block or connector block (110).

[0224] One or more ball detents (220) may be embedded in the wall of the receptacle (200). The ball detent (220) may include a bore (223) formed inside the wall, a hardened steel ball (221) received in the bore (223), and a spring (222) that deflects the ball (221) outward from the bore (223). In a preferred embodiment, the ball (221) may be formed of hardened steel having a hardness of HRC 60 or higher and may be coated with DLC, and the spring (222) may be formed of a SUS316 coil spring to provide a deflection force of 0.5 to 2.0 N. The ball detent (220) may be assembled to the receptacle (200) in a press-fit manner after the completion of the 260°C reflow process, so that thermal damage caused by the high-temperature process can be prevented.

[0225] In addition, in this embodiment, the ball detent (220) can be formed as a three-point holding structure including a first ball detent (220) on the X-direction wall and two second and third ball detents (220) on the Y-direction walls. However, since FIG. 13 is an XZ cross-sectional view, it is preferable to understand that the ball detent (220), ball (221), bore (223), and dimple (214) of the plug body (210) are shown as being representative of the three-point holding structure. The force direction of the ball detent (220) is horizontal and can be orthogonal to the vertical optical axis of the extension beam (180) relay between the first optical block, MPLA block (120), and the second optical block, connector block (110). Accordingly, since the mechanical holding force by the ball detent (220) is separated from the optical alignment interface and the force direction, an alignment-independent holding structure can be realized in which the mechanical holding does not disturb the optical alignment.

[0226] When the plug body (210) is vertically inserted, the guide pin (230) enters the clearance hole (231) to achieve X and Y direction alignment, and the ball (221) is temporarily retracted into the bore (223) by the outer surface of the plug body (210), and the standoff rail of the connector block (110), which is the second optical block, comes into direct contact with the standoff rail of the MPLA block (120), which is the first optical block, thereby forming a Si-Si contact with a mechanical interface gap of substantially 0 μm, and thus a Z-datum and Z-stop can be achieved. Subsequently, the ball (221) is engaged with the corresponding dimple (214) to complete a detachable mechanical retention. At this time, the optical coupling gap is not formed in the standoff rail contact portion itself, but can be formed separately within the recess cavity formed by the recess of the first optical block and the recess of the second optical block. When separating, if a drawing force of about 2 to 4 N is applied in the vertical direction, the ball (221) retracts into the bore (223) along the inclined surface of the dimple (214), so that non-destructive separation can be achieved without direct contact between the lens surfaces.

[0227] Additionally, as illustrated in FIG. 13, the plug body (210) may be formed as a window-frame cradle structure including a first area (211) and a second area (212), and the first area (211) may be formed as Zone A with the top and bottom open to allow optical coupling and rail access of the connector block (110). On the other hand, the second area (212) may be formed as Zone B of a sandwich structure that supports the FAU (160) from above and below by a shelf and a lid. Additionally, the stepped wall (213) of the plug body (210) may be formed to end at the bottom of the connector block (110) or near the rail plane in the first area (211), thereby preventing interference with the upper surface protrusion or T-shape surrounding the MPLA block (120) while maintaining the mechanical support function of the plug body (210).

[0228] In particular, in this embodiment, a sphere, V-groove, or dynamic coupling structure for fine alignment may be absent from the receptacle (200) and the plug body (210). When the diameter of the extension beam (180) of the optical coupling device (100) is approximately 125 μm and the misalignment tolerance is ±15 μm, the optical performance requirements can be satisfied solely by the coarse alignment by the guide pin (230) and the vertical drop-in operation of the plug body (210), so the fine alignment feature may be unnecessary. Accordingly, a separable structure can be implemented in which the ball detent (220) performs only a purely mechanical holding function and optical alignment is achieved entirely by direct contact of the standoff rail. Additionally, since the guide pin (230) maintains a floating state within the clearance hole (231), it does not participate in forming a reference in the Z direction and can perform only auxiliary alignment in the X and Y directions. This alignment-independent retention structure is a unique design principle made possible by the self-alignment characteristics of the expanded beam, and can be distinguished from conventional micro-alignment-dependent retention structures.

[0229] Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention may be implemented in other specific forms without changing its technical concept or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. Industrial applicability

[0230] In addition, the present invention is intended to provide a double-concave inorganic self-aligned optical coupling device and method for CPO applications that can maximize compatibility with mass production processes by reducing reliance on active alignment equipment or complex optical correction procedures during the alignment process of an optical coupling structure and enabling stable assembly with only manual alignment. As such, the invention is industrially applicable as it is not only sufficient for commercialization or business but also clearly feasible in reality. Explanation of the symbols

[0231] 100: Optical coupling device 110 : Connector Block (Block-1) 112 : Block-1 Recessed Lens 114 : Block-1 45° Mirror 120 : MPLA Block (Block-2) 122 : MPLA Recessed Lens 124 : 45° TIR prism 130 : Cavity air gap 140 : Housing 150 : PIC 160 : FAU 170 : DRIE Trench 180 : Expansion beam 190 : Flat reference plane 191 : Fiducial 200 : Receptacle 210: Plug body 211 : Zone A 212 : Zone B 213: Staircase wall 214 : Dimple 220 : Ball detent 221 : Ball 222 : Spring 223 : Bore 230 : Guide pin 231: Clearance Hall

Claims

Claim 1 An optical coupling device for coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a first optical block mounted in a trench of the silicon photonics chip; a second optical block detachably disposed opposite to the first optical block; a first lens and a first beam folding element provided in the first optical block for collimating light from the edge coupler into an expanded beam; and a second lens and a second beam folding element provided in the second optical block for refocusing the expanded beam. A double concave inorganic self-aligning optical coupling device for CPO applications, comprising an optical coupling gap formed between the first lens and the second lens, wherein the first lens and the second lens are recessed within the surface of each corresponding optical block or are positioned oppositely to maintain a mutually non-contact state during coupling and separation, wherein the first beam folding element and the second beam folding element provide double 90° beam folding for the optical path between the edge coupler and the fiber array unit, and wherein all optical components of the optical path between the edge coupler and the fiber array unit are composed of an inorganic material. Claim 2 An optical coupling device for coupling between a silicon photonics chip and a fiber array unit via an expansion beam relay, comprising: a first optical element that converts light from the silicon photonics chip into a collimated expansion beam having a beam diameter of 50 μm or more; and a second optical element disposed opposite to the first optical element with a free-space air gap between them to refocus the expansion beam, wherein the air gap distance is 5% or less of the Rayleigh range z_R = πw² / λ of the expansion beam, where w is the beam radius of the expansion beam and λ is the operating wavelength; wherein the expansion beam diameter satisfies the relationship Loss(dB) = 8.686·Δx² / w², where Δx is the amount of lateral misalignment and w is the beam radius of the expansion beam, and is set to achieve a coupling loss of 1.0 dB or less at a lateral misalignment of ±15 μm; and wherein the first optical element and the second optical element are mutually detachable in the air gap. Mineral self-aligning optical coupling device. Claim 3 A dual concave inorganic self-aligning optical coupling device for CPO applications, comprising: a first silicon optical block including a 45° prism that achieves total reflection without a metal reflective coating; a second silicon optical block including a 45° mirror that achieves total reflection without a metal reflective coating; an inorganic anti-reflective coating formed on the silicon-air interface of the first silicon optical block and the second silicon optical block; and a removable air gap formed between the first silicon optical block and the second silicon optical block, wherein the first silicon optical block is bonded to a silicon photonics chip and withstands 260°C Pb-free reflow, the second silicon optical block is bonded to a fiber array unit, and the total reflection conditions in the 45° prism and the 45° mirror are maintained even when an immersion coolant with a refractive index of 1.80 or less penetrates the air gap region. Claim 4 An optical coupling device for optical coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising: a double recess cavity structure in which an aspherical convex lens received in a first recess on the upper surface of a first optical block and an aspherical convex lens received in a second recess on the lower surface of a second optical block are arranged oppositely with an air gap of 50 to 300 μm; a monolithic MPLA in which a 45° TIR prism and an aspherical convex microlens are integrally formed on a single inorganic substrate to generate a collimated expanded beam of 50 μm or more; an integrated mirror-microlens structure in which a 45° mirror and a microlens are integrally formed; a passive alignment structure in which the MPLA is passively inserted into an etched trench of a silicon photonics chip; a quasi-telecentric self-aligned beam relay in which the expanded beam diameter is set to achieve a coupling loss of 0.5 dB or less at ±15 μm misalignment according to Gaussian beam theory; and an all-inorganic structure in which all optical components within the optical path are composed of inorganic materials. A double concave inorganic self-aligned optical coupling device for CPO applications, characterized by including a horizontal FAU side alignment structure in which a fiber array unit is horizontally bonded to the side of a second optical block, and optical coupling is achieved through a double 90° beam folding of a first beam folding (H→V) by a TIR prism of the MPLA and a second beam folding (V→H) by a mirror of the second optical block. Claim 5 A method for optical coupling between an edge coupler of a silicon photonics chip and a fiber array unit, comprising the steps of: inserting an MPLA block, which includes a monolithic MPLA in which a 45° TIR prism and an aspherical convex microlens are integrally formed on a single inorganic substrate, into a trench of the silicon photonics chip; positioning a connector block, which includes a 45° mirror, opposite the MPLA block so that the lenses of both blocks are positioned oppositely with an optical coupling gap of 50 to 500 μm; forming a quasi-telecentric self-alignment relay by the opposing lenses with an expanded beam diameter of 50 μm or more; and coupling to a horizontal fiber array unit through a double 90° beam folding of H→V folding by the 45° TIR of the MPLA block and V→H folding by the 45° mirror of the connector block, wherein all components of the optical path are composed of inorganic materials. Claim 6 A dual concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, characterized in that the expanded beam diameter is 80 to 130 μm and is determined within the range of 80 to 130 μm by lens manufacturing tolerance and mode field variation of the silicon photonics chip. Claim 7 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, the height of the first optical block is 0.15 to 0.60 mm and the height of the second optical block is 0.40 to 0.70 mm. Claim 8 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 1, the depth of the trench is 50 to 100 μm, the width in the X direction is 100 to 200 μm, and the length in the Y direction is 1,500 to 3,000 μm. Claim 9 A dual concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, characterized by including an array comprising 16 or more channels and a channel pitch of 127 μm. Claim 10 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 1, the inorganic material comprises fused silica (SiO₂), a metal reflective coating comprising aluminum or gold is formed on the 45° beam folding surface, and it has broadband optical transmittance over O-band, C-band, and L-band wavelength ranges. Claim 11 A double concave inorganic self-aligning optical coupling device for CPO applications according to claim 1, wherein the fiber array unit comprises a V-groove based SMF or PM-SMF, is horizontally permanently bonded to the side of the second optical block with epoxy, and the detachable interface is implemented only at the optical coupling gap. Claim 12 A double concave inorganic self-aligning optical coupling device for CPO applications according to claim 1, wherein the first optical block includes a first recess for receiving the first lens, and the second optical block includes a second recess for receiving the second lens, and the first optical block and the second optical block are aligned by a flat reference plane and a fiducial, and the first lens received in the first recess and the second lens received in the second recess maintain a non-contact state during coupling and separation to form a detachable structure. Claim 13 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, the total vertical height of the optical coupling device is 1.50 mm or less. Claim 14 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, a peripheral UV epoxy glue stop for 260°C reflow resistance is formed around the trench. Claim 15 A dual concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, wherein the first optical block is an MPLA, and the MPLA is configured such that a lens array and a 45° prism are integrally formed on a single inorganic substrate by a wafer-level lithography process, and the lens ROC is configured to provide a magnification of 15 to 25 times compared to a PIC MFD (3 to 6 μm) to generate an expanded beam of 50 μm or more. Claim 16 A dual concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, wherein the dual 90° beam folding forms a seven-step optical path of PIC edge coupler output, H→V folding by 45° TIR, MPLA collimation, expanded beam delivery in the detachable cavity air gap, refocusing in the second optical block, V→H folding by 45° mirror, and fiber array unit coupling. Claim 17 A dual concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 1, the expansion beam diameter is 80 μm and a coupling loss of 0.55 dB or less is achieved at a lateral misalignment of 10 μm, or the expansion beam diameter is 100 μm and a coupling loss of 0.51 dB or less is achieved at a lateral misalignment of 12 μm. Claim 18 A double-concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, wherein the silicon photonics chip further comprises one or more grooves surrounding the trench opening on the upper surface of the chip substrate, and the grooves block the capillary flow of the adhesive so that the adhesive is trapped in a reservoir channel before reaching the waveguide cross-section inside the trench, thereby protecting the waveguide cross-section from adhesive contamination during assembly. Claim 19 A dual concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, the refractive index of the anti-reflective coating formed on the lens surface is located between the ideal anti-reflective refractive index for air and the ideal anti-reflective refractive index for immersion cooling fluid, so that the residual reflectance of each coating surface is maintained below a predetermined threshold value in both air and immersion cooling fluid environments. Claim 20 A double concave inorganic self-aligning optical coupling device for CPO applications according to claim 1, wherein the first optical block comprises a first recess for receiving the first lens and a first monolithic standoff rail surrounding the first recess, and the second optical block comprises a second recess for receiving the second lens and a second monolithic standoff rail surrounding the second recess, wherein when coupled, the first monolithic standoff rail and the second monolithic standoff rail are in bidirectional contact to form a labyrinthine sealing geometry that suppresses the penetration of a cooling fluid into the recess cavity formed by the first recess and the second recess. Claim 21 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 1, the first optical block is an MPLA, and an inorganic anti-reflective coating is additionally formed on the silicon photonics chip trench-facing surface of the first optical block to reduce Fresnel reflection on the facing surface. Claim 22 A double concave inorganic magnetic alignment optical coupling device for CPO applications, characterized in that, in claim 1, the first optical block is an MPLA, an inclined surface is formed on the lower part of the trench insertion portion of the first optical block, and a corresponding inclined surface is formed on the upper part of the trench of the silicon photonics chip, so that when vertically dropped in, the inclined surfaces passively guide and seat the first optical block, thereby providing transverse magnetic centering and vertical height determination by wedge coupling. Claim 23 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 22, one or more non-datum inclined surfaces include a clearance or relief area, and the final seating is defined by a part rather than all of the opposing inclined surfaces while manual insertion is guided by the inclined surfaces. Claim 24 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 22, the inclined surfaces form surface contact on substantially all opposing surfaces upon seating. Claim 25 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 22, the vertical landing height is further defined by a shoulder landing structure formed on a silicon photonics chip surface adjacent to the trench. Claim 26 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 2, the air gap distance is 5% or less of the Rayleigh range z_R = πw² / λ of the expansion beam, where w is the beam radius of the expansion beam and λ is the operating wavelength. Claim 27 A dual concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 2, Loss(dB) = 8.686·Δx² / w² is satisfied, where Δx is the amount of lateral misalignment and w is the beam radius of the expansion beam, and the beam radius w is set to 62.5 μm or more to achieve a coupling loss of 0.5 dB or less in lateral misalignment of ±15 μm. Claim 28 A double concave inorganic self-aligned optical coupling device for CPO applications according to claim 3, wherein the first silicon optical block and the second silicon optical block are formed of silicon, the 45° prism and the 45° mirror are implemented without a metal coating by total internal reflection within silicon, and a 164 nm thick Si₃N₄ anti-reflective coating is applied to the silicon-air interface so that residual reflection is less than 0.5%. Claim 29 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 28, the ROC of the silicon microlens is optimized for the change in thermo-optical refractive index at an operating temperature of 105°C and configured to achieve optimal coupling efficiency under full-load operating conditions rather than room-temperature assembly conditions. Claim 30 A double concave inorganic self-aligning optical coupling device for CPO applications according to claim 28, characterized in that it is configured such that a 45° total reflection condition is maintained even when an immersion coolant with a refractive index of 1.80 or less penetrates into the air gap due to the high refractive index of silicon. Claim 31 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 28, the second silicon optical block is formed of silicon, epoxy adhesion with the fiber array unit is provided by a native oxide film on the silicon surface, and epoxy fillet beads are formed on four sides of the fiber array unit bonding line to prevent capillary wicking of the immersion cooling liquid. Claim 32 A dual concave inorganic self-aligned optical coupling method for CPO applications, characterized in that, in claim 5, prior to step (a), the MPLA is fabricated by a wafer-level lithography process, wherein the lens ROC is configured to provide a magnification of 15 to 25 times relative to the PIC MFD (3 to 6 μm). Claim 33 A dual-concave inorganic self-aligned optical coupling method for CPO applications, characterized in that, in claim 5, it further comprises vision-based non-contact manual insertion and peripheral UV epoxy glue stop application. Claim 34 A double concave inorganic self-aligned optical coupling method for CPO applications, characterized in that, in claim 5, step (b) is performed after 260°C Pb-free reflow of a silicon photonics chip in an MPLA mounted state. Claim 35 A double-concave inorganic self-aligned optical coupling method for CPO applications, characterized in that, in claim 5, it further comprises the step of horizontally permanently bonding a fiber array unit to the side of a connector block with epoxy, and the detachable interface is implemented only in an air gap. Claim 36 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, at least one of the first lens and the second lens is formed as a protruding convex lens protruding from the surface of the corresponding optical block, and the optical coupling gap is defined by the protrusion height of the protruding convex lens and a mechanical standoff structure. Claim 37 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 36, an inorganic spacer made of glass, fused silica, or silicon is disposed between the first optical block and the second optical block to define an optical coupling gap between the protruding convex lenses, and the thickness of the spacer is determined by the height of the protruding convex lenses and the target optical coupling gap. Claim 38 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 36, the first optical block comprises a protruding convex lens and the second optical block comprises a recessed convex lens, and the optical coupling gap is determined by the protrusion height of the protruding convex lens, the recess depth of the recessed convex lens, and a mechanical standoff structure. Claim 39 A double-concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 1, the entire optical path through which the optical beam passes is devoid of epoxy, polymer, or resin, and the adhesive is used only in mechanical fixing parts outside the optical beam path. Claim 40 A dual concave inorganic self-aligning optical coupling device for CPO applications according to claim 20, further comprising a receptacle for receiving the first optical block, wherein coarse alignment is achieved by a mechanical guide structure upon insertion of the receptacle, and precise alignment is achieved by contact between the first monolithic standoff rail and the second monolithic standoff rail upon coupling. Claim 41 A double concave inorganic self-aligned optical coupling method for CPO applications, characterized in that, in claim 32, the wafer-level lithography process is performed to form a recess depth of 25 to 30 μm and an aspherical convex lens in a single process without a separate DRIE etching step, and the non-etched periphery forms a standoff rail. Claim 42 A dual concave inorganic self-aligned optical coupling method for CPO applications according to claim 34, wherein the connector block is detachably coupled to the MPLA block after the completion of Pb-free reflow of the silicon photonics chip, and the detachable coupling is achieved without separate active alignment by mechanical contact of standoff rails formed on the MPLA block and the connector block, respectively. Claim 43 A dual concave inorganic self-aligned optical coupling device for CPO applications according to claim 1, characterized in that a plurality of first optical blocks are each mounted in a plurality of trenches of a single silicon photonics chip, or the number of channels of a single first optical block is expanded to 32, 42, 64, 96 or more, and the channel pitch is 127 μm or 80 μm or less. Claim 44 A double concave inorganic self-aligned optical coupling device for CPO applications, characterized in that, in claim 18, the groove is formed by two concentric grooves having a depth of 10 to 20 μm and a width of 20 to 30 μm and spaced 30 to 50 μm apart from each other to form a closing ring surrounding the trench opening. Claim 45 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 20, the first monolithic standoff rail and the second monolithic standoff rail are in contact with a mechanical interface gap of substantially 0 μm when coupled, and the optical coupling gap is formed separately within a recess cavity formed by the first recess and the second recess. Claim 46 A double-concave inorganic self-aligning optical coupling device for CPO applications according to claim 1, further comprising a receptacle permanently mounted on the silicon photonics chip substrate, wherein the receptacle comprises a frame having three or more walls, a bottom opening providing a window into the trench, and one or more spring-loaded ball detents each embedded in the respective wall; a plug body accommodating the second optical block and fiber array unit is detachably received in the receptacle, and the plug body has one or more dimples on its outer surface corresponding to the one or more ball detents; and wherein, upon vertical insertion of the plug body, the spring-loaded ball engages with the dimples to provide a detachable mechanical retention. Claim 47 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 46, the force direction of each ball detent is horizontal and orthogonal to the vertical optical axis of the expansion beam relay between the first optical block and the second optical block. Claim 48 A double concave inorganic self-aligning optical coupling device for CPO applications according to claim 46, wherein the plug body comprises a first region having upper and lower window openings for optical coupling and heat dissipation of the second optical block; and a second region having a sandwich structure having closed upper and lower frame surfaces to hold the fiber array unit substrate; wherein the wall height of the first region has a stepped wall shorter than the wall height of the second region to prevent interference with the upper surface protrusion of the first optical block. Claim 49 A double concave inorganic self-aligning optical coupling device for CPO applications, characterized in that, in claim 46, the receptacle and the plug body are absent from a micro-alignment feature, the optical alignment between the first optical block and the second optical block is achieved by direct contact between standoff rails formed on each of the first optical block and the second optical block, and the ball detent retention is mechanically independent of the optical alignment interface defined by the direct contact between the standoff rails.

Images